lable at ScienceDirect

Biomaterials 32 (2011) 8712e8721

Contents lists avai

Biomaterials

journal homepage: www.elsevier .com/locate/biomateria ls

Effects of chitosan-nanoparticle-mediated tight junction opening on the oralabsorption of endotoxins

Kiran Sonaje a,1, Kun-Ju Lin b,c,1, Michael T. Tseng d, Shiaw-Pyng Wey b, Fang-Yi Su a, Er-Yuan Chuang a,Chia-Wei Hsu a, Chiung-Tong Chen e,**, Hsing-Wen Sung a,*

aDepartment of Chemical Engineering, National Tsing Hua University, Hsinchu 30013, Taiwan, ROCbDepartment of Medical Imaging and Radiological Sciences, Chang Gung University, Taoyuan, Taiwan, ROCcDepartment of Nuclear Medicine and Molecular Imaging Center, Chang Gung Memorial Hospital, Taoyuan, Taiwan, ROCdDepartment of Anatomical Sciences and Neurobiology, University of Louisville, Louisville, KY, USAe Institute of Biotechnology and Pharmaceutical Research, National Health Research Institutes, Zhunan, Miaoli, Taiwan, ROC

a r t i c l e i n f o

Article history:Received 19 July 2011Accepted 29 July 2011Available online 8 September 2011

Keywords:ChitosanTight junctionEndotoxinInsulinParacellular transport

* Corresponding author. Tel.: þ886 3 574 2504; fax** Corresponding author.

E-mail addresses: ctchen@nhri.org.tw (C.-T. Che(H.-W. Sung).

1 The first two authors (K. Sonaje and K. J. Lin) con

0142-9612/$ e see front matter � 2011 Elsevier Ltd.doi:10.1016/j.biomaterials.2011.07.086

a b s t r a c t

Recently, we reported a pH-responsive nanoparticle (NP) system shelled with chitosan (CS), which couldeffectively increase the oral absorption of insulin and produce a hypoglycemic effect, presumably due tothe CS-mediated tight junction (TJ) opening. It has been often questioned whether CS can also enhancethe absorption of endotoxins present in the small intestine. To address this concern, we studied the effectof CS NPs on the absorption of lipopolysaccharide (LPS), the most commonly found toxin in thegastrointestinal tract. To follow their biodistribution by the single-photon emission computed tomog-raphy/computed tomography, LPS and insulin were labeled with 99mTc-pertechnetate (99mTc-LPS) and123iodine (123I-insulin), respectively. The 99mTc-LPS was ingested 1 h prior to the administration of the123I-insulin-loaded NPs to mimic the physiological conditions. The confocal and TEM micrographs showthat the orally administered CS NPs were able to adhere and infiltrate through the mucus layer, approachthe epithelial cells and mediate to open their TJs. The radioactivity associated with LPS was mainlyrestricted to the gastrointestinal tract, whereas 123I-insulin started to appear in the urinary bladder at 3 hpost administration. This observation indicates that the insulin-loaded in CS NPs can traverse across theintestinal epithelium and enter the systemic circulation, whereas LPS was unable to do so, probablybecause of the charge repulsion between the anionic LPS in the form of micelles and the negativelycharged mucus layer. Our in vivo toxicity study further confirms that the enhancement of paracellularpermeation by CS NPs did not promote the absorption of LPS. These results suggest that CS NPs can beused as a safe carrier for oral delivery of protein drugs.

� 2011 Elsevier Ltd. All rights reserved.

1. Introduction

Despite the advances in the development of drug deliverytechnologies, successful oral administration of protein drugs hasremained to be an elusive goal. After oral administration, proteindrugs encounter several difficulties such as rapid pre-systemicdenaturation/degradation and poor absorption in the small intes-tine [1]. Therefore, a delivery system is needed to enhance thebioavailability of such drugs. An ideal delivery system for oral

: þ886 3 572 6832.

n), hwsung@che.nthu.edu.tw

tributed equally to this work.

All rights reserved.

administration of protein drugs should reversibly increase thepermeability of the mucosal epithelium to improve the absorptionof protein drugs and provide the intact drugs to the systemiccirculation [2].

In a recent study, we reported a pH-responsive nanoparticle(NP) system shelled with chitosan (CS) for oral delivery of insulinvia the paracellular pathway [3,4]. CS, a cationic polysaccharide, canadhere to the epithelial surface to impart transient opening of thetight junctions (TJs) between contiguous cells [5]. The resultsobtained in a diabetic rat model indicated that CS NPs couldeffectively increase the intestinal absorption of insulin and producea slower, but prolonged hypoglycemic effect [6]. However, it hasbeen often argued whether CS can also enhance the absorption ofunwanted toxins present in the small intestine [7].

The gastrointestinal (GI) tract is normally exposed to a numberof chemical and bacterial toxins. Some chemical toxins such as

K. Sonaje et al. / Biomaterials 32 (2011) 8712e8721 8713

thallium and lead can enter the systemic circulation with orwithout the disruption of cell membrane integrity. Fortunately,these chemical toxins are not normally present in the intestine,unless ingested accidentally [8].

On the other hand, the enteric bacteria produce two kinds oftoxins, exotoxins and endotoxins [9]. Exotoxins are the proteinssecreted by living bacteria [10]. Since the majority of the orallyingested bacteria are killed by the acidic environment in thestomach, the intestine is seldom exposed to exotoxins. Additionally,the proteolytic enzymes present in the GI tract can furtherneutralize such exotoxins [9]. In contrast, endotoxins are thenegatively charged components (lipopolysaccharide, LPS) from thecell wall of gram-negative bacteria; they are released in the GI tractby the disintegrating or dead bacteria [11]. Thus, LPS is the majorbacterial toxin present in the GI tract at all times [12]. If delivered tothe systemic circulation, LPS triggers a systemic inflammatoryresponse that can progress to endotoxic shock and sometimesdeath [13]. Therefore, it is essential to study the effect of CS NPs onthe absorption of LPS.

To follow their biodistribution, we used 99mTc-pertechnetate(99mTc) to label LPS (99mTc-LPS) and 123iodine (123I) to label insulin(123I-insulin). Employing a rat model, the 99mTc-LPS was orallyadministered 1 h before the ingestion of the 123I-insulin-loaded NPstomimic the natural conditions, inwhich LPS is present in the smallintestine. The biodistribution of the orally administered LPS wasstudied using the single-photon emission computed tomography(SPECT)/computed tomography (CT) and confocal laser scanningmicroscopy (CLSM). The activity of epithelial TJ opening by CS NPswas investigated by transmission electron microscopy (TEM).Additionally, the in vivo toxicity of the orally administered LPS wasexamined in mice.

2. Materials and methods

2.1. Preparation and characterization of CS NPs

CS (MW 80 kDa) with a degree of deacetylation of approximately 85% wasacquired from Koyo Chemical Co. Ltd. (Japan), while poly(g-glutamic acid) (g-PGA)(MW 60 kDa) was purchased from Vedan Co. Ltd. (Taichung, Taiwan). The insulin-loaded NPs were prepared by an ionic-gelation method, using the positively-charged CS and the negatively charged g-PGA in the presence of bovine insulin[4,6,14,15]. The prepared NPs were washed three times with deionized (DI) waterand collected by centrifugation at 8000 rpm for 50 min. The collected NPs wereredispersed in DI water and stored at 4 �C until used. Themean particle size and zetapotential value of the prepared NPs were measured using a Zetasizer (Nano ZS,Malvern Instruments Ltd., Worcestershire, UK); their insulin loading efficiency andcontent were calculated as reported previously [4,6].

2.2. Evaluation of the micelle-forming characteristics of LPS

LPS is an amphiphilic molecule, composed of a hydrophilic oligosaccharide chainwith varying length and a hydrophobic portion known as lipid A [11,16], which mayform micelles in an aqueous environment. To confirm this possibility, differentconcentrations of LPS (Escherichia coli, serotype 0111:B4, Sigma-Aldrich, St. Louis,MO, USA) were suspended in phosphate buffered saline (PBS, pH 7.4). The formationof micelles was validated by measuring the particle size and zeta potential of theresulting suspensions using the Zetasizer.

2.3. Animal studies

Animal studies were performed in compliance with the “Guide for the Care andUse of Laboratory Animals” prepared by the Institute of Laboratory AnimalResources, National Research Council, and published by the National Academy Press,revised in 1996.

2.3.1. Biodistribution studyThe biodistribution of LPS and the insulin-loaded in CS NPs was studied in rats

(male Wistar, 200�250 g) using the SPECT/CT. In the study, LPS was radiolabeledwith 99mTc (emitting 140 keV photons) using a stannous chloride (SnCl2) method[17]; the labeling efficiency of 99mTc to LPS was determined by the instant thin layerchromatography (ITLC) [18]. The insulin was radiolabeled with 123I (emitting159 keV photons) using an iodogen-tube (Pierce Iodination Tubes, Thermo Fisher

Scientific, Rockford, IL, USA) method, as per the manufacturer’s instructions. The123I-insulin was separated from the free-form 123I using a centrifugal dialysis device(MWCO: 3 kDa, Amicon Ultra 4, Millipore, Billerica, MA, USA); its labeling efficiencywas determined by a reversed-phase HPLC system equipped with a gamma counter[19]. The obtained 123I-insulinwas then used to prepare test NPs as described above.

In the biodistribution study, rats were fed with 99mTc-LPS alone or together withthe 123I-insulin-loaded CS NPs (n ¼ 3 in each studied group). For the group receivingboth LPS and test NPs, the 99mTc-LPS was administered 1 h before the ingestion ofthe 123I-insulin-loaded NPs. The detailed protocol used in the image acquisition waspreviously described by our group [18,20]. Animal images were acquired usinga dual modality system (NanoSPECT/CT, Bioscan Inc., Washington DC, USA), which iscapable of detecting two kinds of isotopes simultaneously at a relatively high spatialresolution (approximately 0.6 mm).

Animals were kept under the controlled temperature (37 �C) and anesthesia(1.5% isoflurane in 100% oxygen) during imaging. Dual isotope dynamic SPECTimages were acquired at 30-min intervals up to 24 h after the administration of testsamples. Additional CT images were collected for anatomical references and used toinvestigate the details of radiotracer distribution in rats. The co-registered dynamicscintigraphy and CT images were displayed and analyzed using the PMOD v2.9image analysis software (PMOD Technologies Ltd., Zurich, Switzerland).

The quantitative analysis of SPECT images was performed to evaluate thedistribution of 99mTc-LPS and 123I-insulin within the peripheral tissue/plasma (PP)compartment. The PP compartment was defined as the whole body (WB) excludingthe gastrointestinal (GI) tract and urinary bladder. To calculate the percentage ofinitial dose (% ID) within each region, the corresponding contours were manuallydrawn on the co-registered dynamic SPECT and reference CT images. The bio-distribution data were expressed as % ID using the following formula:

% ID ¼ decay corrected total radioactivity in the target regioningested counts

� 100% (1)

2.3.2. Ultra-structural examination of TJ opening by CS NPsThe opening of epithelial TJs by CS NPswas examined using TEM. In the study, CS

was labeled with quantum dot (QD, CdSe) according to a method reported in theliterature [21]. Briefly, carboxyl QD (40 mL, 0.6 nM, Qdot� ITK�, Invitrogen, USA) wasactivated in the presence of 1-ethyl-3-(3-dimethylaminopropyl) carbodiimide (EDC,60 mL, 50 mM) and N-hydroxysuccinimide (NHS, 30 mL, 25 mM) for 15 min undergentle stirring. The resulting NHS-activated QDwas covalently linked to the primaryamines on CS (QD-CS) at pH 6.0. The reaction was carried out under gentle mixingfor 4 h. The final QD-CS conjugate was purified by the centrifugal spin filtration andthen resuspended in DI water.

QD-CS NPs were then prepared using the method described above. The QD-CSNPs (1 mg/mL, 1 mL) were orally administered to the overnight-fasted ICR mice(33e40 g, n ¼ 3). Animals were sacrificed 3 h later and the intestinal segments weredissected and washed three times with isotonic saline. The dissected intestinalsegments were then fixed in 4% paraformaldehyde (PFA), cut into smaller pieces andsilver-intensified using a commercial silver enhancer kit (Sigma-Aldrich) [22]. Aftersilver-enhancing QD-CS, tissue samples were post-fixed in 1% osmium tetroxide anddehydrated in a graded series of ethanol. Subsequently, the dehydrated sampleswere infiltrated with and embedded in Spurr resinwith overnight polymerization at70 �C to prepare the tissue-embedded TEM blocks. Sections (1 mm in thickness) weremade and stained with toluidine blue and then observed under a light microscope.

To demonstrate the enhancement of paracellular transport by CS NPs, tissuessamples were stained with lanthanum nitrate [23]. For this, the PFA-fixed tissuesamples were washed with s-Collidine buffer and treated with 2% lanthanum nitratefor 2 h at room temperature [23,24]. After washing with s-Collidine and PBS, thetissue samples were processed for TEM as detailed above. Ultrathin sections werethen cut with a diamond knife and loaded onto TEM grids. The sections wereexamined by a Philips CM10 electron microscope (Philips Electron Optics B.V.) ataccelerating voltage of 60 kV.

2.3.3. Intestinal absorption of FITC-LPS and Cy3-insulin-loaded NPsThe FITC (fluorescein isothiocyanate) labeled LPS (FITC-LPS) and Cy3 (Cyanine-3)

labeled insulin (Cy3-insulin) were used to visualize their intestinal absorptioncharacteristics using CLSM (TCS SL, Leica, Germany). The FITC-LPS was obtainedfrom Sigma-Aldrich, whereas the Cy3-insulin was synthesized as per a methoddescribed in the literature [4,14]. Briefly, Cy3 NHS ester (GE Healthcare, Pittsburgh,PA, USA) dissolved in DMSO (dimethyl sulfoxide, 1 mg/mL, 1 mL) was slowly addedinto an aqueous solution of insulin (1%w/v in 0.01 NHCl, 4mL) and stirred overnightat 4 �C. To remove the unconjugated Cy3, the synthesized Cy3-insulin was dialyzedin the dark against 5 L of 0.01 N HCl and replaced on a daily basis until no fluo-rescence was detected in the dialysis medium. The resultant Cy3-insulin waslyophilized in a freeze dryer. Fluorescent NPs were then prepared for the subsequentin vivo CLSM study according to the procedure described above.

FITC-LPS (2 mg/mL, 0.5 mL) alone or in combination with a mucolytic agent (N-acetylcysteine) were administered to the overnight-fasted rats (maleWistar, n¼ 3 ineach studied group). To study the effects of CS NPs on the absorption of LPS, the Cy3-insulin-loaded NPs (2 mg/mL, 0.5 mL) were orally administered 1 h after the

Table 1Particle size, distribution and zeta potential of the micelles formed in aqueous LPSsuspensions at different concentrations (n ¼ 5).

LPS Concentration(mg/mL)

Particle Size(nm)

Polydispersity Zeta Potential(mV)

0.2 135.3 � 3.2 0.2 � 0.0 �9.8 � 2.52.0 220.5 � 7.6 0.5 � 0.1 �8.5 � 1.710.0 545.8 � 14.1 0.7 � 0.1 �4.8 � 1.2

K. Sonaje et al. / Biomaterials 32 (2011) 8712e87218714

ingestion of the FITC-LPS (2 mg/mL, 0.5 mL). Rats were sacrificed 3 h later andintestinal segments were then dissected and washed three times with isotonicsaline. The isolated intestinal segments were fixed using the methanol-Carnoy’sfixative and processed for paraffin-embedding. The embedded sections were dew-axed, hydrated and stained with Alexa-633-labeled wheat-germ-agglutinin andSYTOX blue (Invitrogen, Carlsbad, CA, USA) to visualize the mucus and nuclei,respectively. Finally, the stained sections were examined under CLSM.

2.3.4. In vivo toxicity studyLPS is a well-known inflammatory agent; therefore, a study was performed to

evaluate whether CS NPs can promote the toxicity of the orally administered LPS.Animals (male ICR mice) were randomly divided into four groups (n ¼ 6 for eachstudied group). The experimental groups received once-daily oral doses of LPS(5 mg/kg) with or without CS NPs (10 mg/kg) for 7 consecutive days; the groupwithout any treatment was used as a control. Additionally, a group receiving intra-peritoneal (IP) LPS (5mg/kg) served as a reference for the extent of toxicity producedby the systemic LPS. All animals were fed with normal chows and water ad libitum.Animals were observed carefully for the onset of any signs of toxicity and monitoredfor changes in body weight. At the end of the treatment period, animals wereanaesthetized (tribromoethanol, IP, 240 mg/kg) and blood samples were collectedvia cardiac puncture for the determination of alanine aminotransferase (ALT) andaspartate aminotransferase (AST) using a FUJI DRI-CHEM 3500s serum-chemistryanalyzer. After sacrificed, internal organs of each animal were harvested andobserved grossly. For histological examinations, specimens of liver were fixed in 10%phosphate buffered formalin, embedded in paraffin, sectioned and stained withhematoxylin and eosin (H&E).

2.4. Statistical analysis

Comparison between groups was analyzed by the one-tailed Student’s t-test(SPSS, Chicago, Ill). All data are presented as a mean value with its standard devi-ation indicated (mean � SD). Differences were considered to be statistically signif-icant when the p values were less than 0.05.

3. Results and discussion

The efficacy of oral delivery of protein drugs is often limitedbecause of their inherent instability in the GI tract. Additionally, thehigh molecular weight of this class of drugs coupled with theirhydrophilic nature significantly restricts their transcellularpermeation [1]. Thus, enhancement of the paracellular permeationis an alternative for oral absorption of protein drugs [25]. CS isa well-known mucoadhesive agent with the capability of tran-siently and reversibly opening epithelial TJs [26]. Formulating CSinto NPs has the advantages over the traditional tablet or powderformulations, as NPs can readily infiltrate into the mucus layer anddeliver the protein drugs to the actual site of absorption (i.e., the TJsbetween epithelial cells). CS is generally regarded as a safe materialfor drug delivery. However, its effects on the absorption ofunwanted toxins remain to be understood.

3.1. Characteristics of CS NPs

The prepared CS NPs had a mean particle size of 253.2 � 4.8 nmwith a zeta potential of 28.2 � 1.3 mV; their insulin loading effi-ciency and content were 72.4 � 3.9% and 17.9 � 2.1%, respectively(n ¼ 6 batches). The as-prepared NPs are pH-responsive: they werestable in the pH range of 2.0e7.0; beyond this range, the particlesbecame unstable and disintegrated. Similar characteristics werefound for the NPs prepared with QD-CS or FITC-CS.

3.2. Micelle-forming characteristics of LPS

Due to its amphiphilic nature, LPS may form micelles in anaqueous environment when above its critical micelle concentration(CMC) [12]. The reported CMC values for LPS vary from 10 nM to1.6 mM, depending on the source of LPS [16]. It has been suggestedthat the aggregated form (i.e., micelles) of LPS predominates in therange of concentration usually found in the intestinal lumen[8,12,27]. To evaluate the micelle-forming characteristics of LPS, we

measured the size and zeta potential of the aqueous suspensions ofLPS at different concentrations. As shown in Table 1, with anincrease in concentration, the size of LPS micelles increasedsignificantly (P< 0.05). The LPSmicelles were negatively charged atall concentrations, which could be due to the two phosphate groupspresent on the LPS structure. In the concentration used in thesubsequent animal study (2 mg/mL), LPS formed micelles with anaverage size of 220.5� 7.6 nm and a zeta potential of�8.5�1.7 mV.

3.3. Biodistribution and absorption of the orally administered LPS

The biodistribution and intestinal absorption of the orallyadministered LPS was investigated by the SPECT/CT. In the study,99mTc was used to label LPS and its labeling efficiency was deter-mined by the ITLC. It was found that more than 99% of 99mTc wassuccessfully conjugated onto LPS (Fig. 1a). The SPECT/CT images ofthe orally administered 99mTc-LPS are shown in Fig. 1b. As shown,after oral administration of LPS alone, the radioactivity (99mTc-LPS)propagated from the stomach, small intestine and then to the largeintestine with time. Overall, the 99mTc-labeled LPS appeared to berestricted within the GI tract, with no detectable radioactivitypresent in the PP compartment. These results were in agreementwith the findings reported in the literature that the orally admin-istered LPS was not able to be absorbed into the systemic circula-tion in rabbits [27,28]. Mucus is a viscoelastic gel layer that protectstissues that would otherwise be exposed to the external environ-ment [29]. The mucus layer has been shown to act as a physicalbarrier to the enteric bacteria and hinders their access to theunderlying epithelium [30]. However, the protective function of themucus layer against LPS is still unknown.

To understand how the intestinal epithelium prevents theabsorption of LPS, we investigated the absorption of FITC-LPS in ratsusing CLSM. In the study, the FITC-LPS was orally administered inthe absence/presence of a mucolytic agent (N-acetylcystein). In theabsence of the mucolytic agent, the epithelial mucus layer wasintact; the administered FITC-LPS was found to be restrictedoutside the mucus layer (Fig. 2, upper panels). This is probably dueto the charge repulsion between the negatively charged mucus [31]and the anionic LPS in the form of micelles. Additionally, theintestinal mucus is rich in LPS-binding proteins [32], which mightcontribute to the inability of LPS to penetrate through the mucuslayer. In contrast, in the presence of the mucolytic agent, thethickness of the mucus layer decreased significantly, and theadministered FITC-LPS was able to infiltrate through themucus andaccess the epithelium surface. This led to the absorption of LPS intothe systemic circulation, as indicated by its presence on the baso-lateral side of intestinal villi (Fig. 2, lower panels).

The epithelium lining on the GI tract provides a regulated,selectively permeable barrier between the external environment(the intestinal lumen) and the systemic circulation. It transportsnutrients, ions and fluid transcellularly, but prevents the entry oftoxins, antigens and microorganisms [9]. The paracellular route isrestricted by the presence of TJs at the apical poles of enterocytesthat limit the passage of macromolecules [33]; its permeabilitygenerally depends on the regulation of intercellular TJs by using anintestinal permeation enhancer such as CS [5].

Fig. 1. (a) Scan of the radioactivity of 99mTc-LPS, developed on an instant thin layer chromatography (ITLC) plate; (b) biodistribution of 99mTc-LPS observed in a rat model after oralingestion.

K. Sonaje et al. / Biomaterials 32 (2011) 8712e8721 8715

3.4. Ultra-structural examination of TJ opening by CS NPs

The ability of CS in enhancing the permeability of model drugcompounds across Caco-2 cell monolayers has been investigated in

Fig. 2. Confocal images showing the intestinal absorption of FITC-LPS (green) after its oralpresence of the mucolytic agent, the mucus layer (red) became thinner, and FITC-LPS was obarrows), an indication of the intestinal absorption of LPS. (For interpretation of the reference

numerous studies [5,26]; however, it’s in vivo activity has neverbeen investigated. Fig. 3a shows photomicrographs of a silver-enhanced intestinal segment of an ICR mouse after being treatedwith QD-CS NPs. The silver-enhancement procedure enlarges QD

administration in the absence/presence of a mucolytic agent (N-acetylcystein). In theserved underneath the epithelium (indicated in the superimposed image by the whites to colour in this figure legend, the reader is referred to the web version of this article.)

K. Sonaje et al. / Biomaterials 32 (2011) 8712e87218716

by selective deposition of metallic silver, making them easilyidentifiable by light microscopy [34]. As shown, the orally admin-istered NPs were able to adhere and infiltrate into the mucus layerand approach the surface of epithelial cells. However, we were notable to tell whether the epithelial TJs were opened, due to the lackof electron density in their intercellular spaces.

For this reason, the same tissue samples after being treated withQD-CS NPs were incubated with an aqueous lanthanum.Lanthanum is an electron-dense element with a hydrated radius of0.4 nm and has been widely used to stain the cell surfaces for TEMexamination [24]. It has been reported that the width of TJs, whenfully opened, is less than 20 nm [25]. For the control sample (thatwithout being treated with QD-CS NPs), the TJs were intact and thelanthanum staining was restricted on the mucosal surface (Fig. 3b).In contrast, lanthanum was able to penetrate into the paracellularspaces for the experimental sample, suggesting that the TJs wereindeed opened by QD-CS NPs. These results support our hypothesisthat CS NPs could infiltrate through the epithelial mucus layer andthus may deliver the loaded drugs near the opened TJs to promotetheir intestinal absorption. However, it is also essential to consider

Fig. 3. (a) Photomicrograph of a silver-enhanced intestinal section showing the mucoadhesa rectangle is shown at a higher magnification in the inset; (b) TEM micrographs of the colanthanum through the opened paracellular space (indicated by the blue arrows) in mice treathe reader is referred to the web version of this article.)

the effects of CS-mediated TJ opening on the transport of endo-toxins (i.e., LPS) present in the small intestine.

3.5. Effects of the insulin-loaded CS NPs on the biodistribution andabsorption of LPS

In the study, insulin was radiolabeled by 123I with an efficiencyof about 80%, determined by the reversed-phase HPLC; the free-form 123I was removed using a centrifugal dialysis device.Following oral administration in rats, effects of CS NPs on the bio-distribution and absorption of 99mTc-LPS and 123I-insulin (theloaded drug) were studied using the dual isotope dynamic SPECT/CT. To mimic the physiological conditions, the 99mTc-LPS wasingested 1 h prior to the administration of the 123I-insulin-loadedNPs. As shown in Fig. 4a, the radioactivity associated with LPS wasmainly limited to the GI tract throughout the entire course of thestudy, whereas 123I-insulin started to appear in the urinary bladderat 3 h post administration. These results suggest that the insulin-loaded in CS NPs can traverse across the intestinal epithelium

ion and infiltration of QD-CS NPs (black dots) after oral administration, area defined byntrol intestinal segment incubated with a lanthanum solution; and (c) permeation ofted with QD-CS NPs. (For interpretation of the references to colour in this figure legend,

Fig. 4. (a) Biodistribution of 99mTc-LPS and 123I-insulin in rats orally treated with 99mTc-LPS followed by 123I-insulin-loaded NPs; (b) reconstructed 3D images showing the wholebody (WB) and gastrointestinal tract/urinary bladder (GIT/UB) regions together with their sagittal and coronal views: the plasma/peripheral tissue (PP) compartment is defined asthe WB region (blue contour) excluding the GIT/UB (yellow contour); and (c) % initial dose (% ID) of 99mTc-LPS and 123I-insulin observed in the PP compartment. (For interpretationof the references to colour in this figure legend, the reader is referred to the web version of this article.)

K. Sonaje et al. / Biomaterials 32 (2011) 8712e8721 8717

Fig. 5. Schematic drawing of intestinal villi and confocal images showing the intestinal villi retrieved from rats fed with FITC-LPS (green) followed by the administration of Cy3-insulin-loaded NPs. The white arrows indicate the absorbed insulin (purple) underneath the epithelium, while FITC-LPS (pointed by the blue arrows) was mainly restricted outsidethe mucus layer (red). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Fig. 6. Changes in (a) body weight and (b) serum levels of aspartate transaminase and alanine transaminase in mice treated with LPS alone via the intra-peritoneal route (IP LPS) orthe oral route (oral LPS) or in combination with CS NPs (oral LPS þ CS NPs). The group without receiving any treatment was used as a control. *statistical significance at a level ofP < 0.05.

K. Sonaje et al. / Biomaterials 32 (2011) 8712e87218718

Fig. 7. Photomicrographs of the liver sections obtained in the toxicity study: (a) the control group; (b) the group treated with LPS intraperitoneally (IP LPS); (c) the group orallytreated with LPS alone (oral LPS); and (d) the group orally treated with LPS followed by CS NPs (oral LPS þ CS NPs). The black arrows indicate foci of necrotic cells.

K. Sonaje et al. / Biomaterials 32 (2011) 8712e8721 8719

and enter the systemic circulation; in contrast, LPSwas unable to doso, hence retaining primarily within the intestinal lumen.

To quantify the amount of 99mTc-LPS and 123I-insulin within thePP compartment, contours of the WB and the GI tract/urinarybladder were manually drawn on an averaged SPECT/CT image. The

Fig. 8. Schematic illustrations showing the selective barrier function of the mucus layer, prpositively-charged CS NPs to infiltrate through.

dynamic 99mTc-LPS SPECT/CT images superimposedwith the regioncontours in the sagittal and coronal views are presented in Fig. 4b.The radioactivity counts of 99mTc-LPS and 123I-insulin within the PPcompartment were calculated and normalized to their initiallyingested doses (% ID). As shown in Fig. 4c, the radioactivity of

eventing the oral absorption of anionic LPS in the form of micelles, while allowing the

K. Sonaje et al. / Biomaterials 32 (2011) 8712e87218720

123I-insulin in the PP compartment increased with time andreached to its maximumvalue (w8% ID) by 6 h post administration,while no measurable 99mTc-LPS was observed throughout thestudy. These results suggest that the absorption enhancement by CSNPs was specific for the loaded insulin only. To determine a specificreason for this observation, we used the FITC-LPS and Cy3-insulin-loaded NPs to repeat the study, examining by CLSM.

Fig. 5 shows the CLSM images acquired from a rat ingested withthe FITC-LPS followed by the administration of the Cy3-insulin-loaded NPs. As shown, the Cy3-insulinwas able to infiltrate throughthe mucus layer and was observed underneath the epithelium. Incontrast, the FITC-LPS was still restricted outside the mucus layer.

3.6. In vivo toxicity study

LPS is a potent inflammatory agent, which plays an importantrole in the pathogenesis of endotoxic shock [35]. One of the mostfrequently reported symptoms of endotoxic shock is the hepatocytenecrosis [13,36]. To study whether CS NPs can induce in vivotoxicity in the presence of LPS, the changes in body weight and twoof the liver function indicators (ALT and AST) were evaluated. Asexpected, the daily injections of LPS via the IP route led to decreasein body weight and a significant increase in the level of AST(P < 0.01, Fig. 6a and b). In contrast, the group receiving oral LPSfollowed by CS NPs did not show any significant changes in bodyweight and levels of liver function indicators, as compared to thecontrol (p > 0.05). The histological examination of liver sectionswas performed. In the group treated with LPS via the IP route, theliver section showed broad hemorrhagic necrosis, hepatocyteswelling and degeneration (Fig. 7b). In contrast, the liver sectionsretrieved from the groups receiving LPS via the oral route (Fig. 7cand d) were similar to the control (Fig. 7a). These results furtherconfirm that the paracellular permeation enhancement by CS NPsdid not promote the intestinal absorption of LPS.

The aforementioned findings suggest that CS NPs can adhereand infiltrate into the mucus; the infiltrated NPs become unstableand disintegrate near the epithelial cell surface due to their pH-sensitivity (Section 3.1) and thus release the loaded insulin. Thereleased insulin could then enter the systemic circulation due to theCS-mediated TJ opening (Section 3.4). In contrast, the anionic LPS inthe form of micelles are repelled by the negatively charged mucuslayer lining on the intestinal epithelium, consequently preventingLPS from entering the systemic circulation (Fig. 8).

In our previous report, an in vivo toxicity study was performedto determine whether oral administration of CS NPs was safe [4].The animals were treated with a daily dose of CS NPs for 14 days. Nosignificant differences in clinical signs and body weight betweenthe experimental group and the untreated control group werefound. The measured hematological and biochemical parametersfor both studied groups were within the normal ranges. Moreover,no pathological changes were observed in the histological sectionsof the liver and kidney. These results indirectly pointed out that CSNPs did not promote the absorption of any toxins from the GI tract.

4. Conclusions

The results obtained in the study indicate that CS NPs couldadhere and infiltrate through the mucus layer, mediate to open theepithelial TJs, and enhance the paracellular delivery of the loadedinsulin. However, the enhancement of paracellular permeation byCS NPs did not promote the intestinal absorption of LPS; thisobservation was further confirmed in our in vivo toxicity study. Onthe basis of these results, it can be concluded that the CS NPs can beused as a safe carrier for oral delivery of protein drugs.

Acknowledgement

This work was supported by a grant from the National ScienceCouncil (NSC 99-2120-M-007-006), Taiwan, Republic of China.

References

[1] Khafagy el S, Morishita M, Onuki Y, Takayama K. Current challenges in non-invasive insulin delivery systems: a comparative review. Adv Drug DelivRev 2007;59:1521e46.

[2] Carino GP, Mathiowitz E. Oral insulin delivery. Adv Drug Deliv Rev 1999;35:249e57.

[3] Sonaje K, Lin KJ, Wang JJ, Mi FL, Chen CT, Juang JH, et al. Self-assembled ph-sensitive nanoparticles: a platform for oral delivery of protein drugs. AdvFunct Mater 2010;20:3695e700.

[4] Sonaje K, Lin YH, Juang JH, Wey SP, Chen CT, Sung HW. In vivo evaluation ofsafety and efficacy of self-assembled nanoparticles for oral insulin delivery.Biomaterials 2009;30:2329e39.

[5] Smith J, Wood E, Dornish M. Effect of chitosan on epithelial cell tight junc-tions. Pharm Res 2004;21:43e9.

[6] Sonaje K, Chen YJ, Chen HL, Wey SP, Juang JH, Nguyen HN, et al. Enteric-coatedcapsules filled with freeze-dried chitosan/poly(g-glutamic acid) nanoparticlesfor oral insulin delivery. Biomaterials 2010;31:3384e94.

[7] Junginger HE, Verhoef JC. Macromolecules as safe penetration enhancers forhydrophilic drugs � a fiction? Pharm Sci Technolo Today 2000;3:346e58.

[8] Lu FC, Kacew S. Basic toxicology: fundamentals, target organs, and riskassessment. Chapter 2: Absorption, distribution, and excretion of toxicants. 5th

ed. PA, USA: Taylor & Francis; 2009. p. 15�28.[9] Walker RL, Owen RL. Intestinal barriers to bacteria and their toxins. Annu Rev

Med 1990;41:393e400.[10] Macpherson AJ, Geuking MB, McCoy KD. Immune responses that adapt the

intestinal mucosa to commensal intestinal bacteria. Immunology 2005;115:153e62.

[11] Gutsmann T, Schromm AB, Brandenburg K. The physicochemistry of endo-toxins in relation to bioactivity. Int J Med Microbiol 2007;297:341e52.

[12] Piazza M, Colombo M, Zanoni I, Granucci F, Tortora P, Weiss J, et al. Uniformlipopolysaccharide (LPS)-loaded magnetic nanoparticles for the investigationof LPS-TLR4 signaling. Angew Chem Int Ed Engl 2011;50:622e6.

[13] Li G, Liu Y, Tzeng NS, Cui G, Block ML, Wilson B, et al. Protective effect ofdextromethorphan against endotoxic shock in mice. Biochem Pharmacol2005;69:233e40.

[14] Lin YH, Sonaje K, Lin KM, Juang JH, Mi FL, Yang HW, et al. Multi-ion-cross-linked nanoparticles with pH-responsive characteristics for oral delivery ofprotein drugs. J Control Release 2008;132:141e9.

[15] Sonaje K, Lin KJ, Wey SP, Lin CK, Yeh TH, Nguyen HN, et al. Biodistribution,pharmacodynamics and pharmacokinetics of insulin analogues in a rat model:oral delivery using pH-Responsive nanoparticles vs. subcutaneous injection.Biomaterials 2010;31:6849e58.

[16] Yu L, Tan M, Ho B, Ding JL, Wohland T. Determination of critical micelleconcentrations and aggregation numbers by fluorescence correlation spec-troscopy: aggregation of a lipopolysaccharide. Anal Chim Acta 2006;556:216e25.

[17] Rosenbaum JT, Hendricks PA, Shively JE, McDougall IR. Distribution of radio-labeled endotoxin with particular reference to the eye: concise communica-tion. J Nucl Med 1983;24:29e33.

[18] Chen MC, Wong HS, Lin KJ, Chen HL, Wey SP, Sonaje K, et al. The character-istics, biodistribution and bioavailability of a chitosan-based nanoparticulatesystem for the oral delivery of heparin. Biomaterials 2009;30:6629e37.

[19] Sodoyez JC, Sodoyez-Goffaux F, Guillaume M, Merchie G. [123I]Insulinmetabolism in normal rats and humans: external detection by a scintillationcamera. Science 1983;219:865e7.

[20] Lin KJ, Liao CH, Hsiao IT, Yen TC, Chen TC, Jan YY, et al. Improved hepatocytefunction of future liver remnant of cirrhotic rats after portal vein ligation:a bonus other than volume shifting. Surgery 2009;145:202e11.

[21] Bagalkot V, Zhang L, Levy-Nissenbaum E, Jon S, Kantoff PW, Langer R, et al.Quantum dot-aptamer conjugates for synchronous cancer imaging, therapy,and sensing of drug delivery based on bi-fluorescence resonance energytransfer. Nano Lett 2007;7:3065e70.

[22] Liang RQ, Li W, Li Y, Tan CY, Li JX, Jin YX, et al. An oligonucleotide microarrayfor microRNA expression analysis based on labeling RNA with quantum dotand nanogold probe. Nucleic Acids Res 2005;33. e17.

[23] Friend DS, Gilula NB. Variations in tight and gap junctions in mammaliantissues. J Cell Biol 1972;53:758e76.

[24] Flynn AN, Itani OA, Moninger TO, Welsh MJ. Acute regulation of tight junctionion selectivity in human airway epithelia. Proc Natl Acad Sci USA 2009;106:3591e6.

[25] Salama NN, Eddington ND, Fasano A. Tight junction modulation and its rela-tionship to drug delivery. Adv Drug Deliv Rev 2006;58:15e28.

[26] Smith JM, Dornish M, Wood EJ. Involvement of protein kinase C in chitosanglutamate-mediated tight junction disruption. Biomaterials 2005;26:3269e76.

[27] Beatty WL, Meresse S, Gounon P, Davoust J, Mounier J, Sansonetti PJ, et al.Trafficking of Shigella lipopolysaccharide in polarized intestinal epithelialcells. J Cell Biol 1999;145:689e98.

K. Sonaje et al. / Biomaterials 32 (2011) 8712e8721 8721

[28] Ravin HA, Rowley D, Jenkins C, Fine J. On the absorption of bacterial endotoxinfrom the gastro-intestinal tract of the normal and shocked animal. J Exp Med1960;112:783e92.

[29] Cone RA. Barrier properties of mucus. Adv Drug Deliv Rev 2009;61:75e85.

[30] Johansson MEV, Phillipson M, Petersson J, Velcich A, Holm L, Hansson GC. Theinner of the two Muc2 mucin-dependent mucus layers in colon is devoid ofbacteria. Proc Natl Acad Sci USA 2008;105:15064e9.

[31] Atuma C, Strugala V, Allen A, Holm L. The adherent gastrointestinal mucus gellayer: thickness and physical state in vivo. Am J Physiol Gastrointest LiverPhysiol 2001;280:G922e9.

[32] Vreugdenhil AC, Snoek AM, Greve JW, Buurman WA. Lipopolysaccharide-binding protein is vectorially secreted and transported by cultured intestinal

epithelial cells and is present in the intestinal mucus of mice. J Immunol 2000;165:4561e6.

[33] Ward PD, Tippin TK, Thakker DR. Enhancing paracellular permeability bymodulating epithelial tight junctions. PharmSci TechnoloToday 2000;3:346e58.

[34] Mortensen LJ, Oberdorster G, Pentland AP, Delouise LA. In vivo skin penetra-tion of quantum dot nanoparticles in the murine model: the effect of UVR.Nano Lett 2008;8:2779e87.

[35] Illyes G, Kovacs K, Kocsis B, Baintner K. Failure of oral E. coli O83 lipopoly-saccharide to influence intestinal morphology and cell proliferation in rats:short communication. Acta Vet Hung 2008;56:1e3.

[36] Bertok L. Effect of bile acids on endotoxin in vitro and in vivo (physico-chemical defense). Bile deficiency and endotoxin translocation. Ann N Y AcadSci 1998;851:408e10.