PEGylated insulin loaded complexation hydrogels for protected oral delivery

Abstract

While a number of enteric coatings and pH-sensitive oral delivery vehicles have been developed, they lack the ability to sufficiently protect proteins from proteolytic degradation once released from the carrier. In this work, we show that H-bonded, pH-sensitive poly(methacrylic acid-grafted ethylene glycol) glycol (henceforth designated as P(MAA-g-EG) gels) exhibit great promise as protein carriers, as they utilize poly(ethylene glycol) (PEG) chains to promote mucoadhesion in the small intestine, increasing the chances that the drug is released within the villus of the absorptive intestinal wall. Importantly, PEG was also conjugated to the B29-lysine (LysB29) position of insulin in order to protect the drug from proteolytic degradation once released in the small intestine and adhere the drug to the intestinal epithelium through improved mucoadhesion. PEG-conjugated (PEGylated) molecules were found to actively participate in the carrier loading and release mechanism, with the drug conjugate hydrogen bonding to the MAA while in the collapsed state and subsequently repulse the drug above the polymer's isoelectric point. This effect was enhanced through the evaluation of PEG graft density within the carrier. Cellular transport and changes in transepithelial resistance caused by the PEGylated insulin (PI) in the presence of P(MAA-g-EG) microparticles were analyzed using a 1:1 co-culture of human colon adenocarcinoma (Caco-2) and: the mucus-secreting human colon carcinoma cell(HT-29-MTX). Finally, the in vivo absorption of insulin was measured in Sprague-Dawley rats to ensure that the PEGylated insulin conjugates are biologically active, as well as to compare the bioavailability to control insulin. Collectively, these results lead toward the development of a novel system for improved insulin delivery, with improved stability of insulin through PEGylation.

1. Introduction

Controlled drug delivery uses materials (typically biopolymeric compounds), to protect therapeutics from early degradation and controls their release at prolonged timescales (from hours to months) at relevant concentrations through various mechanisms [1]. “Smart” delivery systems rely on stimuli-responsive materials used to deliver therapeutic agents to precise locations at specific times, where these triggers can be internal (e.g. pH, temperature) or external (e.g. magnetic field). In this work, a smart delivery system for the oral delivery of the protein insulin is developed, utilizing both anionic materials and PEGylation of the insulin molecule as shown in Fig. 1. The technical challenges are immense when dealing with the oral delivery of macromolecules such as peptides and proteins, as their large size and potential for degradation limit their travel through the GI tract. Challenges associated with the delivery of proteins through the gastrointestinal (GI) tract include harsh pH in the stomach, digestive enzymes, and the requirement of the protein to travel through the mucus and epithelial cell layer of the intestinal wall to reach the systemic circulation [2].

Fig. 1.

Schematic of insulin conjugated with P(MAA-g-EG) hydrogel. Created with BioRender.com.

One major area of focus for the protein delivery field is an alternate delivery route for insulin, a macromolecule that is a critical part of diabetes treatment and currently only commercially available via injection. Given the nature of injections, there is reduced patient compliance since it is inconvenient, painful, and must be administered multiple times a day. About 13% of the United States adult population has diabetes [3]. The risk of diabetes is markedly higher in African American men and women by 44.7% and 55.3%, respectively, and 51.8% and 51.5% in Hispanic men and women. Even though rates of diabetes fell between 2005 and 2010, this was mainly observed in Caucasian populations. At the same time, diabetes rates rose from 5.5% to 8.5% in people of Mexican American ancestry [4]. Traditional health complications such as stroke, heart failure, peripheral neuropathy, and kidney disease are characteristic of diabetes. However, there is growing evidence suggesting that there are emerging complications correlated with diabetes including cancer, infections, liver disease, and functional and cognitive disabilities [5]. Considering the prevalent health disparities in addition to the numerous health implications there is an urgent need to create accessible, patient-friendly options.

For the past few decades, a great deal of work has focused on attempts to develop an oral platform for insulin delivery. However, insulin delivered orally will encounter several physical and physiological barriers including the hepatic first-pass effect in the liver, the presence of proteolytic enzymes and the acidic environment in the GI tract, and the presence of the mucous layer, intestinal epithelium, and tight junctions which limit the solubility and permeability of insulin across biological membranes [6]. We believe that the convergence of a number of different technologies is needed in order to address all of these barriers simultaneously, including anionic polymers that are pH responsive to protect the therapeutic through the stomach and release in the intestines, in addition to improving the stability of insulin with chemical conjugation. [7,8]

In our laboratory, we have successfully developed a family of copolymer systems that has the ability to address a number of these barriers including protecting the protein from the harsh stomach environment, releasing the insulin into the small intestine, and improving residence through adhesion to the mucus layer of the intestinal wall [9–11].These systems consists of hydrogels with poly(ethylene glycol) (PEG) grafted on poly(methacrylic acid) (PMAA), designated as P(MAA-g-EG) [12,13]. Bioavailability up to 12.8% of the total dose has been achieved with this delivery system alone [14]. One drawback to this system is that upon release from the hydrogel, the insulin is exposed to the proteolytic enzymes in the small intestine, which serve to degrade the insulin [15,16]. The present work is the first complete and detailed presentation of these systems in the public domain.

To avoid proteolytic degradation and chemical instability, as well as improve solubility and mucoadhesion, we have added additional protection to the insulin molecule and elected to investigate the conjugation of poly(ethylene glycol) (PEG) to the insulin molecule. The conjugation of PEG to the insulin will help improve “stealth” and aid in the protection of the insulin from degradation, in addition to becoming an active participant in the controlled release mechanism, becoming part of the P (MAA-g-EG) mesh in low pH and experiencing swelling under neutral pH. PEG is also known to facilitate adhesion to the intestinal mucosa [17]. As protein function is determined by both composition and conformation, altering the physiochemical properties is not trivial. To be successful, a change in peptide or protein formulation must simultaneously address a large number of issues, which include bioavailability, intrinsic permeability, degradation, clearance, confirmation, and chemical stability [7,8]. Thus a new conjugation mechanism that utilized PEG nitrophenyl carbonate (PEG-NPC) was examined in order to improve homogeneity and yield. The transport of the resulting PEGylated insulin was evaluated in a cell model of the intestinal epithelial layer. Furthermore, the in vivo absorption of insulin was measured in Sprague-Dawley rats to ensure that the PEGylated insulin conjugates are biologically active, as well as to compare the bioavailability to control insulin.

Collectively, these results show successful development of the anionic protection material capable of loading and releasing PEGylated insulin at physiologically relevant rates. This is the first published work that combines the novel approaches using tunable anionic hydrogels for the release of PEGylated insulin. Indeed, PEGylation of the insulin molecule is critical in improving insulin stability upon intestinal delivery.

2. Materials and methods

2.1. Materials

Bovine insulin, bovine serum albumin (BSA), phosphate buffered saline, non-essential amino acids, 2-nitro-5-thiocyanobenzoate (NTCB), dithiothreitol (DTT), α-cyano 4 hydroxycinnamic acid, formic acid, methacrylic acid, tetraethylene glycol dimethacrylate (TEGDMA), PEG monomethylether monomethacrylate (PEGMMA) were purchased from Sigma-Aldrich, St. Louis MO. Acetonitrile, hydrochloric acid, and trifluoroacetic acid, were purchased from Fisher Chemical. PEG-NPC was purchased from Laysan Bio Arab, AL. Caco-2 cells were purchased from American Tissue Culture (Rockwell, MD), while HT29-MTX cells were obtained through a generous donation from Dr. Thecla Lesuffleur (Robert Debré Hospital, Paris, France). Dulbecco's Modified Eagle Medium (DMEM) and penicillin-streptomycin was purchased from Thermo Fisher Scientific. Fetal bovine serum was purchased from Corning. Bovine insulin ELISA was purchased from Mercodia Uppsala, Sweden.

2.2. Synthesis and purification of PEGylated insulin

Bovine insulin (Sigma-Aldrich, St. Louis MO) was added to pH 8 PBS buffer with ethylenediamine tetraacetic acid (EDTA) at a ratio of 5 mg/mL and was dissolved in solution by incubating for 10 min at 37 °C. Once dissolved, a 2:1 M ratio of PEG-NPC to insulin with a molecular weight of 1000 da, 2000 da or 5000 da was added slowly to the mixture (Laysan Bio Arab, AL). The mixture was then reacted for 18 h at 37 °C under gentle agitation. Upon completion of the reaction, unreacted material was removed with a Vivaspin 6 centrifuge concentrator with a molecular weight cutoff of 5000 Da (Sartoriums Stedim Biotech Abagne, France). The reaction mixture was placed in the concentrator and centrifuged at 1691 m/s² for a period of two hours at 4 °C. Once purified, the resulting mixture, conjugated with conjugated with 1000 da, 2000 da, and 5000 da PEG (1KPI, 2KPI and 5KPI respectively), was dried with a centrifuge vacuum drier and stored at −15 °C until use.

2.3. Conjugate evaluation

The purity of the conjugate was determined using a Waters 2695 Separations HPLC with a Symmetry 300^™ C4 column. Briefly, a sample of 20 μL was injected into the column, which ran a mixture of 70% water and 30% acetonitrile with 0.1% trifluoroacetic acid at a flow rate of 1 mL/min. The sample was detected at 220 nm.

Purity and molecular weight was also analyzed via MALDI-TOF (PerSeptive Biosystems). Matrix solution consisted of α-cyano 4 hydroxycinnamic acid in aqueous solution with 2% formic acid.

Enzymatic degradation was used to obtain the relative location of the conjugation. 2000 da PEGylated insulin was dissolved in pH 8 tris base (1 mg/mL) and the selectively broken into identifiable peptide fragments. In order to reduce the disulfide bonds, purified PEGylated insulin was subsequently exposed to dithiothreitol (DTT) (Sigma-Aldrich, St. Louis MO) at a 2:1 M ratio of DTT: number of sulfides on PEGylated insulin in PBS pH 7.4 for two hours at 37 °C. 2-nitro-5-thiocyanobenzoate (NTCB) (Sigma-Aldrich, St. Louis MO) was then used to break cysteinyl residues at the N-terminal side of the amine. A 10-fold molar excess of NTCB was added over the approximate number of cysteines and incubated at 37 °C for 30 min in order to modify the residue. The pH was then raised to 9 with sodium hydroxide and the solution was reacted for 16 h in order to cleave the modified peptide. The peptides fragments were then separated using preparative HPLC and analyzed for molecular weight using MALDI-TOF. The weights of the peptide fragments were calculated with and without possible conjugation. The presence or absence of PEG was determined by examining the resulting spectra for polydispersity and molecular weight.

2.4. Cell transport

Caco-2 cells were purchased from American Tissue Culture (Rockwell, MD), while HT29-MTX cells were obtained through a generous donation from Thecla Lesuffleur (Robert Debré Hospital, Paris, France). Cellular transport and transepithelial resistance (TEER) was investigated in the presence of microparticles. A 50:50 coculture of Caco-2:HT29-MTX cells was cultured on a 12-well Transwell^® in Dulbecco's Modified Eagle Medium (DMEM) supplemented with 10% heat-inactivated fetal bovine serum, 1% non-essential amino acids, and 1% penicillin-streptomycin for 21 days. The medium was removed from the apical and basolateral chambers and replaced with HBSS containing 1.26 mM Ca²⁺. Cells were allowed to equilibrate for 1 h. The HBSS was replaced with a solution containing 0.2 mg/mL bovine insulin, 0.5 mg/mL BSA and 1 mg/mL P(MAA-g-EG) microparticles crushed to >52 μm in HBSS. BSA is used to limit non-specific adhesion. Samples were collected in the apical chamber at the initiation and termination of the experiment; 0 and 3 h respectively. Basolateral samples and TEER measurements were taken at 0, 0.5, 1, 2, and 3 h. Volumes were replaced with prewarmed solutions containing 0.5 mg/mL BSA in HBSS. Finally, insulin concentration was measured using a bovine insulin ELISA (Mercodia Uppsala, Sweden).

$P_{\text{app}}$ was calculated by the equation that follows:

$$P_{app}=\frac{J}{A{C}_o}$$ (1)

Here, $J$ represents the flux and is calculated by determining the slope of the linear plot of drug transport versus time, $A$ is the surface of the cell monolayer and $C_o$ is the initial concentration in the apical chamber.

2.5. Synthesis of insulin loaded hydrogels

Poly(methacrylic acid-g-ethylene glycol) (P(MAA-g-EG)) grafted polymers were synthesized using free radical polymerization as described in previous work [12]. Briefly, a 1:1 M feed ratio of methacrylic acid/PEG monomethylether monomethacrylate (PEGMMA) was added to an equal mass of 50 wt% solution of ethanol in water. Tetraethylene glycol dimethacrylate (TEGDMA) was used as the crosslinker and was added at a ratio of 1 mol% of total monomer. 1-Hydroxycyclohexyl phenyl ketone was used as the initiator. Later, a 3:1 M feed ratios were also studied with PEGMAA molecular weights of 1 kDa and 2 kDa. Finally, methacrylic acid particles without grafts were also synthesized. All hydrogels were crushed and sieved to <43 μm based on results showing that this particle size optimized insulin release [14].

2.6. Complexation hydrogel loading and release

Drug loading was accomplished as described previously by equilibrium partitioning of insulin into the P(MAA-g-EG) microparticles [12]. 10 mg of insulin was dissolved in 2.5 mL 0.1 N HCL. The solution was brought to pH 7.4 and buffered with 2.5 mL 0.1 N NaOH and 15 mL of PBS. In the case of the PI conjugates, the molar equivalent of 10 mg was added directly to 20 mL of PBS. 140 mg of P(MAA-g-EG) microparticles were added and allowed to incubate with the insulin for 2 h under continuous stirring. After two hours, 10 mL of 0.1 N HCL was added to collapse the particles, which were allowed to incubate with the acid for 30 mins. Finally, the particles were washed with DI water and filtered with a 0.1 μm filter.

Insulin release studies were performed by placing 10 mg of loaded microparticles into 20 mL of PBS solution in a siliconized flask to limit nonspecific adhesion. Each sample was maintained at 37 °C and stirred at 100 rpm throughout using dissolution apparatus (Distek 2100B, North Brunswick NJ). The release was measured over a period of 2 h by taking 200 μL samples and analyzing samples with RP-HPLC. Prewarmed buffer was used to replace the sample volume.

2.7. In vivo absorption studies

This research was performed at Hoshi University and complied with the regulations of the Committee on Ethics in the Care and Use of Laboratory Animals. Male Sprague-Dawley rats weighing 180–220 g were purchased from the Tokyo Laboratory Animal Sciences Company. After purchase, rats were housed in climate-controlled rooms with a temperature of 23.1 ± 1 °C and a 55 ± 5% relative humidity level and were allowed free access to food and water. Twenty-four hours prior to experimentation, rats were fasted, yet were allowed free access to water. Just prior to experimentation, rats were weighed and anesthetized with 50 mg/kg pentobarbital (Dainippon Pharmaceutical Company, Osaka, Japan) with additional injections (12.5 mg/kg) administered every hour to maintain anesthesia. The rats were restrained in the supine position on a temperature-controlled plate.

A small incision was made along the abdominal midline to expose the inferior portion of the illeum. A section proximal to ileocecal junction segments (10 cm length) was cannulated with polypropylene. The cannulated segment was washed with 20 mL of 37 °C prewarmed saline delivered at 1 mL/min using a parastaltic pump. The cannulae were removed and the intestinal segments were tightly closed. The rats were allowed to recover for at least 30 min to allow for the injury-induced elevated blood glucose levels to stabilize. A 50 IU/kg dose of 1KPI, 2KPI, 5KPI or insulin suspended in PBS was administered to a 6 cm section of the 10 cm pretreated intestinal segment. 0.2 mL aliquots of blood were drawn directly from the jugular vein at 0, 5, 10, 15, 30, 60, 120, 240 min using 1 mL pretreated heparanized syringes (Tuberculin^®, Becton-Dickenson, Franklin Lakes, NJ). These syringes were prepared 1 day prior to the experiment by aspirating heparin through each syringe, completely expelling the heparin and then allowing the syringe to dry. The blood glucose was immediately measured with a Novo Nordisk Glucose Meter (Johnson & Johnson, Langhorn, PA) and then centrifuged at 13,000 RPM for 1 min to separate whole blood from plasma. The plasma insulin concentration was then analyzed with ELISA immediately following the experiment (Invitron, Monmouth, UK). The bioavailability was calculated relative to the subcutaneous (s.c.) injection as described below.

The subcutaneous bioavailability was determined by dissolving PI or insulin in PBS at 1 IU/kg bodyweight. To maintain the same physical conditions, the ileal loop surgery was performed on the rats that received subcutaneous insulin.

In order to determine the pharmacodynamic effect, the area above the normalized glucose versus time curve (AAC) was calculated by determining the area below the axis between the resultant curve and 1. This was calculated using the trapezoidal rule. In this case the s.c. AAC was determined from the subcutaneous administration of 1.0 IU/kg insulin. Pharmacological availability was calculated according to the equation that follows:

$$PA=\frac{[AAC{]}_{\mathit{oral}}}{{\mathit{dose}}_{\mathit{oral}}}\mathrm{*}\frac{[AAC{]}_{S.C.}}{{\mathit{dose}}_{S.C.}}\mathrm{*}100$$ (9.1)

While pharmacological availability reflects the action of insulin on the glucose level of the bloodstream, it often does not directly correlate with the amount of insulin in the blood. Thus, it is also important to measure the amount of insulin in the bloodstream over time, or bioavailability [18,19]. To determine the change in insulin conjugation, the initial concentration was subtracted from the measured insulin concentrations.

In order to determine the bioavailability (F) of PEGylated insulin, the area under the insulin concentration curve versus time (AUC) from 0 to 240 min was calculated using the trapezoid rule for both subcutaneous and intestinal delivery and normalized by dosage as shown in the following equation:

$$F=\frac{[AUC{]}_{\mathit{oral}}}{{\mathit{dose}}_{\mathit{oral}}}\mathrm{*}\frac{{\mathit{dose}}_{S.C.}}{[AUC{]}_{S.C.}}\mathrm{*}100$$ (9.2)

When studying bioavailability, PEGylated insulin in solution delivered to the small intestine was compared to subcutaneously delivered insulin. The plasma peak level $\left({\mathrm{C}}_{\text{max}}\right)$ and the time taken to reach the plasma peak level $\left({\mathrm{T}}_{\text{max}}\right)$ were determined from the plasma insulin concentration versus time curve.

2.8. Noninvasive imaging of insulin being released from P(MAA-g-EG) microparticles in the digestive tract

Insulin was conjugated to NHS-Cy7^™ dye (GE Healthcare, Hyakunincho Shinjuku-ku Japan), a monofunctional fluorescent dye that absorbs at 747 nm and emits at 776 nm. The conjugation was performed as follows. Insulin was dissolved in pH 7.4 PBS with ethylenediamenetetraacetic acid (EDTA). A desalting column was used to change the buffer to 50 mM NaHCO₃ buffer (pH 7.9) and to remove salt and EDTA. The Lowry Method for protein analysis was used to determine protein concentration and the insulin-containing fractions were combined. A second solution was prepared by dissolving NHS-Cy7^™ in DMSO. This solution was combined with the insulin solution and reacted in the dark at room temperature for 14 h. A 15 μL aliquot of triethyleneamine was added to the solution, which was then reacted for an additional 30 min. The solution was then dialyze to remove any unreacted dye and finally lyophilized. Drug loading of the tagged insulin was then completed as described above. Equimolar amounts of Cy7^™-insulin loaded in 1:1 P(MAA-g-EG) or Cy7^™-insulin were loaded into gel capsules.

Prior to study, rats were fasted for 48 h to ensure that the intestine was clear of obstructions that may inhibit imaging. Gel capsules were administered orally to restrained, shaved, male Sprague-Dawley rats. For this study, animals were not anesthetized. The abdomen of the rat was excited at 747 nm with the LB940 Mirthras Multimode Reader (Berthold Technologies, Bad Wildbad, Germany). Cy7^™ insulin emitted around 776 nm, which was imaged with the NightOwl LB981 camera (Berthold Technologies, Bad Wildbad, Germany). At this wavelength, the camera was able to penetrate the body tissue to image the fluorescent dye directly. The placement of the camera was computer controlled by IndiGo^™software and motor driven, with a reproducibility of ±100 μm.

Images were taken at 30, 60 and 120 min. Rats were restrained while the image was being taken and allowed back in the cage between imaging sessions. During the study, rats were placed in separate cages to avoid being injured by other rats while in a stressed and weakened state.

3. Results and discussion

3.1. Synthesis and characterization of PEGylated insulin

In this study, insulin was conjugated to PEG in order to improve insulin proteolytic degradation, chemical instability, and improve solubility. PEG, notably termed a “stealth” polymer, is flexible and hydrophilic; in an aqueous solution, 2–3 water molecules per ethylene oxide unit will coordinate around the polymer effectively creating a polarized hydration sphere that can protect the PEG and associated molecules [20,21]. Furthermore, this hydration sphere effectively acts to prevent the adhesion of other proteins, which leads to a reduction in fibrillation, antigenicity, proteolytic degradation, and clearance rate from the body, which may decrease the dosage frequency. As a plus, PEG is biocompatible and has been FDA-approved for use as a direct food additive (CFR172.820), and as an excipient, intermediate, or active in pharmaceutical formulations (CFR181 and 310). In addition, the PEG molecule on the insulin is an active participant in the P(MAA-g-EG) controlled release mechanism; becoming part of the tight polymer mesh in low pH conditions, and experiencing swelling under intestinal pH conditions. A schematic depicting the PEG conjugation strategy using nitrophenyl carbonate is shown below in Fig. 2.

Fig. 2.

Insulin conjugated with PEG-NPC in water. Each ethylene oxide group can bind with up to 3 water molecules. This creates a “hydration sphere” assists in protecting the protein from proteolytic degradation. Created with BioRender.com

The insulin molecule has 3 active sites that can be used for functionalization: A1-Glycine (A1Gly), B1-Phenylalanine (B1Phe), and B29-Lysine (LysB29). Previously published formulations utilized anhydride chemistry, which resulted in non-specific conjugation. The resulting conjugates were characterized by heterogeneous results and low yield [22]. Thus, it was necessary to investigate a new conjugation chemistry that could potentially resolve these problems. In this study, a PEG-NPC conjugation scheme was utilized. This reaction was conducted to determine if a more homogeneous conjugate or a higher yield would result. Polymers with molecular weights <5000 were chosen for study as it has been previously demonstrated that longer chain lengths limit bioactivity [23].

The mildly endothermic yielded optimal results at 18 h. The average conjugation rate prior to purification reached 66.86% PEGylation, as verified by RP-HPLC as shown in Fig. 3.

Fig. 3.

RP-HPLC analysis of 2KPI before and after centrifuge filtration. The blue trace represents the spectrum prior to purification with Vivaspin (66.86% PEGylated). The black trace represents the spectrum post purification (84.07% PEGylated).

Purification was carried out using a Vivispin 5000 molecular weight cutoff (MWCO) filter. As verified by RP-HPLC (Fig. 3), much of the unreacted insulin was removed from the product. This was demonstrated with HPLC. The peak on the left centered around 4.75 min represents insulin. The wide peak centered around 7 min represents the PEGylated insulin. This conjugate eluted over a longer period of time and is more irregular due to the polydispersity of the PEG. The trace indicates that centrifuge purification improves the percentage of PEGylation to 84.07%. The molecular weight and polydispersity index were investigated using MALDI-TOF. To perform this analysis, the laser beam serves as the desorption and ionization source. The matrix plays a crucial role in this technique by absorbing the laser light and causing the target to vaporize. The results are shown in Fig. 4.

Fig. 4.

Maldi-TOF spectrum of 2KPI. The wide molecular weight is representative of the polydispersity of the poly(ethylene glycol) with the average molecular weight occurring around 8000 Da.

MALDI-TOF results confirm that not only did PEGylation take place, but it also occurred at a ratio of 1 molecule insulin (5733.49 da) to 1 molecule PEG (2000 da). This is important, because a 2:1 M excess of PEG was used and it was believed that NPC-PEG could react at as many as 3 places on the insulin molecule. Given the disparity between expected and experimental results, this shows that selectivity was achieved as the conjugation is localized to certain amino acids rather than all 3 proposed regions. This is due to the fact that NPC-PEG has slow reaction kinetics making it react with certain amino acid groups of the insulin [24].

From the previously mentioned results, the site of conjugation was investigated. The site of conjugation was verified by predicting the molecular weights of each fragment, separating each fragment, and analyzing the molecular weight with MALDI-TOF. All but one fragment displayed minimal polydispersity, while the polydispersity of the fragment containing PEG was visible. The molecular weight range of the fragment was measured to be 2906.79–3438.78 Da. This value corresponds to LysB29, which has an estimated Mw of 3307, and an estimated Mn of 3208. This is important because this reaction allows for high-yield site specific PEGylation of insulin. Conjugating at the Lys29 position has been shown to not affect the activity of insulin.

3.2. Cellular characterization of PEGylated insulin

Next, we investigated the transport of the conjugated PEG-insulin molecule with the well-established Caco-2:HT29-MTX co-culture [25]. Cultured epithelial cells have been successfully used as tools for the screening of drug permeation and transport during the last two decades. Caco-2 cells are an immortal human adenocarcinoma cell line derived from a 72 year-old male of European descent that has demonstrated physiological relevance. Mature Caco-2 monolayers display many of the characteristics of the human epithelial layer; including the production of a number of relevant enzymes, a brush border, and the formation of tight junctions [25,26]. These cells do not produce mucus and the tight junctions are not as permeable as those found in the small intestine. Human colon carcinoma cell HT29-MTX, a second cell line derived from a 44 year-old female with cancer produces a mucus layer and is highly permeable. The combination of these two lines allows the investigator to “tune” the permeability, allowing the model to be more physiologically relevant and is an established model in the literature as a model to test the intestinal permeability of therapeutics [27,28].

Both hydrophilic drugs and peptides are known to distribute poorly in the cellular epithelium and therefore it is assumed that these drugs travel through the water-filled pores of the paracellular pathway [29]. Because PEG is a hydrophilic molecule, diminished permeability is expected. The apparent cellular permeability of insulin and conjugates is shown in Table 1. As predicted, PEGylation decreased the permeability of the insulin. In addition, permeability appeared to decrease with increasing chain length. These cells do not produce measurable quantities of the enzymes primarily responsible for the degradation of insulin: elastase, trypsin and α-chymotrypsin, and thus may underestimate the value of insulin PEGylation [30].

Table 1.

Apparent cellular permeability of insulin and insulin conjugates in the presence of microparticles.

Conjugate	Papp × 10−9 (cm/s)
1 K PEGylated Insulin	3.28 ± 0.88*
2 K PEGylated Insulin	2.90 ± 0.86*
5 K PEGylated Insulin	2.44 ± 0.06*
Insulin	9.17 ± 2.20

Data: mean ± S.D. (n = 3).

^*P < .5 when compared to insulin without microparticles.

TEER is a function of the integrity of the tight junctions in the cellular epithelium. The decrease of the TEER value indicates disruption of the tight junctions and subsequently that transport is occurring between the tight junctions. Previous study has shown a decrease in TEER with insulin in the presence of microparticles [31]. In this study, the overall decrease in TEER is preserved with both insulin and PEGylated insulin in the presence of microparticles. There is no notable difference between PEGylated insulin and insulin, indicating the transport through tight junctions was not affected by PEGylation (Fig. 5).

Fig. 5.

Changes in TEER with in the presence of P(MAA-g-EG) microparticles and insulin 1KPI, 2KPI or 5KPI. Each data represents the mean ± S.D. n = 3.

3.3. Development of P (MAA-g-EG) carriers for PEGylated insulin

Previous systems for controlled release into the intestines utilize enteric coatings. Common enteric coatings such as acrylic-based coatings are formulated breakdown at an expected pH. For example, Eudragit ^® S100 (Rohm Pharma Polymers) coatings are designed to release at a range of 6.0–7.5 for localized delivery to the ileum of the small intestine. These formulations generalize patient populations in assuming that they share typical intestinal pH ranges [32]. However, intestinal pH can vary between patients due to a variety of factors including lifestyle, microbiome environment, and disease state [33–35]. Therefore, development of tunable pH-responsive materials is needed to account for these differences.

Here, we developed anionic hydrogels made of P(MAA-g-EG) to protect the PEGylated insulin through the acidic pH in the stomach and release in the intestines. In the stomach, the pH will be less than the pK_a of the hydrogel allowing for hydrogen bonds to form between MAA and EG [36]. When pH in the intestines is greater than the pK_a, the carboxylic acid groups will become anionic causing interchain repulsion. The increased spacing caused by this repulsion along with the ionized carboxylic acid groups will allow for water to enter the polymer network and continue the swelling. [32] This swelling induced by pH in the physiological environment of the GI tract changes allows for the targeted release of insulin to the intestine. We note here that unlike enteric coatings, our approach utilizes highly tunable and controllable hydrogels which respond to physiological changes in the intestinal pH via swelling or contraction. The dynamic capabilities of the hydrogel we developed can ensure targeted delivery of insulin occurs in the intestines. The mechanism of swelling of the anionic polymer and release of insulin is shown in Fig. 1.

P(MAA-g-EG) carriers were developed for the PEG insulin conjugates. As it is believed that the interactions between the hydrogel and PEGylated insulin will involve hydrogen bonding, the effects of the grafted PEG chain density and length were evaluated (Table 2).

Table 2.

Amount of insulin or 2 k PEGylated insulin loaded before acidifying sample (pre collapse) and after acidifying sample (collapsed). All polymers aside from the PMAA sample were composed of P(MAA-g-EG). The graft length of each polymer is represented by g.

	Graft Length	g: 1000 Da	g: 1000 Da	g: 2000 Da
Sample	MAA	1:1 MAA:EG	3:1 MAA:EG	3:1 MAA:EG
Pre-Collapse
Insulin	6.93 ± 1.79**	89.16 ± 5.25	61.43 ± 18.29*	65.89 ± 6.92*
2 k PEGylated Insulin	76.76 ± 13.83ΔΔ	28.04 ± 55.95*	81.75 ± 17.54Δ	54.58 + 42.38ΔΔ
Post-Collapse
Insulin	21.04 ± 2.27**	82.36 ± 2.73	50.59 ± 5.82*	53.53 ± 3.89**
2 k PEGylated Insulin	99.73 ± 1.46**,ΔΔ	61.24 ± 23.39	96.01 ± 5.58**,Δ	100.74 ± 4.55**,ΔΔ

Data: mean ± S.D. (n = 3–6)

^*p < .05

^**p < .01 when compared to loading results for insulin loading in 1:1 MAA:EG.

^Δp < .05

^ΔΔp < .01 when 2KPI loaded samples are compared to 2KPI loaded in 1:1 MAA:EG.

As shown in Table 2, for unconjugated insulin, prior to the addition of acid, and subsequent collapse of the particles, 1:1 P (MAA-g-EG) yielded the best loading results, with decreasing loading as the amount of PEG grafting decreased; with almost no loading in methacrylic acid (MAA) particles. After the particles were collapsed, in all of the samples with PEG grafts, the amount of insulin loaded decreased due to the fact that some of the insulin is forced out of the polymer when the PEG is collapsed.

Loading of PEGylated insulin prior to collapse was believed to occur because of diffusion. With respect to PEGylated insulin, once acid was added, hydrogen bonding led to loading percentages >96% in all cases except in the 1:1 MAA:EG polymer. It is believed that the decreased loading for the 1:1 MAA:EG polymer was due to steric hindrance of the PEG chains.

The release characteristics of PEGylated insulin were evaluated and compared to insulin released from 1:1MAA:EG hydrogels (Fig. 6). The effects of the release of PEGylated insulin with varying chain densities was evaluated. With respect to the PEGylated insulin loaded into 1:1 MAA:EG, there was limited release. As the density of the chains decreased, the release of the PEGylated insulin increased. This result agrees with earlier work which showed that increased PEG graft density decreases the diffusion of a drug [37]. In order to help facilitate release of PEGylated insulin, future work could focus on addition of a digestive enzyme degradable crosslinker into the MAA:EG carrier, that would facilitate degradation of the carrier in the intestines.

Fig. 6.

Release of PEGylated insulin from each polymer depicted in Table 2, and unconjugated insulin release from 1:1 MAA:EG. Each data represents the mean ± S.D. n = 3 Key: Red: PEGylated insulin 1:1 MAA:EG g-1000, Green: PEGylated insulin 3:1 MAA:EG g-1000, Dark blue: PEGylated insulin 3:1 MAA:EG g-2000, Purple: PEGylated insulin MAA, and Light blue: insulin 1:1 MAA:EG

3.4. Influence of PEGylation in subcutaneous administration

Insulin and PEGylated insulin were injected subcutaneously into Sprague-Dawley rats, and the resulting glucose change was measured as a function of time. This was done to ensure that the PEGylated insulin conjugates are biologically active. This is necessary to establish the balance between the protective impact of the PEG and the potential loss in binding sites and thus a loss of reactivity. Both insulin and 1KPI were effective at reducing the glucose levels below baseline throughout the testing period. The 2KPI insulin had a lower PA, but also was able to maintain the glucose level below baseline. Finally, the 5KPI had the least effect of the PEGylated insulin samples on blood glucose levels. Fig. 7 depicts the changes in blood glucose from subcutaneously delivered insulin and PEGylated insulin while Table 3 depict the AAC, a value of the hypoglycemic response, and PA, a value that compares the AAC of the insulin conjugates to that of insulin.

Fig. 7.

Changes in the blood glucose level following the subcutaneous delivery of 1 IU/kg insulin or PEGylated insulin in fasted Sprague-Dawley rats. Each value represents the average ± SD, n = 3.

Table 3.

The area above the curve and the pharmacological availability of subcutaneously delivered insulin conjugates. The mean ± SD is recorded, n = 3.

	AAC (% glucose reduction)	PA (%)
Insulin	45.05 ± 16.56	n/a
1KPI	46.61 ± 20.60	103.47 ± 45.73
2KPI	18.59 ± 19.63	41.26 ± 43.26
5KPI	8.59 ± 12.16	19.08 ± 26.98

The 1000 PEG chain may limit the resulting decreased clearance and degradation, while allowing the insulin to retain its bioactivity. In addition, literature has shown that PEGylation with this molecular weight further serve to limit aggregation, fibrillation and degradation due to the non-specific steric hindrance of the intermolecular interactions that occur during fibrillation [38,39]. As the molecular weight increases, the PA decreases suggesting that longer PEG chains may impede the binding of insulin with its receptor. Collectively, these results indicate that the PEGylation did not hinder the biological activity of the insulin.

3.5. Influences of PEGylation on insulin in intestinal closed-loop delivery

The amount of insulin in solution and the reduction in glucose was measured for PEGylated insulin as well as unconjugated insulin (Fig. 8 and Fig. 9). The bioavailability and pharmacological availabilities of insulin and PEGylated insulin in solution were determined when delivered directly to the intestine (Tables 4 and Table 5). To account for the fact that healthy rats were used and the glucose response to injury (such as that associated with the intestinal operation or the needle stick required for the blood draw), each measurement was normalized by the initial insulin value.

Fig. 8.

Changes in blood glucose following the administration of 50 IU/kg PEGylated insulin in the small intestine of fasted Sprague-Dawley rats.

Fig. 9.

Plasma insulin concentration after administering 50 IU/kg insulin or PEGylated insulin solution to the small intestine, n = 3.

Table 4.

Intestinal absorption results based on insulin concentration. This chart depicts the maximum concentration (C_max), the time at which the maximum concentration was measured (T_max), the area under the insulin curve (AUC) and the bioavailability (BA) of 50 IU/kg orally delivered insulin and PEGylated insulin solution. The mean ± SD is recorded, n = 3–4.

	Cmax (μg/mL)	Tmax	AUC (ug*h/mL)	BA (%)
Insulin			12.61 ± 9.85	1.064 ± 0.83
1KPI	17.16 ± 11.70	26.67 ± 29.30	15.39 ± 22.73	1.30 ± 1.92
2KPI	72.57 ± 93.04	15.00 ± 13.23	36.16 ± 39.59	3.052 ± 3.34
5KPI	14.19 ± 13.06	8.33 ± 5.77	17.52 ± 23.45	1.48 ± 1.98

Table 5.

Area above the curve and pharmacological availability of insulin and PEGylated insulin delivered directly to the small intestine, n = 3–4.

	AAC (% glucose reduction)	PA (%)
Insulin	12.61 ± 9.85	0.56 ± 0.44
1KPI	12.81 ± 14.59	0.56 ± 0.65
2KPI	82.30 ± 32.50	3.65 ± 1.44
5KPI	3.34 ± 4.37	0.37 ± 0.48

It was discovered that insulin in solution had the lowest AUC and bioavailability. In addition, the elevation in insulin concentration returned to 0 after 2 h. 1KPI and 5 KPI had a bioavailability that was larger than unconjugated insulin. 2KPI had the largest bioavailability, with a value of 3.05. 2KPI also had the highest maximum concentration; 72.58 μg/mL. This is due to the fact that PEGylation may prevent the binding of the antibody, thus skewing the ELISA results [39,40]. This apparent reduction in concentration of PEGylated insulin when measured with ELISA has occurred in studies with other protein types. The generally accepted cause behind this reduction is that the PEG molecule restricts the accessibility of the monoclonal antibody to the binding domain on the insulin [41]

Fig. 8 depicts the glycemic response following the administration of 50 IU/kg insulin or PI to the small intestine. Unconjugated insulin at this dosage yielded a limited hypoglycemic response, indicating that only a limited amount of insulin is naturally absorbed by the ileal segment of the small intestine. When PEGylated insulin was delivered directly to the ileum, all PEGylated insulin conjugates improved the blood glucose level. 2KPI had the greatest hypoglycemic effect with an AAC of 82.3% and a PA of 3.65%. This conjugate effectively increased the PA by 2.8 times. In addition, the 2KPI continued to produce a hypoglycemic effect at 4 h, almost 3 h after unmodified insulin has exceeded baseline.

1 K PEGylated insulin was not in the bloodstream at a pharmacological availability as high as 2 K PEGylated insulin, but it had the highest levels when delivered subcutaneously. This led to the result that although there was not a tremendous amount of 1 K PEGylated insulin insulin in the blood, the high pharmacological availability led to a reduction in blood glucose that always remained above 0. The 2 K PEGylated insulin had by far the highest peak concentration (C_max), with an average of 72.58 μg/mL when delivered orally. Additionally, 2 K PEGylated insulin had blood insulin concentrations that remained above 0 for the duration of the study and were higher than that of insulin and the other conjugates. This result translates to the fact that the 2 K PEGylated insulin conjugate was able to depress the glucose levels over the course of the study and increase the pharmacological availability over insulin by 2.8 times. These results are believed to have occurred because the 2 K PEGylated insulin optimizes the effect of PEG on the insulin; allowing it to be protected from proteolytic degredation, yet the chain is sufficiently short to avoid inhibiting the action of insulin on glucose levels.

The general improvement in the efficacy of PEGylated insulin presumably occurred for two reasons. PEG-insulin is extremely soluble with respect to insulin and it is known that solubility improves diffusion and subsequently transport across the epithelium. PEGylation limits degradation, aggregation and fibrillation, which means it protects the protein against enzymes in the GI tract, and it limits the size of the insulin (and subsequently the hydrodynamic radius) by means of preventing aggregation and fibrillation [39,40]. The reason the 2 K PEGylated insulin outperformed the other molecular weights is likely because it was large enough to provide sufficient protection from proteolytic enzymes without blocking active sites.

These results may further be explained through results obtained in literature. A number of in situ experiments using insulin have been performed which demonstrate that insulin can be uptaken by the small intestine [42,43]. The use of PEGylated protein delivery was previously validated in a closed-loop intestinal study, which was performed with 2 K PEGylated calcitonin. The biological marker for this protein, calcium, was greatly reduced when PEGylated so much so that the AAC of PEGylated calcitonin was 2.2 times that of calcitonin itself [44].

3.6. Noninvasive imaging of insulin being released from P(MAA-g-EG) microparticles following Oral delivery

Fig. 10 shows the imaging results of insulin and insulin loaded particles (ILP) as it traveled through the digestive system of a Sprague-Dawley rat. The concentration of the drug was qualitatively represented by the colored bar on the images. For insulin without a carrier, the insulin appeared to be in the stomach at 30 min, judging by its higher proximity in the abdomen. It then moved into the small intestine at 60 min and begins to diminish in concentration after 120 min. ILPs also appeared in the stomach at 30 min, and then began to move into the intestine at 60 min. After 120 min, the insulin concentration remained high due to the fact that the insulin was being protected by the microparticles as it was released into the small intestine. This study provides direct evidence that P(MAA-g-EG) microparticles increase the residence time of insulin by protecting insulin from proteolytic degradation and mucoadhesion, which keep the insulin at the absorption site for a longer period of time [42,43,45].

Fig. 10.

In vivo fluorescent imaging of insulin loaded in a P(MAA-g-EG) polymer compared to insulin without a carrier. Key: A was taken at 30 min, B at 60 min and C at 120 min. Insulin is shown in row 1 while insulin-loaded particles are shown in row 2.

4. Conclusions

The results of this study demonstrate that PEGylated insulin is a viable candidate for oral drug delivery. The optimized P(MAA-g-EG) carrier yielded excellent loading and release characteristics, with the best results having been reached with a lower PEG graft density. PEGylation of insulin is a critical next step toward creating an oral delivery platform for insulin as it will improve its stability, protect from proteolytic degradation, and promote mucoadhesion to the intestinal epithelium, and PEG was successfully grafted to the B29-lysine (LysB29) position of insulin using a unique PEG-NPC conjugation scheme. Here, it also is advantageous to use PEGylation combined with the P(MAA-g-EG) carrier, as the PEGylated molecules were found to actively participate in the carrier loading and release mechanism. The PEGylated drug conjugate can bind to the MAA group on the hydrogel through hydrogen bonding while in the collapsed state and subsequently repulse the drug above the polymer's isoelectric point. Finally, in vivo studies demonstrate that PEGylated insulin maintains its biological activity and can improve absorption following delivery in the intestines. 2 K PEGylated insulin was shown to be the best choice for the oral delivery of PEGylated insulin, as it leads to the greatest increase in plasma insulin and the greatest glucose depression when delivered orally. We also showed in preliminary oral delivery studies in rats, that the PEGylated hydrogels increased the residence time of insulin the digestive tract by protecting the insulin from degradation and promoting mucoadhesion.

Altogether, the results of this study lead to an improved system for oral insulin delivery that combine the unique approaches of tunable pH-responsive hydrogel materials with PEGylation of the insulin molecule.

Data availability

Data will be made available on request.