In-vitro release studies of insulin-loaded nanoparticles in simulated gastrointestinal fluids

Trinity College DublinSchool of Pharmacy & Pharmaceutical Sciences

In vitro release studies of insulin-loaded nanoparticles in simulated gastrointestinal fluid

Mena Eskander

Department of Pharmaceutics & Pharmaceutical Technology

P R O J E C T I N F O

1. IntroductIon

1.1 Diabetes mellitus, therapy & the patient experience Once a disease that carried a prognosis of certain death, the use of insulin in diabetes therapy has drastically improved the quality of life and prognosis of individuals suffering from the disease. Arguably, the discovery of the hormone was one of the greatest medical advancements of the 20th century. Insulin not only revolutionised diabetes therapy but made clear its crucial physiologic role and how aberrations relating to its synthesis

and use relates to the pathology of the disease. Indeed, until the discovery of insulin, physicians of the past knew little about diabetes and less on its treatment. The first known record of the disease may possibly be the Ebers papyrus which was written around 1500BC and re-discovered in 1862 AD from an Ancient Egyptian grave in Thebes, Egypt. It describes amidst a host of other ailments a disease characterised by “too great emptying of the urine”. By around 230BC, the term “diabetes” began first to be used and it was clear that polyuria and weight loss were two of the main symptoms of the disease. Over the course

A B S T R A C T

First Drafted17th October 2014

Submitted10th November 2014

SupervisorDr. Lidia TajberCo-supervisorMs. Svenja Sladek

In this work, insulin-loaded nanoparticles comprised of varying ratios of hyaluronic acid (HA) and chitosan (CS) were prepared using polyelectrolyte complexation and characterised. This generally lead to homogenous (polydispersity index generally below 0.2) and stable dispersions with surface potential being generally over |30|mV and particle size ranging from 120-300 nm. A chromatographic purification scheme was developed using a simple size-exclusion chromatography column and validated. A number of potentially suitable direct and indirect release study methods such as centrifugation, centrifugal filtration and dialysis were assessed with dynamic dialysis using a 100 kDa molecular weight cut off being deemed the most suitable of the methods tested. One species of nanoparticles, produced using 5:1 w/w HA257/CS, showed sufficiently low release of insulin at clinically relevant timepoints with only 7.55% insulin release in SGF after 4 hours dialysis. SIF data were incom-plete due to a computer malfunction and it is hoped these experiments will be repeated in the future at a more opportune time. It was observed that the presence of nanoparticles significantly increased the rate of insulin permeation across the dialysis membrane. Though the data obtained through dynamic dialysis is not without some degree of error, the results are sufficient to show promise in the carrier system and support the undertaking of further investigation in to their suitability for use in a novel oral insulin formulation.

TABLE OF CONTENTS1. Introduction.................................................................................................................................................................................................................... 1 1.1 Diabetes mellitus, therapy & the patient experience................................................................................................................................... 1 1.2 Challenges associated with oral delivery of insulin and formulation considerations............................................................................. 2 1.3 Nanoparticle formulations: Considerations for and advantages as oral drug delivery systems............................................................ 2 1.4 Aims and objectives......................................................................................................................................................................................... 32. Materials & Methods................................................................................................................................................................................................... 4 2.1 Materials............................................................................................................................................................................................................ 4 2.2 Synthesis and characterisation of nanoparticles........................................................................................................................................... 4 2.2.1 Ultrasonication and preparation of stock solutions................................................................................................................ 4 2.2.2 Preparation of HA/CS unloaded nanoparticles....................................................................................................................... 4 2.2.3 Preparation of HA/CS nanoparticles loaded with insulin...................................................................................................... 4 2.2.4 Measurement of particle size and zeta potential of nanoparticles......................................................................................... 4 2.2.5 Validation of nanoparticle purification using size-exclusion chromatography................................................................... 4 2.3 Determination of insulin content and release studies................................................................................................................................. 4 2.3.1 Preparation of release media...................................................................................................................................................... 4 2.3.2 Determination of insulin content by high-performance liquid chromatography (HPLC)............................................... 4 2.3.3 Centrifugation.............................................................................................................................................................................. 5 2.3.4 Centrifugal Filtration.................................................................................................................................................................. 5 2.3.5 Dialysis.......................................................................................................................................................................................... 5 2.4 Statistical analyses............................................................................................................................................................................................ 53. Results & Discussion.................................................................................................................................................................................................... 5 3.1 Physicochemical characterisation of unloaded hyaluronic acid/chitosan nanoparticles....................................................................... 5 3.2 Validation of a nanoparticle purification method using size-exclusion chromatography (SEC).......................................................... 7 3.4 Direct release study I: Validation of centrifugation...................................................................................................................................... 8 3.4 Direct release study II: Validation of centrifugal filtration......................................................................................................................... 8 3.5 Indirect release study I: Validation of dialysis............................................................................................................................................. 8 3.6 Indirect release study II: Dynamic dialysis to assess the release profile of HA257/CS nanoparticles in SGF...................................... 9 3.7 Indirect release study III: Dynamic dialysis to assess the release profile of HA257/CS nanoparticles in SIF...................................... 114. Conclusion........................................................................................................................................................................................................................ 12

Mena Eskander In-vitro release studies of insulin-loaded nanoparticles in simulated gastrointestinal fluids

of history, physicians around the world slowly uncovered the nature of and causes of the disease.

By the middle of the 19th century, it was established that hyperglycaemia caused much of the disease’s pathology and that glucosuria was a good indicator and diagnostic criterion for diabetes. Most treatments for diabetes in this era consisted of starvation diets which often proved fatal for patients with type 1 diabetes. Near the turn of the century, it was suggested that an antidiabetic substance that reduced blood glucose levels was being secreted by the pancreas. This secretion was not being made in the course of its exocrine digestive functions but rather direct to the bloodstream from the small islets of endocrine tissue present in the gland.

In the early 1920s, using pancreatic extracts obtained from dogs, the first patients began to be treated using insulin. From the relatively crude and impure extracts from dogs and other animals in the early twentieth century to the present day with the advent of recombinant protein production and biosynthetic analogues, insulin therapy has evolved greatly. [1] However, from a drug delivery perspective, though a great number of advancements have been made over the past few decades in terms of both formulation and injection device design, very few alternative dosage forms have been successfully brought to market. Both clinically and commercially, there is demand for more patient-acceptable alternatives to injectable formulations. Patients with type 1, certain other rare types and type 2 diabetes resistant to oral hypoglycaemics must administer insulin regularly to maintain a fasting serum glucose level of 4-7 mmol/L to minimise the pathological effects of hyperglycaemia.[2]

The international diabetes federation estimates that over 580 million people are currently living with diabetes. [3] As found in a recent epidemiological study in Britain, concurrent with the increasing prevalence of diabetes is a corresponding increase in the number of patients undergoing insulin therapy. These findings are not unique with the American Center for Disease Control and Prevention reporting an increasing trend in insulin use with over 30% of diabetics using insulin to manage their condition in 2011.[4, 5]

Almost all of these patients use formulations of insulin which are invariably, at present at least, delivered through invasive injections. Fear of needles, injection-related anxiety and inconvenience of repeated daily injection leads many patients and prescribers to resist initiation of insulin despite it being clinically appropriate. [6] Though most insulin users would prefer an oral formulation of insulin, they place higher value on improved glucose control and reduced adverse effects such as hypoglycaemia. [7]

Therefore, the focus should not only be to deliver insulin orally but to also ensure that such a formulation is at least as safe and efficacious as currently available parenteral formulations. As the dosing of insulin is specific to each patient with little margin for error, the formulation must have a consistent and predictable pharmacokinetic profile.

Furthermore, though some may dismiss such discussion as premature, it is important to ensure that improving the patient experience remains the ultimate and primary objective of any new formulation. Indeed, that is the greatest impetus driving the immense volume of past and ongoing research and losing sight of this is why, some may argue, previous alternative

insulin formulations such as Exubera failed commercially. Throughout most of the world, patients undergoing insulin therapy now have the option of insulin injector pens rather than the traditional vial and syringe. A significant proportion of patients both prefer and persist longer using pen devices due to a number of factors that must also be taken in to account in development of any novel oral formulation. [8, 9] Unlike a vial and syringe, these devices offer patients a discreet, easy and convenient way to use their medication at a cost that is not prohibitively more expensive than traditional formulations. In addition, the needles of these injector devices are sharp, polished and are acceptably small in size to reduce injection pain and anxiety. Though Exubera was efficacious with a technically ideal inhaler device that successfully delivered a sufficiently fine fraction of particles in to the deep lungs for reliable absorption, it was not a user-friendly medicine. Notwithstanding the relative high cost, the inhaler device was large, difficult to use and inconveniently tedious for delivering large doses of insulin. It was neither discreet, more convenient nor easier to use than injectable formulations and therefore, had no place in the market. [10] This must serve as a lesson to help focus development of all future insulin formulations to ensure their success. For oral formulations in particular, it is important to think ahead of how such a formulation will be presented to the patient and a number of factors must be considered. Fundamentally, this includes the form and strength of a preparation. Whether it is to be a solid or liquid, it is important to consider all aspects of the dose assembly process. Considerations include convenience for the patient, the risk of dispensing or administration errors in case of multiple strengths and safety of administration bearing in mind potential co-morbidities such as glaucoma which may preclude independent use of the medicine.

1.2 Challenges associated with oral delivery of insulin and formulation considerations

Insulin, being a somewhat complex polypeptide consisting of 51 amino acid residues, has thus far proven challenging to successfully formulate in non-injectable dosage forms despite interest and attempts dating back to 1925. [11] Injection-related factors such as anxiety, embarrassment and pain have been found to be associated with and a potential cause for intentional non-compliance with insulin therapy. [12] Notwithstanding the notable exceptions of the now-withdrawn inhalable formulation Exubera (Pfizer) and recently approved inhalable formulation AFREZZA (MannKind) there have been few alternatives to parenteral delivery of insulin that have made it to market despite a great number of attempts having been made. Inhaled insulin, though more convenient than injectable formulations for patients poses its own problems such as potentially poor inhaler technique and a number of contraindications involving pulmonary conditions such as COPD. [13, 14] Advantages of oral insulin delivery extends past patient convenience as oral delivery of insulin more closely mimics endogenous distribution and hence is more physiological. Peripheral hyperinsulinaemia and its associated adverse effects are precluded with insulin being directly absorbed in to the portal vein. This also allows for rapid hepatic insulinisation which will correspondingly reduce hepatic gluconeogenesis.[15, 16] Commercially, due to the fact that sterility requirements for oral formulations are much less stringent than injectable formulations, it may be possible to significantly reduce the cost of insulin therapy in the future.

2


With regard to absorption, the vast majority of the challenges associated with oral delivery of insulin are not unique and are in fact shared by most peptide drugs. Insulin is of relatively high molecular weight, hydrophilic and is

susceptible to degradation through an array of metabolic pathways. Figure 1 depicts a molecular surface model of insulin monomer highlighting the high number of hydrophilic regions on its surface. This hydrophilicity and the protein’s large size make it difficult for insulin to pass through lipid-rich cell membranes. Indeed, like most proteins, physiological secretion of insulin occurs through active exocytosis. [17, 18] The gastrointestinal system as an environment is by design hostile towards polypeptides with a varied milieu of chemical and enzymatic digestive processes that, though necessary for nutrition, are challenging to contend with in drug formulation.[19, 20] Proteolysis begins in the stomach with secretion of gastric acid. This reduces pH to the point that insulin may become denatured and activates pepsin which accounts for 10-20% of total protein degradation. Insulin and its degradation products are then further metabolised by proteases released by the pancreas. These are α-chymotrypsin, elastase, the serine endopeptidase trypsin and two exopeptidase enzymes carboxypeptidases A and B. [21] As such, an oral formulation must shield insulin and ensure it remains intact as it passes through these conditions. Even if insulin successfully reaches the small intestine undamaged, thereafter comes the challenge of absorption. This is difficult due to the fact that with the exception of neonates who can absorb intact proteins (such as immunoglobulins) for a very short period of time post-partum, the small intestine cannot easily absorb intact polypeptides greater than 3 residues in length. [22,- 24] In summary, the ideal oral formulation must shield insulin as it traverses the severe conditions of the stomach, release its payload in a controlled manner in the small intestine and be designed in such a way so as to improve absorption of insulin and provide a consistent and relatively high bioavailability suitable for reliable and effective treatment. [25]

1.3 Nanoparticle formulations: Considerations for and advantages as oral drug delivery systemsThere are a multitude of novel delivery systems in development for insulin in various forms such as tablets, capsules, intestinal patches, hydrogels, microparticles and nanoparticles (NPs). A number of oral insulin formulations are currently undergoing clinical trials. An extensive range of various nanoparticle systems have been investigated previously as insulin delivery vehicles.

Many nanoparticles used in drug delivery consist of one or more polymeric substances and form solid, colloidal particles ranging in diameter from 10 to 1000 nm. Their structure and characteristics depends highly on both the precise composition

and preparation method used. Drugs can be present either protected within a cavity surrounded by a polymeric shell (nanocapsules), throughout the solid particle matrix (nanosphere) or attached to the surface of the nanoparticle.[26] A number of polymeric systems comprised of poly(lactic-co-glycolic acid) PLGA [27], chitosan complexes [28], poly(ε-caprolactone) [29], poly(isobutylcyanoacrylate) [30] amongst many others have been investigated and many have shown some promise in oral insulin delivery. Thus far, none have successfully made it to the market though some are currently in the early stages of clinical trials. For example, Nodlin is an oral formulation of insulin in bioadhesive nanoparticles by Biolaxy, Shanghai but does not appear to have progressed beyond phase I trials. [31] CobOral developed by Access Pharmaceuticals, USA encapsulates insulin in nanoparticles functionalised with a vitamin B12 (cobalamin) analogue which binds with intrinsic factor and allows for receptor-mediated uptake in the intestine. This technology does not however seem to have progressed past Phase I trials. [32]

Nanoparticles produced using natural polymers such as alginates, pectin, gelatine, chitosan are of special interest in medical applications for their non-toxicity, biocompatibility, biodegradability and relatively low cost. [33] In this work, nanoparticles comprised of chitosan and hyaluronic acid, two naturally occurring polymers will be prepared using polyelectrolyte complexation and subsequently analysed in terms of their physicochemical properties and suitability as an oral drug delivery system for insulin with regards to their relese characteristics.

Chitosan is a linear biopolysaccharide composed of randomly distributed β-(1,4)-linked D-glucosamine and N-acetyl-D-Glucosamine and is produced by treating chitin with alkali sodium hydroxide in a partial deacetylation reaction. Chitin is found naturally in the exoskeleton of crustaceans and cell wall of some fungi and yeasts and has an excellent safety profile being “generally regarded as safe” as a food additive by the United States’ regulatory body. Its amino group has a pKa of 6.5 and is protonated in acidic conditions thus making chitosan a pH-sensitive polycationic substance. This is ideal for oral drug delivery as the charge density of chitosan will differ dramatically between the stomach and intestines and this may present an avenue for controlled disintegration of a nanocomplex carrier and subsequent drug release. In addition, this polycationic property also makes chitosan a mucoadhesive substance which may prolong the residence time of the formulation in the gastrointestinal tract. [34] Hyaluronic acid is a polyanionic biopolymer (specifically a non-sulfated glycosaminoglycan) comprised of D-glucuronic acid and D-N-acetylglucosamine linked together via alternating β-1,3 and β-1,4 glycosidic bonds. It is naturally occurring and

Figure 2 Chemical structure of hyaluronic acid

3

Figure 1: Solvent accessible surface of insulin monomer. Blue regions are hydrophilic. Red regions are hydrophobic.

Figure 2 Chemical structure of chitosan


indeed abundant in humans, being found in neural, connective and epithelial tissue. [35] It plays an important role in synovial fluid as a lubricant and in wound healing and is found extensively in the extra cellular matrix. [36] As such, it too has an excellent safety profile in humans and for this reason along with its readily modified chemical structure it has been studied extensively in a number of biomedical applications including drug delivery. [37] The pKa of hyaluronic acid’s carboxylic acid groups are approximately 3. [38]

Ideally, disintegration and release of the drug payload (and other excipients such as absorption enhancers) should only occur under certain conditions. Insulin encapsulated in nanoparticles may be absorbed through the small intestine primarily by three different mechanisms; i) Lymphatic uptake and translocation through the M cells of Peyer’s patches in the ileum ii) Transcytosis via epithelial cells of the intestinal mucosa and/or iii) Paracellular uptake. [39] Despite the fact that chitosan is known to be a potent absorption enhancer through its widening of epithelial tight junctions, its relative insolubility outside of acidic media may hamper this effect in the small intestine. In a future formulation, addition of N-trimethyl-chitosan may be considered as it is soluble under neutral/basic conditions yet still possesses the same absorption enhancing effects of chitosan. [34]

1.4 Aims and objectivesThe primary aim of this work is to develop a method for and to investigate the release of insulin from novel pre-developed polyelectrolyte nanoparticles comprised of hyaluronic acid and chitosan produced using two different mass mixing ratios. It is envisaged that this work will help assess the suitability of this nanoparticle carrier system for the oral delivery of peptide drugs such as insulin. As previously described, this requires the carrier system to release insulin only upon reaching the small intestine of the gastrointestinal tract.

The physicochemical characteristics of the nanoparticle system and the effects of insulin-loading on these properties must be determined. In addition, a suitable method for studying the release of insulin from this delivery system must also be developed and subsequently used for profiling the release of insulin from this carrier system in simulated conditions of the stomach and small intestine.

In summary, this work aims to:

1. Produce and characterise both unloaded and peptide-loaded polyelectrolyte nanoparticles comprised of differing ratios of hyaluronic acid and chitosan

2. Develop a suitable method for the purification of these nanoparticles

3. Develop a suitable in-vitro method for assessment of the release profile of insulin-loaded nanoparticles

2. MaterIals & Methods

2.1 Materials Human insulin (containing zinc) was kindly provided by Sanofi. Hyaluronic acid sodium salt from Streptococcus equi was obtained from Sigma (USA). Chitosan ultrapure chloride salt UP CL113 (Protasan) was obtained from Novamatrix (Norway). In all cases where water was used (including rinsing of glassware), unless otherwise specified, the grade of water was type 1 (Milli-Q water) provided on-site by a Millipore Synergy UV water purification system. All other reagents, unless otherwise specified, were of analytical grade.

Synthesis and characterisation of nanoparticles2.2.1 Ultrasonication and preparation of stock solutionsAqueous solutions of 0.1% w/v hyaluronic acid (HA) and chitosan (CS) were prepared and stirred overnight using a magnetic stirrer to ensure complete dissolution. The HA solution was then ultrasonicated using a 130W ultrasonic processor (SONICS VC130PB, Sonics and Materials Inc., USA). The processor was equipped with a probe of diameter 3mm and the process was carried out for 2 hours at an amplitude of 80 corresponding to a nominal power output of 13W. This produced a hyaluronic acid of approximately 257 kDa molecular weight and will henceforth be referred to as HA or HA257. The purpose of this was to reduce the risk of nanoparticle flocculation associated with use of hyaluronic acid of higher average molecular weight. [44] The CS solution was adjusted to pH 3 using dropwise addition of HCl and a pH meter (Orion - Model 520A equipped with an Orion RossTM 8103SC glass body pH semi-micro electrode).

2.2.2 Preparation of HA/CS unloaded nanoparticlesAqueous solutions of HA and CS were prepared as in section 2.2.1. An aliquot of CS solution of suitable volume was rapidly added to a known volume of HA solution and remained under magnetic stirring for 10 min to stabilise the dispersion. In all cases, this immediately produced a turbid dispersion of total polymer concentration 0.1% w/v. The following mass mixing ratios were used (HA257/CS); 5:1 and 1.25:1 and these will be henceforth referred to as MMR 5 and MMR 1.25 respectively.

2.2.3 Preparation of HA/CS nanoparticles loaded with insulinPreparation of insulin-loaded nanoparticles was largely similar to preparation of unloaded nanoparticles. Human insulin was weighed on a microbalance (Mettler MT5) and dissolved in CS solution. The mass of insulin was calculated to be sufficient to ultimately produce a nanoparticle dispersion containing 100 μg/mL insulin.

2.2.4 Measurement of particle size and zeta potential of nanoparticlesThe intensity-averaged mean particle size (mean particle size) and the polydispersity index (PDI) of the produced nanoparticles were measured using dynamic light scattering (DLS) using 173o backscatter detection. Zeta potential of the nanoparticles was measured by converting the electrophoretic mobility values obtained by Laser Doppler Velocimetry (LDV) using the Smoluchowski equation. The aforementioned measurements were performed using a Zetasizer Nano-ZS ZEN3600 equipped with a 633 nm laser (Malvern Instruments Ltd., UK). Samples were analysed, undiluted, in folded capillary tubes at 25oC with an equilibration time of 1 minute. All measurements were

4

Figure 3 Chemical structure of hyaluronic acid


performed in triplicate. Before sampling, the nanoparticle dispersion being tested was agitated to ensure homogeneity. Where possible, samples were taken from random regions of the vessel.

2.2.5 Validation of nanoparticle purification using size-exclusion chromatographySamples of both negatively charged and positively charged unloaded nanoparticles were fractionated using a size-exclusion chromatography column (Centri-Pure PF100, emp biotech, Germany). The column was first equilibrated using type 1 ultrapure (UP) water. To provide a background measurement, a blank fraction was collected and then assessed using the Zetasizer Nano-ZS for mean particle size and derived count rate (corresponding to quantity of particles present in the eluate). A volume of a stirred nanoparticle dispersion was applied evenly over the top of the column. Type 1 UP water was then used as the mobile phase and the retention volume of the nanoparticles was then measured by collection of a sufficient number of 1 mL fractions, ranging from 15-20 fractions. These fractions were then assessed for mean particle size, size distribution and derived count rate and the data graphed and analysed. The same process was repeated using 10 mL of a 100 µg/mL solution of human insulin in 0.01M HCl.

2.3 Determination of insulin content and release studies2.3.1 Preparation of release mediaTwo media were prepared to represent the physicochemical conditions of the stomach and intestines. Both were prepared to a similar specification as to that set out in the British Pharmacopoeia for simulated gastric fluid (SGF) and simulated intestinal fluid (SIF); respectively pH 1.2 and pH 6.8. Methylcellulose at a concentration of 0.001% (w/v) was also included in an attempt to reduce the level of insulin adsorption on to surfaces. [40] These were prepared without enzymes, hence pepsin was not included in SGF and pancreatic enzymes were not included in SIF.

2.3.2 Determination of insulin content by high-performance liquid chromatography (HPLC)For the determination of insulin content, the following apparatus and method was used. The process employed a Thermo BDS Hypersil Gold C18 5µm (250 x 4.6 mm) column equipped to a Waters HPLC system with a Waters 1525 Binary HPLC pump, Waters 717plus Autosampler operating with a Waters 2487 Dual λ Absorbance Detector. The wavelength (λ) used for detection was 214 nm. 0.01M HCl was used as needle wash. The composition of the mobile phase was 50mM KH2PO4 buffer adjusted to pH 2.3 and acetonitrile in a ratio of 70:30 (v/v). For each analysis, an injection volume of 25 µl was used with a flow rate of 1 mL/min.

For quantification purposes, prior to determination of samples, a number of insulin standards of decreasing concentration were prepared and analysed using the system to allow for graphing of a calibration curve. The insulin standards were prepared in 0.01M HCl and were of the following concentrations; 1, 2, 3, 4, 5, 8, 10, 20, 30, 40, 50, 100 µg/mL.

2.3.3 Centrifugation Two samples of the dispersion to be tested were withdrawn immediately from the vessel after preparation and were

analysed for actual insulin content using the HPLC method previously described. A number of aliquots of the dispersion being tested were dispensed to 1.5 mL Eppendorf tubes and centrifuged for; 0 min, 30 mins, 60 mins. At all timepoints, measurements of mean particle size, polydispersity, derived count rate and insulin content were made using previously described methods. Centrifugation was performed for the specified periods of time at 18,000g using a Thermo Scientific™ MicroCL 21/21R centrifuge. For each sampling procedure, 400 µl was drawn from just below the upper surface of the liquid in each tube taking care not to agitate the tube.

Samples to be assessed for insulin content were analysed as per the method described in section 2.3.2. Samples to be assessed for mean particle size and count rate were transferred to a clear disposable cuvette (Malvern, UK) and analysed using the Zetasizer Nano ZS taking care to use the same laser attenuator setting for all measurements.

2.3.4 Centrifugal Filtration The recovery of insulin in the filtrate was first measured to assess the suitability of the method for performance of release studies. The unit used was an Amicon Ultra-15 Centrifugal Filter Unit (50 kDa MWCO, Merck Millipore, Ireland).

A sample of approximately 100 µg/mL insulin solution in 0.01M HCl was analysed by HPLC for determination of insulin content. An aliquot of the aforementioned insulin solution was then loaded on the filter unit and diluted 1:9 (v/v) with enzyme-free SGF. This mixture was placed in a shaking water bath at 37oC for 1 hour. The filter unit was then removed and centrifuged at 3000g for 30 minutes. The filtrate produced was collected in full and its volume measured. An equivalent volume of SGF was then added to the filter unit which was then shaken and replaced in to the shaking water bath at 37oC with further sampling as per the aforementioned process at 21 hours. Samples were analysed immediately after collection using the HPLC method described in section 2.3.2 for the determination of insulin content.

2.3.5 Dialysis The following procedure was performed using the following two models of dialysis tubes: Float-a-Lyzer G2 (Spectrum Laboratories, USA) of pore sizes 20 kDA and 100 kDA, respectively 1 mL and 5 mL in volume. An experiment using each was performed in duplicate using two dialysis tubes. As per Figure 17 (Page 10), A Haake F3-C circulating water bath set at 37oC circulated heated water to a jacketed vessel in which a clean centrifuge tube was placed. This was filled with a sufficient volume of SGF or SIF to allow for a dilution ratio of 1:9 (v/v) test mixture to release media. A stir bar was placed at the bottom of the tube and the jacketed vessel was placed on a magnetic stirrer and set to 950 rpm. The dialysis tube was then equilibrated using the release media. Thereafter, the internal chamber of the dialysis tube was drained and replaced with an appropriate volume of the mixture being tested. The entire jacketed vessel was then sealed and the following samples were taken.Method validationUsing 100 µg/mL insulin solution in 0.01M HCl, samples of the external compartment were taken at 1, 2 and 21 hours and were assessed for insulin content. Samples of the internal

5


compartment were taken at 0 and 21 hours and were assessed for mean particle size, zeta potential and insulin content. Release studiesUsing a loaded nanoparticle dispersion of 0.1% total polymer concentration and 100 µg/mL insulin, samples of the external compartment were taken at 1, 2, 4, 6 and 24 hours and were assessed for insulin content. Samples of the internal compartment were taken at 0 and 24 hours and were assessed for mean particle size, zeta potential and insulin content.2.4 Statistical analysesStatistical analyses were performed using a combination of Minitab and Microsoft Excel. Where averaged data are presented in tables and figures, this is expressed as mean +/- standard deviation.

3. results & dIscussIon

3.1 Physicochemical characterisation of unloaded hyaluronic acid/chitosan nanoparticlesThree batches of nanoparticles were prepared in the mass ratio of 5:1 and 1.25:1 (HA257/CS). The former MMR of 5 produced particles of negative surface charge (negatively charged nanoparticles) whilst the latter MMR of 1.25 produced particles of positive surface charge (positively charged nanoparticles). The mean particle size (expressed as the Z-Average or intensity-based harmonic mean particle size), polydispersity index and the zeta potential of these particles were measured. After stirring to ensure a homogeneous dispersion, three samples were taken from random areas of the vessels used for preparation of the nanoparticles. All apparatus used were rinsed with type 1 UP water to reduce interference from foreign particulate matter. Measurements obtained were adjusted for viscosity (measured at 25oC) using 1.21 mPa.s and 0.947 mPa.s respectively for particles produced using MMRs of 5:1 and 1.25:1. [41] The adjusted data are presented below in Table 1. The size distribution by intensity and frequency undersize particles is presented in figures 4 and 5.

Mean particle size and zeta potential of unloaded HA257/CS nanoparticles

3 samples were analysed with each measurement being performed in triplicate. The averaged results are presented in

this table. Z-Ave is the intensity-based harmonic mean. PDI is the polydispersity index. ZP is the zeta potential measured.

Mass mixing ratio (HA257/CS)

Z-Ave (d.nm) PDI ZP (mV)

5:1 153 ± 10 0.17 ± 0.04 -47.4 ± 2.1

1.25:1 222 ± 15 0.11 ± 0.04 32.9 ± 1.9Table 1: Mean particle size and zeta potential of unloaded HA257/CS nanoparticles

Figure 4 Intensity-averaged size distribution of HA257/CS 5:1 unloaded nanoparticles

Figure 5 Intensity-averaged size distribution of HA257/CS 1.25:1 unloaded nanoparticles

Particles prepared with an MMR of 5 were of negative charge, approximately -47.4 mV and tended to be of smaller size averaging 153 nm. Particles prepared with an MMR of 1.25 however were of positive charge, approximately +32.9 mV and were generally of larger size. It was noted that the mean particle size of particles produced using the 1.25 MMR increased significantly if the rate at which one polymer solution was introduced to the other was slow. As such it would appear the speed at which both solutions are introduced to one another appears to influence particle size with slower addition generally resulting in larger particles. As such, all future batches were prepared by rapidly decanting the second polymer solution into the other with rapid magnetic stirring throughout. The mixing order, speed, stirring speed, beaker type and position on the magnetic stirring apparatus were kept constant from this point to ensure comparable results. Particles produced using an MMR of 1.25 were generally of larger size. It has been shown in numerous past studies that nanoparticles produced by polyelectrolyte complexation increase in size with increasing amounts of one polymer. [42, 43] As has been shown in previous studies by Umerska et al., there is an association between mass mixing ratio (MMR) and zeta potential with the magnitude decreasing to negligible values at an MMR of 1 due to charge neutralisation. [44] There is a clear difference in surface charge between the two mass mixing ratios with increasing amounts of polycation leading to more positive surface charge. Regardless, the zeta potential values of both species of nanoparticle were of a sufficiently high magnitude so as to produce a stable colloidal dispersion. Two dispersions of nanoparticles loaded with approximately 100 μg/mL insulin were prepared using each MMR. The mean particle size, size distribution and zeta potential of these particles are presented in table 2 and figures 6 and 7. Measurements obtained were adjusted for viscosity (measured at 25oC) using 1.48 mPa.s and 1.0167 mPa.s respectively for particles produced using MMRs of 5:1 and 1.25:1. [41]

6


Mean particle size of insulin-loaded HA257/CS nanoparticles

2 samples were analysed with the averaged results are presented in this table. Z-Ave is the intensity-based harmonic mean. PDI is

the polydispersity index. ZP is the zeta potential measured.Mass mixing

ratio (HA257/CS)Z-Ave

(d.nm) PDI ZP (mV)

5:1 142 ± 7 0.20 ± 0.05 -52.4 ± 1.6

1.25:1 307 ± 10 0.21 ± 0.01 44.6 ± 1.8

Figure 6 Intensity-averaged size distribution of HA257/CS 5:1 insulin-loaded nanoparticles

Figure 7 Intensity-averaged size distribution of HA257/CS 1.25:1 insulin-loaded nanoparticles

Particles prepared with an MMR of 1.25 show a clear difference in mean particle size, polydispersity and zeta potential values between the unloaded and insulin-loaded particles. Particles prepared with an MMR of 5 however only differ in polydispersity and zeta potential values. There is an increase in polydispersity index and magnitude of surface charge in both classes of particles with the greater increase being observed in the positively-charged particles. Noting that the isoelectric point of human insulin is 5.3 and the pH of the dispersions of negatively and positively charged nanoparticles (NPs) are approximately 5.8 and 4.4 respectively, this may possibly be due to the fact any insulin on the surface of the nanoparticle is likely to be charged and to a greater extent in the case of the positively charged NPs. [27, 45] As can be seen in figures 6 and 7, this high magnitude of surface charge is sufficient to discourage aggregation and leads to the formation of a stable dispersion of discrete NPs. [26] The presence of insulin also appears to lead to changes in mean particle diameter with a significant difference only being noted in the positively charged species of particles (t-test, p value of <0.001). Curiously, while the negatively charged particles appeared to very slightly decrease in size (not statistically significant with a p value of 0.19), the positively

charged particles greatly increased in size.

3.2 Validation of a nanoparticle purification method using size-exclusion chromatography (SEC)

To assess if size-exclusion chromatography may be suitable for separation of HA257/CS NPs from free human insulin, dispersions each of negative and positively charged nanoparticles were ran through a Centri-Pure PF100 SEC column (emp Biotech GmbH, Germany) and 1 mL fractions were collected and assessed for particle size and derived count rate using the method described in section 2.2.4. The results were then graphed and compared with data obtained after performing the same procedure using 0.1% (w/v) human insulin in 0.01M HCl.

Figure 10 Size-exclusion chromatography of HA257/CS NPs and human insulin

As the data presented in figure 10 shows, the retention volume of both negative and positively charged NPs differ from the retention of free insulin using this system. This allows for reliable purification of NPs from a mixture including free insulin. It is however rather slow and has the disadvantage of diluting the nanoparticles significantly. In this particular instance, even assuming 100% recovery, the nanoparticles were diluted by a factor of at least 1.6 as a sample volume of 10 mL was applied and the bulk of NPs recovered within a retention volume of 16 mL.

3.3 Direct release study I: Validation of centrifugation

Centrifugation is often used in direct release studies to separate nanoparticles from released insulin. The content of free insulin in the supernatant is then measured. To validate if centrifugation may be appropriate for use to study the release of insulin from the prepared NPs, dispersions of both positively and negatively charged NPs with a theoretical insulin concentration of 100 μg/mL dispersion were analysed using the method described in section 2.4.3. The data are presented in tables 3-4 and figures 11-12.

Figure 9 TEM of unloaded 5:1 HA257/CS NPs

Figure 8 TEM of insulin-loaded 5:1 HA257/CS NPs

Table 2 Mean particle size and zeta potential of insulin-loaded HA257/CS nanoparticles

7


Supernatant characterisation – 5:1 HA257/CS nanoparticles

The averaged results are presented in this table. Insulin content is the concentration of insulin in μg/mL calculated with respect to

the sampling volume.

Centrifugation time Sampling method

Insulin content (µg/

mL)

Control Random sample from synthesis vessel after agitation 103.33

0 minutes Sampling of upper 400 µL of supernatant without agitation 126.06



Table 3 Effect of centrifugation on insulin content of 5:1 HA257/CS nanoparticle dispersion

Supernatant characterisation – 1.25:1 HA257/CS nanoparticles

The averaged results are presented in this table. Insulin content is the concentration of insulin in μg/mL calculated with respect to

the sampling volume.

Centrifugation time Sampling method

Insulin content (µg/

mL)

Control Random sample from synthesis vessel after agitation 116.03




Table 4 Effect of centrifugation on insulin content of 1.25:1 HA257/CS nanoparticle dispersion

Figure 11 Effects of centrifugation on insulin-loaded HA257/CS 1.25:1 nanoparticles. All samples were taken from the upper 400 µL of supernatant.

Figure 12 Effects of centrifugation on insulin-loaded HA257/CS 5:1 nanoparticles. All samples were taken from the upper 400 µL of supernatant.

On visual inspection, for both types of NPs, centrifugation produced a white sediment that could easily be re-suspended. Indeed, as can be seen in figures 11-12, centrifugation successfully clarified the supernatant with dramatic reductions in both derived count rate and mean particle size after only 30 minutes with further centrifugation generally leading to further decreases. The same trend was not observed for insulin content which although showed a very slight decrease, did not show a similar magnitude of decrease. This was not altogether unexpected for the positively charged nanoparticles with an MMR of 1.25 as these have been previously measured to have an association efficiency of only 5% and a peptide loading of only 0.2%. [41] This is reflected in the observations made with 400 µL of supernatant (after 60 minutes centrifugation) containing 96.1% of the insulin present when the agitated synthesis vessel was randomly sampled using the same volume (control). Negatively charged particles prepared with an MMR of 5 were previously measured to have an association efficiency of 80% and peptide loading of 9% whilst positively charged particles prepared with an MMR of 1.25 were measured to have an association efficiency of 5% and peptide loading of 0.2% as previously mentioned. [41] It may be perhaps due to this difference in peptide loading that the decrease in insulin content is somewhat larger in magnitude in the negatively charged nanoparticles with a higher peptide loading. The significance of such a small difference is questionable however. Nonetheless, centrifugation using the method described is not suitable to adequately determine the release profile of these NPs as it would appear that the loaded NPs do not remain intact during the centrifugation process or at least do not retain most (if any) of their loaded insulin.

3.4 Direct release study II: Validation of centrifugal filtration To assess the suitability of centrifugal filtration as a method to profile the release of insulin from the prepared nanoparticles, a 100 μg/mL solution of human insulin was filtered and analysed using the method described in section 2.4.4. This was to assess the permeability of the filter to insulin. The molecular weight cut off of the filter membrane used (Amicon Ultra-15 Centrifugal Filter, Merck Millipore, Ireland) was 50 kDa MWCO. Considering the molecular weight of insulin monomer is slightly over 5,800 Da it would be expected that dissolved insulin monomer and indeed multimers up to the hexamer should pass through the filter membrane. [45] Insulin self-aggregation and oligomerisation to dimers, hexamers and potentially higher-order oligomers is dependent on pH, ionic strength and the concentration of Zn2+ ions in solution. [46] The dodecameric form is approximately

8


70 kDa in size. Though it cannot be completely discounted, stable and lasting presence of this multimer is unlikely to be significant without the further addition of zinc. The possibility of insulin adsorption on to either the walls of the filter unit or on the filter membrane itself cannot be discounted.

The data obtained are presented in figure 13 and show extremely low insulin recovery (9.63% at 21 hours). This would suggest that there is significant adsorption of insulin and/or another factor as previously discussed interfering with permeation of insulin through the filter.

Figure 13 Insulin recovery using centrifugal filtration with a 50 kDa MWCO centrifugal filter

3.5 Indirect release study I: Validation of dialysisTo assess the suitability of the method for assessing the release profile of the prepared nanoparticles, the recovery of insulin across the dialysis membrane was first determined. As described in section 2.4.5, Float-a-lyzer G2 dialysis tubes of two different molecular weight cut offs (MWCO) were used; 20 kDa and 100 kDa. Respectively, the internal compartment of these tubes were 1 mL and 5 mL in volume.

A 100 µg/mL solution of insulin in 0.01M HCl was placed in to a 20kDa dialyser and was suspended with agitation in a sealed tube containing a sufficient volume of SGF to provide sink conditions. After each sample was taken, the external media was replenished with a volume of fresh, 37oC SGF equivalent to the sampling volume. Measurements were corrected for loss of insulin during sampling using the following formula

For any sample n, Ccor is the corrected concentration, Cn is the uncorrected concentration as analytically determined, Rv is the volume of release media and Sv is the sampling volume where it is kept constant.

Figure 14 Validation of dialysis: Insulin recovery using a dialysis membrane with a 20 kDa MWCOAttempts to use dialysis tubes with a MWCO of 20kDa

produced inconsistent data (figure 14). It is likely that these erratic results came about due to a combination of possible damage to the fragile dialysis membrane during set up of the testing apparatus and an unsuitably low MWCO. Considering the low pH (approximately 2), low protein concentration and presence of mineral acid (HCl), the most probable and perhaps most abundant multimer expected is the dimer though higher order multimers and aggregates may also be present. [47, 48] Although the dimeric form of insulin (approximately 10.6kDa) is still below the MWCO of the pores of the dialysis tube (20kDa), higher-order multimers such as the hexamer and dodecamer are well above the MWCO point. In any case, it has been shown previously that using too low of a molecular weight cut off can introduce significant error in to determining the rate of release from formulations. [49] In addition to multimerisation, it is likely that insulin adsorption may be altering important characteristics of the dialysis membrane such as pore size and surface charge. [50] Therefore, a further attempt at validating dialysis as a release study method was made using a similar dialysis tube of a higher MWCO (100kDa) as this was deemed large enough to allow for permeation of insulin multimers without necessarily all owing for the nanoparticle carriers to pass through. Using a dialysis tube of higher MWCO, insulin recovery was much higher with one run exceeding 90% recovery (figure 15, table 5).

Figure 15 Validation of dialysis: Insulin recovery using a dialysis membrane with a 100 kDa MWCO

Insulin content (Run 1)

Time Donor compartment

Acceptor compartment

0 653.42 μg -1 - 50.83 μg2 - 178.98 μg

21 49.41 μg 603.63 μgInsulin content (Run 2)

Time Donor compartment

Acceptor compartment

0 639.51 μg -1 - 5.45 μg2 - 11.60 μg

21 127.46 μg 287.56 μgTable 5 Validation of dialysis: Insulin recovery using a dialysis membrane with a 100kDa MWCO. All data are expressed as content in µgHowever, as in figure 16, due to inconsistencies in stirring pattern and hence agitation of the two tubes used, there is significant variation in the data. This was corrected in subsequent experiments by fabrication of new external vessels for the dialyser tubes that permitted use of individual jacketed vessels and precise control of agitation in each (figure 16). Insulin recovery in this particular experiment was 92.38% and

9


44.48% respectively for run 1 and 2. Approximately 99.4% and 64.9% of insulin introduced to the apparatus was accounted for in run 1 and run 2 respectively.

Figure 17 Schematic depicting the optimised test apparatus

3.7 Indirect release study II: Dynamic dialysis to assess the release profile of HA257/CS nanoparticles in simulated gastric fluid

The release profile of both positively charged (MMR 1.25) and negatively charged (MMR 5) NPs was assessed in SGF without enzymes. Dispersions of each were prepared with a theoretical insulin loading of 100 µg/mL dispersion and characterised for particle size and zeta potential using a similar method to that described in section 2.2.4. Actual insulin content of these dispersions was also determined using the chromatographic method specified in section 2.3.2.

4.4 mL of each dispersion was placed in a 5 mL Float-a-Lyzer G2 dialysis tube with a 100 kDa MWCO. This tube was suspended in a vessel containing 40 mL SGF at 37oC with agitation being provided by a magnetic stirring device set to 950 rpm. Care was taken to ensure the vessel was placed in such a manner that any replicates were subject to the same level of agitation. Samples were drawn from the acceptor compartment at 1, 2, 4, 6 and 24 hours with measurements of insulin content being corrected for loss of insulin and media replenishment as previously described. A sample was drawn from the donor compartment at 24 hours to measure the residual insulin content and mean particle size and zeta potential of any nanoparticles remaining. Each experiment was performed in duplicate and the averaged results presented.

Figure 18 Overlaid release profiles of 100 μg/mL insulin solution, 1.25:1 and 5:1 (HA257/CS) nanoparticles in SGF. Error bars are omitted from the insulin solution due to the high variability in these datapoints. Due to the extremely low association efficiency (5%) and peptide loading (0.2%) of the positively charged nanoparticles, the release profile was expected to be similar to free insulin solution as in section 3.5. Curiously, the concentration of insulin within the acceptor compartment from the positively charged nanoparticles rose more rapidly than insulin solution. This implies that the presence of these nanoparticles and their constituent substances increases the rate at which insulin permeates across the dialysis membrane. It may be the case that components of the nanoparticle system are altering characteristics of the dialysis membrane such as pore size thus decreasing resistance to insulin permeation. Alternatively, due to the fact that the insulin used in this work contains zinc, it may be the case that free hyaluronic acid molecules are sequestering zinc. Zinc is important for the assembly and stabilisation of high-order mutlimers such as the insulin hexamer and dodecamer. [51] Sequestering agents (such as EDTA) are frequently included in rapid-acting injectable insulin formulations to maximise the concentration of the monomeric/dimeric species of insulin to ensure a rapid rate of absorption. [52] This interaction, which may be attributed to HA, CS or their complex interacting with either insulin and/or the dialysis membrane demonstrates that dynamic dialysis may only approximate the release profile of insulin from its nanoparticle carrier. This is due to the fact that the rate of insulin release from the nanoparticle matrix may not necessarily match the rate at which insulin permeates across the dialysis membrane.

The release profile of the negatively charged nanoparticle dispersion is noticeably different with an apparent steady and slow release of insulin. Though 46.02% of the insulin present in the formulation was recovered in the external acceptor compartment after 24 hours of dialysis, this is not of much clinical significance. This is due to the fact that significant clearance of the formulation from the stomach should be achieved within 4-6 hours in healthy individuals. [53, 54] At 4 hours, only 7.55% of the insulin present in the negatively charged nanoparticle dispersion had been released in to the acceptor compartment compared to 68.36% in the positively charged nanoparticle dispersion. Though it is as of yet uncertain if these negatively charged nanoparticles offer sufficient protection from other degradative factors such as enzymatic action, they do not appear to release any significant quantity of insulin at pH 1.2 which is a promising observation.

The effect of dialysis on the physicochemical characteristics of the nanoparticles was also assessed.

Figure 16 Schematic depicting the original dialysis test apparatus

10


Time Mean particle size (d.nm) PDI ZP (mV) Derived

count rate

0 248 +/- 2 0.17 +/- 0.02 40.7 +/- 0.8 89485 +/- 835

24 296 +/- 172 0.44 +/- 0.03 9.85 +/- 0.9 2124 +/- 2013

Figure 19 Dialysis of 1.25:1 (HA257/CS) nanoparticles in SGF

Time Mean particle size (d.nm) PDI ZP (mV) Derived

count rate

0 111 +/- 4 0.18 +/- 0.04 -48.0+/- 1.8 83972 +/- 6157.1

24 227 +/- 18 0.55 +/- 0.05 8.1 +/- 2.6 1566.6 +/- 887.7

Figure 20 Dialysis of 5:1 (HA257/CS) nanoparticles in SGF

As in figures 19-20, the effects of dialysis show similar trends. In both, there is a significant increase in mean particle diameter, increase in PDI and decrease in count rate. Regardless of the original surface charge, in both instances, the zeta potential decreased in magnitude to approximately +8-10 mV. This is approximately neutral leading to instability of the dispersion and flocculation of the matter present. In both species of nanoparticle, on visual inspection of the cuvette used for particle sizing, there were clearly visible flocs of material and sedimentation present at the 24 hour timepoint. Prior to dialysis, the size distributions of both nanoparticle types were monomodal with both transforming to multimodal distributions with broad, unfocused peaks extending to matter below 20 nm (these are likely to be free polymer chains) and micron-sized particles/flocs. This would suggest that these nanoparticle systems are disrupted significantly after 24 hours of exposure to simulated gastric conditions.

3.7 Indirect release study III: Dynamic dialysis to assess the release profile of HA257/CS nanoparticles in simulated intestinal fluid

Unfortunately, due to an equipment fault, only data pertaining to the insulin release profile of MMR 1.25 (positively charged) nanoparticles are available up to the 6 hour timepoint.

Using a similar methodology as that described in section 3.6, the insulin release profiles of dispersions of positive and negatively charged NPs were assessed using dynamic dialysis.

Figure 21 Overlaid release profiles of 100 μg/mL insulin solution in SGF and 1.25:1 (HA257/CS) nanoparticles in both SGF and SIF.

As previously observed in section 3.6, considering the negligible association efficiency and peptide loading of particles prepared with an MMR of 1.25, it would appear that the components of the nanoparticle matrix are indeed increasing the rate of insulin permeation. However it is also apparent the presence of these materials is not the only factor influencing permeation. The rate of permeation of insulin from the positively charged nanoparticle dispersion in SIF appears to be lower than the rate observed when the same system is dialysed in SGF. This is most likely attributed to the difference in pH between the two media. It is possible that the rate of permeation across the dialysis membrane is subject to and hence a function of pH. It is uncertain whether this is due to either alterations in the multimerisation or net charge of insulin, surface aberrations of the dialysis membrane, a combination both or another unknown factor. However, it is well known that pH has a significant effect on the multimerisation of insulin. [47, 48] Therefore, it is possible that an increase in pH, despite possible zinc chelation by hyaluronic acid, may be allowing for an increased presence of higher-order insulin multimers of higher hydrodynamic volume and therefore lower rate of permeation than that observed in SGF.

On visual inspection, there was evidence of very significant flocculation and sedimentation present in the remaining dispersion present in the donor compartment after dialysis of both nanoparticle species. Attempts to measure particle size at the 24 hour timepoint were not successful with the instrument reporting a PDI of ~1 and warning the dispersion contained particles that greatly exceeded the upper limits of the instrument. In all cases, after 24 hours, zeta potential was reduced in magnitude to somewhat negligible charge below |14|mV. Most likely due to the increased pH of the SIF media, regardless of the original zeta potential of the dispersions, nanoparticle dispersions prepared according to both MMR were of slight negative charge after 24 hours of dialysis.

11


4. ConclusionIn this work, two hyaluronic acid/chitosan nanoparticle carrier systems were successfully produced and loaded with insulin. A chromatographic purification scheme for these nanoparticles was developed and a number of direct and indirect release study test methods were assessed for suitability. Though there are many aspects of this delivery system that are as of yet untested such as their ability to protect insulin from proteolysis, the results obtained during the course of this study are promising with one particular nanoparticle system (MMR 5) showing evidence of controlled insulin release with very limited (7.55% at 4 hours) insulin release in enzyme-free SGF. Unfortunately, due to a major failure with the HPLC recording computer, the full set of release data in SIF are not available. It is hoped that these experiments will be repeated at an opportune time in the future.

Though a chromatographic purification scheme utilising a simple size-exclusion column was developed and validated, it is somewhat slow (though may be automated) and of sub-optimal chromatographic efficiency leading to significant dilution of the nanoparticles. As flow rate is difficult to control with a simple chromatographic column, use of a different chromatographic column with beads of smaller particle size may help improve efficiency of the column. This will increase back pressure however and slightly extend run times but this may not be problematic if the process is automated. [55] If this purification scheme can be optimised and if it can be proven that the process does not significantly alter the structure of and/or the characteristics of the nanoparticles, it may be useful in reducing the error inherent with performance of release studies using nanoparticle dispersions already containing free unencapsulated insulin. Another potentially suitable purification method worth considering is affinity chromatography which has in the past proven highly selective for insulin. [56] If free insulin can be isolated from loaded nanoparticles using such a technique prior to studying release, this may significantly reduce the inherent error arising from incomplete peptide loading.

In choosing a suitable method for performance of release studies, neither of the two direct methods were deemed suitable with centrifugation disrupting the nanoparticle system and centrifugal filtration having a very low insulin recovery. The possibility of repeated centrifugation possibly disrupting the nanoparticle system and/or inducing release was also considered and it was deemed that the method was not ideal.

Dynamic dialysis, as used for release studies of many nanoparticle formulations [57], was then investigated as a potential study method and membranes of two different pore sizes were assessed for their insulin permeability with the larger 100 kDa MWCO being deemed suitably high to allow for a reasonably high rate of insulin permeation even in the presence of insulin multimers. After optimisation of the testing apparatus, the method produced relatively consistent and reproducible data and appears suitable for preliminary release studies. It was also identified however that the method is not without its flaws as the rise in insulin content (interpreted as the rate of insulin permeation across the dialysis membrane) in the external acceptor compartment was clearly not just a function of time but appeared to be also dependent on the

presence of HA, CS or their complex. The pH of the external media may also play some role. Substantiation for this may be found by considering that despite the fact that positively charged nanoparticles (MMR 1.25) have a negligibly low association efficiency and are expected to behave similarly to insulin solution, there is a clear difference. Free insulin within these particular nanoparticle dispersions appears to permeate more rapidly in both SGF and SIF compared to insulin solution in SGF. This may be attributable to either physical changes to the surface of the dialysis membrane and/or changes in the multimerisation state of insulin due to zinc chelation by hyaluronic acid as previously discussed.

As such, though dynamic dialysis does appear to be a suitable method for the performance of release studies on this nanoparticle system there is certainly room for improvement. Notwithstanding these potential improvements, due to the fact that insulin must be released from the nanoparticle matrix (a barrier to release) and then must also permeate across a second “barrier” to be measurable, in the event that the data in SIF shows sustained release of insulin it would be imprudent to assume that this will also be the case in-vivo.In the future, it may be interesting to study nanoparticles produced using the same materials but alternative methods such as electrospraying. Electrospraying has in the past proved a sophisticated, facile and robust technique for the production of nanoparticles and is a process that appears to not significantly reduce the bioactivity of insulin. [58, 59] Co-axial electrospray drying, in which two different liquids are sprayed from two concentric nozzles, may be a viable technique for the production of nanocapsules. [60]

In summary, this particular nanoparticle carrier system shows promise as an oral peptide drug delivery system but a number of further investigations will be required to fully characterise the properties of this system and assess its suitability for

further development.

acknowledgeMents

Foremost, I would like to express my sincere gratitude and thanks to Dr. Lidia Tajber and Ms. Svenja Sladek without whom this project would not have been possible. It is difficult to overstate my gratitude for their immense support, patience and unshakeable enthusiasm throughout this project and I am glad to have had the pleasure to work with them.

In addition, I would like to thank Ms. Svenja Sladek for providing me with some supplementary data used throughout this project and Dr. Lidia Tajber, Mr. Ray Keaveny and Mr. Brian Talbot for their assistance in attempting to recover certain data recorded for this project.

references

1. Poretsky, L. (2010). Principles of Diabetes Mellitus. 1st ed. New York: Springer Verlag.

2. Standards of Medical Care in Diabetes--2014. (2013). Diabetes Care, 37, pp.S14-S80.

3. International Diabetes Federation, (2014). Diabetes: facts and figures. [online] Available at: http://www.idf.org/worlddiabetesday/toolkit/gp/facts-figures [Accessed 4

12


Oct. 2014].4. Holden, S., Gale, E., Jenkins-Jones, S. and Currie, C. (2014).

How many people inject insulin? UK estimates from 1991 to 2010. Diabetes, Obesity and Metabolism, 16(6), pp.553--559.

5. CDC.gov, (2012). CDC - Percentage Using Diabetes Medication by Type of Medication - Treating Diabetes - Data & Trends - Diabetes DDT. [online] Available at: http://www.cdc.gov/diabetes/statistics/meduse/fig2.htm [Accessed 11 Oct. 2014].

6. Peyrot, M., Rubin, R., Lauritzen, T., Skovlund, S., Snoek, F., Matthews, D., Landgraf, R. and Kleinebreil, L. (2005). Resistance to insulin therapy among patients and providers results of the cross-national Diabetes Attitudes, Wishes, and Needs (DAWN) study. Diabetes Care, 28(11), pp.2673--2679.

7. Guimaraes, C., Marra, C., Gill, S., Simpson, S., Meneilly, G., Queiroz, R. and Lynd, L. (2010). A discrete choice experiment evaluation of patients’ preferences for different risk, benefit, and delivery attributes of insulin therapy for diabetes management. Patient preference and adherence, 4, p.433.

8. Ahmann, A., Szeinbach, S., Gill, J., Traylor, L. and Garg, S. (2014). Comparing Patient Preferences and Healthcare Provider Recommendations with the Pen Versus Vial-and-Syringe Insulin Delivery in Patients with Type 2 Diabetes. Diabetes technology \& therapeutics, 16(2), pp.76--83.

9. Stockl, K., Ory, C., Vanderplas, A., Nicklasson, L., Lyness, W., Cobden, D. and Chang, E. (2006). An evaluation of patient preference for an alternative insulin delivery system compared to standard vial and syringe*. Current Medical Research and Opinion, 23(1), pp.133--146.

10. FDA, (2006). Exubera Package Insert. [online] Available at: http://www.accessdata.fda.gov/drugsatfda_docs/label/2006/021868lbl.pdf [Accessed 6 Oct. 2014].

11. Gansslen, M. (1925). “Uber inhalation von insulin. Journal of Molecular Medicine, 4(2), pp.71--71.

12. Peyrot, M., Rubin, R., Kruger, D. and Travis, L. (2010). Correlates of insulin injection omission. Diabetes care, 33(2), pp.240--245.

13. FDA Approval Letter (Exubera). (2006). [online] Available at: http://www.accessdata.fda.gov/drugsatfda_docs/nda/2006/021868s000_Approv.pdf [Accessed 4 Oct. 2014].

14. FDA Approval letter (AFREZZA). (2014). [online] Available at: http://www.accessdata.fda.gov/drugsatfda_docs/appletter/2014/022472Orig1s000ltr.pdf [Accessed 4 Oct. 2014].

15. Hoffman, A. and Ziv, E. (1997). Pharmacokinetic considerations of new insulin formulations and routes of administration. Clinical pharmacokinetics, 33(4), pp.285--301.

16. Khafagy, E., Morishita, M., Onuki, Y. and Takayama, K. (2007). Current challenges in non-invasive insulin delivery systems: a comparative review. Advanced drug delivery reviews, 59(15), pp.1521--1546.

17. Generated using ChemBio3D using PDB ID: 2HIU Hua, Q., Gozani, S., Chance, R., Hoffmann, J., Frank, B. and Weiss, M. (1995). Structure of a protein in a kinetic trap. Nature Structural & Molecular Biology, 2(2), pp.129--138.

18. Barg, S. (2003). Mechanisms of Exocytosis in Insulin-

Secreting B-Cells and Glucagon-Secreting A-Cells. Pharmacology & toxicology, 92(1), pp.3--13.

19. Stanfield, C. (2014). Principles of Human Physiology: Pearson New International Edition. 1st ed. Harlow: Pearson Education Limited.

20. Arnott, J. and Planey, S. (2012). The influence of lipophilicity in drug discovery and design. Expert opinion on drug discovery, 7(10), pp.863--875.

21. TenHoor, C. and Dressman, J. (1992). Oral absorption of peptides and proteins. STP pharma sciences, 2(4), pp.301--312.

22. Polin, R., Fox, W. and Abman, S. (2011). Fetal and neonatal physiology. 1st ed. Philadelphia: Elsevier Saunders.

23. Vukavic, T. (1984). Timing of the gut closure. Journal of pediatric gastroenterology and nutrition, 3(5), pp.700--703.

24. Webb, K. (1990). Intestinal absorption of protein hydrolysis products: a review. Journal of Animal Science, 68(9), pp.3011--3022.

25. Yun, Y., Cho, Y. and Park, K. (2013). Nanoparticles for oral delivery: Targeted nanoparticles with peptidic ligands for oral protein delivery. Advanced drug delivery reviews, 65(6), pp.822--832.

26. Wilczewska, A., Niemirowicz, K., Markiewicz, K. and Car, H. (2012). Nanoparticles as drug delivery systems. Pharmacological Reports, 64(5), pp.1020--1037.

27. Hosseininasab, S., Pashaei-Asl, R., Khandaghi, A., Nasrabadi, H., Nejati-Koshki, K., Akbarzadeh, A., Joo, S., Hanifehpour, Y. and Davaran, S. (2014). Synthesis, characterization, and In vitro studies of PLGA-PEG nanoparticles for oral Insulin delivery. Chemical biology & drug design.

28. Mukhopadhyay, P., Chakraborty, S., Bhattacharya, S., Mishra, R. and Kundu, P. (2014). pH-sensitive chitosan/alginate core-shell nanoparticles for efficient and safe oral insulin delivery. International journal of biological macromolecules.

29. de Araujo, T., Teixeira, Z., Barbosa-Sampaio, H., Rezende, L., Boschero, A., Duran, N. and Hoeehr, N. (2013). Insulin-Loaded Poly (ε-Caprolactone) Nanoparticles: Efficient, Sustained and Safe Insulin Delivery System. Journal of biomedical nanotechnology, 9(6), pp.1098--1106.

30. Mesiha, M., Sidhom, M. and Fasipe, B. (2005). Oral and subcutaneous absorption of insulin poly (isobutylcyanoacrylate) nanoparticles. International journal of pharmaceutics, 288(2), pp.289--293.

31. Biolaxy.com, (2009). Biolaxy: Pipeline. [online] Available at: http://www.biolaxy.com/pipeline.html [Accessed 9 Oct. 2014].

32. Accesspharma.com, (2011). Cobalamin Mediated Oral Drug Delivery | Access Pharmaceuticals Inc.. [online] Available at: http://www.accesspharma.com/product-programs/cobalamin-mediated-oral-drug-delivery/ [Accessed 9 Oct. 2014].

33. Soppimath, K., Aminabhavi, T., Kulkarni, A. and Rudzinski, W. (2001). Biodegradable polymeric nanoparticles as drug delivery devices. Journal of controlled release, 70(1), pp.1--20.

34. Sarmento, B. and Neves, J. (2012). Chitosan-based systems for biopharmaceuticals. Chichester, West Sussex: John Wiley & Sons.

35. Stern, R. (2004). Hyaluronan catabolism: a new metabolic

13


pathway. European journal of cell biology, 83(7), pp.317--325.

36. Necas, J., Bartosikova, L., Brauner, P. and Kolar, J. (2008). Hyaluronic acid (hyaluronan): a review. Veterinarni medicina, 53(8), pp.397--411.

37. Yadav, A., Mishra, P. and Agrawal, G. (2008). An insight on hyaluronic acid in drug targeting and drug delivery. Journal of drug targeting, 16(2), pp.91--107.

38. Brown, M. and Jones, S. (2005). Hyaluronic acid: a unique topical vehicle for the localized delivery of drugs to the skin. Journal of the European Academy of Dermatology and Venereology, 19(3), pp.308--318.

39. Woitiski, C., Carvalho, R., Ribeiro, A., Neufeld, R. and Veiga, F. (2008). Strategies toward the improved oral delivery of insulin nanoparticles via gastrointestinal uptake and translocation. BioDrugs, 22(4), pp.223--237.

40. Hwang, S., Maitani, Y., Takayama, K. and Nagai, T. (2000). High entrapment of insulin and bovine serum albumin into neutral and positively-charged liposomes by the remote loading method. CHEMICAL AND PHARMACEUTICAL BULLETIN-TOKYO-, 48(3), pp.325--329.

41. Sladek, S. (2014). [Chitosan/hyaluronic acid nanopartices for oral insulin delivery]. Unpublished raw data from P.hD studies.

42. Hu, Y., Jiang, X., Ding, Y., Ge, H., Yuan, Y. and Yang, C. (2002). Synthesis and characterization of chitosan--poly (acrylic acid) nanoparticles. Biomaterials, 23(15), pp.3193--3201.

43. Teijeiro-Osorio, D., Remunan-Lopez, C. and Alonso, M. (2008). New generation of hybrid poly/oligosaccharide nanoparticles as carriers for the nasal delivery of macromolecules. Biomacromolecules, 10(2), pp.243--249.

44. Umerska, A., Paluch, K., Inkielewicz-Stkepniak, I., Santos-Martinez, M., Corrigan, O., Medina, C. and Tajber, L. (2012). Exploring the assembly process and properties of novel crosslinker-free hyaluronate-based polyelectrolyte complex nanocarriers. International journal of pharmaceutics, 436(1), pp.75--87.

45. Merck Index, 12th Ed., No. 501146. Jameel, F., Hershenson, S., Rathore, A. and Mhatre, R.

(n.d.). Formulation and Process Development Strategies for Manufacturing Biopharmaceuticals. Wiley Online Library.

47. Whittingham, J., Scott, D., Chance, K., Wilson, A., Finch, J., Brange, J. and Guy Dodson, G. (2002). Insulin at pH 2: structural analysis of the conditions promoting insulin fibre formation. Journal of molecular biology, 318(2), pp.479--490.

48. Xu, Y., Yan, Y., Seeman, D., Sun, L. and Dubin, P. (2011). Multimerization and aggregation of native-state insulin: effect of zinc. Langmuir, 28(1), pp.579--586.

49. Moreno-Bautista, G. and Tam, K. (2011). Evaluation of dialysis membrane process for quantifying the in vitro drug-release from colloidal drug carriers. Colloids and Surfaces A: Physicochemical and Engineering Aspects, 389(1), pp.299--303.

50. Abe, M., Okada, K., Ikeda, K., Matsumoto, S., Soma, M. and Matsumoto, K. (2011). Characterization of insulin adsorption behavior of dialyzer membranes used in hemodialysis. Artificial organs, 35(4), pp.398--403.

51. Dunn, M. (2005). Zinc--ligand interactions modulate assembly and stability of the insulin hexamer--a review. Biometals, 18(4), pp.295--303.

52. Heinemann, L. and Muchmore, D. (2012). Ultrafast-acting insulins: state of the art. Journal of diabetes science and technology, 6(4), pp.728--742.

53. Bures, J., Kopacova, M., Vorisek, V., Bukac, J., Neumann, D., Rejchrt, S., Pozler, O., Douda, T., Zivny, P. and Palicka, V. (2004). [Examination of gastric emptying rate by means of 13C-octanoic acid breath test. Methods of the test for adults and results of the investigation of healthy volunteers]. Casopis lekaru ceskych, 144, pp.18--22.

54. Hellmig, S., Von Schoning, F., Gadow, C., Katsoulis, S., Hedderich, J., Folsch, U. and Stuber, E. (2006). Gastric emptying time of fluids and solids in healthy subjects determined by 13C breath tests: influence of age, sex and body mass index. Journal of gastroenterology and hepatology, 21(12), pp.1832--1838.

55. Gel filtration: Principles and Methods. (2011). GE Healthcare handbook.

56. Yu, H., Dong, X. and Sun, Y. (2004). Affinity Chromatography of Insulin with a Heptapeptide Ligand Selected from Phage Display Library. Chromatographia, 60(7-8), pp.379-383.

57. Modi, S. and Anderson, B. (2013). Determination of drug release kinetics from nanoparticles: overcoming pitfalls of the dynamic dialysis method. Molecular pharmaceutics, 10(8), pp.3076--3089.

58. Jaworek, A. and Sobczyk, A. (2008). Electrospraying route to nanotechnology: An overview. Journal of Electrostatics, 66(3-4), pp.197-219.

59. Gomez, A., Bingham, D., Juan, L. and Tang, K. (1998). Production of protein nanoparticles by electrospray drying. Journal of Aerosol Science, 29(5-6), pp.561-574.

60. Xie, J., Ng, W., Lee, L. and Wang, C. (2008). Encapsulation of protein drugs in biodegradable microparticles by co-axial electrospray. Journal of Colloid and Interface Science, 317(2), pp.469-476.

14

In-vitro release studies of insulin-loaded nanoparticles in simulated gastrointestinal fluids

Documents

insulin release

use of insulin

discovery of insulin

insulinloaded nanoparticles

low release of insulin

rate of insulin permeation

dynamic dialysis

novel oral insulin formulation