The novel agent phospho-glycerol-ibuprofen-amide (MDC-330) inhibits glioblastoma growth in mice: an effect mediated by cyclin D1

Summary

Glioblastoma multiforme (GBM) urgently needs new therapeutic strategies. The novel compound phospho-glycerol-ibuprofen-amide (PGIA) is effective, selective toward GBM cells, and cyclin D1 represents a key molecular target. When formulated in polymeric nanoparticles, PGIA crosses the blood-brain barrier, reaching the target tissue.

Abstract

Given that glioblastoma multiforme (GBM) is associated with poor prognosis, new agents are urgently needed. We developed phospho-glycerol-ibuprofen-amide (PGIA), a novel ibuprofen derivative, and evaluated its safety and efficacy in preclinical models of GBM, and its mechanism of action using human GBM cells and animal tumor models. Furthermore, we explored whether formulating PGIA in polymeric nanoparticles could enhance its levels in the brain. PGIA was 3.7- to 5.1-fold more potent than ibuprofen in suppressing the growth of human GBM cell lines. PGIA 0.75× IC₅₀ inhibited cell proliferation by 91 and 87% in human LN-229 and U87-MG GBM cells, respectively, and induced strong G₁/S arrest. In vivo, compared with control, PGIA reduced U118-MG and U87-MG xenograft growth by 77 and 56%, respectively (P < 0.05), and was >2-fold more efficacious than ibuprofen. Normal human astrocytes were resistant to PGIA, indicating selectivity. Mechanistically, PGIA reduced cyclin D1 levels in a time- and concentration-dependent manner in GBM cells and in xenografts. PGIA induced cyclin D1 degradation via the proteasome pathway and induced dephosphorylation of GSK3β, which was required for cyclin D1 turnover. Furthermore, cyclin D1 overexpression rescued GBM cells from the cell growth inhibition by PGIA. Moreover, the formulation of PGIA in poly-(l)-lactic acid poly(ethylene glycol) polymeric nanoparticles improved its pharmacokinetics in mice, delivering PGIA to the brain. PGIA displays strong efficacy against GBM, crosses the blood-brain barrier when properly formulated, reaching the target tissue, and establishes cyclin D1 as an important molecular target. Thus, PGIA merits further evaluation as a potential therapeutic option for GBM.

Introduction

Glioblastoma multiforme (GBM) is the most common and lethal among all gliomas. The current standard of care for GBM is maximal surgical resection followed by radiotherapy and concurrent temozolomide, followed by adjuvant temozolomide (1,2). However, despite recent progress, GBM continues to be associated with a 5-year survival rate <10% (2,3). Thus, there is an urgent need to develop new compounds against GBM.

The use of non-steroidal anti-inflammatory drugs (NSAIDs) is associated with reduced incidence of various human cancers (4). Case–control studies show an inverse association between NSAID’s use, including ibuprofen, and GBM (5). Furthermore, ibuprofen can significantly reduce tumor growth in rat models of glioma (6) and enhance the cytotoxic effects of doxorubicin and vincristine in human malignant glioma cells (7). Similarly, aspirin can enhance the efficacy of temozolomide on human GBM xenografts in mice (8). These data suggest that NSAIDs may be considered as a therapeutic option for GBM treatment.

However, the application of NSAIDs to cancer is hampered by their limited efficacy and significant toxicity, which includes primarily gastrointestinal and renal side effects (9). Prompted by these considerations, we explored approaches to enhance their efficacy and limit their side effects. Therefore, we synthesized phospho-glycerol-ibuprofen-amide (PGIA; Figure 1A), a novel ibuprofen derivative, which seems to meet the dual goal of increased efficacy and reduced toxicity. Of note, the enhanced safety of PGIA is attributed, in part, to its chemical modification because the carboxylic group, present in nearly all NSAIDs, mediates much of their gastrointestinal toxicity (10).

Figure 1.

PGIA inhibits GBM cell growth. (A) Chemical structure of PGIA (MDC-330). (B) IC₅₀ values for glioblastoma cells treated with PGIA or ibuprofen for 24h. These values are representative of three experiments, each performed in triplicates; results were within 10%. (C) Differential cytotoxic effect of PGIA in GBM cells compared with NHA. Cell growth was determined after treatment with escalating concentrations of PGIA for 48h. Results are expressed as % control.

A major hurdle in GBM treatment is the poor access of chemotherapeutic drugs to the brain. To overcome this limitation, over the past decade, there has been an increasing interest in using nanotechnology for drug delivery (11,12). Among these, polymeric-based drug delivery systems, including the poly-(l)-lactic acid (PLLA) polymers, are being developed to improve the diagnosis and treatment of various diseases, including cancer (13). PLLA polymers are of particular interest because they are biodegradable, biocompatible and US Food and Drug Administration-approved for parenteral drug delivery (14). A major breakthrough in the nanoparticle field is the use of hydrophilic polymers, for example poly(ethylene glycol) (PEG), to efficiently coat conventional nanoparticle surfaces (15). Amphiphilic copolymers with PEG, such as PLLA–PEG, form a protective hydrophilic and flexible corona around the polymeric core of the nanoparticles, and these ‘stealth’ nanoparticles repel plasma proteins, avoid opsonization and exhibit prolonged circulation (15). Moreover, coating these PLLA–PEG nanoparticles with polysorbate-80 increases the levels of drugs in the brain (11).

Cyclin D1, a component of the core cell cycle machinery, functions as a cyclin-dependent kinase (CDK) activator (16). Recently, cyclin D1 has been recognized as a proto-oncogene, with evidence indicating that its increased expression contributes to the loss of cell cycle control in many human tumors. Indeed, cyclin D1 levels are abnormally high in many human cancers, including GBM. Specifically, cyclin D1 expression, greatly increased in grade IV astrocytomas, is correlated with poor survival rates (17). Furthermore, knockdown of cyclin d1 induces apoptosis and attenuates cell proliferation and invasive capacity, effects that were reversed when it was overexpressed (18). These data indicate that cyclin D1 represents a potential target for GBM treatment.

In this study, we examined, for the first time, the efficacy of the novel agent PGIA in preclinical models of GBM, its mode of action, and demonstrated that its formulation in polymeric nanoparticles ensures its delivery to the brain. Our data show that PGIA strongly inhibited the growth of human GBM murine xenografts and reduced cyclin D1 levels leading to strong inhibition of GBM cell proliferation.

Materials and methods

Reagents

PGIA (MDC-330) was a gift from Medicon Pharmaceuticals Inc. (Stony Brook, NY). PGIA was prepared as a 400mM stock solution in dimethyl sulfoxide. Ibuprofen was purchased from Sigma–Aldrich (St Louis, MO). Annexin V was purchased from Invitrogen (Carlsbad, CA). All general reagents were of high-performance liquid chromatography (HPLC) grade or the highest grade commercially available.

Cell culture and cell growth assay

A panel of four human GBM (U87-MG, LN-18, LN-229 and U118-MG) cell lines were obtained from the American Type Culture Collection (Manassas, VA). American Type Culture Collection characterizes these cell lines using cytogenetic analysis. These cell lines were grown as monolayers in the specific medium suggested by American Type Culture Collection and supplemented with 10% fetal bovine serum, penicillin (50U/ml) and streptomycin (50 µg/ml), and frozen for future use. Normal human astrocytes (NHA) were obtained from Lonza (Walkersville, MD) and cultured in astrocyte basal growth medium, supplemented with containing 25 µg/ml bovine insulin, 20ng/ml epidermal growth factor, 5% fetal bovine serum, 20ng/ml progesterone and 50 µg/ml transferrin. Before experiments, all cell lines were characterized for cell morphology and growth rate. All experiments were performed within 3 months after cells were thawed. We did not perform additional cell line characterization; however, we routinely test for mycoplasma contamination in every cell line every 3 months. Cell growth was determined in GBM cells and NHA treated with PGIA for 24 or 48h, using the MTT assay (Sigma–Aldrich) (19).

Cell proliferation assay

Following treatment with PGIA 0.5×, 0.75× or 1× IC₅₀ for 24h, cell proliferation was assayed by 5′-bromo-2′-deoxy-uridine labeling, as described previously (19).

Cell cycle analysis

Following PGIA treatment, cell cycle was determined by flow cytometry (19). The percentage of cells in G₀/G₁, S and G₂/M was determined from the DNA content histograms.

Apoptosis

To measure apoptosis, GBM cells and NHA (1.0×10⁵ cells/well) were treated with various concentrations of PGIA for 24h. Apoptosis was determined by staining with annexin V-FITC and propidium iodide and analyzing the fluorescence intensities by FACScaliber (BD Bioscience).

Western blot and immunoprecipitation

Following treatment with PGIA, cells were collected, and total cell fractions obtained with RIPA buffer and western blot analysis were performed as described previously (20). The following antibodies were used [all from Cell Signaling Technologies (Danvers, MA)]: p-Cyclin D1 (Catalog # 3300), cyclin D1 (Catalog # 2978), p-GSK (Catalog # 5558), GSK (Catalog # 5676), cyclin D3 (Catalog # 2936), cyclin B1 (Catalog # 4135) and p21 (Catalog # 2946). β-Actin (Catalog # A1978 from Sigma–Aldrich) was used as a loading control.

Co-immunoprecipitation analysis was performed as described previously (20).

Data mining analysis of CCND1

A search of genome-wide expression study with both glioblastoma and non-tumor brain tissue yielded two independent data sets. NCBI GEO GSE4290 data set contains 23 non-tumor samples and 81 glioblastomas. Normalized expression values from the downloading sites were transformed to linear scale and used for statistical analysis. TCGA GBM Affymetrix U133A data set (2013-12-18 freeze) contains 10 normal tissues and 529 primary glioblastoma tumors. Two-tailed Welch’s t-test and false discovery rate calculation were performed using R and samr bioconductor package. In TCGA GBM data set, 522 glioblastoma samples have expression data information available.

Immunohistochemistry

Immunohistochemical (IHC) staining for Ki-67, p21, p-AKT and cyclin D1 was performed as described previously (21). Scoring: at least five fields per sample (at magnification ×200) were scored. We calculated percentage of positive cells (brown staining) by dividing the number of labeled cells by the number of cells in each field and multiplying by 100.

Overexpression of cyclin D1

Cyclin D1 plasmid (item # EX-B0078-Lv105-10) was purchased from GeneCopoeia (Rockville, MD). Stable transfection was performed by lentiviral infection, following the manufacturer’s instructions. Cells were then selected by cell sorting (GFP expression) and followed by treatment with puromycin for 3 weeks. Cells were continuously grown in the presence of puromycin after selection.

Animal studies

All animal studies were approved by our Institutional Animal Care and Use Committee.

Efficacy study in nude mouse xenografts

Female immune-deficient BALB/c nude mice at 6 weeks of age were purchased from Charles River Laboratories (Wilmington, MA). In a first study, mice were injected bilaterally, subcutaneously with 5×10⁶ U118-MG cells/site in 0.1ml sterile phosphate-buffered saline (PBS). When the resultant tumors reached ~500mm³, mice were divided randomly into three groups (n = 8/group) and treated with vehicle control (PBS), PGIA (20mg/kg) or ibuprofen (20mg/kg), in 100 µl PBS given i.p. 1×/day, 5 day/week for 29 days. To rule out cell-specific effect, in a second study, mice were injected bilaterally, subcutaneously with 1.5×10⁶ U87-MG cells/site in 0.1ml sterile PBS. When the resultant tumors became palpable, mice were divided randomly into two groups (n = 7/group), treated with vehicle control (PBS) or PGIA (20mg/kg), in 100 µl PBS given i.p. 1×/day, 5 day/week for 12 days. Body weight was determined once weekly and tumor size twice weekly. Tumor size was calculated by the formula: [length × width × (length + width/2) × 0.56] in cubic millimeter.

Brain tissue levels in mice

PGIA (10mg/kg), either reconstituted in 2-hydroxypropyl-β-cyclodextrins or formulated in PLLA(5k)–PEG(2k) nanoparticles (described below) was administered to female BALB/C mice as a single dose by intravenous tail vein injection (100 µl total volume injected). Mice were sacrificed at various time points after drug administration, and blood and brain were collected. PGIA level in the brain was assayed by HPLC as described below.

Preparation and characterization of polymeric nanoparticles containing PGIA

We used the single emulsification and solvent evaporation method to prepare the PLLA(5k)–PEG(2k) polymeric nanoparticles from each respective polymer (22). These nanoparticles were further coated with polysorbate-80 by suspending them in a solution of 1.5% polysorbate-80 for 2h at 37°C. Following the adsorption of polysorbate-80 on the surface on the nanoparticles, these were centrifuged and then suspended in water.

The size and ζ (zeta) potential of the nanoparticles were determined using dynamic light scattering and microelectrophoresis, respectively, using a Zeta-Plus Brookhaven instrument (Holtsville, NY). Entrapped PGIA was quantified with HPLC [Waters Alliance 2695 equipped with a Waters 2998 photodiode array detector (220nm) (Milford, MA)] and a Thermo BDS Hypersil C18 column (150×4.6mm, particle size 3 µm; Thermo Scientific, Waltham, MA). We dissolved a small amount of lyophilized nanoparticle suspension into 1ml of acetonitrile to determine the drug loading and encapsulation efficiency of the formulation. The mobile phase followed a gradient between buffer A (H₂O, acetonitrile, trifluoroacetic acid 94.9:5:0.1 v/v/v) and buffer B (acetonitrile).

Statistical analysis

Data, obtained from at least three independent experiments, were expressed as the mean ± SEM. Statistical evaluation was performed by one-factor analysis of variance (ANOVA) followed by Tukey test for multiple comparisons. P < 0.05 was regarded statistically significant.

Results

PGIA inhibits the growth of GBM cell lines more potently than ibuprofen

To study the effect of PGIA on cell growth, we initially determined the 24h IC₅₀ values (drug concentration inhibiting cell growth by 50% at 24h) of PGIA and ibuprofen in various human GBM cell lines (Figure 1B). The IC₅₀ values of PGIA varied among these GBM cell lines (189–447 µM), whereas those of ibuprofen were consistently higher. In all cases, the potency of PGIA was substantially greater than that of its parent compound, being enhanced between 3.7- and 5.1-fold.

We next compared the effect of PGIA on the growth of GBM cells against that of the NHA, by treating them with or without PGIA (2.5–640 µM) for 48h. As shown in Figure 1C, PGIA reduced cell growth in a concentration-dependent manner in all GBM cell lines tested. For instance, PGIA 80 µM at 48h reduced cell growth in LN-18, LN-229 and U118-MG cells by 61.8, 66.3 and 40.1% (P < 0.05, for all), respectively. In contrast, under the same experimental conditions, PGIA 80 µM for 48h had no effect on NHA cells, with 100% of them being viable (Figure 1C). Furthermore, treatment of NHA cells with PGIA 640 µM for 48h was unable to reach an IC₅₀ value. This indicates that PGIA decreases cell growth preferentially in GBM cells compared with NHA.

PGIA has a strong cytokinetic effect on GBM cells, including G₁/S arrest

To explore the mechanism of the growth inhibitory effect of PGIA, we determined its effect on cell proliferation, cell death and cell cycle. LN-229 cells were plated overnight, treated with PGIA for 24h, and cell proliferation was evaluated by the BrdU incorporation method. PGIA 0.75× IC₅₀ reduced BrdU incorporation in a concentration-dependent manor, decreasing it by 91.7% at this concentration (P < 0.01; Figure 2A). This effect was not cell specific since PGIA 0.75× IC₅₀ reduced BrdU incorporation in U87-MG, LN-18 and U118-MG cells by 87, 36 and 38%, respectively, compared with controls (P < 0.05 for all).

Figure 2.

Cell kinetic effect of PGIA in GBM cells. (A) LN-229, U118-MG, U87-MG and LN-18 cells were treated with PGIA for 24h. Cell proliferation assay based on BrdU incorporation into DNA during the S-phase of the cell cycle. The percentage of BrdU-positive cells is shown in the right. (B) Cell death by apoptosis was determined by flow cytometry using the dual staining (annexin V and propidium iodide) in U87-MG cells treated with increasing concentrations of PGIA for 24h. Results are expressed as fold-increase compared with the percentage of annexin V(+) cells in the control group. (C) Differential cytotoxic effect of PGIA in GBM cells compared with NHA. Apoptosis was determined by flow cytometry in NHA, LN-18 and LN-229 cells incubated without or with 200 µM of PGIA for 24h. Results are expressed as fold-increase compared with the percentages of apoptotic cells in the control cells. (D) Cell cycle analysis of cells treated with and without PGIA for 24h.

PGIA also induced cell death by apoptosis. Treatment of U87-MG cells with PGIA for 24h increased the proportion of apoptotic cells in a concentration-dependent manner compared with controls (Figure 2B). This increase became statistically significant (P < 0.05) at 1.5× IC₅₀ (5.5-fold over control) and 2× IC₅₀ (13.1-fold over control). We then compared the apoptotic effect of PGIA on GBM cells to that on NHA. Incubation of LN-18 and LN-229 cells with PGIA 200 µM for 24h generated a 2.9- and 3.4-fold increase (P < 0.05, for both) in annexin V(+) cells, compared with controls. In contrast, no increase in annexin V(+) cells was observed in NHA cells treated with the same concentrations of PGIA (Figure 2C), indicating that PGIA induces apoptosis preferentially in GBM cells.

The cell cycle progression was measured in U118-MG, LN-18 and LN-229 cells treated with PGIA. After 24h of incubation, there was a strong cell cycle arrest in G₁/S phase, evident in all three GBM cell lines tested. For instance, the percentage of cells in G₁ phase for control versus PGIA 0.75× IC₅₀ was 61.8 versus 74.6%; 41.95 versus 69.82% and 59.83 versus 82.28%, for U118-MG, LN-18 and LN-229 cells, respectively (Figure 2D).

PGIA inhibits the growth of human GBM xenografts in nude mice

To assess the in vivo chemotherapeutic potential of PGIA and compared it with ibuprofen, we used a GBM xenograft model. U118-MG cells subcutaneously injected into athymic nude mice gave rise to exponentially growing tumors. Once the tumors reached ~500mm³, the mice were treated intraperitoneally with either PGIA 20mg/kg, ibuprofen 20mg/kg or vehicle. On day 29 of treatment, the tumor volume (mean ± SEM) for the vehicle control, PGIA and ibuprofen groups were 1670±233mm³, 757±86mm³ and 1248±125mm³, respectively. Compared with vehicle-treated controls, PGIA reduced tumor growth by 77% (P < 0.01), whereas ibuprofen, given at an equi-dose, reduced it by 36%, indicating over a 2-fold reduction in the rate of tumor growth for PGIA when compared with ibuprofen (Figure 3A). Compared with controls, both effects are statistically significant (P < 0.01), as is the difference between the effects of PGIA and ibuprofen groups (P = 0.023). To note, PGIA was well tolerated, with the mice showing no weight loss or other signs of toxicity during treatment (Supplementary Figure S1 is available at Carcinogenesis Online).

Figure 3.

PGIA reduces GBM xenograft growth. (A) PGIA inhibits the growth of human U118-MG xenografts. U118-MG tumor volume growth over time for vehicle control (■), ibuprofen (20mg/kg/day 5×/week (●) and PGIA 20mg/kg/day 5×/week (▲) treated mice. *Significantly different compared with control group (P < 0.01), ^#significantly different compared with ibuprofen group (P = 0.023), one-way ANOVA test. (B) PGIA inhibits the growth of human U87-MG xenografts. U87-MG tumor volume growth over time for vehicle control (♦), and PGIA 20mg/kg/day (■) treated mice. *Significantly different compared with control group (P < 0.01), one-way ANOVA test. (C) Ki-67 and p21 immunostaining were performed on tumor sections, and photographs were taken at ×20 magnification. Representative images are shown (left). Results were expressed as percentage of Ki-67+ or p21+ cells ± SEM per 20× field (right). *Significance compared with control group; P < 0.05. (D) Levels of p21 were measured by immunoblot in total fractions isolated from LN-229 and LN-18 cells treated with PGIA. β-Actin = loading control.

To rule out a cell-specific effect, we assessed the in vivo chemotherapeutic potential of PGIA using the human U87-MG GBM xenograft model. In this model, the mice were treated intraperitoneally with either PGIA 20mg/kg or vehicle. On day 12 of treatment, the tumor volume (mean ± SEM) for the vehicle control and PGIA 20mg/kg groups were 2265±277mm³ and 996±180mm³, respectively, presenting a 56% reduction in the rate of tumor growth (P < 0.05; Figure 3B and Supplementary Figure S2 is available at Carcinogenesis Online).

To investigate the mechanism by which PGIA reduced tumor growth, we determined cell proliferation and p21 expression levels in tumor tissue sections from control and PGIA-treated mice (Figure 3C). Compared with controls, PGIA 20mg/kg inhibited cell proliferation (Ki-67 staining) by 41% (P < 0.05) and increase the percentage of p21(+) cells by 2.1-fold (P < 0.05; Figure 3C). The increase in p21 levels was also observed in vitro. PGIA treatment led to a time-dependent increase in p21 expression levels in LN-229 and LN-18 cell lines (Figure 3D).

PGIA reduces cyclin D1 levels in GBM cells and xenografts

Given the strong inhibition of cell proliferation and the robust cell cycle arrest at G₁/S by PGIA, we explored whether PGIA could affect proteins that control the G₁/S cell cycle phase transition. For this purpose, we explored the effect of PGIA on cyclins and CDKs. U87-MG cells treated with PGIA showed a time-dependent inhibition of cyclin D1 and CDK4 expression. As shown in Figure 4A, PGIA reduced cyclin D1 levels while those of cyclin B1 remained unaffected. PGIA also reduced cyclin D1 levels in a concentration-dependent manner, but failed to affect those of cyclin D3 (Figure 4B).

Figure 4.

PGIA reduces cyclin D1 levels in vitro and in vivo. (A) PGIA reduces cyclin D1 levels in a time-dependent manner in U87-MG cells. Immunoblots of cyclin D1, CDK4 and cyclin B1. β-Actin = loading control. (B) PGIA decreases cyclin D1 levels in a concentration-dependent manner in U87-MG cells. β-Tubulin = loading control. (C) Cyclin D1 immunostaining was performed on tumor sections, and photographs were taken at ×20 magnification. Representative images are shown (left). Results are expressed as percentage of cyclin D1+ cells ± SEM per 20× field (right). *Significance compared with control group; P < 0.05. (D) Effect of MG-132 on PGIA-induced cyclin D1 degradation. LN-229 cells were pretreated with MG-132 10 µM for 2h, treated with PGIA for 1h and lysed for immunoblotting. β-Tubulin = loading control. *Significance compared with control group; P < 0.05. (E) Immunoprecipitation with cyclin D1 and immunoblotting analysis for ubiquitin of lysates of LN-229 cells treated with PGIA for 2h.

Using publicly available data sets (NCBI GEO GSE4290 and TCGA GBM Affymetrix U133A 2013-12-18 freeze), we explored the expression status of CCND1 gene (cyclin D1) in human GBM samples. CCND1 gene expression was significantly increased in glioblastoma tumor samples compared with non-tumor samples. The fold-increase of cyclin D1 expression for glioblastoma tumor/normal tissue was 1.9 (false discovery rate < 5%) and 3.8 (false discovery rate < 1%) for each data set, indicating that cyclin D1 is overexpressed in human GBM and could represent a therapeutic target.

We then evaluated the effect of PGIA on cyclin D1 in vivo. IHC studies of U87-MG xenografts revealed that PGIA inhibited cyclin D1 expression in vivo as well. Cyclin D1 expression was reduced by 54.3% in PGIA-treated group, compared with control (P < 0.05; Figure 4C).

Because PGIA reduces cyclin D1 levels rapidly, we evaluated the mechanisms underlying PGIA-triggered cyclin D1 reduction. Pretreatment of LN-229 cells with the proteasome inhibitor MG-132 abrogated PGIA-induced reduction in cyclin D1 levels (Figure 4D). By immunoprecipitation and western blot assays, we showed that PGIA could increase ubiquitinated cyclin D1 in LN-229 cells (Figure 4E). These results indicate that the ubiquitin–proteasome system mediates, at least in part, PGIA-triggered cyclin D1 degradation.

We next explored whether PGIA could affect an upstream event in the cyclin D1 degradation pathway. Phosphorylation and proteolytic turnover of cyclin D1 and its subcellular localization during the cell division cycle are, in part, regulated by GSK3β, which can be inactivated by phosphorylation (23). In GBM cells, PGIA inhibited phosphorylated GSK3β (p-GSK3β) in a time- and concentration-dependent fashion (Figure 5A and B). Moreover, pretreatment of cells with the GSK3β inhibitor LiCl (24), partly suppressed PGIA-induced cyclin D1 degradation (Figure 5C).

Figure 5.

PGIA induces dephosphorylation of GSK3β in GBM cells. (A) LN-229 cells were treated with PGIA for 2h. Phospho-GSK3β and total GSK3α/β levels were analyzed by immunoblotting. β-Tubulin = loading control. (B) PGIA induces dephosphorylation of GSK3β in a time-dependent manner. Immunoblots for phospho-GSK3β and total GSK3α/β from LN-229 cells treated with PGIA. β-Tubulin = loading control. (C) Effect of LiCl on PGIA-induced cyclin D1 degradation. LN-229 cells were pretreated with LiCl 20mM for 1h and then incubated with PGIA for 1h. Immunoblots were performed using indicated antibodies (left). Protein expression was quantified by the densitometry analysis (right). *Significant compared with control group; P < 0.05. (D) p-AKT immunostaining was performed on tumor sections and photographs were taken at ×20 magnification. Representative images are shown (left). The number and intensity of p-AKT positive cells were scored and ranked according to its intensity of each field. p-AKT was expressed as IHC index per field (right). *Significantly different from control group (P < 0.05, one-way ANOVA test; IHC staining, ×20). (E) Cyclin D1 overexpression abrogates cell growth inhibition by PGIA. Control and CCND1 stably transfected LN-18 cells were treated with PGIA 320 µM for 24h. Cell growth was evaluated by the MTT assay. Top: cyclin D1 expression status in whole-cell protein lysates. (F) Cyclin D1 overexpression prevents cells from the cell proliferation inhibition induced by PGIA: control and CCND1 stably transfected LN-229 cells were treated with PGIA for 24h. Cell proliferation assay based on BrdU incorporation into DNA during the S-phase of the cell cycle.

Because GSK3β phosphorylation can be regulated by AKT (25), we investigated the effect of PGIA on AKT phosphorylation in vivo. IHC studies of U87-MG xenografts revealed that PGIA reduced p-AKT expression by 43.2%, compared with control (P < 0.05; Figure 5D).

Overexpression of cyclin D1 abrogates, in part, the anticancer effect of PGIA

To confirm that the inhibition of cyclin D1 is a target of PGIA, we generated LN-18 and LN-229 cells stably overexpressing cyclin D1, using as controls their parent cells that have basal levels of cyclin D1. Cyclin D1 overexpression abrogated, in part, the growth inhibitory effect of PGIA in vitro (Figure 5E). For example, treatment of LN-18 cells with PGIA 750 µM for 24h reduced cell growth by 90%. In contrast, overexpression of cyclin D1 partially prevented the reduction in cell growth induced by PGIA 750 µM (48% of viable cells in cyclin D1 overexpressing cells; Figure 5E). The overexpression of cyclin D1 in LN-18 cells was confirmed by immunoblotting. Furthermore, in LN-229 cells, the reduction of cell proliferation and the cell cycle arrest at G₁/S induced by PGIA were completely abrogated in cyclin D1 overexpressing cells compared with controls (Figure 5F and Supplementary Figure S3 is available at Carcinogenesis Online). These results indicate that cyclin D1 is a key target of PGIA in GBM.

PGIA formulated in PLLA–PEG crosses the blood-brain barrier and is safe in mice

The treatment of GBM remains challenging due to the inability of many drugs to cross the blood-brain barrier. To overcome this limitation, we formulated PGIA in polymeric nanoparticles. PLLA–PEG polymeric nanoparticles can entrap anticancer drugs, improving their bioavailability and antitumor efficacy (12). Thus, we formulated PGIA in PLLA(5k)–PEG(2k), coated the nanoparticles with polysorbate-80 and determined its access to the brain. The basic characteristics of these nanoparticles were spherical with homogeneous distribution; size = 57.5±1.1nm; polydispersity index = 0.116±0.002 and zeta-potential = −14.69±1.14 mV (Figure 6A and B).

Figure 6.

Formulation of PGIA in PLLA(5k)–PEG(2k) polymeric nanoparticles delivers PGIA to the brain. (A) Hydrodynamic diameter distribution of PLLA(5k)–PEG(2k) polymeric nanoparticles determined using dynamic light scattering as described in methods. (B) Scanning electron microscopy image depicting the morphology of the PLLA(5k)–PEG(2k) nanoparticles loaded with PGIA. Scale bar: 200nm. (C) Brain levels for PGIA incorporated in PLLA(5k)–PEG(2k) nanoparticles in mice (n = 2 per time point). PGIA level in the brain was assayed by HPLC. To note: we were unable to detect brain PGIA tissue levels when it was reconstituted in 2-hydroxypropyl-β-cyclodextrins.

These coated PLLA–PEG nanoparticles allowed PGIA to cross the blood-brain barrier. For instance, 5 and 10min following intravenous administration of PGIA formulated in PLLA–PEG, the brain concentration of PGIA was 4.5 and 19.8 nmol/mg of tissue, respectively (Figure 6C). In contrast, PGIA dissolved in 2-hydroxypropyl-β-cyclodextrins was undetected in the brain. We conclude that PLLA–PEG nanoparticles provided a significantly improved pharmacokinetic profile for PGIA and allowed PGIA to cross the blood-brain barrier.

Finally, we evaluated the safety of PGIA formulated in PLLA–PEG. For this purpose, BALB/c mice (five/group) were intravenously administered once a day with progressively increasing doses of PLLA–PEG loaded with PGIA for 1 week. The body weights of vehicle-treated animals and all PGIA-treated animals were comparable throughout the study. All of the treated mice showed no signs of toxicity and were healthy. Thus, PGIA formulated in PLLA–PEG appears to have a high degree of safety with a maximum tolerated dose of at least 50mg/kg/day.

Discussion

GBM represents clinically the most challenging glioma, given its incidence and mortality. Even aggressive treatment has failed to significantly impact survival, which at 5 years is still <10%. Our results establish the novel PGIA as a strong inhibitor of GBM in preclinical models, whose distinct mechanism of action is dominated by inhibition of cyclin D1. In addition to its robust anticancer efficacy, PGIA reaches the brain when formulated in PLLA–PEG and lacks detectable toxicity, likely based on its selective inhibition of cancer cells. All these attributes make PGIA a promising candidate drug for GBM treatment.

The present results establish PGIA as an agent with significantly improved anticancer activity over its parent NSAID, ibuprofen. PGIA is a potent inhibitor of GBM cell growth in vitro (>3.7-fold more potent than ibuprofen) and in vivo (>2-fold more efficacious in reducing GBM xenograft tumor growth). It should be noted that PGIA and ibuprofen were administered at an identical dose, but on a molar basis, the amount of ibuprofen was 267% higher, suggesting the even greater efficacy of PGIA, over ibuprofen.

The anticancer effect of PGIA results from its triple cytokinetic effect: inhibition of proliferation, induction of apoptosis and block at the G₁/S cell cycle transition. The antiproliferative effect of PGIA seems to be the dominant one. For example, PGIA at 0.75× IC₅₀ reduced cell proliferation in GBM cells by up to 91%, with the induction of apoptosis being ~2-fold of baseline. In vivo, PGIA suppressed the growth of subcutaneous U87-MG tumors, and this effect was also characterized by decreased cell proliferation and increased p21 levels.

The molecular mechanism underlying the anticancer effect of PGIA is dominated by its ability to inhibit cyclin D1. The relevance of this effect is emphasized by reports that cyclin D1, commonly overexpressed in GBM (26), is critical for the development and progression of GBM (18). Furthermore, our data mining analysis revealed that cyclin D1 levels in GBM biopsy specimens exceeded those of normal brain tissue. Increased cyclin D1 expression is associated with gene amplification in human glioma specimens (27). The overexpressed cyclin D1 increases the proliferation and invasive potential of GBM cells while reducing apoptosis (18). Finally, the link between cyclin D1 and GBM is strengthened by studies in a number of animal models, in which forced overexpression of cyclin D1 directly enhances tumorigenesis. Taken together, these findings make inhibition of cyclin D1 a potentially fruitful strategy for GBM therapy.

Cyclin D1 can be regulated by transcription and proteolysis modulatory pathways, the latter process being mainly mediated by the ubiquitin–proteasome system (28). PGIA induced a rapid decrease of cyclin D1 levels in GBM cells in vitro and in vivo, by modulating upstream events, which include GSK3β. We demonstrate that the proteasome inhibitor MG-132 can inhibit PGIA-triggered cyclin D1 degradation, indicating that this proteolysis is proteasome dependent. Cyclin D1 turnover is regulated by GSK3β that can be inactivated by phosphorylation (23). PGIA caused dephosphorylation of GSK3β, leading to its activation, which can further inhibit cyclin D1. Consistently, inhibition of GSK3β by LiCl resulted in abrogation of cyclin D1 reduction induced by PGIA. The centrality of cyclin D1 in the reduction of cell proliferation by PGIA was documented by the finding that cyclin D1 overexpression abrogated the anticancer effect of PGIA. Similarly, various chemotherapeutic agents have been shown to induce cyclin D1 degradation (28,29), indicating that targeting cyclin D1 may offer a useful avenue for therapeutic intervention.

The treatment of GBM remains challenging due to the inability of some drugs to cross the blood-brain barrier. Thus, the task of improving therapies relies on the development of technologies that overcome the low accessibility of drugs to the brain. Our approach was to formulate PGIA in PLLA–PEG nanoparticles and coat them with polysorbate-80. These PLLA–PEG were able to traverse the blood-brain barrier and deliver PGIA to the brain. In contrast, when PGIA was unprotected (dissolved in 2-hydroxypropyl-β-cyclodextrins), it was unable to reach the brain. Similarly, PLLA–PEG nanoparticles have been successfully used to entrap other anticancer drugs, achieving improved water solubility, bioavailability and antitumor efficacy via the EPR effect (12). For example, PLLA–PEG copolymers loaded with paclitaxel accumulated in tumors, released the drug slowly and resulted in tumor regression in animal models (30). Clearly, our delivery system allows the drug to cross the blood-brain barrier, reaching the desired target tissue. Thus, this potential clinical application for PGIA appears realistic thanks to its incorporation into PLLA–PEG nanoparticles.

Ideally, an anticancer drug should target specifically the tumor and not the normal surrounding tissue. In other words, it should kill the cancer cells while sparing the adjacent normal cells. PGIA displays such selectivity. Compared with several GBM cell lines, NHA were more resistant to PGIA-induced cell death and to the suppression of their growth. Such selectivity, if broadly confirmed, will be a significant advantage of PGIA.

In conclusion, the novel compound PGIA is an effective anticancer agent in preclinical models of GBM, is selective toward GBM cells versus NHA and cyclin D1 represents a key molecular target for its effect. Furthermore, when formulated in polymeric nanoparticles, PGIA appears safe and crosses the blood-brain barrier, reaching the target tissue. Therefore, PGIA is a promising novel agent for GBM and deserves further investigation.

Supplementary material

Supplementary Figures 1–3 can be found at http://carcin.oxfordjournals.org/

Funding

Stony Brook Cancer Center (to G.G.M) ; National Institute of Health (CA154172 to B.R.).

Conflict of Interest Statement: B.R. has an equity position in Medicon Pharmaceuticals, Inc., and D.K. is affiliated with Medicon Pharmaceuticals, Inc., the company that owns the test compound. All other authors declare no competing financial interest.

Author contributions

Acquisition of data, analysis and interpretation of data: L.E.B., G.M., B.M.V., J.F.L., R.W., J.Z. and G.G.M. Provided the test compound: D.K. Drafting the manuscript: L.E.B., B.R. and G.G.M. Study concept and design: D.K., B.R. and G.G.M.

Glossary

Abbreviations

CDK

cyclin-dependent kinase

GBM

glioblastoma multiforme

HPLC

high-performance liquid chromatography

IHC

immunohistochemical

NHA

normal human astrocytes

NSAID

non-steroidal anti-inflammatory drug

PBS

phosphate-buffered saline

PEG

poly(ethylene glycol)

PGIA

phospho-glycerol-ibuprofen-amide

PLLA

poly-(l)-lactic acid