Donor myeloid derived suppressor cells (MDSCs) prolong allogeneic cardiac graft survival through programming of recipient myeloid cells in vivo

Solid organ transplantation is a lifesaving therapy for patients with end-organ disease. Current immunosuppression protocols are not designed to target antigen-specific alloimmunity and are uncapable of preventing chronic allograft injury. As myeloid-derived suppressor cells (MDSCs) are potent immunoregulatory cells, we tested whether donor-derived MDSCs can protect heart transplant allografts in an antigen-specific manner. C57BL/6 (H2Kb, I-Ab) recipients pre-treated with BALB/c MDSCs were transplanted with either donor-type (BALB/c, H2Kd, I-Ad) or third-party (C3H, H2Kk, I-Ak) cardiac grafts. Spleens and allografts from C57BL/6 recipients were harvested for immune phenotyping, transcriptomic profiling and functional assays. Single injection of donor-derived MDSCs significantly prolonged the fully MHC mismatched allogeneic cardiac graft survival in a donor-specific fashion. Transcriptomic analysis of allografts harvested from donor-derived MDSCs treated recipients showed down-regulated proinflammatory cytokines. Immune phenotyping showed that the donor MDSCs administration suppressed effector T cells in recipients. Interestingly, significant increase in recipient endogenous CD11b+Gr1+ MDSC population was observed in the group treated with donor-derived MDSCs compared to the control groups. Depletion of this endogenous MDSCs with anti-Gr1 antibody reversed donor MDSCs-mediated allograft protection. Furthermore, we observed that the allogeneic mixed lymphocytes reaction was suppressed in the presence of CD11b+Gr1+ MDSCs in a donor-specific manner. Donor-derived MDSCs prolong cardiac allograft survival in a donor-specific manner via induction of recipient’s endogenous MDSCs.


Donor-derived MDSCs suppress alloreactive T cell activation in vitro.
Allogeneic mixed lymphocyte reaction (alloMLR) was performed to examine the regulatory function of donor-derived MDSCs. Co-culturing naive T cells isolated from C57BL/6 mice (H2K b , I-A b ) with BALB/c (H2K d , I-A d ) derived conventional DCs (cDCs) stimulates an alloreactive T cell proliferation. We observed that the addition of BALB/c MDSCs to this alloMLR system significantly inhibited the proliferation of CD4 + and CD8 + T cells compared to conventional myeloid derived cells (cMDCs) (Fig. 1A,B).

Donor-derived MDSCs protect cardiac allografts from acute rejection.
We then examined the in vivo suppressive function of donor-derived MDSCs in the allogeneic cardiac transplantation model. C57BL/6 recipients received a single-dose of intravenous injection of 1 × 10 6 BALB/c MDSCs or BALB/c cMDCs 7 days prior to the cardiac transplantation with BALB/c donors (Fig. 1C). We found that BALB/c MDSCs significantly prolonged allograft survival; in contrast, administration of cMDCs showed no difference to the group receiving PBS (PBS control n = 9, median survival time (MST) 7 days; BALB/c cMDCs control n = 7, MST 7 days; BALB/c MDSCs n = 8, MST 14 days, Fig. 1D).
Allografts were harvested on postoperative day (POD) 7 for hematoxylin and eosin (H&E) staining and immune fluorescence staining. H&E staining revealed the attenuated myocardial lesion as well as the decreased lymphocyte infiltration in the donor MDSCs treated group compared to the cMDC group ( Fig. 2A). Immune fluorescence staining showed that CD3 + T cells were significantly decreased in the MDSC group compared to the cMDC group. Of note, there was no difference in CD11b + cells infiltration between two groups (Fig. 2B).
Graft infiltration lymphocytes (GILs) were isolated from allografts (POD7) for flow cytometry analysis. Gating on the CD4 + FoxP3helper T cells, the proportion of effector T cells (Teff) defined as CD44 + CD62L lo population, was significantly decreased in the MDSCs treated group compared to the cMDCs treated group (Fig. 2C-upper). Within this Teff, the activation level measured by Ki67 was also significantly reduced in the MDSCs treated group (Fig. 2C-lower).

Donor-derived MDSCs induce highly immune suppressive endogenous MDSCs.
For mechanistic studies, spleens from C57BL/6 recipients of BALB/c hearts treated with BALB/c MDSCs were isolated on POD7 for flow cytometry analysis. We observed that a suppressive population CD11b + Gr1 +34 , defined broadly as MDSCs 34 , was significantly increased in the donor-derived MDSCs treated group (Fig. 4A). Gating on this population, we found that the PDL1 expression was markedly up-regulated in the MDSCs treated group compared to the cMDCs treated group (Fig. 4B). These data indicate that the donor-derived MDSCs administration not only increases the number of CD11b + Gr1 + population in recipients, but also program CD11b + Gr1 + cells to upregulate PDL1 expression. PD-L1/PD-1 interactions were reported to deliver co-inhibitory signals leading to attenuation of T cell responses both in vitro and in vivo 35,36 . Furthermore, PDL1 has been shown to be required for suppression of the autoimmune responses 37 . In consistent with this, we observed significantly increased CD11b + Gr1 + population in GILs from recipients treated with donor-derived MDSCs (Fig. 4C). We then hypothesized that CD11b + Gr1 + cells in recipients detected upon administration of donor-derived MDSCs play a crucial role in allogeneic immune suppression. To determine whether the increase of MDSCs (CD11b + Gr1 + ) is from the clonal expansion of the transferred donor-derived MDSCs or from the induction of recipients' endogenous MDSCs, we traced the transferred donor-derived MDSCs using H2K d antibody. 1 × 10 6 BALB/c MDSCs were intravenously injected to C57BL/6 recipients, and their splenocytes were examined at 3, 6 and 24 h post-injection. We found that the transferred donor-derived MDSCs peaked 3 h post-injection (PIH3) and then disappeared at PIH24 (Fig. S7A). In parallel, we analyzed the induction of endogenous MDSCs in recipients' spleens by gating on the recipient type MHC-I, H2K b . We found significant increase of these endogenous MDSCs in recipients starting at PIH24 (Fig. S7B). Next, we measured the MHC-II expression on the endogenous MDSCs as a marker of suppressive function 38 . Indeed, the MHC-II expression was markedly elevated on the endogenous MDSCs (Fig. S7B). These data suggest that while donor-derived MDSCs disappear within 24 h, the functional endogenous MDSCs are expanded in the recipients' lymphoid tissue.
Previous studies have shown that systemic administration of donor cells undergoing apoptosis promote donor-specific immunosuppression in vitro 39 and in vivo 40,41 . To study whether our generated donor MDSCs suppressed alloimmune reaction is related to the apoptotic donor cell mediated suppression, we used 7-AAD and Annexin-V to measure the frequency of apoptotic cells in MDSCs and control cMDCs. Result showed approximately 3.3% early apoptotic and 6.1% late apoptotic cells in generated donor MDSCs, which are significantly lower than in cMDCs (7.0% early apoptotic and 9.9% late apoptotic cells, p < 0.001) (Fig. S5). Taken together, we concluded that the suppressive function of donor-derived MDSCs is independent of apoptotic-cell-induced immune suppression.
As we have identified that donor-derived MDSCs induced recipient's endogenous MDSCs (CD11b + Gr1 + ) cells in recipients play a key role in allogeneic immune suppression (Fig. 5C,D), we then tested the donor-specific suppressive function of endogenous MDSCs by using the alloMLR. Naïve C57BL/6 spleen T cells were stimulated with BALB/c cDCs or C3H cDCs as a third-party. CD11b + Gr1 + cells were isolated by FACS from BALB/c MDSCs treated C57BL/6 recipient splenocytes at POD7 and were added as modulators (Fig. 5C). The proliferation of CD4 + and CD8 + T cells were determined by using violet dye dilution by flow cytometry. As naïve C57BL/6 T cells respond to BALB/c and C3H cDCs at a different rate, we calculated the relative suppression efficiency (SE) and compared the SE between the primary donor type cDCs and the third-party cDCs stimulated group (Fig. 5D). We observed doubled SE of CD4 + proliferation and a tripled SE of CD8 + proliferation in BALB/c cDCs stimulated group compared to C3H cDCs stimulated group (Fig. 5D). This data supports that the immune suppressive function of donor-derived MDSCs induce endogenous MDSCs is antigen-specific.
MDSCs are a heterogeneous population of immature monocytes and granulocytes 21 . In addition to Gr-1, CD11b combined with Ly6C and Ly6G has been utilized to distinguish the subpopulations of
We concluded that the CD11b + Gr1 + population in MDSCs versus cMDCs carry significantly distinct biological property. Only MDSCs derived CD11b + Gr1 + cells were able to suppress the alloimmune reaction.

Discussion
Achieving donor-specific tolerance without compromising the overall immune response is the ultimate goal in transplantation. Regulatory / tolerogenic DC (DCreg/ DCtol) based therapies have been shown to protect allografts and attenuate GvHD [48][49][50] . Previous study by others have demonstrated that in an NHP allogeneic kidney transplant model, donor-derived DCreg 17 administration prolonged the allografts survival, which indicated both the safety and efficacy of a single donor-derived DCreg infusion. However, the suppressive function of DCreg was limited due to the maturation and polarization under the stimulation of cytokines secreted by helper T cells during early phase of allo-reaction. MDSCs are diverse collection of immature myeloid-lineage cells, which show overlapping regulatory or suppressive properties with DCregs 51,52 . As MDSCs show more stable immature www.nature.com/scientificreports/ biological properties as well as enhanced immune suppressive capabilities 53 , we hypothesized that donor-type MDSCs induce donor-specific immune suppression in solid organ transplantation. We generated MDSCs from bone marrow cells within 6 days in the presence of GM-CSF 29 , TGFβ 30 and IL10 31,32 . As IFNγ is essential for the suppression capability of MDSCs via STAT1 signaling activation 54,55 and iNOS and NO production 56 , we added IFNγ on day 5 to promote the suppressive function of MDSCs 33 (Fig. S1A). To optimize the MDSCs generation protocol, we compared the biological properties of GM-CSF/IL10/TGFβ1 generated cells, freshly isolated bone marrow cells and myeloid derived cells generated with GM-CSF only (GM-MDCs). Results showed that only MDSCs generated under our protocol had the capability to suppress alloimmune reaction in vitro and in vivo (Figs. S3, S4). As donor-derived MDSCs require 6 days of culture, the clinical application of our approach is more suitable for living donor transplantation, such as lung, kidney and liver. However, we employed in this study a very stringent, reproducible and well established mouse allogeneic cardiac transplantation to test our hypothesis 57 .
Several conclusions are drawn from the current study. First, the in vitro generated MDSCs showed immune suppressive phenotype at the level of both protein and mRNA. Our generated MDSCs not only indicate the ≥ 85% purity of CD11b + Gr1 + cells but also display a stable immune suppressive phenotype (Fig. S1B). In addition, the low expression of costimulatory molecules, the generated MDSCs display a high level of PD1. Although PD1 is known to be mainly expressed on T cells, evidence has emerged indicating that other non-lymphoid innate cells also express PD1 36 . Recent studies demonstrate that expression of PD1 on myeloid cells reduces proinflammatory cytokine production 58 , diminishes innate immunity against bacterial infection 59 , and suppresses antigen-specific CD8 + T cell proliferation via decreasing the production of IL-2 and IFNγ 60 . It is known that successful prevention of acute allorejection using cellular approaches is dependent on migration of suppressor cells to secondary lymphoid organs 61 . Our MDSCs show a high expression of CX3CR1, that not only directs trafficking 62 , but also promotes migration 63,64 . CX3CR1 binds to CX3CL1, a membrane-bound chemokine that is highly expressed in the spleen and lymph nodes, and provides a strong survival signal to MDSCs under both steady-state and inflammatory conditions 65 . Transferred donor CX3CR1 hi MDSCs may rapidly migrates into spleen and lymph nodes, thus promoting an immune suppressive microenvironment. We also performed RNA-seq of MDSCs and cMDCs to compare the transcriptomic signatures. When we compared the RNAseq of MDSCs and cMDCs by principal component analysis (PCA), we found that MDSCs phenotypically separated from cMDCs, suggesting distinct transcriptional programs (Fig. S1E). Differential gene expression analysis (Fig. S1D, Table S3) showed the marked up-regulation of Tgfbi in MDSCs. Tgfbi is a secreted protein found in the ECM and it has an N-terminal secretory signal, four FAS1 homologous internal domains, and a cell attachment site (RGD) at its C terminus 66 . Tgfbi binds to the ECM through interaction of the YH motif in its FAS1 domains with collagens I, II, IV, and VI 67,68 . Its FAS1 domain interacts with its α3β1, αVβ3, and αVβ5 integrins on the cell surface. Study has shown that recombinant Tgfbi inhibited the proliferation and activation of CD4 + and CD8 + T cells stimulated with anti-CD3 mAb via reducing the production of IFNγ and granzyme B in vitro 69 . The mRNA expression of S100A8 and S100A9 is also significantly up-regulated in MDSCs. S100A9 protein plays critical role in inhibition of dendritic cell differentiation and accumulation of MDSCs via up-regulating reactive oxygen species (ROS) 70 . Study of transplant patients has demonstrated that high expression of S100A9 predicts better graft outcomes 71 . In line with these positive clinical outcomes, DCs treated with S100A8 or S100A9 maintain their immature phenotype, and show significant reduction in their capacity to induce T cell proliferation or to produce IFNγ 71 .
We also demonstrated that systemic administration of donor-derived MDSCs induced the endogenous MDSCs, which showed the donor-specific alloimmune suppressive capability. A significant increased population of CD11b + Gr1 + is observed in donor MDSCs treated recipients' spleen as well as in allografts. Depletion of this population in vivo abolished the donor MDSCs induced allograft protection. Furthermore, the addition of these recipients' endogenous MDSCs showed a powerful immune suppression in primary donor type APC (cDCs) stimulated alloMLR system in comparison to the third-party APC stimulated alloreaction, that indicated the immune suppression function of the recipients' endogenous MDSCs is donor-specific. While the exact mechanism requires more investigation, this observation may be explained in a two-step process. First, pretreatment with donor-derived MDSCs conditions recipients by inhibiting donor-reactive Teff. Pretreatment with donor-type MDSCs acts as an immune suppressive vaccine, which leads to a primary host versus donor antigen response. Recipient's cognate T cells engage with donor MDSCs presenting allo-antigen with poor co-stimulatory signal leading to the suboptimal activation 72 which in turn results in the generation of donorspecific Tregs 73 and anergy of donor-specific Teffs 74 . Our data supports this by showing increase in activated Treg (Fig. S6C) as well as decrease in activated Teff (Fig. 2C). Another important immune reaction take place simultaneously: recipient's endogenous MDSCs are induced with high level of MHC-II (Fig. S1B) and PDL1 (Fig. 4B) expression. Inhibition of antigen-specific Teff depends on the sufficient level of MHC class II 38 and PDL1 35,36 expressed on MDSCs, which was consistent with our Teff analysis results in Figs. 2C and 3.
Step 2: After allografts are implanted, the passenger lymphocytes and the graft tissue itself serve as the permanent resource of donor antigens. The microenvironment of the recipients promotes the donor-specific endogenous MDSCs which leads to the donor-specific allograft protection.
In conclusion, this study suggests that systemic administration of donor-derived MDSCs leads to immune regulation in a donor-specific manner via inducing endogenous MDSCs. Further research is needed to determine the detailed mechanism underlying in vivo MDSCs programming and to confirm these findings in human transplant recipients.
The graphic protocol is shown in Fig. S1A. Detailed reagents information is listed in Table S2.

RNA-seq.
RNAseq library preparations were performed as previously described 75 . Briefly, samples were lysed with RLT Buffer (Qiagen) and RNA was isolated using MyOne Silane Dynabeads (Thermo Fisher Scientific). RNA was fragmented and barcoded with 8 bp barcodes. Primers were removed using Agencourt AMPure XP bead cleanup (Beckman Coulter/Agencourt). Samples were amplified with 14 PCR cycles. Libraries were gel purified and quantified using a Qubit high sensitivity DNA kit (Invitrogen) and library quality was assessed using Tapestation high sensitivity DNA tapes (Agilent Technologies). RNA was sequenced on an Illumina Next-Seq sequencer (Illumina) according to manufacturer's instructions, sequencing 50 bp single end reads. Analysis was performed using the CLC Genomics Workbench version 8.0.1 RNAseq analysis software package (Qiagen). Briefly, reads were aligned (mismatch cost = 2, insertion cost = 3, deletion cost = 3, length fraction = 0.8, similarity fraction = 0.8) to the mouse genome and differential expression analysis was performed (total count filter cutoff = 5.0). Results were normalized to reads per million.
Heterotopic cardiac transplantation. All transplant procedures were performed under anesthesia with isoflurane. Fully vascularized heterotopic hearts from BALB/c or C3H were transplanted into C57BL/6 recipients using a microsurgical technique 14,57 . Graft survival was considered complete at the time of cessation of a palpable heart beating and confirmed visually by laparotomy.
Isolation of lymphocytes from grafts. Tissues were disrupted mechanically in 10 ml digestion solution, which include 0.5 mg/ml collagenase IV (Sigma, 5138), 50 U/ml DNaseI (Invitrogen, 18047019) in RPMI-1640 medium and incubated at 37 °C for 30 min. After that, 10 ml of iced RPMI 1640 with 5% FBS was added. The suspension was filtered through a nylon mesh (70 μm) to remove aggregates. The resulted cell suspension was centrifuged at 800g for 5 min to pellet the cells. The pellet was suspended in 5 ml PBS, loaded onto 5 ml of Lympholyte-M (Cedarlane, CL5035) and centrifuged at 1500 g for 25 min at room temperature. Cells were isolated from the Lympholyte-M interface and washed twice in PBS at 300 g for 5 min and prepared for flow cytometry analysis.

Mixed lymphocyte reaction (MLR).
Allogeneic MLR was performed in duplicate in 96-well, round-bottom plates (Corning, 7007). Nylon wool-eluted spleen T cells (2 × 10 5 /well) were labeled with CellTrace violet (Invitrogen, C34557) and used as responders. Cultures were maintained in the complete medium for 3-4 days at 37 °C in 5% CO 2 in air. The reaction system and other details are shown in the associated figures.
Histopathology. Grafts harvested on POD7 were fixed in 10% formalin solution (Sigma, HT5011) and then embedded in paraffin. Sections of 4 µm were made for hematoxylin and eosin (H&E) staining.
Immuno fluorescent staining. Grafts harvested on POD7 were embedded in OCT compound (Sakura, 4583) and preserved in − 80 °C freezer. Cryo sections of 4 µm were made for immuno fluorescent staining 76 . Antibodies used in this experiment is listed in Table S2.
RNA preparation and quantitative reverse transcriptase-polymerase chain reaction (qRT-PCR). Cardiac grafts were harvested on POD3 and POD7 and submerged in RNAlater stabilization for freezing (Sigma, R0901). Total RNA was extracted from frozen tissue samples using TRIzol method (Invit-