Formulation pH Modulates the Interaction of Insulin with Chitosan Nanoparticles

ZENGSHUAN MA,1 HOCK HIN YEOH,2 LEE-YONG LIM1

1 2

Department of Pharmacy, National University of Singapore, 10 Kent Ridge Crescent, Singapore 119260 Department of Biological Sciences, National University of Singapore, 10 Kent Ridge Crescent, Singapore 119260

Received 30 October 2001; accepted 1 February 2002

ABSTRACT: Previous studies on chitosan-insulin nanoparticles have reported diverse encapsulation efciency and insulin release proles despite similar formulation and preparation procedures. This study examined the efciency and mechanism of association of insulin with chitosan nanoparticles in the pH range of 2.3 to 6.3. Nanoparticles of 237 to 253 nm were prepared by ionotropic gelation of chitosan with tripolyphosphate counterions. Insulin was quantied by an RP-HPLC method. The insulin association efciency (AE) spanned a broad range from 2 to 85%, and was highly sensitive to formulation pH. Highest AE was measured at insulin loading concentrations  4.28 U/mL and pH 6.1, close to the pI of native insulin and the pKa of chitosan. This association, attributed to physical adsorption of insulin through hydrophobic interactions with chitosan, was labile, and the associated insulin rapidly and completely released by dilution of the nanoparticles in aqueous media of pH 2 to 7.4. AE obtained at pH 5.3 was less than half that measured at pH 6.1 at corresponding insulin concentration, but the association at pH 5.3 appeared to be based on stronger interactions, because the release of insulin was pH-dependent and recovery was less than 25% even upon disintegration of the chitosan matrix. Interaction of insulin with the chitosan nanoparticles rendered the protein more susceptible to acid and enzymatic hydrolyses, the effects being more predominant in nanoparticles prepared at pH 5.3 than at pH 6.1. 2002 Wiley-Liss, Inc.
and the American Pharmaceutical Association J Pharm Sci 91:13961404, 2002

Keywords:

chitosan; insulin; nanoparticles; pH; association efciency; recovery

INTRODUCTION
Chitosan is a polysaccharide derived from chitin by alkaline deacetylation. Generally regarded as biocompatible, biodegradable, and nontoxic, chitosan is an interesting biomaterial because of its ability to mediate the permeation of molecules across absorptive epithelia. It has been reported to enhance the nasal absorption of insulin in rabbits1 and sheep,2 and the mucosal permeation of other drug molecules.3 The mechanism has

Correspondence to: Lee-Yong Lim (Telephone: 65-8746137; Fax: 65-7791554; E-mail: phalimly@nus.edu.sg)
Journal of Pharmaceutical Sciences, Vol. 91, 13961404 (2002) 2002 Wiley-Liss, Inc. and the American Pharmaceutical Association

been ascribed to a combination of mucoadhesion and widening of the paracellular transport pathway.4,5 An added advantage is the lack of mucosal damage associated with chitosan,2 unlike more conventional permeation enhancers such as the fatty acids and bile salts.6 As a drug carrier, chitosan has been formulated into different pharmaceutical dosage forms such as tablets,7 beads,8 microspheres,9 and nanoparticles.1 Of these, the nanoparticles were regarded as promising carriers in breaching the epithelial and enzymatic barriers to enhance the bioavailability of biomolecules. Chitosan nanoparticles were rst prepared in 1997 by Alonso et al.10 and the same group later used chitosaninsulin nanoparticles for the nasal delivery of insulin in rabbits.1 We were interested in using chitosan

1396

JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 91, NO. 6, JUNE 2002

INTERACTION OF INSULIN WITH CHITOSAN NANOPARTICLES

1397

nanoparticles to enhance the intestinal permeation and stability of insulin delivered via the oral route. However, we were perplexed to nd that, despite seemingly similar formulation and preparation procedures, a diversity of characteristics were reported of the chitosaninsulin nanoparticles, particularly with regard to the encapsulation efciency (EE) and the in vitro release prole of insulin. In one study,1 impressive EE values that ranged from 87.4 to 96.7% were reported, although the encapsulated insulin was completely released within 30 min into phosphate buffer (pH 7.4). In a follow-up study by the same group,11 equivalent chitosaninsulin nanoparticles were reported to have signicantly lower EE (< 60%), with not more than 40% of the encapsulated insulin released in 2 h into pH 7.4 phosphate buffer. Such discrepancies have serious implications on batch reproducibility and bioactivity, and should be addressed. Among the formulation variables that could account for the diversity in chitosaninsulin nanoparticle characteristics, pH appears to be the most important. pH is known to inuence the amount of protein incorporated into polymer nanoparticles,10,12,13 and to affect chitosan conformation14 and its reaction with the tripolyphosphate (TPP) anions15 used in the preparation of the chitosan nanoparticles. pH also has signicant effects on the chemical and conformational stability of insulin,16 and would govern its interaction with TPP and chitosan. The aim of this study was to evaluate the effect of formulation pH on the characteristics of chitosaninsulin nanoparticles and to elucidate the mechanism of interaction between insulin and the chitosan nanoparticles. The longer term goal is to develop an optimized formulation that will efciently deliver insulin by the oral route of administration. The chitosaninsulin nanoparticles were prepared by the ionotropic gelation method at carefully monitored pH.

Chemical Co.), chitosanase-RD (from Bacillus SP. PI-7S, 0.150.35 U/mg, Pias Corporation, Osaka, Japan) and pentasodium tripolyphosphate (TPP) (Merck, Darmstadt, Germany) were used as received. All organic solvents were of HPLC grade, and all other chemicals were of the highest grade commercially available. Preparation of ChitosanInsulin Nanoparticles Chitosaninsulin nanoparticles were prepared by methods adapted from that reported by Alonso et al.1,10 Chitosan (0.2% w/v) was dissolved in aqueous acetic acid solutions (0.25 to 10% v/v) while TPP (0.1% w/v) was dissolved in deionized water or in dilute sodium hydroxide solutions (0.01 to 0.09 M). Insulin was dissolved in 0.01 M HCl at various concentrations. Insulin solution (1 mL) was premixed with 4 mL of TPP solution or 8 mL of chitosan solution before the addition of the TPP solution dropwise into the chitosan solution under magnetic stirring (1000 rpm, Corning Stirrer, PC-420, USA) at ambient temperature. The nal pH of the nanoparticle suspension was measured (PHM61 Laboratory pH Meter, Copenhagen, Denmark), and the nanoparticles characterized immediately. All experiments were performed in triplicates. Characterization of the Nanoparticles Particle size and average count number (ACN) of each batch of nanoparticles were measured using a 90Plus Particle Size Analyzer (PSA) (Brookhaven Instruments Corporation, NY) equipped with a laser light source (677 nm) operating at 908 to the detector. Samples were diluted (1:3 v/v) with ltered (0.45 mm) deionized water prior to measurements. The ACN reects the signal intensity, which is a measure of nanoparticle concentration in a sample. Nanoparticle morphology was observed under a scanning electron microscope (JEOL JSM5600LV, 7 kV). A drop of nanoparticle suspension placed onto a piece of conductive paper mounted with adhesive on a cuprum stud was air dried overnight, coated with platinum (JEOL JFC-1300 Auto Fine Coater), and observed under the SEM at magnication 17,000 . To determine the association efciency (AE) and loading efciency (LE), triplicate batches of nanoparticles were pelletized at 74,200 g (AvantiTM J-25 centrifuge, Beckman, USA), 208C for 30 min, and the insulin content in the supernatant
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 91, NO. 6, JUNE 2002

MATERIALS AND METHODS

Materials Chitosan (CS) (MW 1.86 ( 0.16) 105, degree of deacetylation 84.71 0.59%, Aldrich Chemical Co., Milwaukee, WI), insulin (from porcine pancreas, 1 mg was equivalent to 27.8 USP units, Sigma Chemical Co., St. Louis, MO), lysozyme (from chicken egg white, 48,800 U/mg, Sigma

1398

MA, YEOH, AND LIM

assayed by RP-HPLC. The pellet was lyophilized and weighed. The AE and LE were calculated by the respective formulae: Total amount of insulin Insulin in supernatant 100% AE Total amount of insulin Total amount of insulin Insulin in supernatant LE 100% Weight of recovered particles

At predetermined time intervals, triplicate samples were pelletized at 74,200 g at 208C for 10 min, and the supernatant assayed for insulin. The insulin content was expressed as a percentage of the insulin associated with the nanoparticles as calculated from the AE. Statistical Analyses Data are presented as mean standard deviation, and analyzed by one-way ANOVA with the Tukey test applied post hoc for paired comparisons (SPSS 10). Values of p < 0.05 were indicative of signicant differences. Regression analysis was performed to determine the correlation coefcient between two variates.

Quantitative Analysis of Insulin Insulin was assayed by reversed-phase liquid chromatography (RP-HPLC)17 using a Shimadzu HPLC system (LC-10ATVP, Kyoto, Japan) equipped with UV detector and a Waters Spherisorb column (ODS1, S10, 4.6 250 mm, Waters Corporation, Milford, MA). Samples of 20 mL were eluted with a mobile phase comprising acetonitrile and phosphate buffer (0.1 M monobasic sodium phosphate adjusted to pH 2.0 with phosphoric acid) at a linear eluting gradient of 26 to 32% v/v acetonitrile in 30 min. The ow rate was 1.5 mL/min and the detection wavelength was 214 nm. Insulin was quantied by peak area measurement. In Vitro Release of Insulin Nanoparticles were pelletized at 25,000 g at 208C for 30 min, and each weighed pellet was reconstituted with 5 mL of dissolution medium at 378C under agitation (100 rpm, Orbit shaker bath, Lab-line, Illinois). The dissolution media were phosphate buffers of pH 7.4, 6.8, and 5.5, and 0.1 M HCl solution. At predetermined incubation time, triplicate samples were pelletized by centrifugation (74,200 g 30 min, 208C) and the supernatants assayed for insulin. The amount of insulin released was expressed as a percentage of the total insulin associated with the nanoparticles as calculated from the AE. The in vitro release of insulin was also evaluated in the presence of enzymes that digest the chitosan matrix. The digesting enzyme solution (DES)18 was prepared by mixing solutions of chitosanase-RD (0.150.35 U/mL) and lysozyme (5,000 U/mL) in 50 mM acetate buffer (pH 5.5) in the ratio 4:1 v/v. Nanoparticle dispersion (1 mL) was pelletized at 74,200 g at 208C for 30 min, washed thrice with distilled water, and incubated with 1 mL of DES at 378C and agitated at 100 rpm.
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 91, NO. 6, JUNE 2002

RESULTS
Chitosaninsulin nanoparticles were prepared in the pH range of 2.3 to 6.3, achieved by careful control of the acidity and alkalinity of the chitosan and TPP solutions, respectively. Nanoparticle formation was effected by mixing the chitosan and TPP solutions. Insulin was premixed with either the TPP solution or the chitosan solution prior to nanoparticle formation, and the resultant nanoparticles are designated as TPP-np and CS-np, respectively. The initial insulin loading concentration was 0.88 to 6.54 U/mL. Increasing the pH in the nal formulation from 2.3 to 6.1 at an insulin loading concentration of 0.88 U/mL did not signicantly enlarge the size of the TPP-np (237 53 nm to 325 45 nm, p 0.12) but the ACN increased eightfold from 24 to 190 kilo count per second (Fig. 1). A linear relationship was observed between pH and ACN (R 0.85, p < 0.01), indicating an increase in the number of nanoparticles formed at higher pH. Mirror trends were seen with the CS-np, and there were no signicant differences in the size or yield between the CS-np and TPP-np prepared under similar conditions (Fig. 1). Under the SEM, the CS-np and TPP-np were indistinguishable, appearing as solid spheroids (Fig. 2). Extension of the formulation pH beyond 6.1 caused the mean particle size to rise sharply to 1100 nm at pH 6.3, suggesting particle agglomeration, while gross precipitation was noted at pH 6.5. Relatively low but strongly pH-dependent AE values were attained at insulin loading concentration of 0.88 U/mL (Fig. 3). Maximum AE of 62.99 1.67% and 57.67 3.24% were observed at

INTERACTION OF INSULIN WITH CHITOSAN NANOPARTICLES

1399

Figure 1. Effects of pH on the mean particle size (A) and average count number (B) of chitosaninsulin nanoparticles prepared with insulin loading concentration of 0.88 U/mL (mean SD, n 3).

pH 6.1 for the TPP-np and CS-np, respectively. The AE fell sharply with small pH changes in the vicinity of 6.1, with values of 13.27 4.91% and 12.67 4.71% measured at pH 5.7 and 6.3, respectively, for the TPP-np. AE values were less varied in the more acidic pH range, although a signicantly higher AE (18.99 1.16% for TPP-np) was determined at pH 5.3, while a minimum AE (7.09 1.03%) was obtained at pH 3.3. The TPPnp had slightly higher AE than corresponding CS-np prepared at the same pH, and the TPPmethodology was adopted for subsequent batches of nanoparticles. AE was signicantly improved by using higher insulin loading concentrations (Fig. 4). However, a linear relationship between AE and insulin concentration was not observed at any specied pH and the maximum AE did not always occur at the highest loading concentration. Regardless of the insulin loading concentration, optimal AE was achieved at pH 6.1. Mean particle size and polydispersity of the nanoparticles were not signicantly affected by the changes in the insulin loading concentrations. For example, the mean

Figure 2. SEM micrographs of (A) TPP-np and (B) CS-np prepared at pH 5.3 with insulin loading concentration of 4.28 U/mL. Nanoparticles were prepared by ionotropic gelation of chitosan with tripolyphosphate anions; TPP-np were obtained by premixing insulin with tripolyphosphate solution, and CS-np by premixing insulin with chitosan solution, prior to nanoparticle formation.

particle size of nanoparticles prepared at pH 5.3 with insulin loading concentrations of 0 to 6.54 U/mL ranged from 243 15 to 271 7 nm, while the polydispersity values were between 0.50 0.01 to 0.57 0.07. The LE, which measured the amount of insulin associated with unit weight of nanoparticle, was less sensitive to formulation pH (Fig. 3). Except at pH 6.1 and 6.3, there was little variation in LE with pH changes, which could be attributed to the larger nanoparticle yield normalizing the more efcient insulin association, because both ACN and AE increased with pH in the pH range of 3.3 to 5.3. At pH 6.1, the AE was sufciently high to
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 91, NO. 6, JUNE 2002

1400

MA, YEOH, AND LIM

Figure 4. Association efciency (AE) of chitosan insulin nanoparticles as a function of formulation pH and insulin loading concentration (mean SD, n 3).

Figure 3. The association efciency (AE) and loading efciency (LE) obtained as a function of pH for chitosaninsulin nanoparticles prepared with insulin loading concentration of 0.88 U/mL (mean SD, n 3).

raise the LE by twofold to 4.23%, while a sharp fall in AE coupled with increasing nanoparticle yield resulted in a very low LE of 0.26% for the nanoparticles at pH 6.3. In vitro release of insulin into various aqueous media was evaluated of TPP-np formulated with an insulin loading concentration of 4.28 U/mL at pH 5.3 and 6.1. To separate the nanoparticles from the free insulin without destroying nanoparticle integrity, the nanoparticles were pelletized at 25,000 g, lower than the force of 74,200 g required to effectively spin down the smallest nanoparticles. Processing at the stronger force resulted in a pellet that could not be redispersed subsequently into nanoparticles in an aqueous medium. The release proles are shown in Figure 5. For ease of discussion, nanoparticles formulated at pH 5.3 and 6.1 are respectively denoted as F5.3np and F6.1np. The release proles of F5.3np showed a burst effect within the rst 30 min, followed by little release of insulin thereafter. The cumulative amounts of insulin released after 6 h in 0.1 M HCl and pH 7.4 phosphate buffer solutions were
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 91, NO. 6, JUNE 2002

similar, and amounted to 20% of the associated insulin, calculated from the AE (Fig. 5A). Lower amounts, equivalent to 11.41 1.8 and 7.43 1.25% of associated insulin, respectively, were obtained over the same time interval in the pH 6.8 and 5.5 phosphate buffer solutions. HPLC analyses of the 0.1 M HCl samples showed additional peaks at retention times lower than that of insulin (Fig. 6A), indicating the presence of degradation products whose content increased with incubation time. Measurement of particle size suggested that the nanoparticles disintegrated in this medium. Both phenomena were not observed in any of the phosphate buffer solutions. Incubation of insulin (4.4 U/mL) for 24 h in 0.1 M HCl at 378C also did not yield any degradation product detectable by HPLC, although prolonged incubation caused the protein to degrade by 7, 20, and 23% after 2, 4, and 5 days of incubation, respectively. Deterioration of insulin in acid solutions has been attributed to deamidation at residue AsnA21, and is reportedly not dependent on the strength of preparation.19 The in vitro insulin release proles of the F6.1np were radically different, with  95% of the associated insulin released in 6 h regardless of the composition of the dissolution medium (Fig. 5B). The nanoparticles had identical release proles in 0.1 M HCl, pH 6.8 and pH 7.4 phosphate buffer solutions, releasing > 83% of the insulin load within the rst 30 min. Although a burst effect was also observed in the pH 5.5 phosphate buffer solution, the initial rate of release was slower and a signicantly lower 63.41% of associated insulin was released in the rst 30 min. No extraneous peak was detected by the HPLC

INTERACTION OF INSULIN WITH CHITOSAN NANOPARTICLES

1401

Figure 5. In vitro release proles of insulin from chitosaninsulin nanoparticles into various aqueous dissolution media at 378C. Nanoparticles were prepared with insulin loading concentration of 4.28 U/ mL at pH 5.3 (F5.3np, A) and pH 6.1 (F6.1np, B) (mean SD, n 3).

Figure 6. HPLC chromatograms of 0.1 M HCl samples incubated for 6 h with F5.3np (A) and F6.1np (B). The insulin peak appeared at 2627 min while peaks for degradation products appeared at between 5 and 15 min.

following 6 h incubation of the F6.1np in all four dissolution media, including 0.1 M HCl (Fig. 6B). Insulin release proles in the digesting enzyme solution (DES), which contained the chitosandigesting enzymes of chitosanase and lysozyme, were also evaluated. This is to account for encapsulated insulin that might not be released into the dissolution medium due to steric hindrance posed by the chitosan matrix. The DES solvent was 50 mM acetate buffer at pH 5.5, chosen to optimize the activity of the enzymes.18 Figure 7 shows that insulin released from the F6.1np gradually reached a maximum of 64.17 1.56% of the insulin load at 4 h incubation, while the maximum amount released from the F5.3np was a miserly 8.75 1.2% of the insulin load at 1 h. Despite the digestion of the chitosan matrix in the DES, the rate of release of insulin from the

F6.1np in this medium was less than those measured in media without enzymes. This could be attributed to experimental differences. For the DES experiments, the nanoparticles were pelletized at a higher centrifugation force (74,200 g vs. 25,000 g), giving a pellet of nondispersable, agglomerated particles. With its smaller surface area and greater thickness, the pellet would release insulin at a slower rate compared to the redispersed nanoparticles in dissolution media without enzymes. Prolonging the incubation time resulted in lower insulin content in the DES for both the F6.1np and F5.3np (Fig. 7). This decline occurred after 4-h incubation for the F6.1np, but was seen much earlier, after 1 h, for the F5.3np. On the other hand, free insulin at 4.12 U/mL showed a decline of 7% only after 6 h incubation in the DES.
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 91, NO. 6, JUNE 2002

1402

MA, YEOH, AND LIM

Figure 7. In vitro release proles of insulin from chitosaninsulin nanoparticles into the digesting enzyme solution (DES) at 378C. Nanoparticles were prepared with insulin loading concentration of 4.28 U/mL at pH 5.3 (F5.3np) and pH 6.1 (F6.1np) (mean SD, n 3).

This suggests that insulin degradation in the DES accounted for the decline in insulin content with time seen in Figure 7, and probably also for the lower recovery of associated insulin in the DES compared to the other dissolution media.

DISCUSSION
The chitosaninsulin nanoparticles were produced by ionotropic gelation with the counterion, TPP. Chitosan, which has an apparent pKa of 6.3, assumes a stiff extended conformation in aqueous media of low pH due to the charge repulsion of highly protonated amino groups, but it becomes increasingly globular as pH increases, and can precipitate at pH 6.6 or lower, depending on the ionic strength of the solution.14 In this study, precipitation was apparent at pH  6.3, while the particle yield was enhanced as pH approached 6.3. The larger particles seen at higher pH could be attributed to less extensive chitosanTPP interactions, the complex assuming a loop conformation in contrast to the compact ladder conformation at low pH.15 The high AE obtained at pH 6.1 must be underlined by favorable interactions between insulin and the chitosan nanoparticles. Proteins can adsorb very efciently (90100%) onto polymers at pH around the isoelectric point (pI) of the proteins20 because of the minimization of electrostatic repulsion, increased conformational stability, coupled with smaller specic surface
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 91, NO. 6, JUNE 2002

area of the protein molecules.12,13 Insulin, a 51-amino acid polypeptide of MW 5800, has a pI of 5.3 in the denatured state, but because of its ability to associate under native conditions, insulin can exhibit an apparent pI of 6.4, presumably due to a masked carboxylate ionization.21 At pH 6.1, therefore, the formulation could contain zwitterionic insulin molecules and chitosan molecules in the globular state, both of which favored hydrophobic association. The association between insulin and chitosan could be driven by hydrophobic interactions and H-bonding, the same forces that caused zinc-free insulin in aqueous media to form noncovalent dimers and tetramers at near neutral pH.22 Such association was relatively labile, and the equlibrium rapidly shifted towards dissociation with small pH changes or by dilution in a wide variety of aqueous media, as evidenced from the high rate and extent of release of associated insulin from the F6.1np in the various dissolution media. Efcient adsorption was achieved with high insulin loading concentrations within a narrow pH range. The associated insulin was evidently not encapsulated in the nanoparticle core because of the high recovery rate of  95% within 6 h in aqueous media, and it was more susceptible than free insulin to degradation in the DES. Interaction of insulin and the chitosan nanoparticles at pH  6.1 is somewhat more complicated. Insulin was less efciently associated to the nanoparticles, which might be related to both insulin and chitosan having net positive charges at pH below 6.1. This association was less sensitive to pH changes in the acid range, because similar LE values were obtained from pH 2.3 to 5.3, but it appeared to involve stronger forces of interaction that were not readily reversible by dilution. Acid pH, combined with agitation, is known to facilitate the dissociation of insulin tetramers/dimers into monomers, and to further cause the monomer to adopt a partially unfolded intermediate conformation.22,23 Acetic acid (8 M) at ambient temperature also increased the propensity of insulin to exist in this monomeric form.24 Although the preparation of the nanoparticles used more diluted acetic acid solutions (< 2 M), the high ionic strength combined with rapid agitation of 1000 rpm might still cause the formation of the partially unfolded insulin intermediate.22,24 This intermediate is driven by hydrophobic interactions to self-aggregate into a critical nucleus and eventually into brils.2224 Fibrillation is promoted in the presence of anions,

INTERACTION OF INSULIN WITH CHITOSAN NANOPARTICLES

1403

which shield the positive charges on the insulin molecule.24 However, the polyanionic TPP was more likely to provide an alternative pathway to brillation for the unfolded insulin intermediate. Capable of attracting several cationic species, the TPP could function as a crosslink that bind the positively charged chitosan and insulin intermediate into an electrostatically stabilized chitosanTPPinsulin complex. Such complex formation has been reported to occur between the insulin intermediate and the negatively charged 1-anilinonaphthalene-8-sulfonic acid.22 Although insulin brillation results from noncovalent interaction of several insulin molecules, it involves partially denatured molecules, and is normally not reversible without dramatic changes in the solvent environment, unlike the association of normal globular insulin into dimers and tetramers.23 The involvement of specic electrostatic and hydrophobic interactions in bril formation25 could have accounted for its irreversibility. In this study, the low recovery of insulin in the F5.3np reected the degree of irreversibility of the interaction in aqueous media. Not more than 25% of intact insulin was detected even with the disintegration of the chitosan matrix in 0.1 M HCl and DES, although the low recovery in these two media could also be attributed to insulin degradation. Insulin in the F5.3np was more susceptible to hydrolysis in 0.1 M HCl and DES than free insulin and insulin associated with F6.1np. In phosphate buffer media that did not dissolve chitosan nor degrade the insulin, recovery of insulin from the F5.3np was equally poor but showed pH dependence, the insulin dissociating in increasing amounts when pH of the medium was raised from 5.5 to 6.8 to 7.4. This would correspond to the transformation of the insulin molecule from having a net positive charge to being a zwitterion to having a net negative charge. Although drug loading could inuence the driving force (concentration gradient) for drug dissolution and diffusion from a polymer matrix, it is not likely to be a predominant factor in determining the in vitro release proles of insulin from the F5.3np and F6.1np. This is because the loading efciencies of insulin in the F5.3np and F6.1np were too low, at 1.9 and 4.2%, respectively, to account for the steep differences in the extent of insulin released from the nanoparticles. Results from this study support the hypothesis that failure to control formulation pH resulted in the diverse reported characteristics of chitosan insulin nanoparticles prepared by ionotropic

gelation.1,11 AE and insulin release prole (in pH 7.4 phosphate buffer) of the F6.1np were very similar to data presented in 1999 by Alonso et al.1 On the other hand, the F5.3np had properties that correlated more closely to the chitosaninsulin nanoparticles reported a year later by the same group.11 The complex F5.3np product remains to be conrmed structurally. NMR, IR, and circular dichroism analyses failed because of the common amino functionality in the insulin and chitosan molecules, while chitosan produced a broad DSC peak spanning several tens of degrees that mask any evidence of a new product (results not shown).

CONCLUSION
Chitosan nanoparticles (237 to 325 nm) with insulin association efciency that varied from 2 to 85% could be prepared by the iontropic gelation method with careful control of the formulation pH from 2.3 to 6.3. The insulin association was very sensitive to formulation pH, and was also dependent on insulin loading concentration. High association ( 80%), attributed to physical adsorption predominated by hydrophobic interactions with chitosan, occurred at insulin loading concentrations  4.28 U/mL and pH 6.1, close to the pI of native insulin and pKa of chitosan. This association was labile, and the insulin rapidly and completely dissociated with dilution of the nanoparticles in aqueous media. Efciency of insulin association at pH 5.3 was 50% or lower compared to that at pH 6.1, but the association at pH 5.3 was based on stronger chargecharge interactions, because the release of insulin was pH-dependent and recovery was less than 25% even upon dissolution of the chitosan matrix. Interaction of insulin with the chitosan nanoparticles was accompanied by unstable conformational transitions that rendered the protein more susceptible to degradation in 0.1 M HCl and enzymatic hydrolyses by chitosanase/lysozyme, the effects being more predominant at pH 5.3 than at pH 6.1. Although the structural conrmation of the chitosaninsulinTPP product remains to be elucidated, this work demonstrates the importance of controlling the formulation pH when manufacturing chitosaninsulin nanoparticles. It also suggests different modes of interaction between insulin and chitosan nanoparticles at pH 5.3 and 6.1, which could impact on the in vivo efcacy of the nanoparticles.
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 91, NO. 6, JUNE 2002

1404

MA, YEOH, AND LIM

ACKNOWLEDGMENTS
This study is supported by a National University of Singapore grant (R148-000-023-112). Zengshuan Ma is grateful to the National University of Singapore for nancial support of his graduate studies.
13.

14. 15.

REFERENCES
1. Fernandez-Urrusuno R, Calvo P, Remunan-Lopez C, Vila-Jato JL, Alonso MJ. 1999. Enhancement of nasal absorption of insulin using chitosan nanoparticles. Pharm Res 16:15761581. 2. Illum L, Farraj NF, Davis SS. 1994. Chitosan as a novel nasal delivery system for peptide drugs. Pharm Res 11:11861190. 3. Kotez AF, de Leeuw BJ, Lueben HL, deBoer AG, Verhoef JC, Junginger HE. 1997. Chitosan for enhanced delivery of therapeutic peptides across intestinal epithelia: In vitro evaluation in Caco-2 cell monolayers. Int J Pharm 159:243253. 4. Schipper NGM, Olsson S, Hoogstraate JA, deBoer AG, Varum KM, Artusson P. 1997. Chitosan as absorption enhancers for poorly absorbable drugs 2: Mechanism of absorption enhancement. Pharm Res 14:923929. 5. Artursson P, Lindmark T, Davis SS, Illum L. 1994. Effect of chitosan on permeability of monolayers of intestinal epithelial cells (Caco-2). Pharm Res 11: 13581361. 6. Carino GP, Mathiowitz E. 1999. Oral insulin delivery. Adv Drug Deliv Rev 35:249257. 7. Upadrashta SM, Katikaneni PR, Nuessle NO. 1992. Chitosan as a tablet binder. Drug Dev Ind Pharm 18:17011708. 8. Chandy T, Sharma CP. 1992. Chitosan beads and granules for oral sustained delivery of nifedipine: In vitro studies. Biomaterials 13:949952. 9. Lim LY, Wan LSC. 1998. Effect of magnesium stearate on chitosan microspheres prepared by an emulsication-coacervation technique. J Microencapsul 15:319333. 10. Calvo P, Remunan-Lopez C, Vila-Jato JL, Alonso MJ. 1997. Novel hydrophilic chitosanpolyethylene oxide nanoparticles as protein carriers. J Appl Polym Sci 63:125132. 11. Janes KA, Alonso MJ. 2000. Effect of polymer molecular weight on chitosan nanoparticles for peptide delivery. Proc Int Symp Control Rel Bioact Mater 27:8043. 12. Elgersma AV, Zsom RLJ, Norde W, Lyklema J. 1990. The adsorption of bovine serum albumin on

16.

17.

18.

19.

20.

21.

22.

23.

24.

25.

positively and negatively charged polystyrene lattices. J Colloid Int Sci 138:145156. Barichello JM, Moriko M, Takayama K, Nagai T. 1999. Encapsulation of hydrophilic and lipophilic drugs in PLGA nanoparticles by the nanoprecipitation method. Drug Dev Ind Pharm 25:471476. Roberts GAF. 1992. Chitin chemistry. London: MacMillan Press, pp 203206. Mi FL, Shyu SS, Lee ST, Wong TB. 1999. Kinetic study of chitosan-tripolyphosphate complex and acid resistive properties fo the chitosan-tripolyphosphate gel beads prepared by in-liquid curing method. J Polym Sci Polym Phys 37:15511564. Brange J, Langkjr L. 1992. Chemical stability of insulin. 3. Inuence of excipients, formulation and pH. Acta Pharm Nord 4:149158. Salem II, Bedmar MC, Medina MM, Cerezo A. 1993. Insulin evaluation in pharmaceutics: Variables in RP-HPLC and method validation. J Liq Chromatogr 16:11831194. Mao HQ, Roy K, Troung-Le VL, Janes KA, Lin KY, Wang Y, August JT, Leong KW. 2001. Chitosan DNA nanoparticles as gene carriers: Synthesis, characterization and transfection efciency. J Control Rel 70:399421. Brange J, Langkir L, Havelund S, Vlund A. 1992. Chemical stability of insulin. 1. Hydrolytic degradation during storage of pharmaceutical preparations. Pharm Res 9:715726. Muzzarelli RA, Barontini G, Rocchetti R. 1976. Immobilized enzymes on chitosan columns: Alphachymotrypsin and acid phosphatase. Biotech Bioeng 18:14451454. Kaarsholm NC, Havelund S, Hougaard P. 1990. Ionization behavior of native and mutant insulins: pK perturbation of B13-Glu in aggregated species. Arch Biochem Biophy 283:496502. Nielsen L, Khurana R, Coats A, Frokjaer S, Brange J, Vyas S, Uversky V, Fink AL. 2001. Effect of environmental factors on the kinetics of insulin bril formation: Elucidation of the molecular mechanism. Biochemistry 40:60366046. Brange J, Andersen L, Laursen ED, Meyn G, Rasmussen E. 1997. Toward understanding insulin brillation. J Pharm Sci 86:517525. Nielsen L, Frokjaer S, Carpenter JF, Brange J. 2001. Studies of the structure of insulin brils by Fourier Transform Infrared (FTIR) spectroscopy and electron microscope. J Pharm Sci 90:29 37. Nielsen L, Frokjaer S, Brange J, Uversky VN, Fink AL. 2001. Probing the mechanism of insulin bril formation with insulin mutants. Biochemistry 40:83978409.

JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 91, NO. 6, JUNE 2002