Received August 17, 2007

Revised October 31, 2007

Introduction

Diabetes is one of the most common metabolic diseases. According to an intensive insulin therapy, up to 4 injections per day are required to achieve an optimal therapeutic control.^{1, 2} As modern delivery technology has a strong appeal toward patients′ friendliness, the oral administration of insulin seems a perfect way to improve patients′ compliance and achieve frequent administration of the therapeutic. A lot of new insulin applications have been brought on the market in the past few years, most favorable was the pulmonary application of insulin.^3–5 Because of the enzymatic degradation during the oral route through stomach and intestine, this way of application seems to be the most challenging yet ideal, as the compliance of especially children is greatly improved if the drug can be offered in a liquid formulation. Most oral peptide applications aim for the lower part of the small intestine.⁶ The peptide has therefore to overcome degradation by salvia, denaturizing by gastric and pancreatic fluid.⁷ As the degradation process is limited by the transport time to the point of absorption, a short transport and quick absorption, for example in the stomach, will lower the interaction with the degradation pool. Another elegant and efficient method to avoid enzymatic interference is a coating with thiomers, which are known to protect peptides from enzymatic degradation.⁸ This masking can be achieved by forming particles that complex insulin in a thiomer matrix. Complex polymer formation between poly(acrylic acid) (PAA) and poly(vinyl pyrrolidone) (PVP) has been reported via template polymerization⁹ and via the solvent diffusion method.¹⁰ The objective of both preparations was the formation of hydrogen bonds between the polymers, which is strongly dependent on the carboxylic moieties of PAA. At a pH below 4.75, the pK_a of poly(acrylic acid), the majority of carboxylic moieties will be protonated, favoring the formation of hydrogen bonds and subsequently particles.

The aim of this study was to establish a simple and effective method for the formation of complex insulin nanoparticles that should be stabilized in a gastric environment and their in vivo evaluation. This was achieved by the pH dependent formation of insulin containing particles consisting of poly(acrylic acid)−cysteine conjugate (PAA−Cys), poly(vinyl pyrrolidone) (PVP), and insulin. This new method offers numerous advantages that reach from time saving preparation over low expense, due to the avoidance of organic solvent, to the drug preserving treatment that is most notable in dealing with degradable peptide drugs. Insulin uptake should be achieved very quickly after administration and elevated even more due to the positive attributes of the thiomer poly(acrylic acid)−cysteine such as mucoadhesion,^{11, 12} permeation enhancement,¹³ and shielding against enzymatic degradation.¹⁴ The system was evaluated by detecting the serum insulin concentration as well as the pharmacological efficacy displayed in the reduction of blood glucose level after administration of solid and liquid formulations to rats.

Experimental Section

Materials. Bovine insulin ELISA was purchased from Mercodia, Uppsala, Sweden. Medisense Xtra plus glucose control strips and a measurement instrument were purchased from Abbott, UK. Poly(acrylic acid) (PAA₁₀₀, average molecular weight: 100 kDa), L−cysteine hydrochloride anhydrous, 1-ethyl-3-(3-dimethylaminopropyl)carbodiimide hydrochloride (EDAC), 5,5′-dithiobis(2-nitrobenzoic acid) (Ellman’s reagent), insulin from bovine pancreas, TRIS buffered saline tablets and trypsin−chymotrypsin inhibitor (Bowman−Birk inhibitor = BBI) were purchased from Sigma Aldrich, Vienna, Austria. Tetra sodium ethylendiamine tetra acetic acid (EDTA) was obtained from Calbiochem, Vienna, Austria. Eudragit L 100-55 was supported from Röhm/Degussa, Darmstadt, Germany. Kollidon 12 PF (poly(vinyl pyrrolidone) = PVP) was received from BASF, Ludwigshafen, Germany. Acetonitrile (ACN) and trifluoroacetic acid (TFA) were purchased from ACROS, Sankt Augustin, Germany. All other chemicals were of reagent grade and obtained from Sigma, Vienna, Austria.

Synthesis and Characterization of Poly(acrylic acid)−Cysteine Conjugate. Synthesis was performed as described previously.^{15, 16} In brief, poly(acrylic acid) with a molecular weight of 100 kDa (PAA₁₀₀−Cys) was hydrated in water. The carboxylic acid moieties were activated by the addition of EDAC to a final concentration of 130 mM for 15 min. Thereafter the pH was adjusted to 6 and L−cysteine was added to the reaction mixture in a weight ratio of 1:1 (polymer: amino acid). The reaction mixture was stirred for 3 h and afterward dialyzed. The resulting thiomer PAA₁₀₀−Cys was frozen at −80 °C and lyophilized (Benchtop 2K, VirTis, NY). The degree of modification was determined using Ellman’s reagent.^{15, 16} Neutralized polymer PAA₁₀₀ was prepared and isolated in the same way as the conjugate but omitting EDAC during the coupling reaction served as a corresponding unmodified polymer within the following studies.

Preparation of Oral Dosage Forms for In Vivo Studies. The compositions of nanoparticulate suspensions and tablets as well as the control solution are listed in Table 1. In addition, the theoretical maximum amount of insulin and the determined amount of insulin bound to formulations is shown in Table 1.

Manufacturing of Tablets. The pH of 100 mL of a 0.05% (w/v) PAA₁₀₀−Cys solution and a 0.05% (w/v) solution of unmodified poly(acrylic acid) with the molecular weight of 100 kDa was adjusted to pH 8.7 and insulin was added to each in a final concentration of 0.04% (w/v). To protect insulin from trypsinic and chymotrypsinic degradation in the small intestine, a trypsin−chymotrypsin inhibitor (Bowman−Birk inhibitor) was added to formulations in final concentrations of 0.0006% (w/v). To ensure complete dissolution of insulin, the mixtures were placed in an ultrasonic bath (USC 300 TH, VWR) at 160 W for 15 min. Thereafter, the pH of the solutions was lowered to pH 3.7 after consecutive assembling of 30 mL of a 1% (w/v) PVP solution at pH 2. Particle suspensions were stirred at room temperature for 24 h to achieve oxidation. As Kollidon 12 PF and insulin display a molecular weight of 2.5 and 5.8 kDa, respectively, dialysis of suspensions to eliminate unbound insulin would likely lead to a loss of PVP within formulations. Moreover, centrifugation as purification resulted in aggregation of particles in suspension. Therefore, a purification process was omitted. Suspensions were frozen at −80 °C and lyophilized (Benchtop 2K, VirTis, NY). Size distribution and ζ potential of particles were determined with a particle sizer (Zeta Potential/Particle Sizer, NicompTM 380 ZLS, Tokyo, Japan). Morphology of particles was monitored with an in-column energy filter transmission electron microscope (TEM) (ZEISS 902, Zeiss AG, Oberkochen, Germany). Therefore, particles were transferred onto a copper grid coated with pioloform or graphite and afterward photographed using global imaging as well as inelastic imaging with a selected energy loss of 50 eV. The degree of modification on the surface of PAA₁₀₀−Cys/PVP/insulin nanoparticles was determined via iodometric titration at pH 3.0 (1 mM iodine solution) as described previously.¹⁶ A possible interaction between PVP and iodine was taken into consideration by subtracting the iodine used to saturate interactions with PVP from the total amount of iodine needed. Thereafter the lyophilized nanoparticulate materials PAA₁₀₀−Cys/PVP/insulin and PAA₁₀₀/PVP/insulin were compressed into 2.5 mm diameter flat-faced tablets of around 25 mg with a height of 2 mm (PW 20 Ro.: 13113-589 Paul Weber, Maschinen- und Apparatebau, Remsholden-Grunbach, Germany). The compaction force was kept constant at 3 kN during the preparation of all tablets. Particle size distribution of disintegrated tablets in 100 mM PBS buffer at pH 7.4 was determined as described above. The compressed tablets were enteric coated with Eudragit L 100-55. For enteric coating, the tablets were dipped in a 3% (w/v) Eudragit L 100-55 acetone solution and air-dried. This coating procedure was repeated four times. The tablets were stored at 4 °C until oral administration to rats.

Evaluation of Polymer/Thiomer/Insulin Interaction. The interaction between each exipient and insulin was studied by the following compositions: PAA₁₀₀−Cys/insulin, PAA₁₀₀/insulin, PVP/insulin, PAA₁₀₀/PVP, PAA₁₀₀−Cys/PVP, PAA₁₀₀−Cys, PAA₁₀₀, and PVP. The components were inserted in the same concentrations as used for the preparation of the tablets. The pH was consecutively lowered from pH 8.7, either with a 1% (w/v) PVP solution of pH 2 or with distilled water of pH 2, to a pH value where turbidity occurred and formation of particles was assumed. Particle formation was evaluated with transmission electron microscopy (ZEISS 902, Zeiss AG, Oberkochen, Germany; use of a copper grid coated with graphite). Interaction between polymer or thiomer and insulin was evaluated via HPLC analysis as described below, determining a possible retention time shift appearing due to interactions. Therefore the mentioned combinations of thiomer or polymer and insulin as well as the nanoparticulate formulations PAA₁₀₀−Cys/PVP/insulin and PAA₁₀₀/PVP/insulin were qualitatively analyzed at pH 3.7.

Manufacturing of Nanoparticulate Suspensions. To provide stable and effective formulations, different amounts of PVP and PAA₁₀₀ as well as PAA₁₀₀−Cys were needed for tablets and suspensions. The pH of 7.7 mL of a 0.05% (w/v) PAA₁₀₀−Cys solution and a 0.05% (w/v) solution of unmodified poly(acrylic acid) (MW: 100 kDa) was adjusted to pH 8.7 and insulin was added to each in a final concentration of 0.6% (w/v). To ensure complete dissolution of insulin, the mixtures were placed in an ultrasonic bath (USC 300 TH, VWR) at 160 W for 15 min. Thereafter, the pH of the solutions was lowered to pH 3.7 after consecutive assembling of 2.3 mL of a 1% (w/v) PVP solution at pH 2. The purification process was, according to the mentioned quotation, omitted. For determination of drug load and release, suspensions were frozen at −80 °C and lyophilized (Benchtop 2K, VirTis, NY). Dosage forms for in vivo application were stored at 4 °C until oral administration of 500 µL to rats.

Preparation of Control Solution. As negative control, insulin and PAA₁₀₀ were dissolved in sterile filtered (pore size 0.2 µm, Sartorius, Göttingen, Germany) 100 mM PBS buffer at pH 7.4 in final concentrations of 6 mg/mL and 0.5 mg/mL, respectively; 500 µL of the solution were orally administered to rats.

Preparation of Insulin Solution for Subcutaneous Injection. Subcutaneous injection (sc) of an insulin solution served as positive control and for determination of relative bioavailability of oral formulations. For sc injection, 0.025 mg/mL of insulin were dissolved in sterile filtered (pore size 0.2 µm, Sartorius, Göttingen, Germany) 100 mM PBS buffer pH 7.4. Rats were subsequently injected with 200 µL of the solution.

Preparation of Insulin Solution for Intravenous Injection. Intravenous (iv) injection of an insulin solution served as positive control and for determination of absolute bioavailability of oral formulations. For iv injection, 0.05 mg/mL were dissolved in sterile filtered (pore size 0.2 µm, Sartorius, Göttingen, Germany) 100 mM PBS buffer pH 7.4. Rats were subsequently injected with 50 µL of the solution.

HPLC Analysis and Reliability. The amount of insulin was determined via high performance liquid chromatography (HPLC).^17–19 A Merck Hitachi Elite La Chrome HPLC system was utilized, equipped with a UV detector (Merck Hitachi Elite La Chrome) and a Multi High Bio RP 18 column (4.6 mm × 150 mm, Göhler HPLC-Analysentechnik). The mobile phase consisted of eluent A, acetonitrile (ACN), and eluent B, 0.1% [v/v] trifluoroacetic acid (TFA). A gradient elution was performed with a flow rate of 1 mL/min over 0–22 min (linear gradient from 91% A/9% B to 39% A/61% B). Insulin was quantitatively and qualitatively analyzed at a wavelength of 254 nm. The reliability of the quantification method was determined by measuring linearity, precision, accuracy, repeatability, and the detection limit of insulin with the given setup. The detection limit of the system was determined by detecting the signal-to-noise ratio of an insulin containing sample (signal) and a 100 mM PBS buffer pH 7.4 containing 5 mM EDTA and 5 mM TRIS HCl (noise). For determining the linearity of the system, eight concentrations in the range from 2 mg/mL to 0.03 mg/mL of insulin were prepared and analyzed. The limit for the correlation coefficient was set to ≥0.999. The precision was determined over a six injections of a 1 mg/mL insulin solution in 100 mM PBS buffer pH 7.4 containing 5 mM EDTA and 5 mM TRIS HCl. The accuracy was detected over three injections of three samples with the concentrations of 1 mg/mL, 0.5 mg/mL, and 0.03 mg/mL, respectively. Six independent weighed samples each containing 1 mg/mL insulin were injected for evaluating the repeatability of the method. The relative standard deviation (RSD) for retention time of insulin after 14.2 min and area under the curve (AUC) served as references for the evaluation of precision, accuracy, and repeatability. The limit for the RSD of the retention time and AUC was set to 5% and 2%, respectively.

Determination of the Drug Load of Nanoparticles for Suspensions and Tablets. To determine the drug load of nanoparticles, 2 mg of lyophilized PAA₁₀₀/PVP/insulin and PAA₁₀₀−Cys/PVP/insulin nanoparticulate material for suspensions as well as tablets were placed in 2 mL Eppendorf tubes on an Eppendorf shaker at 37 °C for 24 h at 700 rpm each containing 2 mL of a 100 mM PBS buffer pH 7.4 in addition of 5 mM EDTA, 5 mM TRIS HCl, and 10% (v/v) DMSO to ensure complete dissolution of insulin. Aliquots of 100 µL were drawn after 0 and 24 h to determine the amount of bound insulin. The withdrawn volume was substituted by release medium equilibrated to 37 °C. The amount of bound insulin was determined by HPLC analysis as described above.

In Vitro Release Studies from Tablets. According to European Pharmacopoeia 5.0, the release profiles of insulin from enteric coated tablets was first evaluated in 100 mM HCl mimicking the gastric environment for 2 h. Tablets were placed in 2 mL Eppendorf tubes, each filled with 2 mL of release medium. The release was performed on an Eppendorf shaker at 37 °C and 700 rpm in order to demonstrate the stability of the coated formulation against gastric acidity. After 2 h, the medium was replaced toward a 100 mM PBS buffer pH 7.4, containing 5 mM EDTA and 5 mM TRIS HCl mimicking the intestinal fluid. This dissolution medium differed from the European Pharmacopoeia 5.0, where a 150 mM phosphate buffer pH 6.8 is used. The system was changed to a 100 mM PBS buffer at pH 7.4, as insulin did not provide sufficient solubility in this system. The solubility of insulin was further improved by the addition of EDTA and TRIS HCl in the given concentrations. Aliquots of 100 µL were withdrawn every 30 min up to 3 h. The withdrawn volume was substituted by release medium equilibrated to 37 °C. To evaluate the interaction between the Eudragit L 100-55 coating and insulin, the release of insulin from uncoated tablets was performed in 100 mM PBS buffer pH 7.4, containing 5 mM EDTA and 5 mM TRIS HCl. The amount of released insulin was determined utilizing the HPLC method mentioned above.

In Vitro Release Studies from Nanoparticles for Suspensions. To determine the in vitro release behavior, 2 mg of lyophilized PAA₁₀₀/ PVP/insulin and PAA₁₀₀−Cys/PVP/insulin nanoparticulate material for suspensions were placed in 2 mL Eppendorf tubes on an Eppendorf shaker at 37 °C for 6 h at 700 rpm each containing 2 mL of release medium. As release medium, the mentioned 100 mM HCl as well as 100 mM PBS buffer pH 7.4, containing 5 mM EDTA and 5 mM TRIS HCl, were chosen, first mimicking the pH in a gastric environment and second the upper part of the small intestine. Aliquots of 100 µL were drawn after 0, 5, 10, 15, 30, 60, 90, 120, 180, 240, 300, and 360 min. The withdrawn volume was substituted by release medium equilibrated to 37 °C. Samples were analyzed via HPLC analysis as described above. The amount of released insulin was detected.

In Vivo Evaluation of the Delivery Systems. The protocol for the studies on animals was approved by the Animal Ethical Committee of Vienna, Austria, and adhered to the principles of Laboratory Animal Care. For in vivo studies, nondiabetic male Sprague–Dawley rats, average body weight 250 g, were used. Rats were obtained from the Institute of Labortierkunde and Genetik, University of Vienna, Austria. Rats were fasted for 12 h prior to experiments. Before the application of formulations, 200 µL of blood samples were withdrawn from the tail vein and collected in 1.5 mL Eppendorf tubes. The blood glucose value was determined immediately using blood glucose control strips and a reader (MediSense Precision Xtra Plus, Abbott, UK). Thereafter, the samples were allowed to clot for 30 min and centrifuged for 10 min at 6400 g. The serum separated as supernatant was withdrawn and stored immediately at −20 °C. The insulin concentration within the serum was determined utilizing an ELISA test kit for bovine insulin (Mercodia, Sweden). Both blood glucose values and serum insulin concentrations determined at time point zero served as reference values for the following experiments. To evaluate the impact of oral dosage forms, cohorts of five rats were dosed, each with nanoparticulate formulation as well as the control formulation. PAA₁₀₀−Cys/PVP/insulin and PAA₁₀₀/PVP/insulin enteric coated tablets compressed out of 25 mg of lyophilized nanoparticulate material were administered to nonanaesthetized animals by placing the tablet deep into the throat with a blunt pair of tweezers. To ensure the swallowing reflex, immediately after tablet administration, 1 mL of 100 mM ascorbic acid solution was administered. For nanoparticulate suspensions (PAA₁₀₀−Cys/PVP/insulin and PAA₁₀₀/PVP/insulin) as well as the negative control, 500 µL of freshly prepared nanoparticulate suspension or insulin solution were orally administered, initiating the swallowing reflex of animals. To ensure complete swallowing of the remainder, 500 µL of 100 mM ascorbic acid solution were applied in addition. The theoretical maximum amount of insulin in enteric coated tablets and nanoparticulate suspensions as well as control solution is 3 mg (Table 1). The exact composition of nanoparticulate suspensions and tablets as well as the control solution is listed in Table 1. The insulin amounts given intravenously (iv) and subcutaneously (sc) were evaluated in preliminary studies. The amounts of 10 µg/kg iv and 20 µg/kg sc proved to be hypoglycaemia save for animals while achieving good detectable values for the determination of blood glucose levels and serum insulin concentrations. For evaluating the relative bioavailability of oral formulations, five rats received a subcutaneous injection of 200 µL of a 25 µg/mL solution in sterile filtered 100 mM PBS buffer at pH 7.4. To determine the absolute bioavailability of oral formulations, five rats received an intravenous injection of 50 µL of a 50 µg/mL solution in sterile filtered 100 mM PBS buffer at pH 7.4. During the study, dosed rats were kept in restraining cages with free access to water. Blood samples from rats that received enteric coated tablets were collected from the tail vein 1, 2, 3, 4, and 6 h after administration. Blood from rats that received the nanoparticulate suspensions and the control solution was withdrawn at 0.5, 1, 2, 3, 4, and 6 h intervals. For subcutaneous and intravenous injection, 15 min, 1, 2, 4, and 6 h intervals and 1 min, 20 min, 1, 2, 4, and 6 h intervals, respectively, were maintained. Blood glucose levels and serum insulin concentrations were evaluated according to the methods described above.

Pharmacokinetic Analysis. Calculations were performed using the computer software OriginPro 7G SR4 version 7.0552. C_maxand t_max were determined from the profiles generated by plotting the concentration of insulin (µg/L) against time (h). The areas under the concentration time curves (AUC) were calculated according to the linear trapezoidal rule. The relative bioavailability was calculated from the absolute dose and areas under curves (AUC) for oral against subcutaneous administration.

Statistical Analysis. Statistical data analysis was performed using the student t test with p < 0.05 as the minimal level of significance unless indicated otherwise.

Results

Characterization of Poly(acrylic acid)−Cysteine Conjugate. Cysteine was bound to poly(acrylic acid) with a molecular weight of 100 kDa (PAA₁₀₀) via amide bond between the carboxylic groups of the polymer and primary amino groups of L-cysteine. The poly(acrylic acid)−cysteine conjugate (PAA₁₀₀−Cys) exhibited 36.2 ± 3.3 µmol thiol groups per g of polymer.

Characterization of PAA₁₀₀/PVP/Insulin, PAA₁₀₀−Cys/PVP/Insulin Nanoparticles and Tablets. Particles containing the conjugate PAA₁₀₀−Cys, PVP, and insulin as well as control particles prepared in the same way but containing PAA₁₀₀were established by complexation as a result of hydrogen bonding and ionic interactions between components. The oxidized PAA₁₀₀−Cys/PVP/insulin particles displayed a mean particle diameter of 255 ± 25 nm and a ζ potential of −11.2 ± 1.0 mV, whereas nonoxidized particles showed a mean particle size of 185 ± 40 nm and a ζ potential −18.3 ± 2.4 mV. Control particles were slightly smaller, with values of 178 ± 31 nm and −15.4 ± 2.5 mV. TEM micrographs of oxidized PAA₁₀₀−Cys/PVP/insulin nanoparticles confirmed spherical shaped particles in the range of 184 ± 39 nm (Figure 1E). The amount of thiol groups per g of nanoparticulate material was determined to be 0.2 µmol for freshly prepared nanoparticles and 0.1 µmol for oxidized particles. Disintegration of tablets in 100 mM PBS buffer pH 7.4 led to the formation of an opalescent suspension in which particles showed an immediate aggregation process. Formed aggregates were easily redispersed. Particle size measurements showed bimodal size distributions with maxima at 123 ± 9 and 3985 ± 129 nm for PAA₁₀₀−Cys/PVP/insulin nanoparticles and mean sizes of 229 ± 51 and 1022 ± 203 nm for PAA₁₀₀/PVP/insulin nanoparticles.

Figure 1. TEM micrographs of PAA100−Cys/insulin (A), PAA100/insulin (B), PAA100/PVP (C), PAA100−Cys/PVP (D), PAA100−Cys/PVP/insulin nanoparticles (E), and crystalline structures observed in PVP/insulin formulations (F). Particles were obtained by pH dependent interaction of the components PAA100−Cys, PAA100, PVP, and insulin. Particles were photographed using global (A, B, C, D, F) as well as inelastic imaging with a selected energy loss of 50 eV (E). Bar represent 200 nm (A, B, C, D, F) and 1 µm (E). Click to Enlarge

Evaluation of Polymer/Thiomer/Insulin Interaction. To determine a possible interaction between the compounds with a resulting formation of particles, the pH of solutions of nanoparticulate exipients PAA₁₀₀−Cys, PAA₁₀₀, PVP, and insulin as well as their combinations PAA₁₀₀−Cys/insulin, PAA₁₀₀/insulin, PAA₁₀₀/PVP, PAA₁₀₀−Cys/PVP, and PVP/insulin were consecutively lowered from pH 8.7 to a pH value where turbidity occurred. Opalescent cloudiness within the systems PAA₁₀₀−Cys/insulin, PAA₁₀₀/insulin, and PVP/insulin already occurred at pH 5.1−5.2. In contrast, turbidity was observed at pH 3.7 for formulations containing PAA₁₀₀/PVP/insulin or PAA₁₀₀−Cys/PVP/insulin (Figure 1E). Particle formation was evaluated with transmission electron microscopy (ZEISS 902, Zeiss AG, Oberkochen, Germany; use of a copper grid coated with pioloform or graphite), where particles in the range of 100 nm were observed for PAA₁₀₀−Cys/insulin (Figure 1A), PAA₁₀₀/insulin (Figure 1B), and PAA₁₀₀/PVP (Figure 1C) nanoparticles. PAA₁₀₀−Cys/PVP (Figure 1D) nanoparticles were in the range of 190 nm, comparable to particles obtained with PAA₁₀₀−Cys/PVP/and insulin as mentioned above (Figure 1E). No particles were observed in suspension containing PVP/insulin, but crystalline structures occurred (Figure 1F). Interactions between insulin and polymer or thiomer were in addition evaluated via the determination of possible retention time shifts with reversed HPLC analysis. A retention time for insulin in the formulations PAA₁₀₀/insulin, PAA₁₀₀−Cys/insulin, PVP/insulin, PAA₁₀₀/PVP/insulin, and PAA₁₀₀−Cys/PVP/insulin in the range from 14.1 to 14.2 was observed, giving no sign of interaction via the used method.

HPLC Method Reliability. Insulin was detected with the given method and instrumental set up after a retention time of 14.2 min. The evaluation of the system showed a detection limit of 0.03 mg/mL for insulin. The assay was linear over a concentration range from 2 to 0.03 mg/mL. Precision showed a RSD of 0.1% for retention time and 0.3% for AUC of insulin. The accuracy of the system is given with a RSD of 0.2% for the retention time of insulin solutions in the concentration of 1 and 0.5 mg/mL. Insulin in the concentration of 0.03 mg/mL displayed an RSD for retention time of 0.1%. For insulin in the concentrations of 1, 0.5, and 0.03 mg/mL, the RSD of AUC were determined to be 0.5%, 0.8%, and 0.3%, respectively. The repeatability of the used HPLC method for the quantification of insulin in a concentration of 1 mg/mL is acceptable, with a RSD of 0.03% for the retention time and a RSD of 1.2% for the detected AUC.

Determination of the Drug Load of Nanoparticles for Suspensions and Tablets. The insulin content bound to PAA₁₀₀−Cys/PVP/insulin and PAA₁₀₀/PVP/insulin lyophilized nanoparticulate material for suspensions as well as tablets was measured over the release of insulin in 100 mM PBS buffer pH 7.4 containing 5 mM EDTA, 5 mM TRIS HCl, and 10% (v/v) DMSO via the described HPLC analysis. At time points 0 and 24 h, 100 µL samples were withdrawn. The withdrawn volume did not show any turbidity. Therefore, the separation of nanoparticles was assumed. The drug load was determined to be 2.4% (w/w) and 2.8% (w/w) for tablet nanoparticulate material containing PAA₁₀₀ or PAA₁₀₀−Cys, respectively. The amount of bound insulin for nanoparticulate suspensions was determined to be 16.5% (w/w) for PAA₁₀₀−Cys/PVP/insulin and 17.2% (w/w) for PAA₁₀₀/PVP/insulin.

In Vitro Release of Insulin from Tablets. Diffusion studies in acidic medium showed that no insulin was released at all from enteric coated formulation within 2 h of incubation. Drug release took place after the enteric coated tablets had been placed in 100 mM PBS buffer pH 7.4 containing 5 mM EDTA and 5 mM TRIS HCl where Eudragit L 100-55 was dissolved. Release profiles of insulin from enteric coated formulations over the period of 3 h at pH 7.4 are shown in Figure 2. A delay in release was detected for both formulations containing either polymer or thiomer within the first 30 min of detection. After 60 min of incubation for PAA₁₀₀−Cys/PVP/insulin tablets and 120 min for PAA₁₀₀/PVP/insulin tablets, a zero-order release could be observed where 37.4 ± 3.1% and 40.1 ± 10.2% of the total insulin content within the formulation was released. After 60 and 120 min, the release profiles of PAA₁₀₀−Cys/PVP/insulin tablets and PAA₁₀₀/PVP/insulin tablets, respectively, reached a plateau phase. Around 60% of the total insulin content within formulations was released from noncoated tablets (data not shown) containing thiomer and unmodified polymer, respectively, over an observation period of 180 min. A steeper dissolution gradient occurred for noncoated solid oral dosage forms compared to enteric coated formulations.

Figure 2. Release of insulin from PAA100−Cys/PVP/insulin (−■−) enteric coated tablets and PAA100/PVP/insulin (−△−) enteric coated tablets performed in 100 mM PBS at pH 7.4 containing 5 mM TRIS HCl and 5 mM EDTA. Indicated values are the means ± SD of at least 3 trials. Click to Enlarge

In Vitro Release of Insulin from Nanoparticulate Suspensions. Release studies of insulin from PAA₁₀₀−Cys/PVP/insulin nanoparticles for suspensions (Figure 3A, B) performed at an acidic pH (Figure 3A) showed an immediate release of 15.0 ± 2.0% from the total insulin amount. Within 10 min, 90.5 ± 22.5% insulin was released from particles followed by the establishment of a plateau phase. PAA₁₀₀/PVP/insulin nanoparticles showed, as seen in Figure 3C, a slightly different profile at pH 1 with 14.8 ± 2.2% of insulin released at time point zero. A burst release of 88.8 ± 20.5% of insulin after 10 min was observed before a plateau was reached. Insulin release profiles from thiomeric nanoparticles in acidic medium, shown in Figure 3A, display a decrease in detected insulin amount after 180 min. The effect was also moderately observed for polymeric nanoparticles in the same release medium after 240 min (Figure 3C). PAA₁₀₀−Cys/PVP/insulin (Figure 3B) nanoparticles released initially 17.1 ± 1.9% of insulin. Compared to the release of insulin observed at pH 1, less protein was detected within the first 10 min. Insulin nanoparticles containing the poly(acrylic acid)−cysteine conjugate showed 65.0 ± 11.0% insulin released after 10 min and an affiliated plateau phase until 120 min. After 240 min of monitoring, a release of 90.2 ± 15.3% insulin led to another plateau phase until the end of observation after 360 min. PAA₁₀₀/PVP/insulin nanoparticles showed a comparable result at pH 7.4 with 16.2 ± 1.5% insulin immediate released, a first plateau after 10 min with 55.6 ± 2.7%, and a second plateau with 96.1 ± 3.9% insulin released after 300 min (Figure 3D). The analysis of insulin via HPLC analysis showed no interaction between polymers and insulin. No additional peaks and retention time shifting appeared within the chromatograms.

Figure 3. Release of insulin from PAA100−Cys/PVP/insulin nanoparticulate suspension (A,B) and PAA100/PVP/insulin nanoparticulate suspension (C,D) in A,C, 100 mM HCl, and B,D, 100 mM phosphate buffer pH 7.4, containing 5 mM EDTA and 5 mM TRIS HCl at 37 °C. Indicated values are the means ± SD of at least 3 trials. Click to Enlarge

In Vivo Studies: Determination of Insulin in Rat Serum. The bioavailability of insulin, delivered via nanoparticulate suspensions or enteric coated tablets, was evaluated by determining the insulin concentrations in serum. The mean serum insulin concentration against time profiles obtained after sc injection, iv injection, oral administration of an insulin solution containing PAA₁₀₀ as well as PAA₁₀₀−Cys/PVP/insulin, PAA₁₀₀/PVP/insulin enteric coated tablets, and suspensions are displayed in Figure 4.

Figure 4. Serum insulin concentrations after administration of controls (A): iv injection (−●−), sc injection (−△−), oral administration of insulin solution (−■−), after oral administration of formulations containing PAA100−Cys (B): PAA100−Cys/PVP/insulin suspension (−◆−), PAA100−Cys/PVP/insulin tablets (−×−), and after oral administration of formulations containing PAA100 (C): PAA100/PVP/insulin suspension (−◆−), PAA100/PVP/insulin tablets (−×−). Rats were fed after 6 h. Indicated values are the means ± SD of at least 4 rats. Click to Enlarge

Subcutaneous and intravenous injection showed an immediate increase after 15 min for sc injection and 1 min for iv injections, resulting in an insulin concentration of 2.3 ± 0.8 µg/L and 8.0 ± 0.2 µg/L, respectively (Figure 4A). Oral administration of insulin solutions showed a negligible effect with an insulin level of 0.1 ± 0.0 µg/L over a time period of 6 h (Figure 4A). No significant difference between solid oral formulations was observed. Enteric coated PAA₁₀₀/PVP/insulin tablets showed a maximum, with 0.4 ± 0.3 µg/L detected after 1 h and another increase in serum insulin with 0.3 ± 0.3 µg/L after 6 h (Figure 4C). Enteric coated PAA₁₀₀−Cys/PVP/insulin tablets showed a maximum of 0.5 ± 0.1 µg/L after 3 h (Figure 4B). Main pharmacokinetic parameters, calculated referring to a theoretical maximum amount of insulin as well as bound insulin, are listed in Table 2. They display a relative bioavailability for PAA₁₀₀−Cys/PVP/insulin tablets of 0.2 ± 0.0% (bound) and 0.1 ± 0.0% (theoretical maximum). PAA₁₀₀/PVP/insulin tablets lead to relative bioavailabilities of 0.2 ± 0.0% by taking the amount of bound insulin into consideration. Profiles obtained in the same way as mentioned for the oral delivery of enteric coated tablets were recorded for the oral administration of nanoparticulate suspensions. Results, displayed in Figure 4B, show a quick and strong uptake, with 1.9 ± 0.3 µg/L for PAA₁₀₀−Cys/PVP/insulin nanoparticulate suspensions after 30 min. A similar but not as distinctive result was achieved for PAA₁₀₀/PVP/insulin nanoparticulate suspensions, with a serum insulin concentration of 1.2 ± 0.2 µg/L also after 30 min of administration (Figure 4C). The relative bioavailability of bound insulin in suspension formulations containing PAA₁₀₀−Cys or PAA₁₀₀ was determined to be 0.7 ± 0.1% and 0.3 ± 0.1%, respectively (Table 2). Taking the theoretical maximum amount of insulin into consideration, relative bioavailabilities of 0.2 ± 0.0% and 0.1 ± 0.0% for PAA₁₀₀−Cys/PVP/insulin and PAA₁₀₀/PVP/insulin nanoparticulate suspensions, respectively, were determined. By classifying all formulations regarding their bioavailability, the following order occurs: PAA₁₀₀−Cys/PVP/insulin nanoparticulate suspensions > PAA₁₀₀/PVP/insulin nanoparticulate suspensions > PAA₁₀₀−Cys/PVP/insulin tablets = PAA₁₀₀/PVP/insulin tablets.

In Vivo Studies: Pharmacological Efficacy. To evaluate the pharmacological efficacy of administered insulin, the decrease of the blood glucose level, from initially detected 3.6 ± 0.3 mmol/L, as a biological response to the administration of insulin, was determined. The hypoglycaemic effect induced by the oral administration of insulin solutions and iv and sc injections is shown in Figure 5A. Administration of insulin solutions led to a slight decrease in blood glucose levels over the first 2 h to 88.9% from an initial value (Figure 5A). However, no peak reduction could be observed over the monitored period of 6 h. Intravenous and subcutaneous injections showed minimum blood sugar values of 48.8 ± 14.0% and 39.8 ± 8.2% from an initial value detected 20 and 60 min after injection (Figure 5A). A slight and moderate effect was observed for PAA₁₀₀−Cys/PVP/insulin tablets (Figure 5B). The minimum glucose level was detected with 81.4 ± 10.2% of the initial value after 4 h. PAA₁₀₀/PVP/insulin tablets showed a decrement to 83.9 ± 9.4% of an initial value after 3 h (Figure 5C). Nanoparticulate suspensions showed a reduction of blood glucose levels in correlation to an initial value of 78.2 ± 4.2% after 3 h for PAA₁₀₀−Cys/PVP/insulin (Figure 5B) and of 80.3 ± 5.2% for PAA₁₀₀/PVP/insulin after 2 h (Figure 5C).

Figure 5. Decrease of the blood glucose level (% of initial value) as a biological response for the insulin administration to fasted rats after administration of control (A): iv injection (−●−), sc injection (−△−), oral administration of insulin solution (−■−), after oral administration of formulations containing PAA100−Cys; (B): PAA100−Cys/PVP/insulin suspension (−◆−), PAA100−Cys/PVP/insulin tablets (−×−), and after oral administration of formulations containing PAA100; (C): PAA100/PVP/insulin suspension (−◆−), PAA100/PVP/insulin tablets (−×−). Rats were fed after 6 h. Indicated values are the means ± SD of at least 4 rats. Click to Enlarge

All formulation showed an extended time delay between the detected maximum insulin levels and the minimum glucose levels from 1 to 2.5 h compared to the positive controls where a delay of 45 and 19 min was observed for subcutaneous and intravenous injection, respectively.

Discussion

Clinical trial status treatments of diabetes reach from injectables, solid oral and buccal dosage forms, over transdermal to pulmonary applications.⁵ Yet a nanoparticulate formulation has failed to enter the market and has not even reached clinical trials. Particulate delivery systems offer the advantage of a better tissue adherence due to their light weight, the display of a smaller dose variation referring to the large number of particles,¹⁰ and the possibility to reach greater mucosal surface areas leading to a comparatively higher drug uptake.²⁰ Within this study, we evaluated the effect of complex stabilized PAA₁₀₀−Cys/PVP/insulin nanoparticulate formulation on the serum insulin concentration and the pharmacological effect in vivo. As it is, to our knowledge, the first time that insulin was to be released from a nanoparticulate system and to be absorbed already within the stomach, we see it regarding to the present results as a first proof of principles.

Earlier attempts have been made to reach the stomach as an absorption area for peptide drugs.²¹ A combination between thiomer and a pepstatin chitosan conjugate was used in order to overcome a degradation process of the model drug salmon calcitonin with a pharmacological efficacy of 1.35%.²¹ The use of the enzyme-inhibitor pepstatin within the presented nanoparticulate formulation was abdicated as another conjugate on the used thiomer, poly(acrylic acid)−cysteine would likely hinder the formation of hydrogen bonding between free carboxylic groups of PAA₁₀₀−Cys and the carbonyl function of PVP. Also, toxic side effects for the use of pepstatin are known; for that reason, a formulation with a systemic application of not covalently bound pepstatin was diminished.²² The comparative parameter in both studies was the use of a thiomeric excipient. Because of the mucoadhesive properties of poly(acrylic acid)−cysteine, nanoparticles should initially adhere to the mucosa after entering the stomach environment. This is believed to guarantee a prolonged residence time of the drug carrier system at the absorption site and improve the drug uptake through the membrane. The insulin oral pharmacological efficacy was in earlier performed studies²³ increased due to the incorporation in a polycarbophil−cysteine matrix. As other mucoadhesive polymers with a poly(acrylic acid) backbone have shown to remain stable in the gastric environment,²⁴ we assume the mucoahesion of the introduced thiomer/insulin nanoparticulate system especially due to the postulated greater mucoadhesive properties of thiolated poly(acrylic acid) compared to nonthiolated PAA.²⁵ Particles were assumed to be established due to pH dependent hydrogen bond²⁶ formation between PAA₁₀₀ or PAA₁₀₀−Cys and PVP (Figure 1C, Figure 1D) as well as ionic interactions between insulin and polymer or thiomer, respectively, (Figure 1A,B). Ionic interactions are likely, as particles out of polymer/thiomer and insulin are formed at pH 5.1–5.2, which is slightly below the isoelectric point of insulin (pH 5.3–5.4), where the peptide already shows a cationic net charge. Poly(acrylic acid) is, with a pK_a of 4.75, at this pH mainly anionic and can therefore interact with insulin. Similar interactions are assumed for poly(acrylic acid)−cysteine due to structural analogy. No particle formation was observed between PVP and insulin. The occurring turbidity of the combined solutions at pH 5.2 is likely due to the precipitation of insoluble insulin within the system. The solubility of insulin in general was found to be very differing from literature if administered to thiomer or polymer solutions in the concentrations of 0.04 and 0.6% (w/v), respectively. Therefore pH 8.7 was used to provide complete solubility of insulin after the use of ultrasonication. The release was performed at a pH closer to in vivo circumstances, but EDTA was added to increase insulin solubility. Only insulin release from enteric coated solid oral formulations showed significant differences between tablets containing thiomer compared to unmodified polymer. The cationic net charge of insulin at pH 3.7 can, due to the buffer capacity of aniogenic polymers even in hydrated tablets, be maintained.²⁷ In earlier studies, it was assumed that the immobilization of L-cysteine on the backbone of poly(acrylic acid) causes lower interactions of cationic moieties of insulin with thiomer compared to polymer due to minor anionic moieties on the polymeric backbone.²⁸ This could yield a steeper release gradient for tablets containing PAA₁₀₀−Cys over an observation time of 60 min. In general, great differences were observed for release profiles of insulin from solid oral formulations compared to nanoparticulate suspensions. Around 40% and 60% of insulin were released from enteric coated tablets (Figure 2) and noncoated tablets (data not shown), respectively. Therefore, only a minor interaction of insulin with the coating material can be assumed, not revealing the loss of around 40% of insulin from solid oral dosage forms. A possible explanation could be the described interactions between insulin and PAA₁₀₀ or PAA₁₀₀−Cys, respectively, elevated due to the compaction force, leading to an insufficient release over 3 h. In addition, a delayed release of insulin from enteric coated formulations within the first 30 min was detected, which is most likely the swelling process of the coating material, as no lag time was observed for insulin release from noncoated tablets.

On the contrary, insulin was completely released from nanoparticulate suspensions within 6 h (Figure 3). Increased moisture expansion leads to a complete, but two cascade release of insulin from nanoparticles at pH 7.4 within 6 h (Figure 3B,D). The good solubility of insulin²¹ and poly(acrylic acid) as well as poly(acrylic acid)−cysteine at pH 1 led to about 90% of insulin released from nanoparticles already after 10 min (Figure 3A, C). The decrease in insulin release detected after 180 and 240 min for formulations containing PAA₁₀₀−Cys, PAA₁₀₀, respectively, might be due to a pH dependent interaction with used polypropylene vessels.

In vivo results display that the polymeric nanoparticulate complex plays an essential role in the absorption of insulin in an acidic environment. Incorporation of insulin during the particle formation at pH < 4 is very likely, as it displays a cationic net charge and is therefore already ionically bound to PAA₁₀₀ or PAA₁₀₀−Cys. However, ultimate evidence is not provided. Particles formed due to hydrogen bonds and ionic interactions can vice versa not be maintained if a certain pH is reached.²⁶ The intestinal environment provides a pH > 5.3, where negatively charged insulin leads to low interactions with anionic PAA₁₀₀ and PAA₁₀₀−Cys. Hydrogen bond formation strongly depends on the carboxylic moieties of poly(acrylic acid) and poly(acrylic acid)−cysteine. At pH > 4.75, the majority of carboxylic moieties will be deprotonated, leading to a minor contribution in hydrogen bonding with PVP and therefore the dissolution of particles. In combination with the observed release of insulin from enteric coated tablets, this fact is probably causing the low serum insulin concentrations for enteric coated solid oral dosage forms and the low observed pharmacological efficacies. On the contrary, nanoparticulate suspensions containing the thiomer poly(acrylic acid)−cysteine display a AUC of 4.89 ± 1.01, which almost reaches the AUC of the subcutaneous injection of 5.35 ± 0.32 (Table 2) and provide a profile comparable to subcutaneous injections of insulin in terms of the unset of action but show a reduced duration of action (Figure 5). To reduce the risk of hypoglycaemia, the amount of insulin that reaches the systemic circulation should be smaller and the peak shorter lived than that observed with injective insulin.²⁹ Therefore, the presented nanoparticulate suspensions seem to be suitable for a hypoglycaemia save oral application. Serum insulin concentration of nanoparticulate suspensions showed a significant difference between formulations containing thiolated and nonthiolated polymer (Figure 4). These results lead to the suggestion that the mucoadhesive properties of PAA₁₀₀−Cys play an essential role in the absorption of insulin from the gastric environment. A blood glucose reduction after 3 h for PAA₁₀₀−Cys/PVP/insulin is fitting in the profile of the subcutaneous injection, although the minimum reduction is not reached (Figure 5). It is assumed that bound insulin is to some extent protected from pepsin degradation, as pepsin with a molecular weight of about 30 kDa has to penetrate the carrier matrix in order to degrade the drug. But due to the comparatively bigger surface of a nanoparticulate formulation, the extent of interaction with the degradation pool is bigger as compared to a tablet formulation,²¹ which will probably result in a partial inactivation of insulin.

It can be concluded that thiomer nanoparticulate suspensions showed in comparison to enteric coated tablets an overall improvement with a rapid onset of action after oral administration, which is assumed to be due to a stronger resistance of the hydrogen bond bearing formulation within the gastric environment.

Conclusion

Within this study, a novel PAA₁₀₀−Cys/PVP/insulin nanoparticulate drug delivery systems has been evaluated in vivo. We assume that the introduced formulation provides a stabilized system within a gastric environment, resulting in a significant serum insulin concentration and a blood sugar reduction after oral administration. The study demonstrates in conclusion that this nanoparticulate formulation is a promising new tool for the noninvasive systemic delivery of insulin or other peptide drugs.

* Corresponding author. E-mail: andreas.bernkop@uibk.ac.at. Telephone: +43-512-5075383 . Fax: +43-512-5072933.

† ThioMatrix GmbH, Research Center.

‡ Institute of Pharmacy, Department of Pharmaceutical Technology, University of Innsbruck.