Extrathymic T Cell Deletion and Allogeneic Stem Cell Engraftment Induced with Costimulatory Blockade Is Followed by Central T Cell Tolerance

Female C57BL/6 (B6: H-2^b), B10.A (B10.A: H-2^a), and A.SW (H-2^s) mice were purchased from Frederick Cancer Research Center (Frederick, MD) or from The Jackson Laboratory (Bar Harbor, ME). Mice were maintained in a specific pathogen–free microisolator environment, as previously described (17).

Age-matched (6–8-wk-old) female B6 mice received 3 Gy whole body irradiation (WBI) and were injected intravenously on the same day (day 0) with unseparated BM harvested from MHC-mismatched female B10.A donors (10–12 wk old). A control group was injected intraperitoneally with depleting doses of rat IgG2b anti–mouse CD4 mAb GK1.5 and anti–mouse CD8 mAb 2.43 on days −5 and −1, as previously described (18). Murine CTLA4Ig was injected intraperitoneally as a single dose (0.5 mg) on day +2, and hamster anti– mouse CD40L mAb (MR1) was injected intraperitoneally on d0 (0.45 mg). CTLA4Ig was a gift from Bristol-Myers Squibb Pharmaceuticals, Seattle, WA; the MR1 hybridoma was provided to us by Dr. Randolph J. Noelle (Dartmouth Medical School, Lebanon, NH). In the experiments indicated, thymectomies were performed 4 wk before BMT, as described elsewhere (19, 20). The completeness of thymectomy was confirmed at the time of death 2 wk after BMT by visual inspection and two-color FACS^® staining (Becton Dickinson, San Jose, CA) (CD4-FITC versus CD8-PE) of mediastinal tissue. Mice showing any evidence of remaining thymic tissue were excluded from analysis.

Flow cytometric analysis (FCM) of multilineage chimerism was performed as previously described (21). In brief, forward angle and 90° light scatter properties were used to distinguish lymphocytes, monocytes, and granulocytes in peripheral white blood cells. Two-color FCM was used to distinguish donor and host cells of particular lineages, and the percentage of donor cells was calculated as previously described (21), by subtracting control staining from quadrants containing donor and host cells expressing a particular lineage marker, and by dividing the net percentage of donor cells by the total net percentage of donor plus host cells of that lineage. Dead cells were excluded using propidium iodide staining. Nonspecific FcγR binding was blocked by anti–mouse FcγR mAb 2.4G2 (22). FITC-conjugated mAbs included anti-CD4, anti-CD8, anti-B220 (all purchased from PharMingen, San Diego, CA), and anti-MAC1 (Caltag Labs., San Francisco, CA). Negative control mAb HOPC1-FITC, with no reactivity to mouse cells, was prepared in our laboratory. Biotinylated anti– H-2D^d mAb 34-2-12 and control mAb HOPC1 were developed with PE-streptavidin.

Peripheral blood lymphocytes were stained with FITC-conjugated anti-Vβ5.1/2, Vβ11, and Vβ8.1/2 mAbs versus PE-conjugated anti-CD4 mAb (all from PharMingen). Nonspecific PE-conjugated rat IgG2a (PharMingen) served as negative control. Two-color FCM analysis was performed on gated CD4⁺ cells. Splenocytes were stained with FITC-conjugated anti-Vβ5.1/2, Vβ11, and Vβ8.1/2 mAbs versus PE-conjugated anti-CD4 mAb (or anti-CD8 mAb; PharMingen) and anti–34-2-12-biotin (BIO) developed with CyChrome-streptavidin (PharMingen). Three-color FCM analysis was performed on 34-2-12–negative, CD4⁺ (or CD8⁺) cells. Thymocytes were stained with FITC-conjugated anti–TCR-β (PharMingen), anti–Vβ5.1/2, Vβ11, and Vβ8.1/2 versus BIO-conjugated KH95 (anti-D^b, PharMingen) developed with PE-streptavidin. Two-color FCM analysis was performed on gated host-type class I (KH95)–high cells, and the percentage of Vβ positive cells in this gate was corrected for the percentage of TCR-high cells in the same gate, as previously described (23). For BIO.A controls, gated 34-2-12-high cells were analyzed in a similar way. Background staining (as determined by nonreactive mAb HOPC-FITC) was subtracted from the percentage of cells staining with each anti-Vβ mAb. P values were calculated using a two-tailed Student's t test.

Full thickness tail skin from B10.A (donor-specific) and fully MHC-mismatched A.SW (third party) mice was grafted onto the lateral thoracic wall, secured with 5-0 silk sutures and bandaids, and followed by visual and tactile inspections daily for 3 wk, and then at least one inspection every week thereafter. Grafts were defined as rejected when <10% of the graft remained viable.

To determine whether blocking the CD28 and CD40 costimulatory pathways could allow survival of fully MHC-mismatched BM and the induction of mixed chimerism and tolerance, B6 mice were treated with 3 Gy WBI and received 1.5 × 10⁷ unseparated BM cells from fully MHC-mismatched B10.A donors. A single dose of MR1 and of CTLA4Ig was given either alone or in combination, on days 0 and +2, respectively. Donor hematopoiesis was assessed at multiple time points after BMT by FCM of peripheral white blood cells. By staining with an mAb specific for donor class I versus various lineage markers, the net percentage of donor cells within these lineages was determined (see also Materials and Methods).

The combined administration of CTLA4Ig plus MR1 led to high levels of chimerism in all hematopoietic lineages, including T, B, and myeloid cells (Fig. 1,D). Donor reconstitution averaged >40% in all lineages by 7 wk, and remained high during the observation period of 34 wk. Especially surprising was the high level of donor representation among CD4 and CD8 cells by 7 wk after BMT, even though the hosts did not receive T cell depletion in their conditioning. Donor T cell levels in the group treated with MR1 plus CTLA4Ig were stable throughout long-term follow-up (34 wk) and were, on average, higher than those in a control group conditioned with anti-CD4– and anti-CD8–depleting mAbs (Fig. 1,A) (21, 24). 15 out of 16 mice treated with the combination of MR1 and CTLA4Ig developed high levels of chimerism (one mouse showed no detectable chimerism and was excluded from further analysis as an outlier). Treatment with MR1 alone led to high levels of donor cells among myeloid lineages and B cells at early time points after BMT, and to lower levels of chimerism among CD4 and CD8 cells. However, the induction of chimerism was less reliable in these mice than that in the group receiving both costimulatory blocking reagents, and donor chimerism was not stable in recipients of MR1 alone (Fig. 1,C). Mice receiving CTLA4Ig alone did not show chimerism in peripheral blood, as detected by FCM, at any time after BMT (Fig. 1 B). Similarly, control animals receiving WBI and BM cells alone failed to show hematopoietic chimerism (data not shown).

These results demonstrate that the simultaneous blockade of the CD28 and CD40 pathways allows the induction of high levels of stable multilineage chimerism, and thus allogeneic pluripotent stem cell engraftment. In previous protocols using immunocompetent hosts, this has only been achieved with regimens that involve exhaustive depletion of host T cells with mAbs or myeloablation by lethal irradiation or cytotoxic drugs (18, 21, 25–28). The toxicity associated with myeloablation is generally considered to preclude its use in organ transplant recipients. The completeness of T cell depletion necessary in experimental protocols to achieve mixed chimerism (21, 26) would also be of great concern in the clinical setting, given that the capacity of the adult human thymus to regenerate T cells might be limited after such treatment (29).

Primary skin grafting is considered the most stringent test of transplantation tolerance. We therefore grafted donor (B10.A) and third party (A.SW) full-thickness tail skin onto recipients at various time points after BMT. Mice that received both CTLA4Ig and MR1, plus 3 Gy WBI and BMT, permanently accepted donor skin grafts placed 3, 6, or 10 wk after BMT (Fig. 2,B), with the exception of two animals that rejected their grafts 57 and 76 d after graft placement, respectively. Third party grafts were readily rejected (median survival time [MST] = 10 d), demonstrating the donor specificity of the tolerance induced. This skin graft survival compares favorably even to the control animals that were conditioned with T cell– depleting antibodies, in which only 60% of donor skin grafts survived more than 100 d (Fig. 2,A). Mice treated with MR1 alone in addition to 3 Gy WBI and BMT demonstrated prolongation of donor skin graft survival (MST = 42 d; Fig. 2 A). However, only two out of nine grafts were accepted for 100 d. In contrast, mice receiving BMT after 3 Gy WBI and CTLA4Ig alone did not show prolonged survival of donor skin grafts (MST = 10 d), consistent with the absence of chimerism.

These results demonstrate the presence of donor-specific tolerance across a full MHC barrier in chimeras prepared with MR1 plus CTLA4Ig. The ability of mixed chimeras prepared with MR1 and CTLA4Ig to rapidly reject third party skin grafts is evidence for their immunocompetence. Although previous studies reported the ability of costimulatory blockade alone to prolong skin graft survival (8, 9, 11), permanent skin graft acceptance and tolerance was not reliably achieved (9, 30).

To examine whether deletion of donor-reactive T cells occurs when chimerism is induced with MR1 and CTLA4Ig, peripheral blood lymphocytes were analyzed for the presence of certain Vβ subunits on their TCRs. The donor strain B10.A expresses I-E, which is required to present superantigens derived from Mammary tumor virus 8 and 9 endogenous retroviruses encoded in the B6/B10 background genome. Developing thymocytes whose TCRs contain Vβ11 or Vβ5.1/2, which bind to these superantigens, are deleted in I-E-positive B10.A mice (31–33), but not in B6 mice, because they do not express I-E (32, 34).

Partial deletion of Vβ5⁺ and Vβ11⁺ peripheral CD4 T cells was observed as early as 1 wk after BMT in mice receiving 3 Gy WBI followed by MR1 plus CTLA4Ig (Fig. 3). The deletion became progressively more complete over the ensuing weeks, and reached similar levels to those in chimeras prepared with T cell depletion (data not shown) (23). Deletion of these Vβ5⁺ and Vβ11⁺ cells was sustained throughout the follow-up period (>6 mo in the first experiment for which chimerism data are shown in Fig. 1). Percentages of Vβ8-bearing CD4 cells, which do not recognize superantigens on the donor or host, were not reduced at any time point, ruling out a nonspecific deletional process. Mice treated with BMT (plus 3 Gy WBI) and MR1 alone showed early partial deletion of Vβ5 and Vβ11, which was only transient in the experiment shown (Fig. 3). However, in the experiment shown in Fig. 1, deletion was still observed at later time points for the group receiving BMT (plus 3 Gy WBI) and MR1 alone, which correlated with the higher initial levels of chimerism observed for this group in this experiment. Control animals receiving 3 Gy WBI plus BMT alone or BMT (plus 3 Gy WBI) with CTLA4Ig failed to show any Vβ5 or Vβ11 deletion. As expected, deletion of Vβ5 and Vβ11 did not occur in control animals receiving WBI and MR1 plus CTLA4Ig without BMT (data not shown). Downregulation of the level of TCR expression instead of deletion seems an unlikely explanation for the reduction in Vβ5⁺ and Vβ11⁺ CD4 T cells in chimeras, since the intensity of the Vβ5 and Vβ11 staining on the cells remaining in the blood at 1 and 3 wk after BMT was similar to that in nontransplanted controls (data not shown). Thus, no evidence for TCR downmodulation was observed.

These results indicate that in BMT recipients treated with MR1 plus CTLA4Ig and 3 Gy WBI, donor-reactive host T cells start to disappear from the periphery very soon after BMT. Neither CTLA4Ig nor MR1 (35, 36) is known to be directly cytotoxic to the T cells to which they bind. 3 Gy WBI causes only transient and mild leukopenia (37) and only partial T cell depletion (Sykes, M., unpublished data). This time course suggests that the deletion observed at 1 wk is not entirely due to intrathymic mechanisms, since a sufficient number of thymocytes would be unlikely to emigrate from the thymus during this period to “dilute” the preexisting peripheral repertoire to an extent that could explain the observed decrease in Vβ5⁺ and Vβ11⁺ CD4 lymphocytes in PBLs.

To directly determine whether peripheral deletion is responsible for the early decline in donor-reactive CD4 T cells in chimeras, thymectomized (ATX) B6 mice received B10.A BM cells after conditioning with 3 Gy WBI and MR1 plus CTLA4Ig. As shown in Fig. 4,A, ATX mice demonstrated partial deletion of Vβ5⁺ and Vβ11⁺ peripheral blood CD4 cells 1 wk after BMT, and the degree of deletion was comparable to that in euthymic recipients. Vβ8⁺ CD4 cells were not diminished, indicating the specificity of the deletion for superantigens presented by the donor. Similarly treated ATX mice not receiving BMT showed no reduction in the percentage of Vβ5⁺ or Vβ11⁺ CD4 cells compared to untreated B6 mice (Fig. 4 A), demonstrating that this peripheral deletion occurred specifically in response to donor marrow in BMT recipients.

Further evidence that peripheral deletion plays a role in the early period after BMT in this model was obtained from chimeras killed 20 wk after BMT and CTLA4Ig plus MR1. Among splenocytes of these mice, the percentage of Vβ5⁺ and Vβ11⁺ cells among CD4⁺ T cells was reduced to similar levels as in normal, control B10.A mice. However, the percentages of Vβ5⁺ and Vβ11⁺ cells were substantially higher among CD8⁺ splenocytes than among CD4⁺ splenocytes. Nevertheless, the percentages of CD8 splenocytes using these Vβ were significantly lower than among those of normal, control B6 mice (Fig. 4 B). As discussed below, this difference most likely reflects the dilution of the peripheral CD8 pool by new thymic emigrants that are tolerized by a central deletion mechanism in the chimeras.

Peripheral deletion has been shown to be one consequence of powerful T cell responses in vivo, but it has only been reported after marked expansion of antigen-recognizing cells (38, 39). Although 1 wk after BMT was the earliest time point at which donor-reactive host T cells were examined, we did not see evidence of such initial expansion. More recently, in vitro evidence has demonstrated that costimulatory signals play a prominent role in preventing apoptotic cell death after TCR engagement (40–42). However, apoptosis induced in vivo by antigen encountered in the presence of costimulatory blockade has not been described previously.

Mature recipient T cells (including both CD4 and CD8 cells) in the thymus showed marked deletion of Vβ5 and Vβ11 when animals were killed 20 wk after BMT (Fig. 4 B), demonstrating that newly developing donor-reactive thymocytes are effectively deleted during maturation in the thymus in long-term chimeras.

Together, our data suggest a model whereby extrathymic clonal deletion occurs early after BMT under cover of costimulatory blockade. Allogeneic pluripotent stem cell engraftment is thus permitted, and subsequent tolerization of newly-developing donor-reactive thymocytes occurs by deletional mechanisms in the thymus similar to that which occurs in animals initially treated with T cell–depleting mAbs (23). This model also explains the discrepancy in the extent of deletion of CD4 cells compared to CD8 cells in the peripheral tissues of long-term chimeras. Vβ5⁺ and Vβ11⁺ CD4 cells are subject to deletion both intrathymically and in the periphery when they recognize superantigen plus donor MHC class II. CD8 cells expressing these Vβ are efficiently deleted intrathymically (43) at the CD8⁺CD4⁺ stage of maturation. However, CD8⁺CD4⁻ cells are not very effectively deleted extrathymically (38, 39, 44), although weak proliferative activity to superantigens presented by class II MHC in the periphery has been described (39, 44). It follows that the substantial difference in the degree of deletion between peripheral CD4 and CD8 cells is most likely due to the more extensive contribution of extrathymic deletion among the CD4 compared to the CD8 cells preexisting at the time of BMT. This conclusion is further supported by the observation that ATX recipients of BMT plus costimulatory blockade demonstrated a significant reduction of Vβ5⁺ and Vβ11⁺ CD4 splenocytes, but not of CD8 splenocytes, at 2 wk after BMT (data not shown).

Previously published reports showed that costimulatory blockade as the sole immunosuppressive treatment can allow the permanent acceptance of vascularized allografts grafted at the same time (8, 11, 12). However, the organ allografts themselves probably play a critical role in tolerance induction in these models, and deletional tolerance has not been demonstrated. In contrast, solid organ (skin) grafting was not required in the induction phase of tolerance in our model. Instead, the permanent engraftment of donor hematopoietic cells ensured the tolerization of preexisting host T cells and of T cells that developed subsequent to the disappearance of the costimulatory blocking agents from the circulation. This later tolerance occurred through intrathymic deletional mechanisms (Fig. 4), presumably as a consequence of the presence of donor-derived APCs in the thymus, as has been demonstrated in long-term mixed chimeras prepared with other regimes that involve initial depletion of the T cell repertoire with mAbs (23). In previous studies, administration of donor lymphocytes in the presence of anti-CD40L has been shown to allow survival of pancreatic islet allografts (45) and to prolong survival of MHC-mismatched skin allografts (9). However, chimerism and deletion were not evident in those studies. Thus, our studies provide the first demonstration that costimulatory blockade leads to peripheral deletion of donor-reactive T cells, then allows the engraftment of fully MHC-mismatched, allogeneic pluripotent stem cells, which induce central tolerance among T cells that subsequently develop in the thymus.

BMT with CTLA4Ig plus MR1 specifically eliminates donor-reactive T cells, while avoiding the nonspecific depletion or suppression of T cells, which is a component of all clinically available immunosuppressive strategies, and can lead to severe complications. This new treatment protocol would be suitable for both cadaveric and living-related organ transplantation, as it allows the reliable induction of deletional tolerance with a nontoxic conditioning regimen beginning on the day of transplantation. Since the peripheral T cell repertoire is not globally depleted by the conditioning and only a low, minimally myelosuppressive dose of WBI is given, the clinical potential of this approach is extraordinarily high.

We thank Dr. Hugh Auchincloss, Jr., and Dr. Henry Winn for critical review of the manuscript; Dr. David H. Sachs for his support and advice; and Diane Plemenos for secretarial assistance.

This study was supported by National Institutes of Health grant R01 HL-49915, and in part by a sponsored research agreement between Massachusetts General Hospital and BioTransplant, Inc. Dr. Wekerle was supported by fellowships from the Max Kade Foundation and the Austrian Science Fund (FWF). Dr. Sayegh is a recipient of the National Kidney Foundation Clinician Scientist Award.