SEARCH

SEARCH BY CITATION

Keywords:

  • Alloreactive T cells;
  • chimerism;
  • graft-versus-host disease (GVHD);
  • liver transplantation;
  • tolerance

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References
  9. Supporting Information

Acute graft-versus-host disease (aGVHD) is a life-threatening complication after solid-organ transplantation, which is mediated by host-reactive donor T cells emigrating from the allograft. We report on two liver transplant recipients who developed an almost complete donor chimerism in peripheral blood and bone marrow-infiltrating T cells during aGVHD. By analyzing these T cells directly ex vivo, we found that they died by apoptosis over time without evidence of rejection by host T cells. The host-versus-donor reactivity was selectively impaired, as anti-third-party and antiviral T cells were still detectable in the host repertoire. These findings support the acquired donor-specific allotolerance concept previously established in animal transplantation studies. We also observed that the resolution of aGVHD was not accompanied by an expansion of circulating immunosuppressive CD4/CD25/FoxP3-positive T cells. In fact, graft-versus-host-reactive T cells were controlled by an alternative negative regulatory pathway, executed by the programmed death (PD)-1 receptor and its ligand PD-L1. We found high PD-1 expression on donor CD4 and CD8 T cells. In addition, blocking PD-L1 on host-derived cells significantly enhanced alloreactivity by CD8 T cells in vitro. We suggest the interference with the PD-1/PD-L1 pathway as a therapeutic strategy to control graft-versus-host-reactive T cells in allograft recipients.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References
  9. Supporting Information

Acute graft-versus-host disease (aGVHD) is a rare but serious complication occurring early after solid-organ transplantation (1–4). Most cases have been documented in patients undergoing orthotopic liver transplantation (LTx) (5). The disease usually presents with fever and skin rash, and subsequently proceeds to diarrhea and pancytopenia. Because of the lack of established therapies, current treatment approaches comprise the intensification as well as the reduction of immunosuppressive medication (6). More than 85% of affected patients die, mostly from infectious complications.

The pathophysiology of aGVHD after LTx resembles that after allogeneic hematopoietic stem-cell transplantation where the pivotal role of donor-derived T cells for the induction of alloreactivity is generally accepted (7). Accordingly, LTx-associated aGVHD is mediated by donor T cells that are introduced by the graft in fairly high numbers estimated between 109–1010 cells (8,9). These T cells are detectable in peripheral blood and organs of patients during the first weeks after LTx (6,8,10,11). On the other hand, the transfer of donor T cells into the recipient appears to promote donor-specific allograft tolerance (12,13). The exact mechanism for this phenomenon has not been elucidated. Several pathways have been proposed including alloreactive T cells to undergo clonal exhaustion and deletion, anergy or active suppression by naturally occurring CD4/CD25/FoxP3-positive regulatory T (Treg) cells (14). Taken together, graft-derived T cells influence the delicate balance between tolerogenicity and alloreactivity after LTx. However, whether these T cells will facilitate graft acceptance or induce deleterious aGVHD is not predictable for the individual transplant recipient.

We herein report on two patients with severe LTx-associated aGVHD who developed a transient donor T-cell chimerism of up to 100%, which is far above that of previously described cases (8). This enabled us to sort donor-derived T cells in sufficient numbers to allow for a detailed analysis of phenotypic and functional features ex vivo. We observed that the invading donor T cells died by apoptosis over time without evidence of rejection by host T cells. We also did not detect any expansion of circulating Treg cells after both patients recovered from GVHD. In fact, our data suggest that alloreactive donor T cells were controlled by an alternative negative regulatory pathway, executed by the programmed death-1 (PD-1) receptor and its major ligand PD-L1.

The PD-1 molecule has originally been described in exhausted T cells prone to apoptosis (15) and belongs to the CD28/CTLA-4 family of immunoreceptors. It is expressed on T, B and myeloid cells upon activation (16). Since the first observations of spontaneous autoimmune diseases in knockout mice lacking PD-1 or PD-L1 expression (17,18), the PD-1/PD-L1 pathway has been postulated to be crucial for the induction and maintenance of peripheral tolerance. We believe that interfering with this pathway might be a novel therapeutic strategy for organ transplant-associated aGVHD, in which early induction of the ‘exhaustion phenotype’ in graft-derived T cells potentially mitigates the course of disease.

Materials and Methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References
  9. Supporting Information

Patients

Patients provided written informed consent for participating in this study. The experimental part was approved by the local institutional review board. The 64-year-old male patient M-1 with alcoholic cirrhosis received a liver allograft from a 58-year-old male. HLA-class-I types of patient M-1 and donor were A*0101/A*0201,B8/B44,Cw7, and A*0201/A*0301, B5/B51, Bw4, respectively. On postoperative day (POD) 26, patient M-1 was admitted with fever, skin rash and progressive pancytopenia. The clinical diagnosis of aGVHD grade 3 (19) was confirmed by skin biopsy (Figure 1A). Immunosuppressive medication was switched from cyclosporine to tacrolimus and prednisolone (2 mg/kg) was added (Figure 2A). Rapidly thereafter, fever and skin rash disappeared. White blood cell (WBC) counts gradually increased, whereas thrombocytopenia and lymphopenia persisted (Figure 2B,C). Subsequently, patient M-1 developed multiple opportunistic infections (Figure 2A). Pathogen-adapted antimicrobial treatment remained ineffective and he died from septic multiorgan failure on POD 159.

image

Figure 1. GVHD and bone marrow aplasia after LTx in patient M-1. (A) The patient was admitted on POD 26 with fever and generalized skin rash. Skin biopsy on POD 30 showed epidermolysis and apoptotic keratinocytes, confirming acute GVHD. By immunohistochemistry, CD8 T cells (×400) secreting the cytotoxic molecule Tia1 (×1000; insert) were detected at the basal epidermal layer. (B) Bone marrow smears on POD 39 showed severe aplasia with scattered lymphocytic infiltrates in standard Pappenheim's staining (×50; left panel). These infiltrates consisted mainly of CD8 T cells of donor origin, as demonstrated by binding the HLA-A3-specific MAb GAP-A3 in flow cytometry (right panel). For microscopy, DMR microscope equipped with DFC camera and IM500 V4.0 software was used (Leica, Wetzlar, Germany).

Download figure to PowerPoint

image

Figure 2. Clinical course, blood counts and lineage-specific chimerism in patient M-1. (A) After skin GVHD and bone marrow aplasia were diagnosed, the patient was treated with high-dose steroids for 2 weeks. This was followed by resolution of skin rash and gradual increase of WBC. Although steroid treatment was stopped, the patient developed several opportunistic infections, finally leading to death by septic multiorgan failure (MOF). (B) WBC and platelet counts in peripheral blood. (C) T- and B-cell counts in peripheral blood (n.d. = not determined). (D) Donor chimerism results for CD4 T cells, CD8 T cells, CD19 B cells and CD15 granulocytes, all selected from peripheral blood by immunomagnetic beads and subsequently analyzed by STR assay.

Download figure to PowerPoint

The clinical course of the second patient has been previously described (20). Briefly, the 67-year-old female patient M-2 developed cutaneous aGVHD 2 months after LTx that she received because of hepatitis B virus-induced hepatocellular carcinoma. GVHD was refractory to tacrolimus, high-dose prednisolone, and the monoclonal antibody (MAb) basiliximab. Because of progression to grade 4 aGVHD, immunosuppressive therapy was escalated by MAb alemtuzumab. Subsequently, patient M-2 was infused twice with ex vivo activated and selected host T cells on POD 112 and 163. This treatment sequence was followed by resolution of GVHD and conversion from mixed to full host chimerism in peripheral blood. Patient M-2 is now 5 years after LTx and shows excellent graft function.

Immunohistochemistry

Skin-infiltrating cells were stained using MAb C8/144B (DAKO, Hamburg, Germany) for CD8 and MAb 26gA10F5 (Immunotech, Marseille, France) for cytotoxic molecule Tia1, respectively. PD-L1 expression was analyzed using MAb 27A2 (MB Laboratories, Nagoya, Japan). Immunohistological labeling was performed on paraffin-embedded tissue sections (21).

Mononuclear cells

Mononuclear cells isolated from patients and healthy donors by standard Ficoll density gradient centrifugation were always frozen and stored in liquid nitrogen for at least 1 month before use in flow cytometry and ELISPOT assays. Freezing and thawing procedures were standardized. Data shown in Figures 3–7 were obtained from samples with a cell viability of 80% or higher upon thawing as determined by trypan blue exclusion.

image

Figure 3. Donor-derived T cells express TCRαβ and are predominantly of memory phenotype. Multicolor flow cytometry was performed on bone marrow-derived mononuclear cells of patient M-1 from POD 39. Donor and host origin of T cells was determined by staining with MAb GAP-A3 that recognizes the donor-specific HLA-A3 allele. T cells were gated according to expression of CD4, CD8 and HLA-A3, and were then analyzed for expression of other cell surface markers. Propidium iodide (PI) and isotype-matched control antibodies were used to exclude dead cells and nonspecific staining, respectively (for details see M&M). Dot blots show representative results obtained with viable CD8 T cells of donor origin. Numbers indicate percentages of positive cells in the donor CD8 T-cell subpopulation.

Download figure to PowerPoint

image

Figure 4. Donor-derived T cells are not rejected by host T cells. Host CD3 T cells were isolated from peripheral blood of patient M-1 on POD 129 and were seeded at 105/well in IFN-γ ELISPOT plates. Target cells (5×104/well) were donor-derived CD3 T cells sorted at POD 53, K562 cells transfected with the donor-specific HLA-A*0301 mismatch allele, and CD3 T cells derived from a third-party individual not matched with the allograft donor for any HLA allele. Host T cells were also stimulated with PHA and with K562/HLA-A*0201 transfectant cells (5×104/well) preloaded with HLA-A*0201-binding peptide epitopes of either CMV pp65, EBV BMLF1, or HIV pol origin. HLA-A types of patient M-1 and the allograft donor were A*0101/A*0201 and A*0201/A*0301, respectively. After incubation over 20h, IFN-γ spots were developed and subsequently evaluated using an automated ELISPOT reading system (25). Results are means of data from triplicate wells and are representative of three independent experiments. The difference between the frequency of anti-third-party T cells and the frequency of antidonor T cells is statistically significant (p-value < 0.05) as determined by student's t-test.

Download figure to PowerPoint

image

Figure 5. Lack of Treg cell expansion early after LTx. By staining for CD25 and intracellular FoxP3 in flow cytometry, only very low numbers of Treg cells were detected in peripheral blood CD4 T cells during acute GVHD and early after the relief of symptoms in patients M-1 and M-2. In contrast, more than 10% of CD4 T cells of patient M-1 had a Treg phenotype during the ongoing opportunistic infections on POD 139. We determined the frequency of CD4/CD25/FoxP3pos Treg cells in peripheral blood of five healthy control persons to range between 4.1–8.7% of total CD4 T cells using the same approach (data not shown). All measurements were performed with previously frozen PBMC that had a cell viability exceeding 80% according to trypan blue exclusion. Dead cells and nonspecific staining were excluded as described in Figure 3.

Download figure to PowerPoint

image

Figure 6. Strong PD-1 expression on T cells of donor origin. (A) PBMC samples isolated from LTx patients M-1 and M-2 at indicated PODs were analyzed for PD-1 expressing CD4 and CD8 T cells by multicolor flow cytometry. For both patients, simultaneous staining with the HLA-A3-specific MAb GAP-A3 allowed to detect T cells with donor (HLA-A3pos) and host (HLA-A3neg) origin, respectively. (B) PD-1 expression was also determined on T cells of six different healthy volunteers using the same procedure and was found on 2–28% of total CD4 or CD8 T cells. Representative results of healthy donor HD-1 are shown. All analyses were performed with previously frozen PBMC showing a cell viability higher than 80% after thawing. Dead cells and nonspecific staining were excluded as described in Figure 3. Numbers in dot blots indicate the percentages of PD-1pos cells among CD4 and CD8 T cells of donor and host origin, respectively. n.a. = not applicable.

Download figure to PowerPoint

image

Figure 7. Alloreactivity is mediated by donor-derived CD8 T cells expressing PD-1. (A) Donor CD8 T cells derived from peripheral blood of patient M-1 at POD 53 were sorted according to PD-1 expression by FACS. PD-1pos and PD-1neg cells were weekly stimulated with irradiated host DC in vitro. Responding T-cell populations were tested for alloreactivity to host DC on day 18 of cultures. Data are from IFN-γ ELISPOT wells each seeded with 104 T cells and 2×104 DC over 20 h. The PD-1pos T-cell responders showed clear IFN-γ production upon DC stimulation (control), whereas reactivity to DC was not observed in PD-1neg-derived T cells. Allo-DC reactivity was blocked by anti-HLA-class I MAb W6/32 (100 μg/mL), but not by anti-HLA-class II MAb L243 (100 μg/mL). Blocking PD-L1 ligand on DC with MAb MIH1 (10 μg/mL) resulted in a significant increase of alloreactivity, not obtained with isotype-matched control IgG1. (B) CD8 T cells derived from patient M-2 on POD 85 were stained for PD-1 and the donor-specific HLA-A3 allele. PD-1pos HLA-A3pos cells were sorted by FACS and stimulated with host DC as described for patient M-1. Again, HLA-class I restricted alloreactivity to host DC was detected in PD-1pos T-cell responders seeded at 104 per IFN-γ ELISPOT well. This alloreactivity specifically increased after blocking PD-L1. Results are representative of three separate experiments. n.d. = not determined.

Download figure to PowerPoint

Flow cytometry analysis

After thawing, cells were immediately stained over 15 min at 6°C with fluorochrome-labeled MAb CD3-PC5, CD4-PC5, CD8-PC5, CD25-PE, CD27-PE, CD28-FITC, CD45RA-PE, CD45RO-FITC, CD62L-PE, CTLA-4-APC, TCRαβ-PE (all Beckman Coulter, Miami, FL), CD69-PE, PD-1-FITC (clone MIH4, IgG1) and Annexin-V-FITC (all BD Biosciences, San Diego, CA). HLA-A3 expression was determined by HLA-A3-specific MAb GAP-A3 (22) and a PE-conjugated secondary Ab. Intracellular FoxP3 was analyzed with a FoxP3 staining kit (eBiosciences, San Diego, CA). We regularly used (i) propidium iodide (PI) to exclude dead cells and (ii) irrelevant isotype-matched control antibodies to exclude nonspecific staining. After gating on PI-negative viable cells, 1–3×104 events per MAb were measured on FACSCanto flow cytometer (BD Biosciences). Using this strategy, we assured that only viable cells specifically binding the fluorochrome-conjugated reagent are included in each analysis and that data from different samples can be compared.

Isolation of leukocyte subsets and chimerism analysis

Purification of leukocyte subsets from peripheral blood was performed using whole blood microbeads and the autoMACS device (Miltenyi Biotec, Bergisch-Gladbach, Germany). The degree of donor-host chimerism was determined by genomic short-tandem repeat (STR) analysis (23).

IFN-γ ELISPOT assay

T cells selected from cryopreserved PBMC using magnetic CD3, CD4 or CD8 microbeads (Miltenyi Biotec) were seeded at 105/well in IFN-γ ELISPOT plates (24). Target cells were added at 5×104/well. After incubation over 20 h, spots were developed and evaluated as described (25). ELISPOT results are means of duplicate or triplicate wells, respectively. The generation of K562/HLA transfectants and loading with peptide epitopes have been previously reported (25).

Sorting, expansion and functional analysis of donor CD8 T cells ex vivo

Donor-derived CD8 T cells were isolated by fluorescence-activated cell sorting (FACS) on flow cytometer FACSAria (BD Biosciences). Donor origin was verified using MAb to HLA-A3 that was exclusively expressed by allograft donors, but not by transplant recipients M-1 and M-2 (20). Sorted T cells were stimulated at a 10:1 ratio with irradiated (35Gy) allogeneic dendritic cells (DC) generated from patients PBMC (26) after donor cells had completely disappeared from circulation. The stimulation was performed in round-bottom 96 well plates in AIM-V medium (GIBCO-BRL, Grand Island, NY) supplemented with 10% human serum, 5 ng/mL IL-7 (R&D Systems, Minneapolis, MN) and 50 IU/mL IL-2 (Chiron, Emeryville, CA). After restimulations with irradiated DC on days 7 and 14, responder cells were tested in IFN-γ ELISPOT assay for alloreactivity to host-derived DC and phytohemagglutinin (PHA)-activated lymphoblasts. For Ab blocking tests, target cells were preincubated for 30 min with 10–100 μg/mL MAb W6/32, an anti-HLA-class-I IgG2a (27), L243, an anti-HLA-DR IgG2a (28) and MIH1, an anti-PD-L1 IgG1 (eBiosciences).

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References
  9. Supporting Information

Donor-derived T cells flood liver allograft recipients

Patient M-1 developed skin aGVHD grade 3 and complete bone marrow aplasia between POD 26 and 35 after orthotopic LTx (Figures 1 and 2). To prove the presence of graft-derived hematopoietic cells, we isolated granulocytes (CD15pos) as well as T (CD4pos, CD8pos) and B (CD19pos) lymphocytes from peripheral blood on POD 35 and performed STR-based chimerism analysis. While purified granulocytes were mainly of host origin (93%), the lymphocytes showed mixed chimerism with 49%, 53% and 79% donor-derived CD4, CD19 and CD8 cells, respectively (Figure 2D). The percentages of donor lymphocytes even increased in the following days leading to complete donor CD8 T-cell chimerism on POD 53. Graft-derived lymphocytes then dropped gradually and disappeared in peripheral circulation until POD 115. These observations suggested that alloreactive donor T cells originally emigrating from the liver graft underwent strong in vivo expansion resulting in severe skin GVHD and bone marrow aplasia in patient M-1.

To investigate the direct involvement of graft-derived T cells in the destruction of host hematopoiesis, we sought for bone marrow-infiltrating donor T cells in patient M-1. We chose the bone marrow aspirate from POD 39 because peripheral pancytopenia was most pronounced at this time point (Figure 2B). Since the allograft donor and recipient were HLA-A3pos and HLA-A3neg, respectively, chimerism was determined using an HLA-A3-specific MAb in flow cytometry. We found that 80% of bone marrow-derived CD8 T cells were of donor origin, suggesting their causative role in the development of myeloaplasia (Figure 1B). We observed a similar preponderance of donor cells also in bone marrow-derived CD4 T cells (data not shown).

The second patient M-2 with LTx-associated aGVHD had a donor T-cell chimerism of 53% in peripheral blood on POD 85, when life-threatening grade 4 GVHD occurred (20). Although GVHD started to resolve after treatment with the lympho-depleting MAb alemtuzumab, the donor proportion in residual T cells increased to more than 95% until POD 110.

Donor-derived T cells are of T-cell receptor (TCR)αβ memory phenotype

The comparably long and intense coexistence of donor and host T cells in patient M-1 gave us the unique opportunity to precisely define their phenotype ex vivo. As a clinical sample enriched with functionally active alloreactive donor T cells, we chose the bone marrow aspirate from POD 39, when peak pancytopenia after LTx was observed (Figure 2B). We also included PBMC of POD 53, which contained the majority of donor T cells in peripheral circulation. We observed that most T cells showed a CD45ROpos CD62Lneg memory phenotype (Figure 3), regardless of their origin (i.e. donor or host) and subset (i.e. CD4 or CD8). All of them expressed TCRαβ, with preferential expansion of single Vβ chain families according to Ab-based typing (data not shown). Substantial numbers of T cells were positive for tumor-necrosis factor receptor (TNFR) family member CD27 and costimulatory molecule CD28, but were negative for immunoinhibitory receptor CTLA-4. They strongly expressed TNFR family member CD95, whereas staining intensity for activation markers CD25 (Figure 3) and CD69 (data not shown) was dim.

Host-versus-donor T-cell reactivity is selectively impaired

Donor lymphocytes completely disappeared in peripheral blood of patient M-1 until POD 115 (Figure 2D). We wondered if these cells were effectively rejected by host T cells in vivo, or if they had died spontaneously. Thus, we isolated peripheral blood T cells from patient M-1 on POD 129 and analyzed them for alloreactivity toward purified donor T cells from POD 53. We did not detect any host-versus-donor reactivity, neither by using host T cells ex vivo (Figure 4) nor after repetitive in vitro stimulations with irradiated donor T cells (data not shown). In contrast to the antidonor unresponsiveness, host T cells showed significant recognition of allogeneic third-party T cells as well as CMV and EBV peptide epitopes (Figure 4). These results were confirmed with host T cells of patient M-1 from POD 121. Similar findings were also obtained in patient M-2 who failed to demonstrate any antidonor lymphocyte reactivity in freshly isolated host T cells ex vivo (20). Hence it appeared very unlikely that graft-derived lymphocytes were effectively rejected by donor-reactive host T cells, even though these lymphocytes were mismatched at almost all HLA alleles.

To further explore the fate of donor-derived T cells, we analyzed them for expression of the early marker of apoptosis, Annexin-V. Flow cytometry analysis on PBMC of patient M-1 from POD 53 showed that the vast majority of donor CD8 and CD4 T cells were positive for Annexin-V, indicating that these T cells were destined to undergo apoptosis in vivo (Figure S1). However, only few donor T cells were already dead, as demonstrated by less than 10% of Annexin-Vpos cells incorporating the necrosis marker PI (data not shown).

Functional inactivation of alloreactive T cells is not accompanied by Treg cell expansionin vivo

We observed a surprising discrepancy between the strong and even increasing donor chimerism in the T-cell lineage and the already mitigated clinical signs of aGVHD and bone marrow aplasia beyond POD 40 in patient M-1 (Figure 2). To investigate a potential impact of immunosuppressive regulatory T cells on host-reactive donor T cells in vivo, the frequency of CD4/CD25/FoxP3pos Treg cells was measured in peripheral blood over time. Interestingly, Treg cells were detected at subnormal numbers during active GVHD and aplasia, and did not increase substantially after patient M-1 recovered (Figure 5). Similar Treg cell data were obtained from patient M-2, who had skin aGVHD on POD 85 and showed remission of GVHD on POD 110 (Figure 5) (20). These results indicated that the persistence of alloreactive donor lymphocytes after resolution of GVHD was not accompanied by an expansion of the circulating Treg cell pool in vivo.

Donor-derived T cells express high levels of PD-1

We searched for an alternative negative regulatory mechanism, which may be capable of inducing peripheral tolerance in alloreactive T cells. We found that the immunoinhibitory receptor PD-1 was expressed by almost all donor-derived CD4 and CD8 T cells in patients M-1 and M-2, respectively (Figure 6A). The intensity of PD-1 staining seemed to be somewhat lower on host-derived CD4 and CD8 T cells during aGVHD. However, these levels still exceeded those that we observed on T cells of healthy control persons (Figure 6B). Along with the resolution of GVHD, donor-derived T cells gradually disappeared in circulation.

The PD-1/PD-L1 pathway regulates alloreactive donor CD8 T cells

To analyze the functional impact of PD-1 on alloreactivity, we sorted donor CD8 T cells from bone marrow (POD 39) and peripheral blood (POD 53) of patient M-1 according to PD-1 expression by FACS. After in vitro stimulation with host DC for 2 weeks, PD-1pos and PD-1neg T-cell populations expanded 2.7-fold (1.9–3.2) and 3.1-fold (2.3–3.7), respectively. Responding T cells were tested by IFN-γ ELISPOT assay for alloreactivity using host-derived DC and PHA blasts as targets. Both latter cell types expressed PD-L1 in flow cytometry. We detected CD8 T cells with alloreactivity to host DC (Figure 7A) and PHA blasts (data not shown) only in cultures derived from the PD-1pos fraction. In contrast, the PD-1neg cultures did not contain antihost reacting CD8 T cells. The alloreactivity was completely blocked by an anti-HLA-class-I MAb, indicating that the PD-1pos cell fraction contained HLA-class I restricted alloreactive CD8 T cells. We also observed that antibody blockade of PD-L1 on target cells significantly enhanced alloreactivity (Figure 7A). Overall, the results obtained with PD-1-sorted donor CD8 T cells from bone marrow and peripheral blood were very similar.

We repeated this experiment using PD-1pos donor CD4 T cells sorted from POD 53 PBMC of patient M-1 as responders. However, neither proliferation nor alloreactivity was observed after repeated in vitro stimulations with host DC (data not shown). It remains unclear whether this reflects an intrinsic characteristic of PD-1pos donor CD4 T cells or suboptimal culture conditions used for their expansion. We also sorted PD-1pos and PD-1neg host CD8 T cells from POD 39 bone marrow cells of patient M-1 and stimulated them with irradiated donor T cells of POD 53 in vitro. T cells did not proliferate in this setting, confirming the lack of any host-versus-donor reactivity observed in PBMC of POD 129 (Figure 4).

We finally performed the same experiments with CD8 T cells isolated from patient M-2 on POD 85 during active GVHD. Again, cultures derived from the sorted PD-1pos donor fraction showed HLA-class-I restricted antihost DC reactivity (Figure 7B). This alloreactivity significantly increased by PD-L1 blockade, suggesting an ongoing PD-1/PD-L1-mediated suppression. Although sorted PD-1neg donor CD8 T cells expanded similarly well compared to the PD-1pos counterparts, the starting populations were too small to reach cell numbers sufficient for testing. We concluded from these functional studies that most, if not all alloreactive donor CD8 T cells expressed PD-1 and were negatively regulated by the PD-1/PD-L1 pathway.

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References
  9. Supporting Information

The mortality rate of patients with LTx-associated aGVHD is unacceptably high (29). To develop more effective therapeutic strategies, a better understanding of the underlying alloreactivity and mechanisms of its immune regulation is clearly needed. We herein describe two patients with LTx-associated aGVHD, in whom allograft-derived T cells constituted up to 100% of the total T-cell pool in peripheral blood and bone marrow. Such an extensive conversion to complete donor T-cell chimerism after solid-organ transplantation has not been described before. We used this unique opportunity to demonstrate by direct ex vivo analyses that graft T cells (i) are TCRαβ T cells (and not liver-resident TCRγδ or natural killer T cells) (9) of memory phenotype, (ii) are not rejected by host-derived T cells once severe aGVHD occurs, presumably due to the previous clearance of donor-reactive host T cells and (iii) can persist at high and even increasing frequencies for several weeks despite remission of GVHD.

Graft-derived T cells are regularly detectable early after LTx, but usually disappear within 1–3 weeks (30). A previous prospective study in 49 LTx patients measured the mean and maximum levels of donor T-cell chimerism in peripheral blood at 5% and 11%, respectively (8). Only two cases with GVHD exceeded this range and showed a striking predominance of donor CD8 T cells. Both patients described herein had a peak T-cell chimerism level >95%, with a preponderance of CD8 (patient M-1) or CD4 T cells (patient M-2), respectively. Later on, graft-derived T cells gradually underwent cell death via apoptosis.

We observed that the invading donor T cells were not rejected by alloreactive host T cells. The host-versus-donor reactivity was selectively impaired, as anti-third-party and antiviral T cells were still detectable in the host repertoire (Figure 4). The substantial reduction or even clearance of antidonor reactivity is supported by findings in a third patient M-3, where we have also been unable to generate donor-reactive host T cells after LTx-associated aGVHD (M.T., unpublished observation), and by the fact that we required extensive in vitro restimulations to generate donor-reactive host T cells in patient M-2 (20). Very recently, donor-specific unresponsiveness has also been observed in T cells from recipients of combined allogeneic bone marrow and kidney transplants and even allowed to discontinue maintenance immunosuppressive therapy (31).

A major mechanism of the immune system to induce allograft tolerance is active immunosuppression by CD4/CD25/FoxP3pos Treg cells (32). It has been recently demonstrated that allosuppressive Treg cells detach from the liver graft and circulate in the recipients at substantial numbers, thereby possibly contributing to chimerism-associated tolerance early after LTx (33). We thus assumed a potential role for Treg cells in the functional inactivation of GVHD-inducing graft T cells. However, CD4/CD25/FoxP3pos Treg cells occurred at lower frequencies in both patients than usually observed in healthy individuals (34), making it very unlikely that they contributed to the silencing of GVHD-mediating T cells in vivo. On the other hand, this result fits with the possibility that the low Treg cell numbers early after LTx were unable to prevent alloreactive graft T cells from inducing GVHD.

We found the immunoinhibitory receptor PD-1 at high expression level on graft-derived CD4 and CD8 T cells. The PD-1 up-regulation was not linked to increased expression of CD25 and CD69, arguing against general T-cell activation as the sole reason for PD-1 over-expression. Recently, up-regulation of PD-1 on antiviral CD8 T cells has been described as a marker of T-cell anergy or exhaustion and correlated with disease progression and viral load (35,36). Most interestingly, blocking PD-1 from interaction with PD-L1 ligand reverted the functional impairment of PD-1 expressing T cells in models of LCMV (37), HIV (35,36,38) and HBV (39) infection. The blockade of PD-L1 has also been described as a means to augment human tumor-specific T-cell responses in vitro (40).

Similar to these reports, we here demonstrate for the first time that human HLA-class-I restricted alloreactive CD8 T cells express PD-1 and are functionally activated through blocking PD-L1 on recipient cells. Hence, even though these cells are exhausted and presumably committed to apoptotic cell death they can still be regulated by targeting PD-1/PD-L1. Several transplantation studies in animals already support this concept by showing that ligation of PD-1 can decrease allograft rejection (41) and GVHD lethality (42), and that expression of PD-L1 contributes critically to the protection of tissues from GVHD (43–45).

By using immunohistochemistry analysis of paraffin-embedded skin biopsies, we detected weak PD-L1 staining on epidermal keratinocytes of both patients during active GVHD comparable to that of noninflamed healthy control samples (E.v.S., unpublished finding). This observation might indicate for the human situation that alloreactive PD- 1pos T cells cause aGVHD due to insufficient PD-L1 expression in the skin during the early attack. Proinflammatory cytokines such as IFN-γ and TNF-α are known to upregulate PD-L1 on various hematopoietic and nonhematopoietic cells (16,40,46) including keratinocytes (47–49). It is conceivable that the inflammatory milieu during aGVHD resulted in a secondary increase of PD-L1 expression on keratinocytes, finally contributing to the functional inactivation of alloreactive PD-1pos T cells. As an alternative explanation, the aGVHD reaction might have been self-limiting by significant PD-L1 expression among PD-1pos skin-infiltrating T cells themselves. However, these hypotheses can only be verified by a time-dependent analysis of PD-L1 expression and donor cell apoptosis in skin biopsies taken repeatedly during the course of aGVHD. Appropriate skin samples were not available in both patients.

In conclusion, our results suggest that graft-derived donor T cells mediating LTx-associated aGVHD selectively eliminate donor-reactive T cells from the host T-cell repertoire, thereby preventing donor cells from subsequent rejection. We also demonstrate that alloreactive graft T cells express high levels of PD-1 and can be negatively regulated by the PD-1/PD-L1 pathway. Further studies should now investigate whether targeting PD-1 by PD-L1 constructs can turn off alloreactive T cells in vitro and may potentially ameliorate the course of aGVHD in vivo (46). Considering the substantial PD-1 expression observed on host-derived T cells, the selective inactivation of graft T cells might be accomplished by fusing the PD-L1 protein to antibodies recognizing donor-specific HLA alleles. We would be also curious to see if the PD-1 pathway is similarly involved in aGVHD after allogeneic transplantation of other solid organs (2–4) or hematopoietic stem cells (50), respectively. In addition, graft rejections by alloreactive host T cells might be a further field, in which the PD-1 pathway and its potential applicability for therapeutic interventions should be investigated.

Acknowledgments

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References
  9. Supporting Information

We thank Dr. J. Wenzel (University of Bonn, Germany) for Tia1 immunohistochemistry. The study was supported by grants SFB432/A13 and KFO183/TP5 from the Deutsche Forschungsgemeinschaft (DFG) to W.H. Marcus Schuchmann and Ralf G. Meyer have contributed equally to this work.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References
  9. Supporting Information

Supporting Information

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References
  9. Supporting Information

Figure S1: Donor-derived T cells are destined to undergo apoptosis. (A) Flow cytometry was used to analyze peripheral blood T cells of patient M-1 from POD 53 ex vivo for Annexin-V expression. At that time point, the donor proportion in CD8 and CD4 T cells was >95% and >50%, respectively. Numbers indicate percentages of Annexin-Vpos cells in total CD8 or CD4 T cells. Data are representative of three separate experiments. (B) PBMC of the healthy control person HD-1 were stained using the same procedure. All measurements were performed with previously frozen PBMC that had a cell viability exceeding 80% according to trypan blue exclusion. Dead cells and nonspecific staining were excluded from analysis using propidium iodide and isotype-matched control antibodies.

Please note: Wiley-Blackwell are not responsible for the content or functionality of any supporting materials supplied by the authors. Any query (other than missing material) should be directed to be the corresponding author for the article.

FilenameFormatSizeDescription
AJT_2401_sm_FigureS1.ppt99KSupporting info item

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.