Endoscopically injectable and self‐crosslinkable hydrogel‐mediated stem cell transplantation for alleviating esophageal stricture after endoscopic submucosal dissection

Abstract Esophageal stricture after extensive endoscopic submucosal dissection impairs the quality of life of patients with superficial esophageal carcinoma. Beyond the limitations of conventional treatments including endoscopic balloon dilatation and the application of oral/topical corticosteroids, several cell therapies have been recently attempted. However, such methods are still limited in clinical situations and existing setups, and the efficacies are less in some cases since the transplanted cells hardly remain at the resection site for a long time due to swallowing and peristalsis of the esophagus. Thus, a cell transplantation platform directly applicable with clinically established equipment and enabling stable retention of transplanted cells can be a promising therapeutic option for better clinical outcomes. Inspired by ascidians that rapidly self‐regenerate, this study demonstrates endoscopically injectable and self‐crosslinkable hyaluronate that allows both endoscopic injection in a liquid state and self‐crosslinking as an in situ‐forming scaffold for stem cell therapy. The pre‐gel solution may compatibly be applied with endoscopic tubes and needles of small diameters, based on the improved injectability compared to the previously reported endoscopically injectable hydrogel system. The hydrogel can be formed via self‐crosslinking under in vivo oxidative environment, while also exhibiting superior biocompatibility. Finally, the mixture containing adipose‐derived stem cells and the hydrogel can significantly alleviate esophageal stricture after endoscopic submucosal dissection (75% of circumference, 5 cm in length) in a porcine model through paracrine effects of the stem cell in the hydrogel, which modulate regenerative processes. The stricture rates on Day 21 were 79.5% ± 2.0%, 62.8% ± 1.7%, and 37.9% ± 2.9% in the control, stem cell only, and stem cell‐hydrogel groups, respectively (p < 0.05). Therefore, this endoscopically injectable hydrogel‐based therapeutic cell delivery system can serve as a promising platform for cell therapies in various clinically relevant situations.

gel system with excellent biocompatibility and tissue adhesiveness. 9 This system can be applied to medical fillers for volumetric augmentation and wrinkle correction by the superior injectability and in vivo mechanical stability. 10 The injectability should be more critical for endoscopy which consists of long and narrow tubing with very small gauge needle. 11,12 Thus, we predicted that our injectable hydrogel system could be effectively introduced in the clinical setting via endoscopy equipment and that incorporating ADSCs into the hydrogel system might enhance the retention and therapeutic effects of ADSCs. This study aimed to investigate whether endoscopic injection of an ADSC-hydrogel mixture can prevent post-ESD esophageal strictures.

| Synthesis of endoscopically injectable and self-crosslinkable hyaluronate
Endoscopically injectable and self-crosslinkable hyaluronate (EISCH) was synthesized by conjugating pyrogallol (PG) group to hyaluronate via carbodiimide chemistry, according to previously reported procedures, with slight modifications. 10

| Preparation and material characterization of EISCH hydrogel
To induce in vivo-mimetic crosslinking of the EISCH, the lyophilized conjugate was dissolved in neutral PBS (pH 7, Sigma-Aldrich) and evenly mixed with different concentrations of horseradish peroxidase (HRP; Sigma-Aldrich) solution. The final concentration of EISCH adjusted to 2 w/v%, and that of HRP was set to predetermined concentrations (0.06, 0.6, and 6 U/mL, respectively). For in vivo crosslinking, the pre-gel solution was prepared by dissolving lyophilized conjugate in PBS at a concentration of 2 w/v% and injected directly into desired tissues in vivo. The formed hydrogel via in vivo crosslinking was retrieved for observation and further analysis. The rheological analyses were performed using a rheometer (MCR 102, Anton Paar, Ashland, VA, USA). Viscoelastic properties of the EISCH hydrogels were examined by measuring storage (G 0 ) and loss modulus (G 00 ) in a frequency sweep mode at a frequency range of 0.1 to 1 Hz. The elastic modulus of the hydrogel was determined by calculating the average of storage moduli of each hydrogel at 1 Hz (n = 3-4). The elasticity (tan δ) of the hydrogels was determined by calculating the ratio of the loss modulus to storage modulus (G 00 /G 0 ) at 1 Hz (n = 3-4).
The images of scanning electron microscope (SEM) were obtained using a field emission scanning electron microscope (FE-SEM; JEOL-7001F, JEOL Ltd., Tokyo, Japan). The samples for SEM were prepared by lyophilizing the hydrogels formed by each condition and coating them with platinum (coating regime: 20 mA current, 1 min) using a Cressington sputter coater 208HR (Cressington Scientific Instruments, Watford, UK).

| Injectability test of EISCH
The injectability of EISCH was evaluated by comparing the force needed for injecting EISCH with that of the previously reported endoscopically injectable shear-thinning hydrogel based on alginate-Laponite (Alg-LP). 13 The Alg-LP hydrogel was prepared by mixing 0.2% of sodium alginate (Pronatal LF10/60, FMC Biopolymer, Philadelphia, PA, USA) solution with 2 mg/mL of Laponite (LAPONITE ® XLG, BYK-Chemie GmbH, Wesel, Germany). The viscoelastic properties of Alg-LP hydrogel were measured in the same way mentioned above. The EISCH conjugate dissolved in PBS at a concentration of 2 w/v% and the Alg-LP hydrogel were individually loaded into a 1 mL syringe equipped with different needle gauges (23, 25, 27, or 29G). The extrusion force was measured by universal testing machine with ILC-50 N load cell (UTM, Mecmesin, MultiTest 2.5-i, Slinfold, West Sussex, UK) (n = 3-8). The plunger of syringe loaded with each material was compressed by a probe equipped on the UTM with a jig for syringe fixation, and the probe was moved at a rate of 3.5 mm/s. The break loose force and dynamic glide force was defined as the peak force required to start the movement of plunger and the average force required to maintain the movement of plunger after the initiation, respectively. The injection force was evaluated using the samples from different batches of material synthesis or preparation. The injectability and selfcrosslink-ability after injection were also checked by loading and injecting the pre-gel solution (2 w/v% of EISCH solution) using a 23-gauge endoscopic injection needle (NM-610 U-0423, Olympus, Tokyo, Japan).

| In vitro examination of biocompatibility of EISCH hydrogel and paracrine effects
Human ADSCs (StemPro ® Human Adipose-Derived Stem Cells, Invitrogen, Carlsbad, CA, USA) were cultured using MesenPRO-RS™ medium (Invitrogen). To evaluate the biocompatibility of EISCH hydrogel in vitro, ADSCs were encapsulated in the hydrogels (1.0 Â 10 6 cells per 100 μL of hydrogel) and stained using Live/Dead viability/cytotoxicity kit (Invitrogen) following the manufacturer's protocol at 1, 3, 7, and 14 days of cell culture. Then, the stained ADSCs were observed using a fluorescence microscope (IX73, Olympus), and the ratio of viable cells (green) to dead cells (red) was quantified by manually counting the number of cells in each condition from the acquired images (n = 10). In Live/Dead analysis, the images for quantification were acquired from different samples through separate cell experiments. The cell viability was also examined after injecting the cell-loaded pre-gel solution using a 1 mL syringe equipped with different gauge of needles (23,25,27, and 29G). For investigation on paracrine effects of stem cells encapsulated in EISCH, ADSCs (8.0 Â 10 5 cells per 80 μL of hydrogel) were encapsulated in the hydrogels formed via two different in vivo-mimetic crosslinking using 0.6 and 6 U/mL of HRP. Each hydrogel containing ADSCs was incubated with MesenPRO-RS™ medium, and the medium used for culture was collected at pre-determined time points (2, 4, 7, 10, 14 days after encapsulation). The retrieved medium was used for quantifying the factors released from the encapsulated cells (n = 3-4), and the concentration of each factor (Vascular endothelial growth factor [VEGF] and Interleukin 10 [IL-10]) was measured using a human VEGF ELISA kit and human IL-10 ELISA kit (R&D Systems, Minneapolis, MN, USA) following the manufacturer's protocol. Independent material and cell samples were used for the ELISA analysis.

| Confirmation of in vivo self-crosslink-ability of EISCH
The experimental protocol for the mouse experiments was approved by the Institutional Animal Care and Use Committee (IACUC) of Yonsei University (approval number: IACUC-A-201908-946-01). To confirm in vivo self-crosslink-ability of EISCH, the pre-gel solution prepared by dissolving the conjugate at a concentration of 2 w/v% was loaded into a syringe with the needle narrower than clinically used endoscopic needle (26-gauge needle), and then injected into subcutaneous tissues of mouse. The injected region was visually observed both immediately and 1 day after injection. Then, the mouse was sacrificed, and the injected region was surgically opened for observing the in vivo-crosslinked EISCH hydrogel. Subsequently, the hydrogel with adjacent tissues was removed together and used for further rheological and histological analyses.

| Cell transplantation test in a mouse model and histological analyses
To preliminarily test cell transplantation, ADSCs were labeled using cal animal experiments using pigs. This is because the growth rate of female pigs is slower than that of male pigs, allowing for observation for a long period of time. In particular, the size of the esophagus and stomach of 20-25 kg pigs is most similar to that of humans. They were fasted for 24 h before the procedure; however, water intake was allowed ad libitum. This study followed the ARRIVE reporting guidelines. Pigs were housed singly in cages and cared for according to the Guide for the Care and Use of Laboratory Animals. In detail, a sufficient space to turn or move freely was provided, and suspended flooring was used to prevent contamination by urine and feces. 14 Pigs were randomly assigned to different three groups (control group, ADSC group, and ADSC-EISCH group, n = 5). To minimize the unnecessary use of animals, five pigs were used in each group

| Endoscopic injection of ADSCs and ADSC-EISCH mixture after ESD
Immediately after ESD, the labeled ADSCs (1.0 Â 10 6 cells per 100 μL) were resuspended in PBS (ADSC group) and the pre-gel solution (2 w/v% of EISCH solution) (ADSC-EISCH group) and injected into residual submucosal tissues using a syringe with a 23-gauge needle. Between 0.2 and 0.5 mL of the ADSC-loaded EISCH solution per dose (total volume, 2 mL) was injected evenly throughout the lesion. Injecting solutions into the remaining submucosa is difficult and dangerous because the submucosa is very thin. Therefore, prior to injection, careful examination of the ESD site is essential. The first injection was done at the point where the remaining submucosal layer was clearly visible. Once the first injection was done, the submucosal layer of the injection area got swollen so that subsequent injections could be done safely at the margins of the first injection. In this way, the injection site was gradually enlarged and was able to cover the ESD site. It is also important to approach the needle diagonally, not vertically, into the submucosa for effective and safe injection. For postoperative care, all pigs were fed a liquid diet 24 h after ESD and a semi-solid diet the following day. Mucosal constriction rate (%) ¼ 1 À length of the short axis at the site of maximal constriction ð Þ = length of the short axis at the normal mucosal site ð ½ on the upper side þ length of the short axis at the normal mucosal site on the lower sideÞ=2 Â 100: To prevent selection bias, the endoscopists, radiologists, and pathologists have been blinded to the group assignments to evaluate the level of strictures endoscopically and with fluorography.

| Histological and immunofluorescence analysis of the porcine model
Esophageal tissues were fixed in a 10% formaldehyde/saline solution, embedded in paraffin, and cut into 5-μm sections. Tissue sections were stained with hematoxylin and eosin (H&E) and Masson's trichrome (MT) to examine histopathological changes, including fibrosis. For image-based quantification, each image was acquired from different tissue samples.

| Statistical analysis
All quantitative data were expressed as the mean ± standard deviation (SD). Statistical analysis was performed to determine statistical significance by calculating the p-values using Student's t-test, Mann-Whitney U test for non-parametric values, and analysis of variance

| Material design and characteristics of EISCH for facile and effective cell delivery
The EISCH was designed from a highly autoxidative chemical moiety (PG, red) inspired by 3,4,5-Trihydroxyphenylalanine (TOPA)containing polypeptides in a marine ascidian capable of rapid selfregeneration ( Figure 1a). 9 HA, one of the most widely used extracellular matrix components in biomedical application, was utilized as a polymer backbone of this hydrogel system due to its great biocompatibility and biodegradability. 17,18 As HA is known to perform various important biological functions during tissue regeneration, it has been used to develop injectable hydrogel systems for wound healing. 19,20 The bioinspired polymer was synthesized by chemically conjugating 5-hydroxydopamine containing PG group to the HA backbone via a carbodiimide coupling reaction. The conjugation of 5-hydroxydopamine to carboxyl group in HA backbone was confirmed by 1 H nuclear magnetic resonance (NMR) spectroscopy. The presence of proton peaks at $6.4 ppm indicating aromatic protons in the PG groups demonstrated successful synthesis of EISCH conjugate ( Figure S1a). The DS of EISCH was measured using 1 H-NMR by calculating the integral area ratio of the peak of aromatic protons in the PG groups to that of the methyl groups of the HA backbone. The DS was determined to be approximately 8%. Furthermore, the absorbance spectrum and its peak at 278 nm using UV-vis spectroscopy also showed the presence of PG group within the polymer ( Figure S1b).
Initially, the EISCH liquid may be prepared as a cell-loaded pre-gel solution which is more suitable for easily passing narrow tubing and needle for injection than already formed hydrogel scaffold (Figure 1b).
Immediately after in vivo injection, EISCH undergoes spontaneous self-crosslinking without any complex procedures like mixing additives for crosslinking owing to the auto-oxidative property of the conjugated PG group, which leads to in situ hydrogel formation, enabling effective cell retention as a scaffold. We confirmed that the system is practically applicable to a clinically used endoscopic equipment

| Injectability of EISCH system
The injectability of hydrogels is critical in terms of practical and clinical availability. 21 It would especially be a key factor when the hydrogel is applied as an injecting solution for an endoscopic device that consists of a tubing and needle with a very narrow gauge. The Alg-LP hydrogel system, which is the previously reported hydrogel system with shearthinning-based injectability for endoscopic injection, was demonstrated with the potential to serve as submucosal injection fluids for cushion development to facilitate polyp removal. 13 The Alg-LP system consists of alginate and smectite clays called Laponite, which is prepared as a crosslinked hydrogel form immediately after mixing the components, in contrast with the EISCH system prepared as a noncrosslinked liquid state (Figure 2a). The rheological properties of the Alg-LP showed its state as a viscoelastic solid (G 0 > G 00 ) having elastic modulus over 200 Pa, while the EISCH showed the rheological behavior as a liquid (G 0 < G 00 ) (Figure 2b,c). Thus, the injection force of the Alg-LP to involve inducing shear-thinning should be higher than those of EISCH; otherwise, larger size of needle should be needed for the injection of Alg-LP. Clinically, high pressure during injection may cause severe complications including vascular/neural damage, and the large perforation resulted from using large needle increases the opportunity of bleeding, pain, and subsequent infection. [22][23][24] Both are not only undesirable, but also especially unsuitable for endoscopic injection using small-gauge needle. To comparatively evaluate the injectability, each hydrogel system was loaded in a syringe equipped with 23, 25, 27, and 29 G needles and pressed by the probe with UTM for measurement of the injection force. In the PBS and EISCH groups, the shape and roughness showed in the force-displacement graphs were relatively smooth, which might be attributed to their rheological properties of liquid state (Figure 2d,e). However, those showed in the graphs in the Alg-LP group were relatively rough because the crosslinked Alg-LP hydrogel should be broken via shear-thinning for passing the syringe needle (Figure 2f). Furthermore, the break loose force and dynamic glide force were evaluated (Figure 2g). Although both break loose force and dynamic glide force in the EISCH and Alg-LP groups were higher than those in the PBS group, those in the EISCH group were significantly lower than those in the Alg-LP group (Figure 2h,i). Quantitatively, EISCH required up to $45% less force for injection compared to Alg-LP using a same size of needle. Altogether, the in situ self-crosslinkable EISCH system exhibited superior injectability than the shear-thinning Alg-LP system, which demonstrates it as an excellent candidate for endoscopic injection leading up to surgeons' technical comfort in actual use, improved accuracy of treatment and thus, better clinical outcomes.

| Material properties of EISCH hydrogel
The EISCH hydrogel could be formed via self-crosslinking due to autoxidative property of PG groups (Figure 3a). 10 This oxidative crosslinking can be accelerated and enhanced by various oxidative molecules existing in in vivo oxidative environment. 25 Thus, we used a peroxidase enzyme (HRP) for in vivo-mimetic gelation since the enzyme was considered as one of the key factors for complex oxidative processes in vivo (Figure 3b). [26][27][28][29] The stability of the formed hydrogel was confirmed by a consistently higher storage modulus (G 0 ) than loss  (Figure 3c). Of that, the hydrogel formed by using 6 U/mL of HRP even exhibited a similar level of physical property of the hydrogel made within an in vivo oxidative condition. Interestingly, several previous studies had shown that the concentration of HRP for enzymatic crosslinking mainly affects the gelation rate rather than the physical properties of the formed hydrogels, but the concentration of HRP comparatively influenced the physical properties of EISCH. 30,31 We assumed that peroxidase could be directly involved in and change the path of reaction of the galloyl groups in EISCH, and this phenomenon could also happen in an in vivo oxidative environment. 32,33 In contrast, hydrogels formed with different concentrations of HRP exhibited a reverse trend in the elasticity (tan δ, G 00 /G 0 ), which decreased as the concentration of HRP was higher due to the mechanical reinforcement and stiffening by peroxidase (Figure 3d).  F I G U R E 3 Legend on next page.

| In vitro cytocompatibility of EISCH hydrogel
First, we examined the cytotoxicity of the EISCH hydrogels in vitro, which is an important factor for stem cell transplantation and longterm maintenance in vivo. Cellular viability was assessed through Live/Dead assay, and the ADSCs in the EISCH hydrogel crosslinked by in vivo-mimetic crosslinking using HRP exhibited good viabilities at early time points (Days 1 and 3), which means the oxidative crosslinking of EISCH does not negatively affect the viability of the encapsulated cells (Figure 3f,g). Furthermore, the encapsulated ADSCs were highly viable even until Day 14 (about 90%), which demonstrates that stem cells could be effectively delivered to in vivo environments by utilizing the EISCH hydrogel system and successfully maintain their conditions for a long time within the hydrogel. Cell viability measured after passing the cell-loaded EISCH pre-gel solution through different gauge of needles (23, 25, 27, and 29G) was not significantly different (more than 90% viability in all groups) from control group ( Figure S3a,b). Thus, the stress during cell injection using EISCH seems to have minimal effect on the viability of the ADSCs, which is likely attributed to the superior injectability of EISCH.
Next, we checked the paracrine effect of the stem cells encapsulated in the EISCH hydrogel, which was expected to contribute to therapeutic effects at transplanted regions (Figure 3h). ADSCs are well-known for their ability to secrete various growth factors and cytokines for regeneration and immune modulation. [34][35][36] We examined the paracrine effects of the 3D-encapsulated ADSCs by measuring two well-known proteins released from the ADSCs encapsulated in EISCH hydrogel which were formed by two conditions of in vivo-mimetic crosslinking. VEGF, one of the representative growth factors important to regenerative process, was released from the ADSCs encapsulated in EISCH over 14 days after seeding although the amount gradually decreased (Figure 3i). Furthermore, interleukin 10 (IL-10), one of the typical anti-inflammatory cytokines, was also secreted from the ADSCs encapsulated in EISCH hydrogel, although the released amount of IL-10 also decreased over time (Figure 3j). These results implies that the EISCH hydrogel could provide cyto-compatible microenvironment for the encapsulated ADSCs enough to perform their own roles in paracrine effects.

| In vivo crosslink-ability and biocompatibility of EISCH hydrogel
For investigation of the injectability and in situ self-crosslinking of EISCH, it was injected to subcutaneous tissue in mice. The pre-gel solution was prepared by dissolving the lyophilized EISCH with PBS at a concentration of 2 w/v%. Then the liquid was subcutaneously injected using a 1 mL syringe with 26-gauge needle. Although the solution was slightly viscous, it was easily released through the 26-gauge needle which is narrower than clinically used endoscopic needles such as 22-to 25-gauge needles. The injected solution did not instantly spread all over the surrounding subcutaneous regions due to its viscosity and maintained its own shape, making the injected region look swollen (Figure 4a). Then, the injected pre-gel solution would be crosslinked by in vivo oxidative processes, maintaining its original shape and volume even 1 day after injection.
Interestingly, the location of the formed hydrogel did not change at all, which might have resulted from the tissue-adhesiveness of PG group and its intermediates during oxidative processes. 9 In fact, the hydrogel was successfully crosslinked in the in vivo environment, and stably adhered to the surface of subcutaneous tissues which was confirmed when the injected region was surgically opened after 1-day post-injection ( Figure 4b). Histological analysis also showed that the hydrogel was tightly integrated into the tissue, which was shown in the H&E staining image (Figure 4c

| In vivo stem cell delivery to subcutaneous tissue in mouse using EISCH hydrogel
Investigation of stem cell delivery capability using EISCH hydrogel was conducted before the experiments using a porcine model. To this end, the Qtracker 655 (red fluorescent Qdot 655; QD655)-F I G U R E 3 Self-crosslink-ability and in vitro cytocompatibility of endoscopically injectable and self-crosslinkable hyaluronate (EISCH) hydrogel. (a) Crosslinking mechanism of the EISCH hydrogel via autoxidative coupling of its pyrogallol groups. (b) Photographs of the EISCH solution before gelation, EISCH hydrogel after in vivo-mimetic gelation using horseradish peroxidase (HRP), and EISCH hydrogel retrieved after in vivo gelation in the subcutaneous tissue of a mouse (SQ gelation). (c) Average elastic moduli and (d) elasticity (tan δ [G 00 /G 0 ]) of EISCH hydrogels spontaneously crosslinked without enzyme (0 U/mL group), crosslinked with different concentrations of HRP (0.06, 0.6, and 6 U/mL groups), and crosslinked in the subcutaneous tissue of a mouse (SQ group) (n = 3-4; *p < 0.05; **p < 0.01; ***p < 0.001, independent samples).   Stricture rates were significantly milder in the ADSC and ADSC-EISCH groups than in the control ( p < 0.05 and p < 0.01, respectively) and milder in the ADSC-EISCH group than in the ADSC group ( p < 0.05) (Figure 5c).

| Histological analysis of the esophagus after endoscopic injection of ADSC-loaded EISCH hydrogels
To trace the ADSC-EISCH injection site, histological analysis of ADSC-loaded EISCH-injected esophageal tissues stained with H&Eand MT were performed using samples from pigs at 7 and 21 days after injection (Figure S4a,b). The injected gel was found to be attached to the esophageal tissue on Days 7 and 21. Confocal microscopy also revealed Q-dots (QDs) in the gel, confirming the presence of ADSCs at the injection site ( Figure S4c). The number of observed QDs within and between the ADSC group and ADSC-EISCH group was compared over time ( Figure S5). QD-labeled cells were found in the submucosal layer in both the ADSC and ADSC-EISCH groups until The ADSC group showed milder stricture than the control group; however, it was more severe than the ADSC-EISCH group. (c) Stricture rates determined by calculating the relative ratio of lengths of constricted and normal mucosal sites (*p < 0.05; **p < 0.01).
ADSC-EISCH group than in the ADSC group. To evaluate the effect of ADSC treatment on neovascularization helpful for tissue regeneration, we evaluated VEGF expression levels. ADSC-EISCH group exhibited increased VEGF expression level compared to the control or ADSC groups on Day 21 ( p < 0.05), which might also imply that the transplanted cells in ADSC-EISCH group resided more in the submucosal layer at the time than those in ADSC group (Figure 6a,b). Moreover, the expression of Ki-67, a cell proliferation marker, was significantly decreased in the ADSC and ADSC-EISCH groups compared with the controls at each time point, suggesting that ADSCs inhibited the excessive proliferation in regeneration processes after ESD (Figure 6c). Although no statistical difference was observed between the ADSC and ADSC-EISCH groups, the modulation effect seemed to be continued longer in ADSC-EICSH group than ADSC group until 21 days (Figure 6d). In addition, MPO activity on Day 21 showed no difference between the control and ADSC groups, but it was decreased in the ADSC-EISCH group compared with the control and ADSC groups (Figure 7a  hypothesis can be supported by studies demonstrating similar or even better tissue regeneration effects via infusion or oral ingestion of MSC-derived conditioned medium compared to MSC transplantation. 8,39,40 However, a potential limitation of these previous studies was that the paracrine effects of MSCs did not last long with a single injection of stem cells and that conditioned media required repeated oral intake. Therefore, we predicted that using ADSC-hydrogel mixtures could overcome these shortcomings. Moreover, it would be better if the hydrogels could exist as a liquid before injection and then change to a gel form for a longer period at the initial injection site.
Furthermore, in our in vitro experiment, the released amount of paracrine factors from the ADSCs encapsulated in the EISCH hydrogel crosslinked using 6 U/mL of HRP were slightly higher than that from cells encapsulated in the EISCH hydrogel crosslinked using 0.6 U/mL of HRP. We assumed that the mechanically stiffer environment provided by the hydrogel with 6 U/mL of HRP might facilitate the encapsulated cells to release larger amount of paracrine factors than the relatively softer environment provided by the hydrogel with 0.6 U/mL of HRP. 41,42 In this respect, the ADSC encapsulated in the in vivocrosslinked EISCH hydrogel which has similar or slightly higher mechanical property compared to the EISCH hydrogel formed by in vivo-mimetic crosslinking might be expected to exhibit better paracrine effects.
MSCs can be isolated from various sources such as bone marrow, amniotic fluid, placenta, adult muscle, and corneal stroma as well as adipose tissue. Actually, several studies previously demonstrated that adipose-derived MSCs (ADSCs) show higher levels of paracrine effects than bone marrow-derived MSCs for the treatment of ischemic diseases. 43,44 Thus, we hypothesized that the paracrine secretion of the ADSCs delivered by EISCH contributes to alleviation of esophageal stricture after ESD. Nonetheless, further studies would be required to compare paracrine effects of ADSCs with those of different sources of MSCs.
Finally, a porcine experiment was performed to determine whether it was effective in an environment similar to the clinical setting. The stricture rates were significantly lower in the ADSC and ADSC-EISCH groups than in the control group. In addition, a significantly milder stricture was observed in the ADSC-EISCH group than in the ADSC group. To investigate the underlying mechanisms by which stenosis is reduced, immunohistochemical/fluorescence staining with various markers, such as VEGF, Ki-67, MPO, and α-SMA, was performed, and the results were compared. VEGF is a growth factor that promotes capillary regeneration in ECs. 45 Therefore, blood flow around the injected area could be restored, facilitating oxygen support and nutrient delivery. The presence of α-SMA-positive myofibroblasts are associated with scar contraction. In healing tissues, fibroblasts are activated to become myofibroblasts and participate in the reparative response by secreting large amounts of extracellular matrix proteins, which may be responsible for wound healing. 46,47 In the present study, the ADSC and ADSC-EISCH groups showed decreased α-SMA expression compared with the controls, with the ADSC-EISCH group showing much lower α-SMA expression than the ADSC group. In addition, fewer Ki-67-positive cells were detected in the ADSC and ADSC-EISCH groups than in the control group. Ki-67 is a proliferation marker used to measure the growth fraction of cells. The significance of Ki-67 expression is still controversial, but decreased Ki-67 expression has been reported in a pulmonary fibrosis mouse model after antifibrotic treatment, which is consistent with our results. 48 Interestingly, the Ki-67 expression level in the ADSC group at Day 21 seemed to be slightly increased compared to that at Day 14, while the level in the ADSC-EISCH group stayed about the same. The optimal time of cell retention at the resection site to obtain the best therapeutic effect is still unclear. Given that many patients develop severe stenosis within 2-3 weeks after extensive esophageal ESD, cell retention during at least 2-3 weeks seems to be required. The tendencies of Ki-67 expression levels over time in the ADSC and ADSC-EISCH groups might suggest the required period of therapeutic stem cells and emphasize the importance of stable hydrogel scaffold to deliver the cells. Taken together, these results imply that ADSCs could be effective in reducing esophageal stricture, and this effect can be augmented using an ADSC-EISCH mixture.
Currently, corticosteroids are considered the standard treatment of esophageal stricture after ESD in various ways (either oral or injection, or both) despite potential adverse events such as delayed esophageal perforation, mediastinal abscess, and steroid-induced side effects.
Treatment of esophageal stricture using polyglycolic acid (PGA) sheets has been reported, [49][50][51] but the mechanism preventing the stricture is still elucidated yet. It has been speculated that the PGA sheet protects the wound from external stimuli, leading to less stricture. Compared with these conventional methods, stem cell transplantation using injectable hydrogel seems to be more effective for prevention of esophageal stricture since MSCs secrete a wide spectrum of paracrine factors for not only inhibiting inflammation and abnormal tissue growth but also enhancing tissue regeneration and angiogenesis. [52][53][54] Nonetheless, regenerative approaches using ADSC or ADSC-hydrogel mixture are still experimental and have not yet been approved for human use because long-term efficacy and safety need to be carefully investigated. Direct comparison study between current treatments such as corticosteroid or PGA sheets and our method is required to evaluate the treatment effect in the future. Further biological analysis like time-course spatial transcriptomics would be required to determine the molecular mechanism underlying the effect of stem cell-hydrogel hybrid system for prevention of esophageal strictures. Additionally, hydrogel-mediated delivery of engineered stem cells to overexpress angiogenic and/or anti-inflammatory factors would enable prolonged cell retention and further enhanced paracrine effects for more effectively preventing esophageal strictures.
This study had several limitations. First, the number of animals used in this study was small. However, a similar number of animals was used in most studies using porcine models. Considering the principles of the 3Rs (replacement, reduction, and refinement) in animal research, the sample size in this study could still be considered appropriate. Second, since this study used small young pigs (20-25 kg) that were still growing, it is unclear whether the results can be inferred for human adults. However, because the stomach and esophagus of pigs of this size and age are similar to those of humans, using pigs of similar sizes is preferred in most studies using pig models. Finally, unlike humans, it was difficult to strictly control the diet of pigs after ESD.
As water was allowed to be consumed freely, it cannot be excluded that a part of the injected ADSCs-hydrogel mixture was washed out, underestimating its therapeutic effect.

| CONCLUSION
In conclusion, our study demonstrated that the acsidian-inspired hyaluronic acid hydrogel system could be applied effectively to the endoscopic injection of therapeutic stem cells owing to its improved injectability attributed to the in situ self-crosslink-ability and superior biocompatibility. Thus, the mixtures containing the endoscopically injectable hydrogel and ADSCs significantly reduced post-ESD esophageal stricture by augmenting the paracrine effects of ADSCs.
Although further investigation and comparative studies are required to confirm the clinical application of this system, our injectable hydrogel-based therapeutic cell delivery system may suggest an innovative platform for cell therapies using familiar and established medical equipment in more clinically relevant situations.