INTRODUCTION

Lung cancer is the leading cause of cancer-related deaths worldwide. Despite advances in early detection and chemotherapy, lung cancer still has a 5-year survival <15% (1). It is clear that new agents against lung cancer are urgently needed.

Nonsteroidal anti-inflammatory drugs (NSAIDs) reduce the incidence of various cancers including lung cancer (2,3). Sulindac and sulindac sulfone have been shown to strongly inhibit lung tumorigenesis in mice, reducing lung tumor multiplicities by 52% and 71%, respectively (4,5). However, the long term use of sulindac is associated with significant gastrointestinal toxicity. We have synthesized phospho-sulindac (PS, OXT-328), a novel sulindac derivative, that shows considerable efficacy towards colon cancer and less gastrointestinal toxicity compared to the parent compound (6–10). However, PS is much more hydrophobic as reflected by its higher partition coefficient (logP, 3.91 vs 2.29), resulting in poor solubility in aqueous milieus. This property potentially limits the bioavailability of PS and thus its effective delivery to tumor site(s) in vivo.

Nanocarriers are emerging tools that have demonstrated great potential in the delivery of lipophilic drugs, by i) dissolution of drug in aqueous solution above their solubility limit, ii) stabilization of drug, iii) specific delivery to target site(s) and iv) overcoming multidrug resistance (11). Solid lipid nanoparticles (SLNs) are colloidal carriers which are being increasingly appreciated as alternative drug carrier systems to traditional polymeric nanoparticles (12,13). SLNs consist of a solid lipid core, in which lipophilic therapeutics, such as PS, could be efficiently entrapped and released in a controlled manner. SLNs have been reported to efficiently improve pharmacokinetic profile (14–16), targeted drug delivery to tumor sites and efficacy, while minimizing systemic side effects (17–19).

In this study, we formulated PS in SLNs, and showed that SLN-PS has potent anticancer activity towards non-small cell lung cancer (NSCLC) in vitro and in xenografts in mice. We also report on its pharmacokinetic properties and its novel mechanism of action against lung cancer.

MATERIALS AND METHODS

Reagents

Phospho-sulindac (OXT-328) was a gift from Medicon Pharmaceuticals, Inc, Setauket, NY. Its partition coefficient (LogP) between octanol and water was determined to be 3.91. Stearic acid, lecithin, chloroform, and Myrj59 were purchased from Sigma-Aldrich. Cell culture reagents were purchased from American Type Culture Collection.

SLN-PS Preparation

SLN-PS was prepared using an emulsion/evaporation technique. Briefly, 450 mg Myrj59 was dissolved in 60 ml deionized water. The aqueous solution was heated to 75°C in a water bath under constant stirring at 700 rpm. The organic phase consisted of 380 mg stearic acid, 200 mg lecithin and 300 mg PS dissolved in 20 ml chloroform. The organic phase was gently injected into the aqueous phase with a syringe. The speed of the stirrer was then increased to 1000 rpm. After 2.5 h (when the volume of the emulsion was reduced to about 5–10 ml), 10 ml of ice-cold water was added and stirring continued for another 2 h at 1000 rpm. The resulting suspension was centrifuged at 7,200 g for 1 h and the supernatant was discarded. The pellet was washed twice by re-suspending in deionized water followed by centrifugation to remove the supernatant. The final pellet was re-suspended in deionized water, frozen at −20°C overnight, and freeze-dried.

Characterization of SLN-PS

The morphology of the SLN particles was determined by transmission electron microscopy. One mg SLN-PS was suspended in 2 mL phosphate buffered saline (PBS, pH 7.4) and one drop of the suspension was mounted onto a slide. The shape and size of the particles were then recorded.

To determine the entrapment efficiency of PS, freshly prepared colloidal samples (from the ice-water chilling step) were centrifuged and non-entrapped PS in the supernatant was determined by high-performance liquid chromatography (HPLC). The % entrapment efficiency of PS in SLNs was calculated using the following equation:

$$ \% \,{\text{entrapment}}\,{\text{efficiency}} = \left( {{\text{n1}} - {\text{n2}}} \right)/{\text{n1}} * {1}00\% $$

where n1 = total amount of PS used; n2 = amount of free PS in the supernatant.

To determine the loading of PS in SLNs, 1.0 mg lyophilized SLN-PS was dissolved in 1.0 ml ethanol. After the appropriate dilution, HPLC analysis was performed to determine the percentage drug loading.

Release Profile of PS from SLNs In Vitro

Freshly prepared SLN-PS suspension samples were dialyzed against 100 ml PBS (pH 7.4) or F-12K cell culture medium under constant stirring at 300 rpm at 37°C. At designated time points (Fig. 1), the dialysis buffer was replaced with fresh buffer. The cumulative percentage release of PS from SLNs was measured by HPLC and plotted against time. Triplicate experiments were performed.

Fig. 1
figure 1

SLN-PS: Characterization and effect on cell growth. SLN-PS nanoparticles were evaluated by TEM (upper left); their average size is ∼55 nm. The cumulative release of PS from SLN-PS into PBS (pH 7.4) or F-12K cell culture medium was monitored for 240 h (upper right); drug levels were determined by HPLC as in Methods. The Table summarizes the 24-h IC50 of sulindac, PS and SLN-PS in two lung cancer cell lines.

Full size image

Cell Culture and Cytokinetic Analysis

Human non-small cell lung cancer (A549, wild type p53) and small cell lung cancer (H510, mutant p53) cells were cultured as recommended by ATCC. Cell viability was measured with the MTT assay (Roche Diagnostics) and cell proliferation with the 5-bromo-2′-deoxyuridine (BrdU; BD Immunocytometry Systems) assay, according to the manufacturer’s instructions. Apoptosis and necrosis were assessed by staining cells with Annexin V and propidium iodide and analyzing them by flow cytometry (20).

Uptake and Intracellular Distribution of SLN-PS

A549 cells were treated with 60 μM SLN-PS or PS, harvested at specified time points and washed with ice-cold PBS. Cell lysates were obtained by sonication. To analyze mitochondrial drug levels, mitochondria were isolated using the Mitochondria Isolation Kit (Thermo Scientific). Drug levels were then determined by HPLC and normalized to the protein content. Intracellular localization was also examined by confocal microscopy, using FITC-entrapped SLN-PS. A549 cells cultured in 35 mm glass-bottomed microwell dishes (MatTek, Ashland MA) were treated with 60 μM SLN-PS-FITC or free FITC. At specified time points, the medium was aspirated, monolayers were washed twice with PBS and examined with a two-photon laser scanning confocal microscope (Zeiss LSM 510 META NLO).

Western Blot

Cell lysates were resolved by sodium dodecyl sulfate electrophoresis and transferred onto polyvinylidene difluoride membranes. Bad, Bid, caspase-9, caspase-3, caspase-7, caspase-8, cytochrome c and p-Bad antibodies (Cell Signaling Technology, Danvers, MA) were used. Loading control: β-actin.

Intracellular ROS and Mitochondrial Superoxide

After drug treatment, cells were trypsinized, incubated with 10 μM DCF-DA for 30 min at 37°C, and the fluorescence intensity was analyzed by flow cytometry (FACSCaliber, BD Bioscience). To assay mitochondrial superoxide anion levels, cells were stained with 5 μM MitoSOX Red for 30 min at 37°C and analyzed on a FACSCaliber.

Trx-2 Overexpression

A549 cells seeded in 100-mm dishes were transfected with Trx-2 plasmid or the empty vector for 48 h using Lipofectamine 2000 (Invitrogen) according to the manufacturer’s protocol.

Mitochondrial DNA Depleted Cells (ρ0)

A549 cells depleted of mitochondrial DNA (ρ0) were generated by incubating cells with 200 ng/ml ethidium bromide and 50 μg/ml uridine for 8 weeks as previously described (21).

Xenograft Tumor Model

A549 cells (2.5 × 106) in 100 μl PBS were injected subcutaneously in both left and right flanks of 5–6 week-old female Ncr nude mice (Taconic, Hudson, NY). When the average tumor volume reached 180 mm3, the mice were treated with vehicle (PBS) or SLN-PS (300 mg/kg/day, i.p.) or PS (150 mg/kg, p.o.), for 5 days per week. Dosage of PS and SLN-PS was equivalent to 70% of their respectively maximum tolerated dose. Tumor size was monitored by measuring their length (L) and width (W) with a caliper, and tumor volume was calculated using the following formula: \( L \times W \times \left( {L + W/{2}} \right) \times 0.{56} \). On day 10, an 18-h urine sample was collected from each mouse using individual metabolic cages. The mice were euthanized on day 22, one hour after drug administration. Blood and the major organs were collected. Tumors were divided into halves, one fixed in 10% buffered formalin and the other stored in liquid nitrogen. Xenograft mitochondria were isolated using a Mitochondria Isolation Kit for Tissues (Thermo Scientific).

Immunohistochemistry

Cell death and proliferation were determined in paraffin embedded tissue sections using the terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL) and Ki-67 immunohistochemical staining, respectively, as previously described (20).

Urinary F-2 Isoprostane Analysis

Levels of F2-isoprostane and creatinine in urine of mice were determined by ELISA (Oxford Biomedical Research (Oxford, MA). F2-isoprostane values were normalized to creatinine levels.

Pharmacokinetics and Biodistribution

BALB/c female mice 5 ∼ 6 weeks old (Taconic, Hudson, NY) were treated with a single dose of PS or SLN-PS (50 mg/kg i.p. or p.o.). Mice were euthanized at designated time points and blood samples were collected by cardiac puncture. After centrifugation, plasma was extracted with acetonitrile. Tissues were homogenized, sonicated, and extracted with acetonitrile. The levels of PS and its main metabolites were determined by HPLC. The results were normalized to protein content.

Statistical Analyses

Data are expressed as mean ± S.E.M. Statistical analyses were performed by ANOVA. P values < 0.05 were considered statistically significant.

RESULTS

Characterization of SLN-PS

SLN-PS particles are round in shape, ranging from 20 to 100 nm in diameter (Fig. 1). The entrapment efficiency and drug-loading capacity of SLN-PS were 72 ± 3% and 33 ± 2%, respectively. The release of PS from SLNs, essentially identical in PBS and F-12 cell culture medium, showed a burst during the first 3 h, reflecting the release of drug adsorbed on the surface of nanoparticles. About 42% of the drug was released by 24 h, 75% by 72 h and 98% by 120 h.

SLN-PS Inhibits the Growth of A549 Lung Cancer Cells and Alters Their Cytokinetics

We evaluated the effect of sulindac, PS and SLN-PS on the growth of two human lung cancer cell lines, A549 (non-small cell lung cancer) and H510 (small cell lung cancer). PS inhibited the growth of these cell lines >11-22 fold more potently than sulindac (Fig. 1, table). Cell death appeared to be p53-independent, as the growth inhibitory effect of PS was observed both in the p53 wild-type (A549) and p53 mutant (H510) cell lines. Incorporating PS into SLNs further enhanced its cell growth inhibitory effect, bringing the overall enhancement over sulindac to >44-111 fold. Of note, plain SLNs (no drug loaded) at concentrations equivalent to up to 3xIC50 of SLN-PS did not appreciably affect cell growth.

The growth inhibitory effect of SLN-PS and PS results from their cytokinetic effect (Fig. 2). SLN-PS dose-dependently decreased cell proliferation, reaching a 52% reduction at its IC50 (Fig. 2a). Equimolar concentrations of PS unincorporated in SLNs had a weaker effect on proliferation (25% reduction), while plain SLNs had only a marginal effect (14% reduction). An equipotent concentration of PS (i.e. one that inhibited 50% of cell growth at 24 h) caused only a 37% reduction of proliferation. These findings indicate that PS, when incorporated into SLNs may have a somewhat different mode of action than PS by itself. At 24 h, SLN-PS increased apoptosis 7-fold compared to control (Fig. 2b). Interestingly, the effect of PS on apoptosis was marginal and of plain SLNs was modest, whereas the sum of the two effects was only 2-fold over control. The effect of an equipotent concentration of PS was about half that of SLN-PS.

Fig. 2
figure 2

The effect of SLN-PS on cytokinetics and mediators of apoptosis in A549 lung cancer cells. Cells were treated with SLN-PS or PS (equimolar or equipotent to SLN-PS in terms of cell growth inhibition) or plain SLNs (no drug loaded) equivalent to SLN-PS. a: Cell proliferation was determined by BrdU incorporation after 6 h treatment with the test compounds. Numbers indicate the percentage of cells in S phase. b: Apoptosis was determined by flow cytometry in cells stained with Annexin-V (abscissa) and PI (ordinate). Apoptotic cells are in the right upper quadrant; their proportion of the total is indicated as a percentage. c: Mitochondrial transmembrane potential (ΔΨm) was determined using the molecular marker JC-1; increased JC-1 green fluorescence indicates decreased ΔΨm. d: Immunoblots of various proteins involved in the apoptosis cascade. A549 cells were treated with SLN-PS for the indicated time periods and SLN-PS, SLNs or PS for 4 h. Arrows: products of hydrolytic cleavage of the corresponding (pro)caspase. These findings establish the activation of the intrinsic apoptosis pathway, shown diagrammatically on the right.

Full size image

Taken together, these findings indicate that SLN-PS has a distinctly more pronounced effect on apoptosis than on cell proliferation compared to PS, which may reflect different modes of action between the two formulations of PS.

SLN-PS Activates the Intrinsic Apoptosis Pathway

Mitochondria are central to cell survival and death (22). SLN-PS activated procaspase 9 but not procaspase 8 (Fig. 2d), suggesting that it induced apoptosis through the intrinsic (mitochondrial) pathway. Collapse of mitochondrial transmembrane potential (ΔΨm) is a key event in mitochondria-mediated apoptosis. We evaluated ΔΨm using the molecular marker JC-1; increased JC-1 green fluorescence indicates decreased ΔΨm.

SLN-PS dissipated ΔΨm potently, increasing JC-1 green fluorescence by nearly 13-fold compared to the control (Fig. 2c). In contrast, an equimolar concentration of PS had no effect on ΔΨm, whereas an equipotent concentration of PS increased JC-1 fluorescence by only 2.8-fold. Plain SLNs without PS increased JC-1 fluorescence by 5-fold. SLN-PS is thus considerably more effective in inducing ΔΨm dissipation than PS.

Downstream of ΔΨm dissipation, SLN-PS caused the release of cytochrome c and the subsequent activation of cell death signaling molecules such as caspase 3 and caspase 7. These effects are time-dependent, evident as early as 1 h post treatment; they also manifest the appropriate temporal sequence (Fig. 2d). The lack of an effect by SLN-PS on Bid and Bad (both upstream of caspase 9) suggests that SLN-PS likely targets the mitochondria directly, leading to activation of caspase 9.

SLN-PS Increases the Levels of Intracellular RONS and Mitochondrial Superoxide Anion: Differential Cellular Uptake and Mitochondrial Targeting

We have previously demonstrated that PS and similar agents induce RONS, an effect that may mediate their anticancer activity (23); mitochondria are the main source of cellular RONS (23). Therefore, we explored the effect of SLN-PS and PS on oxidative stress in A549 cells using DCFDA, a general RONS probe (24) and MitoSox Red, a mitochondrial superoxide anion (O2 •) indicator.

Taking into account the effect of plain SLNs, SLN-PS did not generate increased levels of RONS detected by DCFDA probe compared to PS. Interestingly, SLN-PS increased the amount of mitochondrial superoxide anion by 6-fold over that of the control. Mitochondrial superoxide anion induction was 2-fold higher compared to PS, even when the effect of empty SLNs was subtracted (Fig. 3a). Over-expression of Trx-2, a key protein for redox homeostasis in mitochondria (25), abrogated the induction of mitochondrial superoxide anion and apoptosis by SLN-PS (Fig. 3d). This indicates that the increased levels of mitochondrial RONS are critical for the pro-apoptotic effect of SLN-PS.

Fig. 3
figure 3

The effect of SLN-PS on ROS levels and its uptake by A549 cells. a: Left: ROS levels in A549 cells treated as in Methods for 1 h with SLN-PS 1xIC50, SLNs (equivalent) or PS (equimolar or equipotent). Cells were preloaded with the fluorescent molecular probe DCFDA and examined by flow cytometry. Right: A549 cells were treated with SLN-PS 2xIC50, SLNs (equivalent) or PS (equimolar or equipotent) preloaded with MitoSOX Red, which detects mitochondrial O2 • and its fluorescence was determined by flow cytometry. b: SLN-PS was loaded with FITC (green fluorescence) as in Methods and A549 cells were treated with SLN-PS-FITC 2xIC50 for the indicated time periods. The uptake of SLN-PS was monitored by confocal microscopy. Cells were also pretreated with Mitotracker Red (red fluorescence), a mitochondrion-selective fluorescent probe. Yellow/orange color arises when SLN-PS-FITC is present in the mitochondria (co-localisation). c: Left: Concentration of PS in A549 cells treated with equimolar amounts of SLN-PS or PS at 1–4 h following addition of the drug to the culture medium. Right: Mitochondrial concentration of PS in A549 cells treated with SLN-PS or PS for 2 h. Values are mean ± SD (quadruplicate experiments). d: Left: A549 cells overexpressing Trx-2 were treated with SLN-PS 0, 1, 1.5xIC50 preloaded with MitoSOX Red (detects mitochondrial O2 •) and fluorescence was determined by flow cytometry. Middle: Trx-2 overexpressed A549 cells were treated with SLN-PS 0, 1, 1.5xIC50 preloaded with Annexin-V and PI, determined by flow cytometry. Right: MtDNA-less A549 cells were treated with SLN-PS 1.5xIC50 preloaded with Annexin-V and PI, analysed by flow cytometry.

Full size image

Given that, compared to PS, SLN-PS triggers profound mitochondrial oxidative stress and ΔΨm dissipation, we questioned whether SLN encapsulation can facilitate mitochondrial targeting of PS. Confocal microscopy revealed that SLN-PS started to co-localize with the mitochondria 1 h after incubation with A549 cells, but not in the nucleus (Fig. 3b). Determination of drug levels by HPLC also revealed that compared to PS-treated cells, SLN-PS-treated cells accumulated higher levels of PS in whole cells (3-fold) and mitochondria (4-fold). Indeed, in SLN-PS-treated cells the level of PS in mitochondria was 7.6 times higher than that in whole cell. Such preferential accumulation of PS in mitochondria following treatment with SLN-PS may contribute to its pronounced induction of oxidative stress in mitochondria, compared to PS.

To further substantiate the importance of mitochondrial targeting in the induction of apoptosis by SLN-PS, we generated ρ0 A549 cells (depleted of mitochondrial DNA). Consistent with our hypothesis, we found that these cells were significantly less sensitive to the pro-apoptotic effects of SLN-PS, with 60% less cell death compared to A549 cells with the full mitochondrial complement (1.5xIC50; Fig. 3d).

SLN-PS Inhibits the Growth of Human Lung Cancer Xenografts

To assess the in vivo efficacy of SLN-PS against lung cancer, we treated nude mice bearing A549 xenografts with vehicle, SLN-PS or PS, starting when the average tumor volume was ∼180 mm3. At dose equivalent to 70% MTD, SLN-PS inhibited profoundly the growth of A549 xenografts, leading to tumor regression during the first 10 days (Fig. 4). By day 16, compared to controls, SLN-PS reduced tumor growth by 78% (p < 0.005). In contrast, the effect of PS on the growth of A549 xenograft was marginal and did not reach statistical significance.

Fig. 4
figure 4

Xenografts inhibitory effect and in vivo immunostaining. Upper Left: the inhibitory efficacy on tumor growth of SLN-PS and PS. Upper Right: urinary F2-isoprostane levels of nude mice bearing/not bearing A549 xenografts treated with SLN-PS. Middle Left: H&E, Ki-67 and TUNNEL in tissue sections from A549 xenografts. Middle Right: the quantification of expression of Ki-67 and TUNNEL based on the number of positive cells. Lower Right: Drug levels of PS from the A549 xenografts. Lower Left: Drug levels of PS from the mitochondria of A549 xenografts.

Full size image

We next determined cell proliferation and apoptosis in tumor tissue sections. Compared to controls, SLN-PS decreased Ki-67 levels by 71% (p < 0.001) and induced apoptosis, increasing it 5.7-fold (p < 0.01; Fig. 4). In contrast, PS did not significantly induce apoptosis and only inhibited proliferation modestly (from 61% to 52%, p < 0.03). These findings indicate that the profound growth inhibitory effect of SLN-PS is due to its cytokinetic effect.

SLN-PS also significantly induced oxidative stress in vivo. As shown in Fig. 4, urinary level of F2-isoprostane, a marker of oxidative stress, was much higher in SLN-PS-treated group compared to that in the control group. On the other hand, SLN-PS did not increase urinary F2-isoprostane in mice without tumors, suggesting that SLN-PS specifically induced oxidative stress in tumors.

Pharmacokinetics and Biodistribution

Given the remarkable difference in the anti-tumor efficacy between SLN-PS and PS, we assessed their respective pharmacokinetics in mice. In vivo PS is rapidly metabolized into several metabolites, of which quantitatively most important are sulindac, sulindac sulfide and sulindac sulfone (7).

Compared to PS, SLN-PS consistently generated higher levels of PS and its three main metabolites in the blood of mice (Fig. 5). PS showed a secondary peak following SLN-PS dosing (i.p.), suggesting the possibility of enterohepatic circulation. The peak levels (Cmax) of sulindac and sulindac sulfide (>10-fold) after treatment with SLN-PS (i.p.) were much higher than those with PS alone. The total drug levels (AUC0-24h values) obtained with i.p. injection of SLN-PS were 1.8- and 12.4-fold higher than those with oral gavage and i.p. injection of PS, respectively. Our results suggest that SLN encapsulation significantly improves the bioavailability of PS.

Fig. 5
figure 5

Pharmacokinetics of PS and its metabolites in mouse plasma and its biodistribution after treatment with PS or SLN-PS. (a) Pharmacokinetic curves of PS and its metabolites after treatment with PS or SLN-PS via p.o. or i.p. in 24 h. (b) AUC value of PS and its metabolites after treatment with PS or SLN-PS via p.o. or i.p. (c) Drug level of PS and its metabolites in different organs after treatment with SLN-PS via i.p. ( Biodistribution of PS and its metabolites after treatment with PS is shown in Supplementary Material Fig. S1).

Full size image

We also compared biodistribution of SLN-PS and PS in the A549 xenografts and major organs in mice (Fig. 5 and Supplementary Material Fig. S1). Intact PS and PS-sulfone were detected in the A549 xenografts treated with SLN-PS, however, these intact forms were not found after PS administration. Notably, we observed the specific accumulation of intact PS in tumor mitochondria following treatment with SLN-PS (35-fold higher than treatment with PS), suggesting that SLN encapsulation facilitates mitochondrial targeting of PS in vivo. Apart from intact PS, the levels of sulindac and sulindac sulfide in A549 xenografts were also much higher (∼14-fold) after the administration of SLN-PS. SLN-PS also enhanced drug distribution to the lung and liver, reflecting the higher bioavailability of SLN-PS compared to PS. These data are also in agreement with the in vitro cellular uptake study which showed that intracellular levels achieved by SLN-PS were higher than those by PS, even as early as 1 h post dosing (Fig. 3).

The tumor levels of sulindac, sulindac sulfide and sulindac sulfone were significantly associated with tumor volume (p = 0.003 to 0.025; Fig. 6), cell proliferation (p = 0.0001 to 0.017; Supplementary Material Fig. S2) and apoptosis (p = 0.038 to 0.129, Supplementary Material Fig. S2), with the exception of the correlation between sulindac sulfone and apoptosis. These findings suggest that enhanced efficacy of SLN-PS is closely associated with the improved delivery of PS and its metabolites to the A549 tumors.

Fig. 6
figure 6

Correlation between antitumor effect with tumor drug level and Ki-67 or TUNEL positive cell ratio. Upper panel, sulindac, s. sulfide and s. sulfone tumor levels correlations with tumor volume, all of them have significant correlations with antitumor effect; Middle panel, ratio of TUNEL positive cell is adversely correlated with tumor volume; Lower panel, ratio of Ki-67 positive cells is significantly correlated with tumor volume.

Full size image

DISCUSSION

Our work establishes that incorporating PS in SLNs enhances its efficacy against human lung cancer in pre-clinical models. The superior efficacy of SLN-PS over PS in vitro and in vivo is due to a constellation of effects: a) a remarkable improvement of PK parameters leading to enhanced delivery of PS to the cancer tissue; b) targeting of SLN-PS to the mitochondria, the master regulator of cell death; and 3) a favorable cytokinetic effect dominated by the activation of the intrinsic apoptosis pathway.

SLN-PS was a much more potent inhibitor of human lung cancer cells than PS and sulindac (over 4- and 44-fold enhancement, respectively), and it exerted a strong inhibitory effect in A549 xenografts. Whereas free PS did not significantly inhibit tumor growth, SLN-PS at an equivalent dose was remarkably effective, reducing tumor volume by 78%. A key finding of our study is that SLN-PS exhibits superior PK properties compared to free PS. The phospho-modification of sulindac significantly impacted its physicochemical properties. Most relevant among them is that it increased substantially its hydrophobicity, i.e. it decreased further its already poor water solubility. SLNs, versatile, safe and biocompatible carriers (26), overcame this limitation of PS by solubilizing it, which improved its bioavailability. Incorporation of PS into SLNs led to much higher peak concentrations and AUCs0-24h of PS and its metabolites. More importantly, SLN-PS enhanced by 14-fold the delivery of PS and its metabolites to lung cancer xenografts. In addition to increased bioavailability, it is conceivable that, due to their size (<100 nm) SLN-PS nanoparticles gained preferential access to tumors through the enhanced permeability and retention effect (“EPR effect” (27)). The higher levels of PS and its metabolites in A549 tumors were consequential, as they correlated with reduced tumor volume, reduced cell proliferation and increased apoptosis. These findings support the idea that the improved delivery of PS to the A549 tumors as a result of its incorporation in SLNs contributes to its efficacy against lung cancer in our tumor model.

Our work also reveals that SLNs potently enhanced the mitochondrial targeting of PS, an effect primarily responsible for the induction of apoptosis in vitro and in vivo. This is supported by the finding that the SLN formulation increased 35-fold the content of PS in the mitochondria of lung cancer xenografts compared to those from free PS. Enhanced access of PS to mitochondria by SLN formulation occurred in a cell culture system as well, a finding that, in addition, offered a mechanistic window. When we compared free PS to SLN-PS, we obtained two informative sets of responses: although PS alone exhibited preferential accumulation in the mitochondria, SLN-PS generated much higher mitochondrial PS levels and higher rates of apoptosis compared to equimolar amounts of free PS, suggesting that the accumulation of PS in the mitochondria did affect the fate of the cell. This notion was supported by the finding that equipotent (in terms of cytotoxicity) amounts of free PS and SLN-PS generated both similar mitochondrial levels of PS and similar rates of apoptosis. The critical role of mitochondria targeting of PS by SLN-PS in the induction of apoptosis is further confirmed by the observation that mitochondria-deficient (ρ0) cells were resistant to the cytokinetic effects of SLN-PS.

Mitochondria are a major source of oxidative stress (28) and a central regulator of the intrinsic apoptosis pathway (29). Our data document that, once SLN delivered PS to the mitochondria, PS triggered ROS-dependent cell death. This effect was reflected in a cascade of physicochemical (ROS induction and ΔΨm dissipation) and molecular (sequential activation of caspases) changes that culminated in apoptotic cell death. Underlying these changes was the significant induction of superoxide anion levels in the mitochondria in response to PS. A key player here was Trx-2, the Trx isoform predominantly expressed in mitochondria. Trx-2 is part of the Trx system that participates in antioxidant defense; its other members are NADPH and Trx reductase (30,31). Trx-2 plays a crucial role in the scavenging of mitochondrial ROS. In fact, Trx-2 overexpression is cytoprotective against oxidative stress-induced cell death (25,32), whereas its genetic ablation results in massive apoptosis (31). Consistent with this notion, overexpression of Trx-2 in A549 cells completely suppressed the levels of mitochondrial superoxide anion induced by SLN-PS, which was accompanied by dramatically reduced apoptosis. Trx-2 differs markedly in this regard from Trx-1, the other mammalian isoform of Trx, which can enhance or suppress apoptosis depending on biological context (33). Indeed, we have previously demonstrated the pro-apoptotic role of Trx-1 in response to PS (34). Hence, depending on the cellular location that PS is delivered to, the two Trx isoforms may have completely different roles in response to PS induction of cell death. Apart from cellular compartmentation, other factors such as the redox status of the active site cysteines may also affect the dynamics between Trx and PS (35).

Our data also establish that SLN-PS induced oxidative stress specifically in tumors in vivo, by assaying urinary levels of F2-isoprostane. Urinary levels of F2-isoprostanes, derived from non-enzymatic oxidation of arachidonic acid, provide an accurate assessment of oxidative stress in vivo (36). SLN-PS-mediated induction of ROS in tumors thus leads to selective killing of cancer cells.

CONCLUSION

Our findings suggest that SLNs are an effective drug delivery system for PS, accounting for the observed therapeutic efficacy of PS against human lung cancer xenografts. The SLN formulation not only improved the PK profile of PS, but was also mitochondrial targeting, which triggered cell death. Given the efficacy of SLN-PS, we speculate that PS, when appropriately formulated, is a promising agent for the treatment of lung cancer that merits further evaluation.