New loading process and release properties of insulin from polysaccharide microcapsules fabricated through layer-by-layer assembly

Journal of Controlled Release 112 (2006) 79 – 87 www.elsevier.com/locate/jconrel

New loading process and release properties of insulin from polysaccharide microcapsules fabricated through layer-by-layer assembly Shiqu Ye, Chaoyang Wang, Xinxing Liu, Zhen Tong ⁎, Beye Ren, Fang Zeng Research Institute of Materials Science, South China University of Technology, Guangzhou 510640, China Received 12 April 2005; accepted 20 January 2006 Available online 9 March 2006

Abstract Polysaccharide multilayer microcapsules were fabricated in aqueous media by the layer-by-layer self-assembly of chitosan (CHI) and sodium alginate (ALG) on melamine formaldehyde (MF) microparticles of 2.1 μm diameter as templates, followed by removal of the templates through dissolving at low pH. The loading process was observed with the confocal laser scattering microscope (CLSM) using fluorescence labeled insulin. Insulin was spontaneously loaded into the ALG/CHI microcapsules at pH below its isoelectric point of 5.5 where insulin was positively charged and the loading capacity increased with pH decreasing from 4.0 to 1.0. Because there exited a negatively charged complex of ALG/MF residues inside the microcapsules formed during the MF particle dissolution. A novel two-temperature loading procedure was proposed as loading at 20 °C for the first hour and at a higher loading temperature for the second hour. This procedure was very significant that increasing the second loading temperature from 20 to 60 °C not only increased the insulin loading capacity, but also slowed down its release rate. The release rate of insulin at pH 7.4 was found much faster than that at pH 1.4 due to the positive charges on the insulin. Cross-linking the ALG in the microcapsule shell with calcium ions (Ca2+) or re-sealing the microcapsules with additional layers also remarkably decreased the insulin release rate. The results provide a simple method to control the loading and release of protein molecules within these polysaccharide microcapsules. © 2006 Elsevier B.V. All rights reserved. Keywords: Polysaccharide microcapsule; Insulin loading capacity; Temperature effect

1. Introduction Recently, the layer-by-layer (LbL) self-assembly technique has been applied to micro-encapsulation [1–4]. With the LbL technique, polyelectrolyte multilayer films can be fabricated on various templates through repeated deposition, mainly due to the electrostatic attraction between oppositely charged polyelectrolytes [5]. By removal of the sacrificial core templates, hollow microcapsules with multilayer shell of well defined thickness and composition can be prepared through the deliberate control of the layer number and the polymer species [6,7]. These hollow microcapsules are of both scientific and technological interests because of their potential applications as a new colloidal system in areas such as medicine, drug delivery, micro-reactor and catalysis [8–11]. One of the most important properties of these microcapsules is the selective permeability for different species under tunable ⁎ Corresponding author. Tel./fax: +86 20 87112886. E-mail address: [email protected] (Z. Tong). 0168-3659/$ - see front matter © 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.jconrel.2006.01.015

conditions [8]. The first investigation of hollow microcapsules permeability showed that small polar molecules could readily permeate the multilayer films while macromolecules were excluded [12]. Utilizing this property, monomers or small ions could be loaded into the hollow microcapsules and then macromolecules or ionic substances, such as Fe3O4 and CaCO3 were synthesized inside the microcapsules [4,13–15]. The hollow polyelectrolyte microcapsules were also loaded with macromolecules by adjusting pH value or salt concentration to change the film permeability [16,17]. Another way for microcapsule loading was to employ the mixture of organic solvent and water, which loosened the film structure and encapsulated enzyme into microcapsules [3]. The templates used in formation of hollow microcapsules also influenced the loading process [8]. Many water-soluble substances have been reported to spontaneously cumulate in the interior of polyelectrolyte microcapsules using MF particles as the template, and the driving force is the electrostatic attraction with a negatively charged complex of MF residues in the capsules formed during the dissolving process [18]. Therefore,

80

S. Ye et al. / Journal of Controlled Release 112 (2006) 79–87

the micro-encapsulation method based on the LbL technique and the formation of hollow polyelectrolyte microcapsules has been developed during recent years [19–21]. The release is significant characteristics for the microcapsules loaded different substances. Up to now, most of the investigations concern the release of some loaded water-insoluble dyes or drug microcrystals that were directly coated with polyelectrolytes [22–27]. As for water-soluble substances, Tiourina et al. used the poly(styrenesulfonate)/poly(allylamine) (PSS/PAH) and alginate (ALG)/protamine multilayer microcapsules to load α-chymotrypsin and studied the release rate at pH 1.7 and 8.0 [19,20]. Balabushevich et al. also loaded peroxidase into (dextran sulfate)/protamine microcapsules and found that the release rate became faster at lower pH and slower with additional layers coated on the peroxidase-loaded microcapsules [21]. Recently, Balabushevich et al. investigated the influence of loading temperature, salt concentration and the medium polarity on the catalase loading capacity of dextran sulfate/protamine microcapsules. They found that the loading capacity decreased slightly when either increasing the loading temperature from 20 to 50 °C or adding 1∼2 M NaCl or 30% isopropanol to the catalase bulk solution [28]. In our precious study, the ALG/ chitosan (CHI) sub-microcapsules loaded with acridine hydrochloride were cross-linked and the cumulative release amount from the cross-linked capsules was obviously reduced [29]. Generally, the method to increase the protein loading capacity is to change pH of the protein solution, so that the influence of pH and additional layer on the release also becomes considerably important. Finding some other methods to increase the protein loading capacity and control its release is expected. Here, we intend to investigate the loading and release of water-soluble insulin into the microcapsules fabricated through the LbL self-assembly of natural polysaccharides CHI and ALG. The polysaccharides are chosen for their biodegradability and suitable for medicine applications. Insulin is the most widely used protein drug, and is indispensable for the patients of the type 2 diabetes [30]. The research on insulin delivery and release has attracted many interests, including oral insulin delivery [31]. Insulin is used in this work only as a watersoluble model drug. Two-temperature loading has been used first time inducing a significant change in loading capacity and release rate. The possibility to adjust the release rate by crosslinking the ALG in the capsule shell with calcium ion (Ca2+) has been also observed for the first time. 2. Experimental section 2.1. Materials Insulin (Sigma, MW 6500), chitosan (CHI, Fluka, MW 15,000), and fluorescein isothiocynate (FITC, Amresco USA) were used as received. Sodium alginate (ALG, Kimitsu Chemical Industry Co., Japan) was purified as follows. The aqueous solution of alginate about 5 wt.% was first dialyzed against distilled water using a cellulose tubular membrane (cutoff molecular weight: 14,000 ± 2000) until the conductivity of water outside became constant before and after refreshing.

Then, the solution was filtered and freeze-dried to produce purified dry sample. The molecular weight MW of ALG is 120,000 as determined by gel permeation chromatography (GPC) with a Waters apparatus at 40 °C using 0.1 mol/L of Na2 SO 4 aqueous solution as the elution and narrowly distributed PEO as the standard. Melamine was purchased from Beijing Chemical Plant and used as received. Highly purified water was obtained by deionization and filtration with a Millipore purification apparatus (resistivity higher than 18.2 MΩ cm). Other chemicals were analytical reagents and used as received. 2.2. Preparation of microcapsules Melamine formaldehyde (MF) microparticles with monodisperse diameter were prepared as the template in this work. 2.8g melamine was mixed with 6.0 g formaldehyde under stirring of 200 rpm at 60°C for 20 min to obtain a prepolymer. 0.45g poly(vinyl alcohol) was dissolved in 90 g purified water and the solution pH was adjusted to 4.5 with acetic acid. This solution was kept at 60°C and the prepolymer was subsequently added into the solution under stirring of 300rpm. After 30min, the reaction was stopped by immerging the flask into ice water. The produced MF microparticles were rinsed using the circle of centrifugation (5000 rpm, 5 min)/washing/redisperse in water for three times. The obtained MF microparticles were stored at 5°C, which were stably dispersed for at least two months. Ionic polysaccharides were assembled on the MF microparticles of 2.1μm diameter using the LbL assembly in aqueous solutions. The first ALG layer was deposited with addition of 1mL aqueous ALG solution of 1mg/mL containing 0.5M NaCl into MF suspension. The mixture was incubated for 15min with gently shaking. The excess ALG was removed by two repeated circles of centrifugation (5000rpm, 5min)/washing/re-disperse in water. The following CHI layer was deposited with 1mL CHI solution of 1mg/mL containing 0.3vol.% acetic acid and 0.5M NaCl following the same procedure. Subsequent alternating ALG and CHI layers were deposited in the identical way until the desired number of polyelectrolyte layers was achieved. Hollow microcapsules were prepared by dissolving the core of MF particles with hydrochloric acid solution of pH 1.2 for 10 min and centrifuging at 5000 rpm for 5 min [8]. This dissolving process was repeated two times to ensure the complete removal of the MF template. The change of the suspension from white emulsion to colorless transparent state after the treatment confirmed the removal of the MF cores. Then, the microcapsules were washed two times with water and centrifuged to remove HCl, and finally redispersed in water. The concentration of the microcapsule suspension was estimated directly with an optical microscope by counting the amount of microcapsules in the defined volume and averaging 3 times of count. 2.3. Loading and release experiments 0.5 mL microcapsule suspension (2 × 1011 capsules/L) in water was centrifuged at 5000rpm for 5 min and the supernatant

S. Ye et al. / Journal of Controlled Release 112 (2006) 79–87

was removed. Then, the microcapsules were re-dispersed in 0.5 mL insulin solution at different pH. After being incubated at desired temperature, the mixture was centrifuged and the insulin-loaded microcapsules were washed 2 times with water and used in release experiments at once. The amount of insulin loaded inside the microcapsules was estimated through the difference between the input amount and the amount in the supernatants produced during the washing process. In order to observe loading temperature effect, the loading experiment was conducted in two procedures: one was to incubate the microcapsules at only one desired temperature for 2h, another was to load at 20 °C for the first hour and then at a higher desired temperature for the second hour. For cross-linking the microcapsule shell, 50 μL of CaCl2 solution (0.01 M) was added into 0.5 mL suspension of the insulin-loaded microcapsule and cured at room temperature for 30 min. Then, the cross-linked microcapsule was washed two times with the same procedure. The protein content in the centrifuge supernatant was determined according to the Bradford method [32]. The loading capacity (L), defined as the number of the loaded protein in one microcapsule, was evaluated by the following equation: L ¼ cNA =Mw ½N  where c is the concentration of the loaded protein in mg/mL, approximately evaluated from the concentration difference of the loading solution and the washing supernatants and the volume of a MF particle of 2.1μm diameter. NA is the Avogadro constant, MW is the molecular weight of the loaded protein in g/mol, and [N] is microcapsule concentration in number of microcapsules per litre. The release experiment was carried out by dipping ∼10 8 insulin-loaded microcapsules into either 1 mL of phosphate buffer at pH 7.4 or hydrochloric acid buffer at pH 1.4. After immersed for a desired time, the microcapsules were separated by centrifuge and the protein content in the supernatant was determined by the Bradford method. The relative amount of insulin released was calculated from the protein cumulatively released to the supernatants to that loaded in the microcapsules after washing. The additional ALG and CHI layers were coated on the loaded microcapsules. The same ALG and CHI deposition solutions and procedures as those used in the LbL assembly for the core-shell microparticles were used for coating the insulin-loaded microcapsules. All the supernatants produced during the washing process were collected for insulin quantification by UV absorbance. Through the total insulin amount in the supernatants, we calculated the insulin loss due to the additional layer deposition.

2.4. Characterization Scanning electron microscopy (SEM) of a Philips XL 30 was used and the sample for SEM was prepared by depositing suspensions of bare and coated MF particles on Si slides. After evaporation of water, the sample was coated with a thin gold layer.

81

The confocal micrograph was taken with a MRC 600 (BioRad, America) confocal laser scanning microscopy (CLSM) with a 100× oil immersion objective with a numerical aperture of 1.4. The microcapsule was visualized either by FITC-ALG or FITC-insulin with the excitation wavelength of 488 nm. The insulin labeling was based on a method reported by Lamprecht et al. [33]. 10mL of the 2.5% (w/v) aqueous insulin solution was adjusted to pH 8.5 by sodium hydroxide solution (1M). FITC was dissolved in dimethyl sulfoxide at concentration of 1mg/mL, 10μL of the dye solution was added to the insulin solution and stirred at 40 °C for 1h. The reaction was stopped by adding 5μL of thanolamine and free FITC was removed by precipitating the FITC-insulin at pH 5.5 of the isoelectric point of insulin and then FITC-labeled insulin was dissolved at pH 3. ALG was labeled with FITC using the same procedure. The FITC-ALG was also purified following the same procedure with an additional condition that the fluorescence intensity of water outside became minimal at the end of dialysis. The MF particles were dissolved in pH 1.0 HCl solutions and then 1 mL aqueous ALG solution of 1mg/mL was added to form ALG/MF complex. The resulting suspension was centrifuged and the complex was re-dispersed in water at desired pH and ζpotential of the complex was measured with a Brookhaven zetapotential analyzer. The value was obtained from the average of three successive measurements. UV absorbance at 600 nm was measured with a Hitachi U3010 for insulin quantification. Fluorescence emission was recorded with a Hitachi F-4500 fluorescence spectrometer with 490nm excitation and 515 nm emission. The excitation and emission slits were fixed at 5nm. 3. Results and discussion 3.1. ALG/CHI multilayer shell microcapsules Several types of decomposable templates have been utilized for the LbL assembly to fabricate hollow polyelectrolyte capsules, such as polystyrene latex, silica particles, drug microcrystals, cells, etc [6,9,17,26,34]. Among them, the weakly cross-linked MF colloid particles with monodisperse diameter have been generally employed because of easy dissolution in mild acidic medium [8]. Freshly prepared MF particles are required to ensure the dispersed state and capsule integrity, so MF particles were prepared in our laboratory with desired diameter for the sacrificial template [35]. The SEM image in Fig. 1a shows the MF particles so prepared with almost monodisperse diameter of 2.1μm, which can be easily dissolved in acid medium within the shelf life of at least 2months. The surface roughness of the bare MF particles is higher than other templates, such as polystyrene colloid, which helps the adsorption of deposited polyelectrolytes to form a more stable multilayer shells on the particles [36]. For the LbL assembly of oppositely charged ALG and CHI onto the MF particles, the electrostatic attraction is the dominant driving force. Negatively charged ALG is the first layer on the MF particles because of the positive charge on the MF particle surface. The next layer is positively charged CHI, and this procedure is repeated until

82

S. Ye et al. / Journal of Controlled Release 112 (2006) 79–87

Fig. 1. SEM photos of the bare MF particles (a) and the core-shell particles with 5ALG/CHI bilayers (b) with the insets showing details. The scale bars correspond to 10μm.

Fluorescence intensity (a. u.)

required layer number of the ALG/CHI multilayer film is realized. Direct visualization of the core-shell MF particles with 5 bilayers of ALG/CHI is depicted in Fig. 1b by SEM images. Compared with the bare MF particles (Fig. 1a), the surface of these core-shell particles (Fig. 1b) becomes smooth. This is owing to the irregular stack of polysaccharides CHI and ALG on the particle surface during the alternating deposition. Hollow microcapsules have been produced after removing the MF templates in the core-shell particles by repeatedly dipping into pH 1.2 hydrochloric acid solutions. The weakly cross-linked MF core was allowed to degrade at low pH to small soluble fragments, which could penetrate through the polyelectrolyte multilayer shell from the capsule interior to form freestanding hollow microcapsules [35]. FITC-ALG was used in fabricate the FITC-ALG/CHI microcapsules to visualize the microcapsule directly with the CLSM observation. Fig. 2 shows the confocal image of the microcapsule framed with (FITCALG/CHI)5 multilayer shell maintaining the shape of the MF particle except for some flattening. From the corresponding fluorescence intensity profile, one can see the intensity change along the line at the capsule cross-section. As expected, the main fluorescence emission comes from the (FITC-ALG/CHI)5 multilayer shell of the microcapsule. However, we can observe from Fig. 2 somewhat fluorescence emission appearing inside the capsules, indicating the FITC-ALG macromolecule existing at the capsule interior. This photo was taken just after removing

the MF particles and the FITC-ALG was the innermost layer of the capsules. This phenomenon is of great interesting and will be discussed in the following section. 3.2. Selective loading of insulin into microcapsules Insulin is a protein soluble in both acid and basic aqueous solutions [37]. Therefore, loading insulin into the ALG/CHI microcapsules was carried in mediums either pH 3.0 or pH 8.0 at room temperature. The CLSM images are shown in Fig. 3 with the emission from FITC-insulin. There is no visible insulin accumulated in the microcapsules when deposited at pH 8.0 (Fig. 3a). In contrast, obvious insulin accumulated inside the microcapsules can be observed from Fig. 3b when deposited at pH 3.0. Considering that the isoelectric point of insulin is pH 5.5, the above observation means that only positively charged insulin (at pH 3.0) could be loaded into the microcapsules. Gao et al. reported that some positively charged substances, such as rhodamine, dextran etc., spontaneously accumulated inside the PSS/PAH microcapsules when MF particles were used as templates [18]. They proposed that the negatively charged PSS at the innermost layer formed a complex with the positively charged MF residues produced during the core dissolution and decomposition. The driving force for their spontaneous deposition was considered as the electrostatic interaction between the PSS/MF complex with left negative charge and

80

60

40

20

0 0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

4.0

Position (µm) Fig. 2. CLSM photo of (FITC-ALG/CHI)5 multilayer microcapsules (right) and the fluorescence intensity profile along the line crossing one microcapsule (left).

S. Ye et al. / Journal of Controlled Release 112 (2006) 79–87

a

83

b

Fig. 3. CLSM photos showing the FITC-insulin loaded into the ALG/CHI microcapsules at pH = 8.0 with negatively charged insulin (a) and pH = 3.0 with positively charged insulin after rinse (b).

3.3. Insulin loading capacity Owing to the electrostatic attraction, the concentration of loaded substance inside the microcapsules is possible to be higher than that in the outside bulk solution. Thus, we have examined some factors that affect the insulin loading capacity of the microcapsule. The loading time was fixed at 2 and 1h loading

was long enough for the loading equilibrium as known from preliminary experiments (data not shown here). Fig. 4 illustrates the relationship between the insulin loading capacity and the insulin concentration in the bulk solution at pH 1.0 under fixed microcapsule concentration. With increasing insulin concentration in the bulk solution from 1.0 to 6.0mg/mL, the loading capacity increases first and then reaches an equilibrated value of about 1.6 × 108 molecules/capsule, which is comparable to the 1.4 × 107∼1.0 × 109 molecules/capsule in the literature [19–21]. This insulin concentration of 44mg/mL inside the microcapsules is much higher than 5mg/mL of the bulk solution, unconformable to the diffusion driven by concentration difference. The present fact also indicates the existence of other driving force rather than concentration difference for this insulin loading. However, the interior volume of one microcapsule approximately estimated from the MF particle diameter is only 4.8 × 10− 12 mL, so the insulin in one capsule is as low as 2.13 × 10− 13 g. The ALG/MF complex required in the microcapsules for insulin spontaneous loading will be even lower because both ALG and insulin have many charge sites along one molecular chain. The bulk solution pH influence on the insulin loading capacity of the ALG/CHI microcapsules is demonstrated in Fig. 5 for pH ranging from 1.0 to 4.0. The loading capacity is 1.8

Loading capacity ( x108 insulin molecule/capsule)

the water-soluble positively charged substances. As observed from Fig. 2, some FITC-ALG macromolecules appeared at the interior of the microcapsules after removing the MF particles. According to Gao et al. suggestion [1,18], some FITC-ALG at the innermost layer would form the negatively charged ALG/MF complex with the MF fragments. During the acidic degradation of the MF particles some ALG would remain in the binding state with the MF fragments to form a complex because the ALG in the innermost layer was adsorbed on the MF particles by charge attraction. In an independent experiment, water-insoluble ALG/ MF complex was formed by mixing ALG solution with MF particles immediately after dissolved in a solution of pH = 1.0. The ζ-potential of this ALG/MF complex was −22.6mV at pH 4.0 and decreased to − 14.2mV at pH 1.0. Though the ALG/MF complex has not been directly separated from the interior of the microcapsules yet, it is reasonable to assume its formation inside the microcapsules. Owing to existence of the negatively charged ALG/MF complex inside the microcapsules, the positively charged insulin can be selectively loaded into the ALG/CHI microcapsules (pH 3.0, Fig. 3b) while the negatively charged insulin cannot be loaded into the microcapsules (pH 8.0, Fig. 3a). Thus, the loading of insulin into the ALG/CHI microcapsules made on MF templates can be simply controlled with pH. However, one may argue that the change in the insulin loading capacity could be due to the permeability change of the microcapsule wall with pH. For example, Sukhorukov et al. reported that the microcapsule wall changed the permeability with pH because of the change in the ionization degree of PAH in the wall with pH [16]. However, if this were true for the present case, the loading of positively charged insulin would be excluded at pH 3.0 for CHI in the multilayer wall is positively charged at this pH due to the CHI and ALG with pKa of about 7.0 and 4.0, respectively [38,39].

1.6 1.4 1.2 1.0 0.8 0.6 0.4 0.2 0.0 1

2

3

4

5

6

Insulin concentration (mg/mL) Fig. 4. The loading capacity of the AL/CHI microcapsules as a function of insulin concentration in the bulk solution at pH 1.0.

S. Ye et al. / Journal of Controlled Release 112 (2006) 79–87

0.32 × 108 molecules/capsule at pH 4.0, and increases as the pH is decreased. The loading capacity is 1.58 × 108 molecules/ capsule at pH 1.0, about 5 times as that at pH 4.0. This significant increase in the loading capacity is caused by the increase in ionization degree of insulin with decreasing pH due to departing further from its isoelectric point pH 5.5, leading to a stronger electrostatic attraction between the positive insulin and the negative ALG/MF complex inside the microcapsules. As mentioned in 3.2, the absolute ζ-potential value of the ALG/MF complex at pH 1.0 is decreased to − 14.2mV, still providing attraction to the positively charged insulin. Therefore, pH = 1.0 is chosen for the following insulin loading. Here, we have designed two procedures to investigate the temperature effect on the loading capacity and depicted the results in Fig. 6. The one-temperature loading represents loading at one temperature for a hour and the two-temperature loading means loading at 20 °C for the first hour and then at another temperature for the second hour. For the one-temperature loading, the insulin loading capacity decreases from 1.70 × 108 to 0.85 × 108 molecules/capsule when the loading temperature is increased from 20 to 60 °C. For the two-temperature loading, however, the loading capacity is increased up to 2.63 × 108 molecules/capsule as loaded at a higher temperature for the second hour. To authors’ knowledge, this is the first report for this interesting phenomenon. Increasing temperature has two effects on the insulin loading: enhancing the penetrating movement of insulin through the multilayer wall and accelerating the rearrangement of the polysaccharide chains in the shell to form more compact structure. The latter effect seems to be dominant in the present loading process, blocking the insulin from penetrating through the microcapsule shell at higher temperatures. Balabushevich et al. raised the loading temperature of catalase from 20 to 50 °C and found a 30% decrease in the loading capacity [28]. Ibarz et al. also reported that the penetration of small molecule fluorescein through the PSS/ PAH microcapsule shell was reduced significantly by annealing at 80 °C [40]. As for the two-temperature loading, the first hour loading insulin at 20 °C is sufficient to reach the loading equilibrium, the second hour loading at a high temperature

Loading capacity ( x108 insulin molecule/capsule)

84

2.5

one-temperature loading two-temperature loading

2.0

1.5

1.0

0.5

0.0

20

40

60

Loading temperature (°C) Fig. 6. The loading capacity of the ALG/CHI microcapsules as a function of loading temperature at pH 1.0. The white columns represent the one-temperature loading and the gray columns represent two-temperature loading at the shown second loading temperature.

produces a more compact multilayer shell and reduces the shell permeability. In other words, the second hour loading at a high temperature corresponds to the annealing of loaded capsules at a higher temperature before rinsing. Therefore, the loss of loaded insulin from the capsules during rinse after loading is greatly suppressed. That the loading capacity of the one-temperature loading at 20°C is even lower than that of the two-temperature loading with a higher second temperature appears due to the higher loss of the loaded insulin for the former during rinse. This contrary influence of increasing loading temperature on the insulin loading capacity is interesting and provides a novel method to increase the loading capacity of water-soluble substances into microcapsules. 3.4. Controlled release from microcapsule The insulin release from the ALG/CHI microcapsules was investigated at pH 1.4 and 7.4. Fig. 7 shows the cumulative insulin release curves at these two pH values. The insulin is released very slowly at pH 1.4 and only 14.2% of the loaded insulin can be released after 6 h. While at pH 7.4, a much fast 100

1.6 1.4

Cumulative release (%)

Loading capacity ( x108 insulin molecule/capsule)

1.8

1.2 1.0 0.8 0.6 0.4 0.2

80

60 pH =7.4 pH =1.4

40

20

0.0 1

2

3

4

pH of loading solution

0 0

200

400

600

800

t (min) Fig. 5. The loading capacity of the ALG/CHI microcapsules as a function of pH of the insulin bulk solution.

Fig. 7. Insulin release from (ALG/CHI)5 microcapsules at indicated pH.

100

100

80

80

60

40 two-temperature loading second temperature 20 °C 40 °C 60 °C

20

0

Cumulative release (%)

Cumulative release (%)

S. Ye et al. / Journal of Controlled Release 112 (2006) 79–87

85

60

40 0 additional bilayers 2 additional bilayers 4 additional bilayers

20

0 0

100

200

300

400

500

600

700

800

900

0

100

t (min)

200

300

400

500

600

700

800

900

t (min)

Fig. 8. Release profiles of insulin from (ALG/CHI)5 microcapsules loaded with two-temperature loading at indicated second loading temperature.

Fig. 10. Release profiles of insulin from (ALG/CHI)5 microcapsules coated with additional layers after loading.

release is observed. About 50% insulin has been released within the first 80 min, and then the release rate decreases and the curve gradually levels off as the cumulative release amount reaches ∼80%. This significant difference of insulin release with the change in solution pH is also resulted from the insulin ionization. The loaded insulin is positively charged at pH 1.4 as mentioned above, thus strongly attracted with the negative ALG/MF complex inside the microcapsules, restricting the release. The insulin charge becomes negative at pH 7.4 and preferentially moves into the solution from the microcapsules due to the electrostatic repulsion, leading to a rapid release. The release curves at pH 7.4 for insulin from the (ALG/ CHI)5 microcapsules loaded with the two-temperature loading at different second temperatures are depicted in Fig. 8. The release rate becomes slower as the second loading temperature is higher. When the second loading temperature is 60 °C, the release rate is the slowest and the curve levels off at the cumulative release amount of ∼44%. As mentioned above, higher loading temperature produces a more compact shell wall for the ALG/CHI microcapsules, which blocks the insulin molecules from penetrating to the bulk solution. The present release behavior can be consistently explained with the

temperature effect on the loading capacity. Thus, the loading temperature provides a simple way not only to increase the insulin loading capacity but also to control the release rate of drug loaded in the microcapsules. Recently, We found that the release rate of acridine hydrochloride became slightly slower when the ALG in the ALG/CHI multilayer shell was cross-linked with glutaraldehyde [29]. Here, we investigate the insulin release from the same ALG/CHI microcapsules with the ALG cross-linked after loading. The diazoresin/PSS microcapsule shell cross-linked with UV irradiation has been found to lead to a decrease in the multilayer film permeability [41–43]. But this method requires the synthesis of diazoresin and the multilayer film assembled in dark. Hence, the calcium ion (Ca2+) was chosen as the crosslinker in this work for the ALG in the multilayer film. The release profiles in Fig. 9 indicate the reduction in the release rate and cumulative release amount due to the cross-linking of ALG with Ca2+ in the ALG/CHI multilayer shell. Because the crosslinking restricts the ALG chain moving in the shell and enhances the obstruction to the insulin movement passing through the microcapsule shell. The loaded microcapsules can be coated with additional polyelectrolyte layers to reduce the release rate further [44]. In our experiment, this process caused 34∼46wt.% loss of the insulin already loaded in the microcapsules, which is much smaller than 70wt.% loss reported by Balabushevich et al. for coating the dextran sulfate/protamine microcapsules with additional layers [21]. The insulin release from the ALG/CHI microcapsules with additional deposited polysaccharide layers is obviously slowed dawn compared with that without additional deposited layers as shown in Fig. 10. The additional deposited layers on the loaded microcapsules will give rise to an extra barrier to the insulin molecule movement, thus slow down the release rate.

Cumulative release (%)

100

80

60

40

non-crosslinked ALG crosslinked ALG

20

0 0

100

200

300

400

500

600

700

800

900

t (min) Fig. 9. Release profiles of insulin from (ALG/CHI)5 microcapsules before and after cross-linking ALG with Ca2+ ions.

4. Conclusions Weakly cross-linked melamine formaldehyde microparticles with diameter of 2.1 μm were prepared as sacrificial templates. Alginate and chitosan multilayer microcapsules

86

S. Ye et al. / Journal of Controlled Release 112 (2006) 79–87

were fabricated by the layer-by-layer self-assembly on the MF microparticles followed by removal of the template through dissolving at low pH. Insulin was spontaneously loaded into the ALG/CHI microcapsules when it was positively charged owing to the existence of negatively charged complex of ALG/ MF residues inside the capsules. The insulin loading capacity increased when the solution pH values decreased from 4.0 to 1.0. When the two-temperature loading process was applied, loading insulin at two temperatures of 20 and 60°C (one hour for each) not only increased the loading capacity but also slowed down the insulin release rate due to more compact multilayer shell formed at higher temperature. The release rate of insulin in pH 7.4 buffers was much faster than that in pH 1.4 buffer due to the charge reverse of insulin induced by the pH change. Cross-linking ALG in the capsule shell with calcium ions (Ca2+) or coating the insulin-loaded microcapsules with additional layers remarkably decreased the insulin release rate. These results provide novel methods to control the loading and release properties of protein with these polysaccharide microcapsules. Acknowledgements The authors thank Ms. Shanshan Liu of SCUT for kindly assistance in experiments. The financial support to this work of the NSF of China (No: 90206010, 20574023, 20534020), the NSF of Guangdong province (No: 031380), and the Scientific and Technologic Program of Guangzhou municipal is gratefully acknowledged. References [1] C.Y. Gao, E. Donath, H. Möhwald, J.C. Shen, Spontaneous deposition of water-soluble substances into microcapsules: phenomenon, mechanism, and application, Angew. Chem., Int. Ed. Engl. 114 (2002) 3789–3793. [2] G.B. Sukhorukov, L. Dähne, J. Hartmann, E. Donath, H. Möhwald, Controlled precipitation of dyes into hollow polyelectrolyte capsules based on colloids and biocolloids, Adv. Mater. 12 (2000) 112–115. [3] Y. Lvov, A.A. Antipov, A. Mamedov, H. Möhwald, G.B. Sukhorukov, Urease encapsulation in nanoorganized microshells, Nano Lett. 1 (2001) 125–128. [4] L. Dähne, S. Leporatti, E. Donath, H. Möhwald, Fabrication of micro reaction cages with tailored properties, J. Am. Chem. Soc. 123 (2001) 5431–5436. [5] G. Decher, J.D. Hong, J. Schmitt, Buildup of ultrathin multilayer films by self-assembly process: III. Consecutively alternating adsorption of anionic and cationic polyelectrolyte on charged surfaces, Thin Solid Films 210/211 (1992) 831–835. [6] F. Caruso, R.A. Caruso, H. Möhwald, Nanoengineering of inorganic and hybrid hollow spheres by colloidal templating, Science 282 (1998) 1111–1114. [7] T. Okubo, M. Suda, Synchronous multilayered adsorption of macrocations and macroanions on colloidal spheres: influence of foreign salt and basicity or acidity of the macro ions, Colloid Polym. Sci. 280 (2002) 533–538. [8] A.A. Antipov, G.B. Sukhorukov, Polyelectrolyte multilayer capsules as vehicles with tunable permeability, Adv. Colloid Interface Sci. 111 (2004) 49–61. [9] S. Ye, C. Wang, X. Liu, Z. Tong, Deposition temperature effect on release rate of indomethacin microcrystals from microcapsules of layer-by-layer assembled chitosan and alginate multilayer films, J. Control. Release 106 (2005) 319–328.

[10] Z. Liang, C. Wang, Z. Tong, W. Ye, S. Ye, Bio-catalytic nanoparticles with urease immobilized in multilayer assembled through layer-by-layer technique, React. Funct. Polym. 63 (2005) 85–94. [11] C. He, Z. Liang, C. Wang, X. Liu, Z. Tong, Preparation of nanoparticles loaded superoxide dismutase by layer-by-layer assembly and their bioactivity, Chem. J. Chin. Univ. 1 (2005) 88–92. [12] G.B. Sukhorukov, M. Brumen, E. Donath, H. Möhwald, Hollow polyelectrolyte shells: exclusion of polymers and donnan equilibrium, J. Phys. Chem., B 103 (1999) 6434–6440. [13] R. Ghan, T. Shutava, A. Patel, V.T. John, Y. Lvov, Enzyme-catalyzed polymerization of phenols within polyelectrolyte microcapsules, Macromolecules 37 (2004) 4519–4524. [14] D.G. Shchukin, I.L. Radtchenko, G.B. Sukhorukov, Micron-scale hollow polyelectrolyte capsules with nanosized magnetic Fe3O4 inside, Mater. Lett. 57 (2003) 1743–1747. [15] A.A. Antipov, D.G. Shchukin, Y. Fedutik, I. Zanaveskina, V. Klechkovskaya, G.B. Sukhorukov, H. Möhwald, Urease-catalyzed carbonate precipitation inside the restricted volume of polyelectrolyte capsules, Macromol. Rapid Commun. 24 (2003) 274–277. [16] G.B. Sukhorukov, A.A. Antipov, A. Voigt, E. Donath, H. Möhwald, pHcontrolled macromolecule encapsulation in and release from polyelectrolyte multilayer nanocapsules, Macromol. Rapid Commun. 22 (2001) 44–46. [17] E. Donath, S. Moya, B. Neu, G.B. Sukhorukov, R. Georgieva, A. Voigt, H. Baumler, H. Kiesewetter, H. Möhwald, Hollow polymer shells from biological templates: fabrication and potential applications, Chem. Eur. J. 8 (2002) 5481–5485. [18] C.Y. Gao, X.Y. Liu, J.C. Shen, H. Möhwald, Spontaneous deposition of horseradish peroxidase into polyelectrolyte multilayer capsules to improve its activity and stability, Chem. Commun. 17 (2002) 1928–1929. [19] O.P. Tiourina, A.A. Antipov, G.B. Sukhorukov, N.I. Larionova, Y. Lvov, H. Möhwald, Entrapment of α-chymotrypsin into hollow polyelectrolyte microcapsules, Macromol. Biosci. 1 (2001) 209–214. [20] O.P. Tiourina, G.B. Sukhorukov, Multilayer alginate/protamine microsized capsules: encapsulation of α-chymotrpsin and controlled release study, Int. J. Pharm. 242 (2002) 155–161. [21] G.N. Balabushevich, O.P. Tiourina, D.V. Volodkin, N.I. Larionova, G.B. Sukhorukov, Loading the multilayer dextran sulfate/protamine microsized capsules with peroxidase, Biomacromolecules 4 (2003) 1191–1197. [22] X. Qiu, E. Donath, H. Möhwald, Permeability of ibuprofen in various polyelectrolyte multilayers, Macromol. Mater. Eng. 286 (2001) 591–597. [23] X. Qiu, S. Leporatti, E. Donath, H. Möhwald, Studies on the drug release properties of polysaccharide multilayers encapsulated ibuprofen microparticles, Langmuir 17 (2001) 5375–5380. [24] H. Ai, S. Jones, M. de Villiers, Y.M. Lvov, Nano-encapsulation of furosemide microcrystals for controlled drug release, J. Control. Release 86 (2003) 59–68. [25] D.B. Shenoy, G.B. Sukhorukov, Engineered microcrystals for direct surface modification with layer-by-layer technique for optimized dissolution, Eur. J. Pharm. Biopharm. 58 (2004) 521–527. [26] Z.H. An, G. Lu, H. Möhwald, J.B. Li, Self-assembly of human serum albumin (HSA) and L-alpha-dimyristoylphosphatidic acid (DMPA) microcapsules for controlled drug release, Chem. Eur. J. 10 (2004) 5848–5852. [27] Z.F. Dai, A. Heilig, H. Zastrow, E. Donath, H. Möhwald, Novel formulations of vitamins and insulin by nanoengineering of polyelectrolyte multilayers around microcrystals, Chem. Eur. J. 10 (2004) 6369–6374. [28] N.G. Balabushevich, E.P. Zimina, N.I. Larionova, Encapsulation of catalase in polyelectrolyte microspheres composed of melamine formaldehyde, dextran sulfate, and protamine, Biochemistry(Moscow) 69 (2004) 937–944. [29] S. Ye, C. Wang, X. Liu, Z. Tong, Multilayer nanocapsules of polysaccharide chitosan and alginate through layer-by-layer assembly directly on PS nanoparticles for release, J. Biomater. Sci., Polym. Ed. 16 (2005) 909–923. [30] G.P. Carino, E. Mathiowitz, Oral insulin delivery, Adv. Drug Deliv. Rev. 35 (1999) 249–257. [31] E. Mathiowitz, J.S. Jacob, Y.S. Jong, G.P. Carino, D. Chickering, C. Santos, K.C. Vijayaraghavan, M. Bassett, S. Montgomery, C. Morrell, Biologically erodable microsphere as potential oral drug delivery system, Nature 386 (1997) 410–414.

S. Ye et al. / Journal of Controlled Release 112 (2006) 79–87 [32] M.M. Bradford, A rapid and sensitive method for the quantitation of microgram quantities of protein utilizing the principle of protein-dye binding, Anal. Biochem. 72 (1976) 248–252. [33] A. Lamprecht, U.F. Schäfer, C.M. Lehr, Characterization of microcapsules by confocal laser scanning microscopy: structure, capsule wall composition and encapsulation rate, Eur. J. Pharm. Biopharm. 49 (2000) 1–9. [34] Y. Itoh, M. Matsusaki, T. Kida, M. Akashi, Preparation of biodegradable hollow nanocapsules by silica template method, Chem. Lett. 33 (2004) 1552–1553. [35] C.Y. Gao, S. Moya, E. Donath, H. Möhwald, Melamine formaldehyde core decomposition as the key step controlling capsule integrity: optimizing the polyelectrolyte capsule fabrication, Macromol. Chem. Phys. 203 (2002) 953–960. [36] Y. Lvov, F. Caruso, Biocolloids with ordered urease multilayer shells as enzymatic reactors, Anal. Chem. 73 (2001) 4212–4217. [37] F.A. Dorkoosh, J.C. Verhoef, M.H.C. Ambagts, M. Rafiee-Tehrani, G. Borchard, H.E. Junginger, Peroral delivery systems based on superporous hydrogel polymers: release characteristics for the peptide drugs buserelin, octreotide and insulin, Eur. J. Pharm. Sci. 15 (2002) 433–439.

87

[38] X. Liu, L. Qian, T. Shu, Z. Tong, Rheology characterization of sol–gel transition in aqueous alginate solutions induced by calcium cations through in situ release, Polymer 44 (2003) 407–412. [39] A. Domard, pH and c d measurements on a fully deacetylated chitosan: application to CuII-polymer interactions, Int. J. Biol. Macromol. 9 (1987) 98. [40] G. Ibarz, L. Dähne, E. Donath, H. Möhwald, Controlled permeability of polyelectrolyte capsules via defined annealing, Chem. Mater. 14 (2002) 4059–4062. [41] I. Pastoriza-Santos, B. Schöler, F. Caruso, Core-shell colloids and hollow polyelectrolyte capsules based on diazoresins, Adv. Funct. Mater. 11 (2001) 122–128. [42] Y. Zhang, S. Yang, Y. Guan, W. Cao, J. Xu, Fabrication of stable hollow capsules by covalent layer-by-layer self-assembly, Macromolecues 36 (2003) 4238–4240. [43] H.G. Zhu, M.J. McShane, Macromolecule encapsulation in diazoresin-based hollow polyelectrolyte microcapsules, Langmuir 21 (2005) 424–430. [44] G. Ibarz, L. Dahne, E. Donath, H. Möhwald, Resealing of polyelectrolyte capsules after core removal, Macromol. Rapid Commun. 23 (2002) 474–478.