HO-1/Bmmscs Perfusion Using a Normothermic Machine Perfusion System Reduces the Acute Rejection of DCD Liver Transplantation by Regulating NKT Cell Co-Inhibitory Receptors

Background: Liver transplantation (LT) represents the most effective treatment for many end-stage liver diseases. While donation after cardiac death (DCD) donor livers are used due to organ shortage, acute rejection (ACR) remains an important risk factor affecting the survival of recipients following transplantation. Although immunosuppressive agents can be used, they are associated with complications. Bone marrow mesenchymal stem cells (BMMSCs) are used in the treatment of organ transplantation; however, there is limited colonization in the target organs and a short survival time following BMMSCs application. Thus, an optimized BMMSCs application method is required to suppress immune rejection and promote the long-term survival of allogeneic liver transplant recipients. Methods: BMMSCs were isolated and modied with heme oxygenase 1 (HO-1) gene. HO-1/BMMSCs were perfused into the donor liver in vitro using a normothermic machine perfusion (NMP) system, followed by LT. The severity of ACR was evaluated based on the liver histopathology. Gene chip technology was used to detect differential gene expression, and the ow cytometry was used to analyze changes in natural killer (NK) T cells. Results: NMP can induce BMMSCs to colonize the donor liver during in vitro preservation, and the survival of HO-1/BMMSCs in the liver grafts was signicantly longer than that of BMMSCs. When the donor liver contained HO-1/BMMSCs, the ACR is obviously controlled, and the survival time was signicantly prolonged. The application of HO-1/BMMSCs reduces the number of NKT cells in the liver grafts, increases the expression of the NKT cell co-inhibitory receptors, and reduces the level of NKT cell expression of IFN-γ. Thus, NK cell and CD8 + T cell activation was inhibited, which reduced acute of rejection of the transplanted liver. Conclusions: The NMP system preserves the DCD donor liver in vitro, and also allows large quantities of BMMSCs to colonize the liver. HO-1-modied BMMSCs are able to improve and prolong the local immunosuppressive level our NKT

hours, which regulate the function of many immune cells [23]. For example, interleukin (IL)-2, interferon (IFN)-γ, and tumor necrosis factor (TNF)-α induce pro-in ammatory Th (T helper, Th)1 cell responses [24], whereas IL-4, IL-5, IL-6, IL-10 and IL-13 IL-6, IL-10, and IL-13 promote pro-in ammatory Th2 cell responses [25]. There is increasing evidence suggesting that NKT cells are essential in the regulation of autoimmune responses and IFN-γ production activates NK cell and CD8 + T cell immune responses, which are closely associated with the development of ACR of LT [26]; however, the effects of NKT cells on ACR of LT remain to be elucidated.
In this study, we used NMP to infuse HO-1/BMMSCs into the donor liver through the portal vein during its in vitro preservation under the premise of protecting the DCD donor liver in vitro. HO-1/BMMSCs were colonized in the donor liver and the effect of HO-1/BMMSCs on NKT cells and their inhibitory receptors was observed in the DCD liver grafts in rejection and the associated mechanism was explored. The ndings of this study will provide a novel method by which optimized BMMSCs regulate the immunosuppressive state of transplantation and prolong the survival time of the recipient.

Animals
The experimental animals were provided by Beijing Vital River Laboratory Animal Technology Co. Ltd. (Beijing, China). Male Lewis rats (7 − 8 weeks old, 200 g − 220 g) were used as donors, and male Brown Norway rats (8 − 9 weeks old, 220 g − 240 g) were used as the recipients. The rats were divided into the following six groups according to the different liver treatments: sham operated (Sham) group; SCS group; NMP group; NMP + BMMSCs (BMP) group; NMP + HO-1/BMMSCs (HMP) group; and NMP + FK506 (FK506) group. A total of 36 (6 rats/group) recipients were used for the survival analysis, and the other 72 recipients were used for postoperative (7 d, n = 6; 14 d, n = 6) specimen collection. The animals were housed in a standard laboratory animal room at a constant temperature (22°C ± 1°C), 60% relative humidity, 12 h/12 h light-dark cycle, and free access to water and food. All animal protocols were based on the National Institutes of Health "Guide for the Care and Use of Laboratory Animals" (National Institutes of Health publication 85 − 23, Bethesda, MD). Efforts were made to minimize the number of animals used and any discomfort encountered, and all procedures were approved by the Ethics Committee of Tianjin First Central Hospital (license number: 2016-03-A1).

Preparation and characterization of HO-1/BMMSCs
BMMSCs were prepared in accordance with our previous methods [16,17]. Brie y, the femurs and tibias were removed under aseptic conditions after the rats had been executed and sterilized, and BMMSCs were extracted using a whole bone marrow apposition screening method, and perfused or transfected with HO-1 after culturing to the third generation. HO-1/BMMSCs were prepared by transfecting HO-1 using an adenovirus (Genechem, Shanghai, China) when the cells were in a stable state. The differentiation ability of the HO-1/BMMSCs was determined via in vitro osteogenic and lipogenic differentiation with antibodies against CD29, CD34, CD45, CD90, RT1A, and RT1B (Biolegend, CA, US) staining was used to identify the molecular phenotype using ow cytometry. qRT-PCR and Western blot were used to determine whether HO-1 expression was elevated.

DCD model
The rats were anesthetized and placed on a warming pad, the abdomen was opened along the median abdomen, the liver, inferior vena cava and portal vein were freed, and heparin (Solarbio, Beijing, China) (1 U/g body weight) was injected from the dorsal penile vein. After 10 min, the diaphragm was cut and the heart was compressed to promote cardiac arrest, simulating the process of circulatory death in vivo, and the abdominal temperature was maintained with warm saline at 37°C (range: 35°C − 37°C) for 30 min.
HO-1/BMMSCs combined with NMP for in vitro preservation of donor livers The NMP system primarily consists of an organ compartment, peristaltic pump, lter, membrane oxygenator and oxygen supply system, as well as a temperature and pressure sensor ( Figure S1A and B). The donor liver was perfused with portal vein, and the perfusion temperature was maintained at 36°C − 38°C. The main components of the perfusion uid consist of 20% fetal bovine serum (FBS, Biowest, Nuaillé, France) containing, 60 mL Dulbecco's modi ed Eagle's medium (DMEM)/F12 (Gibco, Thermo Scienti c, Waltham, MA, USA), 20 mL rat blood, 100 U/mL penicillin (Gibco), 100 µg/mL streptomycin (Gibco), and 5 U/mL heparin (Gibco) [17]. All of the donor livers were ushed with 10 mL University of Wisconsin (UW) solution through the portal vein before performing machine perfusion, and the SCS group was subjected to static cold storage with UW solution. NMP livers were simply mechanically perfused, the BMP group was perfused with 1× 10 7 BMMSCs through the portal vein, the HMP group was perfused with 1× 10 7 HO-1/BMMSCs through the portal vein (all BMMSCs were perfused into the liver through a 100-µm pore size lter). The livers in the FK506 group were mechanically perfused and 0.1 mg/kg FK506 was administered intragastrically daily following surgery [27]. The donor livers were stored in vitro for 4 h before transplantation.

Orthotopic LT
The rats were subjected to the orthotopic LT technique based on Kamada's "two-cuff method" [17,28]. The operator was a surgical professional and underwent substantial training in the early stages. The anhepatic phase is controlled at 19 ± 1 min. After transplantation, the rats received 2 mL Ringer's lactate and were rewarmed in a postoperative incubator for 30 min. On days 7 and 10 after LT, the recipient was euthanized by an intraperitoneal injection with an overdose of pentobarbital (150 mg/kg), and the spleen, blood, and liver grafts were obtained.

BMMSCs tracing
To demonstrate that BMMSCs can colonize in the liver grafts, an adenovirus expressing green uorescent protein (GFP) (Genechem) was used to transfect the BMMSCs using NMP perfusion into the liver. The level of uorescence expression in the liver was observed using an in vivo imaging system (PerkinElmer, CA, USA) after LT. After the recipient was sacri ced, the liver grafts were obtained, and frozen sections were prepared to observe BMMSCs colonization in the liver tissue under a uorescence microscope.

Histopathology
The livers were xed in a 10% neutral formalin solution and sections of each of the samples were stained with hematoxylin-eosin (HE) to visualize transplanted liver histopathology or for terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL) in accordance with the methods provided by the manufacturer of a commercial kit. The rejection activity index (RAI) was used to evaluate the degree of ACR.

Enzyme-linked immunosorbent assay (ELISA)
The serum and cell culture supernatant were analyzed for the concentration of cytokines, and the serum content of IL-2, IFN-γ, and TNF-α were detected in the serum in accordance with the manufacturer's instructions of the commercial ELISA kit (MultiSciences Biotech Co., Hangzhou, China).

Western blot
The manufacturer's instructions (Solarbio, Beijing, China) of a commercial kit were used to extract the total cell protein using cell lysates, after which a Western blot was performed. The detailed experimental methods were performed as described in our previously published literature [16]. Using β-actin as a control, the membranes were scanned with an imaging system (Bio-Rad, Hercules, CA, USA).

Gene chip
After extracting the total RNA from the liver samples of each group, it was quanti ed using a NanoDrop ND-2000 (Thermo Scienti c) and the RNA integrity was detected with a Agilent Bioanalyzer 2100 (Agilent Technologies, CA, US). After passing the RNA quality inspection, the total RNA was reverse transcribed into double-stranded cDNA and then further synthesized with Cyanine-3-CTP (Cy3)-labeled cRNA. The labeled cRNA was hybridized with the chip, and the original image was obtained by scanning with an Agilent Scanner G2505C (Agilent Technologies) after elution. Gene chip detection was performed by OE Biotechnology Co., Ltd., (Shanghai, China). Feature Extraction software (version 10.7.1.1, Agilent Technologies) was used to process the original images to extract the raw data, and GeneSpring GX software (version 14.9, Agilent Technologies) was used to quantile the raw data. The standardized data was ltered, each group of samples was used for comparison, and at least 75% of the samples marked as detected probes were saved for subsequent analysis. The P-value and fold-change value of the t-test were used to screen for differential genes. The screening criteria consisted of up-regulation or downregulation in the fold-change value ≥ 2.0 and P-value ≤ 0.05. A gene ontology (GO) analysis and Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis were applied to determine the role of these differentially expressed mRNA. The mRNA microarray data were deposited in the NCBI's Gene Expression Omnibus database (GSE167980).

MNC and HO-1/BMMSCs co-culture
The HO-1/BMMSCs were seeded into the upper chamber and the liver MNCs were placed into the lower chamber, and stimulate with α-Galactosylceramide (α-GalCer) (100 ng/mL) [29]. MNCs and HO-1/BMMSCs were co-cultured at a 1:1 ratio for 48 h, Brefeldin A was used to block cytokine release for the nal 6 h, and the expression of IFN-γ by MNCs was detected by ow cytometry.
Statistical analysis SPSS 13.0 (SPSS GmbH, Munich, Germany), GraphPad 8.0 (GraphPad Software Inc., San Diego, CA, US) were used for the statistical analysis. The data were expressed as the mean ± SD, and multi-group analyses were performed by a one-way analysis of variance (ANOVA) for data assessment. A two-tailed Student's t-test or ANOVA were used for comparisons between groups. For the survival analysis of the rats, Kaplan-Meier survival curves were plotted by counting the survival time and a Log-rank (Mantel-Cox) test. P < 0.05 was considered to indicate statistical signi cance.

Extraction and identi cation of HO-1/BMMSCs
BMMSCs were isolated and cultured to the third generation as previously described [16]. BMMSCs transfected with an HO-1 adenovirus remained unchanged morphologically ( Figure S2A), and had the ability to induce osteogenic ( Figure S2B) and lipogenic ( Figure S2C) differentiation in vitro. Flow cytometric identi cation of surface biomarkers revealed a positive rate of over 99% for CD29, CD90, and RT1A, and a negative rate of over 99% for CD34, CD45, and RT1B ( Figure S2D − F). This demonstrated that the molecular biology of the HO-1 adenovirus-transfected BMMSCs was not affected. The immuno uorescence exhibited signi cantly higher red uorescence intensity in the HO-1/BMMSCs compared to that of the BMMSCs ( Figure S2G and H). qRT-PCR ( Figure S2I) and Western blot ( Figure  S2J) results were also con rmed to exhibit signi cantly higher HO-1 expression in the HO-1/BMMSCs.

The NMP system preserves the donor livers in vitro and allows the colonization of HO-1/BMMSCs
The normal liver is bright red (Fig. 1A), while the DCD liver is a purplish red color with rounded edges and slight edema (Fig. 1B). After the liver had been preserved using the SCS method, the liver was congested with edema, enlarged in size, and rounded at the edges (Fig. 1C). A stable NMP system was established ( Figure S1A and B) and the liver preserved by the NMP method was uniform in color, without edema and congestion (Fig. 1D). After the LT was completed using the "double cuff method", the liver was evenly congested, indicating that the model was successfully established ( Fig. 1E and F). Respectively, 1 × 10 7 GFP/BMMSCs or HO-1/BMMSCs were perfused, and LT was completed. Our previous studies demonstrated that approximately 50% of BMMSCs could be colonized in the donor liver during the perfusion process [17]. The vivo imaging system was used to detect the liver grafts the uorescence intensity of the livers perfused with HO-1/BMMSCs was higher than that of BMMSCs ( Fig. 1G and H). The liver grafts were obtained, frozen sections were prepared, and the number of GFP + cells in the frozen sections of the HO-1/BMMSCs-perfused liver grafts were signi cantly higher than that of the BMMSCs (Fig. 1I − K). This nding con rms that in the in ammatory environment after LT, HO-1/BMMSCs survived in signi cantly higher numbers in the transplanted livers compared to BMMSCs.

HO-1/BMMSCs signi cantly reduced ACR and improved the recipient prognosis
The morphological changes of the liver grafts were observed by HE staining ( Fig. 2A). At 7 d post-LT, the SCS group liver grafts were observed to be severely damaged, a large number of hepatocytes disappeared around the vein, a large number of macrophages and lymphocytes had in ltrated, and focal necrosis of the hepatocytes with eosinophilic homogeneity and hemorrhage were locally observed. The liver grafts in the HMP and FK506 groups were signi cantly improved compared with the SCS, NMP, and BMP groups, there was no obvious necrosis in the liver, and the liver lobular structure was intact, with a moderate amount of lymphocytic in ltration surrounding the bile ducts and portal vein. No rats survived in the SCS group at 14 d post-LT, whereas the liver lobules were more intact in the HBP group, exhibiting neatly arranged hepatocyte strips and low levels of lymphocyte in ltration. The TUNEL results of the liver tissues showed that the apoptotic cells were signi cantly increased in the SCS and NMP groups compared with other groups, whereas there were no signi cant differences between the HMP and FK506 groups and both were reduced compared with the BMP group ( Fig. 2B and C). The acute cellular rejection scores were not signi cantly different between the NMP and SCS groups, but signi cantly increased compared with all other transplantation groups. Moreover, the rejection scores in the HMP and FK506 groups were signi cantly lower than those in the BMP group (P < 0.05); however, there was no signi cant difference between the HMP and FK506 groups (Fig. 2D). These results indicate that HO-1/BMMSCs combined with NMP preserved the DCD livers in vitro, and HO-1/BMMSCs retained in the liver grafts after transplantation improved the pathology and inhibited ACR, similar to the level of FK506 application.
The median survival time after transplantation in the Sham, HMP, and FK506 groups was > 60 d. All animals survived long-term and no statistical differences were observed regarding a comparison in the survival rate between the groups (Fig. 2E). In the survival analysis conducted using a , and SCS group. The survival time of the rats in the SCS group was signi cantly lower than that of the other groups (P < 0.05), and the survival time of rats in the BMP group was longer than that of the NMP group but shorter than that of the HBP and FK506 groups (P < 0.05). This nding indicates that HO-1/BMMSCs perfused via portal vein could signi cantly prolong the survival time of the recipient rats after LT, and the application of HO-1/BMMSCs was more effective than BMMSCs.
ALT, AST, ALP, and GGT re ected the level of liver cell injury, and the levels of ALT, AST, and ALP in the HBP group were signi cantly lower than those in the SCS, NMP, and BMP groups at 7 d post LT (P < 0.05).
The difference with the FK506 group was not statistically signi cant (P > 0.05). At 14 d post-LT, no rats in the SCS group survived, and the levels of ALT, AST, and ALP in the HBP group were signi cantly lower than those in the SCS, NMP, and BMP groups (P < 0.05). The levels of ALB responding to hepatic synthetic function were only observed to be signi cantly lower in the SCS group compared to that of the Sham and HBP groups at 7 d post-LT (P < 0.05). TBil can respond to the excretory function of the liver and to hepatocyte damage. The level of TBIL in the HBP group was higher than that in the Sham group at 7 d post-LT but signi cantly lower than that of all the other groups (P < 0.05). The level of TBIL in the SCS group was signi cantly higher than that in all of the other groups. The level of TBIL in the BMP group was lower than that in the NMP group, but higher than that in the HBP group. These ndings indicated that HO-1/BMMSCs signi cantly improved the condition of the recipients' liver function (Fig. 2F).
Gene chip analysis of the differential gene expression in the liver grafts of each group We detected the level of mRNA expression in the liver grafts of each group, and analyzed the differential gene expression using gene chip technology, focusing on the NMP, BMP, and HMP groups (Table S1 − 3). The combination of the GO analysis and KEGG analysis revealed that the interaction between cytokines and cytokine receptors, as well as the level of gene expression of Th1-and Th2-related cytokines were signi cantly different (Fig. 3A − C and S3A − D). We compared the gene expression of cytokines related to ACR (Fig. 3D) and found that the gene expression of Th1 cytokines (e.g., IFN-γ, TNFα, and IL-2) related to ACR was signi cantly lower in the HMP and BMP groups compared with that of the NMP group. The level of gene expression in the above cytokines was lower in the HMP group compared to that of the BMP group. In combination with the analysis of the interaction of cytokine genes associated with ACR performed in a String database (Fig. 3D), we focused on the role of IFN-γ gene expression in HO-1/BMMSCs in attenuating ACR. The effect of IFN-γ on NKT cells, NK-and Th1-and Th2-related cytokines was assessed.

HO-1/BMMSCs perfused by NMP reduced the proportion of NKT cells and inhibited IFN-γ expression following LT
Under the guidance of the gene chip analysis results, we next sought to determine the cellular source of IFN-γ. IFN-γ is primarily secreted by NKT and NK cells. Simultaneously, a large number of NKT cells in the liver of rats and NKT cells could secrete IFN-γ in large quantities within a few hours of stimulation. In light of the fact that IFN-γ can signi cantly promote ACR, we used ow cytometry to detect the proportion of NKT cells and IFN-γ expression in the different recipient tissues. The results showed that following LT, the proportion of NKT cells in the spleen of the NMP group was increased compared with that of the Sham group (P < 0.05); however, no signi cant difference was observed in the other groups (P > 0.05). While the proportion of NKT cells in the blood of the NMP group was signi cantly higher than that of the other groups (P < 0.05), no signi cant difference was observed between the BMP and the HMP groups (P < 0.05). The proportion of NKT cells in the liver of the NMP group was signi cantly higher than that of the other groups and the HMP group was signi cantly lower than that of the BMP group (P < 0.05) (Fig. 4A).
Intracellular staining was used to detect the changes in IFN-γ levels in the NKT cells (Fig. 4B). The level of IFN-γ in the NKT cells of the NMP group was signi cantly higher than that of the BMP and HMP groups, and the HMP group was signi cantly lower than that of the BMP group (P < 0.05). These results demonstrated that the DCD donor liver was preserved by NMP combined with HO-1/BMMSCs. The proportion of NKT cells and level of IFN-γ in the recipients was signi cantly reduced following LT. HO-1/BMMSCs in the donor livers exhibit a stronger regulatory effect on the NKT cells than BMMSCs.

HO-1/BMMSCs perfusion by NMP can reduce the proportion of NK and CD8 + T cells in the liver grafts, inhibit their function, and reduce the level of in ammatory factors in the recipients
Flow cytometry analysis of the proportion of NK cells in the liver grafts revealed that the proportion of the HMP and BMP groups was signi cantly lower than that of the NMP group, and the HMP group was signi cantly lower than that of the BMP group (P < 0.05) (Fig. 5A). Intracellular IFN-γ staining revealed that the level of IFN-γ expression in the NK cells of the liver grafts of the HMP group and BMP group was signi cantly lower than that of the NMP group. The HMP group was signi cantly lower than that of the BMP group (P < 0.05) (Fig. 5B). The HO-1/BMMSCs in the DCD donor liver could reduce the proportion of NK cells in the liver grafts after LT, and the inhibitory effect of HO-1/BMMSCs on NK cells was greater than that of the BMMSCs.
Since IFN-γ can activate CD8 + T cells, we detected the proportion of T cells in the liver by ow cytometry.
The proportion of CD8 + T cells in the liver grafts of the HMP and BMP groups was signi cantly lower than that of the NMP group, and the proportion of CD8 + T cells in the HMP group was signi cantly lower than that in the BMP group (P < 0.05) (Fig. 5C). The level of granzyme B (GZMB) expression re ected the intensity of the toxic effect of CD8 + T cells, and the level of GZMB expression in CD8 + T cells was signi cantly lower in the liver grafts of the HMP group and BMP group compared with the NMP group, and it was signi cantly lower in the HMP group compared with that of the BMP group (P < 0.05) (Fig. 5D).
An ELISA was used to detect the ACR-related cytokines, IFN-γ, TNF-α, and IL-2, and the results showed that the cytokines in the HMP and BMP groups were signi cantly lower than those of the NMP group, and the HMP group was signi cantly lower than that in the BMP group (P < 0.05) (Fig. 5E). These results suggest that HO-1/BMMSCs can reduce the level of in ammatory factors related to ACR after LT and the effect of HO-1/BMMSCs is greater than that of BMMSCs.

Excessive activation of NKT cells aggravate ACR
To clarify the role of NKT in the ACR of LT, the rats in the NMP and HMP group the relative speci c stimulator of NKT cells were administered α-GalCer (50 µg/kg) by intraperitoneal injection. Our ndings showed severe ACR in the transplanted livers of the NMP group, and the application of α-GalCer had no signi cant effect on the NMP group ( Fig. 6A and B). The histopathology of the liver in the HMP group demonstrated that ACR was signi cantly reduced (Fig. 6A), the ACR score was reduced (Fig. 6B), and the degree of liver apoptosis was signi cantly improved (Fig. 6C). However, when α-GalCer was used to activate NKT cells, the level of acute liver rejection was more severe than that in the HMP group. The liver function (Fig. 6D) analysis showed that the level of liver enzymes and TBil were signi cantly increased under the administration of α-GalCer.

Further detection of NKT cells in the liver grafts revealed that the use of α-GalCer could affect the inhibitory effect of HO-1/BMMSCs on NKT cells and signi cantly increase the level of IFN-γ expression in
the NKT cells (Fig. 7A). In the event that α-GalCer induces NKT cell activation, we tested NK cells and CD8 + T cells closely related to ACR. We found that the level of IFN-γ in the NK cells and level of GZMB in CD8 + T cells were signi cantly increased ( Fig. 7B and C). In addition, compared to the HMP group, α-GalCer caused a signi cant increase in ACR-related cytokines (e.g., IFN-γ, TNFα, and IL-2) (Fig. 7D). Our results indicate that the inhibitory effect of HO-1/BMMSCs on NKT regulates ACR. This also suggests that α-GalCer has no effect on ACR after LT; however, in the case of HO-1/BMMSCs-induced transplantation tolerance, NKT cell activation can cause aggravation of ACR.
Co-culture experiments observed the effect of HO-1/BMMSCs on IFN-γ expression in NKT cells We co-cultured HO-1/BMMSCs and liver MNCs in vitro to verify whether it could regulate the function of NKT cells activated by α-GalCer. Our results showed that the level of IFN-γ expression in NKT cells was signi cantly increased following stimulation with α-GalCer. When co-cultured with BMMSCs or HO-1/BMMSCs, the level of IFN-γ expression was signi cantly reduced in NKT cells, and the HO-1/BMMSCs inhibited IFN-γ expression in NKT cells more signi cantly than in BMMSCs (Fig. 8A). We detected ACRrelated cytokines (e.g., IFN-γ, TNF-α, and IL-2) in the culture supernatant, and found that the co-culture of HO-1/BMMSCs with NKT cells could signi cantly reduce the levels of these cytokines (Fig. 8B). In vitro studies have shown that HO-1/BMMSCs can inhibit IFN-γ expression in the NKT cells and reduce the level of cytokines related to ACR.

HO-1/BMMSCs in the liver grafts affects the co-inhibitory receptor expression of NKT cells
Our previous study con rmed the signi cant inhibitory effect of HO-1/BMMSCs on the level of IFN-γ expression in NKT cells. We further investigated the mechanism of action of HO-1/BMMSCs on NKT cells, the activation of which is triggered by linking semi-invariant TCRs, as well as a series of costimulatory and co-inhibitory receptors. We detected the co-receptor expression on NKT cells in the liver grafts of each group ( Fig. 9A and B), and found that NKT cells constitutively expressed a variety of co-stimulatory receptors; however, few differences were observed between the groups. The examination of the coinhibitory receptors revealed that the expression of NKT cell surface expression of the co-inhibitory receptors, BTLA and CD160, differed signi cantly in each group. In the HMP and BMP groups, BTLA and CD160 expression (proportion of positive cells and MFI) in the NKT cells were signi cantly higher than that of the NMP group ( Fig. 9C and D) and the level of expression in the HMP group was higher than that of BMP group. This nding con rms that the HO-1/BMMSCs in the liver grafts increasing the expression of co-inhibitory receptors BTLA and CD160 on NKT cells, through which inhibitory signals are transmitted to NKT cells, reducing the level of IFN-γ secretion from NKT cells and reducing ACR after LT. Moreover, the HO-1/BMMSCs that colonized the transplanted livers had a greater regulatory effect on the NKT cells than the BMMSCs.

The regulation of NKT cell function by HO-1/BMMSCs is achieved through BTLA and CD160
To further con rm the role of BTLA and CD160 in the regulation of NKT cell function by HO-1/BMMSCs, we extracted MNCs from the normal rat livers using α-GalCer to activate the NKT cells. A blocking antibody was used to block BTLA or/and CD160 in NKT cells, and co-cultured with HO-1/BMMSCs. Our research showed that when BTLA or CD160 was blocked, the level of IFN-γ expression in the NKT cells increased to varying degrees compared with that of the control group (Fig. 10A). The levels of IFN-γ, TNF-α, and IL-2 in the culture supernatant were also signi cantly increased compared to that of the control group (Fig. 10B). The level of IFN-γ expression in NKT cells was further increased when BTLA and CD160 were both blocked (Fig. 10A), the concentration of IFN-γ, TNF-α, and IL-2 were also further increased (Fig. 10B). These results suggest that HO-1/BMMSCs affects the surface expression of BTLA and CD160 on NKT cells, transmitting inhibitory signals to NKT cells and regulating the levels of IFN-γ, inhibiting the ACR of LT. Thus, BTLA and CD160 may be co-inhibitory receptors with non-overlapping functions.

Discussion
ACR following LT remains the primary complication experienced by patients, which directly affects the long-term prognosis. The long-term immunosuppressant regimen is associated with multiple complications, and some patients continue to suffer from ACR under treatment with immunosuppressive agents [9]. In this context, it is necessary to seek new immunomodulatory methods. The interaction between BMMSCs and immune cells can establish a stable and balanced microenvironment through the regulation of innate or acquired immune cells [30]. The intercellular interaction and paracrine response between BMMSCs and immune cells provide a theoretical basis for the treatment of immune-related diseases [31][32][33]. Numerous studies have shown that BMMSCs play an active role in attenuating the ACR of organ transplantation [34,35]. Currently, studies describing the interaction of BMMSCs with NKT cells in LT remain unclear. BMMSCs are less effective following systemic application and their shorter survival time in target organs continues to challenge the progress of cell therapy.
In this study, NMP was used to preserve the DCD donor liver in vitro while perfusing HO-1/BMMSCs into the portal vein of the donor liver, after which LT was performed. The follow-up results of the BMMSCs remaining in the liver grafts following LT showed that HO-1/BMMSCs can be colonized in large numbers in the liver grafts, and the survival time is longer than that of BMMSCs. Our method solves the problem associated with BMMSCs colonization and the short survival time in target organs, and also avoids complications (e.g., pulmonary embolism and thrombosis), which may be caused by an intravenous application of BMMSCs. Moreover, liver grafts carrying HO-1/BMMSCs exhibited signi cantly lower ACR after LT, as demonstrated by liver function and histology.
To explore the impact of HO-1/BMMSCs on ACR, we used gene chip technology for analysis and found that the interaction between cytokines and cytokine receptors, as well as Th1 and Th2 cell-related factor gene expression were signi cantly different. We performed an association analysis of these genes and con rmed that the level of IFN-γ expression played a role in the reduction of ACR by HO-1/BMMSCs. In the process of regulating autoimmune responses, NKT cells are essential for microbial defense and initiating an adaptive immune response. There are a large number of NKT cells in the liver sinusoids, and these numbers increase following transplantation [34,35]. NKT cell activation releases a large amount of IFN-γ, which helps to activate NK cells, CD8 + T cells, and antigen-presenting cells [36,37]. In addition, IFNγ has been shown to play an important role in allograft rejection, indicating that NKT cells may play a critical role in the ACR of LT.
Our research shows that HO-1/BMMSCs carried in the liver grafts reduces the number of NKT cells after LT and inhibits IFN-γ expression, thereby inhibiting NK and CD8 + T cells activation, and signi cantly reducing ACR. It is important to note that NKT cells are required for transplantation immune tolerance in some organs or tissues. One heart transplantation study showed that immunosuppressive regimens were effective in the presence of NKT cells and ineffective after the clearance of NKT cells; however, α-GalCer had little effect on rejection in the absence of immunosuppressive regimens [38]. Despite these ndings, there have been no reports on the effect of α-GalCer on ACR in LT. Since activated NKT cells release both pro-in ammatory and anti-in ammatory cytokines, they have many different functions in the immune response, and thus their role in the ACR of LT remains unclear.
In our study, the application of α-GalCer did not signi cantly alleviate ACR in the rats of the NMP group; however, HO-1/BMMSCs signi cantly alleviated the ACR of LT. The application of α-GalCer could lead to excessive activation of NKT cells, increased IFN-γ expression, and aggravated ACR (Fig. 6A). This indicates that the excessive activation of NKT cells promotes the occurrence of ACR of LT, whereas the use of HO-1/BMMSCs inhibits the NKT cell activation and decreases IFN-γ expression to reduce the ACR of LT. Thus, the application of α-GalCer has no bene cial for reducing ACR, and may even be harmful.
Activation of NKT cells is triggered by linking semi-invariant TCRs or a series of co-stimulatory and coinhibitory receptors [39]; however, the research on the expression of NKT cell co-receptors in LT is poorly understood. We detected the surface co-receptors of NKT cells in the livers of different treatment groups following LT and found that the levels of CD160 and BTLA expression in the HMP group were signi cantly increased (Fig. 8). CD160 is considered to be a marker of T cell depletion [40], while BTLA is structurally expressed by NKT cells as an inhibitory receptor that transmits signals from herpes virus entry mediator [41,42]. CD160 and BTLA share a common ligand that delivers inhibitory signals to NKT cells [43]. Our study revealed that HO-1/BMMSCs reduce IFN-γ expression in NKT cells by increasing their surface expression of co-inhibitory receptors, through which inhibitory signals are delivered to NKT cells, revealing a novel target for the regulation of NKT cell function.
Our research shows that NMP is a reliable method for preserving DCD donor livers and provides a new means of applying BMMSCs. Moreover, HO-1 gene transfection enhances the survival of BMMSCs in liver grafts in a complex in ammatory environment following LT. In this respect, liver grafts carrying HO-1/BMMSCs signi cantly attenuated the occurrence of ACR after LT and improved the recipient prognosis.
We conducted a preliminary investigation of the role of NKT cells in ACR of LT and con rmed that the regulation of NKT cell surface co-inhibitory receptor expression by HO-1/BMMSCs reduces their level of IFN-γ expression. This has a positive effect on the induction of immune tolerance and may provide a novel method of treating ACR of LT.
It also provides a new target of regulation of NKT cells. It should be highlighted that due to the lack of corresponding gene knockout rats, we were unable to verify the mechanism by which ACR of LT induced the complete disappearance of NKT cells. Thus, our animal model may be inadequate to address this issue and further validation in other animal models may be required in future studies.

Conclusions
Our ndings demonstrate that NMP can protect the DCD donor liver in vitro. BMMSCs infused via the portal vein using the NMP system can colonize the liver in large quantities. Moreover, HO-1 gene transfection can signi cantly improve the activity and survival time of BMMSCs in the liver. There was also a signi cant reduction in ACR following transplantation with donor livers populated with HO-1/BMMSCs, and an excessive activation of NKT cells was found to play a key role in the ACR. HO-1/BMMSCs inhibit the expression of co-inhibitory receptors on the surface of NKT cells through which inhibitory signals are transmitted to reduce the level of IFN-γ expression, thereby inhibiting the occurrence of ACR after LT. Together, our results provide a novel method for the application of BMMSCs, as well as a new target for the treatment of ACR and regulation of NKT cells.

Consent for publication
Not applicable.

Availability of data and materials
All datasets generated for this study are included in the manuscript and the supplementary materials.
Competing interests    blood, and liver of recipient rats. IFN-γ expression in the NKT cells in the liver of the BMP and HMP groups was signi cantly lower than that of the NMP group, and the HMP group was lower than the BMP group.
*P < 0.05.     were signi cantly higher than that in the NMP group. *P < 0.05.