Effector memory T cells (TEM) are less capable of inducing graft-versus-host disease (GVHD) compared with naive T cells (TN). Previously, in the TS1 TCR transgenic model of GVHD, wherein TS1 CD4 cells specific for a model minor histocompatibility Ag (miHA) induce GVHD in miHA-positive recipients, we found that cell-intrinsic properties of TS1 TEM reduced their GVHD potency relative to TS1 TN. Posttransplant, TS1 TEM progeny expressed higher levels of PD-1 than did TS1 TN progeny, leading us to test the hypothesis that TEM induce less GVHD because of increased sensitivity to PD-ligands. In this study, we tested this hypothesis and found that indeed TS1 TEM induced more severe skin and liver GVHD in the absence of PD-ligands. However, lack of PD-ligands did not result in early weight loss and colon GVHD comparable to that induced by TS1 TN, indicating that additional pathways restrain alloreactive TEM. TS1 TN also caused more severe GVHD without PD-ligands. The absence of PD-ligands on donor bone marrow was sufficient to augment GVHD caused by either TEM or TN, indicating that donor PD-ligand–expressing APCs critically regulate GVHD. In the absence of PD-ligands, both TS1 TEM and TN induced late-onset myocarditis. Surprisingly, this was an autoimmune manifestation, because its development required non-TS1 polyclonal CD8+ T cells. Myocarditis development also required donor bone marrow to be PD-ligand deficient, demonstrating the importance of donor APC regulatory function. In summary, PD-ligands suppress both miHA-directed GVHD and the development of alloimmunity-induced autoimmunity after allogeneic hematopoietic transplantation.

Allogeneic hematopoietic stem cell transplantation (alloSCT) can cure hematological malignancies and nonmalignant inherited and acquired disorders of blood cells. Mature allograft αβT cells promote engraftment, contribute to immune reconstitution, and can attack malignant cells, mediating the graft-versus-leukemia (GVL) effect. However, alloreactive T cells also attack recipient nonmalignant cells, causing graft-versus-host disease (GVHD). Because of GVHD, all patients receive some form of immunosuppression to diminish its incidence and severity. A primary goal of alloSCT research is to understand and differentiate mechanisms of GVHD and GVL to maximize the positive effects of donor T cells while minimizing GVHD.

Toward this end, we and others discovered in mouse models that naive T cells (TN) induce severe GVHD, whereas effector memory T cells (TEM) fail to induce sustained GVHD, but engraft and can mediate GVL (15). These data support the selective depletion of TN as a method of GVHD prevention. This has shown promise in humans, wherein TN depletion reduced chronic GVHD, without an apparent increase in risk for relapse (6). Additional efforts to understand the mechanisms behind this effect could help to further optimize this approach in humans.

However, why memory T cells (TM) and, in particular, TEM, fail to induce GVHD is incompletely understood. Deciphering the mechanisms might provide strategies to similarly impair GVHD-inducing TN. We initially considered that TEM may be less capable of causing GVHD because they are relatively restricted from key sites of priming such as lymph nodes, but this proved incorrect (7). To test whether TEM fail to induce GVHD solely because of their having a less alloreactive TCR repertoire and/or whether repertoire-independent properties also reduce their ability to cause GVHD, we developed a TCR-transgenic GVHD model that enabled direct comparison of TN and TEM with identical TCRs. In this model, BALB/c CD4+ TCR-transgenic T cells (TS1 T cells), specific for the hemagglutinin (HA)-derived S1 peptides 110–119 (SFERFEIFPK) presented by I-Ed, were combined with BALB/c bone marrow (BM) and transferred into irradiated BALB/c recipients that express HA at a low level in all cells (HA104 mice) (8). TS1 TN induced severe acute GVHD, characterized by weight loss and typical GVHD pathology of skin, liver, and colon. In contrast, TS1 TEM induced only transient disease, demonstrating that TEM have TCR repertoire-independent limitations.

Although TS1 TN and TEM caused very different degrees of GVHD, they nonetheless proliferated and accumulated to a similar extent in secondary lymphoid tissues early posttransplant (3). However, compared with TN, TS1 TEM progeny produced less IFN-γ and accumulated to a lesser degree in the colon, a major site of GVHD in the model. In addition, although progeny of both TN and TEM upregulated PD-1 posttransplant, PD1 expression was higher on the progeny of TS1 TEM relative to that of TS1 TN in both secondary lymphoid tissues and colon (3).

In this study, we have investigated whether the higher level of PD-1 expression on TS1 TEM progeny leads to greater inhibition, which in turn would reduce their capacity to cause GVHD. We found that PD-ligands regulate GVHD directly mediated by TS1 TN and TEM. Surprisingly, and contrary to our expectations, PD-ligands were also critical to prevent the emergence of autoimmune myocarditis triggered by alloimmune TS1. Together these data indicate that PD-ligands not only restrain alloreactive TS1 TEM, but are also critical for blocking subsequent GVHD-dependent autoimmunity, which could masquerade clinically as GVHD and be more difficult to cure.

BALB/c mice were from the National Cancer Institute (Frederick, MD) or The Jackson Laboratory. After being obtained from the following sources, gene-modified mice were bred at Yale and the University of Pittsburgh: BALB/c RAG2−/− (Taconic); BALB/c PD-L1/2−/− and BALB/c PD-L1−/− (A.H. Sharpe); BALB/c TS1 (9) (A. Adler, University of Connecticut); and BALB/c HA104 (10) (Wistar Institute). These strains were intercrossed to produce the following strains: TS1 RAG2−/−, PD-L1/2−/− HA104, PD-L1−/− HA104, PD-L1/2−/− RAG−/− HA104, and RAG−/− HA104. For all experiments, mice were heterozygous for the HA and TS1 alleles. All TS1 cells were RAG2−/− and are referred to as TS1.

TS1 CD4+ T cells were enriched from splenocytes to >80% purity using a CD4+ T cell negative selection (STEMCELL Technologies). TS1 cells (106 cells/ml) were cultured with HA S1 peptides 110–119 (SFERFEIFPK, 10 μg/ml; Keck Facility, Yale University) and T cell–depleted, irradiated splenocytes (3 × 106 cells/ml) for 3 d. A total of 5 × 106 TS1 cells were transferred i.v. into BALB/c RAG2−/− recipients. TS1 TEM were harvested from these “memory mice” for use 8–14 wk later.

Abs were from the following sources: CD4 Alexa 488 or Alexa 405 (GK1.5), TS1 Alexa 647 (6.5), and F4/80 Alexa 488 (HB-198) were laboratory prepared. CD25 [R-PE (7D4)] was from Southern Biotech. FOXP3 (150D), CD62L PE-Cy7 (Mel-14), PD-1 PE (29F.1A12), CD8α PE (53-6.7), and CD44 allophycocyanin-Cy7 (1M7) were from BioLegend. Gr-1 PE (RB6-8C5) and CD11b PE (MI/70) were from eBioscience. CD4 PE (RM4-5; BD) and CD8 PE (YTSI56.7.7; BioLegend), which could bind in the presence of the depleting Abs, were used for confirmation of in vivo T cell depletions. Ethidium monoazide (Invitrogen) or propidium iodide was used to exclude dead cells.

TS1 TN (CD62L+ CD44dim CD25) were sorted from unmanipulated TS1 mice, and TS1 TEM (CD62L CD44+ CD25) were sorted from memory mice. To avoid staining cells with anti-CD4 Ab, which we found to reduce engraftment, TS1 cells were enriched via exclusion of CD11b+, F4/80+, CD8α+, and CD25+ cells. Sorted cells were 40–60% CD4+TS1+, with no contaminating CD8+ TS1+ cells. Purities of CD4+ TN and TEM among TS1 cells were >99%.

Seven- to twelve-week-old HA104 BALB/c hosts were irradiated with 750 cGy via a cesium irradiator and transplanted with 8 × 106 BM cells alone or in combination with 1000 sorted TS1 TN or TEM, except as otherwise noted. Recipients received wet food and Sulfatrim-treated water for 2 wk posttransplant. Weights were taken two to three times per week. Mice that lost 20% of their body weight were euthanized and scored as dead. For euthanized mice, the last recorded weight was carried forward to subsequent time points. All transplants were performed in accordance with Yale University or the University of Pittsburgh Institutional Animal Care and Use Committee regulations.

Skin, liver, and colon histology were scored as described previously (3). Cardiac histology was scored as follows: 0 = no inflammation or necrosis; 0.5 = minimal inflammation and necrosis involving <5% of the total cross-sectional area of the left and right ventricles; 1 = mild inflammation and focal necrosis involving >5–33%; 2 = moderate inflammation and focal necrosis involving >33–66%; 3 = severe inflammation and focal necrosis involving >66%. Tissues were taken for histopathological analysis when mice dropped <80% of their original body weight (between 30 and 60 d posttransplant). Histopathology for mice that maintained body weight >80% was taken at the end of each experiment.

Livers were perfused with PBS, excised, minced, passed through a cell strainer, and incubated in 0.02% collagenase IV (Sigma) at 37°C for 40 min. Lymphocytes were separated on a 25% Optiprep gradient (Accurate Chemical and Scientific). To perfuse the heart, a cut was first made in the right atrium, and PBS with 1% heparin was then injected into the left ventricle until flow of blood ran clear. The heart was then excised, minced, and incubated with 20 μg/ml DNAse (Sigma) and 0.1 mg/ml Liberase (Roche) at 37°C for 30 min. The resulting cell suspension was passed through a 40-μm cell strainer. Lymphocytes were isolated by centrifugation over a Percoll gradient (GE Healthcare Bio-Sciences AB).

Flow cytometric and cell sorting were performed using an LSRII or Fortessa (BD) and a FACSAria II (BD), respectively. Data were analyzed using FlowJo software (Tree Star). For intracellular cytokine staining, splenocytes were cultured for 1 h in the presence of 20 ng/ml PMA (VWR Scientific) and 750 ng/ml ionomycin (VWR Scientific), and for an additional 2 h in the presence of 10 μg/ml brefeldin A (BioLegend). Cells were treated with DNase (Sigma), surface stained for CD4 and TS1, and treated with Cytofix/Cytoperm (BD). Perm/Wash (BD) was used for intracellular staining with IFN-γ PE (XMG1.2; BioLegend). For FOXP3 staining, cells were surface stained for CD4 and TS1, treated with FOXP3 Fix/Perm buffer (BioLegend), and stained for FOXP3 in FOXP3 perm buffer (BioLegend).

To deplete CD8+ T cells, mice received 200 μg TIB105 (Bio X Cell) i.p. on days −6 and −4. Mice were transplanted on day 0 and injected with 100 μg of TIB105 on days 6, 13, 20, 27, 34, and 41. To deplete CD4 cells, mice received 150 μg of GK1.5 (laboratory-made) on days 27, 31, 35, 39, 42, and 46 posttransplant. Additional groups received both anti-CD4 and anti-CD8 or control rat IgG2a (Bio X Cell).

Echocardiographic assessments of cardiac structure and function were performed 38 d posttransplant using a 40-MHz probe and a Vevo2100 Ultrasound system (FUJIFILM; VisualSonics, Toronto, ON, Canada). Mice were lightly anesthetized with 0.5–1.0% isoflurane with continuous monitoring of heart rate, body temperature (kept at normal by a heated platform), and respiratory rate. Two-dimensionally guided M-mode images of the left ventricle were acquired in addition to two-dimensional evaluation of left ventricular size and function. The data were collected in both short- and long-axis views at heart rates >400 beats/min. A total of three to five cardiac cycles per measurement were averaged in each view to obtain functional and anatomical data.

The significance of weight change was calculated by Student t test. The significance of survival differences was calculated by log rank test. The significance of all other comparisons (pathology, cell numbers, percentage of IFN-γ+ cells, and IFN-γ mean fluorescence intensity) was calculated by the Mann–Whitney U test.

To investigate whether PD1 regulates TS1-induced GVHD, we modified the TS1 GVHD model to use BM donors and HA104 hosts that lacked both PD-L1 and PD-L2. We chose to create an environment without PD-ligands instead of using PD-1−/− TS1 TEM to avoid the confounding effects the absence of PD-1 might have on TS1 development and TEM generation. As reported previously (3), transplantation of wild type (WT) BM and TS1 TEM into WT HA104 recipients caused no death and only mild, transient weight loss (Fig. 1A, 1B). In contrast, the transfer of TS1 TEM and PD-L1/2−/− BM into PD-L1/2−/− HA104 hosts caused severe weight loss and death (Fig. 1A, 1B), but with delayed kinetics compared with what is observed with TS1 TN with intact PD-ligands (3) (Fig. 2A). Weight loss began 35 d posttransplant and rapidly progressed, with the majority of TEM recipients dying by day 50. Histological analysis revealed that TEM, in the absence of PD-ligands, induced more severe skin and liver GVHD relative to when PD-ligands were intact (Fig. 1C, 1D). In particular, liver GVHD was more severe than was induced by TS1 TN in a PD-ligand–sufficient environment (p = 0.0064, Fig. 2). In contrast, the absence of PD-ligands did not increase colon GVHD (Fig. 1E), which is a dominant feature of TS1 TN-mediated disease (3) (Fig. 2). Very mild liver disease was also observed in PD-L1/2−/− HA104 recipients of PD-L1/2−/− BM without TS1 TEM, indicating that some aspects of liver disease are TS1 independent (Fig. 1D).

FIGURE 1.

PD-L1/2 limit TS1 TEM-mediated disease in the skin, liver, and heart, but not the colon. (AF) WT or PD-L1/L2−/− BM was transferred into irradiated WT or PD-L1/2−/− HA104 mice, respectively, with or without 1000 TS1 TEM. Shown are survival (A), weight loss (B), and pathology scores for the skin (C), liver (D), colon (E), and heart (F). Data from (A)–(E) were combined from two independent experiments. Data from (F) were combined from two additional independent experiments, for which weight and survival curves are shown in Supplemental Fig. 1. The p values for survival and weight curves are relative to WT HA104 recipients of WT BM and TS1 TEM. (GI) PD-L1/2−/− BM with or without TS1 TEM were transferred into irradiated HA104 PD-L1/2−/− hosts. Thirty-eight days posttransplant, mice were anesthetized and underwent echocardiography. (G) Weights are shown. (H) ECG tracing for one BM control and one TS1 TEM recipient is shown. The ECG tracing for the TS1 TEM recipient was representative of two out of five mice. (I) Heart rate, cardiac output, diastolic volume and systolic volume are shown. Data are from one experiment.

FIGURE 1.

PD-L1/2 limit TS1 TEM-mediated disease in the skin, liver, and heart, but not the colon. (AF) WT or PD-L1/L2−/− BM was transferred into irradiated WT or PD-L1/2−/− HA104 mice, respectively, with or without 1000 TS1 TEM. Shown are survival (A), weight loss (B), and pathology scores for the skin (C), liver (D), colon (E), and heart (F). Data from (A)–(E) were combined from two independent experiments. Data from (F) were combined from two additional independent experiments, for which weight and survival curves are shown in Supplemental Fig. 1. The p values for survival and weight curves are relative to WT HA104 recipients of WT BM and TS1 TEM. (GI) PD-L1/2−/− BM with or without TS1 TEM were transferred into irradiated HA104 PD-L1/2−/− hosts. Thirty-eight days posttransplant, mice were anesthetized and underwent echocardiography. (G) Weights are shown. (H) ECG tracing for one BM control and one TS1 TEM recipient is shown. The ECG tracing for the TS1 TEM recipient was representative of two out of five mice. (I) Heart rate, cardiac output, diastolic volume and systolic volume are shown. Data are from one experiment.

Close modal
FIGURE 2.

PD-L1/2 limit TS1 TN-mediated disease in the skin, liver, and heart, but not the colon. WT or PD-L1/2−/− BM was transferred into irradiated WT or PD-L1/2−/− HA104 mice, respectively, with or without 1000 TS1 TN. Shown are survival (A), weight loss (B), and pathology scores for the skin (C), liver (D), colon (E), and heart (F). Data from (A)–(E) were combined from three independent experiments. Data from (F) were combined from two additional independent experiments, for which survival and weight curves are shown in Fig. 3A and 3B. The p values for survival and weight curves are relative to WT HA104 recipients of WT BM and TS1 TN.

FIGURE 2.

PD-L1/2 limit TS1 TN-mediated disease in the skin, liver, and heart, but not the colon. WT or PD-L1/2−/− BM was transferred into irradiated WT or PD-L1/2−/− HA104 mice, respectively, with or without 1000 TS1 TN. Shown are survival (A), weight loss (B), and pathology scores for the skin (C), liver (D), colon (E), and heart (F). Data from (A)–(E) were combined from three independent experiments. Data from (F) were combined from two additional independent experiments, for which survival and weight curves are shown in Fig. 3A and 3B. The p values for survival and weight curves are relative to WT HA104 recipients of WT BM and TS1 TN.

Close modal

To better pinpoint the cause of clinical disease, we performed repeat experiments (weight loss and survival; Supplemental Fig. 1) in which we examined kidney, lung, stomach, tongue, and small intestine and found no significant pathology (data not shown). However, in the absence of PD-L1/2, TS1 TEM recipients developed cardiac pathology (Fig. 1F), manifested by a myocardial inflammatory infiltrate with cardiomyocyte necrosis (Supplemental Fig. 2). Notably, severe heart disease did not develop in PD-L1/2−/− BM controls.

This inflammation had apparent functional significance. Approximately 5 wk posttransplant, in the absence of PD-ligands, TS1 recipients had a reduction in heart rate (Fig. 1G–I). Two of five such mice had complete heart block and a third had an ectopic atrial focus, consistent with conduction system damage. Echocardiograms demonstrated reduced cardiac output and diastolic and systolic stroke volumes, even in mice without conduction abnormalities, further evidence that the myocarditis was likely to be functionally significant and may have been a cause of death (Fig. 1G–I).

Taken together, these data indicate that PD-ligands limit some, but not all, aspects of TEM function posttransplant, because in their absence TS1 TEM induce skin, liver, and late-onset heart disease. However, in the absence of PD-ligands, TS1 TEM fail to recapitulate the severity of GVHD caused by TS1 TN, particularly colon GVHD, indicating that pathways in addition to those downstream of PD-1 distinguish TEM and TN function posttransplantation.

To determine the role of PD-ligands in regulating GVHD mediated by TS1 TN, we reconstituted irradiated WT HA104 or PD-L1/2−/− HA104 recipients with WT or PD-L1/2−/− BM (respectively), with or without TS1 TN. TS1 TN caused more early weight loss and death in the absence of PD-L1/2 (Fig. 2A, 2B), consistent with prior reports in GVHD models (1113). This contrasts with what was observed with TS1 TEM in a PD-ligand–deficient environment wherein the increases in weight loss and death were delayed relative to what was observed with TN (compare Fig. 1A, 1B with Fig. 2A, 2B; for weight loss, p ≤ 0.003 from day +6 onward; for survival, p = 0.0002). Nonetheless, TS1 TN recipients developed a second wave of weight loss and death that shared the kinetics seen with TS1 TEM. In the absence of PD-L1/2, TS1 TN caused slightly increased skin disease and markedly more severe liver disease, but colon disease was similar to when PD-ligands were present (Fig. 2C–E). Consistent with our prior report (3), in a PD-ligand–sufficient environment, TS1 TN induced more severe histological liver and colon GVHD than did TS1 TEM (p ≤ 0.0024). However, in PD-ligand–deficient environments, TS1 TEM and TN caused similar liver GVHD.

Additional experiments were performed to evaluate heart disease (weight and survival are shown in Fig. 3A, 3B). TS1 TN caused very mild heart disease when PD-L1/2 were intact; but, importantly, they induced severe heart disease when PD-L1/2 were absent (Fig. 2F; representative histology, Supplemental Fig. 3). These data indicate that PD-L1/2 limit TS1 TN-induced skin, liver, and heart disease, similar to what was observed in TS1 TEM recipients. However, unlike with TS1 TEM, PD-L1/2 also limit early TS1 TN-induced weight loss and death.

FIGURE 3.

Role of PD-L1/2 on donor BM and host cells. (AE) TS1 TN and WT BM were transferred into irradiated PD-L1/2−/− HA104 hosts, and TS1 TN and PD-L1/2−/− BM were transplanted into irradiated WT HA104 hosts. As controls, WT BM into WT HA104 hosts and PD-L1/2−/− BM into PD-L1/2−/− HA104 hosts, with or without TS1 TN, were also generated. Survival (A), weight loss (B), and pathology scores for skin (C), liver (D), and heart (E) are shown. Data for Fig. 2F and (E) were generated in the same experiment (two repetitions, data combined) to reduce mouse numbers. Data in (E) from groups in which donor and recipient were both PD-L1/2+/+ or PD-L1/2−/− are therefore the same as in Fig. 2F. (FJ) Mice were transplanted as in (A)–(E), except TN were replaced with TEM. Survival (F), weight loss (G), and pathology scores for skin (H), liver (I), and heart (J) are shown. TN and TEM results were each combined from two independent experiments. The p values for survival and weight curves are relative to HA104 WT recipients of WT BM and TS1 TEM or TN.

FIGURE 3.

Role of PD-L1/2 on donor BM and host cells. (AE) TS1 TN and WT BM were transferred into irradiated PD-L1/2−/− HA104 hosts, and TS1 TN and PD-L1/2−/− BM were transplanted into irradiated WT HA104 hosts. As controls, WT BM into WT HA104 hosts and PD-L1/2−/− BM into PD-L1/2−/− HA104 hosts, with or without TS1 TN, were also generated. Survival (A), weight loss (B), and pathology scores for skin (C), liver (D), and heart (E) are shown. Data for Fig. 2F and (E) were generated in the same experiment (two repetitions, data combined) to reduce mouse numbers. Data in (E) from groups in which donor and recipient were both PD-L1/2+/+ or PD-L1/2−/− are therefore the same as in Fig. 2F. (FJ) Mice were transplanted as in (A)–(E), except TN were replaced with TEM. Survival (F), weight loss (G), and pathology scores for skin (H), liver (I), and heart (J) are shown. TN and TEM results were each combined from two independent experiments. The p values for survival and weight curves are relative to HA104 WT recipients of WT BM and TS1 TEM or TN.

Close modal

We next dissected the individual roles of PD-ligands on recipient and donor-derived cells in modulating TS1-induced GVHD. We transferred TS1 TN or TEM and WT BM into PD-L1/2−/− HA104 hosts and, reciprocally, TS1 TN or TEM and PD-L1/2−/− BM into WT HA104 hosts. Compared with WT HA104 recipients of TS1 TN and WT BM, the absence of host PD-L1/2 alone did not increase skin, liver, or heart disease, but did increase mortality. In contrast, the absence of PD-L1/2 on donor BM alone did not increase mortality but exacerbated skin, liver, and heart disease, with scores as severe as those when PD-L1/2 were lacking on both donor BM and host cells (Fig. 3A–E). Notably, in this experiment, there was less early death than typically seen when donor and host lack PD-ligands. In contrast to the case with TN-mediated GVHD, neither clinical nor histopathological TS1 TEM-induced GVHD was increased when only the recipients lacked PD-ligands (Fig. 3F–J). However, the absence of PD-L1/2 on only donor BM increased weight loss and liver and heart pathology. Nonetheless, GVHD was less severe than when PD-ligands were absent on both the recipient and the donor BM, indicating that PD-L1/2 on both donor BM and host cells modulate TS1 TEM-induced GVHD.

We next determined the contributions of PD-L1/2 on host hematopoietic and radioresistant, primarily nonhematopoietic, cells. To do so, we created WT HA104→PD-L1/2−/− HA104 and PD-L1/2−/− HA104→WT HA104 BM chimeras and used these as recipients in a second transplant in which GVHD was induced by TS1 TEM. As controls, WT HA104→WT HA104 and PD-L1/2−/− HA104→PD-L1/2−/− HA104 chimeras were also transplanted. Because the absence of PD-L1/2 on host cells alone does not enable TS1 TEM-mediated GVHD (Fig. 3), all chimeras received PD-L1/2−/− BM in the second GVHD-inducing transplant. As expected, transfer of TS1 TEM and PD-L1/2−/− BM caused mild heart disease in WT→WT chimeras and severe weight loss, death, and heart disease in PD-L1/2−/−→PD-L1/2−/− chimeras. In comparison with PD-L1/2−/−→PD-L1/2−/− chimeras, disease was similar in WT→PD-L1/2−/− chimeras but was reduced in PD-L1/2−/−→WT chimeras (Fig. 4A–E). Therefore, PD-L1/2 on donor BM and host stromal cells, but not recipient hematopoietic cells, inhibits TS1 TEM-dependent cardiac disease.

FIGURE 4.

PD-1/2 expression on host stromal cells protects the heart from TS1 TEM-induced disease. The following BM chimeras were generated with WT HA104 and PD-L1/2−/− HA104 mice: (PD-L1/2−/−→WT), (WT→PD-L1/2−/−), (PD-L1/2−/−→PD-L1/2−/−), and (WT→WT). Eight weeks later, chimeras were reirradiated and transplanted with PD-L1/2−/− BM and 1000 TS1 TEM. Survival (A), weight loss (B) and pathology for the skin (C), liver (D), and heart (E) are shown. Data were combined from two independent experiments. For weight and survival curves, p values are relative to WT→WT recipients of PD-L1/2−/− BM and TS1 TEM.

FIGURE 4.

PD-1/2 expression on host stromal cells protects the heart from TS1 TEM-induced disease. The following BM chimeras were generated with WT HA104 and PD-L1/2−/− HA104 mice: (PD-L1/2−/−→WT), (WT→PD-L1/2−/−), (PD-L1/2−/−→PD-L1/2−/−), and (WT→WT). Eight weeks later, chimeras were reirradiated and transplanted with PD-L1/2−/− BM and 1000 TS1 TEM. Survival (A), weight loss (B) and pathology for the skin (C), liver (D), and heart (E) are shown. Data were combined from two independent experiments. For weight and survival curves, p values are relative to WT→WT recipients of PD-L1/2−/− BM and TS1 TEM.

Close modal

Whereas PD-L1 is widely expressed and can be induced by IFN-γ, PD-L2 is expressed primarily on hematopoietic cells (14). To test whether PD-L1 is sufficient to modulate TS1-mediated GVHD, we transferred TS1 TEM and PD-L1−/− BM into PD-L1−/− HA104 hosts. GVHD in the absence of PD-L1 alone phenocopied TS1 TEM-mediated GVHD in the absence of both PD-L1 and PD-L2 (Fig. 5), indicating that PD-L1 is sufficient for GVHD suppression.

FIGURE 5.

The absence of only PD-L1 is sufficient for the development of TS1 TEM-induced heart disease. WT or PD-L1−/− BM alone or in combination with TS1 TEM were transferred into irradiated HA104 WT or HA104 PD-L1−/− hosts, respectively. Survival (A), weight loss (B), and pathology scores for the skin (C), liver (D), and heart (E) are shown. Data were combined from two independent experiments. The p values for weight and survival curves are relative to HA104 recipients of WT BM and TS1 TEM.

FIGURE 5.

The absence of only PD-L1 is sufficient for the development of TS1 TEM-induced heart disease. WT or PD-L1−/− BM alone or in combination with TS1 TEM were transferred into irradiated HA104 WT or HA104 PD-L1−/− hosts, respectively. Survival (A), weight loss (B), and pathology scores for the skin (C), liver (D), and heart (E) are shown. Data were combined from two independent experiments. The p values for weight and survival curves are relative to HA104 recipients of WT BM and TS1 TEM.

Close modal

Despite increased TS1 TEM-induced GVHD in the absence of PD-L1/2, TS1 numbers in the spleen and the percentage of IFN-γ+ TS1 at days +14 or +42 were similar (Fig. 6). However, in the absence of PD-L1/2, splenic TS1 at day +42 produced more IFN-γ/cell as measured by intracellular cytokine staining (Fig. 6C). In the liver, TS1 numbers were increased in the absence of PD-ligands only on day +42 (Fig. 6A), consistent with the increased liver GVHD scores. In summary, these data indicate that PD-L1/2 diminishes IFN-γ production by splenic TS1 cells (15), which could result in fewer pathogenic TS1 cells infiltrating GVHD target organs. Alternatively, or in addition, PD-L1/2 could suppress TS1 cells locally.

FIGURE 6.

Characterization of TS1 and polyclonal T cells in the absence of PD-L1/2. (AD) Irradiated WT or HA104 mice were reconstituted with WT or PD-L1/2−/− BM, respectively, with no T cells or with 1000 TS1 TEM. (A) Number of TS1 cells in spleen and liver on days +14 and +42 and the numbers of TS1 from heart on day +42 (D). Data were combined from two independent experiments for a total of 8–12 mice per group. The percentages of splenic TS1 cells on days +14 and +42 that were IFN-γ+ and the IFN-γ mean fluorescence intensity are shown in (B) and (C). Data were combined from two independent experiments for a total of 19 mice per group for day +42 and from one experiment with five mice per group for day +14. (E and F) The number of non-TS1 CD4+ T cells and CD8+ T cells per heart are shown. Data are representative of two independent experiments, each with three to four mice per group.

FIGURE 6.

Characterization of TS1 and polyclonal T cells in the absence of PD-L1/2. (AD) Irradiated WT or HA104 mice were reconstituted with WT or PD-L1/2−/− BM, respectively, with no T cells or with 1000 TS1 TEM. (A) Number of TS1 cells in spleen and liver on days +14 and +42 and the numbers of TS1 from heart on day +42 (D). Data were combined from two independent experiments for a total of 8–12 mice per group. The percentages of splenic TS1 cells on days +14 and +42 that were IFN-γ+ and the IFN-γ mean fluorescence intensity are shown in (B) and (C). Data were combined from two independent experiments for a total of 19 mice per group for day +42 and from one experiment with five mice per group for day +14. (E and F) The number of non-TS1 CD4+ T cells and CD8+ T cells per heart are shown. Data are representative of two independent experiments, each with three to four mice per group.

Close modal

Unexpectedly, the absence of PD-L1/2 reduced the number of TS1 cells in the heart (Fig. 6D). However, non-TS1 CD4+ and CD8+ T cells were increased 3- and 6-fold, respectively (Fig. 6E, 6F). This increase occurred only in TS1 TEM recipients and not in PD-L1/2−/− BM alone controls. These data suggested a model wherein TS1 cells are required for the initiation of a process that leads to heart disease, but that non-TS1 T cells are directly pathogenic. These polyclonal T cells could be donor or recipient BM-derived, radioresistant recipient mature T cells or a combination of these.

To test the hypothesis that polyclonal T cells contribute to heart disease, we developed a system that lacked all non-TS1 T cells. PD-L1/2−/− HA104 or PD-L1/2−/− RAG2−/− (triple knockout [TKO]) HA104 mice were transplanted with PD-L1/2−/− or TKO BM, respectively, with no T cells or with TS1 TEM or TS1 TN. TKO HA104 recipients of TS1 TEM and TKO BM developed early weight loss and liver disease similar to controls; however, TKO recipients suffered less late-onset weight loss and death and, strikingly, no heart disease (Fig. 7A–D). Likewise, both RAG2-sufficient and RAG2-deficient TS1 TN groups developed GVHD, manifested by weight loss (Fig. 7E). Initial weight loss was more severe in RAG2−/− recipients. However, after regaining some weight, the RAG2-sufficient TS1 TN group developed a second wave of weight loss beginning around day +40. We harvested hearts from both groups beginning on day +47. Strikingly, there was severe heart disease in RAG2-intact TS1 TN recipients, whereas cardiac pathology scores in TKO TS1 TN recipients were indistinguishable from negative control PD-L1/2−/− RAG2+/+ BM alone mice (Fig. 7F; representative histology, Supplemental Fig. 4). Consistent with a role for polyclonal CD8 cells in mediating heart disease, they were readily detectable in the spleen (Fig. 7G). Therefore, non-TS1 T cells are required for TS1-induced heart disease, but not for other aspects of GVHD. We again observed low-level liver histopathology in PD-L1/2−/− recipients of PD-L1/2−/− BM without TS1, but not when donors and recipients were also RAG2−/−. This makes it likely that such disease is caused by polyclonal donor- or recipient-derived T cells.

FIGURE 7.

Polyclonal non-TS1 T cells are required for TS1-induced heart disease. (AF) PD-L1/2−/− HA104 or PD-L1/2−/− RAG2−/− (TKO) HA104 mice were transplanted with PD-L1/2−/− or TKO BM, respectively, with no T cells or with 1000 TS1 TEM (A–D) or 300 TS1 TN (E and F). The TS1 TEM experiments had an additional group wherein HA104 RAG2−/− mice were transplanted with RAG2−/− BM with no T cells or with TS1 TEM. Survival (A), weight loss (B), and pathology scores for the liver (C) and heart (D) from the TS1 TEM experiments are shown. Data were combined from two independent experiments.The p values for weight and survival curves are relative to RAG2−/− HA104 recipients of TS1 TEM and RAG2−/− BM and are significant from day +31 onward. Weight loss (E) and heart pathology scores (F) for the TS1 TN experiment are shown. Percentages of CD8+, CD4+ TS1, and CD4+ TS1+ cells in the spleen when mice were sacrificed for pathology are shown in (G). Data are from one experiment. p < 0.05, comparing weights of either TS1 group with its BM alone control from days 9 to 60 after BMT. TKO TS1 recipients had significantly more weight loss (p < 0.05) than PD-L1/2−/− TS1 recipients on days 2, 7–9, 12–21, and 23–58.

FIGURE 7.

Polyclonal non-TS1 T cells are required for TS1-induced heart disease. (AF) PD-L1/2−/− HA104 or PD-L1/2−/− RAG2−/− (TKO) HA104 mice were transplanted with PD-L1/2−/− or TKO BM, respectively, with no T cells or with 1000 TS1 TEM (A–D) or 300 TS1 TN (E and F). The TS1 TEM experiments had an additional group wherein HA104 RAG2−/− mice were transplanted with RAG2−/− BM with no T cells or with TS1 TEM. Survival (A), weight loss (B), and pathology scores for the liver (C) and heart (D) from the TS1 TEM experiments are shown. Data were combined from two independent experiments.The p values for weight and survival curves are relative to RAG2−/− HA104 recipients of TS1 TEM and RAG2−/− BM and are significant from day +31 onward. Weight loss (E) and heart pathology scores (F) for the TS1 TN experiment are shown. Percentages of CD8+, CD4+ TS1, and CD4+ TS1+ cells in the spleen when mice were sacrificed for pathology are shown in (G). Data are from one experiment. p < 0.05, comparing weights of either TS1 group with its BM alone control from days 9 to 60 after BMT. TKO TS1 recipients had significantly more weight loss (p < 0.05) than PD-L1/2−/− TS1 recipients on days 2, 7–9, 12–21, and 23–58.

Close modal

To determine whether recipient T cells were sufficient to cause cardiac disease, we performed GVHD experiments with TKO donor BM and RAG2+/+ PD-L1/2−/− HA104 recipients. Compared with recipients of TS1 TEM and PD-L1/2−/− RAG2+/+ BM, recipients of TS1 TEM and TKO BM developed similar weight loss and liver disease while mortality and heart disease were increased (Fig. 8), indicating that recipient polyclonal T cells are sufficient for myocarditis, whereas donor BM-derived T cells are not required and might be protective. Unexpectedly, heart disease also developed in a small but significant number of PD-L1/2−/− HA104 recipients of TKO BM that did not receive TS1 TEM, but not in TKO HA104 recipients of TKO BM without TS1 TEM (Fig. 7D). Host CD4 and CD8 cells were readily detected in spleens of PD-L1/2−/− recipients of TKO BM, with or without TS1 TEM, at day 60 (Fig. 8F), consistent with a role for host polyclonal T cells in mediating heart disease in both groups. Although PD-L1/2−/− HA104 recipients of PD-L1/2−/− BM alone did not develop severe heart disease, there were also occasional mice with mild disease (see Figs. 2F, 3E, 3J).

FIGURE 8.

Polyclonal recipient T cells are sufficient for TS1 TEM-induced myocarditis. Irradiated PD-L1/L2−/− HA104 mice were reconstituted with 1000 TS1 TEM and RAG2−/− PD-L1/2−/− BM or RAG2+/+ PD-L1/2−/− BM. The following control groups were also transplanted: 1) HA104 mice reconstituted with BALB/c BM and no T cells; 2) PD-L1/2−/− HA104 mice reconstituted with PD-L1/2−/−RAG2−/− BM and no T cells; and 3) WT HA104 mice reconstituted with RAG2−/− BM and 1000 TS1 TEM. Survival (A), weight loss (B), and pathology scores for the skin (C), liver (D), and heart (E) are shown. The p values for weight and survival curves are relative to WT HA104 recipients of TS1 TEM and RAG−/− BM. Percentages of CD8+, CD4+ TS1, CD4+ FOXP3+ TS1, and CD4+ TS1+ cells in the spleen at day 60 posttransplant for two PD-L1/2−/− recipients of TKO BM with and without TS1 TEM are shown in (F). Data for (A)–(E) were combined from two independent experiments. Data in (F) were from one experiment.

FIGURE 8.

Polyclonal recipient T cells are sufficient for TS1 TEM-induced myocarditis. Irradiated PD-L1/L2−/− HA104 mice were reconstituted with 1000 TS1 TEM and RAG2−/− PD-L1/2−/− BM or RAG2+/+ PD-L1/2−/− BM. The following control groups were also transplanted: 1) HA104 mice reconstituted with BALB/c BM and no T cells; 2) PD-L1/2−/− HA104 mice reconstituted with PD-L1/2−/−RAG2−/− BM and no T cells; and 3) WT HA104 mice reconstituted with RAG2−/− BM and 1000 TS1 TEM. Survival (A), weight loss (B), and pathology scores for the skin (C), liver (D), and heart (E) are shown. The p values for weight and survival curves are relative to WT HA104 recipients of TS1 TEM and RAG−/− BM. Percentages of CD8+, CD4+ TS1, CD4+ FOXP3+ TS1, and CD4+ TS1+ cells in the spleen at day 60 posttransplant for two PD-L1/2−/− recipients of TKO BM with and without TS1 TEM are shown in (F). Data for (A)–(E) were combined from two independent experiments. Data in (F) were from one experiment.

Close modal

Taken together, these results support the idea that the absence of PD-ligands during T cell reconstitution in the alloSCT setting is permissive for myocarditis mediated by autoreactive polyclonal T cells. Alloreactive TS1 cells reproducibly trigger this autoimmunity, whereas donor BM-derived polyclonal T cells may suppress it.

If regulatory T cell (Treg) reconstitution in the absence of PD-ligands was diminished, this could have contributed to autoimmune myocarditis. We therefore enumerated Tregs in irradiated WT or PD-L1/2−/− HA104 mice that were reconstituted with WT or PD-L1/2−/− BM, respectively, with or without TS1 TEM (Fig. 9). On day 14 posttransplant, with or without TS1 TEM, PD-L1/2−/− recipients of PD-L1/L2−/− BM had higher percentages and total numbers of Tregs; however, there was no difference in the ratios of TS1 to Tregs. On day +42 posttransplant, a higher percentage of splenocytes were Tregs in the absence of PD-L1/2, although neither the total numbers nor the Treg:TS1 ratios differed. In the liver on day +42, PD-L1/2−/− recipients of PDL1/2−/− BM with and without TS1 TEM had a greater number and percentage of Tregs, but again without a significant increase in the Treg:TS1 ratio. In summary, these data make it unlikely that autoimmunity developed in the absence of PD-L1/L2 because of poor Treg reconstitution.

FIGURE 9.

The absence of PD-ligands does not diminish Treg reconstitution. Irradiated WT or HA104 mice were reconstituted with WT or PD-L1/2−/− BM, respectively, with or without 1000 TS1 TEM. Percentages and numbers of Tregs (CD4+ FOXP3+) as well as Treg to TS1+ T cell ratios in the spleen and liver on day 14 and in the spleen and liver on day 42 are shown in (A)–(D). Day 14 data were combined from two independent experiments, and day 42 data were combined from three independent experiments. TS1 cell numbers from these experiments are shown in Fig. 6A.

FIGURE 9.

The absence of PD-ligands does not diminish Treg reconstitution. Irradiated WT or HA104 mice were reconstituted with WT or PD-L1/2−/− BM, respectively, with or without 1000 TS1 TEM. Percentages and numbers of Tregs (CD4+ FOXP3+) as well as Treg to TS1+ T cell ratios in the spleen and liver on day 14 and in the spleen and liver on day 42 are shown in (A)–(D). Day 14 data were combined from two independent experiments, and day 42 data were combined from three independent experiments. TS1 cell numbers from these experiments are shown in Fig. 6A.

Close modal

To determine the individual roles of polyclonal CD4+ and CD8+ T cells in TS1 TEM-induced cardiac disease, we used depleting Abs in PD-L1/2−/− HA104 recipients of TS1 TEM and PD-L1/2−/− BM. CD8 cells were depleted with anti-CD8 twice in the week before transplant and then weekly thereafter. To avoid depleting TS1 cells during GVHD development, we began CD4 depletion on day +27, before the onset of clinical cardiac disease (manifest by weight loss).

CD4 depletion paradoxically increased mortality and weight loss but had no impact on heart disease (Fig. 10A–C). In contrast, CD8 depletion alone and combined CD8 and CD4 depletion decreased both mortality and heart disease. Enumeration of splenic CD4+ and CD8+ T cells revealed that anti-CD4 treatment slightly decreased CD8+ T cell numbers but drastically reduced TS1 and non-TS1 CD4+ T cells numbers (Fig. 10D). Anti-CD8 treatment had no effect on CD4+ T cell numbers but reduced CD8+ T cells. Taken together, these data indicate that polyclonal CD8 cells, and not CD4 cells, are essential for heart disease. The persistence of heart disease even after profound TS1 depletion suggests that once autoimmune cardiac disease is established, it is independent of further actions by alloimmune cells.

FIGURE 10.

In the absence of PD-L1/2, CD8+ T cells are required for TS1-induced heart disease. WT or PD-L1/2−/− BM, with or without TS1 TEM, were transferred into WT HA104 or PD-L1/2−/− HA104 hosts, respectively. PD-L1/2−/− TS1 recipients were given isotype control, anti-CD8, anti-CD4, or both anti-CD4 and anti-CD8 Abs as follows: 1) isotype control and anti-CD8 were given on days −6, −4, 6, 13, 20, 27, 34, and 41; and 2) anti-CD4 was given on days 27, 31, 35, 39, 42, and 46. Survival (A), weight loss (B), and heart pathology (C) are shown. (D) Number of splenocytes that were CD4+ TS1, TS1+, or CD8+ are shown. Data were combined from two independent experiments. For weight and survival curves, p values are relative to isotype control recipients.

FIGURE 10.

In the absence of PD-L1/2, CD8+ T cells are required for TS1-induced heart disease. WT or PD-L1/2−/− BM, with or without TS1 TEM, were transferred into WT HA104 or PD-L1/2−/− HA104 hosts, respectively. PD-L1/2−/− TS1 recipients were given isotype control, anti-CD8, anti-CD4, or both anti-CD4 and anti-CD8 Abs as follows: 1) isotype control and anti-CD8 were given on days −6, −4, 6, 13, 20, 27, 34, and 41; and 2) anti-CD4 was given on days 27, 31, 35, 39, 42, and 46. Survival (A), weight loss (B), and heart pathology (C) are shown. (D) Number of splenocytes that were CD4+ TS1, TS1+, or CD8+ are shown. Data were combined from two independent experiments. For weight and survival curves, p values are relative to isotype control recipients.

Close modal

It has long been known that TEM cause less GVHD than do TN, a phenomenon that we and others found to be independent of TCR repertoire differences (3, 16). The difference between TN and TM is clinically relevant because a recent human trial has suggested a benefit with TN depletion (6). However, the mechanisms that restrain the GVHD-inducing potential of TM have not been elucidated. In this work, we found that PD-ligands are one component that limits alloreactive TEM. PD-ligands inhibited TS1 TEM-mediated GVHD through two main mechanisms. First, they suppressed HA-directed GVHD in the liver and skin. However, neither colon histopathology, a dominant feature of TS1 TN-induced GVHD, nor early weight loss, were restored to that induced by TS1 TN in a WT or PD-L1/2−/− environment. Therefore, additional properties of TS1 TEM must diminish their ability to cause GVHD relative to TS1 TN. Second, and contrary to expectations, PD-ligands maintained tolerance of polyclonal nonalloreactive T cells, which, in the absence of PD-ligands, was reproducibly broken by alloreactive TS1 TEM. TS1 TN-HA–directed GVHD was also more severe in the absence of PD-ligands. As we observed with TS1 TEM, in the absence of PD-ligands and in the presence of polyclonal T cells, TS1 TN induced myocarditis and late-onset weight loss and death.

The roles of PD-ligands in modifying HA-directed GVHD were complex and context dependent, in part owing to their expression on hematopoietic as well as parenchymal cells. Both TS1 TEM- and TN-induced liver GVHD were suppressed by PD-ligands. TS1 TEM were particularly sensitive to PD-1 suppression because liver scores in the absence of PD-ligands were similar to scores in TS1 TN recipients, with or without PD-ligands. In contrast, colon GVHD was not increased in the absence of PD-ligands in TS1 TN and TEM recipients, which suggests the possibility that PD-ligands do not regulate colon GVHD directly. For TN-mediated GVHD, the absence of PD-ligands on only donor BM-derived cells was necessary and sufficient for maximal acute GVHD intensification. These features were not modified when hosts also lacked PD-L1/2−/−. The absence of PD-ligands on only donor BM was also sufficient to increase TS1 TEM-induced HA-directed liver GVHD; however, skin disease and maximal liver GVHD required that PD-ligands be absent from both donor BM and recipient. The dominant role of PD-ligands on donor BM-derived cells emphasizes the importance of indirect presentation of minor histocompatibility Ags (miHAs; the HA-derived S1 peptide in this case) by donor APCs to donor CD4 cells (17, 18). Although indirect presentation is commonly thought of as being stimulatory and disease-promoting, our data show that indirect miHA presentation by PD-ligand–expressing donor APCs suppresses GVHD. This regulation could occur in secondary lymphoid tissues, peripheral GVHD target organs, or both. PD-ligands on hematopoietic APCs may be especially important in the regulation of CD4-mediated immunopathology, because CD4-mediated GVHD in miHA-disparate MHC-matched models can occur when nonhematopoietic cells are MHCII (19). Therefore, hematopoietic cells could be the only cells, or at least the most common cell, capable of engaging CD4 cells by their TCRs while presenting PD-ligands.

In contrast with our studies, prior work on PD-ligands in GVHD found a dominant role for host PD-ligands and, when examined, on host radioresistant nonhematopoietic cells (12). In these studies, GVHD was in MHC-mismatched models. As such, direct recognition of recipient MHC likely dominated the alloimmune response, which would dictate less of a role for indirect miHA presentation by donor APCs. Also, recipient nonhematopoietic PD-L1 would be expected to be dominant in models with CD8 cells because CD8 cells must make TCR:MHCI contact with nonhematopoietic cells to cause GVHD (19).

Surprisingly, genetic and depletion studies demonstrated that TS1-induced cardiac disease in the absence of PD-ligands required polyclonal, CD8+ non-TS1 T cells. Commensurately, infiltrating T cells were predominantly not TS1 clonotype, with ∼2-fold more CD8+ than CD4+ T cells. That CD4 depletion, which also profoundly reduced TS1 TEM progeny, did not diminish myocarditis suggests that once TS1 cells break tolerance, they are no longer required to maintain heart disease. Surprisingly, CD4 depletion, but not dual CD4 and CD8 depletion, accelerated weight loss and death. Anti-CD4 treatment likely depleted CD4+ Tregs, which may suppress autoimmune CD8 cells.

Of note, there was mild but significant heart disease in the absence of TS1 cells when donors were RAG2−/−PDL-L1/2−/− and recipients were RAG2+/+PD-L1/2−/−. These data further support the notion that PD-ligands suppress the development of autoreactive polyclonal T cells posttransplantation even without GVHD, and that this autoreactivity is reproducibly triggered by alloreactive T cells. It is tempting to speculate that alloimmune attack, as in the system we studied, is just one example of an inflammatory condition that would trigger autoimmunity when PD-ligand engagement is suboptimal. Other examples might include chronic infection or toxic injury, both settings that are associated with autoimmune induction in general.

How do PD-ligands regulate self-tolerance in the context of alloSCT? A major clue is that the absence of PD-L1/2 on the donor BM is sufficient for myocarditis. Therefore, parenchymal expression of PD-L1–suppressing effector cells, as invoked in PD-ligand–dependent tumor immune resistance settings (20), cannot be the only mechanism by which PD-ligands regulate autoimmunity. Instead, the absence of donor hematopoietic PD-L1/2 could lead to the peripheral activation of self-reactive polyclonal T cells, be they mature T cells that survived radiation, donor allograft T cells, or T cells derived from hematopoietic progenitors. In this connection, presentation of self-antigen by APCs is thought to be a natural means of maintaining tolerance (2123). In the absence of PD-ligands, such tolerogenic APCs could become stimulatory instead in a background of immune system activation by alloreactive CD4 cells.

The absence of PD-L1/2 on recipient nonhematopoietic tissues promoted even more extensive TS1 TEM-triggered myocarditis. PD-L1/2−/− thymic parenchyma may have been permissive for the selection of self-reactive T cells. PD-L1 has been demonstrated on cardiac endothelia (24, 25), and therefore PD-ligands could suppress both the generation of autoreactive CD8 cells and their effector activities. Resolving the mechanisms that account for TS1-induced autoimmunity would require both thymus and cardiac transplantation, which can be addressed in future work.

An unanswered question is why we observed only autoimmune heart disease and not autoimmune disease in other organs. One possibility is that, for unknown reasons, autoreactive CD8 cells were reactive only against cardiac-specific Ags. In contrast, autoreactive CD8 cells could recognize Ags with a broader tissue distribution. However, we might not have readily been able to detect their contribution to disease in tissues wherein TS1 already cause significant immunopathology. For example, in the absence of PD-ligands, but without TS1, there was low-level liver histopathology, which was not present when donor and recipient were RAG2−/−. Such mild changes could easily have been masked by HA-directed alloimmune damage.

The PD-1 pathway has not been implicated in protecting the heart in polyclonal models of GVHD. In most of these reports, impairment of the PD-1 pathway led to early death such that later onset myocarditis could not have been evaluated. It will be of interest to determine whether autoimmune myocarditis develops in the absence of PD-ligands in polyclonal GVHD models, which will require establishing conditions such that PD-1 blockade does not result in early lethality, but still intensifies GVHD.

Outside the GVHD literature, there is evidence that PD-1 and its ligands protect the heart against autoimmunity. PD-1−/− BALB/c mice develop fatal dilated cardiomyopathy, although this was found to be Ab mediated (26, 27), whereas disease in our model was caused by CD8+ T cells. PD-1– or PD-L1–deficient MRL-lpr mice developed cardiomyositis with a mixed inflammatory infiltrate along with anti-cardiac Abs (24, 28). PD-1 and PD-L1 also suppressed experimental myocarditis directed against a model Ag (25, 29), which differs substantially from the autoimmune disease we describe.

Recent reports describe rare cases of myocarditis with checkpoint inhibitor therapies, especially when CTLA-4 and PD-1 were combined (3032). PD-1 or PD-L1 inhibition is being considered in patients with relapsed hematologic malignancies after alloSCT and has been applied in a small number of patients, mostly with Hodgkin’s lymphoma relapsed after alloSCT (33, 34). Although prior mouse work clearly demonstrated the risk for increased GVHD with such therapy (1113, 35), our results raise additional concerns that autoimmune manifestations, including heart disease, could be precipitated.

When cardiac dysfunction occurs after alloSCT, it is frequently in the setting of complex medical problems that can include GVHD. Heart pathology is rarely examined, such that cardiac GVHD could be underappreciated. Nonetheless, cardiac GVHD has been reported. Diagnoses included heart block (as we observed), myocarditis, cardiomyopathy, and coronary artery disease, with instances of patients improving with immunosuppressants (36, 37).

Our results bear on the long-standing hypothesis that chronic GVHD has an autoimmune component (3841). Acute GVHD is the single greatest risk factor for chronic GVHD, which suggests that alloimmunity can beget autoimmunity. Our data directly demonstrate that this can occur and that it is important for pathogenesis. It is of a chronic nature in that it occurs only later in the course, after the initial wave of TS1-mediated allorecognition that leads to acute disease. Multiple nonexclusive mechanisms have been suggested for GVHD-associated autoimmunity in other contexts (4144). Our data suggest that if chronic GVHD is at least in part autoimmune, PD-ligands may be an additional mechanism that promotes self-tolerance. In our experiments, residual recipient T cells were sufficient to cause myocarditis. Even with ablative conditioning, a substantial number of patients have residual circulating host-derived T cells posttransplant (4551). Our results suggest that these T cells could be functionally autoimmune with insufficient PD-ligand engagement.

In any case, our results demonstrate a clear role for PD-ligands in regulating miHA-directed GVHD mediated by TEM and TN, and that this occurs at multiple levels including by donor APCs that indirectly present alloantigens. Most intriguingly, we also reveal a clear PD-ligand–dependent regulation of autoimmunity that is triggered by initial alloimmunity. This could explain chronicity of GVHD in some cases and in fact why chronic GVHD can be unremitting. Although we have used genetic knockouts to elicit strong phenotypes, differences in the strength of PD-ligand signals, perhaps genetically encoded, could underlie susceptibility to chronic GVHD, which itself could be partly caused by autoimmune induction downstream of alloimmune-mediated tissue damage.

We acknowledge the animal technicians and the flow cytometry and cell sorting facilities at Yale University School of Medicine and the University of Pittsburgh School of Medicine.

This work was supported by National Institutes of Health Grant R01 HL109599. This work benefited from the use of the special Becton Dickinson LSR Fortessa, funded by National Institutes of Health Grant 1S10OD011925-01.

The online version of this article contains supplemental material.

Abbreviations used in this article:

alloSCT

allogeneic hematopoietic stem cell transplantation

BM

bone marrow

GVHD

graft-versus-host disease

GVL

graft-versus-leukemia

HA

hemagglutinin

miHA

minor histocompatibility Ag

TEM

effector memory T cell

TKO

triple knockout

TM

memory T cell

TN

naive T cell

Treg

regulatory T cell

WT

wild type.

1
Anderson
,
B. E.
,
J.
McNiff
,
J.
Yan
,
H.
Doyle
,
M.
Mamula
,
M. J.
Shlomchik
,
W. D.
Shlomchik
.
2003
.
Memory CD4+ T cells do not induce graft-versus-host disease.
J. Clin. Invest.
112
:
101
108
.
2
Zheng
,
H.
,
C.
Matte-Martone
,
H.
Li
,
B. E.
Anderson
,
S.
Venketesan
,
H.
Sheng Tan
,
D.
Jain
,
J.
McNiff
,
W. D.
Shlomchik
.
2008
.
Effector memory CD4+ T cells mediate graft-versus-leukemia without inducing graft-versus-host disease.
Blood
111
:
2476
2484
.
3
Juchem
,
K. W.
,
B. E.
Anderson
,
C.
Zhang
,
J. M.
McNiff
,
A. J.
Demetris
,
D. L.
Farber
,
A. J.
Caton
,
W. D.
Shlomchik
,
M. J.
Shlomchik
.
2011
.
A repertoire-independent and cell-intrinsic defect in murine GVHD induction by effector memory T cells.
Blood
118
:
6209
6219
.
4
Chen
,
B. J.
,
X.
Cui
,
G. D.
Sempowski
,
C.
Liu
,
N. J.
Chao
.
2004
.
Transfer of allogeneic CD62L- memory T cells without graft-versus-host disease.
Blood
103
:
1534
1541
.
5
Chen
,
B. J.
,
D.
Deoliveira
,
X.
Cui
,
N. T.
Le
,
J.
Son
,
J. F.
Whitesides
,
N. J.
Chao
.
2007
.
Inability of memory T cells to induce graft-versus-host disease is a result of an abortive alloresponse.
Blood
109
:
3115
3123
.
6
Bleakley
,
M.
,
S.
Heimfeld
,
K. R.
Loeb
,
L. A.
Jones
,
C.
Chaney
,
S.
Seropian
,
T.
Gooley
,
F.
Sommermeyer
,
S. R.
Riddell
,
W. D.
Shlomchik
.
2015
.
Outcomes of acute leukemia patients transplanted with naïve T cells depleted stem cell grafts.
J. Clin. Invest.
125
:
2677
2689
.
7
Anderson
,
B. E.
,
P. A.
Taylor
,
J. M.
McNiff
,
D.
Jain
,
A. J.
Demetris
,
A.
Panoskaltsis-Mortari
,
A.
Ager
,
B. R.
Blazar
,
W. D.
Shlomchik
,
M. J.
Shlomchik
.
2008
.
Effects of donor T-cell trafficking and priming site on graft-versus-host disease induction by naive and memory phenotype CD4 T cells.
Blood
111
:
5242
5251
.
8
Shih
,
F. F.
,
D. M.
Cerasoli
,
A. J.
Caton
.
1997
.
A major T cell determinant from the influenza virus hemagglutinin (HA) can be a cryptic self peptide in HA transgenic mice.
Int. Immunol.
9
:
249
261
.
9
Kirberg
,
J.
,
A.
Baron
,
S.
Jakob
,
A.
Rolink
,
K.
Karjalainen
,
H.
von Boehmer
.
1994
.
Thymic selection of CD8+ single positive cells with a class II major histocompatibility complex-restricted receptor.
J. Exp. Med.
180
:
25
34
.
10
Caton
,
A. J.
,
J. R.
Swartzentruber
,
A. L.
Kuhl
,
S. R.
Carding
,
S. E.
Stark
.
1996
.
Activation and negative selection of functionally distinct subsets of antibody-secreting cells by influenza hemagglutinin as a viral and a neo-self antigen.
J. Exp. Med.
183
:
13
26
.
11
Blazar
,
B. R.
,
B. M.
Carreno
,
A.
Panoskaltsis-Mortari
,
L.
Carter
,
Y.
Iwai
,
H.
Yagita
,
H.
Nishimura
,
P. A.
Taylor
.
2003
.
Blockade of programmed death-1 engagement accelerates graft-versus-host disease lethality by an IFN-gamma-dependent mechanism.
J. Immunol.
171
:
1272
1277
.
12
Saha
,
A.
,
K.
Aoyama
,
P. A.
Taylor
,
B. H.
Koehn
,
R. G.
Veenstra
,
A.
Panoskaltsis-Mortari
,
D. H.
Munn
,
W. J.
Murphy
,
M.
Azuma
,
H.
Yagita
, et al
.
2013
.
Host programmed death ligand 1 is dominant over programmed death ligand 2 expression in regulating graft-versus-host disease lethality.
Blood
122
:
3062
3073
.
13
Deng
,
R.
,
K.
Cassady
,
X.
Li
,
S.
Yao
,
M.
Zhang
,
J.
Racine
,
J.
Lin
,
L.
Chen
,
D.
Zeng
.
2015
.
B7H1/CD80 interaction augments PD-1-dependent T cell apoptosis and ameliorates graft-versus-host disease.
J. Immunol.
194
:
560
574
.
14
Keir
,
M. E.
,
M. J.
Butte
,
G. J.
Freeman
,
A. H.
Sharpe
.
2008
.
PD-1 and its ligands in tolerance and immunity.
Annu. Rev. Immunol.
26
:
677
704
.
15
Wherry
,
E. J.
2011
.
T cell exhaustion.
Nat. Immunol.
12
:
492
499
.
16
Zhang
,
P.
,
J.
Wu
,
D.
Deoliveira
,
N. J.
Chao
,
B. J.
Chen
.
2012
.
Allospecific CD4(+) effector memory T cells do not induce graft-versus-host disease in mice.
Biol. Blood Marrow Transplant.
18
:
1488
1499
.
17
Markey
,
K. A.
,
T.
Banovic
,
R. D.
Kuns
,
S. D.
Olver
,
A. L.
Don
,
N. C.
Raffelt
,
Y. A.
Wilson
,
L. J.
Raggatt
,
A. R.
Pettit
,
J. S.
Bromberg
, et al
.
2009
.
Conventional dendritic cells are the critical donor APC presenting alloantigen after experimental bone marrow transplantation.
Blood
113
:
5644
5649
.
18
Anderson
,
B. E.
,
J. M.
McNiff
,
D.
Jain
,
B. R.
Blazar
,
W. D.
Shlomchik
,
M. J.
Shlomchik
.
2005
.
Distinct roles for donor- and host-derived antigen-presenting cells and costimulatory molecules in murine chronic graft-versus-host disease: requirements depend on target organ.
Blood
105
:
2227
2234
.
19
Matte-Martone
,
C.
,
J.
Liu
,
D.
Jain
,
J.
McNiff
,
W. D.
Shlomchik
.
2008
.
CD8+ but not CD4+ T cells require cognate interactions with target tissues to mediate GVHD across only minor H antigens, whereas both CD4+ and CD8+ T cells require direct leukemic contact to mediate GVL.
Blood
111
:
3884
3892
.
20
Chen
,
L.
,
X.
Han
.
2015
.
Anti-PD-1/PD-L1 therapy of human cancer: past, present, and future.
J. Clin. Invest.
125
:
3384
3391
.
21
Steinman
,
R. M.
,
D.
Hawiger
,
M. C.
Nussenzweig
.
2003
.
Tolerogenic dendritic cells.
Annu. Rev. Immunol.
21
:
685
711
.
22
Lechler
,
R.
,
W. F.
Ng
,
R. M.
Steinman
.
2001
.
Dendritic cells in transplantation--friend or foe?
Immunity
14
:
357
368
.
23
Hawiger
,
D.
,
K.
Inaba
,
Y.
Dorsett
,
M.
Guo
,
K.
Mahnke
,
M.
Rivera
,
J. V.
Ravetch
,
R. M.
Steinman
,
M. C.
Nussenzweig
.
2001
.
Dendritic cells induce peripheral T cell unresponsiveness under steady state conditions in vivo.
J. Exp. Med.
194
:
769
779
.
24
Lucas
,
J. A.
,
J.
Menke
,
W. A.
Rabacal
,
F. J.
Schoen
,
A. H.
Sharpe
,
V. R.
Kelley
.
2008
.
Programmed death ligand 1 regulates a critical checkpoint for autoimmune myocarditis and pneumonitis in MRL mice.
J. Immunol.
181
:
2513
2521
.
25
Grabie
,
N.
,
I.
Gotsman
,
R.
DaCosta
,
H.
Pang
,
G.
Stavrakis
,
M. J.
Butte
,
M. E.
Keir
,
G. J.
Freeman
,
A. H.
Sharpe
,
A. H.
Lichtman
.
2007
.
Endothelial programmed death-1 ligand 1 (PD-L1) regulates CD8+ T-cell mediated injury in the heart.
Circulation
116
:
2062
2071
.
26
Okazaki
,
T.
,
Y.
Tanaka
,
R.
Nishio
,
T.
Mitsuiye
,
A.
Mizoguchi
,
J.
Wang
,
M.
Ishida
,
H.
Hiai
,
A.
Matsumori
,
N.
Minato
,
T.
Honjo
.
2003
.
Autoantibodies against cardiac troponin I are responsible for dilated cardiomyopathy in PD-1-deficient mice.
Nat. Med.
9
:
1477
1483
.
27
Nishimura
,
H.
,
T.
Okazaki
,
Y.
Tanaka
,
K.
Nakatani
,
M.
Hara
,
A.
Matsumori
,
S.
Sasayama
,
A.
Mizoguchi
,
H.
Hiai
,
N.
Minato
,
T.
Honjo
.
2001
.
Autoimmune dilated cardiomyopathy in PD-1 receptor-deficient mice.
Science
291
:
319
322
.
28
Wang
,
J.
,
I. M.
Okazaki
,
T.
Yoshida
,
S.
Chikuma
,
Y.
Kato
,
F.
Nakaki
,
H.
Hiai
,
T.
Honjo
,
T.
Okazaki
.
2010
.
PD-1 deficiency results in the development of fatal myocarditis in MRL mice.
Int. Immunol.
22
:
443
452
.
29
Tarrio
,
M. L.
,
N.
Grabie
,
D. X.
Bu
,
A. H.
Sharpe
,
A. H.
Lichtman
.
2012
.
PD-1 protects against inflammation and myocyte damage in T cell-mediated myocarditis.
J. Immunol.
188
:
4876
4884
.
30
Heinzerling
,
L.
,
P. A.
Ott
,
F. S.
Hodi
,
A. N.
Husain
,
A.
Tajmir-Riahi
,
H.
Tawbi
,
M.
Pauschinger
,
T. F.
Gajewski
,
E. J.
Lipson
,
J. J.
Luke
.
2016
.
Cardiotoxicity associated with CTLA4 and PD1 blocking immunotherapy.
J. Immunother. Cancer
4
:
50
.
31
Läubli
,
H.
,
C.
Balmelli
,
M.
Bossard
,
O.
Pfister
,
K.
Glatz
,
A.
Zippelius
.
2015
.
Acute heart failure due to autoimmune myocarditis under pembrolizumab treatment for metastatic melanoma.
J. Immunother. Cancer
3
:
11
.
32
Johnson
,
D. B.
,
J. M.
Balko
,
M. L.
Compton
,
S.
Chalkias
,
J.
Gorham
,
Y.
Xu
,
M.
Hicks
,
I.
Puzanov
,
M. R.
Alexander
,
T. L.
Bloomer
, et al
.
2016
.
Fulminant myocarditis with combination immune checkpoint blockade.
N. Engl. J. Med.
375
:
1749
1755
.
33
Haverkos
,
B. M.
,
D.
Abbott
,
M.
Hamadani
,
P.
Armand
,
M. E.
Flowers
,
R.
Merryman
,
M.
Kamdar
,
A. S.
Kanate
,
A.
Saad
,
A.
Mehta
, et al
.
2017
.
PD-1 blockade for relapsed lymphoma post-allogeneic hematopoietic cell transplant: high response rate but frequent GVHD.
Blood
130
:
221
228
.
34
Herbaux
,
C.
,
J.
Gauthier
,
P.
Brice
,
E.
Drumez
,
L.
Ysebaert
,
H.
Doyen
,
L.
Fornecker
,
K.
Bouabdallah
,
G.
Manson
,
H.
Ghesquières
, et al
.
2017
.
Efficacy and tolerability of nivolumab after allogeneic transplantation for relapsed Hodgkin lymphoma.
Blood
129
:
2471
2478
.
35
Fujiwara
,
H.
,
Y.
Maeda
,
K.
Kobayashi
,
H.
Nishimori
,
K.
Matsuoka
,
N.
Fujii
,
E.
Kondo
,
T.
Tanaka
,
L.
Chen
,
M.
Azuma
, et al
.
2014
.
Programmed death-1 pathway in host tissues ameliorates Th17/Th1-mediated experimental chronic graft-versus-host disease.
J. Immunol.
193
:
2565
2573
.
36
Rackley
,
C.
,
K. R.
Schultz
,
F. D.
Goldman
,
K. W.
Chan
,
A.
Serrano
,
J. E.
Hulse
,
A. L.
Gilman
.
2005
.
Cardiac manifestations of graft-versus-host disease.
Biol. Blood Marrow Transplant.
11
:
773
780
.
37
Ahn
,
J. S.
,
S. H.
Cho
,
Y. K.
Kim
,
D. H.
Yang
,
W. K.
Bae
,
H. J.
Shim
,
J. J.
Lee
,
I. J.
Chung
,
S. W.
Juhng
,
H. J.
Kim
.
2009
.
Polymyositis and myocarditis after donor lymphocyte infusion.
Int. J. Hematol.
90
:
113
116
.
38
Parkman
,
R.
1986
.
Clonal analysis of murine graft-vs-host disease. I. Phenotypic and functional analysis of T lymphocyte clones.
J. Immunol.
136
:
3543
3548
.
39
Parkman
,
R.
1993
.
Is chronic graft versus host disease an autoimmune disease?
Curr. Opin. Immunol.
5
:
800
803
.
40
Chen
,
W.
,
R.
Gress
,
W. D.
Shlomchik
.
2009
.
Animal models of chronic graft-versus-host disese
. In
Chronic Graft-Versus-Host Disease.
G.
Vogelsang
,
S.
Pavletic
, eds.
Cambridge University Press
,
Cambridge, U.K.
, p.
31
45
.
41
Wu
,
T.
,
J. S.
Young
,
H.
Johnston
,
X.
Ni
,
R.
Deng
,
J.
Racine
,
M.
Wang
,
A.
Wang
,
I.
Todorov
,
J.
Wang
,
D.
Zeng
.
2013
.
Thymic damage, impaired negative selection, and development of chronic graft-versus-host disease caused by donor CD4+ and CD8+ T cells.
J. Immunol.
191
:
488
499
.
42
Zhang
,
Y.
,
E.
Hexner
,
D.
Frank
,
S. G.
Emerson
.
2007
.
CD4+ T cells generated de novo from donor hemopoietic stem cells mediate the evolution from acute to chronic graft-versus-host disease.
J. Immunol.
179
:
3305
3314
.
43
Tivol
,
E.
,
R.
Komorowski
,
W. R.
Drobyski
.
2005
.
Emergent autoimmunity in graft-versus-host disease.
Blood
105
:
4885
4891
.
44
Matsuoka
,
K.
,
H. T.
Kim
,
S.
McDonough
,
G.
Bascug
,
B.
Warshauer
,
J.
Koreth
,
C.
Cutler
,
V. T.
Ho
,
E. P.
Alyea
,
J. H.
Antin
, et al
.
2010
.
Altered regulatory T cell homeostasis in patients with CD4+ lymphopenia following allogeneic hematopoietic stem cell transplantation.
J. Clin. Invest.
120
:
1479
1493
.
45
Doney
,
K.
,
M.
Loken
,
E.
Bryant
,
A.
Smith
,
F.
Appelbaum
.
2008
.
Lack of utility of chimerism studies obtained 2-3 months after myeloablative hematopoietic cell transplantation for ALL.
Bone Marrow Transplant.
42
:
271
274
.
46
Mossallam
,
G. I.
,
A. M.
Kamel
,
B.
Storer
,
P. J.
Martin
.
2009
.
Prognostic utility of routine chimerism testing at 2 to 6 months after allogeneic hematopoietic cell transplantation.
Biol. Blood Marrow Transplant.
15
:
352
359
.
47
Saito
,
B.
,
T.
Fukuda
,
H.
Yokoyama
,
S.
Kurosawa
,
T.
Takahashi
,
S.
Fuji
,
N.
Takahashi
,
K.
Tajima
,
S. W.
Kim
,
S.
Mori
, et al
.
2008
.
Impact of T cell chimerism on clinical outcome in 117 patients who underwent allogeneic stem cell transplantation with a busulfan-containing reduced-intensity conditioning regimen.
Biol. Blood Marrow Transplant.
14
:
1148
1155
.
48
Lee
,
H. C.
,
R. M.
Saliba
,
G.
Rondon
,
J.
Chen
,
Y.
Charafeddine
,
L. J.
Medeiros
,
G.
Alatrash
,
B. S.
Andersson
,
U.
Popat
,
P.
Kebriaei
, et al
.
2015
.
Mixed T lymphocyte chimerism after allogeneic hematopoietic transplantation is predictive for relapse of acute myeloid leukemia and myelodysplastic syndromes.
Biol. Blood Marrow Transplant.
21
:
1948
1954
.
49
Reshef
,
R.
,
E. O.
Hexner
,
A. W.
Loren
,
N. V.
Frey
,
E. A.
Stadtmauer
,
S. M.
Luger
,
J. K.
Mangan
,
S. I.
Gill
,
P.
Vassilev
,
K. A.
Lafferty
, et al
.
2014
.
Early donor chimerism levels predict relapse and survival after allogeneic stem cell transplantation with reduced-intensity conditioning.
Biol. Blood Marrow Transplant.
20
:
1758
1766
.
50
Koreth
,
J.
,
H. T.
Kim
,
S.
Nikiforow
,
E. L.
Milford
,
P.
Armand
,
C.
Cutler
,
B.
Glotzbecker
,
V. T.
Ho
,
J. H.
Antin
,
R. J.
Soiffer
, et al
.
2014
.
Donor chimerism early after reduced-intensity conditioning hematopoietic stem cell transplantation predicts relapse and survival.
Biol. Blood Marrow Transplant.
20
:
1516
1521
.
51
Bleakley
,
M.
,
S.
Heimfeld
,
K. R.
Loeb
,
L. A.
Jones
,
C.
Chaney
,
S.
Seropian
,
T. A.
Gooley
,
F.
Sommermeyer
,
S. R.
Riddell
,
W. D.
Shlomchik
.
2015
.
Outcomes of acute leukemia patients transplanted with naive T cell-depleted stem cell grafts.
J. Clin. Invest.
125
:
2677
2689
.

The authors have no financial conflicts of interest.

Supplementary data