Extracellular Matrix-Liposomes: A Novel Vehicle for Tissue Regenerative Wound Healing

The unfolded states of bronectin (FN) subsequently induce the formation of the extracellular matrix (ECM) brillar network, which is necessary to generate new substitutive tissues. Here, we demonstrate that negatively charged small unilamellar vesicles (SUVs) qualify as candidates for FN delivery due to their remarkable effects on the autonomous binding and unfolding of FN, which leads to increased tissue regeneration. In vitro experiments revealed that the FN-SUV complex remarkably increased the attachment, differentiation, and migration of broblasts. The potential utilization of this complex in vivo to treat inammatory colon diseases is also described based on results obtained for ameliorated conditions in rats with ulcerative colitis (UC) that had been treated with the FN-SUV complex. Our ndings provide a new ECM-delivery platform for ECM-based therapeutic applications and suggest that properly designed SUVs could be an unprecedented FN-delivery system that is highly effective in treating UC and other diseases. In addition to performing single-SUV the size distribution of the entire SUV population by using the dynamic light scattering (DLS) method S2). The results showed that the average size of the SUV population increased from 140.3 nm to 178.3 nm and that the curve shifted to the right (larger size), indicating that the size of the SUV had increased after incubation in an FN solution.


Abstract
The unfolded states of bronectin (FN) subsequently induce the formation of the extracellular matrix (ECM) brillar network, which is necessary to generate new substitutive tissues. Here, we demonstrate that negatively charged small unilamellar vesicles (SUVs) qualify as candidates for FN delivery due to their remarkable effects on the autonomous binding and unfolding of FN, which leads to increased tissue regeneration. In vitro experiments revealed that the FN-SUV complex remarkably increased the attachment, differentiation, and migration of broblasts. The potential utilization of this complex in vivo to treat in ammatory colon diseases is also described based on results obtained for ameliorated conditions in rats with ulcerative colitis (UC) that had been treated with the FN-SUV complex. Our ndings provide a new ECM-delivery platform for ECM-based therapeutic applications and suggest that properly designed SUVs could be an unprecedented FN-delivery system that is highly effective in treating UC and other diseases.

Main
Fibronectin (FN) plays an essential role in the wound healing process due to its ability to form an extracellular matrix (ECM) and to regulate cellular activities 1 . In addition to the ability to bind together and form FN brils, FN molecules have various recognition and binding sites in their structure, and these sites favor interactions with various cell types, growth factors, cytokines, and other ECM proteins, such as collagen, laminin, and heparin 2,3,4 . Soluble FN is a folded dimeric protein whose subunits are composed of three types of repeating modules, I, II and III, which are variably expressed due to alternative splicing of RNA 5 . Among the three modules, FN type III (FNIII) is the largest and most common repeat 2,4 . Once in contact with the cell surface, FNIII domains will be recognized by integrin proteins on the cell membrane and subsequently undergo an unfolding process caused by cell-traction forces, which induces FN brillogenesis 3,6 . Unfolding is, therefore, essential in FN functionalization, which is the target of FN-based material fabrication 1,7 . Although the cell binding sites within the FNIII module are exposed on the surface, the active domains, which are known to be phosphorylation sites, are usually hydrophobic. These domains are buried within FNIII modules in the folded state due to hydrophobic interactions and are maintained by intramolecular ionic bonds 8, 9 . Therefore, unfolding FN is crucial not only for enhancing the elasticity of bronectin brils but also for exposing the buried sites that are known to display enzymatic activities and to regulate many cellular activities 10 .
The use of soluble, compact FN has been successful in improving wound-healing treatments. Some examples of wounds considered thus far are diabetes-and radiation-induced cutaneous wounds in rats or mice 11,12 . Corneal epithelial wounds have also been reported to heal more quickly after having been treated using eye drops containing a soluble FN and hyaluronan combination 13 14 . In 2018, Christophe et al. adopted rotary jet spinning to produce unfolded FN bers and con rmed that its use effectively enhanced wound healing 15 . These studies suggest that for higher wound-healing e ciency, FN used in tissue regeneration should undergo certain procedures to induce changes in its conformation.
Some methods for doing so include physical binding on different substrates, modifying surfaces with chemical groups, or using gold nanoparticles 16,17,18,19 . Although unfolded FN has been proven to have improved functions over compact FN, the methods for unfolding FN that have been reported so far have had limited clinical applications due to the complex processes required for the fabrication and chemical modi cation of the substrate.
In this study, to tackle the above issues, we demonstrate a potential FN-unfolding platform using negatively charged, small unilamellar vesicles (SUVs) for FN delivery. These SUVs are simply produced by using a unique composition of lipids so that the surface of the SUV not only enhances direct binding but also induces the unfolding of FN. The effects of FN delivered by our SUV system were investigated using a variety of cell-based assays to evaluate all aspects of cellular activities, including growth, differentiation, and migration. An in vivo wound-healing model of ulcerative colitis (UC) was adopted to evaluate the su ciency of the unfolded FN that was delivered. Collectively, our ndings indicate that negatively charged SUVs represent a new platform for functional FN delivery.
For critical cellular functions to be accessed, compact FN must be unfolded to expose buried domains 20 .
Hence, our rst effort was to determine the composition of lipids that not only enhanced the binding e ciency of FN but also induced conformational changes of FN prior to its delivery to cells (Fig. 1a). The optimized lipid composition of SUVs was found to be zwitterionic:net negative:cholesterol at a ratio of 3:2:1. An appropriate ratio of 1,2-dioleoyl-sn-glycero-3-phospho-L-serine (DOPS) resulted in a net negative charge on the surface of the SUV. The negatively charged SUVs had strong interactions with FN compared to those with only zwitterionic lipids (Fig. S1). An analogous giant unilamellar vesicle (GUV) system was developed with the same lipid composition but a hundredfold increase in size (20-30 µm on average) (Figs. 1b and c). The analogous GUVs allowed us to directly observe the binding by using confocal imaging. FN tended to form insoluble aggregates when it was incubated with neutral vesicles ( Fig. 1b) but was attached evenly on the surfaces of negatively charged GUVs (Fig. 1c). The strongly favorable and selective binding of FN to a negatively charged lipid is obvious in phase-separated GUVs with two lipid phases (Fig. S1a). The liquid order (L o ) phase contained zwitterionic sphingomyelin, and the liquid disorder (L d ) phase contained DOPS and other unsaturated lipids, constructing the negative portion of the membrane. Under the same experimental conditions, FN was observed only in the L d phase (green uorescence) and not in the L o phase of the same GUV (Fig. S1b). This result clearly con rmed that a negatively charged vesicle system selectively enhanced FN binding.
Evaluating the binding of FN on SUVs with an appropriate size for cell delivery was much more sophisticated because SUVs (average radius of 200 nm) are remarkably smaller than GUVs (average radius of 20 µm). Images obtained from cryogenic transmission electron microscopy (Cryo-TEM) showed that the SUVs incubated with FN had a membrane that was signi cantly thicker (thickness d = 8.3 ± 0.6 nm) than the membrane containing only lipids (d = 4.4 ± 0.6 nm) (Figs. 1d and e). The thickness of the membrane comprises the lipid membrane itself, which is approximately 4-5 nm, and an additional FN layer, if any, on the membrane. Therefore, this result implied that FN indeed bound to the SUVs, which was observed as an increase in the membrane thickness. In addition to performing single-SUV thickness measurements, we measured the size distribution of the entire SUV population by using the dynamic light scattering (DLS) method (Fig. S2). The results showed that the average size of the SUV population increased from 140.3 nm to 178.3 nm and that the curve shifted to the right (larger size), indicating that the size of the SUV had increased after incubation in an FN solution.
In addition to enhancing the binding of FN on the surface, negatively charged SUVs induced conformational changes. The stretching of FN was measured by using uorescence resonance energy transfer (FRET) with FN labeled with donor-acceptor dyes (FN-DA). Four molar (4 M) guanidinium chloride (GdnHCl), one of the strongest denaturants, was used as a positive control 21 (Fig. S3). Figure 1f shows that the I A /I D ratio of FN-DA incubated with SUVs was comparable to that of FN-DA incubated in 4-M GdnHCl (0.38 ± 0.06 and 0.34 ± 0.13, respectively) and was signi cantly lower than that of FN-DA incubated in phosphate buffered saline (PBS) used for SUV hydration or sucrose used for GUV hydration (1.18 ± 0.2 and 1.19 ± 0.05, respectively). FRET was also accomplished on GUVs coated with FN-DA; the I A /I D ratio of FN-DA was lower after incubation with GUVs (from 0.92 ± 0.11 to 0.18 ± 0.06 after incubation) (Figs. 1g and h). This result veri ed that stretching of FN had also occurred on the surfaces of GUVs. Taken together, the negatively charged SUV system is a suitable vehicle for delivering FN in an active, unfolded conformation.
FN has been utilized in cell and tissue cultures to strengthen cell attachment, proliferation, and migration 17,19,22,23,24 . We hypothesized that the conformational changes induced by binding to negatively charged SUVs would enhance the above cellular functions of FN to a greater extent than its compact form. We performed in vitro experiments using human neonatal dermal broblast (HDFn) cells, as they play essential roles in tissue repair. The fate of FN after having been delivered to HDFn cells was probed with either green (FN-SUV) or red (FN) uorescent indicators. In both groups, FN brils were found to colocalize with their membrane receptor, integrin α5 (Fig. 2a). However, more FN brils were produced when FN was delivered by using SUVs. We found an interesting event in which a clump of FN-SUVs rapidly burst and expanded when it came into contact with the HDFn cell surface (Fig. 2b). This event was not observed when using FN without SUVs (data not shown), indicating that SUVs promoted the interaction of FN with the cell surface. As we observed that SUVs unfolded FN in advance of the interaction with the cell membrane, we predicted that this process would subsequently increase the adhesion of cells to the culture substrate. Fig. 2c shows that trypsinized HDFn cells had limited attachment to the culture dish. However, when delivered by using the FN-SUV system, the number of active broblasts (seen as spindle-shaped cells with long lamellipodia) increased signi cantly 25 . A quantitative comparison of cell attachment among the experimental groups was performed by calculating the total area of attached cells, as shown in Fig. 2d and Video S4. Moreover, the total cell surface area covered by FN was signi cantly higher when FN was delivered by SUV, implying enhanced FN binding to the cells (Fig. 2e). Although FN promoted cell attachment, stretched FN made the surfaces of GUVs unfavorable for bacteria, particularly Staphylococcus aureus (S. aureus), an organism that often causes opportunistic infections on skin and leads to tissue damage, as described in a previous study 26 . S. aureus was incubated with different types of GUVs (Fig. S4). A large number of bacteria (white) were found on collagen-coated GUVs (COL-GUVs). In contrast to the case with the bare lipid GUVs, the bacteria were hardly present on the FN-coated GUVs, suggesting that using the stretched FN as a wound healing agent had advantages in preventing unwanted bacterial infection.
When tissue is wounded, released cytokines attract broblasts from the surrounding area 1 . During this time, broblasts are activated and migrate towards the wounded area with the recruitment of ECM proteins including FN 27,28 . We tested whether the stretching of FN would promote the migration rate of broblasts, as it has been shown to enhance ECM formation. Figure 3a shows representative images cut from Videos S5-7, illustrating the movements of HDFn-GFP cells as a function of time. Single cells were tracked, and the average migration speeds (µm/min) were compared among groups. As we expected, the migration speed increased when SUVs (green circles) were used for FN delivery (Figs. 3a and b). This enhancement may result from the formation of a denser FN matrix, which facilitates cell adhesion and migration. The expansion of the FN matrix among groups over time can be seen clearly in Fig. S5 and Video S11.
In vitro scratch assays were used to assess the cell migration speed to evaluate cell migration 29 . To ensure the reproducibility of the experiments, we set up an incubation chamber in the confocal microscope system and continuously observed a speci c wounded area for at least 16 h. Representative images at 0 h and 12.5 h for each group are shown in Fig. 3c; the corresponding time-lapse images are shown in Videos S8-10. The results revealed that among the experimental groups, the gap closed at the highest rate in the FN-SUV-treated group, with 60.3 ± 4.3 % of the gap area being covered by HDFn cells after 15 h compared to only 33.5 ± 4.2 % in the FN-treated group (Fig. 3d). Insigni cant proliferation among all groups was observed during the rst 24 h, which showed that the cell migration assays were unbiased, with minimal contribution from cell proliferation (Fig. S6).
The healing effects of FN in vivo have been described previously by using a wide range of wound models in different species, such as rabbits, mice and guinea pigs 12,30,31 . Existing evidence from previous studies indicates that in in ammatory bowel-related diseases, such as ulcerative colitis (UC) and Crohn's disease, the level of FN in the plasma is decreased signi cantly 7,32,33 . In this present study, the healing performance of FN-SUVs was further assessed in vivo using an UC model for rats. An approved protocol using 4% acetic acid was applied to induce UC in the colon, followed by a 10-day treatment strategy with either FN or FN-SUVs (Fig. S7). After treatment, the entire colons of the rats were removed to compare changes in morphology, wet weight and various tissue damage markers. As shown in Fig. 4a, the FN-SUVtreated group showed remarkably thinner and smoother colon walls with less mucus (less tissue damage) than the FN-treated group. In addition, FN-SUV-treated colons had signi cantly lower wet weights than those treated with FN alone, indicating that SUV-FN treatment reduced in ammation and swelling in the colons.
Colon tissues after either FN or FN-SUV treatment were further analyzed semi-quantitatively for in ammatory markers, including tumor necrosis factor (TNF)-α, interleukin (IL)-1β, IL-6, cyclooxygenase (COX)-2 and inducible nitric oxide synthase (iNOS). However, the differences in the levels of TNF-α, IL-1β, and IL-6 expression between the FN and FN-SUV groups were insigni cant (Fig. S8), whereas COX-2 and iNOS expression was downregulated in rats treated with FN-SUV, suggesting that the unfolding of FN via the SUV may expose the domains that communicate with cells via these two pathways (Figs. 4a and b) 34,35 . The inhibition of COX-2 and iNOS expression in our in vivo model clearly showed that in ammation was effectively decreased in the experimental UC model when the FN-SUV treatment was administered 36 .
Many attempts have been made to unfold FN prior to it reaching the cell. However, an appropriate approach to unfolding FN for tissue regeneration applications is still a challenge. The unfolding of FN requires the breakage of intramolecular bonds and hydrophobic interactions, which can be induced by using either chemicals 14,21 or mechanical forces 8, 15 . The pH or salt concentration can induce the stretching of FN, but this reversible change is hard to maintain for cell delivery, and the solution must be similar to the physiological environment. The improper removal of a denaturant will otherwise potentially result in negative effects on live cells and tissues 21 .
The adsorption and unfolding facilitated by chemical modi cations raise a risk of cytotoxicity 16,17,18,19 .
On the other hand, approaches using mechanical forces to unfold FN usually exploit solid substrates and polymers with low biocompatibility. The most promising approach might be to produce nano bers from soluble FN by using mechanical forces 15,37 . However, FN nano ber production remains challenging, as the use of the fabrication technique may not always be possible in laboratories. Nonetheless, FN nano bers are more appropriate for dressing materials and cutaneous wound treatment than for the treatment of bowel diseases. Despite major efforts to deliver ECM or FN to damaged tissues via surfacemodi ed substrates, the fabrication of these materials remains very complex, and these substrates usually have to be removed from the body after having been applied, which is not appropriate for deep, closed wounds 38, 39 . In contrast, negatively charged SUVs are composed of phospholipids and cholesterol, which are highly biocompatible and have negligible cytotoxicity 40 . Accordingly, in our novel system, we managed to develop a negatively charged SUV system that has negligible toxicity 40,41 . In addition, the preparation of self-assembled SUVs in aqueous solution followed by the autonomous binding and unfolding of FN was much simpler and afforded higher yields than any other method.
The adsorption and unfolding of FN on liposomes were once reported by Micheal et al. as an enhanced drug delivery approach using liposomes that can avoid the rapid uptake of the reticuloendothelial system 42 . Both gel-phase and liquid-phase liposomes were considered for FN selective binding, and only liposomes composed of lipids that had high melting temperature (T m ) or existed in the gel phase at room temperature, such as DPPC (T m = 41°C) or DSPC (T m = 55°C), were observed to bind and unfold FN 42 .
Interestingly, we have su cient experimental evidence to prove that FN is able to bind and unfold on liquid-phase SUVs composed of lipids with T m < 0°C under the condition that a negatively charged lipid (DOPS) is added. Due to the low T m values of the lipids used, the fabrication of negatively charged SUVs is simpli ed and can be performed at room temperature. Thus, the negatively charged SUV system has signi cant advantages over previously used materials in terms of fabrication, cytotoxicity, and biocompatibility.
Collectively, we con rmed that a negatively charged SUV can effectively unfold FN bound to its surface. Subsequently, the SUVs enhanced the effects of FN on HDFn migration and proliferation. In vivo experiments showed that rats with UC that had been treated with FN-SUVs improved faster than rats treated with only soluble FN. The abundant amount of FN in the plasma is an advantage of using FN as a potential tissue regeneration material 1,7,32 . FN also possesses an exceptional property for wound healing, as stretched FN disrupts the binding of S. aureus, a common cause of infections in humans 26 .
We conclude that negatively charged SUVs are ideal vehicles for functional FN delivery. The system shows the capability to carry and unfold FN with high biocompatibility and safety and can be used as a platform for protein-small molecule drug incorporation, especially for drugs that accelerate wound healing. Further studies are expected to discover the mechanisms underlying the e cacy of the SUV-FN delivery system.

Preparation of small unilamellar vesicles (SUVs) with negatively charged surfaces
Multilamellar liposomes were rst prepared by blending DOPC:DOPE:DOPS:cholesterol in a ratio of 2:1:2:1 to make a mixture with a total concentration of 5 mg of lipids per ml. One milliliter of the lipid blend in a black glass vial was blown dry using N 2 gas for 2 h to completely remove the solvent. The dried lipids were then rehydrated with 1X PBS (pH 7.4) by using a vortex mixer at room temperature for 60 minutes. The lamellarity of the liposomes was reduced by using 5 cycles of freezing in liquid nitrogen and thawing in a water bath at 40°C to break down the outer bilayers. Finally, the vesicles underwent extrusion by using a mini-extruder set (Avanti, Inc) to pass them through a 200-nm membrane lter 11 times.

Preparation of giant unilamellar vesicles (GUVs)
A lipid mixture with a composition of either DOPC:DOPE:DOPS:cholesterol (2:1:1:1 molar ratio) for negatively charged GUVs or DOPC:DOPE:cholesterol (2:1:1) for neutrally charged GUVs was prepared at a concentration of 1 mg/ml. Approximately 30 µl of the lipid mixture was deposited on ITO glass (30-Ohm sheet resistance) and quickly dried by using owing N 2 gas. The lipids were continuously dehydrated in a vacuum desiccator for approximately 1 h. A silicon ring with a thickness of 1 mm was inserted between two ITO glass slides to make space for a hydrate solution. For observations under a light microscope, the lipids were rehydrated with an inner solution containing 200 mM sucrose and 2 mM MES (pH 4.2) and were later diluted with an outer solution containing 200 mM glucose and 2 mM MES (pH 4.2). A bronectin (FN) solution that had been previously diluted in deionized water was added to make an FN-GUV solution at a nal concentration of 10 µg/ml, followed by incubation at 37°C overnight before performing confocal microscopy.
FN conjugation for uorescence resonance energy transfer analysis Fibronectin (FN) was subsequently labeled with Alexa-546 C5 maleimide on its cysteine residues and with Alexa-488 NHS succinimidyl ester on its lysine residues. Lyophilized FN diluted with PBS to a concentration of 1 mg/ml was denatured in the chaotropic agent guanidine hydrochloride (GdnHCl, 8 M) at a ratio of 1:1 (v/v) and incubated with Alexa 546 maleimide at a molar ratio of 1:30. Excess uorescence dye was removed through dialysis by using a slide-A-lyzer dialysis cassette with a 10,000 Dalton MW cut-off in 0.1 M sodium bicarbonate (NaHCO 3 ) at pH 8.5. The buffer was changed twice every two hours before being stored overnight at 4°C. The initial conformation of labeled FN was con rmed using spectro uorometry. Afterwards, the labeled FN was incubated for one hour with Alexa 488 NHS succinimidyl ester at a ratio of 1:70 (mol/mol). Doubly labeled FN with a donor and an acceptor dye (FN-DA) was separated from the free dyes by using a PD-10 column (pre-equilibrated with PBS). The ratio of the donor intensity to the acceptor intensity per FN dimer (I A /I D ) was determined by measuring the uorescence of FN-DA at 280, 496, and 556 nm. For long-term storage, FN-DA was stored in 10% glycerol at -80°C.

Characterization of FN-SUV incorporation by using cryogenic transmission electron microscopy (cryo-TEM)
FN was incubated overnight with preformed SUVs at a ratio of 1:5 (w/w). Immediately after incubation, samples of either SUVs or FN-SUVs were preserved by vitrification and supported by Quantifoil (R 2/2, 400-mesh holey-carbon grid, Quantifoil Micro Tools). Vitrified samples were prepared by applying a 3-µL drop of sample suspension to a cleaned grid, blotting away the excess with lter paper, and immediately plunging the grid into liquid ethane. Grids were stored under liquid nitrogen until transferred to the transmission electron microscope for imaging. Electron microscopy was performed using a Talos L120C electron microscope (FEI, Hillsboro, OR, USA) operating at 120 kV. Vitreous ice grids were transferred into the electron microscope by using a cryostage that maintained the grids at a temperature below -170°C. An image of each grid was acquired on multiple scales to assess the overall distribution of the specimen. Images were collected using a 4k × 4k Ceta CMOS camera at a search magni cation of 5000 and a nal magni cation of 29,000 to 280,000 at an underfocus of ~ 2.5 μm. After potentially suitable target areas for imaging at lower magni cations had been identi ed, pairs of highly magni ed images were acquired using the UI V1.6 Talos software. ImageJ software was used to analyze the intralamellar distances in the original, high-resolution images collected using the microscope (4096 × 4096 pixels).

Fluorescence resonance energy transfer (FRET) analysis
The uorescence intensity of FN-DA was measured using a standard uorescence spectrophotometer (Hitachi F-7000) with excitation of the donor at 488 nm. The positive controls were measured using a series of GdnHCl concentrations from 0 to 4 M. The FRET measurement over a range of denaturant concentrations was used to plot the correlation of the FRET e ciency with the degree of FN unfolding. FN-DA was incubated overnight with preformed SUVs at a ratio of 1:5 (w/w) with gentle rocking to enhance the binding. Then, cryo-TEM and FRET were exploited to characterize the incorporation and unfolding of FN on the SUVs. The background uorescence was measured using SUVs in solution, and the negative control was measured using FN-DA in PBS without SUVs at the same concentration as in the FN-SUV sample.

Colocalization, attachment area and bril formation imaging
HDFn cells were cultured in DMEM (FBS 10%, 1% Pen-Strep). For integrin observation, FITC-conjugated integrin 5 antibody was used. FN was diluted in PBS and used at a concentration of 10 μg/ml. For cell attachment experiments, HDFn-GFP cells were detached using trypsin-EDTA when the con uence had reached 80% and were then transferred to a new culture dish, treated with non-labeled FN and imaged every hour in an on-stage incubator. For quanti cation of the area of the FN brils, HDFn-GFP cells were treated with lissamine-rhodamine-labeled FN and imaged for 25 min.

In vitro scratch assay
In migration experiments, GFP-HDFn cells were cultured in T75 asks to reach a con uence of 80-90% and were then transferred to 35-mm dishes for continuous culturing over 24 h with 10% FBS for attachment. When cells had formed a con uent monolayer, the medium was replaced with low-serum DMEM (0.4% FBS) every 24 h to minimize the effects of proliferation on the migration speed and to avoid unexpected apoptosis 43 . The distance that a single cell moved was tracked using ImageJ.
A scratch mimicking a wound was created using a p200 pipette tip by scraping a straight line across the center of the dish. Cell debris was then washed away gently with 2 ml of low-serum medium before being Induction of ulcerative colitis by using acetic acid Ulcerative colitis (UC) was induced by using acetic acid, as previously described 44 . The rats were anesthetized with Zoletil 50 ® (10 mg/kg; Vibac Laboratories, Carros, France). Animals were fasted for 48 h for UC preparation. A stainless-steel feeding needle (round tip, 16 gauge) was then carefully inserted into the colon of a rat with the tip being 8 cm proximal to the anus. Then, 4% acetic acid (Sigma Chemical Co., St. Louis, MO, USA) in 0.9% saline (1.0 ml) was injected into the lumen of the colon. The control group underwent intracolonic injection with saline instead. The rats were maintained in a supine Trendelenburg position for 5 min to prevent leakage of the injected acetic acid solution.

UC Treatment
The treatments were administered three days after acetic acid injection. The dosages for the groups were as follows: mesalamine (4 g/100 ml, Asacol Enema, Tillotts Pharma, AG, Swiss), FN-SUVs (20 µg/ml FN, 100 µg/ml SUVs), and FN (20 µg/ml). For drug administration, the rats were con ned in an acrylic cylinder (diameter: 6 cm, length: 20 cm, height: 10 cm) without anesthesia. A stainless-steel feeding needle (round tip, 16 gauge) with a syringe was then carefully inserted through the anus into the damaged site. Then, each drug (1.0 ml) was injected into the lumen of the colon once a day for 10 days after induction of UC.

Western blotting
Colon tissues were homogenized on ice, lysed with lysis buffer, and then centrifuged at 14,000 rpm for 30 min. The amount of protein in each sample was quanti ed using a Bio-Rad colorimetric protein assay kit (Bio-Rad, Hercules, CA, USA). Sodium dodecyl sulfate-polyacrylamide gels were used to separate proteins (30 μg), after which the separated proteins were transferred onto nitrocellulose membranes. Goat COX-2 antibody (1:1000; Santa Cruz Biotechnology, Santa Cruz, CA, USA), mouse iNOS antibody, and mouse βactin antibody (1:1000; Santa Cruz Biotechnology) were used as the primary antibodies. Horseradish peroxidase-conjugated anti-mouse antibodies (1:2000; Vector Laboratories, Burlingame, CA, USA) for iNOS, β-actin, and COX-2 were used as secondary antibodies. Membrane transfer was conducted at 4°C by using a cold pack and prechilled buffer. An enhanced chemiluminescence (ECL) detection kit (Santa Cruz Biotechnology) was used for band detection. Molecular Analyst™ (version 1.4.1, Bio-Rad) was used to analyze the detected bands.     FN-SUV treatment improves the healing process of acid-induced ulcers in rat colons. a, Hypothesized schematic of FN-SUV treatment in the rat colon. b, Gross morphologies of the intestines removed from the rats after 10 days of treatment. In the control group, the healthy colon wall is thin and smooth with no mucus. In the colitis-induced groups, the wall of the colon is swollen and lled with blood and mucus.
Various degrees of colitis can be seen among the different treatments. c, Colon wet weight of the control and colitis-induced groups as described in (a) (n = 10 per group). P values were calculated using Student's t tests. d, e, Expression levels of the in ammatory indicators COX-2 (n = 5 per group) (c) and iNOS (n = 6 per group) (d). P values were calculated using one-way ANOVA. c, *P = 0.0003, **P = 0.009. d, *P = 0.0006, **P = 0.0034. e, *P = 0.0009, **P = 0.0001. c-e, Error bars indicate mean ± SDs.