Oral Delivery of Metformin by Chitosan Nanoparticles for Polycystic Kidney Disease

Abstract

Nanoparticle drug delivery has many advantages over small molecule therapeutics, including reducing off-target side effects and increasing drug potency. However, many nanoparticles are administered parenterally, which is challenging for chronic diseases such as polycystic kidney disease (PKD), the most common hereditary disease worldwide in which patients need continuous treatment over decades. To address this clinical need, we present the development of nanoparticles synthesized from chitosan, a widely available polymer chosen for its ability to improve oral bioavailability. Specifically, we optimized the synthesis parameters of chitosan nanoparticles and demonstrate mucoadhesion and permeation across an intestinal barrier model in vitro. Furthermore, when administered orally to mice, ex vivo imaging of rhodamine-loaded chitosan nanoparticles showed significantly higher accumulation in the intestines compared to the free model drug, as well as 1.3 times higher serum area under the curve (AUC), demonstrating controlled release and improved serum delivery over 24 hours. To test its utility for chronic diseases such as PKD, we loaded the candidate PKD drug, metformin, into chitosan nanoparticles, and upon oral administration to a PKD murine model (Pkd1^fl/fl;Pax8-rtTA;Tet-O cre), a lower cyst burden was observed compared to free metformin, and was well tolerated upon repeated dosages. Blood urea nitrogen (BUN) and creatinine levels were similar to untreated mice, demonstrating kidney and biocompatibility health. Our study builds upon previous chitosan-based drug delivery approaches, and demonstrates a novel, oral nanoformulation for PKD.

1. Introduction

Since the Food and Drug Administration (FDA) approved liposomal doxorubicin (Doxil) for cancer treatment in 1995, now more than 50 nano-pharmaceuticals are available for the clinic [1, 2]. Nanoparticles have been shown to improve drug therapeutic efficacy, reduce toxicity, and increase tissue selectivity compared to small molecule drugs [3]. Moreover, nanoparticles can combine multiple functionalities, including therapeutic and diagnostic capabilities, onto a single nanoparticle platform, [4, 5] and have the potential to provide feedback on treatment effectiveness in real-time [6]. As such, nanomedicine has proven beneficial in the treatment of cancer [7, 8], multiple sclerosis [9], and human immunodeficiency virus (HIV) [10].

Despite these advances, similar developments in nanotechnology to improve the standard of care for chronic diseases is limited. For instance, autosomal dominant polycystic kidney disease (ADPKD) affects up to 12.5 million individuals worldwide [11], but no nanomedicine efforts have been developed [12, 13]. ADPKD is a slowly progressing, irreversible genetic condition characterized by cyst formation and enlargement arising from aberrant tubular epithelial growth and fluid secretion occurring throughout the kidney nephron, ultimately destroying kidney function [14, 15]. Several potential therapies have demonstrated benefits in slowing cyst growth in mice or clinical trials [16, 17], such as metformin (met), an FDA-approved drug for type 2 diabetes [18, 19], and tolvaptan, the only drug that was approved by the FDA in 2018. However, ADPKD preclinical studies administered drugs such as met at a higher dose (300 mg/kg/day) than currently prescribed (maximum 37.5 mg/kg/day for diabetic patients). Subsequently, adverse side effects such as gastrointestinal (GI) discomfort manifest in 25% of patients, and 5% of patients develop complete intolerance for the drug [20–22]. In addition, more serious side effects of met include hypoglycemia and lactic acidosis, that latter of which has been found to be fatal in certain patients [23]. Tolvaptan, which is specifically prescribed to ADPKD patients with a rapidly progressing cyst phenotype [24], can also be a difficult drug to tolerate due to its many off-target side effects such as nausea, polyuria, muscle cramps, and idiosyncratic liver toxicity [25, 26]. The tolvaptan clinical trial dropout rate was significant at 23%, and projections show that after 18 years of continuous tolvaptan treatment, only a modest benefit of 4.9 year delay is achieved until kidney failure [27]. Therefore, new drug delivery approaches that can decrease systemic toxicity without compromising therapeutic efficacy is imperative for chronic diseases such as ADPKD.

Previously, kidney-targeted nanoparticle drug carriers using intravenous (IV) administration have been developed to enhance drug accumulation in the kidney and reduce such systemic side effects [12, 13, 28]. However, IV administration is not practical nor feasible in many cases for chronic diseases that progress over a lifetime, such as ADPKD [29]. Instead, oral drug delivery is the most convenient route of self-administration, and results in the greatest treatment adherence [30, 31], preferred by 70% of patients [32]. Particularly for chronic conditions, oral delivery is attractive as it avoids needle complications such as infection, phlebitis, and pain [33, 34]. Additionally, patients with high needle fear and chronic life-threatening health conditions (e.g., diabetes and multiple sclerosis) have been found to make important treatment decisions based on their aversity to needles over medical expertise [35–37], and hence, achieving a tolerable, self-administrable route is an important aspect to clinical success of future targeted therapies [38, 39].

Although favorable for patient compliance, orally-delivered drugs or nanoparticles must overcome unique physiological barriers that have historically limited their therapeutic efficacy (Figure 1) [40]. These challenges include the acidic pH and enzymes present in the stomach that can degrade pharmaceutically active drugs [41], as well as the intestinal epithelial barrier that acts as a selectively permeable barrier to drugs for systemic circulation [42]. Furthermore, enterocytes within this epithelium secrete a mucosal layer, presenting a continuously recycled barrier hundreds of micrometers thick, that have been found to trap and remove nanoparticles, substantially limiting drug bioavailability [43]. Moreover, even upon reaching the blood after absorption in the intestines, the first pass effect can metabolize up to 70–90% of orally administered drugs, rendering it therapeutically inactive through biotransformation [42, 44, 45].

Figure 1.

Schematic of the physiological barriers for orally delivered nanoparticles. Adapted and reprinted from Smart Servier Medical Art.

Chitosan-based materials have been proposed for oral delivery, as they offer many favorable properties such as biocompatibility, mucoadhesion, and tunability for controlled drug delivery [46, 47]. Chitosan is derived from naturally occurring chitin found in the shells and exoskeletons of many crustaceans and is the second most abundant polysaccharide [48–50]. The purification process of chitin also allows tuning of the resultant chitosan, such as molecular weight, pKa (6–7.5) and degree of deacetylation properties, which provides a biomaterial that can be tailored for a wide range of biomedical applications [51, 52].

Currently, chitosan is used in commercial biomedical products like the AQUANOVA Super-Absorbent Dressing, and is currently under clinical investigation as dental fillers (NCT03237624) and wound dressings (NCT03719261). Chitosan is considered Generally Recognized As Safe (GRAS) and edible by the FDA, but has not been directly approved for any nanoparticle drug delivery usage. The bottleneck may lie in the poor correlation between specific formulations or modifications of chitosan and the predicted in vivo response [53]. Hence, systematic studies assessing chitosan properties such as nanoparticle size and degree of acetylation are still needed to exploit the beneficial properties of chitosan for drug delivery applications in the clinic [34, 54, 55].

Herein, we take advantage of these properties to develop chitosan nanomaterials that have optimal size, stability, and mucoadhesion to navigate through the GI tract and achieve efficient systemic delivery compared to free drugs [56]. Specifically, to form chitosan nanoparticles, several methods have been studied including polyelectrolyte complexation [57], covalent cross-linking [58], complex coacervation [59], and ionotropic gelation [60–62]. We selected ionotropic gelation for our studies as the mild and aqueous processing conditions, non-toxic reagents, and ease of production is suitable for eventual clinical scale-up [63–66]. Moreover, chitosan nanoparticles have been previously shown to successfully deliver therapeutics in vivo, such as insulin [67, 68], cyclosporin A, an immunosuppressant [69], and enoxaparin, an anticoagulant [70], further supporting its clinical suitability.

Due to these benefits, in this study, we synthesized chitosan nanoparticles (CS-NP) to test the hypothesis that CS-NP can be used as an oral delivery platform for ADPKD and other chronic conditions. We investigated the synthesis parameters and its effect on the nanoparticle size, polydispersity, loading efficiency, and degradation rate in pH ranges that are present in the GI tract. We tested the ability of these nanoparticles to permeate an intestinal barrier model and deliver the candidate ADPKD drug, metformin in vitro [71]. Finally, we evaluated the oral delivery of chitosan nanoparticle met (CS-NP met) in a murine model of ADPKD, and demonstrate enhanced therapeutic efficacy compared to the free drug upon oral delivery. These results lay the groundwork for oral drug delivery applications using CS-NP, not only in the context of ADPKD and chronic diseases as provided in this study, but treatments administered IV lacking the ability to overcome enteric barriers.

2. Materials and methods

2.1. Materials

Chitosan with 95% degree of deacetylation and average molecular weight 150 kDa was purchased from Heppe Medical (Germany). Mucin type II from porcine stomach, Rhodamine B, pharmaceutical grade met, and poly-L-glutamic acid were purchased from Sigma-Aldrich (USA). All other reagents were of analytical grade.

2.2. Synthesis of CS-NP

Chitosan (0.5, 1.0, 1.5, 2.0, 2.5, and 3.0 mg/ml) was dissolved in MilliQ water containing 0.5% glacial acetic acid, sonicated and vortexed to obtain homogenous mixtures. Similar concentrations (0.5, 1.0, 1.5, 2.0, 2.5, and 3.0 mg/ml) of poly-L-glutamic acid solutions were prepared in MilliQ water. Chitosan solution was added dropwise to polyglutamic acid under constant stirring in a round bottom flask at 600 rpm; an opalescent solution was seen upon successful formation of nanoscale particles. The final solution was centrifuged at 14,000 rpm for 30 minutes at 14°C. The resulting pellet was serially washed with 20%, 75%, and 100% ethanol. The pellet of nanoparticles was resuspended in MilliQ water or PBS and rapidly pipetted until a homogenous mixture was obtained and was used immediately for further studies. To encapsulate a payload, met or rhodamine B was dissolved at desired concentrations in poly-glutamic acid solution and used during nanoparticle synthesis. Loading efficiency of the payload was calculated by quantifying the amount of unincorporated met or rhodamine remaining in the supernatant. Rhodamine fluorescence was measured at λ_excitation = 553 nm and λ_emission = 630 nm, while met absorbance was measured at 233 nm using a Varioskan LUX plate reader [72, 73] (Thermo Fisher Scientific, Waltham, MA, USA). Loading capacity was determined by dividing the weight of the known amount of encapsulated drug by the weight of the total nanoparticle sample.

2.3. Dynamic light scattering (DLS)

Nanoparticles derived from chitosan (0.5–3.0 mg/ml) and poly-L-glutamic acid (0.5–3.0 mg/ml) concentrations were dispersed in 63 μL of MilliQ water and measured by DLS to confirm size and polydispersity index (PDI). DLS measurements were determined at 163.5° and 532 nm using a Wyatt Technology Möbiuζ system (Santa Barbara, CA, USA, N ≥ 3). All measurements were carried out at 25°C after equilibrating for 5 minutes.

2.4. Zeta potential

The zeta potential of chitosan nanoparticles was measured using the same Möbiuζ system described above. Samples dissolved in MilliQ water were placed in a Quartz cuvette with a polyether ether ketone (PEEK) and platinum dip probe (N ≥ 3) and measurements were carried out at 25°C.

2.5. Transmission electron microscopy (TEM)

TEM samples were prepared by placing 7.0 μL of chitosan nanoparticles in MilliQ water on 400 mesh lacey carbon grids (Ted Pella, Redding, CA, USA) for 5 minutes. Excess liquid was wicked away with filter paper and the grid was washed with MilliQ water before placing 2 wt.% uranyl acetate solution for 2 minutes. After washing once more with MilliQ water, samples were dried and immediately imaged on a JEOL JEM-2100F (JEOL, Ltd., Tokyo, Japan).

2.6. Re-acetylation of chitosan

Chitosan with ~95% deacetylation was re-acetylated to achieve varying degrees of N-deacetylated chitosan. Chitosan (2 mg/ml) was mixed with 200 mM acetic anhydride in a 50:50 methanol/water mixture at 95°C and stirred for 1–12 hours. Confirmation of deacetylation degree was measured from the first derivative of the UV-vis absorption spectra obtained from a Varioskan LUX plate reader as specified by de Silva et al. [74]. Reaction times between 1–6 hours produced chitosan with 95–85% degree of deacetylation, while 6–12 hours and >12 hours resulted in 70–85% and 55–70% deacetylation, respectively.

2.7. Mucin-binding assay

Porcine mucin (PM) in phosphate buffer (pH 7.4) was incubated with CS-NPs of varying degrees of deacetylation verified by UV-vis, (90, 80, 70, 50% deacetylation, 50–300 nm diameter) at room temperature (23°C) for 2 h (1:1, v/v), before centrifugation for 60 min at 14,000 rpm and 14°C. Absorbance of the remaining free PM in the supernatant was measured by UV spectrophotometry at 251 nm. The mucoadhesiveness was expressed as PM binding efficiency calculated by the following equation:

$$\mathit{\text{Binding}}\mathit{\text{efficiency}}=\frac{C_0-C_s}{C_0}\times 100$$ (1)

where C_o is the initial concentration of PM used for incubation (400 μg/mL) and C_s is the measured concentration of free PM in the supernatant after removal of chitosan-bound PM. The standard curve was determined using 50, 100, 150, 200, 250, 300, 350 μg/mL PM solutions.

2.8. Drug release and morphological response to pH

Drug release studies were performed on nanoparticles (2 mg/ml initial chitosan and 1 mg/ml poly-L-glutamic acid) suspended in PBS adjusted to pH 1.2, 2.5, 6.5, or 7.4 with the addition of HCl or NaOH, simulated gastric fluid (SGF) composed of 2.0 g/L sodium chloride and 2.9 g/L HCl (pH 1.3), or simulated intestinal fluid (SIF) composed of 0.62 g/L sodium hydroxide and 6.8 g/L potassium phosphate monobasic (pH 6.8) [75]. Free met released from nanoparticles was quantified at 233 nm using a NanoDrop One microvolume UV-Vis spectrophotometer for up to 6 hours at room temperature (Thermofisher Scientific, Waltham, MA, USA).

To assess the morphology of nanoparticles in response to pH conditions, particles were immersed in each pH condition for 6 hours, and imaged via TEM. All experiments were carried out in triplicate.

2.9. Cell culture

Human colon epithelial cells (Caco-2, ATCC HTB-37, ATCC, Manassas, VA, USA) were cultured following the manufacturer’s recommendations. Cells were expanded in Dulbecco’s Modified Eagle Medium (DMEM, ATCC-30–2003) supplemented with 10% fetal bovine serum (FBS) and 1% penicillin-streptomycin. Cells were seeded at a density of 4.5 × 10³ cells/cm² and subcultured upon 50% confluence [76].

Mouse kidney cortical collecting duct (mpkCCD_c14) cells were expanded in culture media comprised of DMEM/F12 (11054–054, Waltham, MA, USA) supplemented with insulin, dexamethasone, selenium, transferrin, triiodothyronine, glutamine, d-glucose, epidermal growth factor (EGF), HEPES, sodium pyruvate as outlined by Bens et al. [77]. Complete media was filtered before use, and media was changed every two days and subcultures were passaged every 7–8 days. Both cell lines were grown at 37°C in a humidified incubator under 5% CO₂.

2.10. In vitro cell compatibility

Biocompatibility was assessed with an MTS cell proliferation colorimetric assay following the manufacturer’s instructions (BioVision Incorporated, San Francisco, CA, USA). MpkCCD_c14 (5,000 cells/well) or Caco2 (5,000 cells/well) were incubated with either 10, 50, 100, or 500 μM of CS-NP for 24 hours on a 96-well plate before the addition of MTS reagent. Assay fluorescence was measured via a Varioskan LUX plate reader (Thermo Fisher Scientific, Waltham, MA, USA).

2.11. Transepithelial resistance (TER) surveillance

For transport and monolayer resistance experiments, Caco-2 cell monolayers were seeded onto Transwell inserts (Corning, NY, USA; diameter 6.5 mm, growth area 0.33 cm², pore size 0.4 μm), at an initial density of 3 × 10⁵ cell/cm² and maintained for 21 days in complete medium to form a confluent monolayer. The change of TER, representing the tightness of the cell monolayers, was measured by an EVOM2 Epithelial Voltohmmeter (World Precision Instruments, USA). Monolayers reaching steady-state values in the range of 300–400 Ω-cm² were used for studies [76].

2.12. Cellular uptake and transport of CS-NP

To assess the cellular uptake pathways of chitosan nanoparticles, Caco-2 cells were pretreated with medium containing colchicine (transcytosis inhibitor, 10 μM, for 60 min) or wortmannin (micropinocytosis inhibitor, 0.06 mM, for 3 hours) [78, 79] before 100 uL of 100 μM rhodamine-loaded chitosan nanoparticles (CS-NP R) was administered in the apical chamber. The amount of rhodamine fluorescence was measured in the basolateral chamber over the course of 4 hours.

Additionally, the effect of paracellular transport through tight junctions was monitored via TER of Caco-2 cells seeded on Transwell inserts. TER measurements were performed daily for three days prior to treatment with CS-NP R, free Rhodamine (free R), or PBS control to establish baseline measurements. TER was monitored every 6 hours, then again on day 2 and 3 post-administration.

2.13. In vitro therapeutic efficacy of CS-NP met through ELISA and epithelial sodium channel (ENaC) measurements

To assess therapeutic efficacy in vitro, the cellular levels of phosphor-AMPK (Kit #7959) and total AMPK (Kit #7961) were measured via enzyme-linked immunosorbent assays (ELISA, Cell Signaling Technologies, Danvers, MA, USA) according to the manufacturer’s instructions. mpkCCD_c14 cells were treated for 12 hours with 300 μM of met in CS-NP met or free met, and were compared to CS-NP or PBS treated controls. All standards and samples were measured on a Varioskan LUX microplate reader at a wavelength of 450 nm.

To validate the therapeutic efficacy of met on the reduction of ENaC current, TER and potential difference (PD) measurements were made on mpkCCD_c14 cells seeded on Transwell filters as described above [20]. ENaC-dependent equivalent short-circuit currents (I_eq) was estimated by Ohm’s law, dividing the measured PD by the TER value and the area of one Transwell membrane (0.33 cm²). mpkCCD_c14 cell cultures were treated with 100 uL CS-NP met (300 μM met), CS-NP, or PBS at pH 7.4.

2.14. Ex vivo imaging of orally administered CS-NP

To assess the biodistribution of CS-NP semi-quantitatively, 6–7 week old male and female C57BL/6J mice (Jackson Laboratories, Bar Harbor, ME, USA) were orally gavaged with 10 mg/kg of rhodamine loaded in 200 uL of 500 μM CS-NP R, 10 mg/kg of free R, or PBS control. Mice were euthanized after 3, 24, or 48 hours post-injection and organs (e.g., brain, heart, lungs, liver, kidneys, spleen, intestines, and bladder) were excised and imaged ex vivo on an AMI HTX in vivo imaging system (Spectral Instruments Imaging, Tuscon, AZ, USA). The fluorescence signal was quantified via Aura software (Spectral Instruments Imaging, Tuscon, AZ, USA, N ≥ 4), and background was subtracted from the PBS-treated group. The mean radiance (photons/s/cm²/sr) for each organ was quantified as a region of interest, and % of total organ fluorescence was obtained by dividing each organ by the sum of all the organ regions. Urine and blood samples were collected following organ harvest and stored at −20°C until further analysis. All animal procedures followed NIH guidelines for the care and use of laboratory animals and were approved by the University of Southern California’s Institutional Animal Care and Use Committee.

2.15. In vivo half-life of CS-NP

To compare serum half-life, 200 uL of CS-NP R or free R was administrated at a dose of 10 mg/kg loaded rhodamine via oral gavage on 6–7-week-old male and female C57BL/6J mice (N ≥ 4). Blood draws were performed either retro-orbitally or via tail vein at 30 min, 3, 6, 12, 24, 36, and 48 hours post-administration. Fluorescence was measured in serum and quantified using a rhodamine calibration curve developed in mouse serum. Absolute bioavailability was calculated for CS-NP R and free R relative to mice IV injected via tail vein with 10 mg/kg rhodamine dissolved in 100 uL of PBS.

2.16. Histology and immunohistochemistry

Following ex vivo imaging, the brain, heart, lungs, liver, kidneys, spleen, intestines, and bladder were immediately frozen and embedded in OCT (Tissue Tek, Sakura Finetek, Torrance, CA, USA). 10 μm sections were obtained (CM3050 S Cryostat, Leica, Nussloch, Germany) and placed on Superfrost Plus slides (Fisherbrand, Waltham, MA, USA). Tissue sections were stained with hematoxylin & eosin (H&E) and imaged (Leica DMi8, Leica, Wetzlar, Germany, N = 3). For staining of intestinal mucin, tissue sample were processed with an alcian blue 1%, pH 2.5 stain kit (Newcomer Supply, Middleton, WI, USA). Briefly, tissue sections on slides were washed in acetic acid for 3 minutes, incubated in alcian blue for 30 minutes at 4°C in a humidified chamber, and counterstained with Nuclear Fast Red (Vector Laboratories, Burlingame, CA, USA). Samples were mounted using VectaMount™ (Vector Laboratories, Burlingame, CA, USA).

2.17. Therapeutic efficacy in ADPKD mice

To assess the ability CS-NP to enhance the therapeutic efficacy of orally administered drugs in ADPKD, 300 mg/kg met-loaded CS-NP or free met was administered to Pkd1^fl/fl;Pax8 rtTA;Tet-O-Cre mice. Pups were IP injected with doxycycline on postnatal day 10–11 (P10–11) to induce a severe PKD phenotype as previously described [80]. Mice were orally gavaged every two days starting on P12 and euthanized on P22 (N ≥ 4). Kidneys were excised to assess kidney to body weight (KW/BW) ratio and stained with H&E to compare cystic index. Cystic index was defined as the percentage of cystic area divided by total kidney area [81] and determined by ImageJ.

2.18. Kidney health in PKD mice

Serum components, electrolytes, and kidney health markers including sodium (Na), potassium (K), chloride (Cl), ionized calcium (iCa), total carbon dioxide (tCO₂), glucose (Glu), blood urea nitrogen (BUN)/Urea, creatinine (Crea), hematocrit (Hct), hemoglobin (Hb), and anion gap (AnGap) were assessed in diseased Pkd1^fl/fl;Pax8 rtTA;Tet-O-Cre mice. 90 uL of blood taken from the submandibular vein on the day of harvest was analyzed using Chem-8+ cartridges for the i-Stat Handheld Blood Analyzer (Abbott, Chicago, IL, USA).

2.19. Statistical analysis

A Student’s t-test was used to compare means of pairs. Analysis of variance (ANOVA) with Tukey’s multiple comparison test post-hoc analysis was used to determine significant differences among three or more means. A p-value of ≤ 0.05 was considered to be significant.

3. Results and discussion

3.1. Fabrication and characterization of chitosan nanoparticles

Chitosan is an easily procured biomaterial, and several synthesis methods have been investigated to synthesize chitosan nanoparticles [82–85]. Ionic gelation was used in this study as it has been reported to achieve high drug encapsulation and low polydispersity [64, 65], and ionic gelation is based on electrostatic interaction between the amine group of chitosan and a negatively-charged group of a polyanion such as poly-L-glutamic acid [61]. Due to this charge-based interaction, negatively-charged payload drugs can also be easily incorporated during the ionic gelation process [86].

In addition, one of the main advantages of using chitosan for oral delivery is its mucoadhesive properties, as the mucous lining in the GI tract provides a significant barrier to many drugs [87]. Mucous is composed of water (~90 to 98%), salts (~0.5 to 1.0% w/w), proteins (~0.5% w/v), and mucins (0.2–5% w/w) [88] which are the glycoproteins responsible for excluding large micrometer sized particulates by steric hinderance [89]. Without mucoadhesive properties, oral delivery formulations lack the ability to withstand peristalsis movements [90] as well as the extensive washing effect of body fluids, such as GI acids [91], which results in the loss of drug payload available systemically.

In previous studies, chitosan polymers have been shown to bind to mucins more effectively as the degree of deacetylation (DDA) increased [92], as the additional positively-charged amino groups allow for increased interaction with negatively-charged sialic acid residues of mucin [93]. Similarly, in this study, upon deacetylation of the starting chitosan (before nanoparticle synthesis), the zeta potential of the chitosan solutions confirmed increasing % deacetylation increased the positive charge from 15.8 ± 4.2 mV at 50% DDA to 27.3 ± 2.8 mV at 90% DDA (Figure S1). When chitosan polymers of varying DDA (90%, 80%, 70%, and 50% DDA) were tested in a mucin binding assay, the highest DDA (90%) showed the greatest binding efficiency to mucin (75.1 ± 5.0%) (Figure 2a).

Figure 2.

Optimization of CS-NP synthesis parameters and mucin binding efficiency. (a) Binding efficiency of chitosan to mucin increases upon chitosan DDA. (b) Nanoparticle diameter and (c) polydispersity as a function of the starting chitosan and poly-glutamic acid crosslinker concentration. The lowest polydispersity for CS-NP is seen at 2 mg/ml chitosan and 1 mg/ml crosslinker concentrations. (d) Diameter of CS-NP with respect to mucin binding efficiency shows highest binding at approximately 150 nm. (*p ≤ 0.05, **p ≤ 0.01, N ≥ 3) (e) TEM micrographs confirm spherical morphology and a monodisperse population of CS-NP with 90% DDA chitosan at 2 mg/ml and poly-L-glutamic acid crosslinker at 1 mg/ml synthesis conditions.

In order to identify the ideal nanoparticle diameter to adhere and diffuse through the mucosal layer, 90% DDA chitosan was used to synthesize chitosan nanoparticles (CS-NP) of various diameters (50, 100, 150, 200, 250, 300 nm), achieved by altering the starting concentrations of chitosan and poly-L-glutamic acid crosslinker during ionic gelation. Nanoparticle diameter has been found to affect the ability to bind and diffuse through mucus, as the mesh pore size (10–200 nm) of mucus sterically limits nanoparticles larger than 200 nm [94]. Chitosan concentrations were limited between 0.5 mg/ml and 3.0 mg/ml, as higher concentration ranges resulted in immediate aggregation of particulates upon dropwise addition to polyglutamic acid, and lower concentrations resulted in no nanoparticle formation. As shown in Figure 2, CS-NP of approximately 150 nm in diameter demonstrated the highest mucin binding efficiency (74.8 ± 4.0%) (Figure 2d). We observe a decrease in binding for particles beyond 200 nm, which is consistent with the study by S. Bandi et al. demonstrating nanoparticles ≥ 200 nm in diameter had limited diffusivity through mucus due to mucin mesh steric hindrance [95]. Hence, we selected 90% DDA chitosan at 2 mg/ml and poly-L-glutamic acid crosslinker at 1 mg/ml synthesis conditions, which resulted in nanoparticles with the desired diameter of 150 nm and a low PDI of 0.24, to proceed with further studies (Figure 2b,c,e). Increasing mucodiffusion has also been a viable engineering strategy to access underlying enterocytes and increase systemic bioavailability [96]. While not investigated in this study, the mucodiffusive properties of chitosan may play a role in navigating the GI tract, and will be evaluated in future studies.

3.2. Drug release and degradation properties of CS-NP

To successfully deliver substances through the GI tract, enteric delivery systems must protect the payload from degradation and premature release in the low pH environment of the stomach [97]. The amino groups in chitosan (pK_a = 6.5) are protonated to form NH₃⁺ in low pH (pH 1.2–2.5), providing strong electrostatic attractions to the oppositely-charged poly-glutamic acid crosslinker, which allows the particles to remain compacted and remain at their original size and retain the payload [98]. At higher pH values (pH = 6.5–7.4), the amine groups of chitosan exist mostly in the NH₂ form [99]. As a result, the electrostatic interactions between CS and crosslinker are weakened, which favors dissolution [100].

To verify CS-NP have the ability to protect drugs past the gastric environment, we tested met-loaded CS-NP (CS-NP met) in pH environments representative of a fasting stomach (pH 1.5), fed stomach (pH 2.5), duodenum of the small intestine (pH 6.5), circulating blood (pH 7.4), SGF (pH 1.3), and SIF (pH 6.8) [101, 102]. The loading efficiency and loading capacity of met into CS-NP was determined to be 32.2 ± 2.8% and 37.3 ± 3.6%, respectively (Figure S2a,b), likely due to the electrostatic interaction of positively-charged met with the negatively-charged crosslinker during nanoparticle synthesis [103]. To assess met release profiles, we quantified free met release from CS-NP met in the various pH environments for up to 6 hours. We observed that payload release did not exceed 25% at pH conditions 1.2, SGF (pH 1.3), and 2.5, while greater than 50% release was found in pH 6.5, SIF (pH 6.8), and 7.4 at 3 hours (Figure 3a). Consistent with release studies, TEM images showed minimal morphological change at pH 1.2 and 2.5, (Figure 3b), while an increase in diameter (~250 nm) was seen in pH 6.5. At pH 7.4, CS-NP met lost their spherical morphology and fused with adjacent particles. In agreement with observations in other chitosan nanoparticle studies, water is able to intercalate through pores in the polymer matrix, expanding the particle and causing swelling and degradation [104]. This further suggests that CS-NP can remain stable and protect drugs under the low pH conditions found in the stomach, but swell and release drugs upon reaching neutral pH found in the small intestine and systemic circulation.

Figure 3.

In vitro pH response of CS-NP. (a) In vitro release of met from CS-NP under various pH conditions present in the GI tract (pH = 1.2, fasting stomach; 2.5, fed stomach; 6.5 intestines, 7.4 blood; 1.3, SGF; 6.8 SIF) (N ≥ 4). (b) TEM images of CS-NP confirm degradation at pH 6.5 and 7.4 after 6 hours.

3.3. In vitro penetration across intestinal epithelium

In addition to mucoadhesive properties and protection of payloads in the acidic environment of the stomach, chitosan has been suggested to enhance the penetration of the intestinal epithelial cell barrier by opening tight junctions and increase transport of luminal peptides, nutrients, and nanoparticles [105]. To verify this, CS-NP rhodamine (CS-NP R), free rhodamine, or PBS was tested in an intestinal epithelial barrier model consisting of human colorectal Caco-2 cells, on Transwell membranes [106], and tight junction integrity was determined via transepithelial resistance (TER) measurements [79]. TER measurements were made three days before treatment, and again after administration of CS-NP R, free rhodamine, or PBS for up to 3 days. No changes in TER occurred in the PBS or free rhodamine-treated groups, while an 84.8% reduction in resistance to 53.1 ± 32.3 Ohm*cm² was observed for CS-NP R 6 hours after administration (Figure 4a). A recovery to pre-treated baseline resistance levels (355 Ohm*cm²) was seen after 3 days, suggesting the effects on tight junctions are reversible, yet persist enough on the time scale that digestion occurs in the human gut [107].

Figure 4.

In vitro transport mechanisms and therapeutic efficacy of CS-NP (a) TER measurements of a Caco-2 cell layer upon 100 μM CS-NP R, free R, or PBS treatment 3 days before and after treatment show paracellular transport through tight junctions. (b) CS-NP R permeation across a Caco-2 cell layer treated with 100 μM of CS-NP R after 4 hours, pretreated with either colchicine (transcytosis inhibitor), wortmannin (micropinocytosis inhibitor) or no inhibitor. (c) Phosphorylated AMPK to total AMPK obtained via ELISA; and (d) ENaC current measurements of mpkCCD_c14 cell monolayers treated for up to 48 h with CS-NP met (300 μM), free met, CS-NP, and PBS show a significant decrease for the CS-NP met and free met groups, compared to CS-NP, confirming therapeutic activity (****p ≤ 0.0001,***p ≤ 0.001, N ≥ 4).

In addition to paracellular transport, transcellular transport has been reported for nanoparticles passage through the intestinal lining [108]. Previous studies have suggested chitosan nanoparticles can undergo endocytosis (clathrin-mediated) and micropinocytosis in intestinal cells [109]. Specifically, to determine if transcytosis or micropinocytosis is mainly responsible for the transport of CS-NP, Caco-2 cell layers were treated with colchicine (10 μM, for 60 min), wortmannin (0.06 mM, for 3 h), or no inhibitor before 100 μM of CS-NP R incubation, and rhodamine fluorescence was measured in the basolateral chamber over the course of 4 hours. A 27% reduction of transport was seen when the transcytosis inhibitor colchicine was administered, while no reduction was seen when the micropinocytosis inhibitor wortmannin was administered (Figure 4b). These initial findings suggest that transcytosis is a major pathway by which CS-NP are transported across intestinal epithelial layers [79, 110, 111], in addition to increasing paracellular permeation. Importantly, over 90% of cells were viable upon treatment with CS-NP as indicated by an MTS assay, suggesting transport differences are not due to cytotoxic effects of chitosan (Figure S3a). The CS-NP R transport behavior is also distinct from free rhodamine administered in the same conditions (Figure S3b). No statistically significant reduction is observed in transport of free R between the colchicine and wortmannin conditions, while colchicine reduces CS-NP R transport.

3.4. In vitro therapeutic efficacy of metformin-loaded CS-NP

Upon characterizing the biomaterial properties of CS-NP and verifying its potential to deliver drugs through intestinal epithelia, met was loaded into CS-NP for ADPKD applications. We selected met, a first-line therapy already approved for diabetes, due to its secondary benefits in inhibiting ADPKD preclinically, resulting in several ongoing clinical trials repurposing met for ADPKD including METROPOLIS (NCT03764605) and TAME (NCT02656017) [112, 113]. Specifically, met activates the 5’ AMP-activated protein kinase (AMPK) pathway by phosphorylating AMPK. This leads to inhibition of the mammalian target of rapamycin (mTOR) pathway [114], responsible for the expansion of cysts due to the overproliferation of renal tubular cells. Additionally, met inhibits intracellular generation of cAMP via inhibition of adenylyl cyclase [115], a key signaling pathway that drives cystogenesis in ADPKD. In addition, AMPK activation has been found to inhibit fluid secretion into cysts by inhibiting the cystic fibrosis transmembrane receptor (CFTR) channel [116], the key apical membrane chloride secretory route in ADPKD [117]. In ADPKD preclinical murine studies, met was administered at a dose of 300 mg/kg/day which greatly exceeds the dose currently prescribed for patients with diabetes (maximum 37.5 mg/kg/day). Even at low doses, 25% of patients already suffer from GI discomfort and approximately 5% are unable to tolerate met entirely due to these side effects [20–22]. Since the bioavailability of orally taken met is only 40%, it is expected that high doses are needed for PKD efficacy, which may further exacerbate the incidence of side effects [71]. To enhance the bioavailability of met for oral delivery in ADPKD, met was loaded into chitosan nanoparticles (CS-NP met), and the size and charge of CS-NP met was found to be unaltered compared to unloaded CS-NP (Table 2).

Table 2.

Size and charge of unloaded CS-NP compared to CS-NP met.

	Diameter (nm)	Zeta Potential (mv)
CS-NP	145.5 ± 8.2	+27.3 ± 2.8
CS-NP met	144.3 ± 6.6	+27.8 ± 2.1

To first test therapeutic potential in vitro, mpkCCD_c14 cells, derived from the cortical collecting duct, were treated with 300 μM met in CS-NP met, free met, unloaded CS-NP, or PBS for 12 hours and phosphorylated (active) AMPK to total AMPK ratio was measured using an ELISA assay. As shown in Figure 4c, an increase in phosphorylated AMPK to total AMPK ratio in both met-containing groups was seen: 3.0 ± 0.1 for free met and 2.1 ± 0.1 for CS-NP met (p < .005), while no change was found upon PBS and CS-NP blank treatment. While the free drug showed higher therapeutic efficacy at the same dose, we believe this is due to the slow release profile of met from CS-NP at 7.4 pH (Figure 3a), compared to the bolus effect of free met. Additionally, the effect of CS-NP met on the reduction of ENaC current, a measure of the CFTR activity, was analyzed. After 15 min, 3 hours, 24 hours, and 48 hours of CS-NP met, free met, CS-NP blank, or PBS treatment, the CS-NP met and free met groups showed a marked decrease in ENaC current in mpkCCD_c14 cells. The largest change was seen after 48 hours, with CS-NP met reaching 81.1 μA/cm² and free met at 79.2 μA/cm², while unloaded CS-NP ENaC remained at the 117.2 μA/cm² baseline (Figure 4d, p < 0.005). Taken together, these studies confirmed the therapeutic efficacy of CS-NP met was not hindered, as it produced similar AMPK activity and ENaC inhibition compared to free met. As over 90% of cells were viable upon treatment with CS-NP when assayed by MTS, the observed ENaC and AMPK changes are not due to cytotoxic effects of chitosan (Figure S3c).

3.5. Ex vivo imaging of CS-NP in vivo and intestinal localization

Next, to assess the ability of CS-NP to enhance drug bioavailability via oral delivery in vivo, 10 mg/kg rhodamine was first encapsulated into CS-NP (CS-NP R) and C57BL/6J mice were orally gavaged with 200 uL CS-NP R or free R, and after 24 hours, ex vivo imaging was conducted. As shown in Figure 5, ex vivo optical imaging demonstrated the majority of CS-NP R and free R accumulated in the intestines, liver, kidneys, and bladder, and upon quantitative analysis, CS-NP R showed 60.3 ± 11.0% of total organ fluorescence accumulation in the intestines vs. 37.4 ± 15.5% for free R (p<0.005, Figure 5a). A time course for ex vivo imaging using CS-NP R at 3 hours and 48 hours showed similar trends: at 3 hours, CS-NP R had 42 .1 ± 12.0% accumulation while free R had 45.8 ± 14.0% (p < 0.005). At 48 hours, 52.3 ± 10.1% accumulation was found for CS-NP R in the intestines whereas 38.5 ± 15.1% accumulation for free R (p < 0.005, Figure S4). Notably, serum fluorescence showed a higher area under the curve (AUC) ratio of 1.3:1 for CS-NP R compared to free R over the course of 7 days (Figure 5b), demonstrating enhanced depot to systemic circulation.

Figure 5.

Semi-quantitative biodistribution of mice treated with 10 mg/kg rhodamine in 200 uL of CS-NP R and free R 24 hours after oral gavage. (a) Comparison of ex vivo imaging between rhodamine fluorescence levels showed higher accumulation in the intestines for CS-NP R vs. free R 24 hours post-oral gavage. (b) Serum fluorescence of CS-NP R shows a greater absolute bioavailability and an extended release profile for the CS-NP formulation for up to 7 days (76.2% for CS-NP R and 47.9% for free R; ***p ≤ 0.001, N ≥ 4). (c) Representative ex vivo images confirm highest signal in the intestines in the CS-NP R condition 24 hours after oral gavage.

Upon further assessment of CS-NP R localization within the intestines via ex vivo imaging, CS-NP R was found adhered to the jejunum of the intestines (Figure 6a,b). Fluorescence microscopy of intestinal sections also confirmed higher rhodamine signal in the jejunum as well as higher colocalization of CS-NP R with mucin vs. free R (Figure 6c). This is beneficial for oral delivery as the Peyer’s patches located within the jejunum, as well as the larger surface area compared to the duodenum and ilium, are responsible for the majority of nutrient uptake, as well as facilitating nanoparticle transport into systemic circulation [118, 119]. Overall, these finding suggests that through mucoadhesion, CS-NP is retained in the jejunum which allows for sustained drug release and bioavailability Moreover, no morphological differences were found in other organs between treatment and PBS control histologically (Figure S5), confirming safety of CS-NP.

Figure 6.

Quantification of intestinal localization of CS-NP R and free R 24 hours after oral gavage. (a) Ex vivo fluorescence images and (b) quantitative comparison show the majority of CS-NP R adhered to the jejunum, while free R treatment is localized to the duodenum and ilium. (***p ≤ 0.001, N ≥ 4). (c) Alcian blue staining of mucus shows colocalization of the CS-NP R to the intestinal mucosa, demonstration mucoadhesion.

3.6. Therapeutic efficacy of CS-NP met in PKD mice

To confirm the viability of CS-NP to act as an oral delivery vehicle in chronic diseases, CS-NP met was administered in the ADPKD murine model, Pkd1^fl/fl;Pax8-rtTA;Tet-O cre [120]. In this model, a rapidly progressing PKD phenotype can be developed by knockout of the PKD1 gene, induced by doxycycline injection on P10 and P11. Then starting on P12, mice were orally gavaged with 300 mg/kg of met loaded in CS-NP met, control CS-NP, or free met every two days and euthanized on P22 when a severe cystic phenotype is expected. Kidneys were excised to assess kidney to body weight (KW/BW) ratio and stained with H&E to compare cystic index. In CS-NP met-treated mice, a greater decrease in the KW/BW ratio was found compared to free met (10.3 ± 1.1 vs. 13.1 ± 1.0, p ≤ 0.01), confirming enhanced therapeutic efficacy in slowing of cystogenesis of met when delivered via CS-NP (Figure 7a). Moreover, cystic index was statistically lower in CS-NP met-treated mice compared to mice treated with the free drug (57.6 ± 1.2% vs. 66.5 ± 0.8%, p ≤ 0.01, Figure 7b,c). A Cre-mouse serves as healthy control in which the PKD1 gene knockout is not activated and Cre-kidneys represent normal kidney morphology.

Figure 7.

In vivo therapeutic efficacy of CS-NP met. (a) A lower KW/BW ratio (10.3 ± 1.1 vs. 13.1 ± 1.0 (**p ≤ 0.01, N ≥ 4) and (b) cystic index (57.6 ± 1.2 vs. 66.5 ± 0.8; **p ≤ 0.01, N ≥ 4) was seen in the CS-NP met group vs. free drug. (c) H&E staining of whole kidneys shows less severe cystic phenotype in the CS-NP met group. A Cre-recombinase negative control is a non-diseased kidney morphology.

Although a met dose of 300 mg/kg daily has been found to activate AMPK in previous murine models [20], it is higher than what is currently prescribed for patients with diabetes (maximum 37.5 mg/kg/day). Our study administered 300 mg/kg met every two days instead, and confirmed efficacy via oral delivery that was comparable to previous IP delivery studies [20]. Future dose de-escalation studies will be conducted to examine the full benefits of CS-NP in increasing therapeutic efficacy without compromising safety. Regarding renal biocompatibility, kidney health markers including blood urea nitrogen, creatinine, and serum electrolytes were found to remain similar between treatment groups, demonstrating CS-NP formulations do not cause kidney damage (Table 1). The BUN levels correspond to mildly impaired renal function expected in polycystic kidney mice, on the order of 40–80 mg/dL [121]. In sum, CS-NP demonstrated a higher therapeutic efficacy when compared to free drug at the same dose, and is a safe platform that can overcome the physiological barriers of oral delivery. Uniquely, this is the first nanoparticle delivery platform for ADPKD, and our study highlights CS-NPs as a viable oral delivery platform for chronic conditions.

Table 1.

Serum components, electrolytes, and kidney health markers for CS-NP met, free met, and CS-NP treated mice show no significant difference between groups. Measured values include sodium (Na), potassium (K), chloride (Cl), ionized calcium (iCa), total carbon dioxide (tCO₂), glucose (Glu), blood urea nitrogen (BUN)/Urea, creatinine (Crea), hematocrit (Hct), hemoglobin (Hb), and anion gap (AnGap).

	Na [mmol/L]	K [mmol/L]	Cl [mmol/L]	iCa [mmol/L]	tCO2 [mmol/L]	Glu [mmol/L]	BUN/Urea [mg/dL]	Crea [mg/dL]	Hct [PCV]	Hb [g/dL]	AnGap [mmol/L]	Weight [g]
CS-NP met	142.7 ± 1.8	4.5 ± 0.5	109.0 ± 1.5	1.1 + 0.1	24.2 ± 1.5	148.5 ± 19.7	71.2 ± 22.5	< 0.2	23.7 ± 3.8	8.0 ± 1.2	15.2 ± 2.8	9.5 ± 1.1
Free met	142.3 ± 2.6	5.1 ± 1.1	110.0 ± 3.1	1.1 ± 0.1	24.6 ± 2.7	130.5 ± 35.8	79.50 ± 22.7	<0.2	26.5 ± 1.8	9.0 ± 0.6	14.8 ± 2.9	7.3 ± 1.9
CS-NP	142.1 ± 3.6	3.1 ± 0.9	111.0 ± 2.1	1.2 ± 0.1	24.1 ± 2.3	136.5 ± 33.6	75.50 ± 2.7	<0.2	24.5 ± 1.8	8.9 ± 0.6	14.3 ± 2.3	7.4 ± 1.6

4. Conclusion

Chitosan nanoparticles (CS-NP) were investigated as a promising drug delivery platform for oral delivery in chronic kidney disease. CS-NPs were synthesized through ionic gelation and their physiochemical properties were characterized. In vitro, CS-NPs demonstrated effective mucoadhesion, while protecting premature release of the payload in the low pH environment of the stomach. When met-loaded CS-NP were cultured with cells in vitro, therapeutic efficacy was found via AMPK activation and ENaC current reduction. Moreover, upon oral gavage in a murine model of PKD, disease burden was significantly reduced upon met delivery using CS-NP compared to the free drug. While free met at the dosages used in the study did not cause significant toxicities, future studies assessing high met dosages or increasing dose exposure in the slowly developing mouse model of PKD that more closely mimics the chronic nature of the human disease will more fully elucidate the benefits of delivering drugs in the CS-NP system. Furthermore, the observed advantage of CS-NP in increasing systemic delivery may also be more evident upon loading candidate ADPKD drugs with poor oral bioavailability, such as somatostatin or bardoxolone methyl. Our study provides the framework to advance chitosan nanotechnology for PKD; future studies will include additional animal models including slowly progressing PKD models to further mimic the chronic nature of PKD, as well as large porcine animal models as we look towards clinical translation.