Intelligent semi-IPN chitosan–PEG–PAAm hydrogel for closed-loop insulin delivery and kinetic modeling

Abstract

A successful approach to designing an intelligent insulin delivery system could facilitate diabetic patients’ lives especially for type 1 diabetes. The aim of the present investigation was to develop an intelligent closed-loop insulin delivery system for implantation. Glucose-responsive semi-IPN hydrogels were synthesized from free radical polymerization of chitosan (CS), acrylamide (AAm) and polyethylene glycol (PEG). Glucose oxidase (GOx) and catalase (CAT) along with insulin were immobilized and loaded into the hydrogels in order to make an intelligent drug carrier, which is able to play the role of an artificial pancreas. The designed glucose responsive hydrogel acts as a self-regulating insulin delivery system and the insulin release rate is associated with the blood glucose level. The loaded hydrogels were characterized using Fourier transform infrared spectroscopy (FTIR), scanning electron microscopy (SEM), high performance liquid chromatography (HPLC), and thermal analysis (TGA/DTG). Cell culture tests with fibroblast cells were conducted to perform biocompatibility testing for the drug carrier systems. The effect of the incorporated PEG on the swelling ratio (SR), drug loading capacities (DLC), and entrapment efficiency (EE) of the intelligent semi-IPN hydrogels was investigated using HPLC and UV-vis spectroscopy. Optimization of the hydrogel was also investigated using a full factorial design and by changing the amount of PEG. Kinetic studies were performed using different kinetic models to investigate the insulin release rate.

---

Introduction

Diabetes mellitus, a metabolic disease characterized by an abnormal increase of the blood glucose level, resulting from poor insulin secretion by the pancreas and/or defects in the insulin action, is considered as a global epidemic. It represents a major challenge for health systems around the world. According to the World Health Organization (WHO), the worldwide number of people with this disease was 177 million in 2000 and is expected to reach 350 million people by 2025.¹ Conventional insulin therapy does not tightly regulate the glucose level in patients, since the glucose sensing and insulin therapy are not directly coupled. This method of diabetes treatment is known as open-loop insulin delivery. In people with type 1 diabetes, who have lost the ability of insulin secretion, the body's defensive mechanisms against hyperglycemia are impaired during the night, which can increase the risk of hyperglycemia.^2,3 Untreated hyperglycemia causes long-term complications including cardiovascular disease, nerve damage (neuropathy), kidney failure, damage to the blood vessels of the retina (diabetic retinopathy), bone problems, gum infections, and fatal hypoglycemia. Researchers have been investigating chemically controlled closed-loop delivery systems, as an appropriate substitute for electronic devices.^4–14 The various applications and features of polymers make them an appropriate choice to be used as excipients in controlled drug release therapies. Polymeric hydrogels that swell or shrink to adjust the insulin release rate according to variations in the ambient glucose level could be appropriate candidates to perform this function.^15,16 Chitosan is a biopolymer with many structural aspects feasible for mechanical and chemical modification, generating novel applications and properties mainly in the biomedical field. In addition, chitosan has special properties like biodegradability, biocompatibility and non-toxicity besides its antimicrobial activity and low immunogenicity, making it a good choice to be used in drug delivery systems. In the body, chitosan (CS) can be degraded by lysozymes or glycosidases into amino sugars and subsequently cleared from the body.^17,18 Polyethylene glycol (PEG) is a nontoxic hydrophilic polymer extensively used for protein purification and protein conjugation. PEG helps to adapt the hydrogel’s conformation to fit the surface topology of proteins through its uncharged chemical composition and segmental flexibility. The chemical modification of chitosan with PEG not only improves the biocompatibility of chitosan, but also reduces the adsorption of circulating plasma proteins onto the material surface and extends the half-life of drugs.^19–22 The aim of the present work is to find a less traumatic, more reliable, and intelligent insulin delivery system as an alternative to insulin administration. Accordingly, the development of a system for closed-loop insulin delivery that can perform this function is needed. This system should have the abilities of continuous glucose-sensing and regulating the release of insulin, simultaneously. A considerable amount of work has been reported in the literature regarding the application of polymer gels and nanoparticles as drug delivery systems in insulin therapy, however in most of the cases the kinetics and on–off switching of the systems were not investigated (Table 1). Therefore, the design of an intelligent drug delivery system, which has the ability of sensing the glucose concentration using a specific sensor like glucose oxidase, is an important improvement for insulin therapy, which demands further investigation. In addition, factorial design can be used as a powerful tool to study the kinetics of release. In this research, glucose responsive semi-IPN hydrogels were prepared by free radical polymerization of chitosan (CS), acrylamide (AAm) and polyethylene glycol (PEG).

Table 1 Typical polymer gels and nanoparticles as drug delivery systems in insulin therapy

Ref no.	Matrix	On–off switching	Working mechanism	Factorial design	GOx	Kinetic modeling	Release detection
5	Chitosan/PEG hydrogel beads	—	Swelling diffusion	—	—	—	UV-visible
6	Nanoparticle chitosan–alginate polyelectrolyte complexes	—	Destruction with pH	—	—	—	UV-visible
7	Poly dimethylsiloxane (PDMS) hydrogel	—	Swelling diffusion	—	*	—	UV-visible
8	N-Isopropylacrylamid/methacrylic acid nanoparticles	—	—	*	*	—	HPLC, CD
9	Nanoparticle chitosan–TPP	—	—	—	—	—	—
10	Chitosan–PEG hydrogel	—	Swelling diffusion	—	—	—	UV-visible
11	PEG/N-isopropylacrylamide hydrogel	—	Swelling non-Ficken diffusion	—	—	*	UV-visible
12	Hydroxyapatite combined with macrophages	—	Osmotic pressure	—	—	—	UV-visible
4	Chitosan-microgels	—	Swelling diffusion	—	*	—	UV-visible
13	Chitosan–PEG nanoparticles	—	Swelling diffusion	—	—	—	UV-visible
14	Nanoparticle chitosan–alginate modified with dextran	*	Degradation with pH	—	*	—	UV-visible

---

Glucose oxidase (GOx) was immobilized in the hydrogel as a specific glucose biosensor to regulate the insulin delivery. The pH value of the medium will decrease due to the formation of gluconic acid produced from the specific reaction of GOx with glucose, which leads to hydrogel swelling and release of insulin (Fig. 1).^4,15,23 Catalase is added to regenerate oxygen (O₂) to assist the GOx's glucose catalysis reaction and consume undesired hydrogen peroxide (H₂O₂) produced by glucose oxidation, which may be toxic to the body and deactivate GOx. The degradation of chitosan and related hydrogels with H₂O₂ has been reported by other researchers.^24–26 The presence of catalase seems to increase the swelling kinetics, which was reported by Traitel et al.²⁷ The insulin loaded hydrogels were characterized using Fourier transform infrared spectroscopy (FTIR), scanning electron microscopy (SEM), high performance liquid chromatography (HPLC), and thermogravimetric analysis (TGA). In vitro biocompatibility tests of the semi-IPN hydrogel were conducted for the drug carrier systems. The effect of the incorporated PEG on the swelling ratio (SR), the drug loading capacities (DLC), and the entrapment efficiency (EE) of the glucose responsive semi-IPN hydrogels was investigated using UV-vis spectroscopy and HPLC. To optimize the intelligent insulin-loaded hydrogel, the effects of the independent variables, i.e. the drug/hydrogel ratio (X₁) and the stirring speed (X₂), on the dependent variables, i.e. EE (%), SR and the time required for 75% of drug release (t₇₅), were investigated using a full factorial design. Finally, various kinetic models were studied to select the best model for the in vitro controlled release of insulin.

Fig. 1 Schematic representation of insulin release, as a response to high glucose concentration.

Experimental section

Materials

Chitosan (CS) with an approximate molecular weight of 270 000 and a 56% degree of deacetylation was received from Novinshimiar. N,N′-Methylenebisacrylamide (MBA) was purchased from SD Fine. Acrylamide (for the synthesis) and PEG 35000 were obtained from Merck Millipore. Hydrogen peroxide (H₂O₂), glucose oxidase (GOx) and catalase were gifts from the Pishtazteb Zaman research and development Co. Human insulin 100 IU mL⁻¹ was purchased from the Exir pharmaceutical Co.

Synthesis of the semi-IPN chitosan–PEG–PAAm hydrogels

Chitosan–polyethylene glycol semi-IPN hydrogels were prepared through free radical cross-linking polymerization of acrylamide as the hydrophilic non-ionic monomers and chitosan as a cationic polymer. Briefly, 1.00 g of chitosan was dissolved in 15 mL of 1% acetic acid solution. While the chitosan solution was under magnetic stirring at ambient temperature, 10 mL of acrylamide (200 g L⁻¹) and a PEG solution of different concentrations were added. The solution was mixed until it became homogenous. The temperature of the reactor was kept at 80 °C. Then, 30 μL of hydrogen peroxide 30% (v/v) was used as an initiator. Afterwards, 5 mL of MBA, 10% (w/v), was added to the solution and allowed to stir under atmospheric conditions at 80 °C. After formation of the hydrogel, it was allowed to cool at room temperature and then kept in ethanol to allow extraction of the water and soluble fractions. To complete the drying process the hydrogels were incubated in an oven at 40 °C overnight.

Insulin loading and enzyme immobilization

Immobilization of the GOx, CAT and insulin, and loading into the hydrogels were done via a swelling–diffusion method, simultaneously. A solution of the enzymes and insulin in a known amount was prepared using phosphate buffer saline (PBS) 1× at pH = 7; (1× PBS should have a final concentration of 10 mM PO₄³⁻, 137 mM NaCl, and 2.7 mM KCl) and was applied as a loading medium. Then, 0.10 g of the hydrogel powder was suspended in 100 mL of this medium for 24 hours. When the swelling equilibration was completed, the swollen hydrogels were taken out and rinsed thoroughly with PBS. Then, the hydrogels were weighed to calculate the swelling ratio and freeze dried at −40 °C for 24 hours.

Techniques and methods

The loaded intelligent semi-IPN hydrogels were characterized using various techniques as follows: FTIR spectra were recorded using a Bruker Tensor 27. Scanning electron microscopy (SEM), model AIS 2100 from Soren Technology, was performed to obtain morphological information on the intelligent hydrogel surface. TGA/DTG measurements were performed under a nitrogen atmosphere at a heating rate of 10 °C min⁻¹ with the aid of a Perkin Elmer Pyris Diamond thermal analyzer.

Chromatographic separation was performed with the aid of a Knauer 1100 HPLC analyzer with a C18 25 × 4.0 mm analytical column at room temperature. Methanol was used as the mobile phase. The eluent flow rate was 1.0 mL min⁻¹. The volume of the sample injected was 20 μL. The system was equipped with a UV detector at a wavelength of 214 nm. Camspec M 350, a UV-vis spectrophotometer, was applied to detect insulin in a PBS solution at 280 nm.

Cell culture tests

Cell culture tests were conducted using mouse fibroblasts (Mouse C34/connective tissue-L929) in a DMEM culture media with 10% fetal bovine serum, L-glutamine and antibiotics (50 units penicillin, 50 μg ml⁻¹ streptomycin) in an atmosphere with 85% air humidity and 5% CO₂. Upon seeding the samples in twenty four-well culture plates, 2000 cells per ml were used for the MTT cytotoxicity assays.

MTT cytotoxicity assays

The MTT [3-(4,5-dimethylthiazol-2-yl)-2,5-dipenyl tetrazolium bromide] was at a concentration of 5 mg ml⁻¹ in PBS (Phosphate Buffer Solution). After a 48 h growth period, MTT was added to the cell culture and incubated for 4 h at 37 °C. After decanting and washing with PBS, 0.5 mL of DMSO was added to each well to dissolve the product. A sample of the solution was taken for analysis using an Elisa plate reader and the absorbance was determined at 570 nm. The test was repeated three times and the average absorbance was calculated. Monitoring of the cultures was performed after 24 and 48 h using phase contrast inverted microscopy (Nikon TE-100).

Drug loading capacities (DLC) and entrapment efficiency (EE)

The amount of unloaded insulin was determined using HPLC of the remaining loading medium. A UV-vis spectrophotometer at a wavelength of 280 nm was applied as a second technique to measure the unloaded insulin in the loading medium. The swelling ratio (SR), drug loading capacities (DLC), and entrapment efficiency (EE) were calculated as follows:where, W_t and W₀ are the weight of the swollen gels and dehydrated samples, respectively. W_IL is the weight of the insulin loaded in the hydrogel, W_I0 is the total weight of insulin in the medium before loading, and W_HG is the weight of the dry powder gel sample.²⁸

CHN analysis

Elemental analysis was performed using a Perkin-Elmer 2004 (II) CHN analyzer.

Factorial design

A 3² full factorial design was utilized to investigate the effect of independent variables. In this statistical model, which includes interactive and polynomial terms, the drug/hydrogel ratio (X₁) and the stirring speed (X₂) are independent variables, whereas the EE (%), SR (%) and time required for 75% drug release (t₇₅) are dependent variables.

Eqn (4) demonstrates the relationship between the variables as follows:

Y = b₀ + b₁X₁ + b₂X₂ + b₁₂X₁X₂ + b₁₁X₁² + b₂₂X₂² (4)

where, Y is the dependent variable, b₀ is the arithmetic mean response of the nine runs, and b_i is the estimated coefficient for the factor X_i. When two factors are simultaneously changed, the interaction terms (X₁X₂) show how the response changes. To study non-linearity, polynomial terms (X₁² and X₂²) are included.²⁹

Glucose responsiveness of the intelligent drug carrier

The sensitivity of the intelligent drug carrier to glucose concentration was examined with two tests. In the first examination, 0.10 g of the synthesized drug loaded hydrogel was put in PBS with different glucose concentrations. The amount of insulin released was measured after 30 minutes using UV-vis spectroscopy at 280 nm.

In the second test, 0.10 g of the drug loaded hydrogel was immersed in PBS containing 500 mg dL⁻¹ glucose and the buffer was replaced by PBS containing 100 mg dL⁻¹ glucose. The buffer exchange was repeated three times to simulate an on/off switch (Fig. 9).

Drug release kinetics

Drug release has been described using different kinetic models. In this research, the kinetics of the insulin release from the intelligent drug delivery system were interpreted using various models including zero-order, first-order, Higuchi, Peppas and Hixson–Crowell.

Higuchi described the release of drugs from an insoluble matrix as a mathematical model based on Fickian diffusion. The release of the drug, following the Higuchi model, can be expressed with the equation:³⁰

Q_t = k_nt^1/2 (5)

where, Q_t is the amount of drug released in time t, and k_n is the Higuchi constant. For this model, a plot of the percentage of drug released versus the square root of time is linear.

The Hixson–Crowell cube root law eqn (6), describes release from systems where there is a change in the surface area and diameter of the particles.³¹

Q_t = [1001/3 − (kt)]³ (6)

where, Q_t is the amount of drug released in time t and k is the Hixson–Crowell constant. For this model, the percentage of drug release versus the cube root of time is linear.

Peppas et al. derived a simple relationship which describes drug release from a polymeric system, which was reported by Dash et al. and Saeedi et al.^32,33 The model is explained using eqn (7):

Q_t = k_nt^1/2 (7)

where, Q_t is the amount of drug released in time t, k is a constant, and n is the diffusional exponent, which characterizes the type of release mechanism. From this equation, drug transport is classified as Fickian (n = 0.5), case II transport (n = 1), non Fickian or anomalous (0.5 < n < 1), and super case II (n > 1). For the Peppas model, a plot of the percentage drug released versus log t is linear.

Results and discussion

A chitosan–PEG–AAm semi-IPN hydrogel was synthesized by free radical cross-linking polymerization using hydrogen peroxide as an initiator and MBA as a crosslinker. The formation of semi-IPN hydrogels from polysaccharides (e.g. chitosan), acrylic monomers, and PEG is well-known and has been reported by many authors.^34–37 The intelligent drug delivery system for insulin delivery was prepared by immobilizing glucose oxidase, as a specific glucose biosensor, and insulin, and loading them into the hydrogel via a swelling–diffusion method. Catalase was also immobilized in the system, which plays the role of an oxygen generator to support the activity of GOx and reduces the hydrogen peroxide. This insulin delivery system which has the ability to release insulin continuously and intelligently in response to the glucose concentration could be used as an artificial pancreas.

The effect of PEG was investigated to improve the semi-IPN hydrogels. Increase of the PEG content results in a significant increase in the SR, DLC and EE, which leads to an increase in insulin release. The results are summarized in Table 2, and illustrated in Fig. 2 and 3.

Table 2 Effect of PEG on the loading characteristics of the semi-IPN hydrogels

Code no.	Chitosan (g dL^−1)	PEG (g dL^−1)	SR (%)	EE (%)	DLC (%)
a	3.33	1.00	141	79	40.2
b	3.33	0.86	112	66	38.2
c	3.33	0.60	95	59	26.1
d	3.33	0.40	73	47	19.6
e	3.33	0.20	58	33	16.2
f	3.33	0.06	48	28	13.1
g	3.33	0.03	21	23	10.4

---

Fig. 2 Swelling ratios of the semi-IPN hydrogels with different PEG content, in PBS at room temperature.

Fig. 3 In vitro release profile for insulin from the semi-IPN hydrogels with different PEG content at room temperature.

A factorial design was also applied to optimize the hydrogel loading characteristics. The results are depicted in Table 3. It can be seen from the table that G7 shows the optimum loading characteristics for the hydrogel. Comparison of the G7 results with the other data in the table (e.g. for G3), clearly indicates that an increase in the drug/hydrogel ratio along with a decrease in the stirring speed leads to a significant increase in the SR (i.e. from 65.2 to 158), and EE (i.e. from 53.6 to 74.8). This can be related to the diffusion of insulin and oxygen through the hydrogel and polymer solution, respectively. A higher drug/hydrogel ratio has a positive effect, while a higher stirring speed has a negative effect on the SR and EE. Atmospheric oxygen diffusion through the polymer solution with a high stirring speed is responsible for this effect. The adverse effect of oxygen on the polymerization of acrylic systems is well known and referred to as oxygen inhibition. Molecular oxygen has a biradical nature, and in propagation steps the radicals in growing polymer chains can be trapped by the oxygen, resulting in ineffective radical propagation.

Table 3 Results of the factorial design^a

Code no.	Variable level in gel formation			EE (%)	SR (%)	t75 (min)
	X1	X2
a X1: drug/hydrogel ratio (g mL^−1), at low (1 : 1), mid (2 : 1), or high (4 : 1); X2: stirring speed (rpm), at low (400), mid (800), or high (1200). Low = −1, mid = 0, and high = 1.
G1	−1		−1	61.2	77.2	190
G2	−1		0	59.2	65.8	187
G3	−1		1	53.6	65.2	164
G4	0		−1	70.1	128	175
G5	0		0	68.2	107	181
G6	0		1	65.3	84.3	201
G7	1		−1	74.8	158	287
G8	1		0	69.5	111	247
G9	1		1	64.6	104	213

---

FTIR spectra of the chitosan and the chitosan semi-IPN hydrogel before and after the loading of insulin are shown in Fig. 4. The main characteristic absorption bands of pure chitosan appear at 1660 cm⁻¹, 1590–1610 cm⁻¹, 3350–3550 cm⁻¹, and 1040–1150 cm⁻¹, which represent C=O stretching, NH angular deformation, OH hydroxyl groups, and the C–O–C in glycosidic linkages, respectively (Fig. 4a). A new absorption band in the spectrum of the semi-IPN hydrogel at 1680 cm⁻¹ is attributed to the C=O stretching mode of amide groups, verifying the formation of chitosan-g-acrylamide (Fig. 4b). In the spectrum of the insulin loaded semi-IPN hydrogel, the band at 1580–1730 cm⁻¹ for the C=O of amide groups has become broader due to the presence of insulin. Another peak at 1180 cm⁻¹ is related to the C–O stretching mode of the insulin structure (Fig. 4c). The stretching band of –NH overlapped with the –OH stretching band of the chitosan. These observations prove that the insulin was loaded into the hydrogel.

Fig. 4 FTIR spectra of the (a) CS, (b) CS semi-IPN hydrogel, and (c) insulin loaded CS semi-IPN hydrogel.

Thermogravimetric analysis (TGA) results are shown in Fig. 5. The initial weight loss up to 120 °C is due to water evaporation from the hydrogel. The main degradation of the semi-IPN hydrogel starts at temperatures above 300 °C. As can be seen from Fig. 5, about 35% weight loss takes place in the temperature range of 300–500 °C. The values of T₅₀ are 466 °C, and 352 °C for the semi-IPN hydrogel and the insulin loaded semi-IPN hydrogel, respectively. A weight loss at 256 °C in the thermogram of the insulin loaded hydrogel was associated with the temperature of insulin decomposition (Fig. 5b). A similar behavior was reported by Shyong et al.¹² Two main degradation steps were observed in the DTG thermogram of the semi-IPN hydrogel that are related to the decomposition of the semi-IPN hydrogel (Fig. 6a). The DTG observations also indicated that the decomposition of insulin starts at 180 °C and reaches a maximum at 270 °C.

Fig. 5 TGA thermograms of the (a) semi-IPN hydrogel and (b) insulin loaded semi-IPN hydrogel.

Fig. 6 DTG thermograms of the (a) semi-IPN hydrogel and (b) insulin loaded semi-IPN hydrogel.

The results of CHN analysis for the chitosan, hydrogel and drug loaded hydrogel are given in Table 4. The increase of the nitrogen percentage of the hydrogel compared to the chitosan indicates that the acrylamide monomer was grafted onto the chitosan backbone.

Table 4 Elemental analysis of the chitosan, semi-IPN hydrogel, and drug loaded hydrogel

Compound	W(C)%	W(H)%	W(N)%
Chitosan	40.480	7.001	6.690
Hydrogel	40.103	6.900	12.270
Drug loaded hydrogel	36.939	6.181	11.612

---

HPLC results for the optimized hydrogel show that the amount of unloaded insulin in the loading medium was 14% of the initial value, which confirmed that 86% of the insulin was loaded into the hydrogel (Fig. 7).

Fig. 7 HPLC chromatograms of (a) the insulin in the loading medium before loading, (b) an insulin control, 5.0 (v/v%), and (c) the insulin in the loading medium after loading.

The surface morphology of the semi-IPN hydrogel was studied using scanning electron microscopy. The SEM micrographs show that the synthesized hydrogel has a good porosity (Fig. 8a), whereas the pores in the insulin loaded hydrogel were fully occupied (Fig. 8b).

Fig. 8 SEM images of the (a) chitosan semi-IPN hydrogel and (b) insulin loaded chitosan semi-IPN hydrogel at magnification scales of 1000×.

The MTT assay demonstrated that this material provides a viable surface for connective tissue compared to the control. It should be mentioned that these experiments cover only a fraction of the necessary biocompatibility testing, and more investigations are needed for medical application. The cell viability values indicated that the semi-IPN hydrogels are biocompatible, with cell viability values of 75% and over. The results show that the semi-IPN hydrogels have no toxic effects on the cell viability.

Fig. 9 demonstrates the sensitivity of the intelligent system to the glucose concentration using phosphate buffer saline (PBS) 1× at pH = 7. Fig. 9a reveals that an increase in the glucose level of the medium leads to an increase in insulin release. In this in vitro test, the amount of insulin released for a 500 mg dL⁻¹ glucose concentration was 150 units per 0.1 g of hydrogel. If a lesser amount of insulin is needed, we can adjust the amount of the insulin release using the following factors: variation of the PEG content, the crosslinker content, the loading amount, and also the amount of hydrogel in the implant. Fig. 9b clearly indicates that the insulin release is associated with glucose concentration. Like an on–off switch, the system acts as an artificial pancreas in which the release of insulin increases at high concentration and decreases significantly at low concentration of the glucose solution.

Fig. 9 Evaluation of the intelligent insulin delivery system. (a) Sensitivity of the system to the glucose level of the medium. (b) Relationship between the glucose concentration and insulin release, lines (i) and lines (ii) show the insulin release in a 500 mg dl⁻¹ and 100 mg dl⁻¹ glucose medium, respectively.

Kinetic studies were performed to investigate the release of insulin using an appropriate kinetic model, and the results are shown in Table 5. Kinetic modeling for different hydrogels with various amounts of PEG indicated that the best fitting model based on the highest regression is zero-order for the hydrogel with the highest PEG content (code a). Using Peppas model for the hydrogel indicated that the diffusional exponent is equal to 5.37 and demonstrated non Fickian diffusion.

Table 5 Results of kinetic modelling^a

Kinetic model	Parameter	Formulation code
		a	b	c	d	e	f	g
a Codes a–g are related to Table 2; k0, k1, kH, kp and kc are zero-order, first-order, Higuchi, Peppas and Hixon–Crowell rate constants, respectively; r is the regression coefficient and ss is the sum of squares.
Zero-order	k0	5.61	7.66	7.41	7.14	5.57	5.70	6.71
	r^2	0.989	0.983	0.972	0.963	0.953	0.607	0.746
	ss	2990.2	2763.1	2410.9	2391.5	1515.8	1213.0	1585.6
First-order	k1	−0.079	−0.082	−0.065	−0.060	−0.046	−0.023	−0.069
	r^2	0.943	0.960	0.940	0.923	0.974	0.637	0.771
	ss	42.30	49.29	49.20	49.19	49.09	49.13	49.37
Higuchi	kH	29.96	29.93	28.26	27.13	22.06	19.60	25.85
	r	0.949	0.979	0.923	0.909	0.976	0.771	0.727
	ss	2759.5	2717.7	2365.3	2345.9	1470.2	1767.2	3425.5
Peppas	kP	1.37	1.42	1.31	1.29	1.40	0.99	1.46
	r^2	0.937	0.988	0.890	0.881	0.981	0.771	0.576
	n	0.537	0.529	0.560	0.565	0.457	0.412	0.443
	ss	1.05	1.02	1.07	1.28	0.97	1.21	1.12
Hixon-Crowell	kC	−0.210	−0.214	−0.812	−0.171	−0.133	−0.099	−0.184
	r^2	0.965	0.975	0.957	0.942	0.970	0.647	0.772
	ss	51.01	51.01	50.53	50.46	49.18	50.06	51.56

---

Conclusion

An intelligent closed-loop insulin delivery system for implantation was designed in order to develop an insulin delivery system for type I diabetes patients. For this purpose, a semi-IPN hydrogel was synthesized by free radical polymerization using chitosan (CS), acrylamide (AAm), and polyethylene glycol (PEG). To make the system intelligent, glucose oxidase (GOx) and catalase (CAT) were loaded into the hydrogel. Characterization of the intelligent hydrogel was investigated using various techniques such as FT-IR, HPLC, TGA/DTG, and SEM. The cell viability values indicate that the semi-IPN hydrogels are biocompatible, with cell viability values of more than 75%, and have no toxic effects on the cell viability. The results show that an increase in the PEG content results in a significant increase in the SR, DLC and EE. An optimization of the loading characteristics of the hydrogel was also carried out using a factorial design. It was found that the intelligent insulin delivery system has a good responsiveness and sensitivity to the glucose concentration, like an on-off switch. Kinetic modeling for different hydrogels with various amounts of PEG indicated that the best fitting model based on the highest regression is zero-order for the hydrogel with the highest PEG content. Using Peppas model for the hydrogel indicated that the diffusional exponent is equal to 5.37 and demonstrated non Fickian diffusion.

Acknowledgements

The authors thank the Pishtazteb Zaman research and development Co. for the enzyme donations. We are also grateful to the authors of the references cited in this paper, which helped with improvement of the quality.

References

1. A. L. Holanda, F. M. V. Lia and R. Swarnakar, J. Biomater. Nanobiotechnol., 2014, 5, 211.
2. D. R. Owens, Nat. Rev. Drug Discovery, 2002, 1, 529.
3. D. Elleri, D. B. Dunger and R. Hovorka, BMC Med., 2011, 9, 120.
4. Z. Gu, T. T. Dang, M. Ma, B. C. Tang, H. Cheng, S. Jiang, Y. Dong, Y. Zhang and D. G. Anderson, ACS Nano, 2013, 7, 6758–6766.
5. T. Buranachai, N. Praphairaksit and N. Muangsin, AAPS PharmSciTech, 2010, 11, 1128.
6. B. Sarmento, S. Martins, A. Ribeiro, F. Veiga, R. Neufeld and D. Ferreira, Int. J. Pept. Res. Ther., 2006, 12, 131.
7. M. K. L. Chu, J. Chen, C. R. Gordijo, S. Chiang, A. Ivovic, K. Koulajian, A. Giacca, X. Y. Wuand and Y. Sun, Lab Chip, 2012, 12, 2533.
8. K. Zhang, C. Quan, H. Huang, N. Taulier and X. Y. Wu, J. Pharmacol. Pharmacother., 2004, 56, 611.
9. J. R. Azevedo, R. H. Sizilio, M. B. Brito, A. M. B. Costa, M. R. Serafini, A. A. S. Araújo, M. R. V. Santos, A. A. M. Lira and R. S. Nunes, J. Therm. Anal. Calorim., 2011, 106, 685.
10. H. Tan, H. Luan, Y. Hu and X. Hu, Macromol. Res., 2013, 21, 392.
11. M. Adimi, H. Attar, A. Barati, A. Seifkordi and M. Bakhtiari, Int. J. Biosci., 2014, 5, 129.
12. Y. J. Shyong, R. F. Lin, H. S. Soung, H. H. Wei, Y. S. Hsueh, K. C. Chang and F. H. Lin, J. Mater. Chem. B, 2015, 3, 2331.
13. M. Rajan and V. Raj, Int. J. Pharm. Pharm. Sci., 2012, 4, 255.
14. Z. Gu, A. A. Aimetti, Q. Wang, T. T. Dang, Y. Zhang, O. Veiseh, H. Cheng, R. S. Langer and D. G. Anderson, ACS Nano, 2013, 7, 4194.
15. V. Ravaine, C. Ancla and B. Catargi, J. Controlled Release, 2008, 132, 2.
16. K. M. Bratlie, R. L. York, M. A. Invernale, R. Langer and D. G. Anderson, Adv. Healthcare Mater., 2012, 1, 267.
17. R. Jayakumar, D. Menon, K. Manzoor, S. V. Nair and H. Tamura, Carbohydr. Polym., 2010, 82, 227.
18. M. N. Kumar, R. A. Muzzarelli, C. Muzzarelli, H. Sashiwa and A. J. Domb, Chem. Rev., 2004, 104, 6017.
19. S. Sajeesh and C. P. Sharma, J. Biomed. Mater. Res., Part B, 2006, 76, 298.
20. M. Zhang, X. H. Li, Y. D. Gong, N. M. Zhao and X. F. Zhang, Biomaterials, 2002, 23, 2641.
21. M. M. Amiji, Carbohydr. Polym., 1997, 32, 193.
22. Y. Hu, X. Jiang, Y. Ding, L. Zhang, C. Yang, J. Zhang, J. Chen and Y. Yang, Biomaterials, 2003, 24, 2395.
23. K. Ishihara, M. Kobayashi, N. Ishimaru and I. Shinohar, Polym. J., 1984, 16, 625.
24. X. Zheng, Y. Yin, W. Jiang, L. Xing and J. Pu, BioResources, 2015, 10, 2338.
25. K. L. B. Chang, M. C. Tai and F. H. Cheng, J. Agric. Food Chem., 2001, 49, 4845.
26. A. Trusek-Holownia and A. Noworyta, Biotech Rep., 2015, 6, 13.
27. T. Traitel, Y. Cohen and J. Kost, Biomaterials, 2000, 21, 1679.
28. X. Gao, Y. Cao, X. Song, Z. Zhang, X. Zhuang, C. He and X. Chen, Macromol. Biosci., 2014, 14, 565.
29. A. Senthil, T. Sivakumar and V. B. Narayanaswamy, Pharm. Lett., 2011, 3, 496.
30. M. Rani, A. Agarwal and Y. S. Negi, BioResources, 2010, 5, 2765.
31. Y. Fu and W. J. Kao, Expert Opin. Drug Delivery, 2010, 7, 429.
32. S. Dash, P. N. Murthy, L. Nath and P. Chowdhury, Drug Res., 2010, 67, 217.
33. M. Saeedi, J. Akbari, R. Enayatifard, K. Morteza-Semnani, M. Tahernia and H. Valizadeh, Iran. J. Pharm. Res., 2009, 8, 241–249.
34. S. P. Zhao, L. Y. Li, M. J. Cao and W. L. Xu, Polym. Bull., 2011, 66, 1075.
35. D. A. Rennerfeldt, A. N. Renth, Z. Talata, S. H. Gehrke and M. S. Detamore, Biomaterials, 2013, 34, 8241.
36. N. Rodkate, U. Wichai, B. Boontha and M. Rutnakornpituk, Carbohydr. Polym., 2010, 81, 617–625.
37. Z. Amoozgar, T. Rickett, J. Park, C. Tuchek, R. Shi and Y. Yeo, Acta Biomater., 2012, 8, 1849.

---