Fluidic microgel assemblies with enhanced oral delivery of biologics alleviate colonic inflammation in murine and canine models

Abstract

The efficacy of oral delivery of biologic drugs for inflammatory bowel disease (IBD) treatment is substantially compromised by the swift enzymatic degradation and insufficient retention of biologics in the gastrointestinal tract. Here, we report a simple method for fabricating fluidic microgel assemblies (FMGAs) by introducing the hydrogen bonds between the microgel surfaces. These FMGAs with water content of 77% are flowable yet immiscible with water. Upon oral administration, these FMGAs form a large-area physical adhesive coating on the intestinal surface, extending intestinal retention time to at least 48 hours. The intermicrogel spaces within the assemblies can efficiently harbor therapeutic biologics (such as exosomes and TNFα antibody), protect them from the harsh intestinal microenvironment, and enable efficient delivery to intestinal lesion sites to restore intestinal barrier function and microbiota diversity in mouse and beagle models of IBD. We believe that the fluidic microgel assembly strategy holds great promise for oral delivery of labile biologics for the treatment of diverse gastrointestinal diseases.

Water-immiscible fluidic microgel assemblies enhance oral delivery of labile biologics to intestinal lesion sites for IBD therapy.

INTRODUCTION

Biologics represent a notable milestone in the treatment of recurrent inflammatory bowel disease (IBD) (1–4), which imposes a lifelong medical burden on millions of patients worldwide (5–8). The clinical guidelines of the American Gastroenterological Association recommend that adults with moderate to severe IBD initiate biological drug therapy at an early stage rather than after an inadequate response to 5-aminosalicylic acid therapy (9). However, the frequent administration of biologics via subcutaneous or intravenous injection often leads to substantial systemic drug exposure and challenges in patient compliance (10, 11). The oral delivery of a broad range of biologics is highly desirable for the treatment of IBD (12) but generally has limited therapeutic outcomes because of the rapid enzymatic degradation and insufficient retention time of biologics in the gastrointestinal (GI) tract. The substantial fluid flow in the intestine, notable intestinal motility, high protease activity, patchy distribution of IBD lesions among normal intestinal tissues, and easy dilution of drug delivery carriers (i.e., nano and microparticles) (1, 6, 10, 13, 14) due to the severe diarrhea associated with IBD (15–19) all contribute to the difficulties in achieving sustained oral release of orally administered biologics in the lesion sites of patients with IBD.

Maximizing the coverage area and duration of the drug delivery carriers in the intestinal mucosa to increase the probability of the drug reaching the intestinal lesion sites potentially represents a promising drug delivery strategy for IBD treatment. Previously, we reported a liquid coacervate that spreads to form a large-area coating in the colon and mediates the sustained release of small-molecule drugs (20). However, these nanoparticle-assembled coacervates suffer from limited loading capacity for large biologics [i.e., monoclonal antibodies and exosomes (EXOs)], which are generally regarded as the most efficacious modalities for the treatment of IBD. Microgel assemblies, which are macroscopic aggregates assembled from microgels in a bottom-up approach, typically have the capacity for loading and sustaining prolonged release of biologics (21–25). Compared with conventional drug delivery vehicles such as nan/microparticles, microgel assemblies are promising for resisting dilution by bodily fluids to extend the intestinal retention time, thus potentially enabling the sustained release of orally administered biologics in the GI environment. However, to the best of our knowledge, the previously reported solid or semisolid microgel assemblies do not exhibit the macroscopic fluidity and water-immiscibility required for the formation of durable large-area coatings on the intestinal surface (25, 26). Designing microgel assemblies that simultaneously have fluidity and water immiscibility for efficient drug delivery to the GI tract remains a challenge.

Here, we propose a simple and general strategy involving the assembly of microgels into water-immiscible fluidic condensates. EXOs (30 to 200 nm in diameter), which play important roles in anti-inflammatory and immune regulation, were used as a model of a large biologic. Fluidic microgel assemblies (FMGAs), driven by multidentate hydrogen bonding, effectively spread to form a large-area coating on the intestinal surface with a residence time of more than 2 days (Fig. 1A). FMGA can efficiently carry therapeutic EXOs, protect them from the extreme GI environment, and mediate the sustained release of preloaded EXOs at intestinal lesion sites. Using phosphate-buffered saline (PBS)–dispersed EXO as a control, the oral delivery of EXO-laden FMGA markedly enhanced IBD treatment outcomes, improved gut microbiota richness and diversity, and reduced the histopathological features in both mouse and beagle IBD models.

Fig. 1. Design of water-immiscible FMGAs.

(A) Core-shell structure design scheme and assembly strategy. (B) FMGA protected therapeutic biologics (such as EXOs) from the harsh environment of the GI tract and prolonged the duration of biologics function in the gut. (C and D) Assembly of the formed microgels was observed by macroscopic phase separation (C) and laser scanning confocal microscopy (D). Scale bar, 50 μm. (E) Changes in the storage modulus (G′) and loss modulus (G″) in NaCl solution with different concentrations, confirming that hydrogen bonding derived from the TAN induced the microgel assembly. (F and G) FTIR spectra of the GMs and FMGA. (H) XPS of the GM and FMGA.

RESULTS

Core-shell microgels self-assemble to form FMGAs

The core-shell microgel required to fabricate the FMGA assembly contains a shell composed of tannic acid (TAN)–modified branched polyethyleneimine [PEI; M_w (molecular weight) of 25,000 Da] and a gelatin microgel (GM) core (Fig. 1A). First, confocal laser scanning microscope (CLSM) and x-ray photoelectron spectroscopy (XPS) demonstrated the successful preparation of gelatin cores with a size of ~100 μm and a uniform distribution of C, O, and N elements. (figs. S1 to S4). In addition, the gelatin cores remained stable in PBS for several days and then gradually degraded in a collagenase solution (2 μg/ml; figs. S5 and S6). In addition, GM successfully loaded both fluorescein isothiocyanate (FITC)–labeled bovine serum albumin (BSA-FITC) protein molecules and nonprotein macromolecule dextran-FITC (fig. S7). Second, the core-shell structure of the microgel was prepared by grafting TAN-modified PEI onto the microgel (Fig. 1A). These core-shell microgels were then dialyzed with deionized (DI) water at room temperature for 48 hours, thereby inducing the in situ assembly to form a dense condensate phase through macroscopic phase separation (Fig. 1A). The FMGAs can flow like a liquid upon tilting of the test tube (Fig. 1C). The fluidity of the FMGA is determined by the shell of the microgel. The core-shell microgels with a lower TAN content (m_TAN/m_PEI = 0.70) cannot assemble and were miscible with water (table S1 and fig. S8). The core-shell microgels with excessive TAN content (m_TAN/m_PEI = 3.66) self-assembled to form macroscopic hydrogels, which were immiscible with water but had no fluidity (table S1 and fig. S8). Gelatin cores without modification of TAN on their surface failed to form microgel assemblies, thus confirming that the polydentate hydrogen bonds from the pyrogallol end of the TAN drive the formation of FMGAs.

The assembly behavior of the FMGA was further examined by CLSM (Fig. 1D), Fourier transform infrared (FTIR) spectrometer, XPS, and rheological tests (Fig. 1, F to H). The red color in the CLSM image represents the assembled GM cores, and TAN-modified PEI shells acted as flexible polymer linkers that connected the gelatin cores to form stable microgel assemblies. The FTIR spectra of the freeze-dried gelatin cores or FMGA showed that FMGA had an enhanced O═C characteristic peak at 1701 cm⁻¹, a C═C characteristic peak in the aromatic ring at 1615 cm⁻¹, and a C─N characteristic absorption peak at 1030 cm⁻¹, indicating that TANs were successfully grafted onto the PEI on the surface of the gelatin cores. The XPS spectra of the freeze-dried gelatin cores or FMGA showed that the energy spectrum of FMGA N1s was decomposed into two peaks: The first peak at 399.3 eV was attributed to C─N, and the characteristic peak at 400.8 eV was attributed to C═N, indicating the successful Schiff base reaction between TANs and PEI (27). Rheological analysis further confirmed the shear thinning behavior of the FMGA (fig. S9), and the loss modulus (G″) of the FMGA was always greater than the storage modulus (G′), thus confirming that its viscous fluid structure was maintained under high- and low-frequency shear (fig. S10). In addition, the moduli (G′ and G″) of FMGA was frequency dependent because of the presence of supramolecular interactions such as hydrogen bonds from pyrogallol groups (28). In addition, the modulus (G′ and G″) gradually increased with increasing salt concentration, and our FMGA remained stable over a wide pH range for up to 2 days (fig. S11), suggesting that the formation of FMGA can be attributed to hydrogen bonds rather than electrostatic interactions (Fig. 1E).

FMGA efficiently encapsulates EXOs and protects EXO integrity

EXOs, bioactive components secreted by cells, play crucial roles in anti-inflammatory and immune regulation. Current EXO treatments, primarily through intravenous injections, alleviate colitis to some extent but suffer from low bioavailability and potential trauma. We next evaluated the efficacy of FMGA to sustain prolonged release of umbilical cord mesenchymal stem cell–derived EXOs (UMSCs-EXO). EXOs were first characterized by dynamic light scattering (DLS), transmission electron microscopy (TEM), and Western blot (Fig. 2, A to C, and fig. S12). The sizes of UMSCs-EXO ranged from 30 to 200 nm (Fig. 2B), and TEM images revealed the vesicular structures of EXOs (Fig. 2C). The purity of the EXOs was further confirmed by the presence of the EXO-specific marker CD81 and the absence of cell-specific β-actin in the EXO fraction (fig. S12).

Fig. 2. Fluidic bioadhesive microgel assemblies (FMGA-EXO) can prolong the release and preserve the anti-inflammatory properties of loaded EXOs.

(A) Formation and characteristics of EXOs. (B) Distribution of EXO diameters. (C) TEM image indicates the double-membrane vesicular structure of EXOs. Scale bar, 50 nm. (D) EXO encapsulation efficiencies of the GM and FMGA for EXO. (E) CLSM images of FMGA-EXO and GM-EXO. The red fluorescence corresponds to GM, and the green fluorescence corresponds to EXO. Scale bar, 100 μm. (F) Morphology of FMGA soaked in SGF for 2 hours and then soaked in SIF for 2 days. (G and H) Degradation (G) and EXO release (H) of FMGA-EXO or GM-EXO incubated in SGF for 2 hours and then in SIF for several days. (I) Viscosity changes of FMGA-EXO under the frequency sweep (0 to 100 rad/s). (J and K) Rheological frequency sweep [(J) frequency fixed at 10 rad/s] and shear strain sweep [(K) frequency fixed at 10 rad/s] of FMGA-EXO incubated in SGF for 2 hours and then in SIF for 2 days. (L) TEM images of FMGA-treated EXOs that preserved structural integrity. (M and N) Flow cytometry analysis of CD63 and CD9 expression on PBS-treated versus FMGA-treated EXOs. (O) Efficiency of Caco-2 cells in phagocytosing EXO pretreated with FMGA and EXO pretreated with PBS. (P) TEM images of the EXO within FMGA-EXO treated with gastrointestinal fluids, respectively. (Q) The EXO released from GM-EXO and FMGA-EXO after 8 hours of sequential GI exposure, respectively, were cocultured with lipopolysaccharide (LPS)–induced RAW264.7 macrophages for 48 hours. The expression levels of proinflammatory factors (IL-6 and TNFα) and anti-inflammatory factors (IL-10) were detected by RT-PCR. The results of ELISA revealed the extracellular expression of IL-6 and IL-10. Data are represented as means ± SD.

We then quantified the encapsulation efficiency of UMSCs-EXO into FMGA, calculated as the ratio of EXOs successfully loaded into the FMGA to the total EXOs used multiplied by 100%. FMGA achieved a high-loading efficiency of 68%, whereas unmodified GMs lacking the TAN-PEI shell showed significantly lower efficiency (31%) (Fig. 2D). CLSM images revealed the successful loading of EXO by the FMGA and unmodified GM cores, termed FMGA-EXO and GM-EXO, respectively, with red indicating the FMGA and green indicating the EXO (Fig. 2E). We monitored the release kinetics of EXO by adding FMGA-EXO or GM-EXO to the simulated gastroenteric fluid and determining the EXO content in the simulated gastroenteric fluid at selected time points. The FMGA platform enables sustained release of EXOs, as evidenced by in vitro release kinetics in simulated gastroenteric fluid. EXO content in the fluid gradually increased over 48 hours (Fig. 2H and fig. S13). Crucially, stability assays confirmed that the FMGA effectively preserves EXO structural integrity under harsh GI conditions. We further conducted comparative release experiments using BSA and mesalazine as representative macromolecular and small-molecule agents (fig. S14). Under mechanical shaking (80 rpm, 37°C) conditions that mimic intestinal peristalsis, the differences in release rates between BSA and mesalazine remained within ~10%.

The stability of FMGA-EXO was further confirmed by incubation in simulated gastric fluid (SGF) for 2 hours, followed by incubation in simulated intestinal fluid (SIF) at 37°C for 2 days (Fig. 2F). A rheometer was used to test the properties of FMGA-EXO after soaking in gastric or intestinal fluid for 2 hours and 2 days, respectively. FMGA-EXO still maintained the viscous flow characteristics even after incubation in simulated gastric/intestinal fluid with a slight decrease in the modulus and viscosity (Fig. 2, I to K), whereas the GM-EXO quickly degraded in the simulated gastroenteric fluids (Fig. 2G). We systematically assessed vesicle integrity, surface marker preservation, and cargo transfer efficiency. TEM confirmed that FMGA-treated EXOs retained intact morphological structure, with no signs of membrane rupture or aggregation (Fig. 2L). Flow cytometry analysis further demonstrated that surface markers CD63 and CD9 on FMGA-treated EXOs showed no obvious difference compared to PBS-treated controls, confirming FMGA’s negligible impact on EXO surface protein integrity (Fig. 2M). To evaluate cargo transfer efficiency, intestinal epithelial cells (Caco-2 cells) were seeded in 48-well plates and cocultured with EXOs (100 μg/ml; FMGA treated or PBS treated) for 48 hours. Quantification revealed no notable difference in cellular uptake between FMGA-treated and PBS-treated EXOs (Fig. 2, N and O), indicating that FMGA does not interfere with EXO internalization by target cells. To evaluate the ability of FMGA-EXO to protect EXO from the hash environmental in gastric and intestinal fluids, we first collected EXO released from GM-EXO and FMGA-EXO after incubation for 2 hours in the gastric fluid, followed by 6 hours in the intestinal fluid. First, TEM analysis confirmed that FMGA-EXO maintained intact membrane morphology after GI incubation, whereas GM-EXO exhibited severe structural degradation (Fig. 2P). This demonstrates that FMGA’s protective matrix shields EXOs from enzymatic damages, preserving their functional architecture. Crucially, we further validated the therapeutic relevance of this protection by coculturing released EXOs with lipopolysaccharide (LPS)–stimulated RAW264.7 macrophages. Reverse transcription quantitative polymerase chain reaction (RT-qPCR) revealed that EXOs from FMGA-EXO significantly suppressed pro-inflammatory cytokines [interleukin-6 (IL-6) and tumor necrosis factor–α (TNFα)] and enhanced anti-inflammatory IL-10 expression compared to GM-EXO (Fig. 2Q). In contrast, GM-EXO showed diminished anti-inflammatory effects, likely due to structural compromise during 8 hours of sequential GI exposure. These findings underscore that while total EXO release may appear quantitatively similar, FMGA preserved EXO integrity to ensure sustained bioactivity, a factor critical for therapeutic efficacy.

Bioadhesive FMGA-EXO extends EXO retention in the GI tract

Maximizing intestinal mucosal coverage and retention duration of drug carriers represents a promising strategy for IBD treatment, as the patchy distribution of IBD lesions among normal intestinal tissues hinders sustained delivery of oral biologics to the lesion sites of patients with IBD. After oral administration, we hypothesize that the microscale fluidity of FMGA (fig. S15) enables it to conform to intestinal peristalsis and mucosal folds, thereby forming a large-area physical adhesive coating on the intestinal surface to efficiently deliver biologics to intestinal lesion sites. When the FMGA-EXO was deposited on the surface of the upright fresh pig intestinal mucosa, the FMGA adhered to the fresh wet mucosa and flowed steadily downward driven by gravity, leaving behind a trailing adhesive coating (Fig. 3A). Fluorescence microscopy images demonstrate the binding of FMGA-EXO (green) to the porcine intestinal mucus layer (red, stained with wheat germ agglutinin–Cy5) (fig. S16). The FMGA-EXO–coated fresh intestinal mucosa was further immersed in SGF or SIF, respectively. After incubation at 37°C for 2 hours, the FMGA-EXO adhesive coating remained undiluted and adhered to the surface of the intestinal mucosa. We further verified the TAN-mediated adhesion properties of FMGA-EXO and GM-EXO on wet pig skin tissues through universal testing machine and the differences in tissue adhesion between the FMGA group and the GM group in SGFs and intestinal fluids. The data of the universal testing machine indicate that the adhesion performance of FMGA-EXO is significantly higher than that of GM-EXO (Fig. 3, B and D, and fig. S17). Notably, after sequential exposure to SGF (2 hours) and SIF (6 hours), FMGA-EXO almost retained its initial adhesion capacity (Fig. 3C). The adhesive energy of FMGA-EXO was estimated to be ~5.2 J m⁻², which was comparable to the adhesive energy of polymer adhesives (about 2 to 10 J m⁻²) and nanoparticle-based adhesives (Fig. 3C) (29). We further investigated the cytocompatibility of FMGA-EXO and GM-EXO on intestinal epithelial cells (Caco-2 cells). The results show that FMGA-EXO and GM-EXO have good cytocompatibility (Fig. 3, E and F). In addition, the mice were administered FMGA-EXO or GM-EXO every other day for 10 days. Compared with untreated healthy mice, mice treated with oral administration of FMGA-EXO or GM-EXO presented no abnormal symptoms such as diarrhea, no obvious changes in body weight, and no obvious changes in organ tissue structure during the 10 days of gavage (Fig. 3G and fig. S18). In addition, representative images of hematoxylin and eosin (H&E)–stained major organs were not different between the mice that orally received FMGA-EXO and the healthy control mice at day 28 (fig. S19). These results further confirmed that FMGA-EXO and GM-EXO had good biocompatibility.

Fig. 3. FMGA prolonged the retention of orally delivered EXOs in the GI tract.

(A) Fluid FMGA adhered to mucosal surface of fresh wet pig intestines, slowly flowed downward to form the adhesive coating, and remained stable after soaking in SGF and SIF for 2 hours. Scale bar, 20 mm. (B) Shear strength of FMGA-EXO and GM-EXO on wet porcine skin (n ≥ 5). (C) After sequential exposure to SGF (2 hours) and SIF (6 hours), FMGA-EXO almost retained its initial adhesion capacity. (D) Schematic diagram of FMGA-EXO adhesion to tissue surfaces. (E) Representative image of Caco-2 cells stained with calcein AM (green)/PI (red). The images revealed that both FMGA-EXO and GM-EXO had good cytocompatibility. Scale bars, 100 μm. (F) Viability of cells treated with FMGA-EXO or GM-EXO (n ≥ 5). (G) Representative images of H&E-stained major organs were not different between the mice that orally received FMGA-EXO and the healthy control mice at day 10. Scale bar, 200 μm. (H) Fluorescence intensity at excitation wavelengths of 598 nm (FMGA/GM tracking) and 488 nm (EXO tracking) confirmed sustained release of EXOs delivered by FMGA (FMGA-EXO) in the GI tract compared to controls. (I) Cytokine array profiles comparing selected serum cytokines and chemokines between healthy mice and FMGA-EXO–treated mice. Mice were administered FMGA-EXO via oral route at designated intervals of days 1, 3, 5, 7, and 9. The blood of mice was collected for cytokine array analysis on day 10. (J) Multiorgan functional biomarker analysis comparing healthy mice and FMGA-EXO–treated mice. (K) Immunofluorescence staining of tight junction proteins occludin (green) and ZO-1 (red) in Caco-2 cells, demonstrating preserved localization at intercellular junctions. Scale bar, 20 μm. Data are presented as means ± SD.

Normal gastric emptying occurred within 2 hours, and colon arrival occurred 5 hours later (30). Therefore, we further examined the ability of FMGA-EXO and GM-EXO to act as durable bioadhesive coatings in the mice GI tract. The mice were given a single administration of 200 μl of EXO, FMGA-EXO, or GM-EXO [FMGA and GM produced red fluorescence due to the action of genipin, EXO labeled by 3,3′-dioctadecyloxacarbocyanine perchlorate (DIO) cell membrane green fluorescent probe] and euthanized at 0, 6, 12, 24, or 48 hours to assess fluorescence retention in the GI tract (Fig. 3H). Consistent with its in vitro adhesion properties, FMGA-EXO adhered to the GI tract of the mice for at least 48 hours and prolonged the retention time of EXO in the GI tract as evidenced by strong fluorescence signals in the GI tract at excitation wavelengths of 598 and 488 nm. However, the retention of the GM-EXO in the GI tract was limited after 1 day (Fig. 3H). The retention of the EXO in the GI tract was limited after 12 hours (Fig. 3H). The fluorescence intensities of FMGA (598 nm) and EXO (488 nm) in the main digestive tract of the FMGA-EXO group were significantly greater than those of the GM-EXO group. Furthermore, the live imaging of major organs (heart, liver, spleen, lung, kidney, and brain) in colitis mice (fig. S20) revealed no detectable EXO distribution outside the GI tract in FMGA-EXO–treated animals. Free EXOs alone face rapid transit and dilution in the intestinal lumen, limiting their interaction with inflamed tissues. This challenge can be potentially addressed by our FMGAs, which exhibit unique water-immiscible properties (77% water content) to form a robust, large-area adhesive coating on the intestinal mucosa upon oral administration (Fig. 3H).

The cytokine array analysis of FMGA-EXO–treated mice showed no obvious differences in inflammatory mediator levels compared to healthy controls (Fig. 3I), indicating no systemic immune activation. We further tested the biomarkers creatinine (Cr) and urea (BUN), respectively, to evaluate renal function and tested the biomarkers aspartate aminotransferase (AST) and alanine aminotransferase (ALT), respectively, to evaluate liver function. The biomarkers creatine kinase (CK) and C-reactive protein (CRP) were detected to evaluate liver function, and the biomarkers white blood cell (WBC) count and platelet count were detected to evaluate liver function (Fig. 3J). The cytokine array data and multiorgan function biomarker data of mice in the FMGA-EXO group showed no significant differences from those of mice in the healthy group, indicating that FMGA-EXO has good biological safety (Fig. 3, I and J; table S2; and fig. S21). To test the potential impact of FMGA adhesion on intestinal epithelial cell function, we evaluated the effects of FMGA-EXO and GM-EXO on intestinal epithelial integrity using Caco-2 cells, focusing on the expression and localization of critical tight junction proteins: occludin and zonula occludens-1 (ZO-1). These proteins are essential for maintaining intercellular barrier function, regulating signal transduction, and responding to pathological conditions. Immunofluorescence staining revealed that both FMGA-EXO and GM-EXO preserved normal distribution and membrane localization of occludin (green fluorescence) and ZO-1 (red fluorescence) at intercellular junctions, with no disruption to their structural organization (Fig. 3K and fig. S22). The fluidity, water immiscibility, and bioadhesion of our FMGA-EXO allowed them to adapt to notable pH changes, liquid environments, and complex movements in the GI tract and formed durable coatings over large areas of the intestinal mucosal surface to continuously release bioactive drugs (EXO) continuously.

EXO-loaded FMGA effectively alleviates UC in a mouse model

We next evaluated the therapeutic effect of FMGA-EXO in a mouse model of ulcerative colitis (UC) induced by dextran sulfate sodium (DSS) (Fig. 4A). The mice were administered 4.5% DSS in the drinking water for 7 days to induce UC. The evaluation of the body weight, colon length, and disease activity index (DAI) of the mice after 7 days demonstrated successful establishment of the colitis model (fig. S23). After the successful establishment of the mouse colitis model, the mice were orally gavaged with EXO-laden FMGA (FMGA-EXO group), FMGA (FMGA group), EXO-laden GM (GM-EXO group), or an equivalent amount of EXO in PBS (200 μg, EXO group) on days 1, 3, and 5 (Fig. 4A). Untreated C57BL/6J mice with colitis were used as the negative control.

Fig. 4. EXO-loaded FMGA showed enhanced therapeutic effects in a DSS-induced colitis C57BL/6J mouse model.

(A) Colitis was induced in mice given drinking water containing 4.5% DSS for 7 days. Colitis model mice were treated with PBS, EXO, FMGA, GM-EXO, or FMGA-EXO on days 1, 3, and 5, respectively. PBS-treated colitis model mice were used as negative controls. All the mice were euthanized on day 7. n = 6 independent C57BL/6J mice per group. (B) Representative colons from each mouse are shown on the left. The figure on the right shows the quantification of colon length in mice. Compared with the control mice, the EXO-treated mice, FMGA-treated mice, and GM-EXO-treated mice, FMGA-EXO-treated mice, and FMGA-EXO-treated mice presented significant improvements in colon shortening caused by UC. n = 6 independent mice per group. Scale bar, 5 mm. (C) Daily body weight changes in each group of mice. (D) Daily DAI of the mice in each group. (E) H&E staining of the distal colon of mice. Scale bars, 200 μm. (F) On Day 7, all the mice were euthanized, and the polarization of macrophages in the colon tissue of the mice was analyzed. (G) Immunofluorescence staining for the markers F4/80⁺CD86⁺ and F4/80⁺CD206⁺ in M1 macrophages and M2 macrophages, respectively. Scale bars, 50 μm. (H) Histopathological scores. (I to L) mRNA levels of the anti-inflammatory factor IL-10 (J) and the proinflammatory factors IL-6 (I), IL-1β (K), and TNF (L). (M) mRNA levels of tight junction–associated protein (ZO-1) in colonic tissue. (N and O) ELISA results showing IL-6 (N) and IL-10 (O) levels in colonic tissues. (P) Colitis mice were treated with FMGA-TNFα Ab for 7 days. (Q to S) DAI, representative colons, and colon quantification in mice. (T) H&E staining of colon sections. Scale bar, 50 μm. Data are presented as means ± SD.

The mice were allowed unrestricted access to water and a standard diet before and after oral tube feeding. Their body weights, fecal traits, and fecal occult blood were measured every day, and the mice were euthanized on day 7 to further evaluate their colon length, histological severity, transcription levels of related inflammatory factors (IL-6, IL-1β, TNF, and IL-10) in the distal colon, contents of related inflammatory factors, and transcription levels of tight junction–related protein (ZO-1). Our results demonstrated the substantial efficacy of FMGA-EXO therapy in DSS-induced UC. FMGA-EXO reduced weight loss, diarrhea, and blood in the stool in UC induced by DSS (Fig. 4, B to D). Representative images of hematoxylin/eosin (H&E) staining and alcian blue/nuclear fast red staining revealed substantial inflammatory infiltration, severe crypt destruction, and high histopathological scores in untreated mice with colitis. The histological inflammation of the mice with colitis that received FMGA-EXO via oral gavage was substantially improved. The colonic structure and crypts were intact, the number of cup cells was increased, and the histological score was low, indicating that FMGA-EXO could relieve colitis. Mice treated with the same amount of EXO in PBS (EXO group) presented relatively severe crypt defects, fewer goblet cells, and higher histopathological scores. Goblet cell proliferation and inflammatory infiltration were observed in colitis mice treated with GM containing equal amounts of EXO (GM-EXO group) (Fig. 4E and fig. S24).

Macrophage polarization usually includes proinflammatory M1 polarization and anti-inflammatory M2 polarization (31). Macrophages play a key role in maintaining mucosal stability by secreting a variety of cytokines (32). Therefore, we investigated the immune response of mice with colitis by analyzing macrophage polarization in the distal colonic tissues of the FMGA-EXO group, GM-EXO group, EXO group, FMGA group, and untreated group (control group) (Fig. 4, F and G). Studies have shown that EXO can inhibit proinflammatory M1 polarization and anti-inflammatory M2 polarization (33), thereby inhibiting the secretion of proinflammatory factors and increasing the secretion of anti-inflammatory factors. Immunofluorescence staining for F4/80⁺CD86⁺ (M1 marker) or F4/80⁺CD206⁺ (M2 marker) revealed reduced macrophage M1 polarization and increased macrophage M2 polarization in the mice with colitis that received FMGA-EXO via oral gavage (Fig. 4G and fig. S25). In addition, mice with colitis treated with FMGA-EXO recovered the expression of tight junction–associated protein (ZO-1) and demonstrated significant reductions in the expression levels of proinflammatory factors (IL-1β, IL-6, and TNF) in colon tissue compared with those of the untreated group. The expression level of anti-inflammatory factor (IL-10) was significantly increased (Fig. 4, H to O). In summary, the oral delivery of EXO-laden FMGA improved the therapeutic effect of EXO in mice with colitis compared with the administration of the same amount of EXO in aqueous solution form by protecting EXO from the adverse gastrointestinal environment and reducing systemic exposure.

Furthermore, EXOs were used primarily as a representative model of large biologics to validate the functional performance of our FMGA platform. The oral delivery of biologics typically faces limited therapeutic efficacy due to rapid enzymatic degradation and insufficient GI residence time. To rigorously assess the compatibility and generalizability of the FMGA-based delivery system for other biologics, we further conducted systematic evaluations using a TNFα antibody (TNFα Ab), a prototypical macromolecular therapeutic for IBD treatment. We established five experimental groups for IBD treatment: healthy mice, untreated controls (IBD model), TNFα Ab only, GM-TNFα Ab, and FMGA-TNFα Ab (Fig. 4P). The encapsulation efficiency of TNFα antibody in FMGA was calculated as (amount of TNFα Ab loaded in FMGA/total TNFα Ab used) × 100%, yielding a high efficiency of 73%. In contrast, unmodified GMs lacking the TAN-PEI shell achieved only about 31% loading efficiency. Therapeutic outcomes were assessed through DAI, colon length analysis, and histopathology (Fig. 4, Q to T). FMGA-TNFα Ab–treated mice exhibited significantly lower DAI scores (Fig. 4Q), reduced colon shortening (Fig. 4, R and S), and improved mucosal architecture with diminished inflammatory infiltration (Fig. 4T) compared to TNFα Ab only and GM-TNFα Ab groups. Collectively, these findings demonstrate that FMGA markedly enhances the therapeutic efficacy of macromolecular biologics such as monoclonal antibodies by improving their gastrointestinal retention and targeted delivery, consistent with its superior performance in the EXO-based therapy.

FMGA-EXO restores intestinal microbiota

The occurrence and development of intestinal inflammation can lead to an imbalance in the intestinal microbiota, which plays an important role in healthy gut development (34). Restoring the gut microbiota can help treat colitis (18). Fecal samples collected from the mice with colitis on day 7 were analyzed by sequencing the V3-V4 region of the 16S ribosomal RNA (rRNA) gene. Principal coordinate analysis (PCoA) revealed obvious differences in the β diversity between the groups, indicating a notable difference in the gut microbiota composition between the colitis model mice and healthy control mice, which could be modulated by oral FMGA-EXO (Fig. 5B). Species diversity was further analyzed by nonmetric multidimensional scaling (NMDS), and the analysis results were similar to those of PCoA, indicating that species differences existed between different groups and that FMGA-EXO could change the composition of the intestinal microbiota in model mice (Fig. 5C). We subsequently analyzed the sequencing data of the microbiome, and the results revealed that the oral administration of FMGA-EXO increased the bacterial richness [observed operational taxonomic units (OTUs)] (Fig. 5, D and E) and diversity (Chao index) (Fig. 5F) in the mice with UC.

Fig. 5. FMGA-EXO restored the intestinal microbiota.

(A) Intestinal microbiota was analyzed by sequencing the V3-V4 region of the 16S rRNA gene collected from colitis mice. (B and C) Intestinal microbial species diversity was analyzed by PCoA and NMDS. (D to F) Venn diagram analysis, microbiota richness (observed OTUs), and diversity (Chao index) of mice with colitis that received oral gavages of EXO-laden FMGA (FMGA-EXO group) increased compared with those in the EXO group and control group. n = 5 independent C57BL/6J mice per group. (G) Heatmap of the β diversity heatmap of the gut microbiota. The closer the evolutionary tree was, the greater the sample similarity was. The color of the square indicates the evolutionary distance between samples. The red-to-blue range corresponds to the distance to a close range. The larger the index was, the greater the difference between samples was. (H) Species composition clustering heatmap. Longitudinal clustering indicated similarity of species composition among different samples. The closer the distance was, the shorter the branch length was, indicating that the species composition and abundance of the samples were more similar. Horizontal clustering indicated that the abundance of the species was similar across all the samples. Same is in vertical clustering, the shorter the distance was, the shorter the branch length was, indicating that the compositions of the two species were more similar in each sample. (I) Functional difference analysis of FMGA-EXO versus EXO and FMGA-EXO versus control. The histogram of the relative abundance of each group is shown on the left. The log₂ value of the mean relative abundance ratio of the same pathway in both groups is shown in the middle. The figure on the right shows the P value and false discovery rate (FDR) values obtained via the Wilcoxon test. Data represented is as mean ± SD.

In addition, we generated heatmaps of β diversity distance distributions and clustered samples with similar β diversities to reflect similar compositions of the gut microbiota (Fig. 5G). In comparisons among the EXO group, FMGA-EXO group, GM-EXO group, and untreated control group, there was a tighter aggregation of β diversity between the colonic mice treated with FMGA-EXO and the healthy mice, suggesting that FMGA-EXO treatment promoted the recovery of the intestinal microbiota in the colonic mice. This finding was further confirmed based on a heatmap of intestinal flora clusters (Fig. 5H). The results showed an increase in beneficial bacteria (such as Lactobacillus and Alloprevotella) and a decrease in harmful bacteria (such as Paraprevotella and Bacteroides) in the FMGA-EXO group (fig. S26). The functional difference analysis data revealed that the immune and endocrine systems of the colitis mice in the control (untreated) group were down-regulated compared with those of the healthy control mice. Compared with EXO-treated mice (EXO group) and untreated (control group) colitis mice, the FMGA-EXO–treated mice presented an up-regulation of the immune system and endocrine system of the gut microbiota (Fig. 5I). To address whether these effects arise from the FMGA material or the encapsulated EXOs, we conducted additional comparative analyses of the gut microbiota in colitis mice treated with FMGA alone versus untreated controls (fig. S27). To eliminate potential confounding effects from batch-to-batch variability, the same cohort of mice was consistently used across all stages of the IBD modeling process. The results revealed no obvious positive regulation of microbiota by FMGA alone. Venn diagram analysis demonstrated that the FMGA group exhibited fewer unique microbial species compared to the control group (fig. S27). These findings confirm that FMGA itself does not directly modulate gut microbiota.

We set a clinical-used golden standard in IBD as control groups to further demonstrated the advantages of FMGA-EXO (35–37). Therapeutic outcomes were assessed through DAI, colon length analysis, and histopathology (figs. S28 to S30). FMGA-EXO–treated mice exhibited significantly lower DAI scores (fig. S28), reduced colon shortening (fig. S29), and improved mucosal architecture with diminished inflammatory infiltration (fig. S30) compared to untreated controls and mesalazine groups. These results confirm that FMGA-EXO has efficacy superior to the clinical-used golden standard (mesalazine) in IBD.

FMGA-EXO effectively alleviates UC in a canine model

We next evaluated the efficacy of FMGA-EXO to treat colitis in beagle dogs. Endoscopic images of the healthy Beagle dog show the intact structure of a healthy colon (fig. S31). To establish the beagle colitis model, 7% acetic acid was injected into the colon of the beagle dogs on day 0 (Fig. 6A). The colonic mucosa of beagle dogs treated with acetic acid was disrupted, resulting in the release of inflammatory factors (fig. S32). One day later, the colonoscopy of the beagle’s colon revealed obvious bleeding and wounds (Fig. 6D). After successful establishment of the beagle colitis model, the beagles with colitis received oral gavages of FMGA-EXO or GM-EXO on days 1, 3, and 5, respectively (Fig. 6A). On day 7, the beagles treated with FMGA-EXO lost less weight than those beagles treated with GM-EXO, EXO, and PBS (Fig. 6B). Representative H&E stained sections of the major digestive tracts (esophagus, stomach, duodenum, jejunum, and ileum) of beagle dogs in the GM-EXO and FMGA-EXO groups revealed the integrity of different tissue structures (fig. S33). H&E staining of major beagle organs during treatment validated the safety of the GM-EXO group and FMGA-EXO in beagles (fig. S34). To monitor the intestinal retention of FMGA-EXO over time, a series of x-ray images of beagle dogs were taken periodically (Fig. 6C). Gastrointestinal retention assays confirmed that FMGA-EXO formed an extensive adhesive coating on the intestinal surface of beagles with a retention duration of 48 hours (Fig. 6C), whereas free EXO exhibited rapid clearance. The oral administration of FMGA-EXO had a notable effect on intestinal healing in colitis beagles (Fig. 6, D to F and I). FMGA-EXO–treated beagles demonstrated significantly preserved colon length (Fig. 6F), reduced DAI scores (Fig. 6I), and normalized systemic inflammation markers [WBC and high-sensitivity C-reactive protein (hs-CRP); Fig. 6, G and H]. Critically, hepatic and renal safety profiles (AST, Cr, and ALT; Fig. 6, J to L) confirmed FMGA-EXO’s biocompatibility. Furthermore, macrophage polarization analysis revealed a marked shift toward anti-inflammatory M2 phenotypes (F4/80⁺CD206⁺) and reduced pro-inflammatory M1 phenotypes (F4/80⁺CD86⁺) in the FMGA-EXO group compared to GM-EXO, EXO, and PBS (Fig. 6, M to O). Histopathological improvements, including restored crypt architecture, increased goblet cells, and diminished inflammatory infiltration (Fig. 6P and figs. S35 and S36), further validated FMGA-EXO’s ability to promote mucosal healing.

Fig. 6. FMGA-EXO demonstrated enhanced therapeutic effects and restored the gut microbiota in a beagle dog model.

(A) Beagle dogs were treated with 7% acetic acid (2 ml kg⁻¹) for 180 s. Colitis dogs received oral gavages of GM-EXO (10 ml/kg) or FMGA-EXO (10 ml/kg) orally on days 1, 3, and 5, respectively (n = 5 dogs per group). (B) Weight changes in dogs during treatment. The data were normalized as a percentage of body weight on day 0. (C) X-ray images of intestinal retention (indicated by red arrows) of FMGA-EXO and EXO groups. (D) Representative endoscopic images of dogs in Control, EXO, GM-EXO and FMGA-EXO groups over a 7-day period. (E) Representative image of colon of dogs that received different treatments on day 7. (F) The quantification of colon length in dogs. (G and H) Leukocyte (G) and hypersensitive CRP (H) levels in dogs that received different treatments on day 7. (I) DAI of dogs in each group (n = 5 dogs per group). (J to L) AST (J), Cr (K), and ALT (L) levels in dogs that received different treatments on day 7. (M and N) Fluorescence intensity ratio of (M) CD206⁺ (M2 polarized) to F4/80⁺ and (N) CD86⁺ (M1 polarized) to F4/80⁺ on day 7. (O) Immunofluorescence staining of the markers F4/80⁺CD86⁺ and F4/80⁺CD206⁺ for M1 macrophages and M2 macrophages, respectively. Scale bars, 50 μm. (P) AB staining and H&E staining of the distal colon of dogs. Scale bar, 50 μm. (Q) Box plot of the alpha diversity of fecal microbial species, as shown by the Chao1, Simpson, and Shannon indices. (R) Sample similarity heatmap showing fecal microbiome β diversity. (S and T) Functional distribution of the gut microbiota of dogs that received oral gavages of GM-EXO or FMGA-EXO. (U) Functional KEGG pathway enrichment in GM-EXO and FMGA-EXO groups. Data are expressed as means ± SD. tRNA, transfer RNA; UDP, uridine 5′-diphosphate; dTTP, 3′-deoxythymidine 5′-triphosphate.

To further test whether FMGA-EXO can alleviate the imbalance in the gut microbiota of beagles, we used metagenomic sequencing to analyze the gut microbiota of beagles. The box plot of the α-diversity of the fecal microbial species shown by the Chao1, Simpson, and Shannon indices revealed that FMGA-EXO treatment increased the diversity of the intestinal microbial species in beagles (Fig. 6Q). In addition, we generated heatmaps of the distance distribution of β diversity and clustered samples with similar β diversity to reflect similar compositions of the gut microbiota (Fig. 6R). To further explain why FMGA-EXO enhanced the recovery of the intestinal microbiota in colitis beagles, linear discriminant analysis effect size (LEfSe) was performed (figs. S37 and S38). In the analysis of the 300 species with the highest abundance, the number of different species of intestinal microbiota in beagles treated with FMGA-EXO was markedly greater than that in the GM-EXO group, suggesting that FMGA-EXO restored the diversity of the intestinal microbiota in the beagles. Functional Circos diagram (Fig. 6, S and T, and fig. S39) analysis revealed that the FMGA-EXO treatment up-regulated the immune system, amino acid metabolism, nucleotide metabolism, and other functions of the intestinal microbiota. Functional Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis further revealed notable differences in 30 KEGG pathways between the GM-EXO and FMGA-EXO groups (Fig. 6U and fig. S40). In beagles that received oral gavages of FMGA-EXO, the functional differences in the intestinal microbiota genes were concentrated in the pentose phosphate pathway, nucleotide biosynthesis pathway, amino acid (cysteine, tryptophan, etc.) biosynthesis pathway, polyamine biosynthesis pathway, and other inflammation-related pathways. These results together confirmed that the FMGA-EXO treatment restored the gut microbiota by suppressing inflammatory pathways.

DISCUSSION

The development of oral delivery platforms for biologics represents a transformative frontier in treating gastrointestinal disorders. Therefore, the primary focus of this work is to establish the universality of the FMGA for oral delivery of macromolecular biologics, addressing the rapid enzymatic degradation and insufficient retention time of labile biologics in the GI tract. While EXOs (UMSCs-EXO) served as a representative biologic model to validate delivery capabilities of FMGA, their intrinsic anti-inflammatory and barrier-repairing mechanisms in IBD have been reported in prior studies (4, 18, 38). These reports confirm that EXOs act via multicomponent synergy (miRNAs and proteins) to modulate immune responses and promote mucosal healing, albeit with limited therapeutic efficacy due to rapid clearance and instability in the GI environment, a key issue our FMGA aims to resolve.

In summary, we report a simple method for fabricating FMGAs for the oral delivery of labile biologics. The formation of FMGA was attributed to the assembly of core-shell microgels driven by the polydentate hydrogen bonds from the pyrogallol groups of the TAN shell. Unlike conventional solid or semisolid microgel assemblies, the oral gavage of the bioadhesive FMGA adapts to the dynamic gastrointestinal environment, effectively spreading to form a durable, large-area coating on the intestinal surface with a residence time of over 2 days. The FMGA efficiently carries therapeutic EXOs, protects them from the harsh gastrointestinal conditions, and mediates their sustained release at intestinal lesion sites. Furthermore, the oral delivery of EXO-loaded FMGA markedly improved IBD treatment outcomes, enhanced gut microbiota richness and diversity, and reduced histopathological features in both mouse and beagle IBD models. The capacity of FMGA highlights its potential to broaden the therapeutic window for labile biologics, from EXOs to monoclonal antibodies, which have been restricted to invasive administration routes. Therefore, we believe that our FMGA, characterized by macroscopic fluidity and water immiscibility, represents a promising drug delivery carrier for the treatment of diverse gastrointestinal diseases.

MATERIALS AND METHODS

In situ assembly of as-prepared core-shell microgels into microgel assemblies (FMGAs)

The gelatin (type A, Maokang Biotechnology) solution was added to polydimethicone (M_w ~ 25,000 Da, viscosity = 100 cS; Sigma-Aldrich) and heated to 40°C at 1000 rpm. Blend the gelatin-polydimethicone mixture continuously in a blender at 2000 rpm for 30 min. The polydimethylsiloxane containing the sample was placed in an ice bath and stirred at a constant speed for 1 hour. The sample was centrifuged to remove the upper layer and washed three times with PBS solution containing 1% Tween 20, and the sample was collected centrifugally. Subsequently, the collected samples were added to 1% Genipin solution prepared with 10 mM PBS for cross-linking reaction for 48 hours, centrifugally enriched the samples, and repeatedly cleaned with distilled water for three times and isopropyl alcohol for three times. The microgel was then purified by dialysis in distilled water with dialysis membrane [14 kDa molecular weight cut off (MWCO)] for 4 weeks, and the distilled water solution was changed daily. Then, the microgel was purified by dialysis in 0.1 M NaCl solution with dialysis membrane (14 kDa MWCO) for 7 days, and then the microgel was treated with distilled water for 7 days. The final concentration of the purified microgel dispersion (GM) was 1.03 wt %. 1-ethyl-3-(3-dimethylaminopropyl) carbodiimide and N-hydroxysuccinimide were added to 1.03 wt % microgel, the pH of the reaction solution was adjusted to 4.5 with 0.1 M HCl, and the solution was stirred at room temperature for 2 hours. Then, PEI was added, and the pH of the reaction solution was adjusted to 10 by adding 0.1 M NaOH, and the reaction was carried out at room temperature for 48 hours. The modified microgel was purified by dialysis membrane (14 kDa MWCO), and the purified dispersion was enriched to 3.00 wt % concentration. Then, the 3.00 wt % microgel dispersion was purged by N₂ for 30 min, the TAN solution dispersed in tris-HCl (pH = 8.5) was added under continuous agitation at a constant speed, and the reaction was carried out in a shaker at 4°C for 12 hours. The prepared core-shell microgel solution was dialyzed at room temperature in dialysis membrane with 3500 MWCO (standard regenerated cellulose dialysis membrane) for 48 hours in distilled water to form microgel assembly (FMGA).

Characterizations of microgel assemblies (FMGA)

The samples were observed by TCS SP8 Leica laser confocal scanning microscope. The morphology of GM was observed, respectively, under the bright-field channel or the excitation wavelength of 561-nm laser channel, and the size of microgels was calculated. The morphology of FMGA was observed under the excitation wavelength of 561-nm laser channel. After the GM was freeze-dried for 96 hours by a freeze-drying machine, its structure was examined by Zeiss EVO MA15 field emission-scanning electron microscopy, and the accelerated voltage was controlled at 25 kV. After the samples GM and FMGA were freeze-dried in the freeze-drying machine for 96 hours, the spectral scanning of GM and FMGA was performed with the Avatar 380 Fourier transform infrared spectrometer (Thermo Fisher Scientific Nicolet), and the wave number of the scanning was controlled within 500 to 4000 cm⁻¹. The chemical states of FMGA and GM were analyzed by the full spectrum scanning of ESCALAB250 x-ray photoelectron spectrometer (Thermo Fisher Scientific) and high-power scanning of C and N elements.

Preparation of bioactive microgel assemblies (FMGA-EXO) and bioactive microgels (GM-EXO)

GM and FMGA were incubated overnight in α–minimum essential medium (MEM) medium at 4°C.Then, appropriate amount (3.23 mg/ml) of UMSCs-EXO was dripped into GM, then vortexed on a vortexer for 1 hour, swirled for 15 min, and incubated in a shaker at 4°C overnight. The bioactive GM was obtained after repeated cleaning of unloaded UMSCs-EXO with α-MEM medium. Appropriate amount (1.47 mg/ml) of UMSCs-EXO was dropped into FMGA and then vortexed on a vortexer for 1 hour, swirled for 15 min, and incubated in a shaker at 4°C overnight. The bioactive FMGA (FMGA-EXO) was obtained after repeated cleaning of the unloaded EXOs with the α-MEM medium.

Inflammatory macrophage model was used to detect anti-inflammatory activity in vitro

RAW264.7 cells were inoculated into 48-well culture plates (3.25 × 10⁶ cells per well), stimulated by LPS (0.5 μg/ml) for 4 hours and then added with EXO (100 μg/ml; EXO collected after simulated gastroenteric fluid treatment) and cocultured for 24 hours. The cell culture medium was collected and centrifuged at 2000 rpm for 3 min. The supernatant was collected, and the contents of inflammatory factors (IL-6 and IL-10) were detected by enzyme-linked immunosorbent assay (ELISA). The cell underwent a thorough lavage with PBS buffer repeated five times. Following this, the cellular constituents were harvested, and the genetic material RNA was meticulously extracted. Subsequently, the qRT-PCR method was used to ascertain the levels of mRNA expression of related inflammatory factors (IL-6, TNFα, and IL-10).

Stability of FMGA-EXO in SGF and SIF

GM-EXO (1 ml) was added into SGF and incubated in a shaker at 37°C for 2 hours, and the GM-EXO in SGF was centrifuged (2000 rpm, 5 min) and precipitated. After the precipitation was dispersed in SIF and incubated in a shaker at 37°C for a specific time, GM-EXO/SIF was centrifuged (2000 rpm, 5 min). The EXOs from the upper phase were enriched and dispersed in PBS, and their content was quantified using the bicinchoninic acid assay (BCA) method. The precipitation was washed with DI water for three times and freeze-dried. Then, the remaining weight was quantified. The relative weight of GM-EXO at different time points was calculated by the following formula: relative weight (%) = remaining weight/initial weight.

FMGA-EXO (1 ml) was added into SGF and incubated in a shaker at 37°C for 2 hours, and FMGA-EXO in SGF was left for 10 min. The precipitation was dispersed in SIF, incubated in a shaker at 37°C, and set for at least 6 min on FMGA-EXO/SIF. The EXOs from the upper phase were enriched and dispersed in PBS, and their content was quantified using the BCA method. The condensed phase was cleaned by DI water for three times and freeze-dried, quantifying the remaining weight. The relative weight of FMGA-EXO at different time points was calculated by the following formula: relative weight (%) = remaining weight/initial weight.

In vitro cytocompatibility and in vivo safety of FMGA-EXO

The cytocompatibility assessment was conducted in accordance with the methodologies outlined in the literature (39). Caco-2 cells were seeded into a 24-well culture plate at a concentration of 8 × 10⁴ cells per well in α-MEM medium (Gibco) supplemented with 1% penicillin-streptomycin (Gibco) and 10% fetal bovine serum. GM-EXO or FMGA-EXO was incorporated at a final concentration of 3 mg/ml. Neither GM-EXO nor FMGA-EXO was included in the control group. Following a 48-hour incubation period, the Caco-2 cells were visualized utilizing a CLSM (Leica TCS SP8) using live and dead fluorescent staining (Sigma-Aldrich). In the Cell Count Kit-8 assay, caco-2 cells were inoculated into 96-well plates in a growth medium at a density of 1 × 10⁴ cells per well, and cells were cultured with different concentrations of GM-EXO or FMGA-EXO extracts for 2 days. Subsequently, a volume of 10 μl of CCK-8 solution was introduced into each well, followed by a further incubation period of 3 hours. The optical density readings of each well at a wavelength of 450 nm were quantified using a Multiskan spectrophotometer (Thermo Fisher Scientific).

In vivo safety of GM-EXO or FMGA-EXO. The C57BL/6J mice were partitioned into three cohorts by chance: a cohort of untreated healthy mice for control purposes, while the remaining two cohorts were administered GM-EXO or FMGA-EXO via oral route at designated intervals of days 1, 3, 5, 7, and 9, correspondingly. Body mass was meticulously documented on a daily basis, while fecal characteristics were meticulously evaluated. The major organs (heart, liver, spleen, lung, kidney, and brain) were resected for HE measurement. We collected the blood of mice for cytokine array analysis on day 10.

In addition to histopathological assessment via H&E staining, we conducted systemic inflammatory profiling of blood using cytokine arrays and assessed multiorgan functional biomarkers of blood to comprehensively evaluate biocompatibility. The C57BL/6J mice were partitioned into two groups by chance: a group of untreated healthy mice for control purposes and one group that was administered FMGA-EXO via oral route at designated intervals of days 1, 3, 5, 7, and 9. We collected the blood of mice for cytokine array analysis and assessed multiorgan functional biomarkers analysis on day 10.

Adhesion of FMGA-EXO in vivo and in vitro

Prior to in vitro adhesion experiments, fresh pig intestine tissue and pig skin were washed 5 times with PBS buffer (1×). The fluid adhesion behavior of FMGA-EXO induced by gravity at different times was measured by deposition of FMGA-EXO on the surface of upright wet intestinal mucosa. The FMGA-EXO coated intestinal tissues were immersed in SGF and SIF at 37°C for 2 hours, respectively. Photos were taken before and after immersion to monitor the changes of adhesion to FMGA-EXO coating. GM-EXO and FMGA-EXO were applied to the interface of two strips of fresh pig skin tissue. Then, the lap shear test and the standard 180° stripping test were carried out by the electronic universal testing machine.

Before in vivo adherence, C57BL/6J mice were fed drinking water containing 4.5% DSS for 7 days to induce colitis, followed by regular water during treatment. DSS-induced colitis mice were given a single oral dose of 0.2 ml of GM-EXO or FMGA-EXO (DIO cell membrane green fluorescent probe labeling). Then, C57BL/6J mice were euthanized at 0, 6, 12, 24 and 48 hours, and the complete GI tract was collected. The distribution and intensity of fluorescence were measured using in vivo imaging system (IVIS) spectrum, a small animal imaging system with immunoglobulin G (IGG) filter channels.

Mouse colitis model induced by DSS

The animal care, feeding, and research procedures of all C57BL/6J mice were carried out in accordance with the guidelines for the care and use of experimental animals and approved by the ethics committee of the animal center of South China University of Technology. Animal ethics number is 2024021. The 8-week-old female C57BL/6J murine specimen was procured from Specific Pathogen Free Laboratory Animal Center (SLAC) Laboratory Animals LLC. Colitis was induced by feeding mice with 4.5% DSS (50 kDa, Macklin) for 7 days to construct a C57BL/6J mice colitis model. The C57BL/6J mice were randomly divided into five groups. Healthy C57BL/6J mice drank only ordinary water.

Then, EXO (200 μg) dispersed in PBS, GM (GM-EXO) containing an equal amount of EXO, or FMGA containing an equal amount of EXO (FMGA-EXO) was given by oral route to colitis C57BL/6J mice on days 1, 3, and 5, respectively. In addition, colitis C57BL/6J mice were given equal amounts of FMGA by oral route on days 1, 3, and 5. Untreated colitis C57BL/6J mice were used as negative controls. All C57BL/6J mice were allowed unrestricted water and standard laboratory diet before oral administration of the therapeutic material. All C57BL/6J mice were euthanized on day 7, and the entire colon was resected for length measurement, histological score, relevant mRNA expression detection, and ELISA assay (MultiSciences), following the manufacturer’s instructions. On day 5 of treatment, fresh fecal samples were collected and stored in enzyme-free sterile tubules at −80°C for further analysis of the gut microbiota. Dexamethasone ELISA assay (Neogen Corporation) following the manufacturer’s instructions.

Mesalazine (6 mg) or FMGA-EXO was given by oral route to colitis C57BL/6J mice on days 1, 3, and 5, respectively. All C57BL/6J mice were euthanized on day 7, and the entire colon was resected for length measurement and H&E staining.

TNFα antibody (150 μg) dispersed in PBS, GM (GM-TNFα antibody) containing an equal amount of TNFα antibody, or FMGA containing an equal amount of TNFα antibody (FMGA-TNFα) was given by oral route to colitis C57BL/6J mice on days 1, 3, and 5, respectively. Untreated colitis C57BL/6J mice were used as negative controls. All C57BL/6J mice were allowed unrestricted water and standard laboratory diet before oral administration of the therapeutic material. All C57BL/6J mice were euthanized on day 7, and the entire colon was resected for length measurement and H&E staining.

Acetic acid induced colitis in beagle dogs

All beagle care, feeding, and research procedures are carried out in accordance with the guidelines for the care and use of laboratory animals. Animal ethics numbers are 2023044, 2025005, 2025011, and 2025014. The beagles were randomly divided into five groups, and 25 beagles acclimated for 1 week before inclusion in the study. The five healthy beagles were not treated and served as the healthy positive control group. At day 0, the intestines of the beagles were first cleaned by enema before the UC model was established in 20 beagles. After anesthesia, a polyethylene catheter was inserted 20 cm from the anus into the colon and slowly injected with 7% acetic acid solution (2 ml/kg⁻¹). The UC model was established by keeping the head low and the tail high immediately after injection, allowing acetic acid to etch the colon. Last, the beagle’s colon was rinsed twice with 100 ml of saline, and the beagle lay still. The state and fecal characteristics of beagle dogs after waking were observed. A day later, the colitis beagles were divided into four groups (control group, EXO group, GM-EXO group, and FMGA-EXO group). The beagles were allowed unrestricted access to water and a standard diet before and after oral tube feeding. Five colitis beagles in the EXO group were given EXO on days 1, 3 and 5, respectively. Five colitis beagles in the GM-EXO group were given GM-EXO at 10 ml/kg on days 1, 3 and 5, respectively. Five colitis beagles in the FMGA-EXO group were injected with FMGA-EXO at 10 ml/kg on days 1, 3, and 5, respectively. Untreated colitis beagles (control group) were used as negative controls. Beagle weight changes were assessed daily during the 8-day experiment. Endoscopy was performed when the beagle was anesthetized on days 0, 1, and 7. On day 5 of treatment, fresh fecal samples were collected and stored in enzyme-free sterile tubes at −80°C for subsequent analysis of the gut microbiota. On the last day, the beagle was euthanized, blood samples were collected for whole blood group analysis, and the colon and major organs (heart, liver, spleen, lungs, and kidneys) and digestive tract (esophagus, stomach, small intestine, jejunum, and ileum) were removed for histological evaluation and immunofluorescence staining.

In vivo evaluation of intestinal retention of the FMGA-EXO coating through x-ray imaging

Radiocontrast agent (barium sulfate, 0.5 g/ml) was dispersed in the FMGA-EXO and EXO liquid samples. Beagles take FMGA-EXO (10 ml/kg) and EXO (10 ml/kg) orally. X-ray monitoring was conducted to monitor the retention of FMGA-EXO and EXO in the GI tract. To monitor the intestinal retention of FMGA-EXO over time, a series of x-ray images of beagle dogs were taken periodically.

Histological and immunofluorescence analysis

The distal colon tissue was meticulously preserved in 4% paraformaldehyde solution for a period of 48 hours, subsequently encased in paraffin, and finely sliced into sections measuring a mere 5 μm in thickness. Paraffin sections were dewaxed and hydrated. Then, the sections were antigen retrieval with immunohistochemical antigen repair buffer (1×, citrate buffer pH 6.0, Thermo Fisher Scientific), and the sections were incubated with immunostaining sealing solution (Thermo Fisher Scientific) for 1 hour to block nonspecific binding sites. Distal colon sections were stained with H&E and alcian blue (AB) to visualize the structure of colon tissue.

The immunofluorescence staining of colon tissue was performed using antibodies against CD86 (1:100 dilution; M1 marker, Santa Cruz Biotechnology) and CD206 (1:100 dilution; M2 marker, Santa Cruz Biotechnology, ab64693), respectively. After dewaxing and antigen repair of distal colon tissue, the nonspecific binding sites were blocked with immunostaining blocking solution for 1 hour. The tissue sections were incubated with primary antibodies (1:100 dilution) against CD86 or CD206 overnight and secondary antibodies for 1 hour, respectively. Then, the nucleus is stained with DAPI (4′,6-diaminidine 2-phenylindole) (1:500 dilution; Thermo Fisher Scientific). Immunofluorescence images were obtained by using laser confocal scanning microscopy (Leica TCS SP8).

ELISA detects the secretion of related inflammatory factors in colon tissues of mice

Colon tissue was immersed in an optimal volume of radioimmunoprecipitation assay lysate, then delicately homogenized using a tissue homogenizer at a temperature of 4°C, left to stand motionless for a duration of 20 min, subjected to centrifugation at 12,000g for 10 min, following which the resultant supernatant was carefully extracted and preserved at a freezing point of −80°C. The levels of IL-6 and IL-10 in colon tissue were measured according to the instructions.

Statistical analysis

All data were shown as means ± SD. One-way analysis of variance (ANOVA) by Tukey post hoc test was used for statistical analysis between groups. MATLAB (version R2018b) and GraphPad Prism 9.0 were used to analyze the statistical significance of all studies. A value of P ≤ 0.05 was deemed to be statistically significant.

Supplementary Materials

This PDF file includes:

Supplementary Text

Figs. S1 to S40

Tables S1 and S2