Polymeric glabrescione B nanocapsules for passive targeting of Hedgehog-dependent tumor therapy in vitro

Aim: With the purpose of delivering high doses of glabrescione B (GlaB) to solid tumors after systemic administration, long-circulating GlaB-loaded oil-cored polymeric nanocapsules (NC-GlaB) were formulated. Materials & methods: Synthesis of GlaB and its encapsulation in nanocapsules (NCs) was performed. Empty and GlaB-loaded NCs were assessed for their physico-chemical properties, in vitro cytotoxicity and in vivo biodistribution. Results: GlaB was efficiently loaded into NCs (∽90%), which were small (∽160 nm), homogeneous and stable upon storage. Further, GlaB and NC-GlaB demonstrated specific activities against the cancer stem cells. Preliminary studies in tumor-bearing mice supported the ability of NC to accumulate in pancreatic tumors. Conclusion: This study provides early evidence that NC-GlaB has the potential to be utilized in a preclinical setting and justifies the need to perform therapeutic experiments in mice.

Cancer stem cells (CSCs) represent a subpopulation of stem cells within a tumor which drive growth and metastasis. While chemotherapy and radiation therapy are frequently able to eradicate most of the tumor bulk and proliferative cells, these treatments frequently fail in eradicating the quiescent CSCs, which might be dormant for years [1]. Eradication of the CSCs is thought to be essential to achieve complete tumor ablation and prevent cancer relapse [2]. Several signaling pathways that regulate normal stem cell self-renewal cause neoplastic proliferation when dysregulated by mutations [3]. Among those is the Hedgehog (Hh) signaling pathway, which is essential for tissue development and stemness, and whose dysregulation leads to tumorigenesis [4]. Aberrant activation of the Hh pathway is responsible for the tumorigenesis of several human cancers including medulloblastoma (MB) [5], rhab-domyosarcoma [6,7], melanoma [8], basal cell carcinoma (BCC) [9], breast [10], liver [11], lung [12,13], prostate [14,15], brain and pancreas tumors [16][17][18][19]. Thus, this pathway has been widely targeted in recent years for the development of anticancer drugs. A recent study from the co-authors of this paper identified glabrescione B (GlaB) [20], a naturally occurring isoflavone in Derris glabrescens [21], as the first small molecule that binds the Hh modulator Gli1 (a zinc finger) and impairs its activity by interfering with Gli-DNA interaction. GlaB was shown to inhibit the in vitro and in vivo growths of multiple Hh-dependent cancer cells and CSCs, as well as the self-renewal ability and clonogenicity of tumor-derived stem cells [22]. In this previously reported study, the ability of GlaB against MB in the allograft model of MB cells by subcutaneous injection, using a solution of GlaB (75 μmol/kg) prepared in 2-hydroxypropyl-β-Polymeric glabrescione B nanocapsules for passive targeting of Hedgehog-dependent tumor therapy in vitro future science group Research Article Al-Jamal, Botta, Ingallina et al. cyclodextrin:ethanol (3:1), was assessed. During the 18-day treatment, GlaB significantly suppressed the tumor mass compared with the control. Despite the promising findings confirming the therapeutic efficacy of GlaB against Hh-dependent cancer, it was concluded that delivery approaches are required to solubilize GlaB in aqueous solvents so that systemic administration of relatively high doses is feasible.
Nanotechnology-based drug delivery systems offer the possibility of solubilizing poorly soluble drugs, protecting labile molecules and improving on tissue biodistribution [23]. Among the different drug delivery technologies, polymeric nanocapsules (NCs) have attracted growing attention due to the high loading capacity of hydrophobic molecules, as well as longterm stability, offering enhanced bioavailability of the active compound [24,25]. NCs consist of a core-shell structure made of hydrophilic polymer stabilized with lipophilic and/or hydrophilic surfactants surrounding an oily core, in which active molecules are confined [26]. Their preparation is generally cheap, fast and easily scalable.
This work focused on developing GlaB-loaded oilcored polymeric NCs to improve its water solubility, so that it targets Hh-dependent cancer cells in vitro and in vivo. The choice of this formulation was encouraged by our previous studies for the delivery of water-insoluble drugs such as quercetin [27] and curcumin [28]. Physico-chemical properties, including hydrodynamic diameter, polydispersity index (PDI), ζ potential and stability, were determined. Once the desired formulation was identified, the anticancer activity of both GlaB and NC-GlaB was evaluated in vitro against a range of CSCs and non-CSCs, which have been reported to have a strong dependence on the Hh pathway. Finally, the in vivo biodistribution profile and organ histology of intravenously administered NC-GlaB was assessed in pancreatic-tumor-bearing mice. Overall, this study provides early evidence that NC-GlaB has the potential to be utilized in a preclinical setting, aiming at its future use in clinic.

3-iodo-5,7-dimethoxy-4H-chromen-4-one 3
The enamino ketone 2 was dissolved in CH 3 OH (450 ml) and I 2 (11.7 g, 46.2 mmol) was added to the solution. The mixture was stirred at room temperature for 1 h, then the solvent was evaporated. To remove residual I 2 , the crude was treated with a saturated aqueous Na 2 S 2 O 3 solution until the mixture became clear. The mixture was then extracted with CHCl 3 , and the combined organic layers were dried over Na 2 SO 4 and concentrated under reduced pressure. The residue was purified by column chromatography using hexane-EtOAc as eluent to obtain 3-iodo-5,7-dimethoxy-4H-chromen-4-one 3 (  To a solution of 3 (2.5 g, 7.5 mmol), in 1,2-dimethoxyethane/H 2 O = 50:50 (150 ml) were added Na 2 CO 3 (3.18 g, 30 mmol), 3,4-(methylenedioxy)-phenylboronic acid (1.8 g, 11 mmol) and Pd EnCat™40 (937 mg, 5%). The resulting mixture was stirred at 45°C for 2 h and then filtered. The catalyst was washed with H 2 O and CH 2 Cl 2 . The aqueous phase was extracted with CH 2 Cl 2 . The combined organic layers were dried over Na 2 SO 4 and concentrated under reduced pressure. The crude residue was purified by flash chromatography using hexane-EtOAc as eluent to give 4 (1.4 g) as a gray powder.
Gray A mixture of 4 (1.4 g, 4.3 mmol) and Pb(OAc) 4 (7.5 g, 17 mmol, freshly recrystallized from AcOH) in dry C 6 H 6 (100 ml) was stirred at 80°C under argon overnight. After, being cooled to room temperature, the reaction mixture was filtered through a pad of Celite, washed with CH 2 Cl 2 and concentrated under reduced pressure. The crude was diluted with THF/H 2 O = 5:1 (50 ml) and CH 3 COOH (50 ml), and the resulting mixture was stirred at room temperature for 6 h. Afterward, a saturated aqueous NaHCO 3 solution was added until pH 8 and extracted with EtOAc. To the combined organic layers was added a solution of NaOH 0.1 M. Water layer was treated with CH 3 COOH and then extracted with EtOAc. The combined organic layers were dried over Na 2 SO 4 and concentrated under reduced pressure to obtain 5 (570 mg).
White crystal (yield 90%); mp 102-104°C. The spectral data of this product were identical to an authentic sample of GlaB.

Solubility of GlaB in castor oil
The solubility of GlaB in castor oil was evaluated using the simple saturation shake-flask method [30]. Briefly, excess GlaB (10 mg) was transferred to vials containing 1 ml of castor oil, sealed and incubated at 37°C for 48 h with shaking (250 strokes/min). Subsequently, 100 μl of the supernatant was diluted in ethanol, and the amount of GlaB dissolved in the organic solvent was determined using a UV-Vis spectrophotometer (UV-1601 PC, Shimadzu, Japan) by measuring the absorbance at 258 nm, against a GlaB standard curve.

Determination of GlaB encapsulation efficiency
To purify the sample from un-encapsulated drug, NCs were passed through a PD10 desalting column (GE Healthcare). To disrupt the NCs and quantify the total amount of drug in NCs prior and after purification, NC preparations (30 μl) were diluted in ethanol (to a final volume of 2 ml). Subsequently, the amount of GlaB in NCs was determined using a UV-Vis spectrophotometer (UV-1601 PC, Shimadzu) by measuring the absorbance at 258 nm against a standard plot of known GlaB concentrations. The amount of GlaB in the measured samples was then calculated from the standard plots The average size and ζ potential of the NCs were measured by dynamic light scattering with a Nanosizer ZS Series (Malvern Instruments, MA, USA), using disposable polystyrene cells and plain-folded capillary Zeta cells. Dilutions were prepared in deionized water and measurements performed at room temperature. The electrophoretic mobility was used to calculate the ζ potential using the Helmholtz-Smoluchowski equation. Results for the hydrodynamic size are presented as the average of 20 measurements, each performed in triplicate.

Release profile & serum stability in vitro
About 2 ml of NC formulation containing 2 mg (GlaB) in 5 ml was transferred into a 10 kDa dialysis bag, and dialysis was performed against 200 ml of phosphatebuffered saline (PBS; pH 7.4) containing 10% (w/v) Tween ® 80, at 37°C under stirring (250 strokes/min). Samples were taken after 24 h, and drug concentration was determined using a Perkin-Elmer Lambda 35 UV-Vis spectrophotometer by measuring the absorbance at 258 nm.
Shelf life stability NC suspensions were transferred to 7 ml glass vials, sealed and stored at 4°C. The stability of NCs was tested after 0, 7, 14 and 28 days by visual inspection (color and opacity) and by measuring the hydrodynamic size and ζ potential. Results of measurements (n = 3) are presented as average ± standard deviation (SD).
Cell culture SKOV3 human ovary adenocarcinoma cell line was cultured in McCoy media, while PANC1 human pancreatic carcinoma cell line and GL261 murine glioma cell line were cultured in Advanced Roswell Park Memorial Institute (RPMI) media (Sigma, MO, USA), at 37°C in 5% CO 2 . Culture medium was supplemented with 10% fetal bovine serum (FBS), 100 U/ ml penicillin, 100 μg/ml streptomycin and 1% l -glutamine (Sigma). Cells were routinely grown in 75 cm 2 tissue culture flasks and passaged twice a week, after detachment with trypsin/EDTA (Sigma), when reaching 80% confluency.
Human (G7) and mouse glioma-initiating neural stem cell line (IENS) glioblastoma (GBM), as well as their culturing conditions, have been described previously [31][32][33][34]. Briefly, cells were cultured using   . Growth factors EGF (Peprotech, UK, 315-09-500) and FGF-2 (Peprotech, 100-18B-500) (10 ng/ml), as well as laminin (Sigma, L2020, 1 μg/ ml), were added freshly before cell split. Cells were routinely grown to confluence and split typically twice per week after dissociation with Accutase solution (Sigma, A6964) and centrifugation. Differentiation of G7 cells was performed as described previously by Pollard and collaborators [35]. After reaching 70% confluency, G7 cells were washed with PBS and replaced with media containing bone morphogenetic protein (BMP)4 (Peprotech 120-05ET-10, 10 ng/ml), but no EGF or FGF-2. Differentiation media, containing BMP, was replaced every 7 days throughout the time course and supplemented with fresh BMP at 4 days. For each condition, 10,000 events were gated and fluorescence was analyzed using the FlowJo software. Results were expressed as averages of percentage cell populations in each phase of the cell cycle ± SD (n = 3).  (TBS-T) and then incubated with secondary antibodies, horseradish peroxidase-linked goat antirabbit antibodies (#7074, Cell Signaling Technology) at 1:3000 dilution for 2 h at room temperature. The protein bands were detected using chemiluminescent kits (Immun-Star™ Chemiluminescent Kit, BioRad, UK) and imaged by the ChemiDoc MP imaging system and Image Lab software (BioRad). GAPDH was used as an internal protein control. Values for Gli1 were normalized to the total GAPDH and then to untreated IENS, in the case of treatment to untreated PANC1 and SKOV3.

Radiolabeling of NCs
Radiolabeling of NC-PEG-DTPA-GlaB The radioactive probe 111 InCl 3 was purchased from Mallinckrodt Pharmaceuticals (The Netherlands) as an aqueous solution in 0.5 M HCl and used with-out further purification. To radiolabel the NC with indium-111, NCs were prepared in water as described in the previous section except that PLGA-PEG-DTPA was included at 10% (w/w) of total polymer content and concentrated 28-times using rotavap, so that the final drug and polymer concentrations are 1.22 and 71.43 mg/ml, respectively. The NC suspension (700 μl) was incubated with 2 M ammonium acetate (one-ninth of the reaction volume, pH 5.5), to which 10 MBq of 111 InCl 3 was added per injection dose for γ scintigraphy. The reaction was kept at room temperature for 30 min with intermittent vortexing every 10 min. Upon completion, the radiolabeling reaction was quenched by the addition of 0.1 M EDTA chelating solution (one-twentieth of the reaction volume). 111 InCl 3 alone and NCs without DTPA incorporation were also subjected to the same conditions of the labeling reaction as a control.

Radiolabeling efficiency & serum stability studies
To evaluate the radiolabeling efficiency, the labeled NC-111 In was spotted on the TLC strips, which were developed in 0.1 M ammonium acetate containing 25 mM EDTA as a mobile phase. Strips were allowed to dry before being developed and counted quantitatively using a cyclone phosphor detector (Packard Biosciences, UK). The NC-111 In was passed through PD-10 column before injecting into animals to exchange the ammonium acetate buffer (pH 5.5) with PBS (pH 7.4) and to ensure that no free 111 In-EDTA was present in the injected dose. The sample was collected from the column and the radiolabeling efficiency was examined immediately after the collection. As a result, the sample was further diluted two-times to three-times compared with the original sample. Final drug concentration after filtration was 1.22 mg/ml. To determine the stability of the radiolabeled NC-111 In, 20 μl of the sample was incubated with 20 μl of serum and 20 μl of PBS at 37°C. At 24 h, 2 μl of the incubated samples was spotted to the TLC strips, which were developed and quantified as described above. Free 111 In-EDTA was detected at the solvent front while radiolabeled NC was retained at the application point. Indium-labeled NC without DTPA incorporation was included as a control. . Animals were sacrificed after 30 min or 24 h. The tumors and major organs (brain, lung, liver, spleen, kidney and heart) were collected, weighed, and the 111 In activity in the collected tissues was quantified using an automated γ counter (LKB Wallac 1282 Compugamma, Perkin Elmer, UK). A portion of the radiolabeled NCs was also analyzed as a reference. The γ counter corrected for physical radioisotope decay. Radioactivity readings (c.p.m.) were plotted as a % ID/organ or as a % ID/g of tissue.

Organ toxicity evaluation
Twenty-eight days after tumor inoculation, pancreatictumor-bearing SCID/Beige mice (n = 6) were injected, via tail vein, with 200 μl of the NCs at a dose of 78 mg/kg of polymer and 12.5 mg/kg of drug (∼1.56 mg of polymer and ∼250 μg of drug per mouse) once a week for 3 weeks. At the experiment end point, the mice were sacrificed and the major organs (liver, lung, tumor, heart, kidney and spleen) were fixed in 10% neutral buffer formalin as 5 mm 2 pieces. The organs were then paraffin-embedded and sectioned for hematoxylin and eosin according to the standard histological protocols at the Royal Veterinary College. The stained sections were analyzed with a Leica DM 1000 LED Microscope (Leica Microsystems, UK) coupled with charge-coupled device (CDD) digital camera (Qimaging, UK).

Formulation & physico-chemical characterization of NCs
The preparation of non-PEG-containing NC (NC-GlaB) is described in detail in the 'Materials & methods' section. PEG-containing NCs (NC-PEG) were also prepared, due to their prolonged blood circulation time facilitated by PEG. All formulations were prepared using the nanoprecipitation method ( Figure 2). The purified formulations were then characterized for their hydrodynamic diameter, PDI and ζ potential, immediately after purification and over 28 days storage at 4°C (Table 1 & Supplementary Table 1). Castor oil was selected as the core material due to its good oil solubilization ability of the drug (27 μg/μl at 37°C). Importantly, efficient drug encapsulation (∼90%) was achieved for both NC-GlaB and NC-PEG-GlaB (Table 1 & Supplementary Table 1). In contrast to the extremely poor aqueous solubility of GlaB [21], NC-GlaB were well dispersed and formed a homogeneous suspension in aqueous media with PDI values below 0.2 and an average hydrodynamic diameter of approximately 160 nm. All NC formulations were found to be stable over the 28-day storage period, with no significant changes in size, PDI or ζ potential observed over time. No apparent change in color or phase separation was observed in any of the formulations.

In vitro stability of the NCs
To determine the in vitro stability of GlaB in the NC formulation, drug release from NC-PEG-GlaB was assessed over 24 h by the dialysis method, in the presence or absence of FBS. The dialysate used was PBS pH 7.4 containing 10% w/v Tween ® 80 to maintain GlaB solubility in aqueous media. As shown in

In vitro cytotoxicity of GlaB
The MTT assay was performed in order to estimate the toxicity of free or NC-formulated GlaB. Two dif-ferent types of cell lines were selected for this study: non-stem cancer cells that contain a small CSC subpopulation (PANC1, human pancreatic carcinoma; SKOV3, human ovary adenocarcinoma; and GL261, murine glioma) [41][42][43] and pure glioblastoma-derived neural stem cells (GNS; patient-derived G7 human and engineered Ink4a/ARF, EGFRvIII overexpressed 154.3 ± 1.1 0.14 ± 0.01 -46.7 ± 1.7 † Measured with dynamic light scattering. ‡ Expressed as mean ± SD (n = 3). § Analyzed with electrophoretic light scattering using 10 mM NaCl. ¶ Calculated as percentage of initial drug added, which was determined by spectrophotometry. # Calculated as mass of incorporated drug divided by the weight of polymer, which was determined by spectrophotometry. GlaB: Glabrescione B; NC: Nanocapsule; NC-GlaB: GlaB-loaded oil-cored polymeric nanocapsule; PDI: Polydispersity index; SD: Standard deviation.  Polymeric glabrescione B nanocapsules for tumor therapy in vitro Research Article IENS murine) [31]. It is hypothesized that the GNS cell lines will be more sensitive to GlaB as they consist entirely of CSCs and therefore have a stronger dependence on the Hh pathway. Indeed, western blot quantification of Gli1 protein expression, a downstream effector in the Hh pathway, showed higher Gli1 protein levels in pure CSCs (IENS and G7), compared with non-stem cancer cells (PANC1, SKOV3 and GL261) (Supplementary Figure 1).
As demonstrated in Figure 4A & Supplementary  Figure 2, a dose-dependent and time-dependent reduction in cell viability was observed in both treatment conditions (NC-GlaB and free GlaB). Incubation for 96 h with 5 μM of free GlaB/NC-GlaB resulted in a significant reduction in cell viability in IENS (∼27%/14% cell viability; p < 0.05) and G7 (∼20%/13%; p < 0.001/p < 0.01) cells, while a smaller decrease was seen in GL261 (∼67%/84%; p > 0.05) or PANC1 (∼36%/48%; p > 0.05) cells. Light microscopy images obtained during treatment confirmed the dose-dependent cell killing effect verified with the MTT assay. Morphological changes, namely smaller and more round, were observed during incubation with free GlaB or NC-GlaB ( Figure 4C). Cells treated with empty NCs, at equivalent concentrations to those used in NC-GlaB, showed normal cell morphology except at very high concentration, in which slight but significant toxicity was detected (Supplementary Figure 3). IC 50 , the drug amount required to achieve 50% cell kill, was estimated in G7 and PANC1 cells. The values obtained following 96 h treatment with free GlaB/ NC-GlaB were 0.29 μM/0.71 μM (G7) and 1.62 μM/1.49 μM (PANC1) ( Figure 4B). The toxicity of NC-PEG-GlaB was also tested, to assess whether the incorporation of PEG into the NCs has any effect on cell viability. Comparable results were obtained at 96 h with 5 μM GlaB, for NC-PEG-GlaB (∼74%) and NC-GlaB (∼86%) (Supplementary Figure 4).

GlaB effect on cell cycle
Having shown that free or NC-formulated GlaB has an effect on the viability of GNS cells and, to a lesser extent, on non-stem cancer cells, experiments were performed to determine whether the effect could be due to alterations in the cell cycle mechanism. As shown in Figure 5  Since not enough protein could be obtained from IENS or G7 cells treated with GlaB, due to high sensitivity to the drug, PANC1 and SKOV3 were selected to further clarify the influence of GlaB incubation on Gli1 protein levels. In this regard, decreased Gli1 expression was detected in the cells after incubation with 5 μM GlaB for 24 h (∼73.3% relative expression for SKOV3, 88.1% for PANC1) or 48 h (∼33.6% for SKOV3, 44.6% for PANC1), compared with control untreated cells (100%) (Supplementary Figure 6). These results suggest that GlaB treatment suppresses the expression of Gli1.

Effect of GlaB treatment in differentiated CSCs
In order to confirm that the activity of GlaB is Hh dependent (and thus connected to stemness), the differentiation of G7 cells into astrocytes was promoted by incubating these cells with BMP4, as shown previously by Carén et al. [35]. It is expected that differentiated G7 cells are less dependent on Hh pathway and therefore less sensitive to GlaB. Indeed, as shown in Figure 6A & B, and opposed to what was observed with pristine G7 cells, no significant toxicity was seen in differentiated G7 cells incubated with free GlaB for

Radiolabeling of NC-PEG-GlaB & in vitro stability
In order to perform in vivo biodistribution studies, radiolabeled NCs were prepared using the previously synthesized PLGA-NH-PEG-NH-DTPA at 10% (w/w) of total polymer content. The presence of DTPA in the shell of the NCs is necessary for complexing 111 In, a useful short-lived γ emitter used in biodistribution studies. The radiolabeling efficiency, calculated as the percentage of bound 111 In of total added 111 In, was tested using instant thin layer chromatography (TLC) technique after chelation of 111 In. Only [ 111 In]-EDTA or [ 111 In]-DTPA chelates migrated to the solvent front while the radiolabeled polymer remained at the application point. Approximately 90% of radiolabeling efficiency was achieved for NC-DTPA ( Figure 7A). Buffer exchange from ammonium acetate to PBS was also carried out using gel filtration chromatography. By passing the NC suspension through PD10 size exclusion columns, unbound 111 In was retained in the column and the eluted NC suspension (in PBS) was ready for injection to mice. Radiolabeled NCs were found to be stable in solution, with approximately 92 and 72% stability obtained after NC incubation in PBS or 50% FBS (respectively) at 37°C for 24 h.

Biodistribution of radiolabeled NC-PEG-GlaB in mice after single intravenous injection
The biodistribution of radiolabeled NC-PEG-DTPA-GlaB was investigated in pancreatic-tumorbearing SCID/Beige mice. After receiving a single tail vein injection of NCs (∼370 μg GlaB per mouse), the animals were sacrificed at 30 min or 24 h postinjection and the % ID/g of tissue and % ID/organ were calcu-lated. As shown in Figure 7B & C, NCs showed accumulation in major first-pass organs such as liver and spleen at 24 h after injection. Tumor uptake was 1.2 and 0.3% ID/g at 30 min and 24 h, respectively. Mice were not perfused before sacrifice. It is therefore likely that the higher % ID/g tumor obtained at 30 min than 24 h is due to high concentration of the drug in the blood. To evaluate the systemic toxicity of radiolabeled NC-PEG-DTPA-GlaB, pancreatic tumor-bearing SCID/Beige mice were injected with the NCs (∼250 μg GlaB/mouse), via tail vein, once a week for 3 consecutive weeks. Histological analysis, carried out on hematoxylin and eosin stained tissue sections, did not show significant changes in lung, heart, kidney, liver and spleen compared with untreated (naive) tissues (Figure 8). This initially suggests that the NC-PEG-GlaB formulation can be administered via systemic injection. Extensive biocompatibility studies of this nanoformulation need to be assessed in the future.

Discussion
CSCs are emerging as key therapeutic targets in oncology, as their efficient elimination may offer long-lasting tumor remission or even potentially disease-free treatment outcomes for patients. In this regard, a substantial amount of evidence suggests that CSCs rely on ligand-receptor-mediated signaling pathways, namely Hh, Wnt and Notch, for strict control of cellular functions that impart cell fate, namely proliferation, self-renewal and differentiation [44,45]. Aberrant Hh signaling is found in the majority of the human cancers, including brain tumors [18], and is usually caused not only by mutations in Hh pathway components (e.g., receptors) but also by high expression of Hh ligands [13], Hh signaling, which plays an important role in stem cell renewal, is epigenetically regulated in  [46]. Targeting Hh signaling components or epigenetic regulators in CSC-driven tumors could therefore be of therapeutic interest.

CSCs mainly Gli transcription factors
To date, an inhibitor of Hh pathway receptor (Smo) -GDC-0449 -has been approved by US FDA and is currently used in clinics, whereas only a few Gli inhibitors have been identified [47,48]. Therefore, novel drugs targeting Gli family members or downstream effectors would be beneficial for a wide spectrum of patients whose tumors have high Gli protein levels and/or activity. Furthermore, downstream activation of Gli function has been reported [49], with the development of resistance to Smo antagonists being observed fre-quently during therapy. Recent in silico and in vitro screens of a library of natural compounds identified a lead candidate for Gli inhibition -GlaB [22]. GlaB, bearing the 5,7-dimethoxyisoflavone nucleus, was isolated and characterized from the seeds of D. glabrescens, a Brazilian plant, in 1977 by Delle Monache et al. [21]. Since developed extraction methods allow us to get very limited amounts of pure isoflavone, here we provided the first total synthesis of GlaB which foresees a six-step route with an overall yield of 7%.
The in vitro and in vivo GlaB activities were also tested in our previous study, with results showing impaired Hh oncogenic activity via inhibition of Gli1/ DNA interaction [22]. This provided proof of concept for the therapeutic relevance of an Hh-targeting approach focused on a downstream Gli effector rather than on upstream oncogenic deregulated signals. However, the native GlaB molecule is insoluble in water (solubility is 0.02 μg/ml), which will likely translate into poor bioavailability in vivo. In a previous study from our group, the efficacy of GlaB against MB-bearing and BCC-bearing mice was assessed [22]. A solution of GlaB, prepared in 2-hydroxypropyl-βcyclodextrin:ethanol (3:1), was injected subcutaneously, compromising drug bioavailability. In this study, the use of ethanol was replaced with the long-circulating NC-GlaB, allowing intravenously injection of relatively high GlaB dose.
We have previously developed PEG-conjugated polymeric oil-cored NCs to enhance systemic delivery of therapeutic agents and active natural compounds as quercetin and curcumin [27,28]. Increased tumor uptake of folic-acid-conjugated NCs, containing quercetin, was observed after intravenous administration of in mice via passive and active targeting, while passively targeted curcumin-loaded PEG-PLGA NCs enhanced curcumin release into the tumor site after systemic administration, when compared with the free drug [28]. According to these studies, in an attempt to formulate NCs characterized by smallest size, narrowest PDI, higher colloidal stability and drug loading, different oil-cored types were used [28]. Among all, castor oil was found to be the one capable of obtaining the desired NCs [28]; thus, it has been selected as the most promising system to enhance GlaB bioavailability. In particular, GlaB was successfully loaded into NCs composed of castor-oil-cored offering approximately 70-fold increase in GlaB aqueous solubility (solubility in NC: ∼700 μg GlaB/ml of NC). Regarding the choice of the polymer, different PLGA polymer derivatives were used depending on the experiment that needed to be performed. Both PEG-PLGA and DTPA-PEG-PLGA were obtained through organic synthesis [28] and, in accordance with the previous studies [27,28], Polymeric glabrescione B nanocapsules for tumor therapy in vitro Research Article the conjugation of PLGA with PEG leads to a slight increase (less negative) in surface charge of NC-PEG and NC-PEG-GlaB, probably due to neutralization of the negative carboxylate groups of PLGA (Table 1 &  Supplementary Table 1). Importantly, empty NCs and NC-GlaB showed similar hydrodynamic size and stability over time, which suggests that the encapsulation of GlaB does not affect NC size and long-term storage. Comparable biological evaluation results were obtained for both NC and NC-PEG (Supplementary Figure 4) when tested in IENS at the three time points. The good in vitro serum stability data (<20% release over 24 h) suggest that there will be minimal drug release from the NC in blood circulation and that most of the drug will be released at the target site, in other words, the tumor.
A series of in vitro functional assays were then performed to evaluate and compare the anticancer properties of NC-GlaB and free GlaB. Cell lines for these studies were chosen, after literature consultation, on the basis of Hh status (dysregulated Hh pathway) and presence of stem-like cells in the cell population. In vitro results obtained from cell viability, cell cycle analysis and western blotting confirmed that CSC cell lines (G7, IENS) were more sensitive to GlaB than non-stem cancer cells (GL261, PANC1 and SKOV3), an effect that was mediated by Hh regulator Gli, which is more pronounced in CSCs. This was further confirmed by the fact that differentiated G7 cells were drug insensitive. Future studies will be performed to prove unequivocally that Hh signaling is dysregulated in the CSCs used in this study (G7, IENS).
The fact that GlaB was effective in killing CSCs at very low concentrations (∼70% cell killing with 1 μM for 48 h) is very encouraging and suggests a promising in vivo therapeutic outcome if such concentrations can be achieved in vivo following systemic injection. Moreover, these results are consistent with previous in vitro studies, in which GlaB (5 μM) was shown to inhibit proliferation of ex vivo cultured mouse granule cell progenitors, MB stem cells isolated from Ptch1 ± mice tumors and ASZ001 BCC cells, with a strength similar to GANT61 (10 μM) [22]. Indeed, GANT61 has become the reference compound for CSC-targeted therapies as it proved to inhibit the Hh pathway at Gli level in several cancer cells, CSCs and tumor animal models, such as pancreatic [17], prostate [14], lung [12], colon and hepatocellular cancer [50,51], as well as in tumors for which the role of Gli was not elucidated before, including ovarian cancer [52], esophageal adenocarcinoma and melanoma [8,53]. However, GANT61 inhibits Gli-mediated transcription by an unclear mechanism -most likely by inducing posttranslational modifications of Gli -with a reported IC 50 of 5 μM (in HEK293 cell line). In our study, GlaB and NC-GlaB were shown to be more effective than GANT61 in inducing tumor cell death, with IC 50 values in PANC1 of 1.62 and 1.49 μM, respectively.
Following up the positive in vitro studies, it was necessary to overcome the poor water solubility and bioavailability of GlaB so that it can be validated for clinical application. Our previous studies assessed organ distribution of PEG-NCs in a leaky solid tumor model -CT26 colon cancer -and approximately 0.8% ID/g of tissue in tumors was achieved 24 h postinjection [28]. Furthermore, curcumin, used as an anticancer drug in that study, has shown to be delivered to the tumor at therapeutic concentrations leading to significant tumor growth delay [28]. In preparation for preclinical testing of therapeutic activity, we first assessed organ biodistribution in PANC1-tumor-bearing mice. This tumor model is known to be highly resistant and not sufficiently leaky [17]. Such data will give us some guidance on the next stages required to prepare glioma-targeting NC formulation to deliver GlaB to brain cancer. Our results have shown that values of 0.5-1.25% ID/g tissues were achieved in tumors 24 h postinjection. Taking into account the fact that the dose administered was approximately 370 μg GlaB per mouse, one would expect that drug concentrations in the tumor to be approximately 1.9-4.4 μg/g tumor (4-10 nM). These concentrations are below the therapeutic concentrations used in PANC1, and it is therefore suggested that higher or multiple doses of the formulation is required.
The results obtained from this work gave us some guidance on the next stages, to use NCs in a leaky tumor. We are also currently working on reducing the size of our NCs to less than 150 nm as smaller sizes are expected to have better extravasation than 200 nm NCs, before testing them in other tumor models for therapy. Moreover, the therapeutic efficacy of these NCs will be tested against CSCs isolated from other types of tumor, such as pancreas, or Hh-driven tumors such as MB.

Conclusion
The efficient encapsulation of GlaB into oil-cored NCs (NC-GlaB) greatly improved the bioavailability and potential clinical application of this drug. More importantly, NC-GlaB retained the same antitumoral activity of the free drug, both revealing remarkable cell killing activity against Hh-dependent CSC lines and, to a lesser extent, non-stem cancer cells. Substantial evidence is provided for a nanoparticle-based drug with good circulation profile that could be useful as part of a multimodal treatment for patients affected by Hh-dependent CSC-driven tumors.

future science group
Research Article Al-Jamal, Botta, Ingallina et al.

Supplementary data
To view the supplementary data that accompany this paper please visit the journal website at: www.futuremedicine.com/ doi/full/10.2217/nnm-2016-0388

Acknowledgements
The authors would like to acknowledge networking contribution by the COST Action CM-1407 'Challenging organic syntheses inspired by nature - from natural products chemistry to drug discovery.' No writing assistance was utilized in the production of this manuscript.

Ethical conduct of research
The authors state that they have obtained appropriate institutional review board approval or have followed the principles outlined in the Declaration of Helsinki for all human or animal experimental investigations. In addition, for investigations involving human subjects, informed consent has been obtained from the participants involved.

Synthesis of GlaB
• The total synthesis of GlaB in six steps with an overall yield of 7% is reported for the first time.

Physico-chemical characterization of NCs
• GlaB was successfully loaded into NCs composed of castor-oil-cored NCs offering approximately 70-fold increase in GlaB aqueous solubility. • Hydrodynamic diameter, polydispersity index, ζ potential and stability were determined. • In vitro serum stability data suggest that there will be minimal drug release from the NC in blood circulation and that most of the drug will be released at the target site, in other words, the tumor.

In vitro functional assays
• Cell viability, cell cycle analysis and western blotting confirmed that cancer stem cell lines (G7, IENS) were more sensitive to GlaB than non-stem cancer cells (GL261, PANC1 and SKOV3). • Differentiated G7 cells were drug insensitive as they could be less dependent on Hh pathway.

In vivo biodistribution
• The biodistribution of radiolabeled NC-GlaB was investigated in pancreatic-tumor-bearing SCID/Beige mice and 0.5-1.25% ID/g tissues were achieved in tumors at 24 h postinjection.