Research ArticleNanomedicine

Transepithelial Transport of Fc-Targeted Nanoparticles by the Neonatal Fc Receptor for Oral Delivery

See allHide authors and affiliations

Science Translational Medicine  27 Nov 2013:
Vol. 5, Issue 213, pp. 213ra167
DOI: 10.1126/scitranslmed.3007049

Abstract

Nanoparticles are poised to have a tremendous impact on the treatment of many diseases, but their broad application is limited because currently they can only be administered by parenteral methods. Oral administration of nanoparticles is preferred but remains a challenge because transport across the intestinal epithelium is limited. We show that nanoparticles targeted to the neonatal Fc receptor (FcRn), which mediates the transport of immunoglobulin G antibodies across epithelial barriers, are efficiently transported across the intestinal epithelium using both in vitro and in vivo models. In mice, orally administered FcRn-targeted nanoparticles crossed the intestinal epithelium and reached systemic circulation with a mean absorption efficiency of 13.7%*hour compared with only 1.2%*hour for nontargeted nanoparticles. In addition, targeted nanoparticles containing insulin as a model nanoparticle-based therapy for diabetes were orally administered at a clinically relevant insulin dose of 1.1 U/kg and elicited a prolonged hypoglycemic response in wild-type mice. This effect was abolished in FcRn knockout mice, indicating that the enhanced nanoparticle transport was specifically due to FcRn. FcRn-targeted nanoparticles may have a major impact on the treatment of many diseases by enabling drugs currently limited by low bioavailability to be efficiently delivered though oral administration.

INTRODUCTION

Nanoparticles (NPs) have the potential to make an impact on the treatment of many diseases, including cancer, cardiovascular disease, and diabetes. Many NP-based therapeutics are now entering clinical trials or have been approved for use (1, 2), including targeted polymeric NPs (3) (clinicaltrials.gov identifier: NCT01478893) based on technologies that we have described previously (4). However, the impact of NPs in the clinic may be limited to a narrow set of indications because NP administration is currently restricted to parenteral methods. Many diseases that could benefit from NP-based therapeutics require frequent administration. Alternate routes of administration, particularly oral, are preferred because of the convenience and compliance by patients (5). Intestinal absorption of NPs is highly inefficient because the physicochemical parameters of NPs prevent their transport across cellular barriers such as the intestinal epithelium (6). To improve the absorption efficiency of NPs and to make the oral administration of NPs practical in the clinic, new strategies are necessary to overcome the intestinal epithelial barrier.

The neonatal Fc receptor (FcRn) mediates immunoglobulin G (IgG) transport across polarized epithelial barriers (7, 8). It was discovered as the receptor in the neonatal intestine that transports IgG in breast milk from mother to offspring (9). However, FcRn is expressed into adulthood at levels similar to fetal expression in the apical region of epithelial cells in the small intestine and diffusely throughout the colon (10). FcRn is also expressed in the vascular endothelium, blood-brain barrier, kidneys, liver, lungs, and throughout the hematopoietic system (11, 12). FcRn interacts with the Fc portion of IgG in a pH-dependent manner, binding with high affinity in acidic (pH <6.5) but not physiological environments (pH ~7.4) (13). The intracellular trafficking of the IgG-FcRn complex has been conclusively demonstrated in the rat intestine using IgG Fc labeled with 1.4-nm gold as a contrast agent for electron tomography (14). The studies revealed that Fc is transported through a complex pathway involving a network of entangled tubular and irregular vesicles to reach the basolateral surface of the cell.

We hypothesized that targeting NPs to FcRn using IgG Fc fragments would allow orally administered NPs to be transported across the intestinal epithelium in rodents (Fig. 1). In acidic sections of the intestine, such as the duodenum and portions of the jejunum (15), Fc fragments conjugated to NPs [Fc-targeted NPs (NP-Fc)] will bind to FcRn at the apical surface of absorptive epithelial cells, leading to receptor-mediated endocytosis (16). NP-Fc could also be taken up by fluid-phase pinocytosis. During intracellular trafficking, NP-Fc and FcRn in the same acidic endosome compartments will bind with high affinity. FcRn can then guide bound NP-Fc through a transcytosis pathway, avoiding lysosomal degradation (17). On the basolateral side, exocytosis results in exposure to a neutral pH environment in the lamina propria, causing the release of NP-Fc (18). NP-Fc can then diffuse through the lamina propria and enter systemic circulation.

Fig. 1. Schematic of Fc-targeted NP transport across the intestinal epithelium by the FcRn through a transcytosis pathway.

(A) IgG Fc on the NP surface binds to the FcRn on the apical side of absorptive epithelial cells under acidic conditions in the intestine. (B) NP-Fc are then trafficked across the epithelial cell through the FcRn transcytosis pathway in acidic endosomes. (C) Upon exocytosis on the basolateral side of the cell, the physiological pH causes IgG Fc to dissociate from the FcRn, and NP-Fc are free to diffuse through the intestinal lamina propria to the capillaries or lacteal and enter systemic circulation.

Fc fusion proteins have been used to overcome biological barriers: Fc fused with erythropoietin was measured in nonhuman primates (19) and humans (20) after pulmonary administration, indicating that using the FcRn is a valid transport pathway in humans, and follicle-stimulating hormone fused with Fc reached circulation after oral delivery in newborn rats (21). However, NPs offer several potential advantages over fusion proteins including (i) transport of many drug molecules with each transcytosis event; (ii) protection of drug molecules, particularly protein therapeutics, through encapsulation; and (iii) mitigation of the need for drug modification. Recently, 20- to 50-nm fluorescent NPs conjugated to IgG Fc were transported across an in vitro airway epithelial cell monolayer (22). Yet, no studies, to our knowledge, have used drug-encapsulated NPs to target the FcRn for oral drug delivery applications.

Here, we developed polymeric NPs surface-modified with Fc to target FcRn, resulting in transepithelial transport both in vitro and in vivo. NP-Fc demonstrated enhanced transport specifically mediated by FcRn across an intestinal epithelial monolayer in vitro. In mice, NP-Fc were imaged crossing the intestinal epithelium and entering the lamina propria after oral administration. NP-Fc were also measured in several organs after oral administration, indicating that NP-Fc were able to reach systemic circulation. Finally, NP-Fc encapsulating insulin as a model NP-based therapeutic for diabetes were administered orally and elicited a hypoglycemic response. These results demonstrate that NPs targeted to the FcRn are capable of crossing an epithelial barrier in vivo using a transcytosis pathway, making NP-based oral drug delivery formulations readily suitable to deliver therapeutics to the systemic circulation.

RESULTS

Preparation of Fc-targeted NPs

NPs were formed from biodegradable and biocompatible poly(lactic acid)–b-poly(ethylene glycol) (PLA-PEG) block copolymers. PLA is a biodegradable polymer used in many U.S. Food and Drug Administration–approved products and forms the NP core owing to its hydrophobicity. PEG is a biocompatible polymer that remains on the NP surface owing to its hydrophilic nature and forms the NP corona. PLA-PEG was synthesized using ring-opening polymerization with a free terminal maleimide group (PLA-PEG-MAL) to conjugate the Fc portion of IgG.

Studies of NPs have shown enhanced intestinal uptake as particle size decreases (23, 24). However, these previous studies on particle size have mostly focused on nonspecific uptake by M cells in the Peyer’s patches as opposed to specifically targeting the NPs to a transcytosis pathway such as the FcRn-mediated pathway. Here, the nanoprecipitation self-assembly method (Fig. 2A) (25) was used to generate particles with a mean hydrodynamic diameter of 55 nm and a polydispersity of 0.05 (Fig. 2B). Polyclonal IgG Fc fragments were covalently conjugated to PEG using maleimide-thiol chemistry. 2-Iminothiolane was used to modify the Fc with thiol groups (Fc-SH). Fc-SH was incubated with NPs for conjugation (Fig. 2A).

Fig. 2. NP assembly, characterization, and in vitro transepithelial transport.

(A) Schematic of NP-Fc assembly. NPs consist of a biodegradable PLA core for drug encapsulation and a PEG surface coating for particle stability and to reduce phagocytic uptake. NPs were formed using the nanoprecipitation self-assembly method (25) and surface-modified with IgG Fc for FcRn targeting. (B) Dynamic light scattering measurements for nontargeted NPs and NP-Fc. Data are means ± SD (n = 3). (C) IgG Fc ligand density on the NP surface with (Fc-SH) and without (Fc) thiol modification of the IgG Fc. Data are means ± SD (n = 3). (D) In vitro transepithelial transport of nontargeted NPs, NP-Fc, and NP-Fc with an excess of human IgG Fc as a blocking agent for FcRn. Data are expressed as mean basolateral 3H disintegrations per minute (DPM) as a percentage of the initial amount of 3H (±SEM; n = 4 wells per group). *P < 0.05, two-tailed Student’s t test.

The amount of Fc conjugated to the NPs was measured for both Fc-SH and unmodified Fc (Fig. 2C). Unmodified Fc resulted in a lower ligand density than Fc-SH, indicating minimal nonspecific interactions between Fc and the NP surface and that the unbound Fc was successfully separated from NPs using centrifugal filtration. The ligand density for Fc-SH was 32-fold higher than that for unmodified Fc, indicating that Fc was bound on the NP surface. In addition, the hydrodynamic diameter of the NPs increased from 55 to 63 nm after Fc conjugation (Fig. 2B)—an increase consistent with the hydrodynamic diameter of IgG Fc (~3 nm) (26). The surface charge showed only a minor change from −4.3 ± 0.4 mV for NPs to −5.6 ± 1.1 mV for NP-Fc (mean ± SD, n = 3; P > 0.05, Student’s t test).

In vitro transepithelial transport

In vitro NP transepithelial transport studies were conducted using an epithelial cell monolayer model with Caco-2 cells, a human epithelial colorectal adenocarcinoma cell line typically used as a model of the intestine for drug permeability testing. Caco-2 cells endogenously express human FcRn and human β2-microglobulin, and have previously been used for IgG transcytosis studies (8, 27). 3H-labeled NPs were used to measure transport across the Caco-2 monolayer. Using this system with a pH gradient established from the apical to basolateral side of the Caco-2 polarized monolayer (to mimic the physiological pH of the duodenum and enhance apical binding), we measured the transcytosis of nontargeted NPs (control) and NP-Fc. 3H measurements for NP-Fc on the basolateral side were twofold greater than nontargeted NPs after 24 hours (Fig. 2D), indicating that Fc on the NP surface significantly enhanced transepithelial transport in vitro. When NP-Fc was combined with a 50-fold excess of free IgG Fc as a blocking agent for the FcRn transcytosis pathway, NP-Fc transport was significantly reduced, indicating that the enhanced transport of NP-Fc is largely mediated by the FcRn (Fig. 2D).

FcRn expression in mice

FcRn expression throughout the entire small and large intestine of wild-type mice was confirmed by Western blot (Fig. 3A). Quantification of band intensity showed that FcRn expression decreased in the distal sections of the small intestine and the colon (Fig. 3B). Immunohistochemistry showed that FcRn was localized to the epithelium of the intestinal villi of the duodenum of wild-type mice (Fig. 3C).

Fig. 3. FcRn expression and NP intestinal uptake in mice.

(A) Western blot of mouse FcRn (mFcRn) in mouse intestinal tissue. (B) Quantification of Western blot band intensity in (A). The relative band intensity was calculated as the ratio of mouse FcRn to β-actin. (C) Immunohistochemistry on sections of mouse duodenum. Mouse FcRn appears brown. The negative control tissue was stained with polyclonal IgG. (D) Fluorescently labeled NPs (red) were administered to fasted wild-type (WT) mice by oral gavage, and the intestines were collected for sectioning and imaging 1.5 hours later. The panels are confocal fluorescence images of 12-μm sections of mouse duodenum. Cell nuclei were stained with 4′,6-diamidino-2-phenylindole (DAPI) (blue). Images are representative of n = 3 mice.

In vivo absorption and biodistribution

In vivo transport of NP-Fc across the intestinal epithelium was visualized using fluorescently labeled NPs. Fasted wild-type mice were orally administered the fluorescently labeled NPs, and sections of the duodenum were collected and analyzed by confocal fluorescence microscopy. Figure 3D shows representative images for both nontargeted NPs and NP-Fc. For the nontargeted NPs, the fluorescence from the NPs was not observed in the villi, suggesting that the NPs were unable to enter the villi. However, for the NP-Fc, fluorescence was observed inside the villi on the basolateral side of the epithelial cells, indicating that NP-Fc crossed the intestinal epithelium and entered the lamina propria (Fig. 3D).

The biodistribution and absorption efficiency of both targeted and nontargeted NPs were quantitatively measured by radiolabeling the NPs with 14C. Figure 4A shows the biodistribution of nontargeted NPs and NP-Fc over the course of 8 hours after oral administration to fasted wild-type mice. For the nontargeted NPs, a small amount of 14C was measured in the organs. By contrast, a large amount of 14C was measured in the spleen, kidneys, liver, and lungs for NP-Fc, indicating that NP-Fc entered systemic circulation after oral administration and reached several organs known to express FcRn (11). The 14C in the organs was transient, peaking at 2.5 hours after delivery and clearing from the organs at later time points (Fig. 4A). To ensure that the 14C remained with the NPs over the course of the experiment, we measured the release of 14C from the NPs and observed no release over 24 hours (Fig. 4B).

Fig. 4. NP absorption and biodistribution in mice.

(A) Biodistribution of 14C-labeled nontargeted NPs and NP-Fc after oral administration to fasted WT mice. Data are mean percent initial dose (%ID) per gram of tissue ± SEM (n = 5 mice per time point). (B) Release of 14C from 14C-labeled NPs in PBS at 37°C. Data are means ± SD for n = 4 release experiments. (C) Total absorbed 14C over time for nontargeted NPs and NP-Fc after administration by oral gavage. Data are mean %ID measured in all of the organs added together ± SEM (n = 5 mice per time point). **P < 0.01 for comparison of nontargeted NPs and NP-Fc at respective time point, two-tailed Student’s t test.

The total 14C absorbed over time was calculated by summing the 14C measured in each of the organs (spleen, kidneys, liver, heart, and lungs) at a specific time point (Fig. 4C). Significantly higher amounts of 14C were absorbed for NP-Fc at 1.5, 2.5, and 4 hours compared with nontargeted NPs, indicating that Fc targeting enhanced absorption. The area under the curve (AUC) was used to calculate the oral absorption efficiency, which was 1.2 ± 0.2%*hour for nontargeted NPs and 13.7 ± 1.3%*hour for NP-Fc (mean ± SEM, n = 5 mice per time point; P < 0.01, Student’s t test). This difference in AUC suggests that IgG Fc targeting was responsible for an 11.5-fold increase in NP absorption.

Oral delivery of insulin to wild-type mice

NP-Fc encapsulating insulin were developed as a model NP-based therapeutic for diabetes that could be orally administered and evaluated for eliciting a pharmacologic response. Insulin NPs (insNPs) were prepared using nanoprecipitation, resulting in an insulin load of 0.5% (w/w). The nanoprecipitation method allowed the particle size to remain small (mean hydrodynamic diameter = 57 nm) while still allowing insulin encapsulation. The release of insulin from insNPs in phosphate-buffered saline (PBS) at 37°C demonstrated a strong burst release in the first hour followed by a controlled release (Fig. 5A). The insulin release profile was advantageous because it allowed all of the insulin to be delivered before complete clearance of the particles after 10 hours. To determine the bioactivity, we collected and injected the insulin released from insNPs into fasted wild-type mice through the tail vein. The bioactivity was measured by monitoring the blood glucose and comparing the response to an equivalent dose of free insulin solution (3.3 U/kg). The released insulin generated a hypoglycemic response in mice (Fig. 5B), indicating that the encapsulated insulin was bioactive after release from the insNPs.

Fig. 5. Encapsulation of insulin and oral delivery of Fc-targeted insNPs to mice.

(A) Release of insulin from insNPs into PBS. Data are means ± SD (n = 3 per time point). (B) Blood glucose response of fasted wild type (WT) mice to insulin encapsulated and released from NPs in (A) before administration. Fasted WT mice received free insulin (3.3 U/kg) administered by tail vein injection. Data are means ± SEM (n = 3 mice per group). (C) Blood glucose response of fasted WT mice to free insulin solution, NP-Fc containing no insulin, nontargeted insNPs, and insNP-Fc, each administered by oral gavage. Data are means ± SEM (n = 6 mice per group). *P < 0.05 for comparison of nontargeted insNPs and insNP-Fc at corresponding time points, two-tailed Student’s t test. (D) Blood glucose response of fasted WT mice to insNP-Fc, insNP-Fc administered concurrently with excess of IgG Fc, and insNPs with chicken IgY Fc fragments, each administered by oral gavage. Data are means ± SEM (n = 5 mice per group). **P < 0.01 for comparison between insNP-Fc with insNP-Fc + free IgG Fc at the 15- and 19-hour time points and between insNP-IgG Fc and insNP-IgY Fc at the 10-, 15-, and 19-hour time points using a two-tailed Student’s t test. (E) Blood glucose response to equivalent insulin doses (3.3 U/kg) administered by tail vein injection into fasted WT and FcRn knockout (KO) mice. Data are means ± SEM (n = 3 mice per group). (F) Fasted FcRn KO mice blood glucose response to free insulin solution, NP-Fc containing no insulin, nontargeted insNPs, or insNP-Fc, each administered by oral gavage. Data are means ± SEM (n = 5 mice per group). (G) Fasted WT and FcRn KO mice were dosed by oral gavage with nontargeted insNPs and insNP-Fc at two different doses. *P < 0.05 for comparison between insNP-Fc at 1.1 U/kg and each of the other groups at corresponding time points, two-tailed Student’s t test.

The hypoglycemic response generated after oral administration of the targeted insNPs (insNP-Fc) was tested using fasted wild-type mice and compared with the efficacies of nontargeted insNPs, free insulin, and NP-Fc without insulin (Fig. 5C). The free insulin administered orally did not generate a glucose response in the mice [unlike the free insulin injected into the tail vein that was able to generate a glucose response (Fig. 4B)]. The NP-Fc without insulin was also unable to generate a glucose response. The glucose response generated by nontargeted insNPs was not different from that generated by the control groups at any time point. However, insNP-Fc caused a significant hypoglycemic response in the mice, reducing the glucose during the first 10 hours after administration (Fig. 5C). This is consistent with the biodistribution (Fig. 4A) and insulin release data (Fig. 5A), which demonstrated that the particles were cleared and the insulin was released within 10 hours, respectively. The blood glucose level then increased and was similar to that of the control groups by 15 hours. The insNP-Fc insulin dose required to generate the hypoglycemic response was 1.1 U/kg, which is clinically relevant (28) and lower than other oral insulin delivery systems that require 10 to 100 U/kg to generate a glucose response (29). When compared with the glucose response from free insulin administered by tail vein injection (Fig. 5B), the orally administered insNP-Fc resulted in a prolonged (15 versus 1.5 hours) hypoglycemic response (Fig. 5C).

To demonstrate that the enhanced hypoglycemic response generated by insNP-Fc was due specifically to the IgG Fc ligand on the NP surface, we tested several additional control groups. In the first control, insNP-Fc was administered concurrently with a 50-fold excess of free IgG Fc. In the second control, chicken IgY Fc fragments were used instead of human IgG Fc fragments conjugated to insNPs. [Chicken IgY is the functional equivalent of IgG in nonmammalian species such as birds, but does not bind to mouse FcRn (30).] Both control groups had hypoglycemic responses that were significantly less than that with insNP-Fc (Fig. 5D), indicating that the use of the IgG Fc as a targeting ligand was responsible for the enhanced hypoglycemic response.

Oral delivery of insulin to FcRn knockout mice

The role of FcRn in NP transepithelial transport was tested by repeating the efficacy experiment using FcRn knockout mice. FcRn knockout mice had the same insulin sensitivity as the wild-type mice, so the same insulin dose was used for both strains (Fig. 5E). In contrast to the results in the wild-type mice (Fig. 5C), insNP-Fc did not generate a hypoglycemic response significantly different from the other three groups in the FcRn knockout mice (Fig. 5F). In these FcRn knockout mice, the response generated by insNP-Fc resembled the response generated by nontargeted insNPs in the wild-type mice (Fig. 5C), suggesting that the benefit gained from using Fc was specifically due to FcRn.

In vivo glucose response dose dependency

The glucose response in both wild-type and FcRn knockout mice was evaluated for dose dependency using two different doses of insNP-Fc: 0.66 and 1.1 U/kg (Fig. 5G). For the FcRn knockout mice, there was no difference in the glucose response between the two doses of insNP-Fc. However, for the wild-type mice, the glucose response was significantly greater at the higher dose of insNP-Fc, suggesting a possible dose dependence in the wild-type mice.

DISCUSSION

For many diseases, oral administration of therapeutics is the standard of care because daily therapy is required. In some cases, chronic diseases, such as cancer, that have been treated in the clinic with intravenous infusions are now increasingly being treated with oral therapeutics because patients prefer the convenience of oral administration relative to parenteral administration. For NP-based therapeutics to be a practical treatment for many diseases, NP formulations appropriate for oral administration are necessary. The most difficult barrier to the effective oral administration of NPs is the intestinal epithelium, which limits the absorption of NPs. To date, there is no practical solution to this problem.

There have been many attempts to develop oral drug delivery systems that overcome this barrier (29). For example, permeation enhancers have been used to open tight junctions to allow both paracellular and transcellular transport of drugs across the epithelium (31). Mucoadhesive biomaterials have been used to increase the retention time and local concentration of drugs near the apical surface of epithelial cells (32). Many oral NPs have been engineered for uptake by M cells in the Peyer’s patches, although this limits the surface area available for absorption and exposes NPs to underlying immune cells (33). Finally, a few NP formulations have targeted cell receptors, but they still suffer from low bioavailability and require high oral drug dosages (29, 34).

We have taken a novel approach to address this problem by developing drug-encapsulated NPs capable of targeting FcRn for transepithelial transport and enhanced NP intestinal absorption after oral administration. FcRn has been shown to mediate the transcytosis of IgG across several epithelial and endothelial barriers (11), and more recently, the transcytosis of fluorescent NPs was demonstrated across a monolayer of Calu-3 airway epithelial cells in vitro after adsorption of Fc on the NP surface (22). Harnessing the transcytosis pathway to cross the intestinal epithelium offers the advantage of leaving intact the integrity of the epithelial barrier, avoiding potential safety issues and adverse effects associated with manipulating the permeability of the intestine for paracellular or transcellular transport. An additional advantage of targeting the FcRn is that this receptor is expressed throughout the intestine, providing a significant increase in the available absorption surface area for NP-Fc, which is in contrast with other drug delivery systems that target only a specific portion of the intestine such as the Peyer’s patches (6).

We have demonstrated that targeting FcRn for transepithelial transport by modifying the NP surface with IgG Fc provides a successful approach for the oral delivery of NPs. Using a mouse model, we demonstrated that FcRn enabled the NP-Fc to cross the intestinal epithelium and reach systemic circulation. The use of IgG Fc to target the NPs to FcRn resulted in absorption efficiency at least 11.5-fold higher than that for nontargeted NPs, but the efficiency could potentially be higher because some 14C could have been absorbed in other tissues not measured in these studies. Furthermore, insNPs targeting FcRn, a model NP-based therapeutic for the treatment of diabetes, were able to generate a hypoglycemic response after oral administration at a clinically relevant insulin dose that is significantly lower than the doses required by other drug delivery systems designed for oral insulin delivery (29). Moreover, rodents significantly down-regulate FcRn expression in the intestine after weaning (35), but humans continue to express FcRn into adulthood. Therefore, although NP-Fc can increase transepithelial transport in mice, the transport in humans could potentially be even more efficient.

There are some potential limitations to this technology that need to be investigated. For example, FcRn is expressed in many tissues in addition to the intestine. Using IgG Fc to target the FcRn could potentially result in uptake in tissues other than the targeted tissue. However, FcRn expression in tissues such as the vascular endothelium, blood-brain barrier, and lungs could also allow the technology to be expanded for drug delivery across other cellular barriers. Immunological responses to Fc on the NP surface need to be investigated as well. The presence of IgG Fc may result in faster blood clearance of the NPs because of interactions with Fc receptors. NP-Fc may also generate immunological responses and thus could potentially be used for oral vaccine applications (36).

This technology may have a major impact on the treatment of several diseases by enabling NP-based therapies to be orally administered. In addition, the encapsulation of drugs or biologics that are currently limited by low bioavailability into NPs that target FcRn may enable markedly more efficient oral delivery of the therapies. The path forward for translation of this technology to the clinic is to test the NP-Fc in larger animal models such as pigs and nonhuman primates. Some key parameters that need to be understood for success in the clinic are the ideal release profile for an oral insulin formulation and how absorption varies based on diet and patient-to-patient variability.

MATERIALS AND METHODS

Study design

This study aimed to develop targeted Fc-functionalized NPs capable of binding to the FcRn protein, and evaluate whether this approach enabled the transcytosis of the FcRn-NP complex across the intestinal epithelial barrier using both in vitro and in vivo models. Fc-targeted NPs were first tested using an in vitro intestinal epithelium model and the transport across the epithelial barrier was quantified by 3H-labeling of the NPs. Fc-targeted NPs were then tested in vivo using fluorescently-labeled NPs to observe transport across the intestinal epithelium and 14C-labeled NPs to quantify the biodistribution and absorption efficiency after oral administration. Finally, Fc-targeted NPs containing insulin were used for plasma glycemic modulation studies in both wild-type and FcRn knockout mice as a proof-of-concept that targeted NPs were capable of oral delivery of bioactive protein therapeutics. For the imaging and 14C experiments, the mice were selected based on weight and then randomly assigned to different treatment groups. For the plasma glycemic modulation studies, the mice were selected based on initial fasted glucose values and then randomly assigned to different treatment groups. The studies were not blinded. Sample sizes (n = 5 to 6 per group) for all of the experiments were sufficient to detect statistically significant differences between treatment groups, with all measurements included in analyses. Endpoints for glucose experiments were determined by total fasting time, which was limited to 24 h.

Polymer synthesis

d,l-Lactide (Sigma-Aldrich) and MAL-PEG-OH (JenKem Technology) were used to synthesize PLA-PEG-MAL by ring-opening polymerization. d,l-Lactide (3 g, 20.8 mmol) and MAL-PEG-OH (544 mg, 0.16 mmol) were dissolved in 15 ml of anhydrous toluene in a round-bottom flask. Tin(II) ethylhexanoate (38 mg, 0.09 mmol) was then added. The flask with condenser was placed in an oil bath, purged with nitrogen for 10 min, heated to 120°C, and reacted overnight while 4°C water circulated through the condenser. Toluene was then evaporated, and the polymer was precipitated in a 50:50 (v/v) mixture of ice-cold methanol and diethyl ether and vacuum-dried. The PLA-PEG-MAL was characterized by 1H nuclear magnetic resonance (400 MHz), δ = 5.28 to 5.11 [br, -OC-CH(CH3)O- in PLA], 3.62 (s, -CH2CH2O- in PEG), 1.57 to 1.45 (br, -OC-CHCH3O- in PLA). With gel permeation chromatography, polymer Mn = 12.5 kD with Mw/Mn = 1.47 relative to polystyrene standards. PLA (inherent viscosity 0.50 dl/g) with terminal carboxylate groups (Lactel) was conjugated to 14C doxorubicin (PerkinElmer) (PLA-14C) and Alexa Fluor 647 (AF647) hydrazide tris(triethylammonium) salt (Invitrogen) (PLA-AF647) were prepared the same way. PLA (30 mg, 0.83 μmol) in 1 ml of dimethylformamide (DMF) was reacted with N-hydroxysuccinimide (NHS) (0.5 mg, 4.2 μmol) in the presence of 1-ethyl-3-[3-dimethylaminopropyl] (0.8 mg, 4.2 μmol) overnight. PLA-NHS was precipitated in a 50:50 (v/v) mixture of ice-cold methanol and diethyl ether and vacuum-dried. For PLA-14C, PLA-NHS (18 mg, 0.5 μmol) was mixed overnight with 14C doxorubicin (0.27 mg, 0.5 μmol) in 1 ml of DMF. For PLA-AF647, PLA-NHS (50 mg, 1.4 μmol) was mixed overnight with AF647 (1.5 mg, 1.4 μmol) in 1 ml of DMF. Both polymers were precipitated in a mixture of 50:50 (v/v) ice-cold methanol and diethyl ether and vacuum-dried.

Synthesis and characterization of NP-Fc

To prepare the NP-Fc, we dissolved 3 mg of PLA-PEG-MAL in 300 μl of acetonitrile and added it dropwise to 1.5 ml of water. The solution was mixed for 2 hours, and the NPs were purified by filtration with Millipore Amicon Ultra 100,000 NMWL (nominal molecular weight limit). The NPs were washed twice with water and twice with PBS containing 5 mM EDTA. Concurrently, 86 μg of purified human polyclonal IgG Fc prepared by papain digestion (Bethyl Laboratories) or 95 μg of chicken IgY Fc (Jackson ImmunoResearch Laboratories) in PBS containing 5 mM EDTA was reacted with 0.48 μl of 2-iminothiolane (5 mg/ml) (Traut’s Reagent) for 1 hour. The modified Fc was added to the NPs and mixed for 1 hour to allow conjugation at 4°C. The NP-Fc were washed with PBS using Millipore Amicon Ultra 100,000 NMWL. The conjugation of IgG Fc to the NP surface was measured with a protein bicinchoninic acid (BCA) assay from Lamda Biotech. Particle diameter and surface charge (ζ potential) were measured with dynamic light scattering with a Brookhaven Instruments ZetaPALS.

In vitro transcytosis studies

Transepithelial transport studies used Transwell plates (Costar) with a Caco-2 [American Type Culture Collection (ATCC)] cell density of 5.5 × 104 in medium [ATCC-formulated Eagle’s minimum essential medium with aqueous penicillin G (100 U/ml), streptomycin (100 U/ml), and fetal bovine serum (20%)]. On the day of the transport experiment, the medium was changed to Hanks’ balanced salt solution (HBSS) (pH 6.5) in the apical chamber and HBSS (pH 7.4) in the basolateral chamber and allowed to equilibrate for 1 hour at 37°C and 5% CO2. Before and at the end of the experiment, the monolayer integrity was checked by measuring the transepithelial resistance (TEER) with a Millicell-ERS (Millipore). TEER values were 900 to 1000 Ω/cm2 for wells used in transport experiments, and the TEER remained constant throughout the experiments. 3H-labeled NPs were prepared by blending 50 μg of 3H–poly(lactic-co-glycolic acid) (PLGA) (PerkinElmer) with 1 mg of PLA-PEG-MAL in 100 μl of acetonitrile before nanoprecipitation in 500 μl of water and then washed in HBSS (pH 6.5) before the transport experiment. The apical solution was then replaced with a solution of 100 μg of 3H-labeled NPs or NP-Fc in 250 μl of HBSS (pH 6.5). The NP formulations were incubated for 24 hours before measuring the 3H content in the basolateral chamber. The basolateral solution was collected and added to a Hionic-Fluor scintillation cocktail (PerkinElmer) before analysis with a Packard Tri-Carb Scintillation Analyser. At the end of the experiment, the TEER was measured again to verify monolayer integrity. For the IgG Fc blocking experiment, a 50× molar excess of IgG Fc relative to Fc on the NP-Fc surface was added concurrently with NP-Fc to the apical chamber, and the 3H content in the basolateral chamber was measured after 24 hours.

Western blot

Sections of small intestine and colon were removed from wild-type mice after euthanasia. Intestinal epithelial cells were isolated (37, 38), and the protein was extracted. Protein concentrations in the extracts were determined with the BCA assay. Extracts were resolved on a 12% SDS–polyacrylamide gel electrophoresis gel under reducing conditions. Proteins were transferred onto a nitrocellulose membrane. The membrane was blocked with 5% nonfat milk, probed with rabbit anti-mouse FcRn (Santa Cruz Biotechnology) for 1 hour, and then incubated with goat anti-rabbit IgG–horseradish peroxidase (HRP) (Santa Cruz Biotechnology). All blocking, incubations, and washes used PBS-T (PBS with 0.05% Tween 20). Detection was by chemiluminescence. Band intensity was quantified with ImageJ.

Immunohistochemistry

Small intestine tissues were harvested and fixed in 10% formalin overnight. After ethanol dehydration, tissues were paraffin-embedded and cut into 8-μm-thick sections, mounted on slides, and dried overnight. The tissues were then rehydrated with xylene and ethanol. Endogenous peroxidase activity, endogenous biotin, and nonspecific proteins were blocked with 3% H2O2, avidin blocking agent, and 10% goat serum, respectively. The samples were incubated with polyclonal rabbit IgG or anti-mouse FcRn IgG (Santa Cruz Biotechnology) primary antibody overnight and then incubated with biotinylated anti-rabbit secondary antibodies (Santa Cruz Biotechnology) and with streptavidin-HRP, developed with DAPI, and mounted with hematoxylin counterstain.

In vivo fluorescence imaging

All animal studies were conducted under the supervision of the Massachusetts Institute of Technology (MIT) Division of Comparative Medicine in compliance with the National Institutes of Health (NIH) Guide for the Care and Use of Laboratory Animals. Wild-type BALB/c mice (Charles River Laboratories) (n = 3) were fasted overnight before gavage. Fluorescently labeled NPs were prepared by blending 100 μg of PLA-AF647 with 1 mg of PLA-PEG-MAL in 100 μl of acetonitrile before nanoprecipitation in 500 μl of water. Fluorescently labeled NPs and NP-Fc were washed three times in water until the flow-through was clear and suspended in 200 μl of water (7 ml/kg). The suspension was then administered to the mice by oral gavage. After 1.5 hours, the mice were euthanized. Duodenum tissue sections were frozen in Tissue-Tek OCT with liquid nitrogen. Cross sections of the tissue were obtained with a Leica CM1900 cryostat with a thickness of 12 μm. The tissue was air-dried overnight and then stained with ProLong Gold (Life Technologies) antifade reagent with DAPI. Fluorescent images were obtained with a Zeiss LSM 710 NLO scanning confocal microscope under oil immersion at 40× magnification.

In vivo biodistribution

To prepare 14C-labeled NPs, we blended 450 μg of PLA-14C with 3 mg of PLA-PEG-MAL in 300 μl of acetonitrile before nanoprecipitation in 1.5 ml of water. NPs were washed twice with water and twice with PBS. 14C release was measured by preparing a batch of 14C NPs in PBS (pH 7.4) and dividing the batch equally into 500-μg samples for incubation at 37°C. At each time point, samples were collected, washed three times with PBS with Millipore Amicon Ultra 100,000 NMWL, and then added to 15 ml of Hionic Fluor scintillation cocktail. The activity was counted with a Packard Tri-Carb Scintillation Analyser. For the in vivo biodistribution experiments, 6- to 12-week-old wild-type mice were fasted 8 hours before oral gavage of 1.5 mg (0.1 μCi per mouse) of 14C-labeled NPs and NP-Fc in PBS (7 ml/kg). Groups of mice (n = 5 mice) were euthanized at each time point, and the spleen, kidneys, liver, lungs, and heart were harvested. Each organ was placed directly in a scintillation vial except for the liver, which was homogenized, and ~100 mg was analyzed. Each organ was solubilized in 2 ml of Solvable (PerkinElmer) for 12 hours at 60°C and then decolored with 200 μl of 0.5 M EDTA (Invitrogen) and 200 μl of 30% hydrogen peroxide (Fisher Scientific) for 1 hour at 60°C. The activity was counted in 15 ml of Hionic-Fluor scintillation cocktail with a Packard Tri-Carb Scintillation Analyser. To determine 100% dose, we counted vials of 500 μg of NPs and NP-Fc in 15 ml of Hionic-Fluor scintillation cocktail. For the oral absorption efficiency, total 14C counted in all tissues was added at each time point. The AUC of total absorbed 14C versus time was calculated with the trapezoid method and divided by the initial dose to determine the oral absorption efficiency.

Insulin encapsulation and release

insNPs were prepared by blending 900 μg of human recombinant insulin (Sigma-Aldrich), 450 μg of PLGA (50:50 glycolide/lactide, inherent viscosity of 0.20 dl/g) with terminal carboxylate groups (Lactel), and 3 mg of PLA-PEG-MAL together in 525 μl of dimethyl sulfoxide before nanoprecipitation in 2.1 ml of water. Free insulin was removed with Millipore Amicon Ultra 100,000 NMWL by washing with water and with PBS. Insulin encapsulation was measured by heating the NPs to 60°C for 30 min, and insulin was quantified with a BCA assay or insulin ELISA (enzyme-linked immunosorbent assay) kit (Millipore). Insulin release from the NP was measured by dividing a batch of insNPs equally into 24-kD dialysis units (Pierce) and incubating at 37°C in PBS (pH 7.4). At each time point, three samples of insNPs were collected, washed with PBS with Millipore Amicon Ultra 100,000 NMWL, heated to 60°C for 30 min, and measured for insulin with a BCA assay.

Insulin activity

Insulin activity was measured by preparing insNPs and allowing insulin release for 2 hours at 37°C in PBS (pH 7.4). Wild-type mice were fasted for 8 hours, and the mice (n = 3 per group) used were chosen so that the mean initial blood glucose levels were the same for each group. Released insulin (3.3 U/kg) was administered to the fasted mice by tail vein injection. An equivalent dose of free insulin by mass was administered by tail vein injection to another group of fasted mice. The blood glucose level was measured with the Contour blood glucose monitor (Bayer).

In vivo efficacy

Six- to 12-week old wild-type or FcRn knockout (39) mice (Jackson Laboratories) were fasted for 8 hours. Mice (n = 5 to 6) were chosen per group such that the mean initial blood glucose levels were the same per group. insNPs or insNP-Fc (150 or 250 μg) (insulin dose: 0.66 or 1.1 U/kg) in PBS (7 ml/kg) were administered to the mice by oral gavage. For controls, free insulin (1.1 U/kg) and 250 μg of NP-Fc without insulin in PBS (7 ml/kg) were administered by oral gavage. For the excess IgG Fc control, 250 μg of insNP-Fc was formulated with 50× molar excess of IgG Fc in PBS (7 ml/kg) before administration to the mice by oral gavage. The blood glucose level was measured as described above.

Statistical analysis

All data are presented as means with either SD or SEM as indicated. Statistical significance was determined by a two-tailed Student’s t test (α = 0.05) assuming equal variance.

REFERENCES AND NOTES

  1. Acknowledgments: We thank the W. M. Keck Microscope Facility at the Whitehead Institute, in particular W. Salmon for her assistance with the fluorescence microscopy imaging. We thank B. Tang for his help with the confocal microscopy image analysis. Funding: Supported in part by a Koch–Prostate Cancer Foundation Award in Nanotherapeutics (R.L. and O.C.F.); the National Cancer Institute Center of Cancer Nanotechnology Excellence at MIT-Harvard (U54-CA151884); a National Heart, Lung, and Blood Institute Program of Excellence in Nanotechnology Award (contract HHSN268201000045C); NIH grant EB000244; and NIH R01 grant EB015419-01. R.S.B. is supported by NIH DK53056 and Harvard Digestive Disease Center grant DK0034854. E.M.P. is supported by a National Defense Science and Engineering Graduate research fellowship and a Center of Cancer Nanotechnology Excellence graduate research fellowship (5 U54 CA151884-02). Author contributions: E.M.P. and F.A. designed and performed the NP synthesis and characterization, in vitro experiments, and in vivo experiments; analyzed the data; did the statistical analysis; and wrote the manuscript. O.C.F. designed the experiments, analyzed the data, and wrote the manuscript. R.L., R.K., and R.S.B. designed the experiments, analyzed the data, and critically reviewed the manuscript. T.T.K. and E.L.-N. analyzed the data and critically reviewed the manuscript. Competing interests: O.C.F., R.K., and R.L. have financial interests in BIND Therapeutics, Selecta Biosciences, and Blend Therapeutics, which are developing NP therapeutics. O.C.F., R.L., F.A., E.M.P., and T.T.K. are inventors on the patent application related to this technology. BIND and Selecta have an exclusive license from Brigham and Women’s Hospital (BWH) and MIT for this technology. Data and materials availability: Materials are readily available and will be provided under the material transfer policies of the MIT and BWH.
View Abstract

Navigate This Article