BM-MSCs derived from Balb/c mouse and a complete medium set were purchased from Cyagen Biotechnologies (Suzhou, China). Cells were incubated in a complete medium containing 10% fetal bovine serum and 1% penicillin–streptomycin at 37°C in a 5% CO₂ humidified incubator, following the manufacturer's instructions. At 80–90% confluency, the adherent cells were passaged after digestion with 0.25% (wt/v) trypsin/0.02% (wt/v) EDTA for 2–3 min. The culture medium was replaced every 2–3 days. BM-MSCs at passage 8 were used for subsequent transfection and transplantation experiments.

BM-MSCs were transfected with control adenovirus (GeneChem Inc., Shanghai, China) using the enhanced infection solution (GeneChem) or complete medium as the infection system for 8 h with multiplicity of infection (MOI) of 0, 50, 100, 200, and 400 to detect the best MOI. The non-transfected cells were taken as the naïve control. After 48 h of in vitro adenovirus infection, transfected and non-transfected BM-MSCs were harvested. To detect the infection rate and whether adenovirus affects the phenotype of MSCs, the cell morphology and growth state were observed, and the expression of green fluorescence protein (GFP) in mouse BM-MSCs and the cell surface molecules (CD31, CD45, and Sca-1) were detected by flow cytometry. Then, BM-MSCs were infected with mouse Tnfaip6 RNA interference (RNAi) recombinant adenovirus (GeneChem) at the best MOI. The expression level of TSG-6 mRNA after transfection was determined by reverse transcription–quantitative polymerase chain reaction (RT-qPCR), and RNAi adenovirus with the highest interference efficiency at the mRNA level was screened out from three candidate targets. The target sequences of mouse Tnfaip6 RNAi recombinant adenoviruses were tgG​AAG​GTT​TAG​CCA​TCT​ATA, ccC​AAA​TGA​GTA​CGA​TGA​CAA, and ccC​GGC​TTC​AAT​ATT​TAT​ATT, labeled as virus A, B, and C, respectively. Finally, mouse BM-MSCs were infected with the selected RNAi adenovirus, and Tnfaip6 knockdown was confirmed by Western blotting.

To determine the phenotype after infection with control adenovirus, MSCs were suspended in Cell Staining Buffer (BioLegend, San Diego, California, USA) at a cell density of 10⁶ cells/100 μL/tube, stained with PE anti-Sca-1, FITC anti-CD31, and FITC anti-CD45 (BioLegend), at 4 °C for 15 min in the dark, and fixed with 500 µL of 4% paraformaldehyde. MSCs stained with isotype-matched control antibodies (PE anti-IgG2a and FITC anti-IgG2a; BioLegend) were taken as the control. Flow cytometry was conducted using a Navios system (Beckman Coulter, Miami, Florida, USA) and analyzed using Kaluza Analysis 2.0 and Navios software.

To detect TSG-6 mRNA in MSCs, total RNA was extracted from BM-MSCs using a PrimeScript™ RT Reagent Kit with gDNA Eraser (TaKaRa Bio Inc., Shiga, Japan), according to manufacturer's instructions. First-strand cDNA was synthesized using 0.5 μg of total RNA. The cDNA samples were then amplified in the AriaMx Real-Time PCR system (Agilent Technologies, Palo Alto, California, United States) for 40 cycles (95°C for 5 s, 60°C for 20 s, and 72°C for 15 s) with specific oligonucleotide primers (TaKaRa, Dalian, China). Each sample was analyzed in triplicate, with Tubulin used as the housekeeping gene. The relative quantification of the target gene was determined using the ΔΔCT method. The primers used in RT-qPCR are listed in Table 1.

In order to determine the level of Tnfaip6, BM-MSCs were homogenized in radioimmunoprecipitation assay buffer (Solarbio, Beijing, China) mixed with protease inhibitor (Solarbio). Total protein concentration was quantified using a bicinchoninic acid protein assay kit (Solarbio), and the denatured protein extract was stored at −80°C until use. Proteins were resolved by sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE) and transferred to polyvinylidene fluoride membrane (Millipore, Boston, Massachusetts, United States). The membrane was blocked with 8% skim milk in Tris-buffered saline containing 0.1% Tween-20 (TBST) at room temperature for 3 h and probed with rabbit anti-mouse Tnfaip6 antibody (1:1,000, Cell Signaling, Boston, Massachusetts, United States) or tubulin antibody (1:1,000, Shanggong Inc., Shanghai, China) in TBST at 4°C for 12 h. Successively, the membranes were incubated with horseradish peroxidase (HRP)–conjugated goat anti-rabbit secondary antibodies (1:3,000, Abcam, Cambridge, United Kingdom) at 37°C for 1 h. Then, the immunoreactive bands were detected by enhanced chemical luminescence (Biosharp, Hefei, China).

Balb/c mice (female, 8–9 weeks old, 19–21 g) were purchased from Charles River (Hangzhou, China), housed under standard conditions (controlled temperature 26–27°C, humidity 60–70%, and a 12-h light/dark cycle), and maintained for at least 1 week before the experiments began. The animal study was reviewed and approved by the Animal Care and Welfare Committee of Guangxi Medical University.

TNBS-induced colitis is characterized by T-cell–mediated immunity and widely used to explore immune regulation in IBD (Wirtz et al., 2017). This colitis model is more in line with the characteristics of transmural colitis of CD, which is characterized by a female preponderance (Goodman et al., 2020), so female mice were used in this study. Colitis was induced in mice by TNBS (Sigma-Aldrich, Saint Louis, Missouri, United States) as described previously (Wirtz et al., 2017). Briefly, the shaved skin was pre-sensitized with 150 μl of 1% (wt/v) TNBS–olive oil–acetone solution, and then 100 μl of 2.5% (wt/v) TNBS in 50% ethanol was injected into the colon of 24-h fasted mice 7 days after pre-sensitization. Mock-treated mice were administered olive oil–acetone solution and injected an equal volume of phosphate-buffered saline (PBS) through the anus as the normal control. Subsequently, the hair color, activity, weight change, stool consistency, and fecal occult blood test results were recorded daily. One day after enema (i.e., day 2), the mice were injected with a single intraperitoneal dose of BM-MSCs, BM-MSCs with or without Tnfaip6 knockdown infected by recombinant adenovirus (1×10⁶ cells in 100 μl PBS/mouse), or an equivalent volume of PBS as control, while each mouse in the Tnfaip6 group was injected intraperitoneally with 4 μg of recombinant mouse Tnfaip6 daily from days 2–5, as described in the literature (Sala et al., 2015). For this experiment, mice were randomly divided into six groups with 10 mice in each group: control, model (TNBS + PBS), MSCs (TNBS + BM-MSCs), Ad-MSCs (TNBS + BM-MSCs infected with control adenovirus), RNAi-MSCs (TNBS + MSCs infected with Tnfaip6 RNAi adenovirus), and Tnfaip6 (TNBS + Tnfaip6). On day 8, 150 μl of 4% (wt/v) chloral hydrate was injected intraperitoneally to anesthetize the mice. Next, blood samples were collected from the eye, and serum samples were obtained by centrifugation at 3,000 rpm at 4°C for 15 min. Then, the mice were sacrificed by cervical dislocation, and biological samples (colon tissue, mesenteric lymph nodes, and spleen) were obtained for further analysis. Serum samples were assessed by enzyme-linked immunosorbent assay (ELISA). Colon tissue was excised for histological assessment and RT-qPCR analysis. The MLNs and spleen were analyzed by flow cytometry to characterize Tfh and Tfr.

The fecal occult blood tests were performed on the mice from days 2–8 using commercial fecal occult blood test kits (dry chemical) (Ameritek, Jiaxing, China), according to the manufacturer's instructions.

Weight loss and stool consistency were recorded daily during the entire experiment. Colon sections were fixed in 10% formaldehyde for 24 h, embedded in paraffin, and cut into 3-μm sections that were stained with hematoxylin and eosin and assessed for colonic damage microscopically. Histological examinations were performed in a blinded manner. The disease activity index (DAI) and inflammation-related histologic score were evaluated for each mouse. The DAI score was calculated based on the scores obtained for weight loss, stool consistency, and degree of intestinal bleeding: score = 0, less than 1% weight loss, normal stool, negative hemocult; score = 1, 1–5% weight loss, soft but still formed stool; score = 2, 6–10% weight loss, soft stool, positive hemocult; score = 3, 11–18% weight loss, very soft or wet stool, blood traces in stool visible; and score = 4, weight loss more than 18%, watery diarrhea, gross rectal bleeding. The criteria for histological score were graded as follows: 0 point = no evidence of inflammation; 1 point = low level of inflammation, with scattered infiltrating mononuclear cells (1–2 foci); 2 points = moderate inflammation, with multiple foci; 3 points = high level of inflammation, with increased vascular density and marked wall thickening; and 4 points = maximal severity of inflammation, with transmural leukocyte infiltration and loss of goblet cells (Wirtz et al., 2017).

RT-qPCR was used to assess the relative mRNA expression of molecules in the colon, including the Tfh main transcription factor BCL-6, the Tfr main transcription factor Foxp3, and related cytokines including IL-10, IL-21, and TNF-α with 1 μg total RNA as described above, and glyceraldehyde 3-phosphate dehydrogenase (GAPDH) was used as the housekeeping gene. Primers are listed in Table 1.

The serum concentrations of Tnfaip6, IL-10, IL-21, and TNF-α of the mice were detected using commercial ELISA kits, according to the manufacturer's instructions. All ELISA kits were purchased from Shanghai MLbio Company in Shanghai, China, except the Tnfaip6 ELISA kit which was purchased from Signalway Antibody Company (College Park, Maryland, United States).

The spleen cells were collected in sterile ice-cold PBS and passed through a 200-µm cell strainer into a 15-ml tube by mechanical disruption using the plunger of a 5-ml syringe. The cells were pelleted by centrifugation at 1,500 rpm for 5 min and lysed with 4 ml red blood cell lysate buffer (Solarbio) for 5 min; the above filtration, collection, and centrifugation steps were repeated. The MLN cells were isolated by the same procedure except for the step of red blood cell lysis. To characterize Tfh and Tfr, the MLN and spleen cells (1×10⁶/tube) were stained with Pacific Blue™ anti-CD4, PE-Cy™7 anti-CXCR5, and APC anti-PD-1 (BD Biosciences, Franklin Lake, New Jersey, United States), at 4° for 30 min in the dark. Meanwhile, the cells were stained with matched isotype control antibodies (Pacific Blue™ anti-IgG2a, PE-Cy™7 anti-IgG2a, APC anti-IgG2, PE anti-IgG2a; BD Biosciences) and taken as the control. Then, the cells were stained with PE anti-Foxp3 at 4°C for 30 min after permeabilization and fixation using Foxp3/Transcription Factor Buffer Set (eBioscience, San Diego, California, United States). Finally, the cells were suspended in 500 µl of 4% paraformaldehyde, and the percentages of CD4⁺CXCR5⁺PD-1⁺ Tfh and CD4⁺CXCR5⁺Foxp3⁺ Tfr of the MLNs and spleen were detected by flow cytometry.

As shown in Figures 1B–E, 48 h later, BM-MSCs were found to be in good condition under the fluorescence microscope at MOI of 0, 50, 100, and 200. Slight green fluorescence was detected at the MOI of 50 and 100, while strong green fluorescence was observed at MOI of 200 and 400, and morphology retraction was detected in some of the cells at the MOI of 400 (Figure 1). These results indicated that the optimal MOI was 200.

The adenovirus infection rate was about 90% (Figure 2). No obvious changes were observed in cell morphology and growth status postinfection. MSCs hardly expressed CD45 and CD31, and their positive rates were <0.5% after infection with control adenovirus. On the other hand, MSCs highly expressed Sca-1 with a positive rate >99% (Figure 3). The expression of CD31, CD45, and Sca-1 met the MSCs standards on the cell surface.

To determine the knockdown efficiency of transfected BM-MSCs, RT-qPCR and Western blotting were performed. As shown in Figure 4A, RNAi adenovirus C was confirmed with the highest interference efficiency at the mRNA level in the three candidate target sequences. The levels of Tnfaip6 in the transfected BM-MSCs were significantly lower than those measured from the naïve BM-MSCs and BM-MSCs infected with control adenovirus (Figures 4B,C).

These findings suggested that Tnfaip6 RNAi recombinant adenovirus has been successfully transfected into BM-MSCs, leading to a significant knockdown of Tnfaip6 but did not affect the phenotype of MSCs.

On days 6 and 7, one mouse died in the model and Tnfaip6 knockdown MSCs groups, and no mice died in the other groups before blood collection. On the following days of TNBS enema, mice in the model group gradually developed arched backs, lazy movements, reduced food and water intake, and roughened hair and reduced gloss, and soft, loose, mucous, watery, or even bloody stools. Fecal occult blood tests were positive, weight was gradually decreased, and DAI was significantly increased (Figures 5A,B). In the later stage of modeling, due to the body's self-healing tendency, the feces became thin and soft with mucus attached to the surface, the positive rate of the occult blood tests decreased, the weight of some mice stabilized or rebounded slightly, and the DAI decreased gradually (Figures 5A–D). On day 8, mice were sacrificed, and the histology of the colon was assessed. Some colitis mice had an intestinal obstruction and showed adhesion of intestinal tissue to peripheral tissues at the obstructed site. The histological examination revealed infiltrating inflammatory cells, increased vascular density and marked wall thickening, loss of goblet cells, and penetrating ulcer in TNBS-induced colitis mouse colons (Figures 6A–C).

After intraperitoneal injection of BM-MSCs, Ad-MSCs, or Tnfaip6, the weight loss and DAI score were all lower than those in the model group from day 4 (p < 0.05), and no significant differences were detected in the body weight loss and DAI among mice injected with BM-MSCs, Ad-BM-MSCs, or Tnfaip6 (p > 0.05) (Figures 5A–D). After the administration of MSCs, Ad-MSCs, or Tnfaip6, the intestinal obstruction and adhesion of intestinal tissue were significantly improved, no transmural ulcers were observed, glands were more orderly than those in the model group, inflammatory infiltrations were significantly reduced, and inflammation-related histologic score was lower than that of the model group (p < 0.05) (Figures 6A–C). However, after injected with RNAi-MSCs, no statistical differences were detected in the weight loss, DAI, and the inflammation-related histologic score compared to that of the model group (p > 0.05) (Figures 5, 6). The above results indicated that BM-MSC transplantation ameliorated TNBS-induced colitis in mice. The injection of recombinant Tnfaip6 achieved similar effects as BM-MSC transplantation (Figures 5, 6). BM-MSCs lost the therapeutic effect of colitis after Tnfaip6 was knocked down in the IBD mice model (Figures 5, 6).

Next, we evaluated the effect of BM-MSCs on the modulation of inflammatory cytokines and molecules closely related to the differentiation and immune regulatory effects of Tfh and Tfr in TNBS-induced colitis, assessed the relative expression level of Tfh's main transcription factor BCL-6 and Tfr's main transcription factor Foxp3 mRNA by RT-qPCR in inflamed colons, and detected serum IL-21, TNF-α, IL-10, and Tnfaip6 concentrations. The relative expressions of BCL-6, IL-21, and TNF-α mRNA were higher in the inflamed colon of the model group compared to those of the control group (p < 0.05), accompanied by the same tends of serum IL-21, TNF-α, and Tnfaip6, while IL-10 and Foxp3 were slightly increased without statistical differences (p > 0.05) (Figures 7, 8). After intraperitoneal injection of BM-MSCs or Tnfaip6, the expression of BCL-6, TNF-α, and IL-21 were significantly decreased (p < 0.05) (Figures 7, 8), while expression of Foxp3, IL-10, and Tnfaip6 was significantly higher than those of the model group (p > 0.05) (Figures 7, 8). However, inhibition of Tnfaip6 significantly reduced the anti-inflammatory abilities of BM-MSCs to modulate BCL-6, Foxp3, TNF-α, IL-21, and IL-10 in the colon and/or serum (Figures 7, 8).

Finally, we detected the percentages of Tfh and Tfr in CD4⁺ cells in the MLNs and spleen by flow cytometry (Figures 9, 10). The results showed that percentages of CD4⁺CXCR5⁺PD-1⁺ Tfh and Tfh/Tfr ratios in the MLNs and spleen of colitis mice increased significantly compared to those of normal mice (p < 0.05), while no statistical difference was found in CD4⁺CXCR5⁺Foxp3⁺ Tfr (p > 0.05) (Figure 11). Compared to the model group, the percentages of Tfh and Tfh/Tfr ratios in the MLNs and spleen in BM-MSCs and Tnfaip6 groups were significantly lower in the MSCs and Tnfaip6 groups (p < 0.05), while the percentages of Tfr were significantly higher (p < 0.05) (Figure 11). However, the Tfh/Tfr polarization effect of BM-MSCs was abrogated when Tnfaip6 was knocked down (Figure 11).