Novel self assembling nanoparticles for the oral administration of fondaparinux: synthesis, characterization and in vivo evaluation.

Fondaparinux (Fpx) is the anticoagulant of choice in the treatment of short- and medium-term thromboembolic disease. To overcome the low oral bioavailability of Fpx, a new nanoparticulate carrier has been developed. The nanoparticles (NPs) contain squalenyl derivatives, known for their excellent oral bioavailability. They spontaneously self-assemble upon both electrostatic and hydrophobic interactions between the polyanionic Fpx and cationic squalenyl (CSq) derivatives. The preparation conditions were optimized to obtain monodisperse, stable NPs with a mean diameter in the range of 150-200 nm. The encapsulation efficiencies were around 80%. Fpx loadings reached 39 wt.%. According to structural and morphological analysis, Fpx and CSq organized in spherical multilamellar ("onion-type") nanoparticles. Furthermore, in vivo studies in rats suggested that Fpx was well absorbed from the orally administered NPs, which totally dissociated when reaching the blood stream, leading to the release of free Fpx. The Fpx:CSq NPs improved the plasmatic concentration of Fpx in a dose-dependent manner. However, the oral bioavailability of these new NPs remained low (around 0.3%) but of note, the Cmax obtained after oral administration of 50mg/kg NPs was close to the prophylactic plasma concentration needed to treat venous thromboembolism. Moreover, the oral bioavailability of Fpx could be dramatically increased up to 9% by including the nanoparticles into gastroresistant capsules. This study opens up new perspectives for the oral administration of Fpx and paves the way towards elaborating squalene-based NPs which self assemble without the need of covalently grafting the drug to Sq.

Introduction

Since its introduction in the market in 2002, fondaparinux (Fpx, 1, Fig. 1) became the anticoagulant of choice in the treatment of short- and medium-term thromboembolic disease with the unfractionated heparin (UFH) and low molecular weight heparin (LMWH) [1]. Fpx is a synthetic analog of the antithrombin-binding pentasaccharide found in heparin [2]. Although its molecular weight (1728 Da) is much lower than the one of LMWH, Fpx remains a hydrophilic polyanionic macromolecule showing a very low bioavailability (> 1%) by oral route [3]. This low oral absorption is the result of: i) poor transport through the intestinal epithelial barrier; ii) pronounced instability in acidic pH conditions in the stomach and iii) fast enzymatic degradation [4]. Due to these drawbacks, Fpx is only administrated via subcutaneous route [5].

Fig. 1

Chemical structures of fondaparinux sodium (Fpx) and the cationic squalenyl derivatives salts (CSq).

It is not questionable that oral delivery is the preferred route of administration. Few attempts have recently been made to develop emulsions as oral delivery forms of Fpx [6-8]. However, the preparation methods of the Fpx-based emulsions are complicated, need heating and up to five different excipients [6-8]. Moreover, the Fpx loadings were low, less than 1 wt.% [6,7] or 5 wt.% [8]. In this context, the aim of the present study was to develop an innovative oral form of Fpx, able to associate large amounts of the drug by using a simple method with only one excipient.

Thus, to improve the poor oral bioavailability of Fpx, we have envisioned associating it to squalene (SQ), a natural lipid well-known for its excellent oral absorption of more than 60% [9,10]. Moreover, SQ derivatives have the property to self-assemble as stable nanoparticles (NPs) in water [9,11]. It could therefore be expected that Fpx would be protected from degradation by encapsulation into SQ-based NPs. To form NPs, a first approach was to synthesize amphiphilic Fpx-SQ conjugates (data not shown). However, chemical conjugation of Fpx and SQ was a tremendous synthetic challenge and had led to the loss of the anticoagulant properties of Fpx (data not shown). Alternative strategies based on ion-pairing to associate Fpx with SQ derivatives appeared to be much more appealing. Therefore, two lipophilic cationic squalenyl derivatives (CSq) (2–3, Fig. 1) were synthesized here to enable ion-pair formation with polyanionic Fpx.

Ion-pairing has previously been shown to be a potential approach for improving the oral bioavailability of heparin [12-16]. The advantage of this method is that cationic molecules such as polycationic-lipophilic-core dendrons [13], chitosan derivatives [14,15] or deoxycholylethylamine (DCEA) [16] efficiently interacted with heparin without changing its chemical structure, thus avoiding the risk of reducing heparin's anticoagulant activity [16,17]. As an illustration of this strategy, Lee et al. synthesized a cationic bile acid derivative to associate with LMWH through the formation of ion pairs [16]. However, the oral bioavailability of the complex was low (3%) with a high administered dose of 50 mg/kg [16]. Chitosan derivatives self-assembled with LMWH in nanocomplexes that were able to protect heparin from enzymatic degradation in the gastrointestinal tract (GIT) [14,15]. Despite these interesting results and to the best of our knowledge, no study has as yet considered the oral administration of Fpx using the ion pairing approach.

We report here on the synthesis and characterization of new CSq, on the formation and characterization of self-assemblies in aqueous media with high Fpx loadings and on the physicochemical stability of the resulting NPs. Finally, preliminary in vivo studies have investigated the anticoagulant activity of this nanoparticulate system after intravenous and oral administrations in rats.

Materials and methods

Drugs and chemicals

Fondaparinux (Fpx, Arixtra®, 10 mg/0.8 mL) was purchased from GlaxoSmithKline (UK). Squalene (SQ) was purchased from Sigma-Aldrich Chemical Co. (France), lithium chloride and trimethylamine hydrochloride from Alfa Aesar (France). Acetone, absolute ethanol, diethyl ether, dimethylformamide and dichloromethane were obtained from Carlo Erba (Italy). Filtered MilliQ water (Millipore®, France) was used. Glucose, glycerol, trehalose, sodium phosphate dibasic, sodium phosphate monobasic, Nile red and citrate concentrated solution were purchased from Sigma-Aldrich Chemical Co. (France). Hard gelatin capsules (size 9el) and capsule feeding needle were purchased from Harvard Apparatus (France). Eudragit L100® was obtained as a gift sample from IMCD (France).

General

IR spectra were obtained as solid or neat liquid on a Fourier Transform Bruker Vector 22 spectrometer. Only significant absorptions are listed. The 1H and 13C NMR spectra were recorded with Bruker Avance 300 (300 and 75 MHz, for 1H and 13C, respectively) or Bruker Avance 400 (400 and 100 MHz for 1H and 13C, respectively) spectrometers. Mass spectra were recorded with a Bruker Esquire-LC instrument. Elemental analyses were performed by the Microanalysis Service in ICSN–CNRS, Gif-Sur-Yvette — France. Analytical thin-layer chromatography was performed with Merck silica gel 60 F254 glass precoated plates (0.25 mm layer) and Merck aluminum oxide 60F254 neutral sheets. Column chromatography was performed with Merck silica gel 60 (230–400 mesh ASTM) and Fluka aluminum oxide type 507C neutral. All reactions involving air- or water-sensitive compounds were routinely conducted in oven- or flame-dried glassware under a positive pressure of nitrogen. Except as otherwise indicated, all reactions were carried out in distilled solvents. Triethylamine was distilled over calcium hydride. Chemicals obtained from commercial suppliers were used without further purification.

Synthesis and characterization of Sq+2

(4E,8E,12E,16E)-4,8,13,17,21-pentamethyldocosa-4,8,12,16,20-pentaen-1-yl methanesulfonate (5)

To a stirred solution of trisnorsqualene alcohol 4 (582 mg, 1.5 mmol) in anhydrous CH2Cl2 (7 mL) at 0 °C, was added DMAP (10 mg) and dropwise, triethylamine (224 mg, 2.2 mmol) followed by methanesulfonyl chloride (205 mg, 1.8 mmol). The mixture was then slowly raised to room temperature and stirred for 2 h. The reaction was then quenched with brine and the mixture was extracted with CH2Cl2 (4 × 50 mL). The combined organic extracts were washed with brine, dried over MgSO4 and concentrated in vacuo. The crude 1,1′,2-trisnorsqualenyl methanesulfonate (5) (695 mg, 85%) was used directly without further purification.

1H NMR (300 MHz, CDCl3) δ: 5.14–4.96 (m,5H, CH vinyl), 4.12 (t, 2H, J = 6.5 Hz, CH2O), 2.91 (s, 3H, CH3SO2), 2.08–1.90 (m,18H, CH2), 1.90–1.76 (m, 2H, CH2, MsOCH2CH2), 1.58 (s, 3H, HC = C(CH)2), 1.53 (s, 3H, 15H, HC = C(CH)CH2)

13C NMR (300 MHz, CDCl3) δ: 135.1 (Cq), 134.9 (Cq), 134.8 (Cq), 132.8 (Cq), 131.2 (Cq), 125.8 (CH), 124.5 (CH), 124.4 (CH), 124.2 (2 CH), 69.7 (CH2, CH2OMs), 39.7 (2 CH2), 39.6 (CH2), 37.3 (CH3, OSO2CH3), 35.1 (CH2), 28.2 (2 CH2), 27.2 (CH2), 26.7 (CH2), 26.6 (CH2), 26.5 (CH2), 25.6 (CH3, HC = C(CH)2), 17.6 (CH3), 16.1 (CH3), 16.0 (3 CH3), 15.8 (CH3).

(6E,10E,14E,18E)-22-chloro-2,6,10,15,19-pentamethyldocosa-2,6,10,14,18-pentaene (6)

To a stirred solution of trisnorsqualenyl methanesulfonate (5) (500 mg, 1.08 mmol) in anhydrous DMF (5 mL) was added LiCl (456 mg, 10.8 mmol). The reaction mixture was heated at 80 °C for 2 h. After cooling to room temperature, the mixture was concentrated under reduced pressure. The residue was taken in water then extracted with diethyl ether (5 × 12 mL). The combined organic phases were dried over MgSO4, filtered and concentrated in vacuo. The crude product was purified by chromatography on silica gel (cyclohexane/EtOAc, 98:2 v/v) to give the 1-chlorotrisnorsqualene (6) as a colorless oil (377 mg, 86%). IR (neat, cm− 1) v = 2855–2970, 1441, 980, 832.

1H NMR (300 MHz, CDCl3) δ: 5.25–5.02 (m, 5H, CH vinyl), 3.56 (t, 2H, J = 6.7 Hz, CH2Cl), 2.20–1.91 (m, 18H, CH2), 1.85–1.90 (m, 2H, ClCH2CH2), 1.61 (s, 3H, HC = C(CH)2), 1.53 (m, 15H, HC = C(CH)CH2).

13C NMR (300 MHz, CDCl3) δ: 135.1 (Cq), 134.9 (Cq), 134.8 (Cq), 133.0 (Cq), 131.2 (Cq), 125.6 (CH), 124.5 (CH), 124.4 (CH), 124.3 (2 CH), 44.5 (CH2, CH2Cl), 39.8 (2 CH2), 39.7 (CH2), 39.6 (CH2), 36.6 (CH2), 30.8 (CH2), 28.3 (2 CH2), 26.8 (CH2), 26.7 (CH2), 26.6 (CH2), 25.7 (CH3, HC = C(CH)2), 17.7 (CH3), 16.0 (CH3), 15.9 (2 CH3), 15.8 (CH3).

Trimethyl[(4E,8E,12E,16E)-4,8,13,17,21-pentamethyldocosa-4,8,12,16,20-pentaen-1-yl]azanium chloride (2)

A 50% aqueous sodium hydroxyde solution (20 mL) was added dropwise by means of a dropping funnel to trimethylamine hydrochloride (3.0 g, 31.4 mmol) placed in a distilled flask. The evolved trimethylamine gas was passed through a washed bottle containing sodium hydroxide pellets and allowed to bubble through 4.0 g of anhydrous ethanol placed in another wash bottle and exactly weighed. The obtained Me3N (2.2 g, 37.2 mmol) ethanol solution was added to 1-chlorotrisnorsqualene (6) (300 mg, 0.71 mmol) in a screw cap sealed tube equipped with a stirring bar. The reaction mixture was stirred at 100 °C for 72 h. After cooling to room temperature, the mixture was concentrated under reduced pressure to provide the tertiary ammonium salt 2 as a pale yellow oil (294 mg, 98%). The crude product was used without further purification.

1H NMR (300 MHz, [D4] MeOH) δ: 5.26 (t, J = 6.0 Hz, 1H, CH vinyl), 5.25–5.15 (m, 4H, CH vinyl), 3.40–3.30 (m, 2H, Me3NCH2), 3.21 (s, 9H, (CH3)3 N+), 2.20–1.80 (m, 20H, CH2), 1.69 (s, 6H, CH3), 1.62 (s, 12H, CH3).

13C NMR (300 MHz, [D4] MeOH) δ: 136.8 (2Cq), 1.36.7 (Cq), 134.7 (Cq), 132.8 (Cq), 128.1 (CH), 126.45 (CH), 126.4 (CH), 126.3 (2CH), 68.5 (CH2, CH2NMe3), 54.5 (CH3, N(CH3)3), 54.45 (CH3, N(CH3)3), 54.4 (CH3, N(CH3)3), 41.7 (2CH2), 41.6 (CH2), 37.9 (CH2), 30.1 (2CH2), 28.7 (CH2), 28.4 (CH2), 26.8 (CH3, HC = C(CH)2), 23.1 (CH2), 18.7 (CH3), 17.1 (3CH3), 16.7 (CH3).

MS (ESI): m/z (%) = 428 (100).

Anal. calcd for C30H54ClN (%): C 77.62, H 11.73, N 3.02. Found: C 77.21, H 11.71, N 2.90.

Synthesis and characterization of Sq++

(4E,8E,12E,16E)-20-(methanesulfonyloxy)-4,9,13,17-tetramethylicosa-4,8,12,16-tetraen-1-yl methanesulfonate (8)

To an ice-cooled stirred solution of 1.20-bis-trisnorsqualene alcohol (7) (1.2 g, 3.30 mmol) and few crystals of DMAP (10 mg) in anhydrous CH2Cl2 (20 mL) was added dropwise triethylamine (1.27 mL, 9.90 mmol) followed by methanesulfonyl chloride (564 μl, 7.92 mmol). The reaction mixture was stirred at room temperature for 4 h and water (40 mL) was added. The mixture was extracted with CH2Cl2 (4 × 30 mL). The combined organic extracts were washed with brine, dried over MgSO4 and concentrated in vacuo. The residue was purified by flash chromatography on silica gel (AcOEt/cyclohexane 1:2) to provide pure bis-mesylate 8 as colorless oil (800 mg, 67%).

1H NMR (300 MHz, CDCl3) δ: 5.09–5.06 (m, 4H, CH vinyl), 4.10 (t, 4H, J = 6.5 Hz, CH2OMs), 2.93 (s, 6H, CH3SO2), 2.05–1.89 (m, 16H, CH2), 1.73–1.82 (m, 4H, CH2CH2OMs), 1.53 (s, 12H, HC = C(CH)CH2).

Trimethyl[(4E,8E,12E,16E)-4,9,13,17-tetramethyl-20-(trimethylazaniumyl)icosa-4,8,12,16-tetraen-1-yl]azanium dimethanesulfonate (3)

A solution of Me3N (4.0 g, 67.7 mmol) in ethanol (9 mL) was added to bis-trisnorsqualenylmethanesulfonate 8 (650 mg, 1.22 mmol) in a screw cap sealed tube equipped with a stirring bar. The reaction mixture was stirred at 100 °C for 5 days and treated as described above for compound 2 to yield salt 3 (600 mg, 92%).

1H NMR (300 MHz, [D4]MeOH) δ: 5.09–5.29 (m, 4H, CH vinyl), 3.29–3.42 (m, 4H, CH2), 3.18 (s, 18H, J = 6.7 Hz, CH3N +), 2,84 (s, 6H, CH3S), 2.01–2.33 (m, 18H, CH2), 1.82–1.89 (m, 4H, CH2), 1.41–1.76 (m, 12H, CH3).

Anal. calcd for C33H68N2O6S2 (%): C 60.15, H 10.41, N 4.38, O 15.02, S 10.04. Found: C 60.06, H 10.50, N 4.38, O 15.62, S 9.03.

Preparation and characterization of the NPs

Preparation of fondaparinux NPs

In order to prepare Fpx NPs, various amounts of Sq+ (0.33 to 13.43 mg) were dissolved in different volumes (0.1, 0.3, 0.5 and 1 mL) of solvent (acetone or absolute ethanol), whereas Sq++ (0.46 to 5.55 mg) was dissolved in 0.1 mL of absolute ethanol. The Fpx:Sq+ and Fpx:Sq++ NPs were prepared by nanoprecipitation. Briefly, 0.1 to 1 mL of Sq+ or Sq++ organic solutions were added drop-wise under stirring (500 rpm) into 1 mL of aqueous solution of Fpx (2.5 mg/mL). NP formation occurred spontaneously. Solvent was then evaporated using a Rotavapor® (Buchi, France). The NPs were ultracentrifuged at 230,000 ×g for 150 min at 4 °C to isolate the supernatants and determine the amount of non encapsulated Fpx.

The nanoparticle suspensions were stored at 4 °C in water until further use.

Physicochemical characterization of the NPs

Quasi-elastic light scattering (QELS)

The mean size (volume intensity) and polydispersity index of the NPs were determined at 25 °C by QELS using a nanosizer (Zetasizer Nano 6.12, Malvern Instruments Ltd., UK). The measurements were performed in triplicate, after 1/10 dilution of the NPs with MilliQ® water. The zeta potential was determined using a Zetasizer (Zetasizer 4, Malvern Instruments Ltd., UK) after dilution of the NP suspensions in an aqueous solution of KCl (1 mM).

Cryogenic transmission electron microscopy (cryo-TEM)

The morphology of the optimal formulation of NPs (Fpx:Sq+ 1:6) was visualized using cryo-TEM. 4 μL of aqueous NP suspension was placed on 200 mesh R2/2 Quantifoil coated copper grids (Quantifoil, Germany). The excess amount of liquid was then blotted with a Whatman no. 5 filter paper, and the grids were rapidly plunged into a liquid ethane bath at − 180 °C using a Leica EMGP Plunge Freezer (Leica Mikrosysteme GmbH, Austria). Afterward, the grids were constantly maintained in liquid nitrogen and carefully transferred to a Gatan 626 DH cryoholder (Gatan, USA). Preparations were examined at − 180 °C with a cryo-TEM Jeol 2010 F electron microscope (JEOL, USA) operating at an accelerating voltage of 200 kV. Images were recorded with 1–5 μm of underfocus under low-electron-dose conditions on Digital Micrograph 2.0.

Electron microscopy after freeze-fracture (FFEM)

The Fpx:Sq+ NPs (ratio 1:6), previously incubated with glycerol (30% v/v) used as a cryoprotectant, were visualized by TEM after freeze-fracture. Briefly, a drop of NP sample was placed on a Cooper support and immediately frozen in liquid propane cooled with liquid hydrogen, and the NPs kept in liquid nitrogen. Before TEM observations, NPs were fractured and the cut surface was sprayed with platinum (4 nm) and carbon (30 nm).

Small-angle X-ray scattering (SAXS)

The structure of the Fpx:Sq+ NPs (ratio 1:6) was further investigated by SAXS. Suspensions of nanoparticles (2.5 mg/mL) were loaded into quartz capillaries (diameter 1.5 mm, Glas Müller, Berlin, Germany). The top of the capillaries was sealed with a drop of paraffin to prevent water evaporation.

X-ray scattering experiments were performed on the Austrian synchrotron beamline at ELETTRA and on the SWING beamline at SOLEIL. The scattered intensity was reported as a function of the scattering vector q = 4π sinθ/λ, where 2θ was the scattering angle and λ the wavelength of the incident beam. For both instruments the calibration of the q range was carried out with silver behenate.

The Austrian SAXS beamline was operated at 8 keV. SAXS patterns were recorded using a position sensitive linear gas detector, argon-ethane filled, with sample-detector distance of 1 m. Exposure times were typically 300 s. On SWING beamline operated at 11 keV, the data were collected by a two-dimensional CCD detector. Intensity values were normalized to account for beam intensity, acquisition time and sample transmission. Each powder-like diffraction pattern, displaying a series of concentric rings, was then integrated circularly to yield the intensity as a function of q.

Preparation and characterization of freeze-dried Fpx:Sq+ nanoparticles

The freeze-dried NPs were prepared using trehalose as cryoprotectant. Briefly, trehalose at a concentration of 20 wt.% was added to an aqueous solution of Fpx (5 mg/mL). At this aqueous solution, Sq+ (8 mg) dissolved in 0.1 mL acetone was added drop-wise under stirring (500 rpm). NP formation immediately occurred. Acetone was then evaporated using a Rotavapor®. 1 mL of NP suspension was then incorporated in glass vials with flat bottom and frozen at − 8 °C for 15 h. Vials containing the frozen NP dispersions were lyophilized using an Alpha 1-2 LO plus freeze-drier to obtain the freeze-dried Fpx:Sq+ NPs in powder form. The mean particle size and the PDI of the freeze-dried NPs after rehydration were measured by QELS using a nanosizer (Zetasizer Nano 6.12, Malvern Instruments Ltd., UK) as previously indicated. The morphology of the freeze-dried NPs was investigated by cryo-TEM as previously described.

Preparation of Fpx:Sq+ NPs-filled enteric-coated capsules

The optimal formulation of lyophilized Fpx:Sq+ NPs was incorporated within hard gelatin capsules size 9el (Harvard Apparatus, France), suitable for use in rats, using a manual capsule-filling device. The NP-filled capsules were then coated with the enteric coating polymer, Eudragit L100® to achieve the gastroresistance characteristics. Briefly, the capsules were immersed in a 15% (w/v) isopropanol solution of Eudragit® L100. Trace amounts (0.2% v/v) of Nile red, a dye was incorporated in the coating polymer solution to visually observe the homogeneity of the capsule coating and further evaluate its gastroresistance. The stability of the coatings was verified in HCl solution at pH 1.2 (SGF) for 2 h and phosphate buffer at pH 6.8 (SIF).

Stability measurements

The physicochemical stability of the NPs was monitored for their size and polydispersity index by QELS for 5 days at 4 °C.

Determination of the encapsulation efficiency of Fpx into the NPs

In order to determine the amount of Fpx associated with the NPs, NP suspensions were ultracentrifuged at 230,000 ×g for 150 min at 4 °C. The supernatant was withdrawn and the concentration of Fpx was determined (see following paragraph).

The encapsulation efficiency (EE) is defined as:where Q is the percentage of the drug which was associated with the NPs and QFpx is the total amount of Fpx in the colloidal suspension.

Fpx was measured using the Azure A colorimetric method [3,10]. This method is based on the measurement of the metachromatic activity of Fpx with the dye, Azure A. Azure dye tightly binds to Fpx, thereby shifting color from blue to violet. Practically, Azure A was dissolved in water (0.04 mg/mL) and 100 μL of this solution was added to 40 μL of the supernatants recovered after NP ultracentrifugation into 96-well plate. The optical density of the solution was measured with a microplate spectra PR 2100 reader at 490 nm and the extinction was related to the Fpx concentration using a calibration curve.

Elemental analysis

Elemental analyses were performed by the Microanalysis Service of ICSN–CNRS, Gif-Sur-Yvette — France.

Composition of the Fpx:Sq+ NPs (%): C 64.65, H 9.72, N 2.90, S 4.91 which correspond to [C31H43N3Na2O49S8]·[C30H54N]8·NaCl, featuring 1 Fpx for 8 Sq+ (theoretical values: C 64.68; H 9.51; N 3.06; S 5.10).

In vivo studies

The animal experiments were carried out according to the principles of laboratory animal care and European legislation (recommendation 2007/526/EC). The protocol ethics were institutionally approved. Male Sprague-Dawley rats (250–300 g) were purchased from Harlan Laboratories and used for the in vivo studies.

The following formulations were administrated orally to the rats (n = 4) with the help of feeding needles (Harvard Apparatus, France): NPs at 25 mg/kg (eq. Fpx) and 50 mg/kg (eq. Fpx); Fpx aqueous solution at 50 mg/kg and isotonic sodium chloride solution were used as controls. The volume of the administrated NPs was 1 mL per 100 g of animal body weight. Additionally, a saline solution of Fpx and Fpx:Sq+ NPs, both at 200 μg/kg (eq. Fpx) were injected intravenously to the rats (n = 3) through the tail vein.

Blood samples (450 μL) were taken from the jugular veins and mixed directly in the syringe with 50 μL 3.8% (m/v) aqueous sodium citrate solution to prevent coagulation. Plasma was collected after centrifugation at 2300 ×g for 15 min and stored at − 20 °C until analysis. The anti-factor Xa activity in plasma was determined using the chromogenic substrate assay (STA Rotachrom® Heparin 8, Diagnostica Stago, France). This assay is based on the FXa binding with the Fpx-antithrombin complex. The FXa activity was measured using a specific chromogenic substrate in the kit STA Rotachrom® Heparin 8 (Diagnostica Stago, France). In brief, 10 μL of plasma from treated or control rat was introduced in each well of a 96-well plate, before adding 95 μL of substrate (830 μM). After incubation at room temperature for 3 min, the reaction was initiated by the addition of 75 μL FXa (to a final concentration of 4 nM in a total volume of 180 μL) and was monitored continuously at 405 nm using a microplate spectra PR2100 reader for 20 min. The Fpx plasma concentration in the sample was directly deduced from a calibration curve. Statistical tests were performed using Student test.

In order to prevent the NP degradation in gastric medium, freeze-dried Fpx:Sq+ NPs were filled into Eudragit L100® coated capsules. The capsules were administrated orally in fed male Sprague-Dawley rats as described above.

Results and discussion

Synthesis and characterization of CSq derivatives

Trimethyl (squalenyl) ammonium chloride salt (Sq+, 2) and 1,20-bis-trimethylammonium (hexanorsqualenyl) dimethanesulfonate salt (Sq++, 3) were synthesized from SQ according to Fig. 2. Briefly, to obtain Sq+ 2, the trisnorsqualene alcohol 4 was prepared by sodium borohydride reduction of the corresponding aldehyde obtained by periodic acid cleavage of 2,3-epoxysqualene according to van Tamelen [18]. Alcohol 4 was activated as methanesulfonate by reaction with methanesulfonyl chloride in the presence of triethylamine. The latter was then converted into 1-chlorotrisnorsqualene 6 by nucleophilic substitution with lithium chloride in dimethylformamide. Finally, heating of chloride 6 with a large excess of trimethylamine at 100 °C in ethanol in a sealed tube gave the desired trimethylammonium chloride salt (Sq+) 2 in 98% yield. In a similar manner, the known 1,20-hexanorsqualene diol 7 available from SQ via 2,3,22,23-dioxidosqualene [19] was converted to Sq++ 3, via the symmetrical bismesylate 8 upon treatment with trimethylamine in 92% yield as described above.

Fig. 2

Synthesis of the CSq. Reagents and conditions: a) CH3SO2Cl, Et3N, CH2Cl2, 0 °C, 2 h, 85%; b) LiCl, DMF, 80 °C, 2 h, 86%; c) Me3N, EtOH, squalenyl, 100 °C, 72 h, 98%; d) CH3SO2Cl, Et3N, CH2Cl2, 0 °C, 2 h, 67%; e) Me3N, EtOH, 100 °C, 120 h, 92%.

The 1H NMR spectra of Sq+ and Sq++ showed single peaks at 3.21 ppm and 3.18 ppm respectively, attributed to the trimethylamino group and additionally for Sq++ the mesylate peak was observed at 2.84 ppm. The presence of these signals confirmed the successful synthesis of the two CSq derivatives. These two salts were fully characterized by 1H, 13C NMR, mass spectrometry and their purity was assessed by elemental analysis including dosage of sulfur and oxygen for salt 3.

Optimization of the formulation of Fpx: Sq+ and Fpx: Sq++ NPs

NPs based on the formation of an ion-pair complex between CSq and polyanionic Fpx were prepared by nanoprecipitation (Fig. 3). Nanoprecipitation is the most common technique used to prepare NPs. It combines the advantages of a one-step preparation, a facile scale-up procedure and the use of less toxic solvents as compared to other fabrication methods [20].

Fig. 3

Schematic representation of the formation of Fpx-Sq+ NPs by nanoprecipitation. The organic phase containing SqC is added dropwise into the aqueous phase containing Fpx, leading to the instantaneous formation of nanoparticles. Solvent is removed using a rotary evaporator, forming stable suspensions of nanoparticles.

The synthesized Sq+ and Sq++ cationic derivatives were comparatively used for the formation of Fpx-loaded NPs. It was observed that the two SqC alone were not able to self-assemble as NPs in water, but formed large aggregates (when used alone in the nanoprecipitation technique). However, when used in combination with Fpx, the spontaneous formation of NPs in water was observed (Fig. 3), likely due to electrostatic CSq/Fpx and hydrophobic CSq–CSq interactions.

Particle size is an important parameter determining the in vivo fate of the NPs after oral administration and as a rule thumb, sizes smaller than 500 nm are considered to facilitate the interactions with the epithelia [15,21]. In this respect, the NP preparation was optimized to obtain stable monodisperse colloidal suspensions together with high drug loadings. For this purpose, different parameters were tested: i) the nature of the organic solvent (acetone or ethanol), ii) the volume ratio of the two phases (organic/aqueous phases) and iii) the concentration of CSq.

The first parameter needed to be identified was a solvent of CSq, also miscible with water. Since Sq++ was insoluble in acetone, NPs could be obtained only by using ethanol. In the case of Sq+, the best solvent to obtain small (around 200 nm) and monodisperse NPs was acetone.

The influence of the volume ratio acetone/water was further investigated for optimization of Fpx:Sq+ NPs (see Supporting information, § 1). The lowest volume ratio organic to aqueous phase of 1:10 was found to be the best to obtain monodisperse NPs (polydispersity of 0.07 ± 0.01) with smaller mean diameter (145 ± 6 nm). Thus, the volume ratio of 1:10 was chosen for further evaluation of the influence of the amount of CSq (Sq+ and Sq++) on the formation of the NPs. Table 1 reports the physicochemical characteristics for the two types of NPs at various Fpx–CSq ratios. In the case of NPs prepared using Sq+, whatever the experimental conditions, nanoprecipitation was successful, leading to the formation of NPs with mean diameter lower than 200 nm and polydispersity index lower than 0.2. Zeta potential values were approximately − 60 ± 5 mV. Above the threshold of molar ratio Fpx:Sq+ of 1:6, the zeta potential reached values close to zero. Taking into account that Fpx has ten negative charges, at molar ratios Fpx:Sq+ of 1:10 and 1:20 (corresponding to charge ratios of 1:1 and 1:2, respectively), the number of Sq+ per Fpx chain was likely sufficient to neutralize the Fpx charges, leading to a neutral global charge of the resulting Fpx:Sq+ NPs. As a consequence, at these values, the stability of the system was poor (see for further details Supporting information, § 2). (See Table 2.)

Table 1

Physicochemical characterization of the NPs of Fpx:Sq+ and Fpx:Sq++ with different molar ratios (mean values ± standard deviation, n = 6): measurement of mean diameter (d), zeta potential (z) and polydispersity index (PDI).

Type of NPs	Molar ratio	Charge ratio	z [mV]	d [nm]	PDI
Fpx:Sq+	1:0.5	10:0.5	− 54 ± 5	118 ± 2	0.12 ± 0.01
	1:1	10:1	− 61 ± 5	129 ± 1	0.15 ± 0.01
	1:2	10:2	− 61 ± 1	139 ± 1	0.14 ± 0.03
	1:3	10:3	− 67 ± 4	145 ± 3	0.14 ± 0.01
	1:4	10:4	− 58 ± 4	146 ± 5	0.15 ± 0.02
	1:5	10:5	− 60 ± 1	143 ± 3	0.12 ± 0.02
	1:6	10:6	− 60 ± 3	149 ± 1	0.13 ± 0.01
	1:10	10:10	− 1 ± 12	147 ± 1	0.11 ± 0.01
	1:20	10:20	− 0.5 ± 5	184 ± 5	0.18 ± 0.06
Fpx:Sq++	1:0.5	10:1	− 52 ± 10	133 ± 3	0.14 ± 0.02
	1:1	10:2	− 60 ± 5	191 ± 2	0.17 ± 0.04
	1:2	10:4	− 60 ± 2	382 ± 12	0.11 ± 0.03
	1:3	10:6	− 30 ± 2	519 ± 8	0.28 ± 0.02
	1:4	10:8	− 10 ± 5	725 ± 87	0.36 ± 0.03
	1:6	10:12	− 5 ± 10	772 ± 151	0.92 ± 0.10

Table 2

Physicochemical characterization of the Fpx:Sq+ NPs at a molar ratio of 1:3 as a function of the volume ratios (acetone/water) (mean values ± standard deviation, n = 6): measurement of mean diameter (d) and polydispersity index (PDI).

Volume ratio (acetone/water)	d [nm]	PDI
0.1	145 ± 3	0.14 ± 0.01
0.3	153 ± 9	0.09 ± 0.02
0.5	170 ± 23	0.08 ± 0.01
1	372 ± 32	0.20 ± 0.05

In contrast to Sq+, Sq++ failed to produce monodisperse NPs except at a ratio below 1:2. Indeed, aggregates or NPs with large sizes up to 772 nm were obtained in other cases (Table 1). Moreover, the NPs prepared with Sq++ were less stable than the Sq+ ones over time and precipitated within two days at 4 °C (see Supporting information, § 2).

Determination of the amount of Fpx associated with the NPs

Extraction studies

In a first approach, to determine the amount of incorporated Fpx, Fpx:Sq+ NPs were submitted to destructuration studies in the presence surfactants such as sodium cholate, a bile salt (see Supporting information, § 2). Whatever the experimental conditions, the nanoparticles could not be destructured, but only broken into smaller ones (10 to 50 nm). Therefore, the amount of Fpx inside the particles was determined by assessing the amount of non-encapsulated Fpx in the supernatants after NP centrifugation (for more details see the Materials and methods section).

Encapsulation efficiency of the Fpx:Sq+ NPs

The Azure A assay was an effective method to determine the amount of non-associated Fpx in the supernatants after NP ultracentrifugation. Fig. 4 presents the encapsulation efficiencies (EE) obtained using this assay. The best EE were obtained with molar ratios (Fpx to Sq+) of 1:6 and 1:10, showing that 85% and 87%, respectively, of Fpx was successfully associated with Sq+ under the form of NPs. Calculated corresponding loadings reached 38.5% at the ratio 1:6, which is the optimized ratio in this study. Indeed, although the NPs with the 1:10 ratio encapsulated large Fpx amounts too, they were unstable as already discussed before (see also Supporting Information, § 2).

Fig. 4

Encapsulation efficiency (EE) of Fpx within the Fpx:Sq+ NPs. The studies were carried out at a Fpx concentration of 2.5 mg/mL. Data represent mean values ± standard deviation (n = 6).

To confirm these results, a direct method (elemental analysis) was developed to determine the amount of Fpx associated with the NPs in the pellets obtained after ultracentrifugation. Elemental analysis enabled determining four elements (carbon, hydrogen, nitrogen and sulfur) which were assayed in the pellet and in the supernatant. Carbon, hydrogen and nitrogen are common to both Sq+ and Fpx, while only sulfur is present in Fpx. In the elemental analysis of the optimized NPs (ratio 1:6), all the four elements were present, which clearly confirmed the presence of Fpx. Calculations have shown that 80% of Fpx was associated with the NPs, whereas 20% remained in the supernatants. This is in agreement with the dosage of Fpx in the supernatant using the Azure A assay (ie. 85%, Fig. 4). From these data, it could be concluded that in the NPs, one Fpx molecule was associated with 8 Sq+ molecules, whereas in the supernatants, one Fpx molecule was associated with one Sq+ (see the Material and methodssection, § 2.5.5). Hence, the existence of strong interactions between Fpx and Sq+ could be confirmed, as these two components were always associated, even in the supernatants.

Encapsulation efficiency of Fpx:Sq++ NPs

For all the experimental conditions studied using the second CSq, Sq++, less than 60% of Fpx was effectively incorporated into the NPs (Fig. 5) and the resulting NPs were less stable especially at ratios > 1:3. In conclusion, among the two CSq synthesized here, Sq+ was the most adapted to encapsulate efficiently Fpx (EE over 80%).

Fig. 5

Encapsulation efficiency (EE) (%) of Fpx within the Fpx:Sq++ NPs with different ratios. Fpx concentration in the NP suspension was 2.5 mg/mL. Data represent mean values ± standard deviation (n = 6).

Supramolecular organization of the Fpx:Sq+ NPs

The morphology of the optimized Fpx:Sq+ NPs (molar ratio Fpx to Sq+ of 1:6) was investigated by cryo-TEM (Fig. 6a). The Fpx:Sq+ NPs had a regular spherical shape with an inner “onion-type” structure. Synchrotron small-angle X-ray scattering (SAXS) was used in conjunction to cryo-TEM to provide a better insight into the supramolecular organization of the NPs. A (repeat order) correlation peak at q = 0.13 Å− 1 and a very weak second order at about q = 0.26 Å− 1 were observed on the scattering curve of the NPs (Fig. 6c). Supposing a lamellar organization, a thickness of about 48 Å of the lamella could be estimated from the peak at 0.13 Å− 1. This pattern is consistent with the interlamellar distance measured on the cryo-TEM images (45–50 Å, Fig. 6a). Taking into account these considerations, the hypothesized supramolecular organization of the Fpx:Sq+ NPs is represented in Fig. 6b. Likely, the “onion”-like structure consists of hydrated layers of Fpx interacting with layers of Sq+. Presumably, both Fpx-Sq+ electrostatic interactions and Sq+-Sq+ hydrophobic interactions play a role for maintaining the cohesion of these assemblies. However, further studies would be needed to fully unravel the supramolecular organization of these NPs.

Fig. 6

Supramolecular organization of Fpx:Sq+ NPs at a ratio 1:6: A) cryo-TEM image; B) electron microscopy after freeze-fracture; C) SAXS pattern and D) hypothetical organization of Fpx (dark blue) and Sq+ (pink) in aqueous phase showing ion pair formation (Fpx:Sq+) and stabilization by interaction between lipophilic squalenyl derivatives. Bars represent 100 nm.

In vivo anticoagulant activity

The anticoagulant activity of the optimized Fpx:Sq+ aqueous nanoparticulate suspensions was evaluated in vivo after oral and intravenous (IV) administration in male Sprague-Dawley rats. Blood samples were periodically taken and the plasma concentration of active Fpx was estimated using an anti-FXa assay.

Fig. 7A shows the plasma concentration–time profile of free Fpx (saline solution) as compared to encapsulated Fpx after IV administration. Both profiles are very similar, with maximum plasma concentration (Cmax) of 0.16 mg/L and 0.12 mg/L for free Fpx and encapsulated Fpx, respectively. These results showed that the Fpx associated to the NPs could be released quickly in the blood, resulting in an anticoagulant activity similar to free Fpx. These data prove that Fpx could be effectively dissociated from the Sq+ moieties in the blood stream.

Fig. 7

Plasma Fpx concentration versus time profiles of rats (A) after intravenous administration of saline solution of Fpx (200 μg/kg) () and aqueous suspension of NPs (200 μg/kg) () (n = 3 for each studied group) and (B) after oral administration of saline solution of Fpx (50 mg/kg) (), aqueous suspension of NPs at 25 mg/kg () and 50 mg/kg () and control solution () (n = 4 for each studied group).

(C) Plasma Fpx concentration versus time profiles of rats after oral administration of Eudragit® L100-coated capsules filled with: i) free-form of Fpx (10 mg/kg) (), fasted rats; ii) freeze-dried NPs in fasted () and fed rats () (n = 6 for each studied group); control groups ().

Fig. 7B shows the plasma concentration–time profile of free Fpx (saline solution) as compared to encapsulated Fpx after oral administration. As expected, Fpx in saline solution was not absorbed. The concentration-profiles were similar to those of the control group. By contrast, administration of nanoparticulate suspensions at two distinct concentrations resulted in an increase in plasma concentration of Fpx (Fig. 7B). The NPs containing Fpx doses of 25 mg/kg and 50 mg/kg showed Cmax of about 0.11 mg/L and 0.22 mg/L, respectively with a time to maximum plasma concentration (Tmax) of around 60 min in both cases. These results revealed a dose-dependent increase in Cmax whereas the time to reach maximal concentration (Tmax) was dose-independent. Of note, the Cmax obtained after oral administration of 50 mg/kg Fpx:Sq+ was close to the prophylactic plasma concentration needed in venous thromboembolism [3].

It was concluded that the newly developed nanoparticulate system could improve the oral delivery of Fpx as compared to the free Fpx. However, the absolute bioavailability still remained low (around 0.3%) probably due to the instability of Fpx:Sq+ at the acidic pH of the stomach. To avoid this, the nanoparticles were entrapped into Eudragit L100®-coated capsules, resistant to the pH in the stomach. For this, nanoparticles have been lyophilized in the presence of trehalose, recovering the same features as before (“onion-type” structures, mean diameter and zeta potential preserved, as assessed by cryo-TEM and DLS, respectively). Fig. 7C shows the concentration–time profiles of free-form of Fpx and Fpx-loaded nanoparticles administered orally in enteric-coated capsules. A very low plasma Fpx concentration was observed in the rats orally treated with the Eudragit L100®-coated capsules filled with the free-form of Fpx. By contrast, oral administration of enteric-coated capsules filled with Fpx-loaded nanoparticles enabled reaching a maximum plasma concentration of 0.43 mg/mL at 3 h post-administration (Fig. 7C). The plasma concentrations of Fpx in the rats were increased significantly with nanoparticles (p < 0.05 at 10 mg/kg) compared with those in the control or free Fpx. In the case of nanoparticles, the area under the plasma Fpx concentration versus time curve, AUC (0–5 h) was 38.73 ± 19.01 mg·min/mL, which corresponds to a bioavailability of 9.01 ± 2.60%.

In conclusion, when administered orally to fed rats, the Fpx bioavailability was dramatically increased up to 9.01% with nanoparticles as compared to less than 0.5% for free Fpx (Fig. 7C). Significant differences (p < 0.05) were found between fasted and fed rats. Food increases around 48% the maximal plasma concentration of Fpx.

With further studies, this nanoparticulate system holds promises for the oral delivery of Fpx.

Conclusion

This study proposes a new approach to the oral administration of the anticoagulant Fpx using self-assembled NPs obtained by electrostatic interactions between cationic squalenyl derivatives and the polyanionic pentasaccharide. Optimized nanoformulations were monodisperse, stable over storage and possessed a mean diameter in the range of 150–200 nm suitable for the oral route. A multilamellar supramolecular organization has been shown by cryo-TEM and SAXS.

The absolute bioavailability of Fondaparinux could be increased up to 9% and the described nanoparticles allowed improved pharmacokinetics in a dose-dependent profile and therapeutic range. This approach opens up challenging perspectives for the oral delivery of Fpx in the future and paves the way towards elaborating Sq-based NPs which self assemble without the need of covalently grafting the drug to Sq.