INSERM, U823, Oncogenic Pathways in the Haematological Malignancies, Institut Albert Bonniot, Grenoble Cedex 9, France
Pôle de Recherche et Pôle de Biologie, Cellular and Molecular Haematology Unit, Plateforme Hospitalière de Génétique Moléculaire des Tumeurs, Centre Hospitalier Universitaire de Grenoble, Cedex 9, France
We recently reported long-term organ allograft survival without ongoing immunosuppression in four of five patients receiving combined kidney and bone marrow transplantation from haploidentical donors following nonmyeloablative conditioning. In vitro assays up to 18 months revealed donor-specific unresponsiveness. We now demonstrate that T cell recovery is gradual and is characterized by memory-type cell predominance and an increased proportion of CD4+CD25+CD127−FOXP3+ Treg during the lymphopenic period. Complete donor-specific unresponsiveness in proliferative and cytotoxic assays, and in limiting dilution analyses of IL-2-producing and cytotoxic cells, developed and persisted for the 3-year follow-up in all patients, and extended to donor renal tubular epithelial cells. Assays in two of four patients were consistent with a role for a suppressive tolerance mechanism at 6 months to 1 year, but later (≥18 months) studies on all four patients provided no evidence for a suppressive mechanism. Our studies demonstrate, for the first time, long-term, systemic donor-specific unresponsiveness in patients with HLA-mismatched allograft tolerance. While regulatory cells may play an early role, long-term tolerance appears to be maintained by a deletion or anergy mechanism.
Chronic rejection and other complications from long-term immunosuppressive therapies could be overcome by the induction of a state of donor-specific tolerance. Mixed chimerism has shown promise in animal models as an approach to achieving transplantation tolerance (1–4). We have recently reported the successful achievement of long-term acceptance of kidney allografts without maintenance immunosuppression using nonmyeloablative conditioning to achieve durable or transient chimerism in two series of patients. The first series of six patients, affected by multiple myeloma, received combined kidney and BMT (CKBMT) from HLA-identical related donors (5). Subsequently, renal allograft acceptance induced by mixed chimerism was reported from another center in a single patient without malignancy receiving CKBMT from an HLA-identical related donor (6). Our second series involved five patients with end-stage renal failure without malignant disease, who received bone marrow and kidney transplantation from haploidentical related donors following nonmyeloablative conditioning (Supporting Table S1). Immunosuppression was discontinued approximately 8–14 months post-transplant in all except patient 3, who had panel-reactive antibodies pretransplant and lost the graft at day 10 to acute humoral rejection. Despite loss of measurable chimerism in all blood lineages by day 21 post-transplant, long-term renal allograft acceptance (now at 4–8 years) without immunosuppression was achieved in four of five patients, though low levels of anticlass II donor-specific antibody (DSA) persist in patient 4. Renal allograft tolerance was associated with donor-specific unresponsiveness in cell-mediated lympholysis (CML) and mixed lymphocyte responses (MLR) assays performed up to 18 months in all four patients (7), suggesting that a state of tolerance was achieved throughout the immune system (systemic tolerance). To gain insight into the mechanisms underlying donor-specific tolerance in these patients, we characterized the phenotype of recovering lymphocytes, followed functional responses to donors and assessed the possible role of regulatory cells.
Material and Methods
Patients and treatment
Five patients with end-stage renal disease were enrolled in this IRB-approved (Massachusetts General Hospital) Immune Tolerance Network (ITN)-sponsored (protocol ITN036ST, IRB approval 2008-P002007). As previously described (7), conditioning included cyclophosphamide, MEDI-507 (Medimmune), and, in patients 4 and 5, pretransplant rituximab and thymic irradiation followed by kidney and BM transplantation from the same HLA-haploidentical donor. Post-transplant immunosuppression included cyclosporine or tacrolimus (7) and was discontinued within 8 to 14 months (Supporting Table S1) in all except patient 3, as noted above.
Fresh blood cells were stained with fluorochrome-labelled mAbs against CD3, CD4, CD8, CD19, CD45RO, CD45RA, CD62L and CD25 (Beckton Dickinson [BD], San Jose, CA, USA). For Treg analysis, cryopreserved patient PBMC were thawed and stained with fluorochrome-conjugated mAbs against CD4, CD25, CD127, HLADR, CD45RO and CD45RA (BD), subsequently permeabilized, fixed and then stained with FOXP3 mAb (eBioscience, San Diego, CA, USA). Polychromatic flow cytometric analysis was performed on a custom 7-laser high resolution LSRII (BD) as described (8) and further analyzed with FlowJo (Tree Star, Inc., Ashland, OR, USA) software.
NK cell depletion
Fresh PBMC were depleted of NK cells and subsequently cryopreserved as previously described (5).
Renal tubular epithelial cell (RTEC) culture
RTECs obtained before transplant from the donor kidney for patients 1, 2 and 5 were cultured as previously reported (5). Epithelial origin was confirmed by expression of cytokeratin.
Mixed lymphocyte response (MLR) and cell mediated lysis (CML) assays
PBMC were aliquoted and cryopreserved or underwent further separation of T cells, CD56− cells and CD25− cells before cryopreservation and use in MLR and CML assays as previously described (5,9,10). The number of responder cells was reduced by 50% when isolated T cells or CD25− T cells were used instead of whole PBMC so that approximately similar numbers of T cells were present in each culture. Briefly, MLR involved 5-day triplicate cocultures of 2×105 responder cells with equal numbers of 30 Gy-irradiated stimulator cells in AIM V medium supplemented with 10% AB serum, followed by pulsing with 3H thymidine. CML involved cocultures of similar cell combinations for 7 days in RPMI 1640 with 6% FCS and standard additives, with the addition of IL-15, 5U/mL (Immunex, Seattle, WA, USA) on day 5. Serially diluted responders were incubated with 4000 51Cr-labelled PHA blast targets and lysis was determined by supernatant gamma counting.
When RTECs were used as targets, CML were performed as previously described, after incubating RTECs with rhIFN-γ to upregulate HLA-class I (otherwise expressed at only low levels on RTEC) (5).
Stimulated suppression assay
These assays were performed in two sequential phases. In phase 1 (priming phase, days 0–6), 4 × 106 PBMCs per well from tolerant patients were cultured with 4 × 106 irradiated donor or third-party PBMC for 6 days at 37 °C using 24-well flat-bottom plates (Costar, Cambrige, MA, USA). Bulk cultures were harvested, washed and resuspended in fresh medium on day 6 and allowed to rest overnight. In phase 2 (coculture phase, days 7–13), 1 × 105 fresh pretransplant PBMC per well from tolerant patients were added in 96 well plates with 2 × 105 per well-irradiated donor or third-party stimulators and 1 × 105 primed responders from phase 1 bulk cultures. After 6 days’ culture at 37 °C, cytotoxicity was tested as described (9).
Limiting dilution assays (LDA)
LDA to quantify cytotoxic T-lymphocyte precursor (CTLp) frequencies and IL-2-producing helper T cell (Th) frequencies were performed as described on replicates of 12–24 wells per dilution, using 51Cr-labelled lymphoblast targets and proliferation of the IL-2-dependent CTLL-2 cell line, respectively, as readouts for activity in each well (10). IL-15 was added to the CTLp cultures after removal of supernatants for IL-2 analysis on day 2 and cultures were continued for 7 additional days before cytotoxicity was assayed. LDA analysis was performed as previously described (11); briefly, deviation from the single hit Poisson model (SHPM) was tested by a standard Wald test and only frequencies fitting the SHPM were reported.
Statistical analyses were performed using the nonparametric Mann–Whitney U Test. Analyses were performed on triplicate values for each point for comparisons of unfractionated versus separated cell populations in CML and MLR assays.
Lymphocyte subset recovery
Flow cytometry was used to assess lymphocyte subset recovery. CD3+CD4+ counts recovered slowly, while CD3+CD8+ cells recovered more quickly, resulting in inversion of the normal CD4/CD8 ratio throughout the 3-year follow-up period. The peripheral blood CD19+ cell counts recovered more rapidly in the first two patients, who did not receive Rituximab, than in patients 4 and 5, whose conditioning included Rituximab (Figure 1).
While ‘memory-type’ CD45RO+ cells were most prevalent among CD4+ cells, recovery of ‘naive-type’ CD4+ CD45RA+ cells, presumably arising from the recipient thymus, was variable, beginning to increase between about 150 and 350 days post-transplant, and eventually reaching the normal range in all evaluable patients (Figure S1). Absolute numbers of naive-type and memory-type cells are shown in Supporting Figure 2. In one patient (patient 5), CD45RA+CD4+ cells persistently expressed CD45RO making naive/memory phenotype uninterpretable. We have observed this apparently stable CD45RA/RO ‘double positive’ phenotype in a small proportion of donors in many phenotypic studies performed over the years (M. Sykes et al. unpublished data). Among CD3+CD8+ T cells, the proportion of ‘naive-type’ CD45RA+RO−CD62L+ cells fluctuated in patients 1 and 5, showed a steady increase over time in patient 2, and remained low for the duration of follow-up in patient 4 (Figure S3).
As shown in Figure 2, a very high proportion of initially recovering CD3+CD4+ T cells in all patients expressed CD25. The reduction in CD25+ cells in patient 5 between days 21 and 42 occurred in association with treatment with rabbit anti-thymocyte globulin (ATG), plasmapheresis and rituximab for a suspected acute humoral rejection episode. Figure 2A shows that the percentage of CD25high cells was also elevated compared to normal to varying degrees in all four patients during the first 150 days post-transplant and remained elevated for the duration of follow-up in patient 4. In patient 5 at POD 45, all gated CD3+CD4+ cells were CD25high. The most markedly increased percentages of CD25high cells were seen in the context of the most marked CD4 T cell lymphopenia (Figure 1) and probably do not represent increased absolute concentrations over pretransplant values, which were not available. Absolute counts of CD25+CD4 T cells are shown in Supporting Figure S4.
To determine whether the above changes indicated enrichment of Treg cells, we characterized samples that had been cryopreserved at approximately POD 120 and 365 for expression of CD127, FOXP3, CD45RO, CD45RA and HLA-DR. At POD 120, the patients showed significantly increased frequencies of CD25+CD127−FOXP3+ regulatory T cells (Treg) within the CD4 population compared to healthy subjects (Figure 2B and C). By 1 year post-transplant, when only three patient samples were evaluable, the increased frequencies of Treg in patient CD4 cells narrowly missed achieving statistical significance compared to normal controls (p = 0.057) (Figure 2C). Expression of CD45RO, CD45RA and HLA-DR on CD25+CD127−CD4+ T cells was variable (not shown) but not significantly different from that in Tregs in healthy controls. A significantly increased proportion of CD4+CD25−CD127− cells compared to normal was also detected in the patients at POD 120 (p < 0.05) (data not shown).
Role of suppression in donor-specific CML and MLR unresponsiveness
As previously published, in vitro assays for CD8 and CD4 T cell-mediated alloreactivity (CML/MLR) demonstrated donor-specific unresponsiveness by 3 months after transplant in patients 2, 4 and 5, and by 9 months in patient 1, which persisted up to 9–18 months post-transplant, with recovery of third-party responses (7). Further follow-up (to 3 years post-transplant) revealed the donor-specific unresponsiveness to be durable in all four patients (Figure S5). To assess a possible role for regulatory cells in this unresponsiveness, whole PBMC, isolated T cells and CD25-depleted T cells were compared in MLR and CML assays. As shown in Figure 3A, depletion of CD25+ T cells revealed an antidonor response in patient 1 at POD 367 in both MLR and CML assays (Figure 3A-B). In contrast, antidonor responses did not become apparent among CD25− T cells at POD 540 (Figure 3A-B). Nevertheless depletion of CD25+ cells enhanced the anti-third-party CML and MLR response at both time points (Figure 3A-B). Similarly, CD25+ depletion failed to reveal an antidonor response at POD 1080 in this patient (data not shown).
An additional assay, which we term a ‘stimulated suppression assay’, described elsewhere (12), was performed on patient 1. Following restimulation in vitro with donor antigens, this patient's POD 187 post-transplant PBMC completely suppressed the antidonor response (Figure 3C, left panel) of pretransplant cells, but did not markedly suppress the anti-third-party response (Figure 3C, right panel). Pretransplant patient cells did not have donor-specific suppressive activity (Figure 3C), suggesting that donor-specific regulatory cells may have been specifically activated or expanded following the transplant. Priming of POD 187 post-transplant PBMC with third-party stimulators also led to specific suppression of the pretransplant antidonor response, with no significant suppression (compared to pretransplant cells) of the anti-third-party response (Figure 3C).
In patient 4, donor-primed POD 185 PBMC suppressed the pretransplant antidonor response completely. In contrast, neither the donor-primed pretransplant sample nor the third-party-primed POD 185 patient PBMC suppressed this response (Figure S6). A similar assay on patient 5 at POD 240 did not provide evidence for suppression.
Depletion of CD25+ cells did not reveal an antidonor alloresponse in patient 2 or 5 at similar early time points (POD 182 and 275, respectively), or in patient 4 at POD 540, despite enhancement of anti-third-party responses (Figure 4A and B). CD25+ cell depletion also failed to reveal an antidonor response at POD 1080 in MLR and/or CML in these patients (data not shown).
Absence of killing of donor RTECs by recipient PBMC
In patients 1, 2 and 5, RTECs from the donor kidney were cultured and tested as targets following stimulation with NK cell-depleted donor PBMC. In patient 2, loss of killing activity against donor RTECs was observed post-transplant compared to pretransplant levels (Figure 5A). In patient 1 and 5, for whom a pretransplant sample was not available, we also observed an absence of killing of donor RTECs post-transplant (Figure 5B and C).
To quantify donor-reactive T cells and probe further for evidence of suppression, LDA were performed. As shown in Figure 6A, antidonor CTLp were undetectable at 3 months and 3 years post-transplant in patient 1. Consistent with bulk CML assays, the anti-third-party CTLp frequency was low at 3 months but had recovered at 3 years. Patients 2, 4 and 5 similarly showed specifically reduced antidonor CTLp frequencies (often below the limit of detection) compared to pretransplant levels, with recovery of anti-third-party CTLp (Figure 6A). All of these assays demonstrated single-hit kinetics, with no evidence for a suppressive pattern.
Patient 1 showed low frequencies of IL-2-producing cells (T helper cells, Th) recognizing third party at 3 months but recovered responses at 9 months, 1 year, 18 months and 2 years. Antidonor responses were undetectable at all of these time points (Figure 6B). Patient 2 showed a markedly reduced antidonor Th frequency with preserved third-party response at 3 months and undetectable antidonor responses at 4, 8, 9 months and 2 years with weak but measurable anti-third-party responses. No evaluable data are available for patient 4. Patient 5 revealed marked reductions in donor-reactive IL-2-producing cells at 6 and 9 months post-transplant (Figure 6B). The fit with the single hit Poisson model is consistent with deletion or anergy and not suggestive of suppression as a mechanism of tolerance (13).
We analyzed patient 5 samples in LDA with and without Treg depletion. As shown in Figure 7A and B, Treg removal did not reveal an increased frequency of donor-reactive Th or CTLp at relatively early time points (6 and 9 months), when donor-reactive T cells had not yet undergone complete deletion.
We present here the first long-term demonstration of complete, systemic donor-specific unresponsiveness in human organ allograft recipients. These data were obtained in a unique series of patients exhibiting allograft tolerance. Our mechanistic data suggest that long-term tolerance in CKBMT recipients is due to deletion or anergy of donor-reactive T cells. While Tregs were enriched in initially-recovering CD4 cell populations of all patients, evidence for active suppression of antidonor reactivity was obtained in only two of four patients at 6–12 months and in zero of four patients after 1 year.
Following transplantation, T cell lymphopenia was prolonged, but none of our patients developed serious opportunistic infections. Most CD4 cells had a previously activated/memory phenotype, but CD45RA+RO- ‘naive-type’ CD4 cells eventually recovered. The predominant ‘memory’ T cell phenotype is consistent with our previous results in patients receiving a similar nonmyeloablative BMT regimen without kidney transplantation for the treatment of hematologic malignancies (10) and reported in other protocols involving T cell depleting reagents, both with (14,15) and without (16–18) conditioning for hematopoietic cell transplantation (HCT). However, patients with malignancies had received multiple prior chemotherapies and radiotherapies and often experienced infectious complications following transplant (19). Recovery of CD45RA+CD4+ cells in the current patient series, which did not occur in the patients with hematologic malignancies (10), most likely reflects more robust thymic recovery (15,20–23).
Early predominance of ‘memory’ T cells may reflect resistance to mAb-induced depletion of memory cells (24) or expansion from a small residual pool of peripheral T cells escaping depletion with conditioning therapy. While T cells undergoing lymphopenia-driven expansion have been reported to have stronger rejection capacity (25) and to resist tolerance induction in mice (26), our patients have achieved tolerance of an HLA-mismatched organ allograft despite presumed lymphopenia-driven expansion.
We considered a possible role in tolerance induction for the enriched Treg population, which has been previously noted in patients receiving HCT alone following a similar nonmyeloablative regimen (10), and was not observed in patients receiving alemtuzemab (16). Rabbit ATG causes Treg expansion and/or conversion of conventional T cells to Tregs (27–29) and spares Tregs in vivo in mice (30,31).
The proportions of CD25+ cells were markedly greater than the proportion of CD25high CD4+ cells in most of our patients. Unfortunately our phenotypic studies on fresh PBMC did not include stains for FoxP3 and CD127, so the latter studies were performed after cryopreservation, which reduces the level of surface CD25 expression (unpublished data), preventing distinction of CD25 high and low populations. Ongoing studies on fresh PBMC of a new patient series demonstrate higher proportions of Tregs early post-transplant than those reported here (T. Morokata et al. unpublished data). It is also possible that the high proportions of CD25low CD4 T cells in the initially reconstituting T cell populations reflect T cell activation.
The current studies demonstrate persistent donor-specific unresponsiveness of both proliferating and cytotoxic T cells for at least 3 years in all patients (Figure S5). Moreover, LDA studies demonstrate both an absence of IL-2-producing cells and of CTL recognizing donor antigens in long-term tolerant patients. The absence of a ‘sawtooth’ pattern (13,32) in these analyses suggests a deletion or anergy, rather than a suppressive, mechanism of tolerance. The LDA pattern produced by anergic T cells would be indistinguishable from that of a deleted population. However, our CTLp LDA and CML assays include IL-15, which has been shown to overcome anergy of human cytotoxic T cells even more effectively than IL-2 (33,34).
While some, but not other (35) studies have demonstrated relative hyporesponsiveness to the donor in MLR (36–38) and LDA (39) in patients with stable allograft function on immunosuppressive therapy, ours is the first to achieve complete, specific and persistent unresponsiveness to donor antigens in humans. In one study of CKBMT without specific conditioning for BMT or withdrawal of immunosuppresion, half of the patients achieved donor-specific unresponsiveness in MLR and CML assays, but in most cases unresponsiveness was demonstrated at only one time point (40). In contrast to our results, a renal allograft recipient studied 12 years after successful immunosuppression withdrawal demonstrated a vigorous MLR response to the donor (41). The specific, complete and persistent absence of antidonor reactivity in LDA, MLR and CML assays in our patients denotes a state of systemic tolerance.
One of our patients demonstrated suppression of persistent antidonor responses by regulatory cells at 1 year post-transplant. This suppression was mediated largely by CD25+ cells, presumably Tregs. A non-T cell suppressive population may also have been present at this point, as purification of T cells revealed a weak antidonor MLR (Figure 3). At 6 months post-transplant, the patient's cells specifically suppressed antidonor responses in a ‘stimulated suppression assay’, whereas pretransplant PBMC lacked such activity. The suppression was detected following exposure in vitro to donor or third-party antigens. Patient 4 PBMC also suppressed the antidonor response at 6 months, and in this case third-party stimulation of these cells did not induce the suppressive activity. Collectively, these data suggest that suppression was specific for donor antigens and that the third party used to stimulate patient 1 cells may have shared antigens with the donor, whereas that used to stimulate patient 4 cells did not.
CD25 depletion assays performed in the first year did not reveal suppressed antidonor responses in patient 2 or 5. These patients developed CML unresponsiveness more rapidly than patient 1 and the expansion, activation and/or survival of Tregs may require a persistent alloreactive population. Moreover, by 18 months, an antidonor response was no longer apparent in Treg-depleted PBMC of patient 1 and was also lacking in patient 4, suggesting that deletion (or anergy) of donor-reactive cells was complete. Collectively, evidence for suppressive mechanisms was obtained in two of four patients in the first year and in zero of the four patients after 1 year. While the possible role of Tregs in this tolerance is thus unclear, preliminary data suggest that Tregs may be in the renal allografts in a new series of patients, consistent with elevated FoxP3 mRNA levels noted in protocol biopsies of patients on this trial (7). We have not addressed a possible role for michrochimerism, which has been associated with CD8 cell-mediated suppression of antidonor reactivity in patients achieving spontaneous tolerance to HLA-identical kidneys (42).
The loss of chimerism in the present series of patients occurred within a few weeks of transplantation, when T cells were markedly depleted by the conditioning. Since we have observed early development of donor-specific IL-2-producing cell unresponsiveness in a patient in our newer series (43), this loss of chimerism may not reflect a T cell-mediated immune response but instead may reflect inadequate donor hematopoietic stem cell engraftment.
Patients receiving mismatched HCT alone for hematologic malignancies following a similar MEDI-507 conditioning regimen showed generally weak alloresponses but nevertheless had stronger antidonor than anti-third-party responses following loss of chimerism (10), in marked contrast to the specifically reduced antidonor responses observed in the current CKBMT patients. In combination, our results are consistent with a role for the kidney in the tolerance achieved in CKBMT recipients, through mechanisms that remain to be determined. Revealing the mechanism by which CKBMT leads to tolerance of the kidney allograft is of extreme importance for developing more widely applicable clinical protocols.
We thank Dr. Susan Saidman for selecting mAbs to be used in chimerism determination and Ms. Gena Coleman and Shavree Washington for secretarial assistance.
Funding: American Recovery and Reinvestment Act funding NIH R01AI084074, NIH ITN N01AI15416, Multiple Myeloma Research Foundation. G.A. was supported by the American Society of Transplantation and Umberto Veronesi Foundation.
Author contributions: G.A. wrote the manuscript, performed flow cytometry and in vitro assays, analyzed data and prepared specimens. M.C., J.S. and P.C. coordinated sample acquisition, prepared specimens, performed in vitro assays. S.A.L. and T.M. performed in vitro assays and assisted with data analysis. F.I.P. provided expert advice on and performed flow cytometry. K.K. supervised and coordinated research sample acquisition. T.B. aided in the analysis and interpretation of limiting dilution assays. A. B.C. T.K., B.R.D., N.T.R., T.S. and D.H.S. designed, supervised and carried out the clinical protocol and patient care. M.S. participated in the design of the clinical protocol, cowrote the paper and participated in and oversaw design of all laboratory studies, data analysis and interpretation, and preparation of the manuscript.
The authors of this manuscript have no conflicts of interest as described by the American Journal of Transplantation.