Alloreactive memory CD4 T cells promote transplant rejection by engaging DCs to induce innate inflammation and CD8 T cell priming

Significance

Solid organ transplantation is the only currently available curative approach for patients with end-stage organ failure. However, recognition of donor MHC molecules (alloantigens) by host T cells can lead to transplant rejection anywhere from days to years after transplantation. Recently, inflammatory cytokines commonly associated with microbial infections have also been implicated in transplant rejection suggesting innate immune activation. Here, we illustrate a role for memory CD4 T cells specific to alloantigens in driving innate inflammation and transplant rejection. TNF superfamily ligands expressed on memory CD4 T cells engage their receptors on dendritic cells to drive both innate inflammation and alloreactive CD8 T cell priming, suggesting that blocking these molecules might be a sound strategy for prolonging graft survival.

Abstract

Alloreactive memory T cells have been implicated as central drivers of transplant rejection. Perplexingly, innate cytokines, such as IL-6, IL-1β, and IL-12, are also associated with rejection of organ transplants. However, the pathways of innate immune activation in allogeneic transplantation are unclear. While the role of microbial and cell death products has been previously described, we identified alloreactive memory CD4 T cells as the primary triggers of innate inflammation. Memory CD4 T cells engaged MHC II–mismatched dendritic cells (DCs), leading to the production of innate inflammatory cytokines. This innate inflammation was independent of several pattern recognition receptors and was primarily driven by TNF superfamily ligands expressed by alloreactive memory CD4 T cells. Blocking of CD40L and TNFα resulted in dampened inflammation, and mice genetically deficient in these molecules exhibited prolonged survival of cardiac allografts. Furthermore, myeloid cell and CD8 T cell infiltration into cardiac transplants was compromised in both CD40L- and TNFα-deficient recipients. Strikingly, we found that priming of naive alloreactive CD8 T cells was dependent on licensing of DCs by memory CD4 T cells. This study unravels the key mechanisms by which alloreactive memory CD4 T cells contribute to destructive pathology and transplant rejection.

Organ transplantation is a critical requirement for patients suffering from end-stage organ failure. Successful transplantations date back to the 1950s; however, survival of donor grafts remains challenging due to alloreactive adaptive immune cells (1). Alloreactive host T and B cells that interact with donor histocompatibility antigens can lead to both cellular and humoral immune responses and subsequent graft rejection (2, 3). Donor dendritic cells (DCs) in the allograft can either directly present donor antigens to host T cells, or host DCs will trogocytose and cross-dress to present donor MHC–antigen complexes to recipient T cells (4, 5). Despite substantial progress in human leukocyte antigen (HLA) typing, immunosuppressant drugs, anti-T cell antibody infusion, and leukocyte depletion in donor organs, a large proportion of transplant recipients eventually experience immunological graft rejection due to persistence of alloreactive effector and memory T cells (6–9). In addition to T cell persistence, these preventative regimens make transplant patients susceptible to a variety of infections which can be life threatening (10, 11).

Clinical evaluation of transplant recipients at different stages of rejection has uncovered substantial evidence for innate cytokines as a critical inflammatory component in graft rejection (12–15). Innate inflammatory cytokines such as IL-1β, IL-6, IL-12, and TNFα have all now been characterized as acute phase mediators of transplant rejection in allograft recipients (16, 17). Production of innate inflammatory cytokines is usually an outcome of exposure to microbial signals such as pathogen-associated molecular patterns (PAMPs), which induce transcriptional activation of these cytokines in cells of the innate immune system (18). Production of proinflammatory cytokines in the initial stages of allograft implantation has largely been attributed to the release of damage-associated molecular patterns (DAMPs) potentially contributed by ischemia reperfusion injury (19–21). However, attempts to build meaningful therapies targeting DAMPs have failed so far. Moreover, comparable infiltration of monocyte-derived DCs (mono-DCs) into heart allografts of WT, Myd88^−/− and Nalp3^−/− mice underscored that innate allorecognition might involve mechanisms apart from recognition of DAMPs (22). Further complicating the understanding of innate inflammation in transplant rejection is the release of proinflammatory cytokines at later stages (1 to 6 mo posttransplantation) when the occurrence of DAMPs is scarcely reported (12, 13).

The innate immune system forms a critical component in the alloresponse and subsequent transplant rejection (22–26). How innate responses are induced during allorecognition, where eukaryotic determinants from genetically dissimilar donors are recognized and presented as antigens to host T cells, remains unclear. DCs accompanying the graft can initiate graft-specific T cell responses by homing to the host’s secondary lymphoid organs (27, 28). Nonetheless, depletion of donor DCs from the transplant before engraftment failed to preserve graft survival in the host (29). Elegant studies from multiple groups have now shown that host DCs rapidly replace their donor counterparts and start presenting intact donor MHC–peptide complexes by a phenomenon termed cross-dressing (5, 30, 31). These host DCs can subsequently engage in stable, cognate interactions with effector T cells present in the graft (30, 32). However, the participating DC subsets, the interacting pathways, as well as the outcome of these interactions remain underexplored.

Recently, the concept of innate immune memory to previously encountered allogeneic MHC-I molecules has also been investigated. Recognition of donor MHC-I molecules by paired immunoglobulin-like receptor A (PIR-A) molecules on host Ly6C^hi monocytes and macrophages was reported to lead to an alloresponse (33). While the infiltration and activation of donor monocytes and monocyte-derived DCs (mono-DCs) and their activation were observed, the presence of innate cytokines in graft recipients could not be attributed to this phenomenon (33). The adverse effects of innate immune activation to graft tolerance were also pointed out in studies involving heart allografts into Rag2^−/− γc^−/−mice that lacked T and B cells. Significant monocyte infiltration into these grafts and their differentiation into IL-12p40⁺ DCs led to their eventual decline (22). Even though Mhc locus independent pathways were found to be involved, the activation of NFκB and secretion of innate cytokines in this case could also not be explained (34, 35).

Multivariate analysis of organ transplants during acute cellular rejection indicates that a major barrier to successful transplant survival is the presence of alloreactive memory T cells residing in peripheral nonlymphoid organs (36–39). It is well documented that in mice and humans, about 1 to 10% of the memory T cell repertoire has allogeneic specificity (27–29). This small proportion of recipient T cells have the potential to acquire an alloreactive tissue-resident memory (T_RM) phenotype postengraftment (40–42). A plethora of factors can lead to the generation of a T_RM cell subset that has the potential to interact with alloantigen presented on host or recipient DCs (42, 43). These alloreactive memory T cells also have a much lower activation threshold compared to their naive counterparts and can become resistant to glucocorticoid-mediated immunosuppression and costimulatory blockade (44). Nevertheless, how T cells in their effector or effector memory states can engage in a cross talk with DCs has not been completely explored. Previous studies from our group have shown that antigen-specific as well as self-reactive memory CD4 T cells can engage DCs via cognate TCR–MHC interactions, leading to inflammasome-independent secretion of IL-1β and production of IL-6 and IL-12 (45, 46). Here, we investigated whether alloreactive memory CD4 T cell interaction with DCs of an allogeneic origin led to innate inflammation and whether such sterile activation of the innate immune system by T cells was responsible for allograft rejection.

Results

Naturally Arising Effector Memory CD4 T Cells Can Induce Proinflammatory Cytokine Production by DCs Presenting Mismatched MHC Molecules.

The frequency of alloreactivity of polyclonal CD4 T cells, both naive and memory, is estimated to be around 1 to 10% (37). In contrast to naive T cells, memory T cells, regardless of their origin, display lower activation threshold and can be quickly reactivated to perform effector functions upon encountering cognate antigen–MHCII complexes (47). Additionally, recent work from our lab established that antigen-specific memory CD4 T cells can activate DCs presenting cognate peptides to drive innate inflammation (45). Since donor-specific memory T cells are associated with rejection of allografts, we were interested in interrogating whether interactions between polyclonal memory T cells and DCs presenting mismatched MHC molecules would result in activation of DCs. We, therefore, examined whether endogenously arising T_EM cells (CD44^hi CD62L^lo) interact with DCs of allogeneic origin to drive innate inflammatory cytokine production. To test this, we cultured CD4 T_EM cells from the spleens and lymph nodes of naive BALB/c mice (i.e., allogenic) or B6 mice (i.e., syngeneic) (Fig. 1A and SI Appendix, Fig. S1A) at a 20:1 ratio of T cells with DCs (CD11c⁺) enriched from the spleens of C57Bl/6J (B6) mice (SI Appendix, Fig. S1B). Of note, we focused on isolating CD11c⁺ cells which could be a mix of classical DCs, monocytes, and monocyte-derived DCs. For the sake of simplicity, we will from here on refer to them collectively as DCs. We detected both IL-1β and IL-6 in the supernatants of cultures only when MHC-mismatched DCs were cultured with CD4 T_EM cells of allogeneic origin (Fig. 1B). This suggested that naturally arising T_EM cells of unknown specificity can interact with DCs presenting mismatched MHC–peptide complexes to drive innate inflammatory cytokine production. Activated CD4 T cells or T_EM cells are present at extremely low frequencies in unimmunized mice, especially when housed under specific pathogen-free conditions. To gain further mechanistic insights into the interaction between alloreactive CD4 T cells and DCs, we activated CD4 T cells in vitro to generate alloreactive effector memory T cells. Briefly, naive CD4 T cells (CD62L^hi CD44^lo) isolated from the spleen and lymph nodes of BALB/c mice were cultured with B6 DCs at a high frequency of interaction (1T cell :2 DCs ratio) to prime naive alloreactive CD4 T cells. After 5 d of culture, total CD4 T cells were isolated and rested in the presence of IL-2 to generate CD4 T_EM (Fig. 1C). These alloreactive T_EM cells generated in vitro (hereafter referred to as allo T_EM) from BALB/c mice were cultured with freshly isolated DCs from B6 mice at different ratios (SI Appendix, Fig. S1C). Since a ratio of 1:1 for MHC-mismatched DC interacting with allo T_EM was found to generate the most optimal amounts of secreted IL-6 in the supernatant (SI Appendix, Fig. S1C), we used this ratio of DC to T cells for all our future experiments (Fig. 1C). Interestingly, further analysis of CD11c⁺ cells revealed that pro-IL-1β induction was specifically elevated in the CD11c⁺CD11b⁺ population and not in the CD11c⁺CD11b⁻ population (SI Appendix, Fig. S1 B and D), suggesting that allo T_EM cells are primarily interacting with CD11c⁺CD11b⁺ cells. Moreover, a 1:1 ratio of DCs to allo T_EM cells was seen to induce the most robust intracellular pro-IL-1β (SI Appendix, Fig. S1E). For this reason, we specifically analyzed pro-IL-1β on the CD11c⁺CD11b⁺ DC population for all future experiments in this study.

Fig. 1.

Interaction of CD4⁺ T_EM cells with DCs expressing mismatched MHC molecules leads to IL-1b and IL-6 production. (A) Schematic representation of the DC: T cell coculture system. Briefly, CD62L⁻CD44⁺CD4⁺ T (T_EM) cells were magnetically sorted from the spleen and lymph nodes of aged (25 to 35 wk old) BALB/c mice and cocultured at a 20:1 ratio with splenic DCs from B6 mice. Lineage-specific antibodies for T, B, NK cells, erythrocytes, yolk-sac-derived macrophages, and neutrophils (i.e., CD90.2, CD19, NK1.1, TER119, F4/80, Ly6G) were added for depletion to enrich DCs from spleens of B6 mice. Syngeneic T_EM cells were sorted from the spleen and lymph nodes of aged B6 mice. (B) IL-1β and IL-6 secreted into the supernatants were measured by ELISA at 24 h. (C) Schematic representation of the experimental setup for in vitro generated BALB/c allo T_EM and B6 splenic DCs. (D) ELISA for IL-1β and IL-6 secreted into the supernatants from coculture described in (C) as measured at 24 h. (E) Pro-IL-1β as measured by intracellular staining on the live CD11c⁺CD11b⁺ CD90.2⁻ population after 6 h of culture with allo T_EM cells. (F) IL-1β and IL-6 as detected and quantified by ELISA at 24 h from cultures of B6 DCs and BALB/c allo T_EM cells. Cultures were treated with an MHC-II blocking antibody where indicated. Error bars indicate mean ± SEM. n = 3 biological replicates for (B) and n = 3 to 4 biological replicates for (D–F). Data were generated from three independent experiments for (B) and (D–F). P values were determined by ordinary two-way ANOVA (*P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001; ns, not significant).

In agreement with our findings from experiments using naturally arising endogenous CD4 T_EM, interactions between DCs and in vitro–generated allo T_EM cells led to production of large quantities of IL-1β and IL-6 (Fig. 1D). Additionally, pro-IL-1β production was induced in CD11b⁺CD11c⁺ DCs cultured with allo T_EM cells, but not with T_EM cells of syngeneic origin (Fig. 1E).

DCs are known to present allogeneic antigens to T cells via a direct, indirect, or semidirect pathway (4, 48). To further establish that the inflammatory cytokine production was dependent on interaction between MHC Class-II and alloreactive TCRs, we used a blocking antibody against MHC Class-II and found that this led to significant reduction of both IL-1β and IL-6 in the supernatants (Fig. 1F). These data firmly establish that T_EM cells expressing alloreactive TCRs interact with allogeneic MHC Class II molecules on DCs resulting in DC activation and secretion of innate cytokines.

Proinflammatory Cytokine Production by MHC-mismatched DCs Is Independent of Engagement of Major Families of PRRs.

DCs are immune sentinels and are equipped with the ability to recognize microbial signatures through various pattern recognition receptors (PRRs). Engagement of PRRs, such as toll-like receptors (TLRs), can lead to downstream activation of the NFκB and MAP kinase pathways, resulting in increased expression of proinflammatory cytokines. To test whether the induction of proinflammatory cytokines in our assay was a result of TLR engagement either due to generation of endogenous TLR ligands or inadvertent contamination by microbial products, BALB/c allo T_EM cells were cultured with splenic DCs from B6 mice lacking the expression of TLR2 and TLR4 as well as endosomal TLRs (TLR2/4xUnc93b1^3d/3d) (49). TLR-deficient B6 DCs secreted near identical levels of IL-6 and IL-1β in the culture supernatants when compared to WT B6 DCs cultured with allo T_EM cells (Fig. 2A), suggesting that TLR activation was not involved in the production of innate inflammatory cytokines. Induction of pro-IL-1β was also not affected by TLR deficiency on MHC-mismatched DCs (Fig. 2B). The absence of functional TLRs was confirmed by lack of IL-6 and IL-1β secretion upon LPS stimulation of these DCs (SI Appendix, Fig. S2).

Fig. 2.

Innate inflammatory cytokine production by MHC-mismatched DCs cocultured with allo T_EM cells is independent of PRR engagement. (A) ELISA for IL-1β and IL-6 secreted into the supernatants of B6 WT and Tlr2/4xUnc93b1^3d/3dDCs cultured for 24 h in the presence or absence of allo T_EM derived from BALB/c mice. (B) Pro-IL-1β expressing cells as measured by flow cytometry at 6 h postculture as mentioned in (A). (C) ELISA for IL-1β from supernatants of WT or Casp1^Δ10DCs cultured for 24 h in the presence or absence of allo T_EM cells derived from BALB/c mice. (D) IL-1β ELISA from supernatants of B6 DC and BALB/c allo T_EM cultures. Cultures were treated with the indicated caspase inhibitors as denoted. Error bars indicate mean ± SEM. n = 3 biological replicates for (A–D). Data were generated from three independent experiments. P values were determined by the unpaired t test (A–C) or ordinary two-way ANOVA (D) (*P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001; ns, not significant).

While engagement of PRRs leads to synthesis of pro-IL-1β (50), its processing into a bioactive form requires cleavage by caspase 1 (50, 51). To further examine the caspase responsible for cleaving pro-IL-1β in allogeneic DCs, allo T_EM cells were cultured with MHC-mismatched DCs from caspase 1–deficient (caspase 1 delta 10; Casp1^Δ10) mice (52). Caspase 1–deficient B6 DCs secreted similar levels of IL-1β as WT B6 DCs when cultured with allo T_EM cells (Fig. 2C). Apart from caspase 1, caspase 8 is an effector caspase known to facilitate the processing of pro-IL-1β into its bioactive form (51, 53). We used Z-IETD-FMK (IETD), a specific inhibitor for caspase 8 activity, to examine whether IL-1β secretion was a result of caspase 8 activation. Indeed, IL-1β secretion was found to be dependent on caspase 8 activation demonstrated by the reduction of bioactive IL-1β levels in the presence of IETD (Fig. 2D). Consistent with our observations with the caspase 1–deficient DCs, the caspase 1 inhibitor Z-YVAD-FMK (YVAD) did not affect IL-1β secretion (Fig. 2D), validating that IL-1β secretion from DCs interacting with alloreactive CD4 T_EM cells was independent of caspase 1 activation and possibly inflammasome activation. Taken together, our observations suggest that T_EM-driven inflammatory cytokine production by allogeneic DCs was independent of certain classical PRRs.

TNF Superfamily Member Engagement between DCs and Allo T_EM Cells Facilitates Proinflammatory Cytokine Production by DCs.

Next, we wanted to gain mechanistic insights into how interaction with alloreactive T_EM cells modulated the transcriptional profile of DCs. We therefore cultured B6 DCs with B6 (syngeneic) or BALB/c (allogeneic) T_EM cells for 3 h before sort-purifying the B6 DCs and examining their transcriptomic profile via bulk mRNA-sequencing (RNA-seq). Principal component analysis (PCA) based on gene expression values across unstimulated DCs (B6 DC), DCs sorted from coculture with syngeneic T_EM (B6 DC+B6 T_EM), and those from allogeneic T_EM (B6 DC+BALB/c T_EM) coculture showed clearly separated clustering, indicating distinct transcriptional profiles (Fig. 3A). Differential expression analysis revealed changes in 580 genes in DCs previously cultured with allo T_EM cells as compared to unstimulated DCs, of which 428 were up-regulated and 152 were down-regulated (Fig. 3B). Il1b, Il6, and Il12b were up-regulated in the B6 DC+BALB/c T_EM group, which was further validated by qPCR analysis (Fig. 3 C and D). DC activation markers Cd80, Cd83, Nfil3, and Irf4, which are critical for maintenance of a conventional DC2 (cDC2) signature, were also up-regulated specifically in DCs that were sorted from cultures with allo T_EM (Fig. 3B). Interestingly, Tlr9, Clec9a, and Id2, which are predominantly expressed by conventional DC1 (cDC1s), and Spib and Ifnar1, which are more characteristic of the plasmacytoid DC (pDC) lineage, were found to be down-regulated on DCs upon interacting with allo T_EM. This was indicative of subset-specific interactions between DCs and allo T_EM. Prominent among the genes specifically up-regulated in B6 DCs cocultured with allo T_EM cells were TNFSF-related genes, along with TGF-β signaling, IL-6 signaling, and other inflammatory response pathways (Fig. 3E and SI Appendix, Table S1).

Fig. 3.

Alloreactive CD4⁺ T_EM cells use TNF and CD40L to drive innate cytokine production by DCs of MHC-mismatched origin. (A) PCA of transcriptomic profiles of B6 DCs sorted from cultures with or without allogeneic (BALB/c) or syngeneic (B6) T_EM as indicated. (B) Heatmap showing scaled expression (row-wise z-score) of differentially expressed genes (FDR < 0.05) in B6 DCs cocultured with allo T_EM relative to unstimulated DCs. Each cluster of columns shows gene expression in the different conditions as indicated on the top. Genes associated with proinflammatory signatures and DC subtypes are highlighted. (C) FPKM values from bulk RNA-seq analysis for Il1b and Il6 expression in DCs sorted post-coculture with TEM cells as indicated. (D) q-PCR analysis of Il1b and Il6 expression in DCs sorted post-coculture with TEM cells as indicated. (E) Overrepresentation of hallmark pathways among differentially expressed genes (FDR < 0.05) in DCs sorted from cultures with syngeneic or allo T_EM cells relative to unstimulated DCs. (F) ELISA measurements of IL-1β and IL-6 in supernatants from 24-h 1:1 coculture of in vitro generated BALB/c TEM cells with B6 DCs in the presence of indicated blocking antibodies. (G) ELISA measurements of IL-1β and IL-6 in supernatants from 24-h 1:1 coculture of in vitro generated BALB/c TEM cells with WT, CD40, or TNFSF1/2 receptor–deficient B6 DCs. Error bars indicate mean ± SEM. n = 3 biological replicates for (A–G). Data were generated from three independent experiments P values were determined by ordinary two-way ANOVA (*P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001; ns, not significant).

As a result, we directly analyzed whether TNFSF members were mediating interactions between MHC-mismatched DCs and allo T_EM cells through the use of neutralizing antibodies against TNFα, CD40 ligand (CD40L, Tnfsf5) and Fas ligand (FasL, Tnfsf6). Blockade of TNF–TNFR interaction led to an observable decrease in bioactive IL-1β and IL-6 (Fig. 3F). The reduction in secreted IL-1β may be the direct result of reduced induction of pro-IL-1β itself (SI Appendix, Fig. S3 A and B). On the other hand, CD40–CD40L interaction between DCs and allo T_EM cells was also required for both IL-1β and IL-6 secretion by DCs correlating to reduction in pro-IL-1β (SI Appendix, Fig. S3 A and B). Surprisingly, FasL–Fas interaction had no contribution to production of lL-1β by DCs (Fig. 3F). Our observations with in vitro neutralization of TNFSF members were further corroborated by lower IL-1β and IL-6 secretion from MHC-mismatched DCs from TNF receptor 1/2–deficient and CD40-deficient B6 mice cultured with allo T_EM cells (Fig. 3G), further demonstrating that TNF receptor and CD40 pathways in DCs were critical for production of innate inflammatory cytokines in response to allogeneic memory CD4 T cell interactions.

TNFα and CD40 Are Critical for Allograft Rejection in a Model of Heterotopic Heart Transplantation.

Thus far, our findings suggested that interaction between allo T_EM cells and DCs leads to profound innate inflammation. While TLRs and caspase-1 were dispensable, we posited that TNFSF-induced innate inflammation by allo T_EM cells is likely responsible for graft rejection in solid organ transplantation. To test this hypothesis, we used a model of heterotopic heart transplantation, in which B6 WT (H-2^b) or Tnf ^−/−-or Cd40l^−/− (both H-2^b) recipient mice received a CB6 (BALB/c x B6) F1 (H-2^b/d) heterotopic heart graft using the cervical engraftment technique (54). The use of B6 recipient mice lacking either TNFα or CD40L allowed us to assess whether expression of these TNFSF members on host T cells was required to initiate allograft rejection. Graft survival was monitored by daily palpitation of the implanted heart, with end-stage rejection confirmed by echocardiography. B6 WT recipients rejected the semiallogeneic heart graft within a week posttransplantation (Fig. 4A). However, both Cd40l^−/− and Tnf^−/− recipient mice failed to reject the heart allograft, with strong long-term palpable heterotopic heart function persisting until day 66 when the experiment was terminated (Fig. 4A). Graft rejection is typically associated with significant host immune cell infiltration and tissue remodeling. Interestingly, when we analyzed the graft infiltrating lymphocyte population 5 d posttransplant, before rejection of the graft by WT mice (SI Appendix, Fig. S4A), a significantly higher lymphocytic infiltrate was found in the engrafted hearts in the WT recipients when compared to the engrafted hearts from either Tnf^−/− and Cd40l^−/− recipients (Fig. 4B and SI Appendix, Fig. S4B). Specifically, there was a significantly higher infiltration of T cells in the graft tissue of WT recipients when compared to either the Cd40l^−/− and Tnf^−/− recipients (Fig. 4C and SI Appendix, Fig. S4C). Interestingly, we found that while both CD4 and CD8 T cells were seen to populate the implanted graft in WT recipient mice at day 5 postimplant (Fig. 4D), there was a much higher proportion of CD8 T cells on day 7 (SI Appendix, Fig. S5). Heart grafts in WT recipients showed increased infiltration of myeloid cells, characterized as macrophages (CD11c⁻CD11b⁺), as well as myeloid (CD11c⁺CD11b⁺) and lymphoid DCs (CD11c⁺CD11b⁻) (Fig. 4E and SI Appendix, Fig. S4D), all of which were markedly reduced in both Cd40l^−/− and Tnf^−/− recipients. Histological analysis revealed increased foci of infiltration and increased pathology scores in grafts from WT recipients when compared to the Cd40l^−/− and Tnf^−/− recipients (Fig. 4 F and G). Additionally, serum from recipients of heart transplants demonstrated elevated presence of IL-6 and IL-12p70 in WT recipients when compared to Cd40l^−/− and Tnf^−/− recipients (Fig. 4H). These data validate our in vitro findings that implicate CD40L and TNFα in driving allo T_EM-induced innate inflammatory cytokines by DCs. Interestingly, heterotopic hearts transplanted into Tlr2/4^−/−recipient mice were rejected by day 10, at a rate slightly delayed but at very similar kinetics to WT recipients (Fig. 4I) further corroborating our in vitro data that TLRs have a limited role in driving innate inflammation associated within transplant rejection.

Fig. 4.

Allogeneic heart transplants survive longer in TNFa and CD40L KO recipient mice and exhibit reduced immune cell infiltration as compared to WT recipients. (A) Survival of the transplanted CB6F1 hearts into B6 WT vs Tnf^−/−and Cd40l^−/−recipients was determined by measuring heart palpation daily. (B–E) Infiltration of leukocytes (represented as cell numbers) as quantitated by flow cytometry into the transplanted F1 heart at day 5 postimplant. (F) Representative histology images showing immune infiltration (marked by white arrows) into CB6F1 hearts at day 5 postimplant into the indicated recipient mice. (G) Histology scores determined using the International Society for Heart Transplantation ACR scoring system by comparison of histological sections upon H&E staining of transplanted hearts at day 5 posttransplant. (H) IL-6 and IL-12p70 measured by Luminex Multiplex bead assay from serums collected from heart transplant recipients on Day 5. (I) Survival of the transplanted CB6F1 hearts into B6 WT vs Tlr2/4^−/−mice was determined by measuring heart palpation daily. Error bars indicate mean ± SEM. For (A), n = 8 WT, 4 Tnf^−/−, and 6 Cd40l^−/− mice; for (B–H), n = 4 B6 WT, 4 Tnf^−/−, 4 Cd40l^−/−, and 2 syngeneic B6 mice; for (I), n = 3 Tlr2/4^−/− and 4 WT mice. For A and I, P values were determined by the Gehan–Breslow–Wilcoxon test. For B–E, G, and H, P values were determined by ordinary two-way ANOVA (*P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001; ns, not significant).

CD4 T_EM Cells Engage DCs through CD40L–CD40 Interaction to Prime Alloreactive CD8 T Cells.

Analysis of the lymphocyte infiltration into the heart allografts of B6 WT recipients revealed only modest CD4 or CD8 T cell infiltration at day 3 posttransplant, but this changed on day 5 and day 7 posttransplantation (Fig. 5 A and B and SI Appendix, Fig. S5). Interestingly, on day 5 the T cell infiltrate was a mixed population of both CD4 and CD8 T cells, but by day 7, when most WT recipients had rejected the allograft, the infiltrate was dominated by CD8 T cells (Fig. 5B and SI Appendix, Fig. S5). One potential explanation is that in the early stages of infiltration, CD4 T cells are critical for priming of DC populations to activate CD8 T cells that contribute to transplant rejection at later time points. To test this, we utilized our in vitro system of allo T_EM cell–driven activation of allo-MHC presenting DCs (Fig. 5C). We specifically asked whether DCs interacting with allo-CD4 T_EM cells would be better equipped to prime naive alloreactive CD8 T cells.

Fig. 5.

Allo CD4⁺ T_EM cells can license DCs to prime alloreactive naive CD8⁺ T cells. (A) Kinetics of CD4⁺ T cell infiltrating into the transplanted CB6F1 heart in WT B6 mice at indicated days posttransplant. (B) Kinetics of CD8⁺ T cell infiltrating into the CB6F1 transplanted heart in WT B6 mice at indicated days posttransplant (depicted as proportions of total CD45⁺ CD90.2⁺ cells and absolute numbers). (C) Schematic representation of the DC: allo CD4⁺ T_EM cell culture setup used for CD8⁺ T cell priming. (D) Flow cytometry analysis showing live CD90.2⁺CD8⁺ cells on day 5 postculture to examine priming of BALB/c naive CD8⁺ T cells cultured with B6 DCs or with B6 DCs + BALB/c CD4⁺ T_EM cells. (E) Flow cytometry plots at day 5 showing BALB/c CD8⁺ T cell priming when cultured with B6 WT or Cd40^−/− DCs and BALB/c CD4⁺ T_EM cells. (F) Intracellular staining and analysis of IFNÉ£ in BALB/c CD8⁺ T cells at day 5 of culture with B6 WT or Cd40^−/− DCs as indicated in the presence of BALB/c CD4⁺ T_EM. (G) Intracellular staining and analysis of Granzyme B in BALB/c CD8⁺ T cells at day 5 of culture with B6 WT or Cd40^−/− DCs as indicated in the presence of BALB/c CD4⁺ T_EM. (H) Quantitation of cell numbers for indicated populations from flow cytometry of BALB/c CD8⁺ T cells primed by B6 WT or Cd40^−/− DCs at day 5. Error bars indicate mean ± SEM. n = 2 for Day 3 and n = 3 each for Day 5 and Day 7 analysis for A and B and data are representative of four biological replicates for D–G and n = 4 biological replicates for (H). P values were determined by ordinary two-way ANOVA (*P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001; ns, not significant).

We found that priming of naive alloreactive CD8 T cells by DCs was completely dependent on the presence of alloreactive CD4 T_EM cells (Fig. 5D). This is very surprising and indicated that alloreactive memory CD4 T cells are indeed critical for licensing DCs to prime alloreactive CD8 T cells. While the identity of these DCs was not completely assessed, the populations we used were CD11c⁺CD11b⁺ and were largely devoid of cDC1s. However, it is hard to ascertain that memory CD4 T cells are indeed licensing cDC2s to prime CD8 T cells. Future studies in cDC1-deficient Irf8+32^−/− mice will be needed to specifically identify the DC subset being licensed by allo T_EM. Nevertheless, these data have uncovered an entirely unknown role for DCs in priming of CD8 T cells by virtue of alloreactive memory CD4 T cell help. Recent studies on priming of antitumor CD8 cytotoxic T lymphocytes (CTLs) have demonstrated the engagement of CD40L–CD40 between CD4 T cells and DCs as critical for cDC1 licensing (49, 50). To further elucidate whether CD40L–CD40 interactions also play a role in activating DCs to further prime CD8 T cell responses, we used the three-cell coculture system described above to compare the ability of WT and Cd40^−/−DCs to prime naive alloreactive CD8 T cells (Fig. 5E). We found that Cd40^−/−DCs were completely compromised in their ability to prime naive alloreactive CD8 T cells (Fig. 5 E and H). CD8 T cells primed by licensed WT DCs were highly competent in production of IFNÉ£ and Granzyme B and both were reduced to negligible levels in CD8 T cells primed by Cd40^−/− DCs (Fig. 5 F–H). This underlines a critical role for CD40L–CD40 engagement between alloreactive CD4 T_EM and DCs, respectively, to not only drive innate inflammation but also prime allogenic CD8 T cells. It is likely that the two CD40-dependent events, namely inflammatory cytokine production and alloreactive CD8 T cell priming, collectively drive rejection of the transplanted organ.

Discussion

The findings described in this study shed light into the cellular and molecular mechanisms by which memory CD4 T cells drive acute transplant rejection. In a previous study, graft infiltrating DCs have been shown to make stable engagement with effector T cells in the graft. Here, we demonstrate that interaction of alloreactive CD4 T_EM cells with MHC-mismatched DCs leads to both profound innate inflammation as well as de novo priming of naive alloreactive CD8 T cells. While MHC–TCR interactions between DCs and alloreactive T cells are obligate and fundamental, we present a specific role for TNFα and CD40L expressed by memory CD4 T cells in inducing DC activation and innate inflammation. Transcriptional profiling of MHC-mismatched DCs exposed to allo T_EM cells displayed a pronounced inflammatory signature, providing insights into the ability of T_EM cells to influence the transcriptome of MHC-mismatched DCs during allorecognition. We further demonstrate that this alloreactive CD4 T_EM cell–induced activation of the DCs through CD40L–CD40 signaling is critical for priming of naive alloreactive CD8 T cells. Our study thus highlights the importance of TNFα and CD40L in mediating T cell–driven innate inflammation and graft rejection.

A critical barrier to the success of organ and hematopoietic transplants is presented by the extraordinary magnitude of alloresponse by T cells. The unusual strength of the alloresponse is dictated by the high frequency of polymorphic MHC alleles (55). CD4 T cells have now been demonstrated to be initiators of an allogeneic T cell response in Graft versus Host Disease (GVHD) and graft rejection has been correlated to the frequency of donor-specific IFNÉ£ producing T cells (56, 57). It is now also understood that graft tolerance is significantly hindered by the presence of memory T cells which also are refractory to regulation by regulatory T cells (58). To overcome the alloresponse, histocompatibility typing and immunosuppressive regimens like corticosteroids and calcineurin inhibitors are used in transplant recipients (10). Despite these advances, a large population of patients exhibit eventual decline of transplants postremoval of immunosuppression. Additionally, immunosuppression also makes patients more prone to infections and malignancies.

The innate immune system forms a critical component in the priming of alloreactive T cells (22–25, 59). Host DCs can cross-dress and present intact donor MHC–peptide complexes and by the semidirect pathway activate alloreactive T cells (5, 27). Many cases of rejection are also attributed to the elevated levels of innate cytokines, specifically IL-1β, IL-6, and TNFα in the serum of transplant recipients (12, 16, 17). However, the nature of the upstream signals that result in the activation of DCs leading to a proinflammatory phenotype and the role of DC:T_EM interactions has not been explored.

We previously described a pathway for DC activation by instructions received from antigen-specific as well as self-reactive T_EM cells (45, 46). These required cognate MHC–TCR interactions in addition to dependence on TNFSF ligation, leading to NFκB activation and cytokine secretion by DCs (45, 46). In the current study, we find that this pathway of innate inflammation appears to be broadly conserved, irrespective of the nature of T cells interacting with antigen-presenting DCs. Specifically, alloreactive memory CD4 T cells appear to largely use TNFSF ligands to engage mismatched MHC–peptide presenting DCs. We demonstrated that TNF–TNFR and CD40L–CD40 engagement between alloreactive T_EM cells and DCs are required to drive IL-1β and IL-6 release from MHC-mismatched DCs using an antibody-based blockade and a genetic deficiency in TNFSF receptors. These data are further strengthened by experiments done to investigate the survival of heterotopic heart transplants. Increased viability of transplants in TNFα– and CD40L-deficient mice confirm the crucial role played by these two pathways in driving transplant rejection.

A previous study showing the infiltration of IL-12-expressing mono-DCs and macrophages into the heart allograft in Rag2^−/− γc^−/−mice has indicated that infiltration of myeloid cells is a contributing factor to graft dysfunction and eventual rejection (22). However, the mechanism of activation of NFκB and secretion of innate cytokines cannot be explained by these studies. It has also been proposed that pretransplantation preservation of the graft can cause ischemia/hypoxia leading to cell death and release of DAMPs. Reperfusion of the graft may lead to release of DAMPs and subsequent recognition by PRRs leading to innate activation (60, 61). DAMPs such as HMGB1, hyaluronan, s100 proteins, and uric acid have been suggested to be ligands for TLR2 and TLR4 thus leading to inflammation and acute transplant rejection (21, 62–64). While it is possible that there is a role for TLR2 or TLR4 signaling in recognition of such moieties, both our in vitro and in vivo data suggest a minor or nonexistent role for the TLR signaling pathway in allo-recognition mediated inflammatory responses as well as transplant rejection. Specifically, we found that the secretion of innate cytokines by MHC-mismatched DCs was completely independent of TLR engagement following interaction in vitro generated alloreactive T_EM cells. Moreover, allogeneic hearts implanted into Tlr2/4^−/− recipient mice were rejected within 10 d of implant, a kinetics similar to that observed in WT recipients.

In addition to finding a critical role for TNFα and CD40L in driving production of inflammatory cytokines by DCs, we also found that the cleavage of pro-IL-1β was independent of caspase 1, thus ruling out inflammasome activation following release of ATP or other damage associate molecular patterns (65, 66). We instead found that the IL-1β production was largely dependent on caspase 8 function. While our previous work with antigen-specific CD4 T cells implicated Fas-caspase 8 signaling pathway in cleavage and release of bioactive IL-1β (45), we surprisingly found that alloreactive T_EM cell–induced production of IL-1β was completely independent of Fas ligation, suggesting involvement of a different upstream receptor capable of activating caspase 8. Future studies focusing on the role of different death receptors engaging between alloreactive T_EM and MHC-mismatched DCs are likely to provide deeper insights into the role of death receptor pathways that result in IL-1β release during graft intolerance.

Rejection of kidney transplants and skin grafts in humans has been attributed to the presence of CD8 CTLs (31, 67). In a model of heterotopic heart transplant into WT recipient mice, we found that the proportion of CD4 T cells declined while CD8 T cell proportions increased by day 7 postimplantation in the transplanted heart. It is well established that priming of CD8 T cells against pathogens and tumor antigens is dependent on the licensing of cDC1s by CD4 T cells (68, 69). CD40L–CD40 engagement between CD4 T cells and cDC1s, respectively, has been shown to be critical for licensing cDC1s, inducing upregulation of the CD70/CD27 signaling axis and subsequent priming of CD8 T cells (68, 69). Licensing of alloreactive macrophages in GVHD has been shown to be critically dependent on CD40/CD40L interactions between host macrophages and CD4 T cells (70). Surprisingly, we found that alloreactive CD8 T cells were being primed by DCs licensed in the presence of alloreactive CD4 T_EM cells. Blocking the CD40–CD40L pathway in an in vitro culture system containing alloreactive CD4 T_EM cells, DCs, and naive CD8 T cells, resulted in significantly reduced CD8 T cell priming. Licensing of DCs by alloreactive CD4 T_EM cells reveals a unique function for memory CD4 T cells as well as implicating DCs in priming of naive alloreactive CD8 T cells. Previously, lung inflammatory DCs have been implicated in priming antiviral CD4 and CD8 T cell responses (71). While cDC1s are critical for priming of CD8 T cells against pathogen-derived antigens and tumor antigens, the current study highlights the potent role of CD11c⁺ CD11b⁺ DCs in both initiating alloreactive T_EM cell–induced innate inflammation as well as de novo priming of alloreactive CD8 T cells that might ultimately be responsible for transplant rejection. It is important to note here that we have not fully assessed the identity of these DCs and they could be monocytes or other myeloid cells that express CD11c. Further studies would be required to firmly establish the identity of the DC subsets involved in priming alloreactive CD8 T cells. Blocking CD40 and CD40L has been previously demonstrated to improve prognosis of grafts (72, 73). Antagonistic antibody to CD40L had even reached stage II in human clinical trials for renal and liver transplants (73). However, thromboembolic events due to the presence of CD40L on platelets have led to failure in approval of this antibody as a potential therapy (73). Conceivably, a more targeted approach would be to block this pathway in pretransplant donor organs to subside systemic thromboembolic events in transplant patients.

Overall, our study highlights a detrimental role for alloreactive memory CD4 T cells to the survival of transplants. In addition to initiating the innate cytokine storm seen in transplant patients through the TNFR and CD40 signaling pathway, the preexisting alloreactive memory CD4 T cells also appear to license a DC population to prime CD8 T cells against alloantigens. These observations strengthen the existing paradigm of memory/antigen-experienced T cells posing as a double-edged sword where on one hand they are crucial to protect the host against pathogenic insults but in cases such as organ transplants, can lead to detrimental outcomes.

Methods

Mice.

C57BL/6 (Jax:00064) and BALB/c (Jax:000651), B6.129P2-Cd40^tm1Kik/J; Cd40^−/− (Jax:002928) and B6.129S-Tnfrsf1b^tm1Imx Tnfrsf1a^tm1Imx/J; Tnf^−/− (JAX: 003243) mice were purchased from the Jackson Laboratory. TLR2/4xUnc93b1^3d/3d mice were a gift from Dr. Gregory Barton. Casp1^Δ10 mice were gifts from Drs. Russell Vance and Isabella Rauch (University of California, Berkeley, CA) and were maintained on Jackson Laboratory C57BL/6 background. All mice were bred and maintained at Cincinnati Children’s Hospital Medical Center under specific pathogen–free conditions in accordance with protocols approved by Institutional Animal Care and Use Committee (IACUC). Age- and sex-matched mice between 6 and 12 wk were used for all experiments. For enrichment of endogenous T_EM cells from B6 mice, mice aged 25 to 35 wk old were used. Both males and females were used for all experiments. CB6F1 (JAX: 100007) were purchased from Jackson Laboratory at 3 to 4 wk of age and used as heterotopic heart donors between 4 and 6 wk of age.

Purification of Immune Cells from Lymphoid Organs.

Splenic DCs were isolated from the spleens of B16-Flt3L melanoma-injected mice. Spleens were harvested from 6- to 12-wk-old mice. Single-cell suspension was prepared by dissociation between frosted slides and then passing through a 70 μm cell strainer. After red blood cell lysis and blocking Fc receptor (anti-mouse CD16/CD32, Tonbo), cells were stained with the anti-mouse CD90.2 (53-2.1), anti-mouse NK1.1 (PK136), anti-mouse CD19 (6D5), anti-mouse Ter119 (Ter119), anti-mouse Ly6G (1A8), anti-mouse F4/80 (BM8), anti-mouse CD317 (937), and lineage-specific biotinylated antibodies for 30 min. Cells were washed and subsequently incubated with MojoSort Streptavidin magnetic beads (BioLegend). Unstained cells were obtained using the MojoSort kit protocol (BioLegend). The purity of isolated splenic DCs was at >95% CD11b⁺ CD11c⁺, CD11b⁺CD11c⁻, and CD11b⁻CD11c⁺. For purification of naive CD4 T cells, the spleen and mesenteric lymph nodes were harvested and made into single-cell suspensions. After RBC lysis, naive CD4 T cells were isolated according to the MojoSort Kit Protocol (BioLegend). For priming alloantigen-specific effector T cells, naive CD4 T cells from Balb/c mouse were cocultured at a 1:2 ratio with B6 splenic DCs for 5 d. After differentiation, total CD4 T cells were isolated using the MojoSort protocol (BioLegend) and replated at 1 × 10⁶/mL in the presence of IL-2 (10 U/mL) and rested for two additional days to allow differentiation into T_EM cells.

Naturally arising T_EM (ex vivo derived) cells were obtained from the spleen and pooled lymph nodes (Inguinal, mesenteric, axillary, brachial, and superficial cervical) of 25- to 35-wk-old BALB/c mice or C57Bl/6J mice. Following RBC lysis of single-cell suspension of the spleen and pooled lymph nodes, total CD4 T cells were enriched using the MojoSort kit (BioLegend). Cells were then labeled with anti-CD62L-biotin conjugated antibody followed by anti-biotin microbeads. Negative selection was performed using magnetic sorting (AUTOMACS, Miltenyi) to obtain CD44^hi CD62L^lo population for interaction with MHC-mismatched DCs.

In Vitro Cocultures of Splenic DCs with Allo T_EM Cells.

A ratio of 1:1 for splenic DC:T_EM cocultures was used. X-VIVO 15 serum-free medium (Lonza) was used for all cocultures to avoid T cell activity to bovine serum proteins. For blocking experiments, T cells were preincubated with anti-MHCII (10 μg/mL; BioLegend) anti-TNFα (20 µg/mL; BioLegend, MP6-XT22) or anti-CD40L (20 µg/mL; BioLegend, MR1) for 30 min before coculture with DCs. For caspase 1, 8, or pan-caspase inhibition, DCs were incubated with 20 μM of Z-YVAD-FMK, Z-IETD-FMK, or Z-VAD-FMK, respectively, for 30 min prior to coculture with allo T_EM cells.

RNA Sequencing or Transcriptome Analysis.

Raw reads from purified RNA of DC sorted cells were processed using the nf-core/rnaseq pipeline version 3.4 (https://dx.doi.org/10.1038/s41587-020-0439-x). Briefly, adapter sequences and low-quality reads were filtered and trimmed using FastQ and Trim Galore Filtered reads were mapped to the mouse GRCm39 reference genome using STAR (10.1093/bioinformatics/bts635). Duplicate reads were removed using Mark Duplicates. Then, Salmon (10.1038/nmeth.4197) was used to generate a count matrix from genes with mapped reads. The full description of the pipeline is available at https://nf-co.re/rnaseq. Differential gene expression between DCs cultured with allogeneic/syngeneic T_EM and unstimulated DCs was evaluated using the DESeq2 R package (https://doi.org/10.1186/s13059-014-0550-8), incorporating the batch information into the model to account for technical variation. Hierarchical clustering of DEGs across all samples revealed a cluster of T cell–related genes expressed in the coculture samples that were likely due to sorting contamination and were removed from further analysis. For PCA and heatmap visualization, batch effects were removed using the removeBatchEffect() function from the limma R package (https://doi.org/10.1093/nar/gkv007). The fgsea R package (10.1101/060012) was used for further gene enrichment analysis using the Hallmark geneset. Heatmaps, dotplots, and PCA plots were generated using Complex Heatmap (10.1093/bioinformatics/btw313) and ggplot2 (ISBN: 978-3-319-24277-4) R packages.

Statistics

Data were statistically analyzed using GraphPad Prism version 10.1.0. Multiple comparisons were performed using ordinary one-way ANOVA using Dunnett’s multiple comparisons test. Statistical comparisons were performed using two-tailed Student’s t test or two-tailed paired t test. P values less than 0.05 were considered significant. All error bars indicate mean ± SEM or mean ± SD wherever specified.

Study Approval

The mouse studies were approved by the IACUC of Cincinnati Children’s Hospital and Medical Center (IACUC No. 2021-0010 and IBC No. 2021-0010).

Supplementary Material

Appendix 01 (PDF)

Data, Materials, and Software Availability

RNA-Sequencing data have been deposited in GEO (GSE253059) (74).