Evaluation of bioadhesive polymers as delivery systems for nose to brain delivery: In vitro characterisation studies

Journal of Controlled Release 118 (2007) 225 – 234 www.elsevier.com/locate/jconrel

Evaluation of bioadhesive polymers as delivery systems for nose to brain delivery: In vitro characterisation studies S.T. Charlton a , S.S. Davis a,b , L. Illum c,⁎ a

School of Pharmacy, University of Nottingham, University Park, Nottingham NG7 2RD, UK Cosmas-Damian Ltd, 19 Cavendish Crescent North, The Park, Nottingham NG7 1BA, UK c IDentity, 19 Cavendish Crescent North, The Park, Nottingham NG7 1BA, UK

b

Received 14 August 2006; accepted 18 December 2006 Available online 23 December 2006

Abstract There is an increasing need for nasal drug delivery systems that could improve the efficiency of the direct nose to brain pathway especially for drugs for treatment of central nervous system disorders. Novel approaches that are able to combine active targeting of a formulation to the olfactory region with controlled release bioadhesive characteristics, for maintaining the drug on the absorption site are suggested. If necessary an absorption enhancer could be incorporated. Low methylated pectins have been shown to gel and be retained in the nasal cavity after deposition. Chitosan is known to be bioadhesive and also to work as an absorption enhancer. Consequently, two types of pectins, LM-5 and LM-12, together with chitosan G210, were selected for characterisation in terms of molecular weight, gelling ability and viscosity. Furthermore, studies on the in vitro release of model drugs from candidate formulations and the transport of drugs across MDCK1 cell monolayers in the presence of pectin and chitosan were also performed. Bioadhesive formulations providing controlled release with increased or decreased epithelial transport were developed. Due to their promising characteristics 3% LM-5, 1% LM-12 pectin and 1% chitosan G210 formulations were selected for further biological evaluation in animal models. © 2006 Elsevier B.V. All rights reserved. Keywords: Chitosan; Pectin; Nose to brain delivery; Drug release; Drug transport

1. Introduction The blood–brain barrier (BBB), segregating the brain interstitial fluid from the circulating blood, and the blood– cerebrospinal fluid barrier (BCB), separating the blood from the cerebrospinal fluid (CSF) that encircles the brain, provide efficient barriers to the diffusion of drugs from the blood stream into the central nervous system (CNS) especially of polar drugs such as peptides and proteins. Hence, these barriers prevent the utilization of many novel therapeutic agents, for example neuropeptides, for treating CNS disorders such as Parkinson's and Alzheimer's diseases [1]. It has been shown in animal and in human studies that after nasal administration drugs can be transported directly from the nasal cavity to the CNS via the olfactory epithelium and/or the trigeminal nerve system thereby bypassing the BBB and the ⁎ Corresponding author. E-mail address: [email protected] (L. Illum). 0168-3659/$ - see front matter © 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.jconrel.2006.12.014

BCB [1,2]. However, as shown in animal studies, the amounts of drug that can be transported via these pathways into the CNS are normally minimal; less than 1% of the drug administered. It is likely that the efficiency of these pathways would be less in humans, due for example to the much smaller olfactory region in this species. Furthermore, the olfactory region in man is situated in the upper part of the nasal cavity, an area that is difficult to reach with presently available nasal spray or powder devices. This is contrary to many animals, such as rats and dogs, in which the olfactory region constitutes about 50% of the surface area of the nose and is situated in the upper and posterior part of the nasal cavity and hence it is easier to reach. Nasal spray or powder devices are now being developed (e.g. OptiNose and ViaNase) that claim to enable the targeting of formulations to specific sites within the nasal cavity in man or to provide a larger deposition area that includes the olfactory region. However, not withstanding these claims once the olfactory region is reached it is then necessary, through formulation approaches, to enhance the transport across the

226

S.T. Charlton et al. / Journal of Controlled Release 118 (2007) 225–234

olfactory epithelium. It has been suggested [1] that an approach, that combines a bioadhesive formulation with vectors that specifically target receptors on the olfactory region and possibly together with an absorption enhancer, may not only target and retain the formulation on the site of absorption for prolonged periods, but also enhance the transport through the membrane by transcytosis (e.g. receptor mediated) or paracellular mechanisms. We have evaluated the use of chitosan and pectins, which have previously been described in the literature as suitable excipients for nasal drug delivery formulations for the improvement of pharmacokinetic parameters [3,4]. Pectin is a natural polysaccharide, consisting of methylated esters of polygalacturonic acid, linked by α- [1–4] glycosidic bonds and with carboxyl and methyl ester side chains. Pectin is commercially available in three different forms each varying in their substitution degree; high methoxyl (HM), low methoxyl (LM) and low methoxyl amidated (LMA) pectin. LM pectin is of most interest for nasal delivery of drugs because a LM pectin solution formulation is able to gel upon contact with the nasal mucosa without the addition of exogenous Ca2+ [4]. Furthermore, LM pectin has been shown to be highly mucoadhesive. In a gamma scintigraphy study in sheep almost half (45%) of a 2% pectin formulation was still present in the nasal cavity 60 min after application, as compared to 15% for a simple solution without pectin. It was further shown that pectin did not increase the bioavailability of nasally applied calcitonin and hence did not act as an absorption enhancer [4]. However, a 1% pectin solution was able to significantly modify the in vitro release characteristics of a drug (fexofenidine), producing a slow release formulation. This beneficial effect has also been shown for other drugs and has been exploited in the development of controlled release nasal formulations (personal communications). Chitosan is a natural polysaccharide present in fungi but is normally derived from chitin of crustacean origin by a process of deacetylation. Chitosan consists of copolymers of glucosamine and N-acetyl-glucosamine. The degree of deacetylation (DDA) of the chitosan is determined by the amount of the acetyl groups present. Amine groups present in the structure of chitosan can be protonated when in solution, thereby giving the molecule a positive charge. Chitosan is available in a broad range of molecular weights and salt forms. Chitosan does not gel in contact with Ca2+ ions nor is it a thermogelling polymer. Chitosan has been shown to be mucoadhesive in the nasal cavity [5]. It increased the half-life of nasal retention from 15 min, for a non-bioadhesive control solution, to 40 min when administered in solution to human volunteers. Chitosan is also a very effective absorption enhancer that is able to transiently open tight junctions in nasal epithelial cells by interaction with the Protein Kinase C pathway [1,6]; Hence, if delivered directly to the olfactory region, it is believed that chitosan could increase the retention of the formulation at this site and at the same time promote the transport of drugs from the olfactory region to the CNS. The purpose of the present study was to identify one or more polymer formulations with ideal characteristics for olfactory delivery. The polymer should be bioadhesive, produce solution viscosities high enough to provide retention on the nasal

mucosa and low enough to allow administration with a simple nasal spray device, provide controlled release of drugs in solution and if required provide absorption enhancement. For this purpose the characteristics of two selected LM pectins with 5% (LM-5) and 35% (LM-12), esterification and chitosan G210 (a medium molecular weight chitosan nominal 210 kDa) were evaluated. The polymers were characterised in terms of molecular weight, gelling ability and viscosity, and the release of a model drug from the polymer formulations was evaluated using a Franz diffusion cell apparatus (FDC). Furthermore, the absorption enhancing effects of the pectins and the chitosan across MDCK1 cell monolayers were evaluated using mannitol and propranolol as model hydrophilic and lipophilic drugs. 2. Experimental 2.1. Materials Genu pectin LM-5 CS (BN91837) and Genu pectin LM-12 CG (BN G1451) were kindly donated by Hercules (Hercules Inc., Salford, Lancashire, UK). Chitosan Seacure G210 (BN604-583-08B1.1)(Now known as Protosan UP G213) (glutamate salt) was donated by Pronova Biopolymer (now known as Novomatrix, FMC Biopolymer, Drammen, Norway). For the Franz diffusion cell studies, anhydrous calcium chloride (C-1016, LOT 55H0291), sodium chloride (S-7653, LOT 68H1130) and potassium chloride (P-9333, LOT 29H00321) were used to manufacture Simulated Nasal Electrolyte Solution (SNES), and were purchased from Sigma Chemical Company, Poole, Dorset, UK. The model drug candidate GR138950X (BN C1798/210/1), a zwitterionic molecules (pKa1 3.6, pKa2 5.5), was kindly donated by GlaxoWellcome, Stevenage, Hertfordshire, UK (Fig. 1). Deionised water (Elga, High Wycombe, Buckinghamshire, UK)

Fig. 1. Molecular structure of GR138950.

S.T. Charlton et al. / Journal of Controlled Release 118 (2007) 225–234

was used for making up solutions. Whatman glass microfibre filter paper (GF/A) (1.6 μm retention size, 21 mm diameter, product code 1820 021) was used as the model membrane, and was purchased from Fisher Scientific UK Ltd, Loughborough, UK. For the cell transport studies, minimum essential medium (MEM) supplemented with Earle's salts and L-glutamine (31095-029); Hanks balanced salt solution (HBSS) without phenol red (14025-050); and L-glutamine (25030-024) were purchased from Gibco BRL (Invitrogen Ltd, Paisley, UK). Foetal calf serum (F4135), hydroxyethylpiperazine-N-2-ethanesulfonic acid (HEPES) buffer (H0887), 14C-mannitol and 3Hpropranolol were purchased from Sigma (Poole, Dorset, UK). All other reagents used were at least of analytical grade. 2.2. Molecular weight determination The molecular weights of the two pectins were determined using size exclusion chromatography/multi angle laser light scattering (SECMALLS) [7]. One percent phosphate buffered saline was pumped at a flow rate of 0.8 ml/min through a column system consisting of TSK-G5000PW, TSK-G4000PW and TSK3000PW analytical columns protected by a guard column. Solutions of each pectin sample were prepared as 1% solutions in standard phosphate buffered saline (pH 7, ionic strength 0.1, RI = 1.333). 100 μl of the sample was injected onto the column at ambient temperature (sample filtered through 0.45 μm membrane filter to remove any insoluble material or dust prior to injection). The eluting fractions were monitored using a Dawn DSP multiangle light scattering photometer (Wyatt Technology, Santa Barbara, Ca, USA) fitted with a 5 mW He–Ne laser and a differential interferometric refractometer (Optilab 903, Wyatt Technology, Santa Barbara, Ca, USA). Apparent weight average molecular weights (Mw,app) were obtained using the so-called Debye plot method as described previously [7]. Low speed sedimentation equilibrium was used to determine the molecular weights (weight averages) of the chitosan sample. A Beckman XL-1 Analytical Centrifuge was used for chitosan solutions of 0.5, 0.8 and 1.0 mg/ml in 0.2 M acetate buffer, at a rotor speed of 8000 rpm and a temperature of 20.0 °C as previously described in Fee et al. [7]. The data collected from the Rayleigh interference optical system were analysed using the MSTARI algorithm [8] to give Mw,app. A plot of Mw,app against concentration for each formulation was extrapolated to zero concentration to determine the weight average molecular weight (Mw). 2.3. Gelling ability A range of calcium chloride solutions were prepared over the concentration range 0.001–0.05 M. 15 μl calcium chloride solution was added to 15 μl of polymer solution in the well of a standard 96 well microtiter plate. The polymer concentrations evaluated were: • Pectin LM-5, 0.5, 1.0, 2.0 and 3.0% in water • Pectin LM-12, 1.0 and 1.5% in water • Chitosan G210—1.0% in water

227

The samples were left for 5 min and the solutions then probed with a pipette tip and visually inspected for gel formation. Each reaction was subjectively rated for degree of gel formation, with 0 points for no visible gelling, 1 point for limited gel formation and 2 points for extensive gelling. Three replicas were used and the scores combined to give a rating from 0 to 6. 2.4. Viscosity A Bohlin Instruments ‘Controlled Stress Rheometer CS10’ with a DG40/50 geometry attachment (cup and cylinder) and a PP40 geometry attachment (parallel plate) was used for measurements of viscosity of the polymers. Rheoscale software (Brookfield Engineering Laboratories) was used to control the rheometer and to capture the data. All three polymers were initially analysed using concentrations of 0.2, 0.4, 0.6, 0.8 and 1.0% w/v in water for injection (WFI). This was done using 30 ml of polymer solution with the DG40/50 cup and cylinder over a shear rate range of 1–100 s− 1. 100 s− 1 was the limit of the equipment. The sample stage was held at 20 °C using the integrated water circulation system coupled to a water bath. This temperature was chosen to reflect normal room temperature when using a nasal device. It has been estimated that a solution delivered from a nasal spray device is subjected to shear rates in an approximate range of 24 000–240 000 s− 1 [9]. Therefore the PP40 parallel plate with the upper limit of shear rate of 18 000 s− 1 was used for evaluation of the viscosities of the polymer solutions given below: • Pectin LM-5—1.0, 3.0 and 6.0% in water • Pectin LM-12—1.0 and 3.0% in water • Chitosan G210—1.0% in water The concentrations used for these studies included some selected from those used in Section 2.3 and higher concentrations in order to evaluate the impact on viscosity on a wider scale. 2.5. In vitro drug release studies The Franz diffusion cell used in the drug release studies was a ‘FDC-200 Flo-Thru’ cell, manufactured by Crown Glass Company Inc. (Somerville, NY, USA). A Beckman DU 640 UV/Vis Spectrophotometer (Beckman Coulter (UK) Ltd., High Wycombe, Bucks, UK) was used to measure drug concentration. The FDC enables in vitro analysis of drug movement across a membrane using a two-compartment model. The donor compartment contains the dose, and a non-rate limiting membrane separates the compartments and supports the dose. Thus, drug-release profiles can be produced for controlled-drug release formulations. The equipment was used to determine the rate and quantity of drug released from formulations containing differing concentrations of pectin or chitosan. A model drug, GR138950 was used as the test substance, and was quantified using UV photospectroscopy.

228

S.T. Charlton et al. / Journal of Controlled Release 118 (2007) 225–234

SNES mimics the extracellular fluid of the nasal epithelium with respect to salt content and concentration [10], and is composed of 8.77% (w/v) sodium chloride, 2.98% (w/v) potassium chloride and 0.45% (w/v) calcium chloride in aqueous solution. SNES was used as the receptor solution to mimic calcium ion concentration on the surface of the nasal mucosa, thus enabling pectin to form a gel on the membrane. The ionic strength of the SNES solution was calculated as 0.404 M and the pH was measured as 5.2. The FDC was connected inline with the UV spectrophotometer using a flow through cuvette (0.2 cm path length). A magnetic stirrer was placed under the FDC to rotate the stirring bar within the receptor compartment. The FDC was maintained at 37 °C by running heated water through the water jacket of the cell. A peristaltic pump was used to move the receptor solution through the FDC and spectrophotometer at a flow rate of 6 ml min− 1. SNES was added to the system and allowed to circulate for 10 min. This gave time for the temperature to equilibrate and for air bubbles to circulate out. The FDC alone held a volume of 5 ml, and the whole setup held a total volume of 7 ml. The microfibre filter membrane was submerged in SNES before being placed on the flange of the FDC. The cap was then placed on the top and was secured using a clamp. Care was taken to avoid trapping air under the membrane. GR138950 adsorption onto the membrane was examined by performing the experiment with and without the membrane and using GR138950 solution. Drug concentration in the receptor solution was comparable (less than 5% difference) after 6 h, indicating that only a small amount of drug was adsorbed onto the membrane. The pectins and the chitosan formulations tested for drug release were the same as described under Section 2.4 except that 6% pectin LM-5 was not included and that the formulations all contained 3 mg/ml of GR 138950. A drug control solution was also included. 30 μl was dosed into the FDC cap for each formulation. At equilibrium, the receptor solution would contain approximately 13 μg/ml GR138950 – this is within the linear region of the standard plot, therefore the absorbance was directly proportional to the concentration. All drug release studies were performed in triplicate. 2.6. In vitro cell permeability studies The effects of the pectins and chitosan on drug transport across an epithelial monolayer and ability to open cell tight junctions were investigated by measuring transepithelial movement of mannitol and propranolol, and measuring transepithelial electrical resistance (TEER). Madin–Darby canine kidney (MDCK1) epithelial cells were used. These cells have been used previously for testing the effect of excipients on membrane permeability [11,12]. These cells were obtained from the GlaxoSmithKline cell bank, and were at passage 25. The study was performed in Transwells™ 12 well plates (Costar UK Ltd, High Wycombe, Bucks, UK), and electrical resistance was measured using an EVOM Epithelial Tissue Voltohmmeter (World Precision Instruments, Inc., Sarasota, Florida, USA) and an Endohm-12 Chopstick electrode (World Precision Instruments, Inc., Sarasota, Florida, USA). Cell cultures were kept in maintenance medium when not used in TEER experiments. This medium was made by mixing

500 ml MEM, 50 ml fetal calf and 5 ml glutamine solution. The medium was filtered through a membrane filter (0.22 μm) prior to use. During TEER experiments the maintenance medium was replaced with transport medium. This was made by adding 5 ml of 1 M HEPES buffer to 500 ml HBSS. The pH was adjusted to pH 7.4 using concentrated sodium hydroxide solution. The medium was filtered through a membrane filter (0.22 μm) prior to use. Polymer formulations of 3% pectin LM-5, 1% pectin LM-12 and 0.5% chitosan G210 were studied. The polymers were dissolved in transport medium, and were left stirring over night. Transport medium without polymer was used as a control. 3 ml aliquots of each formulation were spiked with 3 μl of 10 mM 14C-mannitol and 3 μl of 10 mM 3H-propranolol prior to the addition to the cell cultures. The MDCK1 cells were seeded onto Transwells plates at 80,000 cells per cm2 and were incubated for 3–7 days to allow a confluent monolayer of cells to form before use. The maintenance medium was then removed, and transport medium was added to the monolayers (1.5 ml to basolateral compartment and 0.5 ml to apical compartment). The cells were then incubated for an hour to equilibrate. TEER was measured using chopstick electrodes that were placed at the apical and basolateral compartments of each well and the resistance was measured with the voltohmmeter in order to determine membrane integrity. The transport medium was removed from the Transwells and the wells were inserted into a new cluster plate containing 1.5 ml transport medium in each well. 0.5 ml dose solution or transport medium (control) (pre-heated to 37 °C) was then added to the apical compartment at time zero. The TEER was measured at 15 min pre-dose, 0 (immediately after addition of the test formulation), 15, 30, 45 and 60 min post-dose. After TEER measurements taken at 15, 30 and 45 min, the Transwells were transferred into new cluster plates, and the remaining basal samples were transferred into vials for analysis using liquid scintillation. The cells were stored in the incubator (37 °C) between measurements. After the final measurement the cell culture inserts were removed from the wells. The apical and basal samples were collected for liquid scintillation analysis, and the filters from the Transwells were analysed for radioactive content. Each test formulation was tested in triplicate. The apparent permeability (Papp) for each polymer solution was determined. The Papp is the rate that a compound moves through a membrane or monolayer. Papp ðnmd s−1 Þ ¼

V dc  A  C0 dt

Where: V A Co dc dt

Volume of basal solution (nm3) Area of monolayer or membrane (nm2) Initial concentration of test substance in the apical solution (μM) Change in concentration of the test tube substance in the basal solution over a set time period (μM) Time (s)

S.T. Charlton et al. / Journal of Controlled Release 118 (2007) 225–234

229

2.7. Statistical methods Statistical analysis was performed by GraphPad Prism version 3.02 for Windows (GraphPad Software, San Diego, California, USA) using two-way ANOVA and Bonferroni post-tests. 3. Results and discussion 3.1. Molecular weights Table 1 shows the molecular weight averages and polydispersities for pectin LM-5 and pectin LM-12 as measured by SECMALLS. Fig. 2 gives the dependence of molecular weight on concentration of chitosan G210 as measured by sedimentation equilibrium. Mw,app is plotted against chitosan concentration to determine the weight average molecular weight (Mw) as the intercept on the Y-axes. Mw was found to be 12.720 · 104 g/mol ± 1.0. The Mw values for the three polymers are compared in Table 2. It can be seen that the two pectins are similar in size whereas the chitosan is larger. The manufacturer does not provide values for Mw of these particular pectins and hence the derived values cannot be compared with the results obtained elsewhere. However, pectins with a higher Mw average of 150 kDa produced from depolymerisation of a commercially available HM pectin have been discussed in the literature [13]. It should be noted that according to Fee et al. [7] molecular weights obtained by sedimentation equilibrium are normally somewhat lower than those measured with for example SECMALLS. This can explain the discrepancy between the expected molecular weight of around 200 kDa of the chitosan as indicated by its designated number. 3.2. Gel formation Both types of pectins formed gels with CaCl2. Table 3 shows the total scores given to each combination of polymer and Ca2+ concentration. A score of 6 signifies extensive gel formation and a score of 0 indicates no gel formation. The degree of gelling increased for the two pectin samples with increasing concentration of polymer and increasing Ca2+ concentration. This is in agreement with results by Löfgren et al. [14] who studied the gel

Table 1 Molecular weight averages and polydispersity for pectin LM-5 and pectin LM-12 Polymer

Pectin LM-5

Replicate

1

Number average molecular weight, Mn (×104) Weight average molecular weight, Mw (×104) Z-average molecular weight, Mz (×105) Polydispersity, Mw / Mn

Pectin LM-12 2

1

2

5.400 ± 0.8

3.906 ± 0.7

5.070 ± 0.5

4.845 ± 0.5

9.674 ± 1.0

7.750 ± 0.8

8.605 ± 0.4

8.374 ± 0.5

Fig. 2. Molecular weight concentration dependence for chitosan G210. Mw(app) is plotted against concentration for three concentrations of the polymer. The plot is extrapolated to zero concentration to determine the weight average molecular weight (Mw). Mw = 127,200 g/mol, SD = ±10,400.

forming and rheological properties of a pectin similar to LM-12. As expected chitosan G210 did not form a gel with Ca2+. On the basis of the measured molecular weights of the pectin it can be calculated that a LM-5 molecule has an average number of carboxyl groups of 471 whereas a LM-12 molecule (due to the higher degree of esterification) has only 294. For the lower CaCl2 concentrations and the same concentration of pectin LM-5 and LM-12 (e.g. 0.005 M and 1%, respectively), LM-5 produced a stronger gel structure which is suggested to be due to the larger amount of carboxyl groups available on the LM-5 polymer. The calcium ion concentration in the extracellular fluid in the nasal cavity is approximately 0.004 M. At this concentration LM-5 should form extensive gels at polymer concentrations of 1% or greater, whereas a concentration of 1.5% LM-12 should be used for effective gel formation. 3.3. Viscosity The viscosity of a formulation is of relevance with respect to its administration and clearance from the nasal cavity. A solution should have sufficiently low viscosity to permit delivery through the narrow orifice of a spray device or dosing cannula, but be sufficiently viscous to retain the dose in the nasal cavity. This has been shown to be the case for various commercial nasal spray suspensions (e.g. Beconase, Nasacort, Flixonase and Nasonex). It was shown that all of the nasal suspensions were shear thinning and also thixotropic to various degrees. The absence, at 5 min, of significant thixotropic recoveries for all of the sprays suggested that thixotropy is not necessarily the controlling factor for the prolonged residence in the nasal cavity but rather the high viscosities present in the nasal sprays, even after structural breakdown [15]. Table 2 Weight average molecular weights of pectins and chitosan

4.185 ± 2.0

1.1791 ± 0.369

2.891 ± 0.8

1.984 ± 0.413

1.999 ± 0.2

1.697 ± 0.183

2.076 ± 0.3

1.728 ± 0.195

Polymer

Weight average molecular weight, Mw (g/mol)

Pectin LM-5 Pectin LM-12 Chitosan G210

87,120 ± 6000 84,895 ± 5000 127,200 ± 10,400

230

S.T. Charlton et al. / Journal of Controlled Release 118 (2007) 225–234

Table 3 Gelling capacity of polymer solutions at a variety of concentrations mixed with calcium chloride solution over the concentration range 0.001–0.05 M CaCl2 concentration (M)

LM-5

LM-12

G210

0.5%

1%

2%

3%

1%

1.5%

1%

0.001 0.005 0.01 0.02 0.03 0.04 0.05

2 5 6 6 6 6 6

6 6 6 6 6 6 6

6 6 6 6 6 6 6

6 6 6 6 6 6 6

0 3 4.5 6 6 6 6

0 1 2.5 5.5 6 6 6

0 0 0 0 0 0 0

Scores are subjective, based on observation. n = 3.

Increasing the residence time of the dose on the nasal epithelium may have an effect on the absorption of drugs across the nasal epithelium although this has not been proven incontrovertibly [16]. Here the apparent viscosities have been determined for the LM-5 and LM-12 pectins and the G210 chitosan at various concentrations and at low and high shear rates. This would give an indication of the optimal concentration for a nasal formulation for dosing and increased residence time. For all three polymers, at shear rates up to 100 s− 1, it was found that as expected the viscosity profile increased with polymer concentration (data not shown). At 1% polymer concentration, chitosan G 210 displayed the highest viscosity (∼ 65 vmPa s), followed by pectin LM-12 (∼ 18 mPa s) and then pectin LM-5 (∼ 4 mPa s). It may have been hypothesised that the LM-5 pectin with the higher charge density (when dissociated) would take up a more extended and rod like conformation and hence have a higher apparent viscosity than LM-12, which has a similar molecular weight. However, Smidsröd and Haug [17] demonstrated that chain flexibility of pectins with degrees of esterification below 80% was independent of stochiometric charge density. Large side-chain groups along the polymer molecule have a steric effect and prevent the molecule from coiling up and therefore an extended conformation is adopted. LM-12 has a greater number of esterified side chains than LM-5 and hence exhibited a greater apparent viscosity in solution. Chitosan G210 has a considerably higher molecular weight (larger polymer chains) than both pectins, therefore collision and intertwining between the molecules is greater, resulting in a higher apparent viscosity of 65 mPa s. For both pectins and the chitosan the apparent viscosity was directly related to the concentration of the polymer (at low shear rate, 4 s− 1) as shown in Fig. 3. For the lower concentrations of polymers (LM-5 up to 1%, LM-12 up to 0.8% and G210 up to 0.4%) the solutions behaved in a Newtonian manner i.e. shear rate had no effect on the apparent viscosity (data not shown). With increasing concentrations of the polymers, the solutions exhibited shear thinning profiles (pseudoplastic), i.e. the apparent viscosity decreased with increasing shear rate (Fig. 4). It has been suggested that solutions with an apparent viscosity of 76 mPa s cannot be delivered accurately using a

Fig. 3. Apparent viscosity of polymer solutions related to concentration when subjected to a shear rate of 4 s− 1. n = 1.

standard nasal spray device [18]. However, as discussed above solutions administered with such a device are subjected to a shear rate of 24 000–240 000 s− 1. Hence, as indicated in Fig. 4 it should be possible to successfully administer the chosen polymer concentrations using a standard nasal spray device. 3.4. In vitro drug release studies Thirty minutes after application of GR138950 on the membrane in the FDC, 19.36 μg of the drug had been released from the control solution into the receptor phase, which is equivalent to 21.5% of the dose introduced into the donor chamber (Fig. 5A). The membrane had a retention size of 1.6 μm, therefore GR138950 would have been able to pass through freely. As expected, each of the polymer formulations tested exhibited a slower release rate of GR138950 compared to the control (Fig. 5A,B). The difference was not statistically significant for the 1% pectin LM-12 formulation (p = 0.2182); however, the other formulations exhibited significantly slower release rates (p = 0.05). The difference in the release profiles for 1% and 3% pectin LM-5 was not statistically significant (p = 1.0000); whereas 1% pectin LM-12 released over two and a

Fig. 4. Apparent viscosity values for pectin LM-5, pectin LM-12 and chitosan G210 solutions over a shear rate range of 0–1800 s− 1, n = 1.

S.T. Charlton et al. / Journal of Controlled Release 118 (2007) 225–234

231

There are a number of factors that influence the release of drug from the polymer formulations. Two of the primary mechanisms of drug retention are possible attractive forces between the drug and polymer molecules, and physical hindrance caused by the network of intertwined polymer molecules, and, in the case of the pectins, a gel matrix. GR138950 is a zwitterionic drug (pKa1 3.6, pKa2 5.5). The presence of both charges enables GR138950 to form ionic attractions with both pectin solutions (pH 2.7 and 5.5) and chitosan solution (pH 4.5) under suitable conditions. (Pectin is negatively charged at pH N 3.6 (pKa 3.6) and chitosan is positively charged at pH b 6.5 (pKa 6.5)). When GR138950 was mixed with pectin LM-12 (pH 2.6– 2.9), the drug molecules were predominantly positively charged; therefore an electrostatic attraction was present between the drug and polymer molecules. The 3% pectin LM12 formulation displayed the slowest rate of GR138950 release, which can reasonably be attributed to the electrostatic attraction. The Stokes–Einstein equation shows the relationship between the rate of diffusion of a drug molecule, which is relative to the diffusion coefficient, and the viscosity of the medium that the molecule is in. D¼

Fig. 5. Release of GR138950 from control and pectin solutions (A) and from 1% chitosan solution (B) using a FDC. Each test sample contained 90 mg of drug. Error bars represent SEM, n = 3.

half times more drug over 30 min compared to the 3% solution ( p = 0.0204). This is also shown by the difference in release rates (Table 4). 1% chitosan showed the second slowest release of drug, with the 3% LM-12 solution displaying the slowest rate of drug release (Fig. 5B, Table 4). The results shown here are in good agreement with the work of Murata et al. [22] who evaluated the release of diclofenac sodium salt from pectin beads. The pectin used was of similar molecular weight to those of the LM-5 and LM-12 pectins used here. With a pectin concentration of 3% (DE b 60%) the release of drug was about 60% after 60 min. The rate of release decreased with a higher pectin content. At the end of each experiment, when the FDC was dissembled, the membrane was visually examined. It was noted that each of the pectin formulations had formed a gel on the membrane; whereas the chitosan formulation was in a liquid state. The drug release rate for each formulation and the area under the curve (AUC) for each plot are detailed in Table 4. The drug release rates were calculated using linear regression (by GraphPad Prism 3.02 for Windows) to give an average release rate over the duration of the experiment and to simplify comparison. Each of the polymer formulations examined showed potential for use as a controlled release system for delivering drug nasally.

kT 6kga

Where D = diffusion coefficient; k = Boltzmann constant; T = absolute temperature; η = viscosity of the medium/solvent; a = molecular radius. Applying the Stokes–Einstein equation to this study, the diffusion coefficient describes the movement of GR138950, in the polymer solution and the medium. According to the equation, an increase in viscosity of the formulation should lead to a decrease in the diffusion coefficient, which means that the drug is released more slowly. A high concentration of polymer molecules in the formulation results in a dense network of entangled polymer molecules and an increased hindrance of drug molecules as they move within the formulation. However, a number of research groups have reported unchanged diffusion rates over a range of measured viscosities [19–21]. If a formulation contains a relatively low concentration of polymer, and the incidence rate of drug-polymer collisions is low, then the drug molecule is primarily influenced by solvent viscosity. This is described as ‘microviscosity’. It is possible for the polymer molecules in a solution to produce a high ‘macroviscosity’, in which case the overall formulation exhibits a slow

Table 4 GR138950 release rate for each test formulation Formulation

Rate of drug release (μg min− 1)

r2

Area under the curve

Control 1% pectin LM-5 3% pectin LM-5 1% pectin LM-12 3% pectin LM-12 1% chitosan G210

0.6606 0.3117 0.3071 0.3737 0.1470 0.2563

0.8234 0.6819 0.7951 0.7832 0.8242 0.7380

303.6 140.6 144.3 170.8 65.4 111.8

232

S.T. Charlton et al. / Journal of Controlled Release 118 (2007) 225–234

flow rate, in addition to having a low ‘microviscosity’, therefore drug molecules are able to diffuse out of the formulation relatively quickly. It was shown above that, at equal concentrations in aqueous solution, chitosan G210 exhibited the greatest apparent viscosity, followed by pectin LM-12 and then pectin LM-5. Analysis of the gelling activity of pectin LM-5 and pectin LM12 showed that both polymers formed a gel at polymer concentrations of 1% and above, with physiological levels of free calcium ions. Interestingly, the 1 and 3% pectin LM-5 formulations generated similar drug release profiles. 1% pectin LM-5 is able to form an extensive gel structure throughout the formulation, therefore, it can be concluded that an additional polymer content does not significantly affect drug diffusion and that the two formulations have similar ‘microviscosities’. Electrostatic interaction and gel formation both have separate, though additive, effects on drug retention. Pectin LM-5 has greater gelling capacity than LM-12, though LM-12 has greater electrostatic interaction with GR138950. The gel formed by 3% pectin LM-12 should have caused drug impedance to an equal or lesser degree than 3% pectin LM-5 based on gelling potential; but the LM-12 formulation released GR138950 at a slower rate. It is reasonable to conclude that the difference was due to the greater electrostatic interaction between the dissociated carboxyl groups of LM-12 and GR138950. However, 1% LM-12 released GR138950 at a faster rate than 1% LM-5, indicating that gel formation has a greater effect on drug release rate at this lower polymer concentration. One percent chitosan G210 released GR138950 at a faster rate than 3% pectin LM-12, but at a slower rate than the pectin LM-5 formulations (Fig. 5B). It is suggested that the longer chain length of chitosan molecules resulted in a greater ‘microviscosity’ and a slower drug diffusion rate than for the pectin LM-5 formulations. The additional effect of electrostatic attraction between LM-12 and GR138950 is accountable for the

Fig. 6. Transepithelial resistance across MDCK cells with and without the presence of pectin and chitosan. Error bars represent the SD, n = 3.

Table 5 TEER (Ω cm2) values across MDCK epithelial cells with and without the presence of pectin and chitosan Time (min) −15 0 15 30 45 60

Control

3% pectin LM-5

1% pectin LM-12

0.5% chitosan G210

Mean

SD

Mean

SD

Mean

SD

Mean

SD

4238 2690 3489 5899 6272 6247

901 593 592 213 791 616

6419 1729 1375 1700 1787 1668

751 106 73 118 182 171

6187 991 1155 102 1145 1164

940 100 38 24 27 42

5897 155 33 3 −5 − 10

1179 42 2 2 6 6

n = 3.

slower rate of drug release from the 3% LM-12 formulation compared to the 1% chitosan G210 formulation, despite the hypothesis that the chitosan formulation had a lower ‘microviscosity’. 3.5. In vitro cell permeability study The TEER values at 15 min pre-dose for the three groups of cells that were exposed to polymer formulations were similar; all above 1000 Ω cm2 indicating good cell tight junction integrity [12]. The application of the test polymer solutions caused significant decreases in TEER values indicating opening of the tight junctions (Fig. 6 and Table 5). Application of the control solution to the cells also caused a small decrease in TEER, however, this rapidly recovered. The pectin formulations caused a larger reduction in TEER compared to the control (p b 0.0001). Two-way ANOVA indicated that 1% pectin LM-12 caused a significantly greater reduction in TEER than 3% LM-5 (p b 0.0001). However the Bonferroni post-test indicated that the difference was not significant. The 0.5% chitosan G210 formulation produced a greater decrease in TEER compared to the other formulations ( p b 0.0001), resulting in a resistance of 155 Ω cm2 immediately after dosing; indicating that the cell tight junctions were opened significantly. TEER measurements did not recover significantly over 60 min post-dose in the cell cultures that were exposed to the polymer formulations. Application of the 0.5% chitosan G210 formulation to the MDCK cells resulted in the highest mannitol permeability (p b 0.0001) and low TEER values, indicating that chitosan has a significant ability to open paracellular junctions (Fig. 7). This is in agreement with published data on the ability of chitosan to open cell tight junctions [23,24]. Mannitol permeability increased over the duration of the experiment, suggesting that the effect elicited by chitosan on the cell tight junctions increased over the duration of the experiment. The 3% pectin LM-5 and 1% LM-12 formulations produced a higher mannitol concentration in the basal compartments compared to the control (p b 0.0001), but not to the same extent as chitosan (Fig. 7). Both pectin formulations caused a drop in TEER compared to the control, indicating that cell tight junctions were opened. Pectin has been shown to chelate free calcium ions, and the application of EDTA (a calcium chelator)

S.T. Charlton et al. / Journal of Controlled Release 118 (2007) 225–234

to cell membranes has been shown to cause cell tight junctions to open: therefore, it can be deduced that pectin is causing cell tight junctions to open by virtue of its calcium chelating ability. There was no statistically significant difference in mannitol transport and a marginal difference in TEER values observed with each of the two pectin formulations. The mannitol permeability and TEER values both indicate that chitosan has a significantly greater ability to open cell tight junctions compared to pectin. Although both pectin formulations produced a significant reduction in TEER compared to the control, mannitol transport increased by a relatively small amount; whereas chitosan caused a large change in TEER and mannitol transport compared to the control. It is proposed that the pectin formulations caused the cell tight junctions to open sufficiently to allow a significant increase in ion flow through the pores, and thus a reduction in TEER, but the cell tight junctions did not open wide enough to allow free movement of mannitol molecules. Studies examining the ability of EDTA to enhance paracellular transport of fluorescein isothiocyanate-dextrans (FITC-dextrans) with various molecular weights showed that cell tight junctions were opened sufficiently to cause a reduction in TEER and to permit passage of molecules up to approximately 4000 Da, but larger molecules were unable to cross the membrane [25] — although this example uses different molecules, it illustrates that chelating agents can cause cell tight junctions to open up to a limited amount. All three polymer formulations exhibited lower apparent permeabilities for propranolol (a lipophilic compound) compared to the control (Fig. 8). Compared to the control, approximately 33% less drug crossed the membrane from the chitosan formulation over 60 min, approximately 89% from pectin LM-5 and approximately 98% from pectin LM-12 ( p b 0.0001). Pectin and chitosan are not known to modify cell membranes in a way that makes them less pervious to lipophilic molecules; therefore it is proposed that the reduction in propranolol flux was due to reduced drug release from the polymer formulations. Propranolol (pKa 9.4) carries a net positive charge in the test samples and the medium, therefore an

Fig. 7. Apparent permeability of mannitol across MDCK epithelial cells with and without the presence of pectin and chitosan. Mannitol movement represents paracellular transport. Error bars represent the SD, n = 3.

233

Fig. 8. Apparent permeability of propranolol across MDCK epithelial cells with and without the presence of pectin and chitosan. Propranolol movement represents transcellular transport. Error bars represent the SD, n = 3.

electrostatic attraction will occur between the propranolol molecules and the negatively charged carboxyl groups on the pectin molecules, resulting in substantial retention of propranolol. Physical hindrance by polymer molecules will also inhibit movement of propranolol as discussed above. Chitosan exhibits a low degree of protonation under conditions approaching neutral pH, therefore electrostatic repulsion with propranolol was low, and the microviscosity and retention effect of the polymer molecules was prominent. Hence, electrostatic interaction was most likely responsible for the pectin formulations retaining propranolol more than the chitosan formulation; and the increased retention of propranolol by the LM-12 formulation compared to the LM-5 formulation may be attributed to greater ‘microviscosity’. 4. Conclusion The present study set out to identify formulations with characteristics that potentially improve the direct transport of drug from the nasal cavity to the CNS. Candidate polymers were characterised in terms of relevant physiochemical properties. The low esterified pectins and chitosan were chosen mainly due to their known bioadhesiveness. The pectins were also able to gel in the nasal cavity and chitosan had the ability to open tight junctions. The polymers in solution (N 0.4% chitosan and N1% pectin LM-5 and N 0.8% LM-12) were found to display shear thinning properties and the concentrations tested showed apparent viscosities in the range suitable for nasal application in a simple nasal spray device. Each of the polymer-drug formulations provided slower rates of release for a model zwitterionic drug GR138950, compared to a control solution. This suggests that these formulations can be exploited as nasal controlled release vehicles. Slow release of the drug is suggested to be mainly due to ionic interaction between the polymers and the

234

S.T. Charlton et al. / Journal of Controlled Release 118 (2007) 225–234

zwitterionic drug rather than a diffusional barrier caused by a polymer network. MDCK1 cell monolayers were used for drug transport studies to evaluate the effects of the polymers on a confluent cell membrane and the resultant transport of two model drugs, mannitol and propranolol. Chitosan significantly decreased TEER values, indicating an opening of tight junctions between the cells. A much smaller effect was seen for the pectins. These changes in TEER were reflected in a greatly increased transport of mannitol with the chitosan formulation and a minimal increase with the pectin formulations. All three polymer formulations exhibited reduced transport of the lipophilic (positively charged) drug propranolol, which is most likely due to drug retention by the polymer formulations, especially for the pectin systems. The studies show that it is possible to produce simple polymer formulations that have the potential for increased retention at a drug absorption site in the nasal cavity (olfactory region) together with the ability to provide sustained release of drugs (and for chitosan also improvement in membrane transport). Specific pectin and chitosan formulations have been selected for further evaluation in animal models and in man. Acknowledgements The authors would like to thank GlaxoSmithKline, Stevenage, Hertfordshire, UK for the provision of a case award to S. Charlton and for help and advice with the work. References [1] L. Illum, Is nose-to-brain transport of drugs in man a reality? J. Pharm. Pharmacol. 56 (2004) 3–17. [2] R.G. Thorne, G.J. Pronk, V. Padmanabhan, W.H. Frey II, Delivery of insulin-like growth factor-I to the rat brain and spinal cord along olfactory and trigeminal pathways following intranasal administration, Neuroscience 127 (2004) 481–496. [3] L. Illum, Nasal drug delivery — possibilities, problems and solutions, J. Control. Release 87 (2003) 187–198. [4] L. Illum, Improved delivery of drugs to mucosal services, European patent EP0975367B1, 2000. [5] R.J. Soane, M. Frier, A.C. Perkins, N.S. Jones, S.S. Davis, L. Illum, Evaluation of clearance characteristics of bioadhesive systems in humans, Int. J. Pharm. 78 (1999) 55–65. [6] J.M. Smith, M. Dornish, E.J. Wood, Involvement of protein kinase C in chitosan glutamate-mediated tight junction disruption, Biomaterials 26 (2004) 3269–3276. [7] M. Fee, N. Errington, K. Jumel, L. Illum, A. Smith, S.E. Harding, Correlation of SEC/MALLS with ultracentrifuge and viscometric data for chitosans, Eur. Biophys. J. 32 (2003) 457–464.

[8] H. Cölfen, S.E. Harding, MSTARA and MSTARAI: interactive PC algorithms for simple, model independent evaluation of sedimentation equilibrium data, Eur. J. Biophys. 25 (1997) 333–346. [9] G.M. Eccleston, N.E. Hudson, The use of capillary rheometer to determine the shear and extensional flow behavior of nasal spray suspensions, J. Pharm. Pharmacol. 52 (2000) 1223–1232. [10] R. J. Soane, Bioadhesive polymers as intranasal drug delivery systems for peptide and protein drugs, University of Nottingham, Thesis 1999. [11] M.E. Taub, L. Kristensen, S. Frøkjær, Optimised conditions for MDCK permeability and turbidimetric solubility studies using compounds representative of BCS classes I–IV, Eur. J. Pharm. Sci. 15 (2002) 331–340. [12] C. Boehm, M. Gilbert, B. Geny, M.R. Popoff, P. Rodriguez, Modification of epithelial cell barrier permeability and intercellular junctions by Clostridium sordelli lethal toxins, Cell. Microbiol. 8 (2006) 1070–1085. [13] A.T. Hotchkiss Jr., B.J. Savary, R.G. Cameron, H.K. Chau, J. Brouiette, G.A. Luzio, M.L. Fishman, Enzymatic modification of pectin to increase its calcium sensitivity while preserving its molecular weight, J. Agric. Food Chem. 50 (2002) 2931–2937. [14] C. Löfgren, P. Walkenstrom, A.M. Hermansson, Microstructure and rheological behavior of pure and mixed pectin gels, Biomacromolecules 3 (2002) 1144–1153. [15] G.M. Eccleston, M. Bakhshaee, N.E. Hudson, D.H. Richards, Rheological behavior of nasal sprays in shear and extension, Drug Dev. Ind. Pharm. 26 (2000) 975–983. [16] L. Illum, Nasal clearance in health and disease, J. Aerosol Med. 19 (2006) 92–99. [17] O. Smidsröd, A. Haug, Estimation of the relative stiffness of the molecular chain in polyelectrolytes from measurements of viscosity at different ionic strengths, Biopolymers 10 (1971) 1213–1227. [18] R. Dua, H. Zia, T. Needham, The influence of tonicity and viscosity on the intranasal absorption of salmon calcitonin in rabbits, Int. J. Pharm. 147 (1997) 233–242. [19] C. Alvarez-Lorenzo, J.L. Gomez-Amoza, R. Martinez-Pacheco, C. Souto, A. Concheiro, Microviscosity of hydroxypropylcellulose gels as a basis for prediction of drug diffusion rates, Int. J. Pharm. 180 (1999) 91–103. [20] K.G. Nelson, A.C. Shah, Mass-transport in dissolution kinetics. 1. Convective diffusion to assess the role of fluid viscosity under forced flow conditions, J. Pharm. Sci. 76 (1987) 799–802. [21] A.C. Shah, K.G. Nelson, Mass-transport in dissolution kinetics. 2. Convective diffusion to assess role of viscosity under conditions of gravitational flow, J. Pharm. Sci. 76 (1987) 910–913. [22] Y. Murata, M. Miyashita, K. Kofuji, E. Miyamoto, S. Kawashima, Drug release properties of a gel bead prepared with pectin and hydrolysate, J. Control. Release 95 (2004) 61–66. [23] P. Artursson, T. Lindmark, S.S. Davis, L. Illum, Effect of chitosan on the permeability of monolayers of intestinal epithelial cells (Caco-2), Pharm. Res. 11 (1994) 1358–1361. [24] V. Dodane, M.A. Khan, J.R. Merwin, Effect of chitosan on epithelial permeability and structure, Int. J. Pharm. 182 (1999) 21–32. [25] A. Yamamoto, T. Morita, M. Hashida, H. Sezaki, Effect of absorption promoters on the nasal absorption of drugs with various molecular weights, Int. J. Pharm. 93 (1993) 91–99.