Introduction

Hemophilia B, a hereditary bleeding disorder caused by the deficiency of clotting factor IX (FIX), affects 80,000 males worldwide.8 Protein replacement therapy by intravenous injection or continuous infusion can prevent spontaneous bleeding episodes and restore hemostasis, significantly improving the patient’s lifespan and quality of life. Prophylactic treatments, requiring multiple injections per week, are recommended for 75% of hemophilia B patients, classified as severe (<1% FIX clotting activity) or moderate (1–5% FIX clotting activity).8,21 However, current treatments are limited by noncompliance, increased risk of needle injury, costly medical care, and potential complications (e.g., severe allergic reactions).

A major challenge for prophylactic regimens, especially in children, is the need for a surgically implanted venous access port, which have risks of infection and thrombosis.14 As a result, such treatments are not readily available worldwide, and without treatment the median life expectancy of a hemophilia B patient drops to 11 years.3 Therefore, a non-invasive alternative is highly desirable to overcome such limitations and promote the global availability of hemophilia treatment.

Oral delivery of coagulation proteins can improve the patients’ quality of life by ensuring safety, convenience, and ease of administration. In this work, we focus on developing an oral delivery system for a therapeutically effective prophylactic treatment for hemophilia. While there is currently no commercially available oral delivery method of FIX for protein replacement therapy, there are promising results from a number of investigators. For example, previous work by Verma et al. investigated the oral administration of FIX fusion proteins in chloroplasts of transgenic tobacco plants for immune tolerance induction (ITI) therapy. Treatment by this method demonstrated suppression of an immune response and prevention of anaphylactic reactions to the systemically delivered proteins in humanized hemophilia B mice.20 FIX fusion proteins in the transgenic plant cells were non-functional, therefore they only serve as antigens to prevent inhibitor responses and cannot function as a protein replacement therapy.12,20 This approach only offers an alternative to ITI therapy, but does not treat or prevent bleeding episodes and other symptoms associated with hemophilia.

This work focused the development of pH-responsive biomaterials to enable the oral route of human FIX (hFIX). Previous work by Peppas and collaborators has focused on the use of pH-responsive anionic complexation hydrogels, specifically ones consisting of a poly(methacrylic acid) (PMAA) backbone with poly(ethylene glycol) (PEG) tethers (designated as P(MAA-g-EG)) crosslinked with PEG dimethacrylate (PEGDMA), as vehicles for oral protein delivery.13,15,16,22,23 Anionic complexation hydrogels have been engineered to overcome the challenges of an oral administration route by protecting proteins from the harsh environment of the stomach and delivering them to the small intestine. Their pH-responsive swelling, imparted by ionization of acid pendant groups, exploits the physiological changes along the gastrointestinal (GI) tract (Fig. 1). At a low pH, hydrogen bonding between the carboxyl group of MAA and the etheric oxygen of the PEG chains results a collapsed polymer network with a small mesh size, ξ. As the environmental pH increases above the pKa of MAA (~4.8), deprotonation disrupts these interpolymer complexes and causes ionization and electrostatic repulsion of the acid groups of MAA, resulting in hydrogel swelling and increased mesh size.17,19 Such hydrogels have been evaluated for the oral delivery of insulin (5.8 kDa),13,15,22,23 calcitonin (3.4 kDa),2,7 interferon beta (23 kDa),7 and growth hormone (22 kDa).2 However, a polymeric carrier system has not been optimized for high molecular weight coagulation proteins, such as hFIX (57 kDa).

Figure 1
figure 1

Complexation hydrogels are engineered to exploit the pH changes in the GI in order to (a) protect proteins from the harsh gastric conditions and (b) deliver them to the small intestine, where they can be absorbed into the bloodstream.

Full size image

In this article, we designed and evaluated pH-responsive biomaterials as oral delivery vehicles for hematological FIX.18 The mesh size of the hydrogel networks was tailored to optimize the amount of active protein loaded and released. Additionally, loading conditions were modified to further improve the therapeutic delivery potential. Synthesis, pH-responsive swelling, and cytocompatibility of P(MAA-g-EG), as well as an evaluation of the microcarrier performance, are detailed herein.

Materials and Methods

Materials

Methacrylic acid (MAA) and Irgacure 184® (1-hydroxy-cyclohexyl-phenylketone) were purchased from Sigma-Aldrich (St. Louis, MO). Poly(ethylene glycol) (MW = 1000) mono methyl ether monomethacrylate (PEGMMA1000), poly(ethylene glycol) (MW = 400) dimethacrylate (PEGDMA400), PEGDMA600 (MW = 600), and PEGDMA1000 (MW = 1000) were purchased from Polysciences (Warrington, PA). Plasma-derived human FIX (hFIX) was purchased from Haematologic Technologies Inc. (Essex Junction, VT). Human FIX ELISA kits were purchased from AssayPro LLC. (St. Charles, MO), and the BIOPHEN FIX activity assays were purchased from HYPHEN Biomed SAS (Neuville-sur-Oise, France). FaSSIF, FeSSIF & FaSSGF Powder was purchased from Biorelevant.com (London, England, UK). All reagents were used as received. All other solvents and buffers were purchased from Fisher Scientific (Waltham, MA).

Polymer Synthesis

P(MAA-g-EG) hydrogels were synthesized by bulk UV-polymerization. MAA and PEGMMA1000 were added at a 2:1 molar ratio of hydrogen bonding groups to a 1:1 (w/w) mixture of deionized water and ethanol to result in a 1:1 (w/w) ratio of total monomer to solvent. A PEGDMA crosslinking agent, varying in molecular weight (MW of PEG = 400, 600, or 1000), was added at 1.0 mol% of total monomers, and the Irgacure 184® photoinitiator was added at 1.0 wt% of total monomers. After the addition of all components, the pre-polymer mixture was sonicated for 20 min in a Bransonic® CPX3800 ultrasonic digital bath (Branson Ultrasonics Corporation, Danbury, CT).

In an oxygen-deficient environment of a MBRAUN LABmaster 130 glove box (Garching, Germany), the pre-polymer mixture was needle purged with nitrogen for 5 min to remove residual oxygen (a free radical scavenger). The mixture was then dispensed between two quartz glass plates, which were separated with a 0.7 mm Teflon® spacer, and then polymerized for 30 min under 35 mW cm−2 UV light using a UV flood source. After synthesis, the resulting polymer was removed from the glass plates, and 10 mm diameter disks were punched for swelling studies. The polymer was washed in 1 L of deionized water, changed daily 10 times, to remove any unreacted monomer. After washing, the polymer was air dried for 3 days, followed by drying at 37°C under vacuum for 3 days. Dried films were crushed using a mortar and pestle and sieved into microparticles, sized 30–45 µm.

Scanning Electron Microscopic (SEM) Studies

Scanning electron microscopy was used to examine the surface morphology and size distribution of the microparticles. For SEM sample preparation, vacuum-dried microparticles were dusted onto carbon tape-covered aluminum stubs. Samples were coated with 10 nm of Pt/Pd using a Cressington 208 Benchtop sputter coater (Watford, England, UK). SEM images were obtained using a Zeiss Supra 40VP Scanning Electron Microscope (Oberkochen, Germany).

Dynamic and Equilibrium Swelling Studies in Relevant Media

Swelling studies were conducted to quantify the water uptake into the networks during the initial contact process and determined the associated structural changes which affect the overall loading and release behavior. Dried 10 mm disks were weighed prior to all swelling studies. Dynamic swelling studies were performed using a series of 0.1 M 3,3-dimethylglutaric acid buffers between pH 1.2 and 7.6 at constant ionic strength at 37°C. The disks were stepped through the buffers in order of increasing pH, spending 30 min in each buffer. Between each step, the disks were blotted and weighed.

For equilibrium swelling, dried disks were placed in a pH 2.0 buffer solution at 37°C for at least 24 h, then blotted and weighed. The same disks were then placed in a pH 7.4 buffer solution at 37°C for at least 24 h, then blotted and weighed. Equilibrium swelling studies were conducted using buffers of two different ionic strengths—5 mM and 167 mM. The lower ionic strength buffers were 5 mM NaH2PO4 and the pH was adjusted using 1 N NaOH and 1 N HCl as necessary. For the higher ionic strength buffers, the pH 2 buffer was 167 mM NaCl and the pH 7.4 buffer was 1× PBS. The buffered solutions at pH of 2.0 and 7.4 were selected because these values represent the loading (pH 7.4) and collapse (pH 2.0) conditions.

Turbidimetric Studies

The timescale of microparticle swelling, which is critical to drug release, was measured by kinetic turbidity. Microparticles suspensions of 5 mg mL−1 were prepared with 1× PBS with 2% (v/v) 1 N NaOH at 37°C. Immediately following suspension, 200 µL of solution, including all particle suspensions and buffer only, was added in quadruplicate to a 96-well plate. The absorbance was then measured at 520 nm in 1 min intervals over 60 min at 37°C, with 20 s of shaking prior to each measurement, using a BioTek Cytation 3 multi-mode reader (Winooski, VT).

In Vitro Cytocompatibility Studies

To examine the cytocompatibility of the newly developed systems with biological systems, we conducted basic studies of compatibility with well-known cell lines. Human epithelial colorectal adenocarcinoma (Caco-2) and mucus-secreting goblet-like human colorectal adenocarcinoma (HT29-MTX) cells were obtained from American Type Culture Collection (ATCC, Rockwell, MD). Both cell lines were cultured in modified Dulbecco’s Eagle Medium (DMEM) supplemented with 10% fetal bovine serum (FBS). Cells, either Caco-2 cells (at passages 65–75) or HT29-MTX cells (at passages 15–20), were plated on a 96-well plate at an initial seeding density of 105 cells per well and then cultured to 80% confluence. Separately, microparticles were sterilized under UV exposure for 10 min and then added to DMEM without phenol red with 2% FBS at concentrations of 1.25, 2.5, and 5 mg mL−1 and allowed to swell overnight. After reaching 80% confluence, growth media was removed, and 100 µL of solution—microparticle suspensions (n = 4 for each formulation and concentration), media (positive control, n = 8), or water/lysis solution (negative control for MTS/LDH, n = 8)—was applied to the cells. The cells were then incubated for 6 h.

For the CellTiter 96® Aqueous One Solution Cell Proliferation (MTS) assay (Promega Corporation, Madison, WI), MTS reagent (20 µL well−1) was added, and then incubated for an additional 90 min. The absorbance was read at 490 nm for MTS and 690 nm for background using a BioTek Cytation 3 multi-mode reader (Winooski, VT). For the CytoTox ONE Homogeneous Membrane Integrity (LDH) assay (Promega Corporation), lysis solution (2 µL well−1) was added to the negative control wells after the 6-h incubation. After equilibrating to room temperature, cell solution (50 µL well−1) was removed and added to a black-wall 96-well plate with CytoTox ONE reagent (50 µL well−1). After a 30-s shaking period and a 10-min incubation, the fluorescence was measured at 560 nm excitation and 590 nm emission.

Protein Stability

In order to optimize loading, the stability of the hFIX was evaluated in two different buffers (pH 7.4)—1× PBS and 5 mM NaH2PO4—at three different temperatures—4, 20, and 37°C—over a 7-day period. For the study duration, 0.5 mg mL−1 hFIX solutions in either buffer were stored at the designated controlled temperature under constant end-to-end rotation. hFIX activity was quantified at days 1, 2, 5, and 7, as compared to a freshly prepared stock of hFIX in the same buffer, using hFIX chromogenic activity assays according to the manufacturer’s protocol (BIOPHEN FIX activity assay, HYPHEN Biomed SAS).

Protein Loading Studies

Dried particles were added to a 0.5 mg mL−1 hFIX solution (1× PBS with 2% 1(v/v) N NaOH) to a final concentration of 5 mg mL−1 and pH of 7.4. hFIX loading was carried out by equilibrium partitioning for 24 h with particle/protein suspensions under constant end-to-end rotation at 37°C. After 24 h, particles were collected by Buchner filtration using Whatman 50 filter paper. Particles were then collapsed with 0.1 N HCl and then triple rinsed with 0.1 N HCl, DI water, and 0.1 N HCl. Samples were collected after each step—end of loading, collapse, and rinse—for protein quantification using hFIX ELISAs. Protein-loaded particles were then lyophilized overnight.

To optimize loading, the loading parameters—buffer ionic strength, length, and temperature—were modified, while other parameters remained constant. Loading was increased to 5 days at 4°C using two different buffers (pH 7.4)—1× PBS and 5 mM NaH2PO4.

Protein Distribution Studies

To visually confirm the distribution of protein within the loaded microparticles, loading studies were completed using fluorescein isothiocyanate (FITC)-conjugated hFIX. hFIX was labeled with FITC using Thermo Scientific™ Pierce™ FITC Antibody Labeling Kits according to the manufacturer’s protocol. The molar ratio of FITC to protein was calculated from the absorbance measurements at 280 and 495 nm.

Loading studies were conducted as previously described at 37°C for 24 h. Following the collapse and rinse steps, microparticles were collected and then added to microscope slides with Molecular Probes™ ProLong™ Gold Antifade Mountant, which was allowed to cure for 24 h at room temperature. Confocal images, both fluorescent and bright field channels, were captured using an Olympus All-in-one Confocal Laser Scanning Microscope FV10i (Shinjuku, Tokyo, Japan).

Protein Release Studies

Protein release studies were conducted following a two-stage dissolution procedure using biorelevant media. Biorelevant media were prepared using a FaSSIF, FeSSIF & FaSSGF Powder which contains bile salts and phospholipids. Fasted-state simulated gastric fluid (FaSSGF) was prepared at a 1× concentration with a final pH of 1.6 according to the manufacturer’s protocol (Biorelevant). Fasted-state simulated intestinal fluid (FaSSIF) was prepared at 2× concentration with a final pH of 6.9. For the two-stage dissolution procedure, a 1:1 (v/v) FaSSGF:FaSSIF mixture at a final pH of 6.5 is required for the intestinal condition.

Dissolution testing was performed using an USP apparatus 2 (paddles) setup for small volume (100 mL vessel) with low evaporation mini vessel covers using a Distek 2100B dissolution apparatus (North Brunswick, NJ). A 5 mg sample of hFIX-loaded microparticles was added to a Sigmacoted (i.e., siliconized) vessel containing 30 mL of SGF (1×, pH 1.6) at 37°C. After 30 min, 30 mL of FaSSIF (2×) at 37°C was added, raising the pH to 6.5. The solution was continuously stirred at 75 rpm and maintained at 37°C for both stages. The FaSSIF stage was carried out for up to 6 h. Over the course of the dissolution, 1 mL samples were taken using a 1/8 cannula sample probe (Distek, Inc.) with a 10 µM polyethylene filter tip (Agilent Technologies, Santa Clara, CA). Samples were replaced with an equal volume of appropriate prewarmed media. hFIX released was quantified by hFIX ELISAs and its activity was measured by hFIX chromogenic activity assays.

Endotoxin Depletion

For in vitro use, hFIX was endotoxin depleted using Thermo Scientific™ Pierce™ High-Capacity Endotoxin Removal Spin Columns according to the manufacturer’s protocol. After endotoxin removal, the amount of endotoxin in hFIX samples was measured using a Thermo Scientific™ Pierce™ LAL Chromogenic Endotoxin Quantification Kit. To verify the lack of product inhibition, a spiked hFIX sample was included in the assay.

In Vitro Transport Studies

In vitro transport studies using an intestinal epithelial model were conducted to determine the effect of the presence of the microcarriers on protein transport. Caco-2 and HT29-MTX cells were cultured with DMEM with 10% FBS in separate flasks. For the transport studies, cells (1:1 Caco-2:HT29-MTX mixture) were seeded on the apical side of 12-well Transwell® plates (12 mm diameter, 0.4 µm pores, Corning, Corning, NY) at 105 cells per well. Monolayers were cultured for 21–25 days to allow the cells to fully differentiate and reach confluence. Every other day, media (DMEM supplemented with 10% FBS, 1% l-glutamine, and 1% penicillin/streptomycin) was changed, and the transepithelial electrical resistance (TEER) was monitored with an EVOM2 epithelial voltohmmeter and a chopstick electrode (World Precision Instruments, Sarasota, FL).

One hour prior to the transport experiment, media was removed and replaced with Hank’s balanced salt solution (HBSS) at 37°C with 0.5 mg mL−1 bovine serum albumin (BSA) to allow the cells to equilibrate. After removing the HBSS/BSA solution from the apical side, the cells were exposed to one of the following conditions: HBSS/BSA, neat protein (0.04 mg mL−1 hFIX), and microparticles (0.875 mg mL−1 = 0.391 mg cm−2) with protein (0.04 mg mL−1 hFIX). Over the course of the 4-h experiment, the TEER was measured, and then 200 µL samples were collected from the basolateral side and replaced with HBSS/BSA at 37°C for each time point. After 4 h, final samples from both the apical and basolateral sides were collected, then cells were rinsed with media, and fresh media was added to both sides. At 24 h post study completion, final TEER measurements were collected to determine the recovery of the monolayers. The amount of hFIX on the basolateral side was quantified using hFIX ELISAs.

Statistical Analysis

Data are reported as mean ± standard error. Comparisons between groups were performed using a Student’s two-tailed t test, and comparisons among multiple groups were performed using an ANOVA. Significant differences were considered when *p < 0.05 and **p < 0.01. All statistical analysis was computed using GraphPad Prism 6.

Results

SEM Studies

P(MAA-g-EG) hydrogels were successfully synthesized and crushed into microparticles. SEM micrographs of dried P(MAA-g-EG) microparticles (35–45 µm) shows the irregular morphology due to the crushing process (Fig. 2a). Additionally, the wide polydispersity of microparticle size can be attributed to sieving, which only defines two of the three dimensions. The difference in formulation shows no observable effect on the physical characteristics of the microparticles.

Figure 2
figure 2

(a) SEM images of crushed, dried P(MAA-g-EG) microparticles shown for each formulation—PEGDMA400, PEGDMA600, and PEGDMA1000 (scale bar = 50 µm); (b) dynamic swelling curves show the weight swelling ratio of hydrogel disks in response to change in buffer pH (n = 3, mean ± SE); (c) relative swelling curves show the timescale of microparticle swelling (n = 4, mean ± SE).

Full size image

pH-Responsive Swelling Studies

Dynamic swelling studies were conducted to determine effect of buffer pH on the hydrogel’s swelling response on a short timescale (30 min per buffer). Weight swelling ratios, q, were calculated at each pH condition according to:

$$q = \frac{{W_{\text{S}} }}{{W_{\text{D}} }}$$
(1)

where W S is the swollen weight (mg) and W D is the initial dry weight (mg) of the disk. All formulations remained collapsed at low pH (1.2–4.2), which is critical for protecting the payload in the stomach. As the pH increased to ~5.2, all formulations started to swell, with q increasing as pH increased (Fig. 2b). The effect of the PEGDMA crosslinking agent type (i.e., the length of the PEG unit) was not apparent on the short timescale (30 min intervals) for the macroscopic level (10-mm diameter disks).

Timescale of Microparticle Swelling

The timescale of microparticle swelling, which affects drug release, was quantified by kinetic turbidimetric measurements.9,10 As microparticles swell, the dispersions become more optically transparent, and the refractive index of the swollen microparticle dispersion approaches that of the solvent. Therefore, the swelling of microparticle dispersions reduces the turbidity. Theoretically, turbidity is defined as:

$$\tau = \frac{{cQ_{\text{ext}} }}{2d\rho }$$
(2)

where c is the mass concentration of particles, Q ext is the Mie extinction coefficient, d is the particle diameter, and ρ is the particle density. The value of Q ext is a function of the relative refractive index of the particles and solvent, n p /n 0.9 As the microparticles swell, the relative refractive index decreases (i.e., approaches one), reducing the turbidity. The degree of microparticle swelling is related to the relative turbidity of the particle dispersion to the solvent, defined as:

$$\tau = - \ln (I_{t} /I_{0} )$$
(3)

where I t is the transmittance of the sample at time t and I 0 is that of the pure solvent. With A being the absorbance, transmittance is defined as:

$$I = 10^{(2 - A)}$$
(4)

where, the turbidity can be expressed as:

$$\tau = [A_{t} - A_{0} ]\ln (10)$$
(5)

To monitor the change in turbidity over time for an individual sample, the relative turbidity is defined as:

$$\tau_{\text{rel}} = \frac{{\tau_{t} }}{{\tau_{t = 0} }}$$
(6)

By converting the turbidity data to relative swelling, as defined as:

$$\text{Re} {\text{lative}} \;{\text{swelling}} = \frac{{\tau_{t} - \tau_{t = 0} }}{{\tau_{\infty } - \tau_{0} }}$$
(7)

where τ ∞ is the average τ values at long times (e.g., t = 50–60 min). P(MAA-g-EG) microparticle swelling occurred within the first 5 min, with relative swelling values remaining stable for the reminder of the 60 min (Fig. 2c).

Differences in the relative turbidity values across the different formulations cannot be easily discerned due to the irregular particle morphology and wide polydispersity in size (see Supplementary Fig. S1).

Cytocompatibility Studies

Microparticle cytotoxicity was assessed in Caco-2 and HT29-MTX cells to screen for potentially harmful carriers. Both the effect on cellular metabolic activity (MTS assay) and membrane integrity (LDH assay) were evaluated to determine the concentration dependence of microparticle cytotoxicity after a 6-h exposure, where the average transit time in the small intestine is 2–6 h. For this evaluation, 80% cell viability is considered cytocompatible. At a low concentration (1.25 mg particles mL−1), there is no significant effect on the metabolic activity across all formulations for both cell types (Fig. 3a, c).

Figure 3
figure 3

Cytocompatibility of P(MAA-g-EG) microparticles (6 h exposure) was evaluated in Caco-2 cells using an MTS assay and an LDH assay (a and b, respectively) and in HT29-MTX cells using MTS assay and an LDH assay (c and d, respectively) (For all, n = 3, mean ± SE).

Full size image

The metabolic activity is reduced due to increasing the microparticle concentration, with particularly inhibitory effects observed for 5 mg mL−1 for Caco-2 cells. However, one limitation of these cytotoxicity studies is that the microparticles have settled into a stagnant layer on the cells, possibly creating a barrier to nutrient transport. As a secondary method, the cell membrane integrity was assessed. For all conditions, cell viability is approximately 100% indicating that a 6-h exposure to the microparticles is not membrane disruptive (Fig. 3b, d).

Protein Loading and Release Studies

For a high molecular weight protein like FIX, we investigated the effect of the mesh size, by varying the length of the PEGDMA crosslinking agent, on loading and release. The final amount of protein loaded was calculated based on the concentration of the supernatant after incubation, collapse, and triple rinse. Protein lost during the collapse and rinse was likely surface loaded. The loading level (µg hFIX mg−1 particles) was calculated as:

$${\text{Loading}} = \frac{{c_{0} V_{0} - c_{\text{f}} V_{\text{f}} }}{{c_{\text{p}} V_{0} }}$$
(8)

where c 0 is the initial protein concentration (µg mL−1), V 0 is the initial volume (mL), c f is the final protein concentration (µg mL−1) after rinsing, V f is the final volume (mL) after rinsing, and c p is the particle concentration (mg mL−1).

As shown in Fig. 4a, increasing the PEGDMA length increased the amount of hFIX loaded, with significantly higher loading achieved with both the PEGDMA600 and PEGDMA1000 formulations than the PEGDMA400 formulation. To further investigate the effect of PEGDMA length on loading, FITC-hFIX (0.37 mol FITC per mol hFIX) was loaded to visualize the protein distribution within the microparticles. As confirmed by confocal microscopy (Fig. 4b), hFIX was surface loaded on microparticles for both the PEGDMA400 and PEGDMA600 formulations. However, hFIX was loaded uniformly throughout the microparticles for the PEGDMA1000 formulation. (See Supplementary Fig. S2 for brightfield and overlay confocal images.) Protein loading within the microparticle can be achieved by increasing the mesh size (i.e., longer PEGDMA crosslinking agent), which can lead to improved protection of the protein.

Figure 4
figure 4

Loading amount and distribution of hFIX is improved by increasing the PEGDMA crosslinking agent length of P(MAA-g-EG) hydrogels. (a) The loading of hFIX is dependent on the PEGDMA crosslinking agent length (n = 3, mean ± SE, **p < 0.01); (b) confocal images of FITC-hFIX loaded microparticles (orthogonal view, fluorescent channel) show that hFIX surface loaded for the PEGDMA400 and PEGDMA600 formulations and hFIX loaded throughout the microparticles for the PEGDMA1000 formulation (scale bar = 50 µm).

Full size image

The release of hFIX from P(MAA-g-EG) microparticles was tested using a two-stage dissolution procedure in biorelevant media. For all formulations, hFIX release was minimal in the simulated gastric conditions (FaSSGF), while release significantly increased in the simulated intestinal conditions (FaSSIF) (Fig. 5a). The cumulative release in the FaSSIF conditions was higher for the PEGDMA400 and PEGDMA600 formulations, as compared to PEGDMA1000, due to hFIX being surface loaded and therefore released faster. Release from the PEGDMA100 formulation is expected to be sustained over longer periods as the hFIX must diffuse out of the hydrogel network. While the cumulative protein release showed the desired release profile, the activity of the released protein is critical for its therapeutic use. The activity of released hFIX is better maintained as the length of the PEGDMA crosslinking agent increases (Fig. 5b). Protein released from P(MAA-g-EG) microparticles crosslinked with 1 mol% PEGDMA1000 maintained approximately 100% of its activity, which was significantly higher than the other two formulations. For the PEGDMA1000 formulation, hFIX loading within the microparticle improved protection as confirmed by the activity of released protein. To compare the overall performance, the amount of active hFIX released (Fig. 5c) was calculated as the product of the total release of hFIX and the activity percentage of released hFIX at a given time point. The PEGDMA1000 formulation is the most promising due to significantly higher release of active hFIX.

Figure 5
figure 5

(a) Release of hFIX from P(MAA-g-EG) microparticles using two-stage dissolution in biorelevant conditions shows the desired release profile; (b) activity of released hFIX is better maintained as the length of the PEGDMA crosslinking agent increases; (c) the amount of active hFIX released (the product of the total release and its activity) shows that the PEGDMA100 formulation is the most promising candidate. (For all, n = 3, mean ± SE, *p < 0.05, **p < 0.01).

Full size image

To further optimize the loading efficiency, three parameters—ionic strength of the buffer, time, and temperature—were modified. As previously described,1 reducing the ionic strength of the buffer increases the swelling of anionic hydrogels. Increasing swelling can result in higher protein loading.11 Additionally, increasing both time and temperature can increase the rate of diffusion to increase the amount of protein loaded. However, increasing time and temperature can also reduce the protein stability. A balance between increasing loading and maintaining the protein stability/activity is necessary for determining improved loading conditions.

In order to analyze these parameters, two buffers—1× PBS (162.7 mM) and 5 mM NaH2PO4—were evaluated for their effects on hydrogel swelling and protein stability. Equilibrium swelling studies using the PEGDMA1000 formulation showed minimal swelling at pH 2 and significantly increased swelling at pH 7 (Fig. 6a). At pH 7, the weight-swelling ratio was ~2.3-fold higher for the reduced ionic strength buffer. Additionally, the effect of temperature and time on the stability of hFIX in both buffers at pH 7 was assessed to determine the time and temperature that maintained at least 95% of its activity. After 5 days at 4°C, hFIX was stable in both buffer conditions; however, at higher temperatures, the stability of hFIX dropped over time down to 12% of its relative activity (in 5 mM NaH2PO4 at 37°C) (Fig. 6b).

Figure 6
figure 6

Optimized loading conditions (5 days at 4°C) significantly improved hFIX loading into P(MAA-g-EG) microparticles. (a) Hydrogel disks show increased equilibrium swelling in a reduced ionic strength buffer; (b) time, temperature, and ionic strength effect hFIX stability; (c) loading for 5 days at 4°C significantly increased hFIX loading, and a reduced ionic strength buffer further improved loading. (For all, n = 3, mean ± SE, *p < 0.05, **p < 0.01).

Full size image

Loading studies were conducted at 4°C for 5 days in both buffers to determine if hFIX loaded can be improved for the PEGDMA1000 formulation. Compared to loading at 37°C for 1 day, the new loading conditions resulted in significantly higher hFIX loading levels (Fig. 6c). Reducing the buffer’s ionic strength also improved hFIX loading. Release of hFIX from P(MAA-g-EG) particles showed the desired profile with minimal release in gastric conditions and significantly higher release in intestinal conditions (Fig. 7a).

Figure 7
figure 7

Improved hFIX loading resulted in significantly higher release of active hFIX in intestinal conditions, which can improve the efficacy of an orally delivery therapy. (a) Release of hFIX from microparticles of the PEGDMA1000 formulation using two-stage dissolution in biorelevant conditions shows the desired release profile; (b) activity of released hFIX is maintained in intestinal conditions; (c) improved loading resulted in significantly higher release of active hFIX. (For all, n = 3, mean ± SE, *p < 0.05).

Full size image

The particles loaded in the reduced ionic strength buffer (5 mM NaH2PO4) have higher delivery potential likely due to a higher loading level compared to those using a higher ionic strength buffer (1× PBS). Additionally, released hFIX maintained approximately 80% of its activity in intestinal conditions (Fig. 7b). Overall, improving the amount of hFIX loaded in P(MAA-g-EG) microparticles (PEGDMA1000 formulation) increased the amount of active hFIX released in intestinal conditions (Fig. 7c), which can lead to improved bioavailability of orally delivered therapies.

Factor IX Transport Studies

In vitro transport studies are used to screen for potentially successful candidates and determine the effect of the presence of microparticles on the hFIX absorption. Co-cultures of Caco-2 cells and mucus-secreting goblet-like HT29-MTX cells are commonly used as an intestinal epithelial model for predicting oral drug absorption, particularly paracellular transport in vivo.5 A co-culture model better mimics the physiological conditions of the intestine, containing three major absorption barriers—tight junctions, membranes of absorptive epithelia, and a mucus layer. Additionally, HT29-MTX cells modulate the tight junction geometry, resulting in a “tightness”, indicated by TEER values (~300 Ω cm2), closer to in vivo conditions which range from 20 to 100 Ω cm2 (Supplementary Fig. S3).5 While the co-culture model is more physiologically relevant, the drug permeability is typically lower due to higher TEER values.

Transport studies were conducted to determine the permeability of hFIX across an intestinal epithelial monolayer. Prior to the transport studies, hFIX was endotoxin depleted (final endotoxin concentration <2.5 EU mL−1). For experimental groups, a suspension of 0.4 mg mL−1 hFIX and 0.875 mg mL−1 microparticles (the same surface area coverage of 0.391 mg cm−2 as the 1.25 mg mL−1 concentration used in the cytocompatibility studies and determined as non-cytotoxic) was initially added to the apical chamber, and a 0.4 mg mL−1 hFIX solution was also included as a control. The presence of the P(MAA-g-EG) microparticles promoted the apical-to-basolateral transport of hFIX as compared to the hFIX only (Fig. 8a). Among the microparticle formulations, there was no significant difference in hFIX transported. The permeability of the protein across the monolayer was calculated by:

$$P_{\text{app}} = \frac{dQ(t)}{dt} \times \frac{1}{{A \times C_{\text{AO}} }}$$
(9)

where Q(t) is the cumulative protein release (ng) at time t (s), A is the area of the cell monolayer (cm2), and C AO is the initial protein concentration (ng). hFIX permeability was calculated after the first 2 h, where the amount of hFIX transported over time follows a linear relationship. The overall permeability (4 h) was also calculated; however, values are decreased due to the plateauing of hFIX transported after 2 h, similar to previous reported trends.4 The presence of the microparticles resulted in permeability values at least 2.4-fold higher than hFIX only (Fig. 8b). Previous studies showed the same trends for insulin, where microparticle presence improved permeability, possibly due to Ca2+ chelation.23 Ichikawa et al. hypothesized that PMAA-based microparticles can chelate Ca2+, leading to an opening of tight junctions and increasing permeability.6

Figure 8
figure 8

The presence of the P(MAA-g-EG) microparticles enhanced the permeability of hFIX in vitro. (a) The amount of hFIX transported across the intestinal epithelial monolayer over time shows that the presence of microparticles improved hFIX transport (n = 3); (b) permeability of hFIX (2 h and 4 h) are higher with the presence of microparticles compared to hFIX only (n = 3); (c) TEER remained constant during the transport study, indicating that the tight junctions remained intact (n = 4). (For all, mean ± SE, *p < 0.05, and **p < 0.01).

Full size image

Additionally, the TEER was monitored for the study duration to determine the effect of the microparticles on the monolayer integrity. Similar to previous studies,23 the TEER values remained relatively constant over the 4-h exposure, indicating that the tight junctions remained intact (Fig. 8c). As hypothesized by Wood et al., the microparticles may have a Ca2+ chelating effect, but the mucus layer provides barrier between the cells and microparticles, reducing the effect of the microparticles on TEER.23 The monolayers also recovered after the removal of the microparticles, as indicated by the 24-h TEER measurements.

Discussion

Addressing the medical need, we have developed a delivery system based on pH-responsive P(MAA-g-EG) hydrogel networks for the oral delivery of hematological FIX. Such hydrogels exhibited the desired characteristics—pH-responsive swelling and cytocompatibility—necessary for therapeutic applications. The pH-responsive behavior provides protection of hFIX in gastric conditions and allows for targeted intestinal release.

Tailoring the biomaterial formulation by varying the length of the PEGDMA crosslinking agent resulted in improved performance of the microcarriers. Increasing the mesh size increased loading and distribution of hFIX within the microparticles, offering more protection in gastric conditions. Furthermore, improved hFIX protection resulted in higher release of active protein in intestinal conditions. Based on the results presented, P(MAA-g-EG) crosslinked with 1 mol% of PEGDMA1000 is the best performing formulation. Further optimization of loading conditions (i.e., reduced ionic strength of buffer and increased time to 5 days at 4°C) significantly improved the loading efficiency and the amount of active hFIX released.

The ability of the microcarriers to modulate the oral absorption of hFIX further shows the potential of such oral delivery systems. The presence of the microparticles increased hFIX permeability without causing damage to the in vitro intestinal epithelial layer. The combination of well-maintained hFIX activity and its enhanced permeability is expected to promote overall bioavailability.

The promising work presented here can lead to an orally administered treatment for hemophilia that can replace needle-based options, offering improved quality of life for patients and global access to therapy.