Enhancing oral delivery of plant-derived vesicles for colitis

Abstract

Plant-derived vesicles (PDVs) are attractive for therapeutic applications, including as potential nanocarriers. However, a concern with oral delivery of PDVs is whether they would remain intact in the gastrointestinal tract. We found that 82% of cabbage PDVs were destroyed under conditions mimicking the upper digestive tract. To overcome this limitation, we developed a delivery method whereby lyophilized Eudragit S100–coated cabbage PDVs were packaged into a capsule (Cap-cPDVs). Lyophilization and suspension of PDVs did not have an appreciable impact on PDV structure, number, or therapeutic effect. Additionally, packaging the lyophilized Eudragit S100-coated PDVs into capsules allowed them to pass through the upper gastrointestinal tract for delivery into the colon better than did suspension of PDVs in phosphate-buffered saline. Cap-cPDVs showed robust therapeutic effect in a dextran sulfate sodium-induced colitis mouse model. These findings could have broad implications for the use of PDVs as orally delivered nanocarriers of natural therapeutic plant compounds or other therapeutics.

Introduction

Many edible plants are known to have compounds that can be beneficial for many diseases, including inflammatory diseases and cancer. However, consuming a sufficient quantity of plants to get an effective dose of such compounds (e.g., phytochemicals) may not be feasible because many vegetables are composed of >90% water, ~5% carbohydrates, and ~2.5% fiber[1]. Moreover, variability in preparation of plant-based meals (e.g., canned, blended, cooked) may further affect the availability and loss of beneficial compounds[1]. Therefore, achieving a biological effect with the use of a plant-based diet alone may be challenging.

Recently, it is shown that plant-derived vesicles (PDVs) can be successfully isolated in large quantities and at a low cost. Some PDVs have been shown to have intrinsic therapeutic benefits, especially the amelioration of inflammatory conditions. These benefits include improving the maintenance of intestinal stem cells and shaping the gut microbiota that enhance the gut barrier function to alleviate colitis[2]. Thus, PDVs are an attractive potential treatment strategy for bowel-related diseases.

Although PDVs have been proposed as potent therapeutics, convenient and effective administration methods have yet to be established. Researchers have suspended PDVs in phosphate-buffered saline (PBS) for gavage administration in mice[3, 4]; however, whether PDVs can survive the hostile environment of the upper digestive system is not well understood. Indeed, PDV particle size decreases with exposure to stomach and intestinal pH[4]. To fully understand the effect of the digestive system on PDVs and identify a reliable delivery system, we carried out a series of in vitro and in vivo experiments to characterize the effects of the upper digestive system on PDVs and developed a capsular formulation of Eudragit S100–coated PDVs for reliable applications.

Methods

Cabbage, mushroom, spinach, and ginger PDV isolation.

Whole cabbage leaves were washed sequentially with 10% Clorox bleach and diluted antibacterial soap; the leaves were then thoroughly rinsed with tap water. The same disinfecting and cleaning process was applied to all blenders, containers, and centrifuge tubes used during the isolation of vesicles. The cabbage leaves were blended with a Ninja blender (Model BL610) for about 2 minutes to generate tiny leaf slices. The tiny leaf slices were then subjected to a cold press juicer (Kuhaus) and filtered through a mesh nylon filter bag (Groupcow) with a pore size of 8 × 12 μm. The resultant cabbage juice was centrifuged sequentially at 700 × g for 10 minutes and then 2,000 × g for 20 minutes at 4° C with an Eppendorf benchtop centrifuge (Model 5810), rotor A-4-81. The remaining tissue fragments and large cellular debris were removed using a Beckman Optima XE Ultracentrifuge (rotor Ti45) at 4° C, 10,000 × g for 30 minutes. The remaining supernatant was centrifuged at 100,000 × g for 2 hours to obtain a pellet. The pellets were suspended in PBS and then centrifuged at 4° C at 10,000 × g for 30 minutes to remove the aggregated PDVs and contaminates. The supernatant was subjected to a final centrifugation at 4° C, 100,000 × g for 2 hours. The pellets were resuspended in PBS (Cytiva) as the final product, filtered through a 0.22-μm syringe filter unit (Millipore), and stored at −80° C. The method schema is shown in Supplementary Fig. 1. Spinach and mushroom PDV isolation followed this method; ginger PDV isolation followed that described in a previous publication[3].

Untargeted and targeted liquid chromatography-mass spectrometry (LC-MS).

Untargeted LC-MS analysis was performed on a Q Exactive Plus Orbitrap Mass Spectrometer (Thermo Fisher Scientific) coupled to a binary pump HPLC (UltiMate 3000; Thermo Fisher Scientific). Full MS spectra were obtained at 70,000 resolution (200 m/z) with a scan range of 50-750 m/z. Full MS followed by ddMS2 scans were obtained at 35,000 resolution (MS1) and 17,500 resolution (MS2) with a 1.5-m/z isolation window and stepped normalized collision energy (20, 40, 60). Samples were maintained at 4° C before injection. The injection volume was 10 μL. Chromatographic separation was achieved on a Synergi Fusion 4-μm, 150-mm × 2-mm reverse-phase column (Phenomenex) maintained at 30° C using a solvent gradient method. Solvent A was water (0.1% formic acid). Solvent B was methanol (0.1% formic acid). The gradient method used was 0-5 minutes (10% B to 40% B), 5-7 minutes (40% B to 95% B), 7-9 minutes (95% B), 9-9.1 minutes (95% B to 10% B), and 9.1-13 minutes (10% B). The flow rate was 0.4 mL/minute. Sample acquisition was performed using Xcalibur software (Thermo Fisher Scientific). Data analysis was performed with Compound Discoverer 3.1 (Thermo Fisher Scientific). For statistical analysis of LC-MS compounds, three biological replicates were tested, and differences in relative abundance between these three replicates were expressed as ranges.

Mass spectrometry for proteomics.

Cabbage PDVs were dissolved in 50 mM ammonium bicarbonate containing 1 mM calcium chloride to prepare lysate. The samples were snap-frozen in liquid nitrogen and thawed twice followed by boiling at 95° C for 2 minutes with a short vortex every 30 seconds. The lysate was digested using LysC+ trypsin proteases at 37° C overnight. The enzyme reaction was neutralized using 10% formic acid and the peptides were measured using the Pierce Quantitative Colorimetric Peptide Assay (Thermo Scientific). The tryptic peptides were subjected to a simple C18 cleanup using a C18 disk plug (3M Empore C18) and extracted using 50% acetonitrile containing 0.1% formic acid, then dried in a speed vac (Savant). The samples were analyzed as two technical replicates using a nano-LC 1200 system (Thermo Fisher Scientific) coupled to Orbitrap Exploris 480 (Thermo Fisher Scientific) mass spectrometer. Then, 2 μg of peptide was loaded on a pre-column of 2 cm × 100 μm I.D. switched in-line with an in-housed 20 cm × 75 μm I.D. column (Reprosil-Pur Basic C18, 1.9 μm, Dr. Maisch GmbH, Germany). The peptide elution was done using a 110-minute discontinuous gradient of 90% acetonitrile buffer (B) in 0.1% formic acid at 200 nL/minute (2-30% B: 86 minutes, 30-60% B: 6 minutes, 60-90% B: 8 minutes, 90-50% B: 10 minutes). The eluted peptides were directly electro-sprayed into an Orbitrap Exploris 480 (Thermo Fisher Scientific) mass spectrometer operated in the data-dependent acquisition mode acquiring HCD fragmentation spectra of the top 50 strongest ions. The full MS scan was acquired in Orbitrap in the range of 350-1400 m/z at 120,000 resolution followed by MS2 at 15,000 resolution (HCD 32% collision energy) with 15-second dynamic exclusion time. The MS raw files were searched using Proteome Discoverer 1.4 software (Thermo Fisher) with the Mascot algorithm (Mascot 2.4, Matrix Science; percolator against the Brassica oleracea var. oleracea protein database from NCBI refseq). The peptides identified from the Mascot result file were validated with 5% false discovery rate. Dynamic modification of oxidation, protein N-terminal acetylation, and deamidation on asparagine and glutamine was allowed. The precursor mass tolerance was confined within 20 ppm, with fragment mass tolerance of 0.02 Dalton, and a maximum of two missed cleavages with trypsin enzyme was allowed.

PDVs treated in gastrointestinal (GI) tract–mimicking conditions.

The cabbage PDV stock concentration was 1.0 × 10¹³ particles/mL; 80 μL of 1:100 diluted PDVs was added to 3.20 mL of simulated stomach acid (NaCl 2.1 g/L, 0.155M HCL, pH = 1.0–2.0). Samples were then incubated for 30 minutes with light shaking at 37° C. After the incubation, 1 mL of the PDVs in stomach acid were mixed with 60 mg sodium bicarbonate (NaHCO3) to neutralize pH to 7.0 – 7.4. The same process was performed for cabbage PDVs in the presence of digestive enzymes pepsin (Sigma, 1.0 mg/mL in stomach acid pH = 1.0 - 2.0), pancreatin (Sigma, 2.0 mg/mL in PBS, pH = 7.4), or pepsin then pancreatin (Sigma, stomach acid pH = 1.0 - 2.0 then neutralized stomach acid, pH = 7.0–7.4) [5]. Samples were visualized by transmission electron microscopy (TEM) and nanoparticle tracking analysis (NTA).

Eudragit S100–coated PDVs stock concentration was 1.0 × 10¹² particles/mL; 80 μL of 1:100 diluted PDVs was added to 3.20 mL of simulated stomach acid (NaCl 2.1 g/L, 0.155M HCL, pH = 1.0–2.0) or pepsin (Sigma, 1.0 mg/mL in stomach acid, pH = 1.0–2.0). Samples were then incubated for 30 minutes at 37° C with light shaking and then visualized by TEM.

NTA and Z-average size analysis.

The size and number of nanoparticles were determined using NTA, performed with a NanoSight NS500 instrument (Malvern) with a green laser 532, SCMOS model. Details of the analysis were described in a previously published paper[6]. Samples were diluted 1:10,000 in PBS, resulting in each view having 20-100 particles. Three sequential standard measurements were performed for 1 minute each at 15.8°–15.9° C (viscosity: 1.108-1.111 cP). All samples used a camera level of 13, a detection threshold of 6, and a screen gain of 10. Measurements were done in triplicate.

Z-average size was assayed by dynamic light scattering (Zen 3600 Zetasizer Malvern Instruments) operating at a wavelength of 633 nm, a scattering angle of 173°, refractive index of 1.34, and viscosity 1.0568 (25° C). Measurements were performed three times, with similar results.

Cryogenic electron microscopy (Cryo-EM).

Cryo-EM studies were conducted according to procedures described in detail in a previously published article[6]. Briefly, 3-μL PDV samples were placed on a glow-discharged Lacey carbon 400-mesh Cu grid, after which the grid was plunge-frozen using a Vitrobot specimen-preparation unit (Thermo Fisher Scientific). The freezing conditions were as follows: 100% humidity, 4° C, 5-second wait time, 2-second blot time, and 0 blot force. Micrographs were collected on a FEI Titan Krios operating at 300 kV coupled with a Gatan K2 direct electron detector via Serial-EM software (Mastronarde 2005). Each exposure image was collected at 18,000 × nominal magnification, resulting in a pixel size of 1.36 Å/pixel in the counted mode, using a dose rate of 13 e–/pixel per second and 200 ms exposure per frame, for a total of 40 frames. The total dose in the EM data collection was 55 e–/Ų. The nominal defocus used was −2.0 μm.

For statistical analysis of cryo-EM data, we took 49 images from random spots. PDVs were counted, and intact and broken PDVs were identified. We set the control PDV group as 100% and compared the number of vesicles at pH = 1.5 (treatment) with that in the untreated group to determine the percentages.

Lyophilization.

Cabbage PDVs were suspended in 25mM D-trehalose (Sigma) before lyophilization to avoid disruption from the higher osmotic pressure caused by the increase in salt concentration during the lyophilization process[7]. After isolation, PDVs were preserved by directly freezing them overnight (about 16 hours) in cryopreservation tubes wrapped in heat-insulating material at −80° C. The mixture was vortexed, and 2 mL of the mixture was placed in a borosilicate glass vial to be frozen in an acetone dry ice bath (−78° C). The vials were tilted during freezing to uniformly freeze the sample on the vial walls and facilitate the freeze-dry process. Lyophilization was performed with a Labconco dry system FreeZone 4.5 plus, according to the manufacturer’s guidelines. Lyophilized powder sample was scraped from the tube walls and then centrifuged at 10,000 × g for 10 minutes to gather the powder at the bottom of the tube.

Negative staining and TEM.

Negative staining and TEM were performed according to a method described previously[6]. Briefly, formvar/carbon-coated 100-mesh copper grids (Sigma) were treated with poly-L-lysine for approximately 1 hour and then PDV samples were placed on the grids. Samples were negatively stained with Millipore-filtered aqueous 1% uranyl acetate for 1 minute. The stain was blotted dry from the grids with filter paper, and the samples were allowed to dry. The samples were then imaged using a JEM 1010 TEM (JEOL, USA, Inc) at an accelerating voltage of 80 kV. Digital images were obtained using the AMT Imaging System (Advanced Microscopy Techniques Corp).

Capsule filling and storage.

Capsules (size A-CT M; Braintree Scientific Inc) were filled using a supplied funnel kit (Capsule Filling Funnel, Braintree Scientific Inc), avoiding static electricity during the entire filling operation.

Coating of cabbage PDVs with Eudragit S100.

Under magnetic stirring at 900 rpm, 3 mL of Eudragit S100 (Evonik) solution (2 mg/mL) was slowly added to 3 mL of ~1.0 × 10¹³ PDVs suspension[8], stirring was continued for 12 hours in 4° C. Lyophilization and encapsulation was performed afterwards.

Eudragit S100 coating efficiency on cabbage PDVs.

Eudragit S100 was labeled with sulfo-Cy3-amide-C6-amine following previous research[9]. 0.1 g of Eudragit S100 and 3.86 mg of N,N,N’,N’-tetramethyl-O-(N-succinimidyl) uronium tetrafluoroborate (ThermoFisher) was dissolved in 2.5 mL of dimethylformamide, and 4 μL of N,N-diisopropylethylamine (ThermoFisher) was added and stirred for 2 hours at 27° C for activation. Next, 1 mg of sulfo-Cy3-amine dye (Broadpharm, 0.0012 mmol, Mw: 867) was added and stirred overnight at 27° C. Particles were washed with deionized water and dried with a vacuum machine. We dissolved this product in 0.5 mL of dimethylformamide, then purified by Sephadex LH-20 column. Cy3-labeled Eudragit S100 conjugate fractions were collected and dried with a vacuum machine. Cy3-labeled Eudragit S100 was used to coat CellMask Deep Red dyed PDVs. Incorporation efficiency was assessed via Cytek Aurora system.

Mouse experiments.

Female C57BL/6 mice aged 4-8 weeks were purchased from The Taconic Bioscience. All mice were housed at The University of Texas MD Anderson Cancer Center animal facility under specific pathogen-free conditions. All animal-related experiments were approved by the Institutional Animal Care and Use Committee of MD Anderson.

Positron emission tomography (PET) imaging.

PET scanning methods followed previous research[10]. PDVs were reacted with amine-reactive NOTA (p-SCN.Bn.NOTA; Fisher) with a weight ratio of 10:1 (weight of proteins in the PDVs/weight of NOTA) at pH = 8.5 for 2 hours, then purified with PD SpinTrap G-25 desalting column (Cytiva) using PBS (without calcium and magnesium) to remove free NOTA. Next, ⁶⁴CuCl₂ (74 MBq, MD Anderson Cyclotron Radiochemistry Facility) was diluted in 300 μL of 0.1M sodium acetate buffer (pH = 5.5) and mixed with PDV-NOTA at 37° C for 30 minutes with constant shaking. The resulting ⁶⁴Cu-NOTA-PDVs were purified by PD-10 desalting columns (Cytiva) using PBS as the mobile phase. The radiolabeling stabilities were determined by incubating each with 25% mouse serum for 24 hours. The radiolabeling yield and stability were tested with thin layer chromatography using 50mM EDTA (pH = 5.5) as the mobile phase (n = 3), measured by the Bioscan AR-2000 Radio-TLC Imaging Scanner (Eckert & Ziegler).

Mice were given oral gavage with ⁶⁴Cu-NOTA-PDVs (50 μCi/mouse, 200 μL), and then serial PET scans were performed on an Albira PET/SPECT/CT scanner (Bruker) at 3 hours, 24 hours after gavage. Quantitative data from region-of-interest analysis on tumors and other organs were presented as the percentage dose per gram of tissue (%ID/g). After the final scan at 24 hours after gavage, organs were collected and weighed. Radioactivity was measured using PerkinElmer (Packard) Cobra II Gamma counter (PerkinElmer).

Mouse oral gavage with capsules.

Capsules were loaded into the tip of our lab-modified capsule gavage needle, and the tip was lubricated with petroleum jelly (Supplementary Fig. 2A). The capsule was placed into the esophagus of anesthetized mice (Supplementary Fig. 2B) by gavage (Supplementary Fig. 2C), followed by about 0.1 mL of water. Each mouse was monitored in a closed container with black tissue paper underneath (Supplementary Fig. 2D) until the mouse was fully awake. If the capsule was spit out, it was administered to the mouse again. The parallel treatment groups were subjected to anesthesia to avoid bias.

PDV dose.

The weight of lyophilized PDVs per capsule was 1.6 mg, total number of particles was ~10¹², and protein quantity was 0.5 mg. PDVs were dosed based on findings that 3.3 mg/kg indole-3-carbinol showed biological effectiveness in prior clinical and pharmacokinetics studies[11, 12]; 1.6 mg PDVs contain ~70 μg indole-3-carbinol. Each mouse was dosed once a day with 1.6mg PDVs at an equivalent indole-3-carbinol of 3.3 mg/kg.

We used 2.5% DSS to induce colitis, which represents a higher dose compared with that used by some previous studies (e.g., 1.5% DSS); this could have resulted in more severe colitis (Supplemental Figure 9)[3]. This is one of the reasons why we administered a higher dose of cabbage PDVs compared to the dose that was used with oral ginger PDVs. We recognize that it is hard to directly compare doses of different species of PDVs since there are likely different small compound combinations and macromolecules, such as small RNAs, DNAs and proteins.

Biodistribution.

C57BL/6 mice were used for biodistribution experiments. Three conditions were prepared: 1.6 mg of dyed PDVs suspended in PBS, 1.6 mg of dyed lyophilized PDVs loaded into capsules (Cap-PDVs), and 1.6 mg of dyed lyophilized PDVs coated with Eudragit S100 and loaded into capsules (Cap-cPDVs), and each condition was tested in 3 mice. Cabbage PDVs were dyed with CellMask Deep Red (Thermo Fisher Scientific). To dye PDVs, CellMask Deep Red was added to each sample at a 1:1,000 dilution, then incubated at 37° C for 30 minutes. The samples were then centrifuged at 4° C, 100,000 × g for 2 hours. The pellets were resuspended in 4 mL of PBS to wash away residual free CellMask dye and then recentrifuged at 100,000 × g for 2 hours. The washed pellets were resuspended in 400 μL of PBS or 400 μL of 25mM D-trehalose. Six hours after PDV administration, the mice were euthanized and the total GI system was imaged using the Xenogen IVIS-200 in vivo imaging system. The tissues were then placed in optimal cutting temperature embedding media, and frozen sections were generated for microscopy. Tissue sections were placed in cold acetone for 10 minutes and transferred to PBS for three washes. We first fixed the slice with 4% paraformaldehyde (Electron Microscopy Science) for 15 minutes followed by three PBS washes. Then, 4',6-diamidino-2-phenylindole (Perkin Elmer) working solution was added to each slice. After the 15-minute staining period, the slice was washed three times with PBS. Images were acquired with the Zeiss LSM 800 microscope.

Cabbage PDVs in a dextran sulfate sodium (DSS)-induced colitis model.

C57BL/6 background mice were used for these experiments. Four groups of mice were tested (n=5 per group). Treatment groups were 1) drinking water (negative control); 2) 2.5% DSS (MP Biomedicals) in water and treated with 100 μL PBS (DSS+PBS); 3) 2.5% DSS water and treatment with cabbage PDVs suspended in PBS (DSS+PDVs); and 4) 2.5% DSS water and treatment with Cap-PDVs or Cap-cPDVs (DSS+Cap-PDVs or DSS+Cap-cPDVs). In all DSS-induced mice, 2.5% DSS in drinking water was provided for 7 days to establish colitis; treatments were started along with DSS treatment (during the 7 days of DSS treatment). For PDVs in PBS, Cap-PDVs, and Cap-cPDVs, 1.6 mg of PDVs daily was used for each treatment. Mice were euthanized on day 13 for Figure 3; day 14 for Figure 6. Mice were euthanized at day 8 for Supplementary Fig 9 to confirm that colitis was established. Colon length was determined with freshly excised tissue. One-third of each tissue sample was analyzed by hematoxylin and eosin staining and imaging. Another third was used for RNA isolation (RNeasy Mini Kit from Qiagen) to assess mRNA levels of interleukin (IL)-1β, IL-6, and tumor necrosis factor alpha (TNF-α) transcripts using real-time PCR (Applied Biosystems); 36B4 was used as an endogenous control gene. Relative expression level was compared with samples from control tissue.

Fig.3∣. Cabbage PDV in PBS and Cap-PDVs therapeutic effect to colitis.

A, Average mouse body weight after treatment with PDVs in PBS (DSS+PDV) and Cap-PDVs (DSS+Cap-PDV) with DSS+PBS and control group, error bars indicate SEM. Mice were euthanized on day 13. B, Colon image and length after treatment. C, Hematoxylin and eosin staining (H&E) and histoscore after treatment (scale bars=100μm). D, Expression levels of cytokines in the colon relative to 36B4 after treatments (dose=1.6mg; n=5 per group). Error bars indicate SD in B-D.

Fig.6∣. Cap-cPDVs therapeutic effects on DSS-induced colitis compare with PDVs suspended in PBS.

A, Average mouse body weight after treatment with DSS+PDVs and DSS+Cap-cPDVs, error bars indicate SEM. B, Colon length comparison with control, DSS, DSS+PDVs,and DSS+ Cap-cPDVs. C, Histocore and H&E comparison between the four groups (scale bars=100μm). D, Cytokine expression levels in the colon relative to 36B4 after treatments (dose was 1.6mg; n=5 per group). B-D Error bars indicate SD.

Histologic assessment.

Histologic assessment was performed using criteria for colitis evaluation from the pathology center in the Department of Veterinary Medicine and Surgery at MD Anderson adapt from ref (Supplementary Table 1)[13].

Statistical analysis.

The operator was blinded to the experimental groups for statistical analyses. Data were analyzed by Student t test or two-way analysis of variance using GraphPad Prism 8. p values >0.05 were considered nonsignificant (ns) and p ≤ 0.05 was designated as *, p ≤ 0.01 as **, p ≤ 0.001 as ***, and p ≤ 0.0001 as ****.

Results and conclusions

PDVs in the digestive system.

We first carried out a series of experiments to determine the effect of conditions that mimic some aspects of the GI tract on PDVs isolated from cabbage. TEM and NTA showed that the number of PDVs did not change substantially following a 30-minute exposure to 37° C simulated stomach acid (pH ≈ 1.5) compared with 37° C PBS at pH 7.5; however, the number of PDVs was reduced by −30% when PDVs were incubated in stomach acid that was then neutralized to the pH of the proximal small bowel (Fig. 1A, B). To detect any vesicle shape changes due to the stomach’s acidic pH, we performed cryo-EM, and the results showed that most PDVs suspended in control PBS (pH = 7; n = 170) were intact and round; only 8.9% of these PDVs were not completely sealed (Fig. 1C). When PDVs were exposed for 30 minutes to 37° C simulated stomach acid at pH = 1.5, the total number of PDVs was reduced by only 8.8%, but PDV shape was changed substantially, from round to more irregular shapes, and 36.1% of PDVs appeared broken (n = 155; Fig. 1C). TEM images of PDVs following exposure to pepsin (pH = 1.5) showed that the number of vesicles was reduced to 21% of that in the control group, and most of the PDVs were broken (Fig. 1D). The number of intact PDVs was reduced to only 18% of the original number upon treatment with pepsin followed by exposure to pancreatin (in neutralized stomach acid at pH = 7.5); most of the PDVs were digested into small particles (Fig. 1D). The number of intact PDVs was reduced to 74% of the original number upon treatment with pancreatin (Supplementary Fig. 3). These findings strongly suggest that most of the PDVs would be damaged as they pass through the stomach and proximal small bowel environment.

Fig.1 ∣. Plant-derived Vesicles (PDVs) exposed to conditions mimicking the gastrointestinal (GI) system.

A, Transmission electron microscopy (TEM) imaging and quantification of PDVs exposed to different GI pH mimicconditions as indicated (red arrowheads indicate broken vesicles; n=3). B, Nanoparticle tracking analysis (NTA) showing the number of particles in samples exposed to changes in pH levels mimicking the GI tract (n=3). C, Cryo-EM comparative analysis of PDVs exposed to pH 7.5 versus pH 1.5 for 30 minutes at 37 °C. Red arrowheads show damaged PDVs. Bars=50nm, n=49. D, TEM imaging and quantification of PDVs exposed to changes in GI pH and digestive enzymes. Red arrow heads indicate broken and destroyed PDVs. Error bars indicate SD (n=3; scale bars=50nm).

Compounds of interest in PDVs.

For proof-of-concept studies, we considered several plant species and focused on those with known anti-inflammatory properties. Among the most consumed vegetables, cabbage, mushroom, spinach, and ginger have been shown to contain anti-inflammatory substances[14-17]. To identify whether compounds of interest could be detected in PDVs, we performed LC-MS on PDVs derived from cabbage, mushroom, spinach, and ginger. In these experiments, 19 anti-inflammatory compounds were found in cabbage PDVs, 15 in mushroom PDVs, 14 in spinach PDVs, and 8 in ginger PDVs (Supplementary Table 2). Anti-inflammatory compounds found in cabbage PDVs included the glucosinolate-related metabolites sulforaphane and indole-3-carbinol, cruciferous vegetable biomarker S-methyl-L-cysteine-S-oxide, jasmonic acid precursor 12-oxo-phytodienoic acid, trigonelline, and many amino acids (Supplementary Table 2, Supplementary Fig. 4A). Indole-3-carbinol and choline were further verified by standards for comparison; 10¹³ PDVs contain ~0.7 mg Indole-3-carbinol (Supplementary Fig. 4B). Furthermore, the yield of cabbage PDVs was highest among the four plants studied (Supplementary Fig. 5), cabbage was selected for further investigation. Mass spectrometry proteomic analyses were performed to further characterize cabbage PDVs (Supplementary Table 3). These results showed that cabbage PDVs have the same protein marker as mammalian extracellular vesicles (Supplementary Fig. 6)[18].

Formulation of cabbage PDVs for oral administration.

Given the extensive destructive effects of the GI tract on PDVs, we next focused on developing a strategy to protect the PDVs for successful oral delivery. To accomplish this goal, we lyophilized PDVs and loaded them into capsules (Cap-PDVs). TEM and NTA showed that the number and shape of PDVs were similar after lyophilization and resuspension. Resuspension after lyophilization caused a slightly increase the particle size, as in lyophilized extracellular vesicles (Fig. 2, Supplementary Fig. 7)[19]. For capsule administration in mice, we used a steel tube inside soft polyurethane feeding tubes (13 Ga × 88 mm; Supplementary Fig. 2A). For oral delivery of Cap-PDVs in mice, we used anesthesia; a detailed description is included in the Methods section (Supplementary Fig. 2B-D).

Fig. 2∣. Comparison of structure and number of PDVs suspended in phosphate-buffered saline (PBS) to PDVs subjected to lyophilization followed by resuspension in PBS.

A, TEM, n=3. B, NTA, Error bars indicate SD (n=3; scale bars=50nm).

Cabbage PDVs alleviate colitis.

To assess the biodistribution of orally delivered PDVs, we used ⁶⁴Cu-NOTA labeled cabbage PDVs for PET imaging. The PDVs administered by gavage were predominantly detectable in the stomach, colon, and liver (Supplementary Fig. 8).

Next, to evaluate the potential therapeutic application of Cap-PDVs, we used a mouse model of DSS-induced colitis. Different concentrations of DSS for inducing colitis were tested, using 2%, 3%, or 4% DSS supplied in drinking water for 7 days. The body weight of mice dropped by 7-22%, depending on the dose, by day 8 (Supplementary Fig. 9A). Consistent with induction of colitis, the colon length was significantly reduced in all three treatment groups (Supplementary Fig. 9B). Histologic evaluation showed different colitis scores within the different concentrations (Supplementary Fig. 9C); 2.5% DSS was selected for subsequent experiments.

Next, we sought to determine whether treatment with cabbage PDVs could reduce colitis in mice subjected to DSS. Mice were divided into four groups: 1) drinking water (control); 2) DSS+PBS (negative control); 3) DSS+PDVs (suspended in PBS); and 4) DSS+Cap-PDVs. Body weight and colon length in the DSS+PDVs and DSS+Cap-PDVs groups showed alleviation of colitis symptoms compared with the DSS+PBS group (Fig. 3A, B). Histologic assessment showed only modest improvement in loss of goblet cells, neutrophil infiltration in the lamina propria, mixed leukocytes in the submucosa, and crypt necrosis with suppurative inflammation in the treatment groups compared with the control groups. The total histoscore was not significantly different between the DSS+PBS, DSS+PDVs, and DSS+Cap-PDVs groups. Levels of pro-inflammatory cytokines IL-1β, IL-6, and TNF-α showed modest improvement in the treatment groups compared with the control group. There was some variability in cytokine levels between the DSS+PDVs and DSS+Cap-PDVs groups, but the difference was not statistically significant (Fig. 3C, D). The modest effects of PDV treatment may reflect the high likelihood that some PDVs may not remain intact through the upper GI tract.

To enhance the delivery of intact PDVs to the colon, we considered Eudragit L100 and S100 coatings because they dissolve above pH 6, which is desirable for delivery into the colon[20-23]. We compared the biodistribution of lyophilized PDVs in uncoated capsules with that of lyophilized PDVs in capsules coated with Eudragit L100 or S100. Colonic delivery was significantly higher with Eudragit S100 coating (Supplementary Fig. 10). Therefore, we selected the Eudragit S100 for further assessment. Because mice have slower gastric emptying, the transit of capsules in the GI tract is also slower[24]. Therefore, we considered coating the PDVs with Eudragit (cPDVs) to protect them from upper GI tract contents.

TEM images of the PDVs revealed that the average size of cPDVs was approximately 100 nm before coating and 110 nm after coating, and the number of vesicles after coating did not change (Fig. 4A). NTA showed similar results (Fig. 4B). This size increase is similar to that observed with other Eudragit-coated particles[23]. Eighty-one percent of PDVs were coated by Eudragit S100 (Fig. 4C). Seventy-one percent of the cPDVs were intact following exposure to pepsin (pH = 1.5) (Fig. 4D). In contrast, the number of vesicles was reduced to 21% in the uncoated PDVs group (Fig.1D); these findings indicate that cPDVs survive better in conditions that mimic the upper GI tract than do uncoated PDVs (Fig. 4D).

Fig.4∣. Characterization of PDVs coated with Eudragit S100 (cPDVs).

A, TEM imaging and quantification of PDVs after Eudragit S100 coating. Error bars in SD (n=3; scale bars=100nm). B, NTA for size and particle number changes after Eudragit S100 coating. Error bars in SD (n=3).C, Cytek Aurora flow cytometry testing Eudragit S100 coating efficiency on PDVs. D, TEM imaging and quantification of cPDVs exposed to upper GI pH level and pepsin. Error bars indicate SD (n=3; scale bars=100nm).

We compared optical coherence tomography slices and fluorescence intensity in a region of interest within the stomach, small intestine, and colon between PDVs in PBS, Cap-PDVs, and Eudragit S100–coated PDVs were lyophilized and loaded into capsules (Cap-cPDVs). These results showed that Cap-cPDVs had greater delivery to the colon than PDVs in PBS and Cap-PDVs (Fig. 5).

Fig.5∣. Tissue and organ distribution of PDV in PBS compared with Cap-PDVs and Cap-cPDVs.

A. Blue fluorescence 4′,6-diamidino-2-phenylindole (DAPI) staining shows nuclei, Red fluorescence shows CellMask Deep Red PDVs (scale bars=5μm). B. IVIS imaging, for a does of 1.6mg (n=3).

For testing the effects of Cap-cPDVs in the DSS-induced colitis model, mice were divided into four groups: 1) drinking water (control); 2) DSS+PBS; 3) DSS+PDVs (suspended in PBS); and 4) DSS+Cap-cPDVs. The DSS+PDVs and DSS+Cap-cPDVs groups showed alleviated symptoms of colitis, reflected by body weight and colon length. Histologic assessment showed that mice treated with Cap-cPDVs reached near complete resolution of all assessed microscopic features of colitis compared with the control group (Fig. 6A-C). These results were further confirmed through an analysis of the expression of IL-1β, IL-6, and TNF-α cytokines (Fig. 6D).

Collectively, our findings indicate that Cap-cPDVs were effectively delivered to the colon following oral intake and resulted in robust activity in reducing colitis. This method thus represents a potential new approach for effective delivery of PDVs.

Discussion

Anti-inflammatory (e.g., 5-amino salicylic acid, steroids) or immunosuppressive drugs have been used for the treatment of colitis over the past decade[25, 26]. Despite the effectiveness of these drugs, their nonspecific effects on the immune system can result in undesirable short- and long-term side effects. Targeted therapeutic approaches (e.g., Toll Like Receptor 4) are also being used, but these can result in serious side effects[27]. Thus, the development of targeted therapeutics with low toxicity represent an unmet need.

Plants contain many beneficial phytochemicals for the treatment of colitis, but it is difficult to consume adequate amounts of such compounds with raw plants. PDVs are isolated from plant juice that contains substantial amounts of phytochemicals and thus PDVs could enable delivery of desirable quantities of such compounds. Ginger-derived nanoparticles were found to have effects on colitis and colitis-associated cancer[3]. Grape-, broccoli-, and grapefruit-derived nanoparticles have also been studied for biomedical applications, especially for inflammatory bowel disease[4, 28, 29].

However, a major concern for oral consumption of PDVs is their possible breakdown in the harsh environment of the upper GI tract[30]. We found that about 80% of PDVs were broken down in conditions that mimic the upper GI tract, similar to human salivary extracellular vesicles[31]. Even though about 80% of the PDVs were destroyed in the upper gastrointestinal tract, a portion of the anti-inflammatory compounds in PDVs may still be delivered to the colon, which could explain some reduction in the colitis characteristics observed. However, compared to the DSS+PDVs group, the DSS+Cap-cPDVs group showed improvements in epithelial cell death, goblet cell loss, and crypt architecture alterations induced by DSS. In addition, body weight, colon length and cytokines in the DSS+Cap-cPDVs group are improved compared to the DSS+PDV group. Overall, the DSS+Cap-cPDVs group had better therapeutic effects than the DSS+PDV group; these results suggest that additional benefit may derive from intact cPDVs arriving at damaged sites in the colon. Our formulation could have broad applications for other diseases, including alcohol-induced liver damage[32], wound healing[33], and anticancer therapy[34, 35].