Biomaterials 34 (2013) 9678e9687

Contents lists available at ScienceDirect

Biomaterials
journal homepage: www.elsevier.com/locate/biomaterials

Intestinal mucosa permeability following oral insulin delivery using core shell corona nanolipoparticles
Xiuying Li a, b, Shiyan Guo a, Chunliu Zhu a, Quanlei Zhu a, Yong Gan a, *, Jukka Rantanen b, Ulrik Lytt Rahbek c, Lars Hovgaard d, Mingshi Yang b, **
a

Shanghai Institute of Materia Medica, Chinese Academy of Sciences, 501 Haike Road, Shanghai 201203, China Department of Pharmacy, Faculty of Health and Medical Sciences, University of Copenhagen, Copenhagen 2100, Denmark ADME Department, Novo Nordisk A/S, Maalov 2760, Denmark d Oral Formulation Development, Novo Nordisk A/S, Malov 2760, Denmark
b c

a r t i c l e i n f o
Article history: Received 28 July 2013 Accepted 19 August 2013 Available online 7 September 2013 Keywords: Oral protein delivery Core shell corona Insulin Mucus penetrating particles Cellular uptake

a b s t r a c t
Chitosan nanoparticles (NC) have excellent capacity for protein entrapment, favorable epithelial permeability, and are regarded as promising nanocarriers for oral protein delivery. Herein, we designed and evaluated a class of core shell corona nanolipoparticles (CSC) to further improve the absorption through enhanced intestinal mucus penetration. CSC contains chitosan nanoparticles as a core component and pluronic F127-lipid vesicles as a shell with hydrophilic chain and polyethylene oxide PEO as a corona. These particles were developed by hydration of a dry pluronic F127-lipid lm with NC suspensions followed by extrusion. Insulin nested inside CSC was well protected from enzymatic degradation. Compared with NC, CSC exhibited signicantly higher efciency of mucosal penetration and, consequently, higher cellular internalization of insulin in mucus secreting E12 cells. The cellular level of insulin after CSC treatment was 36-fold higher compared to treatment with free insulin, and 10-fold higher compared to NC. CSC signicantly facilitated the permeation of insulin across the ileum epithelia, as demonstrated in an ex vivo study and an in vivo absorption study. CSC pharmacological studies in diabetic rats showed that the hypoglycemic effects of orally administrated CSC were 2.5-fold higher compared to NC. In conclusion, CSC is a promising oral protein delivery system to enhance the stability, intestinal mucosal permeability, and oral absorption of insulin. 2013 Elsevier Ltd. All rights reserved.

1. Introduction Despite rapid progress made in the development of modern drug delivery technologies, an efcient oral delivery of therapeutic proteins and peptides remains to be achieved. Oral delivery of protein drugs faces several barriers, including pre-systemic degradation, limited mucosal diffusion, and poor intestinal epithelial membrane permeability [1e3]. A variety of innovative approaches have been developed to tackle these challenges, including the use of small molecule permeation enhancers, enzyme inhibitors, and the encapsulation of protein drugs into microspheres or nanoparticles [4,5]. The development of nanoparticles with biodegradable and non-toxic polymers has become one of the major focal areas in this eld due to their ability to protect proteins
* Corresponding author. Tel.: 86 21 20231000 1424; fax: 86 21 20231000 1425. ** Corresponding author. Tel.: 45 35336141. E-mail addresses: simm2122@vip.sina.com (Y. Gan), mingshi.yang@sund.ku.dk (M. Yang). 0142-9612/$ e see front matter 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.biomaterials.2013.08.048

from degradation in harsh pH environment, by enzymes in the gastrointestinal (GI) tract [6e9], and their ability to modulate physicochemical characteristics, drug release, and biological behavior [10,11]. Among those polymeric nanoparticles, chitosan and chitosan derivative based nanocarriers (NCs), appear to be particularly promising [12]. They can signicantly enhance the oral absorption of peptides [13] and have excellent capacity for protein entrapment and low toxicity [14]. They enhance the oral absorption of proteins by several mechanisms, such as bioadhesion with the negative charged cell membrane, transient widening of tight junctions and increasing cell permeability by affecting paracellular and intracellular pathways without changing junctional morphology [15]. However, they cannot effectively transport therapeutic proteins across the mucus barrier due to the electrostatic interaction between the anionic mucus gel and cationic nanoparticles [16]. It also has been reported that the binding of NCs to the surfaces of epithelial cells and subsequent absorption-enhancing effects are signicantly reduced by mucus [16e18]. Although the adhesion of

X. Li et al. / Biomaterials 34 (2013) 9678e9687

9679 total amount of insulin added free insulin total amount of insulin added

NCs to mucus can slow particle transit time through the GI tract, and thus enhance drug absorption, the absorption enhancement is limited by mucus turnover [19]. Recent studies have revealed that nanoparticles coated with hydrophilic polymers exhibit decreased adhesion to mucus components [19]. Our previous work also shows that Pluronic F127 modied lipid vesicles have superior mucus penetration, compared to unmodied lipid vesicles [20]. To further improve the mucus penetration of chitosan nanoparticles, we attempted to design a core shell corona nanolipoparticles (CSC), using chitosan nanoparticles as a core component and F127-lipid vesicles as a shell with polyethylene oxide PEO chains as the corona. By enveloping the F127-lipid shell, the positively charged chitosan nanoparticles (NC) would be shielded, ensuring free diffusion of NC in mucus. Additionally, protein nested in the CSC core would be better protected from enzymatic degradation, and therefore the efcacy of drug delivery could be further enhanced. To the best of our knowledge, the CSC particles, which were designed to feature advantages of both NC and Pluronic F127-lipid vesicles, are the newly designed nano-size lipoparticles for oral protein delivery with signicantly enhanced mucus penetration. In the present study, insulin was chosen as a model biomolecule for testing the CSC system. Insulin has currently attracted great interest in developing protein delivery approach, due to its wide clinical use, but limited routes of administration. Herein, CSC was prepared by hydration of a polymer-lipid lm with NC suspension to form coreeshell structure and they were then characterized in terms of particle size, zeta potential, and morphology. Their ability to protect the encapsulated insulin from enzymatic degradation, mucus penetration property, and cell uptake efciency and permeability were evaluated in HT29-MTX-E12 (E12) cells. Pharmacological and pharmacokinetic studies were conducted in streptozotocin (STZ) induced diabetic Sprague-Dawley rats.
2. Materials and methods 2.1. Materials Protasan UP CL113 was purchased from Nova Matrix (Drammen, Norway). Egg phosphatidylcholine (EPC) was purchased from Q.P. Corp (Tokyo, Japan). Pluronic F127 was donated by BASF (Ludwigshafen, Germany). RPMI1640 medium, Alexa Fluor-555-labeled wheat germ agglutinin and 0.25% trypsin-0.53 mmol/L EDTA were purchased from Invitrogen (Ontario, Canada). Paraformaldehyde was purchased from Sinopharm Chemical Reagent Co., Ltd. (Shanghai, China). 2-(4amidinophenyl)-6-indolecarbamidine dihydrochloride (DAPI) was purchased from Beyotime Institute of Biotechnology (Jiangsu, China). All the other reagents were of analytical grade. Alexa 488 labeled insulin was synthesized by Novo Nordisk A/S (Copenhagen, Denmark). HT29-MTX-E12 (E12) cells (52nde56th passages) cultured for 14e18 days were supplied by the ADME Department of Novo Nordisk A/S. Caco-2 cell lines were obtained from the American Type culture collection (ATCC, Manassas, VA, USA). Cells from the 40th 43rd passages were used in the present study. 2.2. Preparation and characterization of nanocarriers 2.2.1. Preparation and characterization of chitosan nanoparticles Chitosan nanoparticles (NC) were prepared using a previously reported procedure [21], based on the ionotropic gelation of chitosan with sodium tripolyphosphate (TPP) where the positively charged amino groups of chitosan interact with the negative charged TPP. Briey, TPP (1.0 mg/mL) was added to chitosan solution (1.0 mg/mL) under magnetic stirring at room temperature to produce NC at a nal chitosan/TPP weight ratio of 6:1. The insulin-loaded NC were obtained by mixing the insulin solution in 0.01 M NaOH (3.75 mg/mL) with NC solution at a theoretical content of 30% (w/w) of insulin. Self-assembled NC were collected and washed thrice with distilled water by centrifugation at 16,000 g on a 10-mL glycerol bed for 30 min. The centrifuged NC were then re-dispersed in distilled water and stored at 4  C until use. The association efciency and loading capacity of insulin in NC were determined by subtracting the amounts of free insulin in supernatants quantied using high performance liquid chromatography (HPLC) from the amounts added, using the following equation:

Association efficiency Loading capacity

(1)

total amount of insulin added free insulin weight of nanoparticles

(2)

2.2.2. Preparation of core shell nanolipoparticles The homogeneous F127-lipid lm was prepared by drying a chloroform solution containing egg phosphatidylcholine with F127 (4:1, mol/mol) and then hydrating the lm with NC suspensions (lipids/NP 6:1, w/w) for 30 min at room temperature, followed by 6 rounds of extrusion through a polycarbonate membrane with 200-nm pores (Avestin Inc., Canada). As a control, core shell nanolipoparticles (CS) without hydrophilic corona was prepared by encapsulating NC in a pure lipid vesicle without F127. 2.2.3. Characterization of nanocarriers The mean particle sizes and zeta potential values of different nanocarriers were measured using a Malvern Zetasizer NanoZS (Malvern Instruments, London, U.K.). Each measurement was made in triplicate. A transmission electron microscope (TEM, CM-200, Philips, Netherlands) was used to observe their morphology. The samples were stained with an aqueous solution of phosphotungstic acid (1%, w/v) before observation and the images were taken at 160 kV. The stability of these nanocarriers in stimulated gastric uid (SGF) and stimulated intestinal uid (SIF) were tested by measuring the changes in the particle size and polydispersity index (PDI) after a 2-h incubation. 2.3. In vitro enzymatic degradation assay The stability of nanocarriers against enzymatic degradation was carried out using a procedure described previously [22]. In brief, trypsin (2500 IU/mg) and chymotrypsin (800 IU/mg) were dissolved in Tris buffer (pH 8.0) containing 1 mM CaCl2 at nal concentrations of 250 IU/mL and 50 IU/mL, respectively. Each enzyme solution (50 mL) was separately incubated with 950 mL of test formulations (NC and CSC) containing 5 IU/mL of insulin. These formulations were pre-incubated at 37  C for 15 min. 50 mL of each sample was taken at pre-determined intervals and the enzyme activity was terminated by addition of 50 mL ice cold acetonitrile solution containing 0.1% triuoroacetic acid. Samples were subsequently treated with Triton X-100 to remove the CSC and CS lipid shell to release insulin. The samples were then analyzed using HPLC to determine the amount of insulin. Plain insulin solution and insulin-loaded NC suspensions were used as control and treated samples under the same experimental conditions. 2.4. Cell-based assays 2.4.1. Cytotoxicity of nanocarriers Cytotoxicity of various nanocarrier suspensions in Caco-2 cells was evaluated using the MTT assay. Briey, Caco-2 cells were seeded onto 96-well plates at a density of 1.0 104 cells per well. After a 48-h culture, nanocarrier suspensions were added to the culture media. After a 2-h incubation at 37  C, the suspensions were replaced by 100 mL of MTT solution (0.5 mg/mL in HBSS, pH 7.4) and incubated for an additional 3 h at 37  C. Subsequently, the MTT medium was removed and 150 mL of DMSO was added to dissolve the formazan crystals. The absorbance of the resultant solutions was measured at 490 nm using a microplate reader (Bio-Rad, USA). Each sample was analyzed in ve replicates. 2.4.2. Transport study in E12 cell monolayers The transport study was conducted following a previously reported procedure [23]. The E12 cell monolayers were washed thrice with pre-warmed 1 Dulbeccos phosphate buffered saline (PBS, pH 7.4) and equilibrated in the 1 HBSS (with Ca2 and Mg2, 25 mM D-glucose) at 37  C and 95% relative humidity for 30 min. Alexa 488 insulin stock solution (0.5 mg/mL) and the Alexa 488 insulin loaded nanocarrier suspensions were diluted with PBS. On the day of the experiment, the donor (apical) solutions were prepared by diluting an aliquot of these solutions into HBSS to make a nal Alexa 488 insulin concentration of 40 mg/mL. For all the transport experiments, a total of 200 mL of each acceptor (basolateral) sample was removed at 0, 15, 30, 45, 60, or 90 min after drug administration. For each acceptor sample taken, 200 mL of fresh HBSS was added to the mixture to maintain a constant volume. The samples were transferred onto a 96-well titer plate and Alexa 488 insulin was detected using a microplate uorometer (Model 680, Bio-Rad, USA) at an excitation lter of 485 nm and an emission lter of 530 nm. To eliminate the effects of mucus on the transport of free Alexa 488 insulin, the E12 cells were incubated with 10 mM N-acetyl cysteine (NAC) in HBSS for 30 min to remove the mucus prior to the transport study. Apparent permeability coefcient (Papp) and enhancement ratio (R) were calculated using the following equations. Papp dQ 1 dt A C0 (3)

9680

X. Li et al. / Biomaterials 34 (2013) 9678e9687

Table 1 Particle size, polydispersity index (PDI), z-potential, association efciency (AE), and loading efciency (LE) of NC, CS, and CSC. Formulations NC CS CSC Mean Diam (nm) 210.5 45.3 202.8 22.9 195.3 32.9 PDI 0.311 0.075 0.175 0.069 0.151 0.048

z potential (mV)
36.6 4.5 7.1 3.2 4.3 5.4

AE 76.6 5.8%

LE 39.0 4.2%

Papptest Pappcontrol

(4)

Sakura Fine technical Co., Ltd.). Frozen ileum sections (20 mm) were cut using a cryostat (Leica CM 1950) and then stained with DAPI. The tissue-sections were then visualized using a confocal microscope. 2.5.3. Pharmacological and pharmacokinetic studies Diabetes was induced in Male Sprague-Dawley rats weighting 180e200 g by an injection of streptozotocin (65 mg/kg) dissolved in citrate buffer as previously described [26]. The blood glucose level was determined using a glucose meter (On Call EZ II, Acon Biotech.Co.Ltd., Hangzhou, China). Rats were considered to be diabetic when their fasting blood glucose level was higher than 250 mg/dL one week after streptozotocin treatment. Sprague-Dwaley rats were fasted overnight before experiments, but allowed free access to water. Various formulations of insulin were administered to the diabetic rats by gavage of insulin solution (50 IU/kg), insulin loaded NC (50 IU/kg), or insulin loaded CSC (50 IU/kg) or by subcutaneous injection (SC) of insulin solution (5 IU/kg). Blood samples were collected from the tail veins of rats prior to drug administration and at various times (1, 2, 4, 6, 8 and 10 h) after dosing. The blood glucose levels were analyzed and the area under concentrationtime curve (AUC) over 12 h was calculated for each group. For the analysis of serum insulin level, blood samples were centrifuged (3000 rpm, 5 min) and subsequently quantied using an appropriate insulin ELISA kit (R&D system, Inc, MN, USA). The total decrease (D%) in serum glucose levels were calculated using a modied method as follows: D% AUCInsulin solution AUCtest 100 AUCInsulin solution (5)

Where dQ/dt is the ux of insulin from the apical side to the basolateral side, C0 is the initial concentration of insulin in the apical compartment, and A is the membrane area (cm2). Papp-test is the Papp of different groups, while Papp-control is the Papp of those treated with insulin solutions. 2.4.3. Mucus penetration in E12 cells analyzed by confocal microscopy In order to assess the interactions between Alexa 488 insulin and the mucus, the E12 cells grown on Transwell lter inserts for 14e17 days were treated as described above. The apical incubation buffer, containing Alexa 488 insulin loaded nanocarriers, was removed after 60 min of incubation with the formulations. The E12 monolayers were then washed with PBS and the mucus layer was stained with Alexa Fluor 555 labeled wheat germ agglutinin (Alexa Fluor 555-WGA, 10 mg mL1) for 10 min at 37  C [24]. The membranes supporting the cell layers were washed with PBS, and the cell layers supporting membranes of Transwell inserts were cut from the plastic support without xation, mounted onto microscope slides, and covered with coverslips. The slides were immediately observed under a confocal microscope (LSM 5 Pa, Zeiss, USA) using a 63 oil objective lens. Alexa Fluor 488 and Alexa Fluor 555 were excited with a 488 nm argon laser and 543 nm helium-neon laser, respectively. Image visualization and processing were performed using the LSM 5 Pa software. To observe the interactions between the different formulations with mucus in a larger view, a 2D image in the middle of the mucus was taken under a confocal microscope at 5 magnication. 2.4.4. Qualication of mucus entrapment and cellular uptake of Alexa 488 insulin To differentiate Alexa 488 insulin trapped in the mucus from bound insulin and insulin that had been taken up by cells, the E12 cells were treated with the different particle formulations for 1 h. The mucus layer was removed by washing the cell monolayers twice with 300 mL of 4% (v/v) formalin solution in PBS (0.1 M, pH 7.4), followed by the washing procedure as reported previously [17]. Then the cell layers were disrupted with lysis buffer (RAPI), and cell-associated uorescent protein was determined by uorescence spectroscopy. The amount of cell-associated Alexa 488 labeled insulin was reported as a percentage (%) of the initial amount of Alexa 488 labeled insulin. 2.5. Animal study 2.5.1. Animal care Male Sprague-Dawley rats, 6e8 weeks old, were provided by the Animal Experimental Center of Shanghai Institute of Materia Medica, Shanghai, China. The animals had free access to rat chow and tap water. All of the animal experiments were carried out according to the Institutional Animal Care and Use Committee (IACUC) guidelines of Shanghai Institute of Materia Medica. 2.5.2. Absorption studies in the ligated intestinal loops 2.5.2.1. Permeation studies ex vivo. The transport of Alexa 488 insulin loaded CSC and NC across the epithelial mucosa of rat ileum was monitored ex vivo, as described by Yin et al. [25]. Briey, after rats were sacriced, 5-cm sections of the ileum were excised and incubated in 10 mL Kreb-Ringer buffer at 37  C with gentle agitation. CSC and NC (0.5 mL) were loaded into the loops using a syringe. At various times (0.5, 1, 1.5 and 2 h), 200 mL of the incubation buffer was collected for determination of Alexa 488 labeled insulin using uorescence spectroscopy. The same volume of fresh buffer was added at each time point. All of the experiments were performed in triplicates. 2.5.2.2. Absorption assay in vivo. The in vivo uptake of CSC and NC were evaluated using the ligated intestinal loops model following the method described by Yun et al. [9]. Sprague-Dwaley rats were fasted overnight before experiments, but allowed free access to water. The rats were anesthetized with pentobarbital sodium (0.04 mg/kg). After the abdomen was exposed, 5-cm loops of ileum were made by ligation at both ends and the tissue was washed with physiological saline. CSC and NC (0.5 mL, 100 mg/mL of Alexa 488 insulin) were injected into the loop. To study the effect of mucus on the absorption of free Alexa 488 insulin, ligated ileum loops were preincubated with saline containing 10% N-acetyl cysteine (NAC) for 10 min before the absorption study to remove mucus, and then 0.5 mL CSC and NC suspensions were injected into the loop. Rats were sacriced after 0.5 h or 2 h and the sections of each loop were removed and extensively washed using PBS. Subsequently, the removed loops were xed by 4% paraformaldehyde for 2 h and immersed in 30% sucrose at 4  C overnight. Samples were embedded and frozen at 20  C (OCT,

The relative bioavailability (F%) of CSC after oral administration was calculated using the following formula. F% AUCoral Dosesc 100 AUCsc Doseoral (6)

2.6. Statistical analysis Comparisons between the two groups were performed using Students t-test (SPSS, Chicago, IL) and P < 0.05 was considered statistically signicant.

3. Results 3.1. Characterization of nanocarriers As shown in Table 1, the naked NC were around 210 45 nm in size and had a positive zeta potential of approximately 36.6 4.5 mV. The mean association efciency of insulin to NC was 76.6 5.8%, and the mean loading efciency was 39.0 4.2%. The incorporation of NC into vesicles did not have a remarkable effect on the size, but led to a signicant changes in zeta potential (P < 0.05), which were reduced to e 7.1 3.2 and e 4.3 5.4 mV for CS and CSC, respectively. The conversion of the zeta potential suggested that the NC were encapsulated in the vesicles. As shown in Fig. 1, the core shell corona structures were observed for CSC under TEM, indicating that chitosan nanoparticles were encapsulated into the vesicles under the current experimental conditions. A schematic presentation of the core shell corona structure is shown in Fig. 1A. CSC formed by this method remained stable in SGF and SIF, and the size and PDI did not change signicantly. CS was stable in SGF but the size greatly increased after a 2-h incubation in SIF. Whereas, NC was soluble in SGF and no particle size change could be observed in SIF (Data not shown). 3.2. Protection of insulin against enzymatic degradation No signicant difference was observed between CS and CSC with respect to protection of insulin from enzymatic degradation by trypsin and a-chymotrypsin. As shown in Fig. 2, CS and CSC

X. Li et al. / Biomaterials 34 (2013) 9678e9687

9681

Fig. 1. Design and evaluation of NC and CSC. A: Schematic illustration of the formation of NC and CSC; B: TEM images of NC and CSC. Red line, the boundary between NC and polymer-lipid shell. (For interpretation of the references to colour in this gure legend, the reader is referred to the web version of this article.)

protected 40% and 70% of encapsulated insulin, respectively, from degradation by trypsin and a-chymotrypsin after 1 h of incubation with the intestinal enzyme solutions in vitro, whereas plain insulin was almost completely degraded under the same conditions. This suggested that the core shell structures could prevent insulin from being exposed to the enzymes by shielding the protein in the core of the nanolipoparticles. NC alone, in this case, failed to protect the associated insulin from enzymatic degradation. Its degradation proles were similar to those of naked insulin. 3.3. Results from studies with cell-based models 3.3.1. In vitro cytotoxicity As shown in Fig. 3, all the three nanocarriers were found to have no signicant cytotoxicity (P > 0.05) at the tested concentrations (0.2, 0.5 and 1.0 mg/mL), although there was a slight reduction in cell viability treated with 1.0 mg/mL of NC. 3.3.2. Transport of Alexa 488 insulin loaded nanocarriers across E12 cells The apparent permeability coefcient (Papp) and enhancement ratio (R) were used to assess the ability of the nanocarriers to

promote transport of insulin across the E12 cell monolayers. As shown in Table 2, the Papp obtained with NC was quite similar to those obtained with insulin HBSS solution. However, after NAC treatment, the insulin Papp was increased 2-fold, indicating that the mucus formed a barrier for the transport of insulin. The use of CSC was found to signicantly enhance insulin Papp across the E12 monolayers. An R-value of 2.42 was obtained with CSC. This was comparable to the results obtained using plain Alexa 488 insulin HBSS solution in E12 monolayers after mucus depletion by NAC. The unmodied CS also enhanced insulin transport to a less extent as compared to CSC, with an R-value of 1.63. 3.3.3. Mucus penetration of Alexa 488 insulin in E12 cells In order to determine the mucus penetration properties of Alexa 488 insulin, the E12 cell monolayers treated with different formulations were imaged without xation to show the diffusion of Alexa 488 insulin. The red uorescence derived from Alexa Fluor 555 represents the mucus that covers the surface of the cell, and the green uorescent spots derived from Alexa 488 correspond to Alexa 488 insulin. Few green uorescent spots were observed in the monolayer treated with Alexa 488 insulin HBSS solution, but stronger green uorescence was observed in monolayers treated

9682

X. Li et al. / Biomaterials 34 (2013) 9678e9687

A 100
% undegradated insulin 80 60 40 20 0 0 20 40

S NC

CS CSC

B 100
% undegradated insulin 80 60 40 20 0

S NC

CS CSC

60 80 Time (min)

100

120

20

40

60 80 Time (min)

100

120

Fig. 2. Enzymatic degradation proles of insulin as a function of time A: Trypsin; B: a-chymotrpsin.

with NC, CS, and CSC (Fig. 4A). After NC treatment, the mucus layer was discontinuously distributed and showed strong yellow spots, but the mucus layers treated with CS and CSC were similar to those treated with Alexa 488 insulin solution, except for stronger green uorescence. The 2D images from the middle position of the mucus on the surface of the E12 cell monolayer (Fig. 4B) showed that NC closely interacts with mucus, forming aggregates with some of its components (big yellow spots) and some NC self-aggregates (large green spots). This phenomenon was seen much less in E12 cell monolayers treated with Alexa 488 insulin loaded CS and CSC and the aggregates were much smaller than those seen in the NC group. 3.3.4. Concentrations of Alexa 488 insulin in the mucus layer and cell layers To differentiate the fraction of Alexa 488 insulin trapped in the mucus layer and the fraction associated with, or internalized in the cells, the mucus layer was washed away using the formalin solution. As shown in Fig. 5, the CSC had the strongest interaction with E12 cells, reaching a maximum of about 9.3% of the initial Alexa 488 insulin in the mucus layer and cells. Among these particles, 5.4% were inside in or attached to the E12 cells. The CS had an effect similar to that of the CSC, with 8.6% of the initial Alexa 488 insulin in the mucus layer or the E12 cells, but only 1.9% was inside or attached to the E12 cells, far less than that with the CSC. The NC also

showed stronger association with the entire cell monolayer compared to the insulin solution, as suggested by the observation that 4.3% of the initial Alexa 488 insulin was detected in the E12 monolayers prior to removal of mucus, but only 0.5% of the initial Alexa 488 insulin was also detected inside the cell. These results suggested that the enhanced association efciency of insulin with cells was not as signicant as the association with the entire cell monolayer with intact mucus. 3.4. Results from the animal studies 3.4.1. The accumulative transport of insulin ex vivo The transport of NC and CSC across the intestinal epithelia was investigated using ex vivo in ligated ileum loops. As shown in Fig. 6, compared to NC, CSC increased the accumulative amount of Alexa 488 insulin permeated through the rat ileum. The permeated Alexa 488 insulin from CSC at 1 h was 21.75 0.86 ng, which was 1.53fold higher than that from NC (14.18 0.59 ng) and the accumulative amount of Alexa 488 insulin at 2 h increased to 61.74 7.39 ng, which was 1.76-fold higher compared to NC. These results revealed that absorption enhancement of CSC across the epithelia was associated with improved mucus penetration and cellular uptake. 3.4.2. Absorption studies in villi The absorption of NC and CSC was qualitatively observed by a confocal microscope to be located in the villi of ileum loops. Fig. 7 shows the absorption of Alexa 488 insulin from NC and CSC at 0.5 and 2.0 h, respectively. The absorption of Alexa 488 insulin increased with time, and stronger green uorescence was observed in loops treated with CSC. Alexa 488 insulin loaded CSC permeated deeply into villi after 2 h, indicating their effective absorption in vivo. With NAC pre-treatment to remove the mucus, both NC and CSC showed strong green uorescence intensity. The effect of
Table 2 Apparent permeability coefcient (Papp) and enhancement ratio (R) of insulin permeability across E12 cell monolayers. * P < 0.05, compared with S group. Formulations S S NAC NC CS NP CSC NP Papp*107(cm/s) 4.73 7.71 4.58 7.29 11.46 0.35 0.72* 0.53 0.89* 1.58* Enhancement ratio e 2.21 0.97 1.63 2.42

120 100 Cell Viability (%) 80 60 40 20 0

NC

CS

CSC

Control

0.2 mg/mL

0.5 mg/mL 1.0 mg/mL

Fig. 3. Effect of different formulations on the Caco-2 cell viability.

X. Li et al. / Biomaterials 34 (2013) 9678e9687

9683

Fig. 4. Mucus and cell-based evaluation of NC and CSC. A: Confocal microscope images of E12 cell layers treated with different formulations. A: 3D image of mucus penetration. B: 2D images of E12 monolayers. Red: mucus covering the cell surface stained with Alexa Fluor 555-WGA. Green: Alexa 488-insulin. (For interpretation of the references to colour in this gure legend, the reader is referred to the web version of this article.)

9684

X. Li et al. / Biomaterials 34 (2013) 9678e9687

Insulin bound to and within cells and mucus (% of initial)

12 10 8 6 4 2 0

mucus+cell cell

*
#

*
##

concentrations at 1e6 h post administration. As shown in Table 3, the AUC(0e12 h) for CSC was 232.5 35.5 mIU*h/mL, with a relative bioavailability of 7.8%; both parameters were signicantly greater compared with NC and plain insulin. 4. Discussion Pre-systemic enzymatic degradation and poor transmucosal permeability are the major obstacles to an efcacious oral protein and peptide delivery [3], which can be overcome by the use of nanoparticle carriers [7,10]. In the current study, we designed core shell corona structured nanolipoparticles (CSC), based on a particle-in-particle approach. The system was comprised of chitosan nanoparticles as a core component and F127-lipid vesicles as a shell, with hydrophilic chain polyethylene oxide (PEO) as a corona. These particles protect fragile biomacromolecules, such as insulin, in the polymer nanoparticle cores, enhancing mucus-penetrating properties and membrane transportation. CSCs were prepared by hydration of polymer-lipid dry lm with NC suspensions. NC were designed to be covered by the vesicles via hydrogen bonds between the particle surface and lipid polar heads and electrostatic interactions between lipids and the hydrophilic surface [27]. As expected, most NC particles were entrapped in the vesicle as observed by TEM, where the dense black cores of NC were surrounded by gray lipid shells (Fig. 1B). The sizes of CSC were around 200 nm, as shown in TEM. This was also in close agreement with the particle sizes indicated by dynamic light scattering measurements. Upon adsorption of lipid or polymer-lipid layers onto NC, the strong positive zeta potential of NC changed from 36 mV to negative charges (7.1 and 4.3 mV, respectively), which also showed that the cores of NC were encapsulated into the polymer-lipid vesicles and formed nanolipoparticles. Similar assembly of nanoparticles and vesicles has been reported by other groups [28,29]. However, most other carriers reported in the literature are macro-sized and do not feature a hydrophilic polymer chain on the surface, which are unstable in the GI tract and are not ideal for intestinal absorption. Fragile biomacromolecules, such as insulin, packaged in the nanolipoparticles with solid cores are expected to be able to withstand the harsh physiologic environment due to the shielding shell of the polymer-lipid layers, preventing direct contact of the biomacromolecules with enzymes. Our in vitro forced degradation study conrmed this hypothesis. The naked NC failed to protect insulin from enzymatic degradation; whereas, the enzymatic degradation of insulin associated with chitosan nanoparticles was signicantly decreased when they were encapsulated in vesicles. These results suggested that nanolipoparticles with core shell structures may improve the integrity of fragile biomolecules during oral delivery. Whether mucus is also a barrier to the diffusion of protein and peptide drugs is still debatable. The traditional theory is that the pore size of mucus physically constrains the diffusion of macromolecules [30,31]. However, in recent years, some emerging evidence has supported the opposite theory, that mucus is not a diffusion barrier for protein drugs [32,33]. In our study, only very little Alexa 488 insulin was detected in the mucus. As shown in the confocal microscope images (Fig. 4A), very weak green uorescence was visible, indicating that it was difcult for Alexa 488 insulin to pass through the mucin network. The qualitative data (Fig. 5) also conrmed that minimal Alexa 488 insulin was present in the mucus. In addition, the Papp value of insulin solution was signicantly increased after the mucus was removed by NAC treatment. These results supported the notion that mucus is a diffusion barrier to insulin and that it is necessary to consider drug delivery strategies that would overcome the physical barrier of the mucus layer.

NC

CS

CSC

Fig. 5. Relative amounts of Alexa 488 insulin internalized/attached in E12 cells. *P < 0.05, compared with NC group; #P < 0.05, compared with NC group; ##P < 0.01, compared with the NC group.

mucus on the absorption of insulin loaded into NC was much more signicant than those loaded into CSC. For the NC group, the green uorescence was much stronger with NAC treatment, whereas the difference after NAC treatment was less for the CSC group. 3.4.3. Hypoglycemic effect in vivo The pharmacological effects of NC and CSC were evaluated in diabetic rats after oral administration. As shown in Fig. 8A, the plain insulin solution failed to reduce the blood glucose level; NC slightly reduced the blood glucose level; and CSC exhibited stronger hypoglycemic effects than the other two treatments. Compared with the baseline, the blood glucose level decreased to 77% and the hypoglycemic effect lasted 10 h. The D% of CSC was 2.52-fold higher compared to NC. The serum insulin concentration-time proles are shown in Fig. 8B. The rats subcutaneously treated with plain insulin solution resulted in a rapid increase in serum insulin concentration, whereas oral administration of CSC demonstrated a slower rise in serum insulin concentration, reaching a maximum serum concentration at 4 h after treatment. Compared to NC and plain insulin solutions, CSC treatment showed signicant higher serum insulin

75
Amount of insulin permeated (ng)

NC CSC

50

25

0 0.5 1.0 Time (h)

Fig. 6. Accumulative permeated amount of Alexa 488 insulin across rat ileum from NC and CSC (Mean SD, n 3).

1.5

2.0

X. Li et al. / Biomaterials 34 (2013) 9678e9687

9685

Fig. 7. Distribution of Alexa 488 insulin loaded in NC or CSC in ileum villi at 0.5 and 2 h. Green uorescence refers to Alexa 488 insulin and blue uorescence refers to cell nucleus.

It has been reported that the absorption of insulin can be enhanced by using fusogenic liposomes (FLs) as carriers when they are directly administered to colonic and rectal loops [34]. However, the enhancement in absorption is signicantly decreased when administered to the ileum [34]. It has been speculated that the mucus layer overlying the ileal epithelia may impair or compromise the fusion of FL to the intestinal mucosa. It also has been reported that trimethyl chitosan (TMC) and chitosan do not enhance the absorption of insulin in the ileum because of the thick mucus layer [35]. In the current study, we investigated the mucus-penetrating properties of chitosan nanoparticles, which have strong electrostatic interactions with mucus. Strong green uorescence was observed in the mucus network of E12 cells after NC treatment, suggesting that NC could increase the amount of insulin trapped in mucus. However the ability of NC to enhance cellular uptake and cellular transport was less signicant than its ability to enhance mucus entrapment. These results suggested that NC could not readily improve the absorption. This could be because most of the NC is trapped in the mucus due to the strong electrostatic interactions between chitosan and mucin [16]. NC were unable to reach the epithelial surface, and so failed to improve the absorption of encapsulated proteins. Considering the mucus turnover in vivo, such a protein delivery system could be rapidly cleared together with the mucus. Therefore, an ideal pharmacological efcacy will be difcult to obtain. Unlike NC, CSC not only signicantly enhanced the mucus penetration of insulin but also greatly enhanced the efciency of cellular associations. As shown in the LSM images, the mucus of E12 cells treated with CSC was homogenously distributed. The cell monolayers treated with CSC showed strong green spots, indicating that more protein molecules were able to reach the cell surface and promote cellular uptake. In addition to their ability to enhance mucus penetration, CSC may also enhance cellular uptake of insulin by surface modication of the nanolipoparticles with F127 polymers. F127 has been reported to stabilize and seal damaged lipid bilayer membranes [36e 38]. This may reinforce the association of the nanolipoparticles with the cell membrane. The increased level of cellular insulin

uptake observed with CSC in E12 cells showed 10-fold higher uptake compared to NC. Although CSC have been found to improve insulin transport through E12 cells as compared to insulin solution and naked NC (Table 2), their impacts on the Papp values are not comparable to their effects on the efciency of cellular uptake. Similar results have been reported by Anchalee et al. and Thompson et al.: the chitosan nanocomplexes could improve cellular uptake of insulin with E12 cells, but no insulin was detected on the basolateral sides of the E12 cell monolayer [23,39]. In the present ex vivo transport study across ligated ileum loops, we found that CSC signicantly enhanced the permeability of Alexa 488 insulin compared with NC. When reaching the small intestine, NC were mostly immobilized in the mucin network, but CSC could penetrate through the mucus and thus more Alexa 488 insulin could reach the epithelium surface and be transported across the intestinal epithelium via the paracellular pathway, transcytosis or receptor-mediated transcytosis. The absorption enhancement effect of CSC was further conrmed by in vivo absorption. Alexa 488 insulin could clearly be observed in the villi after 0.5 h treatment of ligated ileum loops treated with CSC, while minimal Alexa 488 insulin could be seen in the NC group. The green uorescence in the CSC group was signicant stronger compared to the NC group at 2 h post administration of CSC. With NAC pre-treatment to remove the mucus, ligated loops treated with NC showed signicantly stronger green uorescence intensity compared to the non-NAC treated NC group, while only slightly stronger uorescence was observed in the CSC group compared to non-NAC treated CSC group. These results indicated that the effect of mucus on the absorption of insulin loaded in NC was much more signicant than those loaded in CSC, suggesting enhanced mucus penetration of NC by polymer-lipid vesicle encapsulation. The in vivo pharmacological and pharmacokinetic results showed good correlations with the improved absorption and transport of insulin in the ligated intestinal loop and cell studies in vitro. As expected, the oral administration of plain insulin solution did not lead to signicant hypoglycemic effects, because of the rapid degradation in the gastric uid and poor permeability through

9686

X. Li et al. / Biomaterials 34 (2013) 9678e9687

A 120
Blood glucose level (% of initial) 100 80 60 40 20 0 0 2 4

S S-SC

NC CSC

B 160
Serum insulin level (IU/mL) 140 120 100 80 60 40 20 0 0 2 4

6 8 Time (h) S S-SC NC CSC

10

12

administration and lasted for 10 h. During this experiment, the fasting food blood glucose did not return to the initial level after 10 h, similar as other previous reports, which might be the consequence of the combined effects of hunger and hypoglycemic agents [9,21]. The quick absorption of insulin suggested the fast diffusion of CSC through the mucosal barrier while the long-acting effect revealed the slow release of insulin from the core shell corona structure. Furthermore, the animals treated with CSC exhibited signicantly higher serum insulin levels with a signicantly increased relative bioavailability compared with the NC and free insulin groups. In the clinical practice, sc. administration of insulin can be associated with severe hypoglycemic shock and patient inconvenience. Our results demonstrated that insulin encapsulated in CSC exhibited prolonged hypoglycemic activity without the initial hypoglycemia seen with sc. administration, suggesting that CSC may be a promising strategy for clinical treatment of diabetes. 5. Conclusions In the present study, we designed core shell corona nanolipoparticles to improve the intestinal mucosal permeability of insulin. The insulin loaded chitosan nanoparticles formed the core, and polymer-lipid vesicles formed the shell, with hydrophilic chain PEO as the corona. Experiments performed with mucus-secreting HT29-MTX-E12 cells suggested that mucus might constitute a diffusion barrier for insulin molecules and prevent them from reaching the cell membrane. The naked NC were found to increase the amount of protein entrapped in the mucus, while CSC improved the ability of insulin to penetrate the mucus and increased cellular uptake of insulin. Enhanced permeability of insulin was conrmed by a transport study using ex vivo intestinal tissue. Enhanced absorption of insulin in intestinal villi by CSC was also observed in in vivo absorption study. The CSC treatment resulted in an elevated serum insulin concentration and improved hypoglycemic effect in diabetic rats, suggesting that NC encapsulated in F127-lipid vesicles could be a promising nanocarrier for oral delivery of insulin, as well as other peptides and proteins. Acknowledgments This work was supported by the Novo Nordisk-Chinese Academy of Science (CAS) Research Foundation (No. NNCAS-2009-10) and the National Science and Technology Major Project, Key New Drug Creation and Manufacturing Program (2012ZX09301001-001). We thank Erik Wisaeus and Pia Wahlberg of the Danish Technological Institute, Denmark, for their excellent assistance on TEM, which is supported by The Innovation Consortium NanoMorph (952320/ 2009) funded by The Danish Council for Technology and Innovation. We also thank the ADME department of Novo Nordisk A/S, Denmark, for excellent assistance in HT29-MTX-E12 cell studies. References

6 8 Time (hour)

10

12

Fig. 8. A Blood glucose levels in diabetic rats following oral administration of insulin loaded NC, CSC, and insulin solution (S, 50 IU/kg) and subcutaneous injection of insulin solution (S-SC, 5.0 IU/kg) (Mean SD, n 4). *P < 0.05, compared with S group; **P < 0.01, compared with S group; #P < 0.05, compared with the NC group. B: Serum insulin level in diabetic rats following oral administration of insulin loaded NC, CSC and insulin insulin solution (S, 50 IU/kg) and subcutaneous injection of insulin solution (S-SC, 5.0 IU/kg) as positive control (Mean SD, n 4). *P < 0.05, compared with the NC group.

epithelia. Insulin loaded NC had limited efciency to reduce the blood glucose level, partially because of the instability of NC as seen in in vitro studies. Strong adhesion to the mucus, instead of penetrating through the mucus, also contributed to the ineffectiveness of insulin loaded NC. Whereas, CSC exhibited improved stability in the GI tract, enhanced mucus penetration, and membrane transport, leading to signicantly more potent and prolonged pharmacological efcacies. The onset of hyperglycemic effect occurred at 2 h post

Table 3 Pharmacokinetic parameters of insulin in diabetic rats after subcutaneous injection of insulin solution (5.0 IU/kg) and oral administration of insulin solution, NC and CSC (50 IU/kg). Cmax: maximum serum concentration; Tmax: time at which Cmax is attained; FR: relative bioavailability. INS solution (s.c.) Dose (IU/kg) AUC (mIU*h/mL) Cmax (mIU/mL) Tmax (h) FR % 5.0 299.3 49.2 128.6 22.5 1 100% CSC (i.g.) 50.0 232.5 35.5 40.0 10.1 3.3 7.8% CS (i.g.) 50.0 111.4 19.3 16.3 3.7 3 3.7% INS solution (i.g.) 50.0 36.3 4.9 6.0 0.5 2.7 1.2%

[1] Khafagy ES, Morishita M, Onuki Y, Takayama K. Current challenges in noninvasive insulin delivery systems: a comparative review. Adv Drug Deliv Rev 2007;59:1521e46. [2] Soares S, Costa A, Sarmento B. Novel non-invasive methods of insulin delivery. Expert Opin Drug Deliv 2012;9:1539e58. [3] Hamman JH, Enslin GM, Kotze AF. Oral delivery of peptide drugs-barriers and developments. Biodrugs 2005;19:165e77. [4] Park K, Kwon IC, Park K. Oral protein delivery: current status and future prospect. React Funct Polym 2011;71:280e7. [5] des Rieux A, Fievez V, Garinot M, Schneider YJ, Preat V. Nanoparticles as potential oral delivery systems of proteins and vaccines: a mechanistic approach. J Control Release 2006;116:1e27. [6] Card JW, Magnuson BA. A review of the efcacy and safety of nanoparticlebased oral insulin delivery systems. Am J Physiol Gastroint Liver Physiol 2011;301:G956e67.

X. Li et al. / Biomaterials 34 (2013) 9678e9687 [7] Woitiski CB, Carvalho RA, Ribeiro AJ, Neufeld RJ, Veiga F. Strategies toward the improved oral delivery of insulin nanoparticles via gastrointestinal uptake and translocation. Biodrugs 2008;22:223e37. [8] Colon JL, Diaz A, David A, Perez R, Gonzalez ML, Baez A, et al. Nanoencapsulation of insulin into zirconium phosphate for oral delivery applications. Biomacromolecules 2010;11:2465e70. [9] Jin Y, Song Y, Zhu X, Zhou D, Chen C, Zhang Z, et al. Goblet cell-targeting nanoparticles for oral insulin delivery and the inuence of mucus on insulin transport. Biomaterials 2012;33:1573e82. [10] Huang X, Li L, Liu T, Hao N, Liu H, Chen D, et al. The shape effect of mesoporous silica nanoparticles on biodistribution, clearance, and biocompatibility in vivo. ACS Nano 2011;5:5390e9. [11] Bakhru SH, Furtado S, Morello AP, Mathiowitz E. Oral delivery of proteins by biodegradable nanoparticles. Adv Drug Deliv Rev 2013;65:811e21. [12] Chen MC, Sonaje K, Chen KJ, Sung HW. A review of the prospects for polymeric nanoparticle platforms in oral insulin delivery. Biomaterials 2011;32: 9826e38. [13] Garcia-Fuentes M, Alonso MJ. Chitosan-based drug nanocarriers: where do we stand? J Control Release 2012;161:496e504. [14] de la Fuente M, Ravia M, Paolicelli P, Sanchez A, Seijo B, Alonso MJ. Chitosanbased nanostructures: a delivery platform for ocular therapeutics. Adv Drug Deliv Rev 2010;62:100e17. [15] Canali MM, Pedrotti LP, Balsinde J, Ibarra C, Correa SG. Chitosan enhances transcellular permeability in human and rat intestine epithelium. Eur J Pharm Biopharm 2012;80:418e25. [16] Schipper NGM, Vrum KM, Stenberg P, Ocklind G, Lennerns H, Artursson P. Chitosans as absorption enhancers of poorly absorbable drugs: 3: inuence of mucus on absorption enhancement. Eur J Pharm Biopharm 1999;8:335e43. [17] Behrens I, Pena AI, Alonso MJ, Kissel T. Comparative uptake studies of bioadhesive and non-bioadhesive nanoparticles in human intestinal cell lines and rats: the effect of mucus on particle adsorption and transport. Pharm Res 2002;19:1185e93. [18] Sogias IA, Williams AC, Khutoryanskiy VV. Why is chitosan mucoadhesive? Biomacromolecules 2008;9:1837e42. [19] Lai SK, Wang YY, Hanes J. Mucus-penetrating nanoparticles for drug and gene delivery to mucosal tissues. Adv Drug Deliv Rev 2009;61:158e71. [20] Li XY, Chen D, Le CY, Zhu CL, Gan Y, Hovgaard L, et al. Novel mucuspenetrating liposomes as a potential oral drug delivery system: preparation, in vitro characterization, and enhanced cellular uptake. Int J Nanomed 2011;6:3151e62. [21] Sonaje K, Lin YH, Juang JH, Wey SP, Chen CT, Sung HW. In vivo evaluation of safety and efcacy of self-assembled nanoparticles for oral insulin delivery. Biomaterials 2009;30:2329e39. [22] Chalasani KB, Russell-Jones GJ, Yandrapu SK, Diwan PV, Jain SK. A novel vitamin B12-nanosphere conjugate carrier system for peroral delivery of insulin. J Control Release 2007;117:421e9.

9687

[23] Jintapattanakit A, Junyaprasert VB, Kissel T. The role of mucoadhesion of trimethyl chitosan and PEGylated trimethyl chitosan nanocomplexes in insulin uptake. J Pharm Sci 2009;98:4818e30. [24] Mura S, Hillaireau H, Nicolas J, Kerdine-Romer S, Le Droumaguet B, Delomenie C, et al. Biodegradable nanoparticles meet the bronchial airway barrier: how surface properties affect their interaction with mucus and epithelial cells. Biomacromolecules 2011;12:4136e43. [25] Yin L, Ding J, He C, Cui L, Tang C, Yin C. Drug permeability and mucoadhesion properties of thiolated trimethyl chitosan nanoparticles in oral insulin delivery. Biomaterials 2009;30:5691e700. [26] Sarmento B, Martins S, Ferreira D, Souto EB. Oral insulin delivery by means of solid lipid nanoparticles. Int J Nanomed 2007;2:743e9. [27] Troutier AL, Ladaviere C. An overview of lipid membrane supported by colloidal particles. Adv Colloid Interface Sci 2007;133:1e21. [28] Jain S, Sharma RK, Vyas SP. Chitosan nanoparticles encapsulated vesicular systems for oral immunization: preparation, in-vitro and in-vivo characterization. J Pharm Pharmacol 2006;58:303e10. [29] Grenha A, Remunan-Lopez C, Carvalho EL, Seijo B. Microspheres containing lipid/chitosan nanoparticles complexes for pulmonary delivery of therapeutic proteins. Eur J Pharm Biopharm 2008;69:83e93. [30] Sood A, Panchagnula R. Peroral route: an opportunity for protein and peptide drug delivery. Chem Rev 2001;101:3275e303. [31] Chang M, Alsaigh T, Kistler EB, Schmid-Schonbein GW. Breakdown of mucin as barrier to digestive enzymes in the ischemic rat small intestine. Plos One 2012;7(6):e40087. [32] Aoki Y, Morishita M, Asai K, Akikusa B, Hosoda S, Takayama K. Regiondependent role of the mucous/glycocalyx layers in insulin permeation across rat small intestinal membrane. Pharm Res 2005;22:1854e62. [33] Wood KM, Stone GM, Peppas NA. The effect of complexation hydrogels on insulin transport in intestinal epithelial cell models. Acta Biomater 2010;6:48e56. [34] Goto T, Morishita M, Nishimura K, Nakanishi M, Kato A, Ehara J, et al. Novel mucosal insulin delivery systems based on fusogenic liposomes. Pharm Res 2006;23:384e91. [35] Mao S, Germershaus O, Fischer D, Linn T, Schnepf R, Kissel T. Uptake and transport of PEG-graft-trimethyl-chitosan copolymer-insulin nanocomplexes by epithelial cells. Pharm Res 2005;22:2058e68. [36] Maskarinec SA, Hannig J, Lee RC, Lee KYC. Direct observation of poloxamer 188 insertion into lipid monolayers. Biophys J 2002;82:1453e9. [37] Chang LC, Lin CY, Kuo MW, Gau CS. Interactions of pluronics with phospholipid monolayers at the air-water interface. J Colloid Interface Sci 2005;285: 640e52. [38] Maskarinec SA, Wu GH, Lee KYC. Membrane sealing by polymers. Ann NY Acad Sci 2005;1066:310e20. [39] Thompson C, Cheng WP, Gadad P, Skene K, Smith M, Smith G, et al. Uptake and transport of novel amphiphilic polyelectrolyte-insulin nanocomplexes by caco-2 cells-towards oral insulin. Pharm Res 2011;28:886e96.