Three-dimensional culture method enhances the therapeutic efficacies of tonsil-derived mesenchymal stem cells in murine chronic colitis model

Tonsil-derived mesenchymal stem cells (TMSCs) showed therapeutic effects on acute and chronic murine colitis models, owing to their immunomodulatory properties; therefore, we evaluated enhanced therapeutic effects of TMSCs on a murine colitis model using three-dimensional (3D) culture method. The expression of angiogenic factors, VEGF, and anti-inflammatory cytokines, IL-10, TSG-6, TGF-β, and IDO-1, was significantly higher in the 3D-TMSC-treated group than in the 2D-TMSC-treated group (P < 0.05). At days 18 and 30 after inducing chronic colitis, disease activity index scores were estimated to be significantly lower in the 3D-TMSC-treated group than in the colitis control (P < 0.001 and P < 0.001, respectively) and 2D-TMSC-treated groups (P = 0.022 and P = 0.004, respectively). Body weight loss was significantly lower in the 3D-TMSC-treated group than in the colitis control (P < 0.001) and 2D-TMSC-treated groups (P = 0.005). Colon length shortening was significantly recovered in the 3D-TMSC-treated group compared to that in the 2D-TMSC-treated group (P = 0.001). Histological scoring index was significantly lower in the 3D-TMSC-treated group than in the 2D-TMSC-treated group (P = 0.002). These results indicate that 3D-cultured TMSCs showed considerably higher therapeutic effects in a chronic murine colitis model than those of 2D-cultured TMSCs via increased anti-inflammatory cytokine expression.

www.nature.com/scientificreports/ sulfate sodium (DSS)-induced acute and chronic murine colitis models demonstrated therapeutic efficacy in terms of lowering disease activity index (DAI) scores, colon length recovery, and by decreasing the expression of pro-inflammatory cytokines 14,15 . Additionally, TMSC-conditioned medium showed similar efficacy as TMSCs in acute and chronic murine colitis models, thereby suggesting that TMSCs exhibit therapeutic effects through the anti-inflammatory mechanism.
To improve the therapeutic effect of MSCs, several strategies, including pretreatment with pro-inflammatory cytokines or hypoxic condition, have been evaluated 16 . The effect of these pretreatments is thought to stimulate MSCs to maintain anti-inflammatory properties in inflamed tissues under in vivo stress conditions. Recently, a three-dimensional (3D) culture method was reported to improve survival and enhance the therapeutic properties of MSCs 17 . In an acute kidney injury model, 3D-cultured human adipose tissue-derived MSCs (ADSCs) showed increased secretion of anti-apoptotic and anti-oxidative cytokines when compared with 2D-cultured MSCs, thereby demonstrating enhanced survival and therapeutic effects. However, the therapeutic effects of 3D-cultured TMSCs in a colitis model remain unclear.
In this study, we aimed to evaluate whether 3D-culture method could improve the therapeutic efficacy of TMSCs in a chronic murine colitis model.
Histopathological improvement after 3D-cultured TMSCs treatment. As shown in Fig. 8, the colonic structure in the DSS-induced chronic colitis model was disrupted and characterized by the loss of crypts, diffuse mucosal and submucosal edema, infiltration of mononuclear cells, and ulcerations. The histological scoring index (HSI) values, which indicate the severity and extent of inflammation and crypt damage, in the 3D-TMSC-treated group (7.5 ± 2.8) were significantly lower than those in the colitis control group (5.7 ± 2.8; P = 0.01; Fig. 8). Histological analysis revealed that treatment with 3D-cultured TMSCs mitigated the DSS-induced severe inflammatory cell infiltration, loss of crypts, and ulceration (Fig. 8). The HSI values in the 3D-TMSC-treated group (5.7 ± 2.8) were also significantly lower than those in the 2D-TMSC-treated group (7.6 ± 2.7; P = 0.002).
TMSC localization in the peritoneum of mice using immunofluorescence. White spherical aggregates were observed in the euthanized mice. We assumed that the injected TMSCs formed an aggregate in the perito-  . Weight recovery after treatment with 3D-cultured TMSCs. Body weight loss in the 3D-TMSC-treated group was significantly mitigated when compared with that in the DSS + PBS, the DSS + HEK and 2D-TMSCtreated groups. *On day 18, the body weight loss in the 3D-TMSC-treated group was lower than that in the DSS + PBS (P < 0.001), the DSS + HEK (P = 0.188), and 2D-TMSC-treated groups (P = 0.005). **On day 30, the body weight loss in the 3D-TMSC-treated group was significantly lower than that in the DSS + PBS (P = 0.001), the DSS + HEK (P = 0.027), and 2D-TMSC-treated groups (P = 0.005). 3D, three-dimensional; 2D, twodimensional; TMSCs, tonsil-derived mesenchymal stem cells; DSS, dextran sulfate sodium; PBS, phosphatebuffered saline; HEK, human embryonic kidney. www.nature.com/scientificreports/ neal space of mice. Some aggregates were adjacent to the mouse intestine, whereas other aggregates were not attached to any organ and were observed in the omentum and mesentery (Fig. 9a). Spheroids were observed in the 3D-TMSC-treated group but not in the 2D-TMSC-treated group. To confirm the location of TMSCs in vivo, the cells were tracked with anti-human nuclear antigen antibodies using immunofluorescence. Fluorescence microscopy analysis revealed that the previously observed aggregates were stained with green-colored anti-human nuclear antigen antibodies, which indicated that the transplanted 3D-cultured TMSCs formed a cluster in the peritoneum irrespective of colon inflammation (Fig. 9b). In addition, the expression levels of human DNA in the peritoneal lavage fluid of each groups and aggregates were measured. The expression levels of human DNA were the highest in the aggregates, followed by the peritoneal lavage fluid of 3D-TMSC-treated and 2D-TMSC-treated groups (Fig. 9c). This indicates that these whitish aggregates in the mouse peritoneal cavity comprise TMSCs.

Discussion
In the present study, we evaluated the therapeutic effects of 3D-cultured TMSCs and compared them with those of 2D-cultured TMSCs in a chronic murine colitis model. 3D-cultured TMSC treatment ameliorated the clinical symptoms, including fecal blood, body weight loss, and shortening of colon length in a chronic murine colitis model. Additionally, histological improvement was achieved with 3D-cultured TMSCs, which was not observed through 2D-cultured TMSC treatment. Moreover, we showed that the 3D-culture method could enhance the expression of anti-inflammatory cytokines and growth factors in TMSCs and the survival of TMSCs in vivo after transplantation. Initially, MSCs were thought to infiltrate into inflamed tissues and engage in tissue regeneration by differentiating into mature intestinal epithelial cells 18 . However, recently published studies showed that the paracrine effect is a more important mechanism of MSCs irrespective of their location in vivo 19 . Although MSCs are considered as a promising new therapeutic modality in various diseases, including IBD, clinical trials showed controversial outcomes 20 . Reduced cell viability of MSCs after in vivo transplantation has limited its usability. Moreover, to maintain effective paracrine activity of MSCs, the transplanted cells should endure harsh microenvironments (e.g., oxidative, inflammatory, and hypoxic conditions), which reduce the rate of cell engraftment and survival 21 .
Recently, 3D-culture method has been evaluated as a modality to improve the efficacy of MSC therapy. More recently, organoid culture methods, which are physiologically more relevant, have been developed to overcome the shortcomings of MSC therapy 22 . www.nature.com/scientificreports/ In our previous studies, we evaluated the therapeutic effect of TMSCs in acute and chronic colitis models 14,15,23 . IP injection of TMSCs significantly ameliorated DSS-induced colitis; however, no histologically significant effects were observed. Additionally, in acute and chronic colitis models, TMSC-conditioned medium showed similar therapeutic efficacy as TMSCs. However, histological improvements were not achieved through treatment with TMSC-conditioned medium. Moreover, we tried to improve the therapeutic efficacy of TMSCs pretreated with ascorbic acid and metformin, however, we could not achieve the desired results (data not shown). In the current study, the 3D-culture method showed significant enhancement in the therapeutic efficacy of TMSCs in addition to histological improvement.
The efficacy of 3D-culture method for MSC therapy has been evaluated for several disease models, including those of inflammation, ischemic injury, and cancer 21,[24][25][26] . In a previous study, 3D-cultured ADSCs showed enhanced inhibitory effects on liver cancer cells when compared with 2D-cultured or sphere-cultured ADSCs 26 27 . However, this study only evaluated the therapeutic effect of 3D-cultured BMSCs, administered through limited injection methods, in an acute colitis model, and comparison with 2D-cultured MSCs was not appropriately performed 27 . In the current study, we evaluated enhanced therapeutic effects in a chronic colitis model, which are clinically similar to chronic human IBD.
Enhanced cell survival of 3D-cultured TMSCs was demonstrated in our study, which corresponds with the results obtained in the previous studies. Upon tracking the location of transplanted 2D-and 3D-cultured TMSCs on day 31, 3D-cultured TMSCs, which were injected at least 14 days before euthanizing the mice, were observed to form aggregates in the peritoneum independently of the inflamed colon, however, this effect was not observed upon 2D-cultured TMSC administration. Interestingly, the transplanted 3D-cultured TMSCs were detected in the peritoneum, but not incorporated in the colon tissues, implying that the therapeutic effects of 3D-cultured TMSCs were not caused by gut-homing. Our results are in accordance with a previous study by Sala et al., who reported that intraperitoneally injected BMSCs do not localize in the intestine, instead they form aggregates in the peritoneum and produce cytokines 19 . The mechanisms underlying the survival of transplanted human 3D-TMSCs in immunocompetent mice have not been elucidated. However, long-term survival of human  29 . Although spontaneous cell death occurs after the transplantation of 3D-TMSCs, this process may be delayed by the immunomodulatory function of 3D-TMSCs, which is associated with the secretion of immune-modulatory cytokines, including IDO-1, TSG-6, and transforming growth factor β (TGF-β). In addition, the expression level of CXCR4 in the 3D-cultured TMSCs was higher than that in the 2D-cultured TMSCs. CXCR4 is reported to increase the viability and migration of MSCs 30,31 . Therefore, CXCR4 may play a critical role in the enhanced survival of 3D-TMSCs. Paracrine secretion of therapeutic cytokines, not tissue regeneration by gut-homing, may play an important role in the treatment mechanism of TMSCs. Although the exact mechanism of 3D-cultured TMSC-mediated attenuation of chronic murine colitis is unclear, enhanced paracrine secretion of these beneficial cytokines may be involved in imparting this effect. In our study, the expression of anti-inflammatory cytokines, including IL-10 and TSG-6, was significantly increased via 3D-culture method (P < 0.05). In a previous study, TSG-6 was reported to be a key cytokine in MSC therapy and promoted the expansion of regulatory macrophages that expressed IL-10 and inducible nitric oxide synthase, and reduced serum levels of interferon-6, IL-6, and TNF-α 19 . Moreover, in experimental myocardial infarction model, treatment with MSCs reduced tissue damage by producing TSG-6 32 . Importantly, TMSCs highly expressing TSG-6 showed therapeutic effect in an acute graft-versus-host disease (GVHD) mouse model 33 . In our study, TSG-6 production was markedly increased by 3D-culture method, thereby, enhancing the therapeutic effect of TMSCs. Moreover, the expression levels of cytokines associated with immunomodulatory effects, including TGF-β and IDO-1, were significantly upregulated in the 3D-cultured TMSCs. Additionally, the expression levels of IL-1β, IL-6, and IL-17 were downregulated in the colon tissues of the 3D-TMSC-treated group, which may be due to the production of cytokines associated with immune-regulatory functions, such as TSG-6, TGF-β, and IDO-1. The upregulated expression of VEGF in the 3D-cultured TMSCs may also promote the repair of damaged intestinal mucosa by enhancing TMSC migration and proliferation and accelerating the growth of vessels 34 .
In summary, the present study demonstrated that 3D-cultured TMSCs significantly ameliorated chronic colitis by reducing clinical symptoms, recovery of colon shortening, and histological improvements. We further demonstrated enhanced survival of 3D-cultured TMSCs in the peritoneum. Additionally, we demonstrated that the increased paracrine effect of anti-inflammatory cytokines, including IL-10, TSG-6, TGF-β, and IDO-1, may play a critical role in mediating the therapeutic effect of 3D-cultured TMSCs. Thus, the 3D-culture method provided a novel approach to enhance TMSC function and, thereby, could have therapeutic applications in IBD. www.nature.com/scientificreports/

Methods
Isolation and expansion of TMSCs. Subjects aged less than 18 years were recruited in this study. Written informed consent was obtained from a parent and/or patients' legal guardians for studies on tonsil tissues. This study was approved by the Ewha Womans University Medical Center institutional review board (ECT 11-53-02). All experiments were performed in accordance with the institutional ethical guidelines and the Declaration of Helsinki. All tonsil tissues used in this study were obtained from a single donor. TMSCs were isolated and cultured as described in our previous studies 15,35 . Briefly, tonsil tissues were obtained during tonsillectomy in patients younger than 10 years of age. The tonsil tissues were minced and digested in RPMI 1640 medium (Invitrogen, Carlsbad, CA, USA) supplemented with 210 U/mL collagenase type I (Invitrogen) and 10 g/mL DNase (Sigma-Aldrich, St. Louis, MO, USA) at 37 °C for 30 min. Digested tissues were washed using Dulbecco's modified Eagle's medium-high glucose (DMEM-HG; Welgene, Daegu, South Korea) supplemented with 20% fetal bovine serum (FBS; Invitrogen), and washed again using DMEM-HG supplemented with 10% FBS. From the prepared tonsil tissues, we isolated mononuclear cells using Ficoll-Paque (GE Healthcare, Little Chalfont, UK) density gradient centrifugation. The isolated mononuclear cells were then cultured in cell culture plates, and non-adherent cells were removed after 8 h of seeding. The remaining adherent cells were further cultured for 2 weeks and passaged. The passaged cells (hereafter referred to as TMSCs) were stored in liquid nitrogen for future experiments. The MSC characteristics of TMSCs according to the minimal criteria for defining multipotent mesenchymal stromal cells 36 were confirmed in our previous study 15 and other studies 37 . In brief, the immunophenotypic surface marker assay results revealed that TMSCs were negative for hematopoietic cell markers, such as CD14, CD34, and CD45 and positive for common MSC markers, such as CD73, CD90, and CD105. In addition, cell-specific staining assays were performed to examine the differentiation of TMSCs into adipocytes, chondrocytes, and osteoblasts. TMSCs could differentiate into these three cell types upon induction with commercially available differentiation media 15 .  www.nature.com/scientificreports/ protocol approved by the Ethics Committee for Animal Research of Ewha Womans University (EUM19-0464, ESM18-0415). This study was performed according to the standards articulated in the ARRIVE guidelines 38 . Chronic colitis in mice was induced by oral administration of 1.5% DSS (MP biochemical, Irvine, CA, USA) for 5 days, followed by an additional 5 days of tap water feeding; overall 3 such cycles (total 30 days) were performed. We used PBS or human embryonic kidney 293 (HEK293) cells as a sham control for colitis.

Experimental design.
In the chronic colitis model, mice were randomly assigned to five groups: (1)  Assessment of therapeutic effect of TMSCs. DAI scoring. For each group, the body weight and DAI scores, including weight change, stool consistency, and occult or fecal blood, were determined as previously reported (Supplementary Table 2) 39 .
Measuring colon length and histopathological scoring. On day 31 of chronic colitis induction, mice were euthanized with CO 2 gas inhalation, and colon specimens were acquired from the proximal and distal parts of the dissected colon and fixed with 10% formalin, followed by paraffin sectioning and hematoxylin-eosin (H&E) staining. Two specimens from each proximal and distal colon were evaluated using a previously reported histological colitis scoring system (Supplementary Table 3) 40 . The severity and extent of inflammation, the level of crypt damage, and the damaged portion (%) of the whole colon were evaluated. Finally, the averages of the HSI values were compared between the groups.
Quantitative reverse transcriptase polymerase chain reaction (qRT-PCR) for cytokine expression. qRT-PCR for determining the expression of pro-inflammatory cytokines, namely, IL-1β, IL-6, TNF-α, IL-17, and the antiinflammatory cytokine IL-10 in colonic tissues was performed using the QuantStudio 3 real-time PCR system (Applied Biosystems, Waltham, MA, USA). Detailed procedures are described in the Supplementary methods.
Cytokine array analysis. The levels of cytokines in the colonic tissue were analyzed using the Proteome Profiler Mouse Cytokine Array Panel A kit (R&D Systems, Minneapolis, MN, USA), following the manufacturer's instructions. In brief, the tissues were lysed using radioimmunoprecipitation assay buffer (containing protease inhibitor cocktail). The protein concentration in the lysate was measured using a BCA Protein assay kit (Thermo Scientific, Waltham, MA, USA). Tissue protein diluted in array buffer was incubated with the ready-to-use precoated array membranes overnight at 4 °C on a rocking platform shaker. The membrane was washed and incubated with streptavidin-horseradish peroxidase (HRP) buffer for 30 min. Next, the membrane was washed and incubated with the Chemi Reagent mixture at 23-27 °C for 1 min. The membrane was analyzed using the LAS-300 system (Fujifilm, Tokyo, Japan). Dot density was analyzed using Multi Gauge 3.0.
Immunofluorescence analysis for visualizing TMSC localization and determination of human DNA expression levels in the mouse peritoneal cavity cells. Colon specimens were fixed as paraffin blocks and 4 μm wide sections were resected. After deparaffinization and hydration, the specimens were blocked with mouse IgG reagent using a mouse on mouse (M. O. M) kit (Vector Laboratories, CA, USA). TMSCs were then stained with human nuclear antigen monoclonal antibody (MyBioSource, CA, USA). After fluorescence staining with Avidin DCS (Vector Laboratories, CA, USA), the specimens were mounted with Vectashield phalloidin (Vector Laboratories, CA, USA), and observed using fluorescence microscopy. The expression levels of human DNA in the peritoneal lavage fluid of each treated groups and aggregates were measured. The mice were euthanized and sprayed with 70% ethanol. The outer skin of the peritoneum was dissected using scissors and forceps and gently pulled back to expose the inner skin surrounding the abdominal cavity. The peritoneal lavage fluid was collected by injecting 5 mL of ice-cold saline into the peritoneal cavity