Combining Exosomes Derived from Immature DCs with Donor Antigen-Specific Treg Cells Induces Tolerance in a Rat Liver Allograft Model

Allograft tolerance is the ultimate goal in the field of transplantation immunology. Immature dendritic cells (imDCs) play an important role in establishing tolerance but have limitations, including potential for maturation, short lifespan in vivo and short storage times in vitro. However, exosomes (generally 30–100 nm) from imDCs (imDex) retain many source cell properties and may overcome these limitations. In previous reports, imDex prolonged the survival time of heart or intestine allografts. However, tolerance or long-term survival was not achieved unless immune suppressants were used. Regulatory T cells (Tregs) can protect allografts from immune rejection, and our previous study showed that the effects of imDex were significantly associated with Tregs. Therefore, we incorporated Tregs into the treatment protocol to further reduce or avoid suppressant use. We defined the optimal exosome dose as approximately 20 μg (per treatment before, during and after transplantation) in rat liver transplantation and the antigen-specific role of Tregs in protecting liver allografts. In the co-treatment group, recipients achieved long-term survival, and tolerance was induced. Moreover, imDex amplified Tregs, which required recipient DCs and were enhanced by IL-2. Fortunately, the expanded Tregs retained their regulatory ability and donor-specificity. Thus, imDex and donor-specific Tregs can collaboratively induce graft tolerance.

To further improve allograft survival time, we co-injected 20 μ g of donor imDex with donor-specific Tregs, as imDCs have been reported to work with Tregs in vivo and in vitro 8,[17][18][19] . While the MSTs of recipients from the imDex and Tregs-alone treatment groups reached 37 d and 34 d, respectively, the combination of donor imDex with donor-specific Tregs induced the long-term survival of liver allografts (6 recipients survived over 100 d, p = 0.007 compared with the 20 μ g imDex group, n = 9; p = 0.0083 compared with the BN-specific Tregs group, n = 9) (Fig. 3D/ Table 1). Furthermore, we also combined donor imDex with fresh isolated Tregs as a control group, and the MST of this group reached 47 d (p = 0.0386, n = 9, compared with the experimental group; p = 0.283, n = 9, compared with the 20 μ g imDex group) ( Fig. 3D/Table 1), which suggested that it matters that the combined Tregs were donor-specific, for the extended survival time in the experimental group compared with the imDex alone group.

Infiltrating cells and rejection symptoms in allografts were reduced.
To evaluate the degree of allograft rejection, we harvested grafts on days 0, 10, 35 and 100 after transplantation. The rejection activity index (RAI) according to the Banff schema was used as the standard to evaluate rejection severity in each group. Haematoxylin and eosin (HE) staining showed that the levels of infiltrating inflammatory cells varied in different groups and on different days (Fig. 4A). The 0-day grafts from each group showed normal results.
On the 10 th day, the graft pathology results showed different degrees of degeneration and necrosis in liver parenchyma cells in the four groups. HE staining of the untreated group revealed obvious inflammatory cell infiltration, hepatic lobule structural disorder and severe intravascular dermatitis (Fig. 4A). The RAI was 8.3 ± 0.3, and the rejection was Severe (Fig. 4B). Recipients in the 20 μ g imDex treatment group and the donor-specific Tregs treatment group (Fig. 4A/B) showed similar lymphocytic infiltration in portal areas; the Banff classification was Moderate (RAI of the imDex group was 5.0 ± 0.6 and of the Tregs group was 5.3 ± 0.3). The 20 μ g BN imDex/BN-specific Tregs co-treatment group (Fig. 4A/B) exhibited a very small amount of mononuclear cell infiltration and the Banff grade was Undefined. The RAI (0.7 ± 0.3) was significantly reduced (Fig. 4B) compared with the other three groups (untreated group: p < 0.001, n = 3; imDex treated group: p < 0.001, n = 3; donor-specific Tregs group: p < 0.001, n = 3).
On the 35 th day, a large number of infiltrating cells and chronic rejection symptoms such as biliary atresia and cholestasis appeared in the imDex and Tregs treatment groups. However, no chronic rejection symptoms appeared in the co-treatment group (Fig. 4A) on the 35 th day. In this group, inflammatory cell infiltration into the grafts was reduced compared with the imDex or Tregs treatment groups. The RAI was 6.7 ± 0.3 (co-treatment vs. imDex alone, p < 0.01, n = 3; co-treatment vs. Tregs alone, p < 0.05, n = 3) (Fig. 4C). These results showed that 20 μ g of BN imDex combined with BN-specific Tregs could reduce immune rejection in liver allografts.
On the 100 th day, the co-treatment group grafts exhibited fibrous regeneration and hepatic lobule structural disorder, but mononuclear cell infiltration was reduced (Fig. 4A). The Banff grade was Undefined, and the RAI was 1.3 ± 0.3, which was significantly reduced (Fig. 4D) compared with the 35-day grafts (RAI = 6.7 ± 0.3, p < 0.001, n = 3). The 10-, 35-and 100-day pathology changes suggested that co-treatment reduced rejection and helped receipt livers regenerate after undergoing slight acute rejection. Recipient immune responses were suppressed in a donor-specific manner. To explore the immune statue in recipients, we assessed the anti-donor cellular response in recipients after transplantation. As described in the Methods, 5 × 10 4 purified T cells (isolated from splenocyte harvested 10 d after transplantation from recipients) were incubated with 5 × 10 4 irradiated SDC (stimulator cells) derived from either donors (BN) or other allogeneic donor (F344). As shown in Fig. 5A/B, total T cells from co-treated rats displayed a significant decrease in proliferation against BN SDCs compared with T cells from untreated recipients (p < 0.001, n = 3), 20 μ g BN imDex-treated recipients (p < 0.001, n = 3) and BN-specific Tregs-treated recipients (p < 0.001, n = 3). However, total T cells from all four groups proliferated almost equally with the F344 SDCs used as stimulator cells (Fig. 5A/B). Furthermore, we observed that the IFN-γ levels in the T cell/donor SDC co-culture supernatant were similarly inhibited in a donor-specific manner (co-treatment group vs. untreated group: p < 0.01, n = 3, BN SDCs as stimulators, Fig. 5C; co-treatment group vs. untreated group: p > 0.05, n = 3, F344 SDCs as stimulators, Fig. 5C). These results demonstrated that the immune response of T cells in co-treated recipients was inhibited. Tregs distribution in the recipients. To study the distribution of Tregs in vivo, we designed an independent test with CFSE-labelled Tregs. Twelve extra recipients accepted liver transplantation and four different treatments in the same dosage regimen described above, except that CFSE-labelled Tregs were used. On the 10 th day after transplantation, liver grafts, mesenteric lymph nodes and spleens were harvested for immunofluorescence. As shown in Fig. 6A-C, CFSE-labelled Tregs were distributed in these organs in the BN-specific Tregs treatment group and the co-treatment group. In liver grafts, we found that the number of CFSE + cells in the co-treatment group was significantly higher than in the BN-specific Tregs treatment group (p = 0.0413, n = 48, Fig. 6D). In mesenteric-draining lymph nodes and spleens, similar results were observed (lymph nodes, p = 0.0105, n = 48; spleens, p = 0.0310, n = 48; Fig. 6E/F). Therefore, imDex might promote the proliferation of Tregs, which will enhance regulatory effects of Tregs in vivo, similar to imDCs 17,19 . Exploring the mechanism of the synergistic effects of donor imDex and donor-specific Tregs with in vitro and in vivo assays. We hypothesized that the infused exosomes were the reason why CFSE-labelled Tregs were increased in the co-treatment group. To verify this, we implemented in vitro Tregs proliferation assays. We used 2 × 10 4 Lewis SDCs as "assistance cells" and 200 units/ml IL-2 as an "assistance cytokine".
As shown in Fig. 7A, no considerable proliferation was observed in the untreated group, imDex group or Lewis SDC group. However, Treg proliferation was increased in the imDex/Lewis SDC group compared to the imDex group (p < 0.001, n = 3) or the Lewis SDC group (p < 0.001, n = 3), indicating that imDex can amplify Tregs in vitro and that DCs are essential for this effect. When IL-2 was added to the administration protocol, Treg proliferation increased compared to the imDex/Lewis SDC group (p < 0.05, n = 3), suggesting that IL-2 improved the ability of imDex to expand Tregs.
Next, we performed Treg suppression assays with the expanded Tregs. A dose-dependent suppression of CD8a + T cell proliferation is shown in Fig. 7C/D. The expanded Tregs were still donor-specific, as dose-dependent suppression was not seen when F344 SDCs were used as stimulators (Fig. 7C/D). In the IFN-γ inhibition assays, we observed similar results (Fig. 7E). These results suggested that imDex-expanded Tregs maintain their regulatory ability and do not lose donor specificity.
ImDex was injected into Lewis rats via the caudal vein along with CFSE-labelled Tregs to detect proliferation of exogenous Tregs in vivo. On the 10 th day after transplantation, CD4 + CD25 + T cells were isolated from spleens with magnetic sorting and analysed by FCM (Fig. 7F/G). The CFSE fluorescence intensity was undetectable in the no treatment and 20 μ g BN imDex treatment group. The percentage of divided CFSE-labelled Tregs was analysed. In the imDex/Tregs co-treatment group, the Treg proliferation rate was higher than in the Tregs treatment group (p = 0.0025, n = 3, Fig. 7G). These results suggested that imDex also amplifies Tregs in vivo.

Discussion
Our results confirm that the combined use of proper doses of donor imDex and donor antigen-specific Tregs can induce rat liver allograft tolerance without the need for immunosuppressive agents. This active tolerance could be detected in cell mixing experiments in vitro (Fig. 5) and survival analysis/pathology analysis in vivo (Figs 3/4). Meanwhile, we found that exogenous Tregs were widely distributed in liver grafts, spleens, and mesenteric lymph nodes (Fig. 6) and that imDex could amplify Tregs. Recipient DCs were essential for this imDex function, and IL-2 was also helpful (Fig. 7). Fortunately, the expanded Tregs retained their regulatory ability and specificity, remaining tenable in the in vivo assay, which may explain the synergistic effect and the induction of tolerance by Tregs and imDex (Fig. 7).
ImDCs can inhibit immune rejection 20,21 , and exosomes have many advantages, including their stable nature and easy storage. We therefore added imDex to our treatment protocols to verify whether imDex can function similarly to imDCs in liver recipients. We found that the most effective imDex dosage (20 μ g at one of three time points) prolonged the rat liver survival time, which is consistent with previous reports 11 . However, the optimal exosome dosage varies between studies and is not even described in some reports 6,22 ; these differences may be due to the use of different animals, diverse models, various exosome sources or different dose gradient designs. In the in vivo assay, we verified that only donor-derived imDex (20 μ g at one of three time points) prolonged recipient survival time, which is consistent with previous imDex studies 8,23 . However, this finding appears contradictory to those using DCs, as it was reported that infusion of either donor-20,24-26 or recipient-derived 27,28 DCs with tolerogenic properties prolonged allograft survival time. Considering that there are at least two properties underlying the tolerogenic function of DCs, including "inherently tolerogenic properties" (clonal deletion, inhibition of T effector cells, and the expansion or induction of Tregs) and "negative cellular vaccines" 21 (donor-derived tolerogenic DCs have donor antigen but do not induce rejection), we believe that donor-derived tolerogenic DCs with the "negative cellular vaccine" property may have some advantages and may work with relatively low cell numbers. Indeed, our results indicated that donor-derived imDex may also possess the "negative cellular vaccine" property. However, we did not compare imDex and mDex in our study, which may be a limitation.
After magnetic bead isolation and incubation with donor SDCs, the FOXP3 + rate slightly increased, consistent with previous studies 17,19 . However, the CD127 + rate decreased ( Fig. 2A/B). While CD8a + cytotoxic T cells play an important role in the cellular immune response to transplantation, we observed that SDC-expanded Tregs could inhibit these responses, including proliferation and the production of inflammatory cytokines, with donor specificity (Fig. 2C-E), indicating that SDC-expanded Tregs have the potential to inhibit the allograft rejection reaction. Although there are some controversies surrounding Treg specificity [29][30][31] , we found that donor-specific Tregs did prolong the liver allograft survival time in a donor-specific manner (Fig. 3C/Table 1). Therefore, we argue that donor specificity is important for Tregs in both the in vitro suppression response and the in vivo regulation of immune rejection for transplantation models.
Combined treatment with donor imDex and donor antigen-specific Tregs led to long-term survival (six of nine recipients in the co-treatment group survived over 100 d). However, in the untreated group, the MST was 10 d. In the imDex or Tregs treatment groups, the MST was 37 or 34 d, respectively (Fig. 3, Table 1). These results suggest that in the co-treatment group, acute rejection was delayed and relatively slight compared with the other groups and the findings of another study 32 . The pathological results of combined treatment also showed a relatively delayed and slight acute rejection (Fig. 4), and at 100 d after transplantation, we observed considerable regeneration. We then isolated total T cells from recipients and found that their proliferation and IFN-γ production were both reduced in a donor-specific manner. Together, these results demonstrate that allograft tolerance was achieved.
To explore exogenous Tregs distribution, we administered CFSE-labelled Tregs to recipients and found they were widely distributed in liver grafts, spleens and mesenteric lymph nodes. Interestingly, in the imDex/ CFSE-labelled Tregs group, the number of exogenous Tregs was significantly higher than in the CFSE-labelled Tregs group (Fig. 6). We speculate that imDex amplified the exogenous Tregs, and this phenomenon may explain the synergistic effects in the co-treatment group. To verify this hypothesis, we performed in vitro Treg expansion assays (Fig. 7A/B). In these assays, imDex amplified Tregs only when SDCs were incubated in this system ( Fig. 7A/B), confirming our and Morelli's conjecture. Morelli reported 4 that imDex works with DCs in vivo and speculated that this mechanism underlies the effects of imDex in vivo. These results also suggest that imDex expands Tregs in a similar manner to DCs, as donor DCs function in recipients with recipient DC assistance, and deletion of recipient DCs deters the therapeutic effect of donor DCs, as was reported by Wang. However, the mechanism of interaction between exosomes and assistant DCs has not been fully clarified. Genally speaking, there are two types of hypotheses: i) exosomes directly fuse with the membranes of assistant DCs or bind on the surface of DCs [33][34][35] , so that intact donor MHC molecules can present on recipient DCs and directly interact with effector cells by means of a "direct recognition pathway" 19 ; ii) exosomes are endocytosed by recipient DCs, and donor MHC molecules and antigen peptides will be presented by assistant DCs 36,37 . In this way, exosomes only interact with effector cells by means of classic "indirect recognition pathway". It remains unknown which pathway is dominant in donor imDex inducing recipient Tregs proliferative response.
Next, we found that IL-2 improved the ability of imDex to expand Tregs, which is similar to the DCs expanding assays reported by Yamazaki 16 showing that IL-2 improved the ability of allogeneic DCs to amplify Tregs. Furthermore, we analysed the proliferation rate of CFSE-labelled Tregs in vivo and found that Treg proliferation increased with imDex treatment. This result confirms that imDex can expand Tregs both in vitro and in vivo. However, due to a lack of effective methods, we could not verify the mechanism of interaction between imDex and recipient DCs. We also did not explore whether imDex interacted with a wider range of cells.
In conclusion, imDex plays an immune regulatory role in rat liver transplantation. Our results identified the optimal dose for imDex administration and further verified that Treg antigen specificity is critical for immune regulation in allogeneic rat transplantation. Furthermore, the combined use of both imDex and Tregs induced liver allograft tolerance, and imDex amplified Tregs most likely through binding/fusing with recipient DCs and being presented by recipient DCs, which could be enhanced by IL-2. Synergistic effects between imDex and Tregs may explain why the co-treatment group survived longer than imDex or Tregs treatment groups. As liver allograft tolerance was induced and imDex possesses some advantages over imDCs, this study provides a new method for the regulation of transplantation immunity for clinical work.

Methods
Animals. Male BN RT1 n , F344 RT1 lv and Lewis RT1 l rats (250-300 g) were purchased from Vital River, Inc. (Beijing, China). All rats were bred in a specific-pathogen-free animal facility. The research protocol was approved by the Animal Experiment Administration Committee of the Fourth Military Medical University and the research was carried out in accordance with the approved contents. Anaesthesia during liver transplantation and specimen procurement was maintained with ethyl ether, and all efforts were made to minimize suffering. Isolation of CD4 + CD25 + T cells and CD8a + T cells. CD4 + CD25 + cells were isolated from Lewis rat spleens with a MagCellect* Rat CD4 + CD25 + Regulatory T Cell Isolation Kit (R&D, Minneapolis, MN, USA, off the shelf now) according to instructions. Positive selection of CD8a + cells from rat spleens was performed using anti-CD8a beads.
Bone marrow DCs and spleen DCs. DCs were generated from bone marrow (BM) cells as previously described 8,11,23 with some modification. BM cells were cultured in complete medium (endotoxin-free, 10 ng/ml rat IL4 and 6 ng/ml murine GM-CSF were purchased from Peprotech (Rocky Hill, NJ, USA) for a total of 11 d. On the 6 th day, the medium was refreshed while the original medium supernatant was harvested. From then on, some of the cells were cultured in medium containing LPS (100 ng/ml, Sigma, St. Louis, USA) for 1 d and the medium was then refreshed. At the end of 11 d culture, immature DCs, mature DCs and the corresponding supernatants were harvested. Spleen DCs were isolated from donors or recipients with anti-DC (ox62) beads as per the instructions.
Exosome preparation. Exosomes were isolated from mDC and imDC supernatants (imDC supernatants at 6 and 11 d were pooled to obtain greater output) using an exosome isolation kit (Life Technologies, Carlsbad, California, USA) according to the manufacturer's instructions. The pellet was then re-suspended with saline. Approximately 15 μ g of exosomes were harvested from 1 × 10 7 mDCs or imDCs. The exosome amount was evaluated based on the amount of protein using the Bradford assay. Then, exosomes were stored at −80 °C for future use.
Flow cytometry. As previously described 8 , except that aldehyde/sulfate latex beads were purchased from Thermo Fisher Scientific (Waltham, MA, USA). The beads attached to the two types of exosomes were stained with FITC-MHC class I, FITC-MHC class II, PE-CD80 or PE-CD86 antibodies. Beads stained with FITC and PE isotype antibodies formed the control groups. CD4 + CD25 + T cells and CD8a + T cells were stained with FITC-CD4, PE-CD25, APC-FOXP3, unconjugated CD127, Alexa Fluor 680-conjugated donkey anti-rabbit IgG and FITC-CD8 antibodies as described in the product specifications. Then, the beads or cells were assessed by FCM using a FACSCalibur (BD Immunocytometry Systems, Franklin Lakes, NJ, USA), and data were analysed with FlowJo 7.6. Electron microscopy. As described by Meckes 39 , the sample was diluted ten times, loaded onto a copper grid at room temperature, and examined with a CM120 Philips Biotwin electron microscope (Phillips Electronic Instruments).
Orthotopic liver transplantation, infusion therapy and survival time recording. Surgical procedures were performed as Kamada 40 described. In this procedure, we used BN rats as liver donors, F344 rats as other allogenic donors and Lewis rats as allograft recipients. Cuffs were used. No immunosuppression was given to recipient rats in this study. Different doses of imDex and/or exogenous Tregs (2 × 10 6 per time-point), either unlabelled or labelled with 10/20 μ M CFSE (eBioscience, San Diego, CA, USA), were transferred via caudal injection 7 d before, the day of, and 7 d after transplantation. Recipients that died within 3 d were regarded as technical failures and excluded from further analysis. On day 0, 10, 35 and 100 after transplantation, three recipients were sacrificed for HE analysis and/or anti-donor cellular response assays. These animals were excluded from the survival analysis. If needed, extra transplantation was carried out to ensure 9 recipients in each group. Recipient survival times were recorded and analysed, and the experiment ended after 100 d or 60 d.

Mixed leukocyte reactions.
In the Treg suppression assay, Fresh Tregs, Lewis-derived BN/F344 antigenspecific Tregs and CFSE (5 μ M)-labelled CD8a + T cells were obtained and co-cultured at various Treg/CD8a + T cells ratios (BN SDCs as stimulator cells, irradiated by 20 Gy/2,000 rad γ -ray) in 200 μ l of T cell medium in a 96-well round-bottom plate at 37 °C. Starting cell numbers were 5 × 10 4 Tregs, 5 × 10 4 CFSE-labelled CD8a + T cells, and 5 × 10 4 stimulator cells. Different numbers of Tregs were added later. Five days later, CFSE intensity was measured by FCM, and the rate of divided CFSE-labelled cells was analysed with FlowJo 7.6 and recorded as proliferation rate. The expanded Treg suppression assays were carried out as mentioned above, except that the expanded Tregs were used as suppressor and BN/F344 SDCs were used as stimulator. In the recipient leukocyte proliferation assay, as described by Peche 11 , 5 × 10 4 Lewis total T cells, purified from recipients' spleencytes by Rat T Cell Enrichment Column kit (R & D, Minneapolis, MN, USA), were labelled with 5 μ M CFSE and were added to 96-well round-bottom plates with equal numbers of irradiated SDCs derived from BN/F344. Five days later, CFSE intensity was measured and analysed as mentioned above. The IFN-γ levels in this culture supernatant were determined using the Rat IFN-γ ELISA kit purchased from NeoBioscience (Beijing, China).
Scientific RepoRts | 6:32971 | DOI: 10.1038/srep32971 Histological examinations. Liver grafts harvested on day 0, 10, 35 and 100 after transplantation were fixed for 72 h, embedded in paraffin, sectioned at 6 μ m and stained with HE for histological examination. The RAI according to the Banff schema 32,41 was used to evaluate allograft rejection. An independent test was designed with CFSE-labelled Tregs (20 μ M CFSE, 10 min at room temperature) in which 12 extra recipients received liver transplantation and four different treatments (no treatment, 20 μ g of imDex, CFSE-labelled BN-specific Tregs or imDex/CFSE-labelled Tregs co-treatment. Each group contained 3 recipients.). On the 10 th day after transplantation, sample fragments were snap-frozen in OCT (Tissue-Tek, Elkhardt, IN, USA), cut into 8-μ m sections, and stained with DAPI.
Treg proliferation assays in vitro and in vivo. We cultured 1 × 10 4 Lewis-derived BN-specific Tregs (5 μ M CFSE labelled) in 96-well plates and incubated them with 0.5 μ g of BN imDex, 2 × 10 4 Lewis SDCs and/or IL-2 (200 units/ml, recombinant rat IL-2 purchased from Peprotech, Rocky Hill, NJ, USA). Seven days later, proliferation was analysed. In the in vivo assays, Tregs were labelled with 10 μ M CFSE. After 10 min at room temperature, the reaction was stopped by the addition of an equal volume of FCS, followed by washing in PBS. A total of 2 × 10 6 Tregs were injected as described above. Ten days after transplantation, CD4 + CD25 + T cells were isolated from recipient splenocytes by magnetic sorting. Gated on the CD4 + CD25 + CD127 − CFSE + cell subset, the percentage of divided CFSE-labelled Tregs was analysed.
Statistics. The data points in the line graphs and bar graphs represent the mean ± SEM or mean + SEM, respectively. Data were expressed as the means ± standard error of means (SEM) and analysed using Student's two-tailed t test for comparison between two groups, One Way ANOVA for comparison among more than two groups or the log-rank test for survival analysis. Values were analysed using GraphPad Prism 5.0 (GraphPad, San Diego, CA, USA). The results were deemed statistically significant if the p-value was < 0.05. Statistical significance is indicated in each figure where applicable.