Bone marrow-derived mesenchymal stem cells (BMSCs) repair acute necrotized pancreatitis by secreting microRNA-9 to target the NF-κB1/p50 gene in rats

Acute pancreatitis (AP) is a common acute abdominal disease, 10–20% of which can evolve into severe AP (SAP) causing significant morbidity and mortality. Bone marrow-derived mesenchymal stem cells (BMSCs) have the potential of repairing SAP, but the detailed mechanism remains unknown. We demonstrate here that microRNA-9 (miR-9) modified BMSCs (pri-miR-9-BMSCs) can significantly reduce the pancreatic edema, infiltration, hemorrhage, necrosis, the release of amylase and lipase. Meanwhile, decreased local/systemic inflammatory response (TNF-α↓, IL-1β↓, IL-6↓, HMGB1↓, MPO↓, CD68↓, IL-4↑, IL-10↑, and TGF-β↑) and enhanced regeneration of damaged pancreas (Reg4↑, PTF1↑, and PDX1↑) are also promoted. But these effects diminish or disappear after antagonizing miR-9 (TuD). Besides, we find that miR-9 is negatively correlated with AP and miR-9 agomir which can mimic the effects of pri-miR-9-BMSCs and protect injured pancreas. Furthermore, we investigate that BMSCs deliver miR-9 to the injured pancreas or peripheral blood mononuclear cell (PBMC), which can target the NF-κB1/p50 gene and inhibit the NF-κB signaling pathway (p-P65↓, NF-κB1/p50↓, IκBα↑, IκBβ↑). Taken together, these results show that miR-9 is a key paracrine factor of BMSCs attenuating SAP targeting the NF-κB1/p50 gene and suppressing the NF-κB signaling pathway.

of DAMPs has high potential for blocking the progression of AP. Nuclear factor κ light chain enhancer of activated B cells (NF-κB) is a key transcription factor regulating the productions of pro-inflammatory mediators to determine the inflammatory response and also possesses other non-inflammatory biological features, including proliferation, differentiation, apoptosis, invasion and survival [12][13][14][15] . The mammalian NF-κB family is composed of P50 (NF-κB1), P52 (NF-κB2), REL (cREL), REL-A (P65) and REL-B. NF-κB is a heterodimeric complex in which the p50-p65 heterodimer is the most common form. The NF-κB is restricted to the cytosol in the inactive state via combining with members of the inhibitor of NF-κB (IκB) family (IκBα, IκBβ, IκBε). In response to infection or injury, IκB kinase induces the phosphorylation of IκB leading to the degradation of IκB, which allows translocation of NF-κB to the nucleus. In the nucleus, NF-κB binds to a consensus sequence (5′-GGGACTTTCC-3′) to promote the transcription of pro-inflammatory cytokines. Studies have demonstrated that overactivation of NF-κB can aggravate AP 5,16 . Na-Taurocholate (NaT), a drug commonly used for the establishment of AP model, induces the translocation of NF-κB to the nucleus, which is responsible for the producing of DAMPs during NaT-induced AP 7,17,18 . Therefore, NF-κB has been considered as a key signaling molecule during AP. Mesenchymal stem cells (MSCs) have been widely studied for their potential applications in tissue engineering, autoimmune disease and gene delivery vehicle because of their properties of self-renewal, differentiation, immunosuppression, migration, paracrine and so on 19 . In addition, these cells are of low immunogenicity and can also be easily isolated from multiple tissues or organs. Up to date, MSCs have been reported as a cell-base therapeutic strategy for autoimmune, ischemic, and inflammatory diseases 20 . Recently, studies including ours have found that infused MSCs from bone-marrow or umbilical cord can attenuate SAP by inhibiting local and systematic inflammatory response, secreting cellular growth factors to promote angiogenesis and decreasing the apoptosis of PACs [21][22][23][24][25][26] . However, the mechanisms by which MSCs achieve these effects have not been elucidated clearly. In addition, despite the efforts 21,24,26 , necrotized pancreatic tissues cannot be repaired completely in the early stage of SAP. As a result, it is so urgent that an optimal cell-base therapeutic method should be proposed and the potential mechanism unveiled. MicroRNAs (miRNAs) are a class of endogenous non-coding RNA with a length of 18-23 nucleotides. miRNAs exhibit inhibitory effect on their target genes mostly by binding to the 3′ untranslated regions (3′UTRs) leading to translational repression or mRNAs degradation. miRNAs are found to be involved in the process of embryo development, stem cell fate, virus defense, hematopoiesis, organ formation, cell proliferation, inflammatory response and apoptosis, lipid metabolism and so on more [27][28][29][30] . Recently, microRNA-9 (miR-9) has been reported to have the potential of suppressing inflammatory response induced by lipopolysaccharide (LPS) through inhibiting the expression of NF-κB1/p50 gene in human polymorphonuclear neutrophils (PMN) and monocytes [31][32][33][34] . Moreover, NF-κB1/p50 has been identified as a target gene of miR-9 35,36 . Therefore, miR-9 may function as a factor of anti-inflammatory response by targeting the NF-κB1/p50 gene to reduce the heterodimeric complex of p50-p65 (NF-κB). In AP, the role of miR-9 has not been studied. Besides, whether miR-9 is involved in the process of transplanted MSCs repairing AP remains unknown. Recent evidences have shown that MSCs produce miRNAs to deliver to other cells by exosomes or microvesicle to influence their biological functions. Therefore, we propose that miR-9 may be a small RNA molecule involved in the occurrence and progression of AP, and infused MSCs deliver miR-9 to the pancreas so as to inhibit the inflammatory response and repair the necrotized pancreatic tissues. To test the above hypothesis, we conduct this study to investigate the relationship between miR-9 and SAP and to reveal the possible mechanism of MSCs promoting the repair and regeneration of necrotized pancreatic tissues.
BMSCs could deliver exogenous miR-9 to the damaged pancreas and PBMC. To investigate whether BMSCs can deliver miR-9 to the injured pancreas, we transfected synthesized Cy3-miR-9a-5p into GFP-BMSCs and observed their distributions in vivo by the fluorescence microscope (Fig. 1X). The result showed that Cy3-miR-9a-5p could be released by GFP-BMSCs to the liver, lung, spleen, and pancreas, of which the amount in liver and spleen was more than that in lung and pancreatic parenchyma (Fig. 1X). Interestingly, we found that a lot of Cy3-miR-9 accumulated in pancreatic lymph node (Supplementary information Figure 8/ Supplementary picture). Besides, in vitro, we investigated that Cy3-miR-9a-5p could be transferred by GFP-BMSCs into PBMC (Fig. 6L).
NF-κB1/p50 was validated as the target of miR-9a-5p. The transcript of NF-κB1/p50 gene and the sequence of miR-9 have eight bases pairing at both putative target sites (Fig. 6A). The expression of NF-κB1/p50 in PBMC could be markedly repressed by miR-9a-5p transient overexpression ( Fig. 6M,P,Q and K). To further confirm that miR-9a-5p can target the NF-κB1/p50 gene, we constructed the vectors of dual luciferase reporter, wild-type (wtUTR) or mutation (mutUTR) of NF-κB1 3′UTR harboring predicted binding sites of miR-9a-5p ( Fig. 6B-D). The results showed that the relative activity of firefly luciferase in HEK293T cells not only significantly declined after the transfection of miR-9a-5p mimics, but also presented a downtrend as the concentration of miR-9a-5p mimics increases (Fig. 6E). Besides, the activity of firefly luciferase inhibited by miR-9 could be rescued by anti-miR-9 (TuD) or the mutation of 3′UTR of NF-κB1/p50 gene (Fig. 6F,G). In addition, the expression of miR-9 in PBMC could be induced by LPS (Fig. 6M,N). Taken together, the above results indicated that NF-κB1/p50 was the target gene of miR-9a-5p.
(F) The expressions of miR-9 in damaged pancreas and serum were negatively correlated with pathological scores of AP by Pearson correlation analysis (p < 0.05). AP, acute pancreatitis, H&E, hematoxylin eosin, NC, normal control, NaT, sodium taurocholate, SD, standard deviation, miR-9, microRNA-9. (G) miR-9, produced by infused BMSCs, can target NF-κB1/p50 gene and suppress the activation of NF-κB signaling pathway in PBMC/Macrophage to reduce the release of the pro-inflammatory cytokines and prevent the occurrence of SIRS and MODS, which can promote the repair and regeneration of necrotized pancreatic tissues.

Discussion
The pathogenesis of SAP is very complicated, the common view on which is three theories: self-digestion of pancreatic enzyme, overactivation of white cells, and cascade amplification of inflammatory response.
To date, it is still impossible to prevent SAP or to capture the disease process completely. Besides, its therapeutic effect is still discouraging 41 .
MSCs is an important member of the family of Stem Cells owning multiple properties, such as self-renewal, multi-lineage differentiation, immunosuppressive, directional migration, and paracrine, etc 20 . Meanwhile, MSCs can be used as the vector of delivering the exogenous genes or biomaterials to the injured tissues or organs for disease therapy 42 . Previous studies including ours have found that MSCs has the potential of reducing SAP, but the detailed mechanism has not been clarified thoroughly. Jung et al. 21 found that MSCs could reduce SAP by suppression immune injury response. In our previous study 26 , we investigated that MSCs could attenuate SAP by secreting cellular growth factors to promote angiogenesis. Recent research reveals that miRNAs are involved in the pathologic process of SAP, in which miR-126a and miR-126b 43,44 have been taken as the diagnostic markers. In this study, we explored the possible mechanism of BMSCs in repairing SAP by over-expressing (pri-miR-9-BMSCs) or antagonizing (TuD-BMSCs) the expression of miR-9 and the results suggested that the damaged pancreatic tissues were repaired by pri-miR-9-BMSCs and miR-9 agomir. However, SAP could not be repaired by TuD-BMSCs or miR-9 agomir control. Meanwhile, it had been investigated that the expressions of miR-9 in pancreatic tissues could be up-regulated by pri-miR-9-BMSCs and miR-9 agomir. Besides, we found that BMSCs could deliver miR-9 to the injured pancreas or PBMC, which could repress the NF-κB signaling pathway. Consequently, it was suggested that miR-9 was a key regulatory factor of BMSCs in treating SAP.
What's more, we validated that NF-κB1/p50 gene was the target gene of miR-9 by performing the dual luciferase reporter assay. To further prove the above result, we conducted two other tests which showed that miR-9 reduced the activity of firefly luciferase with a dose-dependent effect by binding to 3′UTR of NF-κB1/p50 gene, which could be rescued by TuD or mutUTR plasmids and BMSCs could deliver exogenous miR-9 to PBMC, which could inhibit the expression of NF-κB1/p50 gene.
The NF-κB signaling pathway is playing a key role in AP 5-7, 16, 45, 46 , responsible for the producing of pro-inflammatory mediators, and therefore the inhibition of NF-κB signaling pathway can ameliorate AP 16,45 . Besides, decreasing the productions of pro-inflammatory mediators, such as TNF-α, IL-1β, IL-6 or HMBG1 can protect the damaged pancreas in experimental animal models of AP 10,11,47,48 , and the administration of exogenous anti-inflammatory cytokines (IL-4 and IL-10) can also attenuate AP 49,50 . Thus, it is believed that anti-inflammation method will be quite helpful and effective to the curing of SAP.
In our study, we investigated that miR-9 could be delivered by pri-miR-9-BMSCs and miR-9 agomir to the injured pancreas, which could inhibit the activation of NF-κB signaling pathway, decrease the levels of pro-inflammatory cytokines (TNF-α, IL-1β, IL-6, and HMBG1) and increase the levels of anti-inflammatory cytokines (IL-4, IL-10, and TGF-β). Besides, in vitro, we observed that BMSCs delivered exogenous miR-9 to PBMC, which repressed the activity of NF-κB signaling pathway. Thus, it is concluded that pri-miR-9-BMSCs and miR-9 agomir repair SAP resulting from the miR-9's inhibitory effect on the NF-κB signaling pathway.
Recently, researchers start to concentrate on pancreatic regeneration so as to alleviate the pains of patients and eradicate pancreatic diseases. Reports have found the capacity of MSCs to repair and regenerate the damaged pancreatic tissues is a promising cell-therapy strategy. However, most of the researches about pancreatic regeneration focus on the regeneration of insulin-producing β-cells 51 rather than the regenerative process of digestive enzyme-producing acinar cells and thereby little is known about the regeneration of acinar cells. Several reports showed the regeneration of acinar cells is probably relevant to the transcription factors (TFs) 51,52 . Pancreatic transcription factor 1 complex (PTF1) is a 48-kDa class B basic helix-loop-helix (bHLH) protein acting as the DNA binding subunit of the trimeric transcription factor resulting in the differentiation of pancreatic precursors into exocrine cells 51,53 . In this study, we observed that the expressions of PTF1 in necrotized pancreatic tissues were significantly up-regulated by pri-miR-9-BMSCs and miR-9 agomir, suggesting pancreatic regeneration. Meanwhile, Regenerating Islet-Derived Protein 4 (Reg4), able to repair and regenerate damaged pancreatic tissue 38 , was also significantly up-regulated by pri-miR-9-BMSCs and miR-9 agomir. Interestingly, pancreatic and duodenal homeobox 1(PDX1), a protein involved in the regeneration of islets and pancreatic development 51,54 , was also significantly increased by pri-miR-9-BMSCs and miR-9 agomir. These above results strongly suggested that miR-9 could promote the regeneration of damaged pancreatic tissues, which also explained why SAP was alleviated significantly by pri-miR-9-BMSCs and miR-9 agomir.
To further investigate whether BMSCs repair SAP by direct contact with damaged pancreatic cells, we assayed the migration of BMSCs in vivo by labeling BMSCs 25,26 . The results showed that the CM-Dil or SPION labeled BMSCs could migrate to the pancreas, but the number was less than those migrating to lung at day 3 after transplantation. Moreover, the amount of cells migrating to injured pancreas were similar among BMSCs, pri-miR-9-BMSCs, Empty virus BMSCs and TuD-BMSCs groups. Hence, we concluded that infused BMSCs repaired injured pancreatic tissues depending on paracrine rather than the direct interaction. Moreover, we observed that transplanted BMSCs could deliver miR-9 to the liver, spleen, lung and pancreas, suggesting that it is possible to repair SAP through the secretions of BMSCs. In this study, we found that a lot of Cy3-miR-9 was released from BMSCs and accumulated in pancreatic lymph node rather than pancreatic parenchyma. The new finding helps us understand how miR-9 could repair SAP, which may be related with modulation of local/systematic inflammatory/immune response. In the study of Jung et al. 21 , they also revealed that MSCs repaired SAP by suppressing immune response. Taken together, these results indicate that miR-9 released by BMSCs can mimic the roles of BMSCs to repair SAP, as a result of which it is not necessary for BMSCs migrating to the injured pancreas. Besides, we also validated the relationship between miR-9 and AP. The results showed that the mild systematic inflammatory response could induce the expression of miR-9 as observed in Sham group without showing any pancreatic injury. The up-regulation of miR-9 could inhibit the inflammatory response in PBMC and PMN as described by Bazzoni et al. 34 . Thus, miR-9 can be regarded as a anti-inflammatory molecular. Further, we found that the expression of miR-9 was markedly decreased in injured pancreas and serum showing a negative correlation with AP, which was consistent with the result of microRNA microarray (GSE61741). Besides, we investigated that the expression of miR-9 in 3% NaT-induced severe AP was significantly higher than that in Caerulein-induced mild AP. Consequently, it can be concluded that the up-regulation of miR-9 expression in SAP may be the compensatory mechanism which antagonizes the uncontrolled inflammatory response and prevents the deterioration of SAP. Though our study reveals that MSCs is effective in SAP treatment, we also acknowledge that there is still a long way to go for the final clinical application. Moreover, the risk of tumorigenicity of MSCs needs to be given cautious consideration and be avoided to the greatest extent in the future 55 . In our study, we just observed the therapeutic effect of MSCs at day 3 after cell transplantation so the side-effect of MSCs may not be observed fully. Next, we will focus on the systematic evaluation of MSCs treating SAP including the short-term and long-term complications.
To sum up, BMSCs ameliorate SAP and promote the regeneration of necrotized pancreatic tissue by releasing miR-9 to injured pancreas and inhibiting the NF-κB signaling pathway (Fig. 7G). Cell and cell culture. Bone marrow-derived mesenchymal stem cells (BMSCs) isolated from 3-4 weeks of Sprague-Dawley (SD) rats were cultured in DMEM-LG complete medium as previously described 25 . HEK-293T cells (human embryonic kidney-293 cells expressing the large T-antigen of simian virus 40) were purchased from the cell bank of Chinese Academy of Sciences and cultured in DMEM-HG medium supplemented with 10% FBS, 100 U/ml penicillin and 100 µg/ml streptomycin. Cells were digested and passaged by 1:5 when reaching 80% of confluence. The peripheral blood mononuclear cells (PBMC) were purified by density gradient centrifugation from healthy rats as previously described 56 . Briefly, 5 ml of peripheral blood was collected in the anticoagulant tube and diluted with 5 ml of phosphate buffer saline (PBS). Then 5 ml of Ficoll was added into the above mixture slowly, which was centrifuged with 2500 rpm for 20 min at room temperature. Finally, the white fog-like liquid of the middle layer (PBMC) was collected, washed by PBS for three times, resuspended in the RPMI-1640 medium supplemented with 10% FBS, 100 U/ml penicillin and 100 µg/ml streptomycin, and cultured in a humidified incubator with 5% CO 2 at 37 °C. miRNAs Targeting Prediction. The prediction of miRNAs targeting genes was performed by the algorithms of TargetScan 57 , PicTar 57 , microRNA org 58 and miRWalk Targets 59 . The results were intersected by MatchMiner 60 , suggesting that the NF-κB1/p50 gene was a potential target gene of miR-9.

General PCR (gPCR) and Quantitative Real-time PCR (qRT-PCR). Total RNA was extracted by
TRIzol or TRIzol LS Reagent from the cells, frozen pancreatic specimens or serum. First-strand cDNA was synthesized by PrimeScript TM Reverse Transcriptase Reagent Kit. The interested genes were amplified by General PCR (gPCR) using Taq enzyme according to the manufacturer's instructions: 98 °C 10 s, 53 °C 30 s, 72 °C 30 s, 30 cycles and the agarose gel was scanned by Gel Doc ™ XR+ Imager (Bio-Rad, CA, USA). The levels of mRNAs were detected by qRT-PCR using the KAPA Kit (Kapa Biosystems, Boston, USA) and the Applied Biosystems 7500 Real-Time PCR system (Thermo Fisher Scientific, CA, USA) as previously described 26 . The primers were synthesized by Beijing Genomics Institute (Beijing, China). GAPDH and U6 were taken as the endogenous control. The sequences of primers are listed in Table 1. Finally, Quadruplicate cycle threshold (CT) values were analyzed by the SDS software (Applied Biosystems, CA, USA) and the levels of mRNAs targeted genes were quantified by the comparative CT method. The procedure was replicated more than three times. Each measurement was set three repeats.
Immunoblotting and immunohistochemistry. The procedure of immunoblotting was depicted in our previous study 26 . In brief, the total proteins were extracted by RIPA lysis buffer supplemented with PMSF(1:100) and protease inhibitor Cocktail Tablets (Roche Applied Science, Shanghai, China) and quantified by the BCA method. Then, the proteins were transferred to nitrocellulose membrane and incubated with primary and secondary antibodies. Finally, the nitrocellulose membrane was detected by the Odyssey 3.0 analysis software (LI-COR Biotechnology, Nebraska, USA). The experiment was repeated more than three times. Besides, the immunohistochemistry was also introduced in this study for measuring the expressive levels of inflammatory signaling proteins, of which the procedure was described in our previous study 26 .
The construction of miR-9 and anti-miR-9 vectors. Rat genomic DNA was exacted by DNA purification kit following the manufacturer's instructions. A 368 bp of DNA fragment containing the miR-9-1 sequence (NC_005101.4) was amplified from genomic DNA by Primer STAR Max DNA Polymerase (98 °C 10 s, 53 °C 5 s, 72 °C 5 s, 30 cycles) using the following primers: sense, 5′-GACAGCTAGCTCTCGTCGTGCTAGTGCGTG-3′ and antisense, 5′-GTCAGGATCCTGGCTGAGCTGAGCAACCCT-3′. Then, the amplified fragment and PCDH-CMV-MSCs-EF1-GFP-T2A-Puro vector (PCDH) (System Biosciences, CA, USA) were digested by Nhe I and BamH I enzymes to produce the sticky ends respectively and connected by T4 DNA ligase at 16 °C overnight to generate the recombinant plasmid (pri-miR-9-PCDH). Finally, the recombinant plasmid was transformed into DH5α to replicate for 16 hours (hr), extracted by TIANprep Mini Plasmid Kit and identified by using gPCR, double enzyme digestion and sequencing analysis (Beijing Genomics Institute, Beijing, China). Besides, we constructed the plasmid of anti-miR-9 by adopting RNA tough decoy (TuD) technique as previously described 61 . In brief, the decoy sequence of anti-miR-9 was designed as follows: 5′-TCATACAGCTAGATCTATAACCAAAGA-3′ and 5′-TCTTTGGTTATAGATCTAGCTGTATGA-3′ and synthesized by Beijing Genomics Institute. Then, the above oligonucleotide pairs and PLKO.1 vector (System Biosciences, CA, USA) were digested by Age I and EcoR I enzymes and connected overnight by using T4 DNA ligase to generate PLKO.1-TuD recombinant plasmid. Finally, the recombinant plasmid was transferred into DH5α to replicate, extracted by TIANprep Mini Plasmid Kit and verified by sequence analysis.
BMSCs were infected by lentivirus. Recombinant lentivirus encoding miR-9 or TuD was produced by lentivirus packaging system (System Biosciences, CA, USA) following the manufacturer's instructions. First, the vectors of pri-miR-9-PCDH, PLKO.1-TuD or PCDH (8 µg/plate), pCMVΔR8.74 expressing HIV gag/pol, Rev and tat (5.3 µg/plate), and pMD2.G expressing VSV-G (2.65 µg/plate) were co-transfected into HEK293T cells by Lipo2000 as previously described 62 . Second, the supernatants were collected at 24 and 48 hr respectively and concentrated by PEG-it Virus Precipitation Solution (System Biosciences, CA, USA) or supercentrifugation (75,000 × g, 2 hr). Third, the viral tires were measured as previously described 63 and BMSCs were infected by pri-miR-9-, empty-, and TuD-lentivirus at a Multiplicity of Infection (MOI) of 50 under assistance of polybrene (8 µg/ml) to construct the cell lines of pri-miR-9-BMSCs, Empty virus-BMSCs, and TuD-BMSCs. Finally, the mRNAs were exacted and the expressions of miR-9 were detected by gPCR and qRT-PCR. All the viral experiments were performed in a biological safety cabinet. Transfection with Cy3-miR-9a-5p mimics and detection of NF-κB Activity. Empty virus-BMSCs stably expressing GFP were transfected by Lipo2000 with Cy3-miR-9a-5p mimics (50 nM) or miR-9a-5p control (50 nM) as previously described 64 . 24 hr later, they were co-cultured with PBMC stimulated by LPS (5 μg/ ml) for 24 hr. Finally, the mRNAs and proteins were extracted by TRIzol reagent and RIPA lysis buffer at 48 and 72 hr respectively. The expressions of miR-9 were measured by qRT-PCR and gPCR. Meanwhile, the expression of NF-κB1/p50 was detected by western-blotting and qRT-PCR and the NF-κB activity was assayed by Dual Luciferase Reporter Assay System (Promega). The pGL6-TA-luc plasmid was selected as a template for constructing the reporter vector of NF-κB-luc containing the response element of NF-κB: 5′-GGG AATTTCCGGGAATTTCCGGGAATTTCCGGGAATTTCC-3′. PBMC was firstly co-cultured with BMSCs of miR-9a-5p transfection for 24 hr and then co-transfected by Lipo2000 with NF-κB-Luc reporter vector (0.1 µg) and Renilla luciferase (pRL-TK, 0.1 µg). 6 hr later, PBMC was simulated by LPS for 48 hr at a concentration of 5 μg/ ml and harvested by passive lysis buffer (Promega, Beijing, China). The luciferase activity of NF-κB was measured and assessed by Dual Luciferase Reporter Assay System. The experiments were repeated for more than three times.
In addition, to reveal the relationship between miR-9 and AP, we established several AP models as follows: NC (n = 3), Sham (n = 3), Caerulein (n = 3), 3% NaT (n = 3). Moreover, to demonstrate that miR-9 could reduce SAP, we administrated miR-9a-5p agomir (1 µM) and miR-9a-5p control (1 µM) (Biotend Company, Shanghai, China) to SAP rats through the tail vein following the manufacturer's instructions (http://www.biotend.com/miRNA). Then, these rats were humanly killed at day 3 after the treatment or at postoperative day 4. Finally, the serum was collected by the centrifugation of 8000 × g at 4 °C for 20 min and stored at −80 °C. The tissues were obtained by surgical vehicles and stored in liquid nitrogen or −80 °C or fixed in 4% paraformaldehyde.
Hematoxylin-eosin (H&E) staining. The H&E staining of paraffin-embedded pancreatic tissues was performed for investigating the severity of AP as previously described 25 .
ELISAs and amylase, lipase and MPO activities assays. The levels of serum IL-1β, IL-4, IL-6, IL-10, TNF-α, TGF-β and HBMG1 were detected by ELISAs kit as previously described 26 . The activity of serum amylase and lipase was measured by the amylase and lipase assay kit as previously described 25 . The activity of MPO in pancreatic tissues was determined by MPO Detection Kit (Jiancheng Bioengineering, Nanjing, Jiangsu Province, China) as previously described 66 .
TUNEL. The terminal deoxynucleotidyl transferase dUTP nick-end labeling staining (TUNEL) was used for detecting cell apoptosis of damaged pancreatic tissues by using the One Step TUNEL Apoptosis Assay Kit (Beyotime Biotechnology, Nantong, Jiangsu Province, China) following the manufacturer's manual as previously described 67 . Apoptotic cells were observed as green fluorescence particles and counted in randomly selected five fields at ×200.

CM-Dil-/SPION-labeled BMSCs and in vivo their distributions.
CM-Dil, a kind of red fluorescent dyes, was selected for labeling BMSCs to track their migrations in vivo as previously described 25,26 . In brief, pancreas and lung were collected, fixed in 4% paraformaldehyde for 24 hr and dehydrated by 30% sucrose solution for more than 2 hr. Then, these tissues were embedded in Tissue-Tek O.C.T. Compound (SAKURA, USA) and solidified into tissue blocks at −80 °C for 10 min. Finally, these tissue blocks were cut into frozen sections with the thickness of 5 μm and observed under the fluorescence microscope. The red particles were counted in randomly selected five fields at ×200. Meanwhile, superparamagnetic nanoparticle (SPION) was also used for labeling BMSCs to trace their distributions in vivo as previously described 26 .
In Situ Hybridization. To analyze the expressions of miR-9 in paraffin-embedded pancreatic tissues, we designed a probe of 5′-digoxigenin-labeled oligonucleotide (5′-ATACAGCTAGATAACCAAAGA-3′) for hybridizing with miR-9 in situ by using Enhanced Sensitive ISH Detection Kit (Boster biology company, Wuhan, Hubei Province, China) following the manufacturer's instructions as previously described 68 .
The distribution of Cy3-miR-9a-5p transfected Empty virus-BMSCs in vivo. The liver, heart, spleen, lung, pancreas, kidney, duodenum were collected at day 3 after the transplantation of Cy3-miR-9a-5p transfected Empty virus-BMSCs, and fixed in 4% paraformaldehyde for 24 hr. Then, these organs were dehydrated by 30% sucrose solution and embedded by Tissue-Tek O.C.T. Compound. Finally, frozen sections were observed and photographed by fluorescence microscope.