Cross‐presentation of dead‐cell‐associated antigens by DNGR‐1+ dendritic cells contributes to chronic allograft rejection in mice

The purpose of this study was to elucidate whether DC NK lectin group receptor‐1 (DNGR‐1)‐dependent cross‐presentation of dead‐cell‐associated antigens occurs after transplantation and contributes to CD8+ T cell responses, chronic allograft rejection (CAR), and fibrosis. BALB/c or C57BL/6 hearts were heterotopically transplanted into WT, Clec9a−/−, or Batf3−/− recipient C57BL/6 mice. Allografts were analyzed for cell infiltration, CD8+ T cell activation, fibrogenesis, and CAR using immunohistochemistry, Western blot, qRT2‐PCR, and flow cytometry. Allografts displayed infiltration by recipient DNGR‐1+ DCs, signs of CAR, and fibrosis. Allografts in Clec9a−/− recipients showed reduced CAR (p < 0.0001), fibrosis (P = 0.0137), CD8+ cell infiltration (P < 0.0001), and effector cytokine levels compared to WT recipients. Batf3‐deficiency greatly reduced DNGR‐1+ DC‐infiltration, CAR (P < 0.0001), and fibrosis (P = 0.0382). CD8 cells infiltrating allografts of cytochrome C treated recipients, showed reduced production of CD8 effector cytokines (P < 0.05). Further, alloreactive CD8+ T cell response in indirect pathway IFN‐γ ELISPOT was reduced in Clec9a−/− recipient mice (P = 0.0283). Blockade of DNGR‐1 by antibody, similar to genetic elimination of the receptor, reduced CAR (P = 0.0003), fibrosis (P = 0.0273), infiltration of CD8+ cells (p = 0.0006), and effector cytokine levels. DNGR‐1‐dependent alloantigen cross‐presentation by DNGR‐1+ DCs induces alloreactive CD8+ cells that induce CAR and fibrosis. Antibody against DNGR‐1 can block this process and prevent CAR and fibrosis.


Introduction
Chronic rejection ultimately leading to allograft fibrosis and dysfunction constitutes a major constraint to long-term graft survival in transplantation [1][2][3][4][5]. In contrast to early post-transplant acute rejection episodes, which can be effectively treated with standard immunosuppressors to prevent activation and proliferation of alloreactive T lymphocytes, chronic rejection cannot be controlled sufficiently by these immunosuppressive therapies [6][7][8][9]. Activation of the immune system in the chronic setting is likely mediated through recognition of non-infectious damaged tissues [10][11][12]. This connection between cell injury and allograft rejection was proposed many years ago and posits that ischemia-reperfusion injury and other traumas during transplantation results in death of allograft cells, which expose intracellular molecules that trigger defensive immune responses in the host [13]. These intracellular molecules are termed damage-associated molecular patterns (DAMPs) and include molecules such as HMGB-1, ATP, uric acid, DNA, and others, which can in many instances engage innate immune receptors and promote inflammatory and adaptive immune responses [14,15].
Given the link between cell death and immunity to allotransplants, the present study aimed to test a possible involvement of DNGR-1 and of cDC1 in cross-priming CD8 + T cells against allograft antigens in a model of chronic rejection. Using a combination of genetic approaches and antibody blockade, we report that chronic allograft rejection and fibrosis can be prevented by inhibition of the DNGR-1 receptor and/or loss of DNGR-1 + DCs. These findings identify DNGR-1 as an important point of control in immunity to allotransplants that could be targeted therapeutically to ameliorate chronic allograft rejection.

Cell necrosis in cardiac allografts exposes DAMPs and leads to infiltration by DNGR-1 + DCs
To investigate if cardiac allogeneity impacts early immune responses, BALB/c allografts transplanted into CD4 + T cell depleted C57BL/6 mice, which is a model of chronic allograft rejection and shows median allograft survival of 27 days, were evaluated by histology and RNA analysis at early times after transplantation (Fig. 1A). Immunohistochemistry revealed early cell infiltration, and necrosis of cells within allografts (nucleus loss in cardiomyocytes, eosinophilic cytoplasms) at day 3 with further increases at day 5 after transplantation in comparison to syngeneic grafts ( Fig. 1B; Supporting Information Fig. S1A and B). Additionally, in allografts, expression of death receptor and necrosis related genes such as TNF, myelin-associated glycoprotein (MAG), Fas ligand (Fasl), cytochromes, and CD40 was upregulated beginning at day 1 and increasing until day 5 after transplantation (Supporting Information Fig. S1C) [38][39][40][41][42]. This was accompanied by a higher CD11c + DNGR-1 + DC infiltration of the allografts (BALB/c→C57BL/6) in comparison to syngrafts as demonstrated by double immunofluorescence staining ( Fig. 1C and D; p < 0.001; Supporting Information Fig. S2). In contrast, no difference in DC subset composition was seen in the spleens of mice transplanted with allografts and syngrafts (Supporting Information Fig. S3).
Comparable to the results in mice, human cardiac allografts undergoing rejection showed greater leucocytic cell infiltration and collagen deposition at day 20 after transplantation than allografts without signs of rejection ( Fig. 1E; mouse data, Fig. 3B and C). Notably, the human samples from rejecting hearts were infiltrated by CD11c + DNGR-1 + DCs whereas these DCs were absent in human biopsies from healthy transplants ( Fig. 1F and G; p < 0.001).

Selective apoptosis and DNGR1 −/reduces CD8 + T cell allorecognition and allograft rejection
To formally assess the CD8 T cell alloimmune response within our model in vivo, we performed additional experiments with cytochrome C injection of recipient mice. In previous functional studies, both in vivo and in vitro it has been shown that cytochrome c profoundly abrogates OVA-specific CD8 T cell proliferation through its apoptosis-inducing effect on cross-presenting DCs [43]. In these experiments, in vivo injection of cytochrome C abolished the induction of cytotoxic T lymphocytes to exogenous antigen and reduced subsequent immunity to tumor challenge. Importantly, this model allows assessment of cross-presentation that is totally in vivo that is ideal for our setting of alloimmunity in transplantation [43].
In detail, we injected C57BL/6 recipient mice that were transplanted with BALB/c hearts with three dosages of cytochrome C and analyzed analogous to our original experiments. Cytochrome C injected recipient mice showed significantly decreased immune cell infiltration and histology rejection score of allografts compared to WT recipients (p = 0.0055; Supporting Information Fig. S6A). This was accompanied by a significant reduction of collagen I deposition in allografts both in Masson's trichrome staining and PCR (p = 0.0152; Supporting Information Fig. S6B) and also reduced CTGF levels (p = 0.0260; Supporting Information Fig. S6C). These results were corroborated by immunofluorescent detection of phosphorylated Smad3 in allografts transplanted into WT recipient animals but not into cytochrome c injected mice (Supporting Information Fig. S6D).
Additionally, in all our experimental groups that show reduced allograft rejection due to either reduced BATF3 dependent DCs or impaired DNGR-1 receptor function (Cleac9a −/− and DNGR-1 antibody treated mice) reduced IFN-γ levels were detected. This altogether underlines that indeed DNGR-1 + DCs cross-prime CD8 + T cells and that these primed CD8 T cell effector responses contribute to allograft rejection and fibrosis.

Blockade of DNGR-1 receptor prevents chronic allograft rejection and fibrosis
Finally, blockade of DNGR-1 with specific mAb was tested for the ability to phenocopy genetic loss in allograft recipients and diminish chronic allograft rejection and fibrosis. Notably, mAb treatment significant prolonged allograft survival (p < 0.0001; Fig. 1A), reduced numbers of allograft-infiltrating cells (p = 0.0003; Fig.  7A) and collagen I deposition into allografts (P = 0.0273; Fig. 7B). Similar to genetic disruption of Clec9a, the profibrotic cytokines were significantly reduced in mAb treated compared to the control group (active TGF-β 1 p = 0.0035; CTGF p = 0.0267; Fig. 7C). Further, mAb treatment significantly reduced the infiltration of allografts by CD8 + T cells (p = 0.0006; Fig. 7D), as well as the We corroborated these results in the fully immunocompetent Bm12→C57BL/6 heart transplantation model. In this model, temporary CD4 + T cell depletion is not necessary to induce chronic allograft rejection and fibrotic organ remodeling. Antibody blockade of DNGR-1 also resulted in reduction of chronic rejection and fibrosis, similar to the extent seen in the BALB/c into C57BL/6 with CD4 + T cell depletion model (Supporting Information Fig. S7A-E). Therefore, DNGR-1 blockade is a useful means of ameliorating chronic rejection in two mouse models of allotransplantation.

Discussion
A transplanted organ experiences various traumas, from physical manipulation to varying oxygen tension, that can cause death of graft cells and consequently cause release of DAMPs [15,44]. DAMPs have been proposed to contribute to the inordinate immunogenicity of allografts through engagement of DAMP receptors on host immune cells but the mechanisms involved remain unclear [12,15]. Here, we establish a connection between DAMP recognition and chronic CD8 + T cell-mediated responses leading to cardiac allograft fibrosis. We show that a subset of specialized DCs with a superior capacity to cross-prime CD8 + T cells,  cDC1s, is absolutely crucial for CD8 + T cell-mediated allograft rejection. Importantly, we further reveal a critical role for the DAMP receptor, DNGR-1, expressed by the cDC1 subset in mediating this rejection and show that blockade of this receptor contributes to allograft acceptance. Our findings help illuminate the immune mechanisms involved in allograft rejection and highlight how increased understanding of such processes can suggest potential avenues for therapeutic intervention.
Allograft rejection is the result of a complex interplay of mechanisms and factors [5]. Rejection episodes are mainly driven by CD4 + Th1 cells that produce IFN-γ and TNF [45,46]. In the absence of a Th1-mediated alloimmune responses, CD4 + Th17 cells can also drive a pro-inflammatory response that accelerates chronic allograft rejection with IL-17A as a key cytokine [47]. In this study, we utilized depletion of CD4 + cells to reveal an important yet underappreciated role also for CD8 + T cells in the development of chronic organ rejection [48][49][50]. The fact that CD8 + T cells can contribute to such rejection highlights their importance as potential targets in immunotherapies aimed at provoking long-term graft acceptance.
The most straightforward interpretation of our experiments is that DNGR-1 + cDC1s infiltrate allogeneic grafts and acquire allograft antigens. Some of those cDC1 then migrate to draining lymph nodes where they cross-present the graft antigens on MHC class I and cross-prime CD8 + T cells. The primed T cells can then home to the graft and produce effector cytokines such as IFN-γ and IL-33. As a consequence, active TGF-β 1 and CTGF are induced locally and lead to allograft fibrosis. The conundrum is that, leaving aside the possibility of cross-dressing [51], the primed CD8 + T cells are restricted by host MHC and, therefore, cannot be restimulated directly by allograft cells. It is possible, therefore, that they are restimulated within the graft once again by cDC1 locally cross-presenting alloantigens, which would reinforce the observed dependence on that DC subtype and on the DNGR-1 receptor. An analogous role for cDC1 has been proposed in tumor immunity whereby cDC1 prime an antitumor CD8 + T cell response in tumor draining lymph nodes but also restimulate effector CD8 + T cells within the tumor itself [52][53][54]. The fact that CD8 + T cells would utilize the indirect presentation pathway to respond to alloantigens within grafts may also explain why their primary contribution in this setting appears to be production of cytokines leading to fibrosis rather than direct acute destruction of allograft cell targets as the latter cannot be directly recognized.
DAMP release and inflammation due to transplantation should be the same in syn-and allografts yet, in our study, cDC1 accumulated to a much greater extent in allografts than in syngeneic grafts. One point that remains unclear is why and how allogeneity is detected at these early time points to control accumulation of cDC1s. Early accumulation of alloreactive T-cells could be involved and those cells could produce chemokines such as XCL1/2 that attract cDC1 in a positive feedback cycle [55,56]. However, we saw no difference in cDC1 accumulation in allografts transplanted into mice lacking CD8 + T cells (data not shown). Alternatively, allogeneic determinants might be recognized early after transplantation not by T cells but by host NK cells that produce XCL1/2 and CCL5 to recruit cDC1 into grafts as recently demonstrated for tumors [57]. Further experimentation will be necessary to assess these possibilities.
In summary, our experiments suggest that immune activation leading to chronic allograft rejection and fibrosis can be triggered by DAMPs, including F-actin, that are detected by graft-infiltrating host cells. F-actin is recognized by DNGR-1, a receptor that is expressed on cDC1s that develop under the control of the transcription factor Batf3. DNGR-1 is a universal marker of mouse and human cDC1s in lymphoid and non-lymphoid tissues [18][19][20][21][22]24] and targeting antigens to DNGR-1 receptor has been used to induce specific immune responses in vaccination modalities [18,58,59]. In this current study, antibodies to DNGR-1 were used instead to block the receptor and shown to be effective at preventing chronic graft rejection in two different mouse heart transplantation model. These experiments suggest that antibody blockade of DNGR-1 can prevent the immune cascade leading to chronic allograft rejection and fibrosis and could therefore be exploited in humans in a therapeutic setting.

Mice
Eight to 10 weeks (w) old female BALB/c (H-2 d ) and Bm12 (H2-Ab1 bm12 ) as donors and 12-14 weeks old WT and deficient (-/-) Batf3 tm1Kmm/J (Batf3 -/-) and Clec9a tm1.1Crs/J (Clec9a -/-) recipient female mice on C57BL/6 (H-2 b ) background were purchased from Jackson Laboratory (Bar Harbor ME, USA). Mice were then bred in the animal facility of the Department of Surgery (University Medical Center Regensburg, Germany). All mice were housed under pathogen-free conditions and handled according to the local institutional guidelines. The experiments were approved by the local animal committee (TVA DMS-2532-2-102).

Heterotopic heart transplantation
Cardiac allografts from donors BALB/c and Bm12 mice were heterotopically transplanted into WT and C57BL/6 -/as previously established by Corry et al. and adapted in our laboratory [2,5,60,48]. Donor hearts were perfused through the abdominal vena cava with 3 mL of cold 0.9% saline containing 500 IE heparin (Ratiopharm, Ulm, Germany) then harvested and placed in 4°C saline until transplantation. Abdominal palpation was regularly carried out to ensure the beating of the allograft. Functioning hearts were then harvested for analyses. In the BALB/c→C57BL/6 model transient CD4 + T cell-depletion was used for generation of chronic allograft rejection [48][49][50]. This model further allows investigation of allograft rejection in absence of CD4 + T cells. The Bm12→C57BL/6 model is a fully immunocompetent model also with the development of chronic allograft rejection.

Human samples
Chronic rejected and control human cardiac graft specimens were obtained from surgical biopsies (Department of Cardiac Surgery, Bad Oeynhausen, Germany). The paraffin-embedded tissue samples were sectioned at 3-4 μm. This was approved by the local ethics committee.
Cytochrome C depletion in vivo: Cytochrome C (Cyt C) is known to be able to induce an Apaf-1-dependent apoptosis selectively in cross-presenting DCs www.eji-journal.eu and consisted of 5 mg doses of Cyt C in PBS administered intravenously (i.v.) on day 1 prior heart transplantation then subcutaneously (s.c.) on day 2 and 7 post heart transplantation. Recipient receiving only PBS solution (100 μL/mouse) were used as control groups.

Cardiac cell isolation
Cells were isolated from cardiac grafts as previously described [48]. Briefly, tissue was minced with scalpels in the presence of sterile RPMI 1640 medium containing 10% FCS, 600 U/ml collagenase II (Roche Diagnostics, Mannheim, Germany) and deoxyribonuclease I (DNAse, from bovine pancreas; Sigma, Munich, Germany). The mixture was slowly shaken at room temperature (RT) for 2 h. The supernatant (SN) was obtained and filtered through a 100-μm-nylon cell strainer. Remaining tissue was again digested in 5 mL of same solution at 37°C and then strained. Filtered material was centrifuged and red blood cells were lysed with ACK lysis buffer (BioWhittacker, Lonza, MD, USA). The pellet containing cells was passed through a 40-μm-nylon cell strainer. For the CD8 + T cell isolation, negative selection procedure was performed as recommended by manufacturer (MACS, Miltenyi Biotec: mouse CD8a + T Cell Isolation Kit). After elution of MicroBeads conjugated cells through magnetic MACS LS column, the fraction containing CD8 + T cells was recovered and washed by centrifugation.

Histology and immunohistochemistry
Formalin-fixed and paraffin-embedded (mouse and human) or frozen (mouse) cardiac graft specimens were prepared and sectioned (3-4 μm

Statistics
All groups are shown as mean ± SD of the mean and were compared using a one-tailed Student's t-test or Mann-Whitney U-test (GraphPad Prism, version 5.00). The log-rank test was used to compare graft survival between the groups. Difference among groups was considered statistically significant (*) if p was < 0.05.