Donor CD8+ T cells play a critical role in mediating graft-vs-leukemia (GVL) activity, but also induce graft-vs-host disease (GVHD) in recipients conditioned with total body irradiation (TBI). In this study, we report that injections of donor C57BL/6 (H-2b) or FVB/N (H-2q) CD8+ T with bone marrow cells induced chimerism and eliminated BCL1 leukemia/lymphoma cells without clinical signs of GVHD in anti-CD3-conditioned BALB/c (H-2d) recipients, but induced lethal GVHD in TBI-conditioned recipients. Using in vivo and ex vivo bioluminescent imaging, we observed that donor CD8+ T cells expanded rapidly and infiltrated GVHD target tissues in TBI-conditioned recipients, but donor CD8+ T cell expansion in anti-CD3-conditioned recipients was confined to lymphohematological tissues. This confinement was associated with lack of up-regulated expression of α4β7 integrin and chemokine receptors (i.e., CXCR3) on donor CD8+ T cells. In addition, donor CD8+ T cells in anti-CD3-conditioned recipients were rendered unresponsive, anergic, Foxp3+, or type II cytotoxic T phenotype. Those donor CD8+ T cells showed strong suppressive activity in vitro and mediated GVL activity without clinical signs of GVHD in TBI-conditioned secondary recipients. These results indicate that anti-CD3 conditioning separates GVL activity from GVHD via confining donor CD8+ T cell expansion to host lymphohemological tissues as well as tolerizing them in the host.
Allogeneic hemopoietic cell transplantation (HCT)4 is a curative therapy for hemological malignancies. However, graft-vs-host disease (GVHD) remains a major obstacle in classical allogeneic HCT, where recipients are usually conditioned with total body irradiation (TBI) (1). The conditioning procedures cause tissue damage and release of proinflammatory cytokines and chemokines, which lead to the activation of host APCs. The activated APCs subsequently activate donor alloreactive T cells and the latter cells expand and differentiate into Th1 and cytotoxic T (Tc) 1 cells, which migrate to GVHD target tissues to cause GVHD (2, 3, 4, 5).
GVHD is a relatively organ-specific disease and the major target organs are gut, skin, liver, and lung (2). It is theorized that donor anti-host T cells enter specific tissues not only because they “see” an alloantigen but also because they posses the requisite combination of homing receptors and chemokine receptors to engage the endothelium at the tissue(s) (6). At present, three tissue-specific lymphocyte homing receptors have been identified: L-selectin for homing to lymph nodes (LN), α4β7 integrin for homing to gastrointestinal tract and gut-associated lymphoid tissues (GALT), and cutaneous lymphocyte Ag (CLA) for homing to skin (7, 8). Multiple chemokine receptors (i.e., CCR7, CCR9, CCR4, and CCR10) have also been demonstrated to mediate T cell tissue-specific homing (9). CCR7 and L-selectin are routinely expressed at high levels among naive lymphocytes and central memory lymphocytes. L-selectin ligand (peripheral LN addressin) and CCR7 ligand (CCL21) are constitutively expressed on high endothelial venules, enabling naive and central memory T cells to move frequently between lymph and blood (8, 10, 11, 12).
However, L-selectin and CCR7 expression are characteristically low on effector and effector memory lymphocytes; alternatively, these cells bear high-level expression of tissue-specific homing receptors and chemokine receptors. For example, high endothelial venules of GALT constitutively express α4β7 ligand (mucosal addressin cell adhesion molecule 1) and CCR9 ligand (CCL25) (8, 9, 10, 11, 12); in contrast, dermal microvessels constitutively express CLA ligands (E and P selectin), CCR4 ligand (CCL17) and CCR10 ligand (CCL27) (8, 9, 10, 11, 12). Expression of these ligands is strongly up-regulated by inflammation (13, 14). Thus, lymphocytes expressing α4β7 and CCR9 or CLA, CCR4, and CCR10 preferentially migrate to gut or skin tissues.
The important roles of homing receptors and chemokine receptors in allogeneic donor T cell migration to GVHD target tissues have recently been demonstrated (15, 16, 17, 18, 19, 20, 21, 22, 23, 24). For example, blocking or gene knockout of CD62L and/or α4β7 integrin markedly reduced donor T cell migration to gut tissues (17, 18); gene knockout of chemokine receptors (i.e., CCR2, CCR6, and CXCR3) has been shown to significantly reduce donor T cell migration to GVHD target tissues in TBI-conditioned recipients (19, 22, 23).
Graft-vs-leukemia (GVL) activity plays a critical role in preventing leukemia relapse and alloreactive donor CD8+ T cells are the most important mediator of GVL activity. However, they also induce severe GVHD in TBI-conditioned recipients in animal models as well as in humans (25, 26, 27, 28). This is because GVL activity is initiated by host APCs and alloantigens that also initiate GVHD (4, 29, 30). Therefore, one of the effective approaches to separate GVL activity from GVHD is to avoid alloreactive donor T cell migration to GVHD target tissues. As we mentioned above, alloreactive T cells tend to infiltrate inflammatory tissues, so a conditioning regimen that does not cause inflammation in GVHD target tissues should not cause GVHD. We recently reported that, in anti-CD3-conditioned recipients, donor CD8+ T cells facilitated induction of mixed chimerism without infiltration of GVHD target tissues, but the same doses of donor CD8+ T cells induced severe GVHD in TBI-conditioned recipients (31).
In the current study, we tested whether anti-CD3 conditioning separated GVL activity from GVHD and explored the mechanisms of GVHD prevention in anti-CD3-conditioned recipients. We observed that donor CD8+ T cells mediated GVL activity without clinical signs of GVHD in anti-CD3-conditioned recipients and that the separation of GVL activity from GVHD was due not only to confinement of donor CD8+ T cells to host lymphohematological tissues, but also to tolerization of host-reactive donor CD8+ T cells.
Materials and Methods
C57BL/6 (H-2b, CD45.2), congenic C57BL/6 (H-2b, CD45.1), BALB/c (H-2d), FVB/N (H-2q), and B10. A (H-2a) mice were purchased from The Jackson Laboratory. The luciferase-expressing (luc+) transgenic FVB/N line was generated as previously described (32). All animals were maintained in a pathogen-free room at City of Hope Animal Research Facilities. Mice at an age of 6–12 wk were used in the current studies. Animal use protocols were approved by the institutional review committee.
Flow cytometric analysis and cell sorting
31, 32, 33+ T cells from donor spleen were positively selected with anti-CD8-FITC and anti-FITC micromagnetic beads (Miltenyi Biotec), as described previously (31). Donor CD8+ T cells from anti-CD3- or TBI-conditioned recipients were first enriched with anti-CD8-FITC and anti-FITC micromagnetic beads, then stained with anti-H-2b and anti-TCRαβ again, and sorted with flow cytometry as described previously (26). FITC-conjugated anti-BCL1-Id (6A5.1) mAb was a gift from Dr. S. Strober (Stanford University School of Medicine, Stanford, CA) and used as described previously (34).4β7 (DATK32), CXCR3 (220803), CCR7 (4B12), H-2q (KH114), H-2b (AF6-88.5), and H-2d (34-2-12). FACS was performed with a four-laser MoFlo Immunocytometry System (DakoCytomation) and data were analyzed with FlowJo software (Tree Star), as described previously (
Conditioning of recipients, HCT, and in vivo and ex vivo bioluminescent imaging (BLI)
Production of anti-CD3 mAb (145-2C11), and anti-CD3 or TBI conditioning of BALB/c mice were described previously (31Limulus amebocyte lysate QCL-1000R (Cambrex). Mice were injected i.v. with anti-CD3 at a dose of 20 μg/g body weight. Serum levels of anti-CD3 mAb were measured with an in vitro blocking assay, in which untreated mouse spleen cells were incubated with serum from the anti-CD3-treated mice, and then stained with FITC-conjugated anti-CD3 (145-2C11). Although serum from mice 3 days after anti-CD3 injection showed clear blocking, serum from mice 5–7 days after injection showed little blocking, indicating there is little mAb in the serum. Thus, HCT was usually performed 7 days after anti-CD3 injection. For induction of chimerism, donor CD8+ T cells were coinjected with whole donor bone marrow (BM) cells. For secondary transfer, donor CD8+ T cells were injected with T cell-depleted (TCD)-BM cells, in which T cells were depleted with Miltenyi Biotec columns after being stained with biotinylated anti-TCRαβ, anti-CD4, anti-CD8, and anti-biotin micromagnetic beads. The purity was >98%. In vivo and ex vivo BLI of BALB/c recipients transplanted with sorted CD8+ T cells from luc+ FVB/N donors and BM from nontransgenic FVB/N donors were performed as described previously, using an IVIS100 charge-coupled device imaging system (Xenogen) (35).
MLR and cytotoxicity assay
Responder cells (0.25–0.5 × 106/well) were cocultured with stimulator cells (0.25–0.5 × 106/well, irradiated with 3000 rad) in complete RPMI 1640 medium with 10% human serum in a 96-well plate for 5 days and proliferation was measured with [3H]TdR incorporation as described previously (31, 33). Cytotoxicity activity of recipient spleen cells was measured with a JAM test in a 96-well plate (36) after a 4-day in vitro culture with irradiated target BCL1 cells (5 × 106 spleen cells and 1 × 106 BCL1 cells/well in a 24-well plate).
Histopathology of skin and intestine
Measurement of cytokines in culture supernatants
Culture supernatants were from a 48-h culture of 0.2 × 106 sorted CD8+ T cells stimulated with plate-bound anti-CD3 mAb (145-2C11) and 2 μg/ml soluble anti-CD28 (37.51; BD Pharmingen). Cytokines were measured using Cytokine 10-Plex for Luminex (catalog no. LMC0001; BioSource International), as described previously (31, 33).
Comparison of survival groups was analyzed using the log-rank test with GraphPad Prism version 3.0 (GraphPad Software). Comparison of two means was analyzed using the unpaired two-tailed Student t test.
Donor CD8+ T cells mediated GVL activity without clinical signs of GVHD in anti-CD3-conditioned recipients
We recently reported that injections of donor BM with CD8+ T cells induced mixed chimerism without clinical signs of GVHD in recipients conditioned with anti-CD3 mAb, but induced lethal GVHD in recipients conditioned with sublethal TBI (31). Our previous studies showed that induction of complete chimerism was associated with elimination of residual leukemia/lymphoma cells in allogeneic recipients (34). In the current study, we tested whether donor CD8+ T cells could induce complete chimerism and mediate GVL activity without clinical signs of GVHD in anti-CD3-conditioned recipients.
BALB/c mice (7–10 wk old) were injected i.p. with 500 luciferase-transfected BCL1 (BCL1/luc) leukemia/lymphoma cells as previously described (26, 37). One day after BCL1/luc inoculation (day −7), the mice were conditioned with anti-CD3 mAb at a dose of 20 μg/g bodyweight, and 7 days after conditioning (day 0), the BCL1/luc bearing mice were injected i.v. with 100 × 106 BM and 20 × 106 CD8+ T cells from C57BL/6 donors, and the CD8+ T cells were injected again on days 5 and 10 after the first injection. The control group was injected with PBS only. An additional control group was the sublethally (650 rad) TBI-conditioned BALB/c recipients given the same dose of donor BM and CD8+ T cells. To avoid the complication of mortality caused by both GVHD and tumor growth, these control mice were not given BCL1/luc cells.
The recipients were carefully monitored for tumor growth and clinical sings of GVHD and checked for chimerism as previously described (26, 31, 37). We found that BCL1/luc cells grew rapidly in control mice given anti-CD3-conditioning only as revealed by in vivo BLI (Fig. 1⇓A), and all (eight of eight) of them died of tumor growth around day 45 after tumor cell injection (Fig. 1⇓B). In contrast, BCL1/luc tumor cells in the anti-CD3-conditioned recipients transplanted with donor BM and CD8+ T cells only grew a little initially, and were eliminated in all (eight of eight) of the recipients, which survived for >120 days (p < 0.001; Fig. 1⇓, A and B). Flow cytometric analysis with anti-BCL1-Id mAb at 35 days after tumor cell injection showed a high percentage (>20%) of BCL1 tumor cells in the peripheral blood of control mice, but BCL1 tumor cells were undetectable in the anti-CD3-conditioned recipients transplanted with donor BM and CD8+ T cells (Fig. 1⇓C).
The TCRαβ+ T, B220+ B, MAC+/Gr-1+ macrophage/granulocyte cells in the anti-CD3-conditioned recipients were all donor-type 8 wk after HCT, and the ratio of CD4+ vs CD8+ T was close to normal (2:1) (Fig. 1⇑D), indicating the recipients were all complete chimeras. The anti-CD3-conditioned chimeric recipients showed healthy appearance without weight loss or hair loss and all survived for >120 days (Fig. 1⇑, B, E, and F). In contrast, the TBI-conditioned recipients given the same dose of donor BM and CD8+ T cells showed severe clinical signs of GVHD, including weight loss, hair loss, and all died by day 60 after HCT (Fig. 1⇑, B, E, and F). In additional experiments, ∼50 days after HCT, the histopathology of skin and colon tissues of the anti-CD3- or TBI-conditioned recipients was studied. We observed intestinal epithelium apoptosis and lymphocyte infiltration, as well as hyperplasia in epidermis and lymphocyte infiltration in the dermis in TBI-conditional recipients, but no damage was found in the intestine and skin of the anti-CD3-conditioned recipients (Fig. 1⇑G).
The thymus is a sensitive target for GVH response and GVHD recipients often show severe thymic damage (38). The generation of CD4+CD8+ thymocytes has been used to evaluate the function of the thymus of GVHD recipients (38, 39). Although CD4+CD8+ thymocytes were <5% in TBI-conditioned recipients, it was >65% in the anti-CD3-conditioned recipients (Fig. 1⇑G). In addition, the yield of thymocytes of anti-CD3-conditioned recipients was ∼28-fold higher than that of TBI-conditioned GVHD recipients (p < 0.01, 28 ± 3.8 × 106 vs 1 ± 0.7 × 106). The percentage of CD4+CD8+ thymocytes and the yield of total thymocytes of the anti-CD3-conditioned recipients were similar to that of TBI-conditioned recipients injected with BM cells only, which did not develop GVHD (data not shown). We previously reported donor bone marrow alone did not cause GVHD in mice (26, 40). These results indicate that donor CD8+ T cells mediate GVL activity without damaging the thymic regeneration function in anti-CD3-conditioned recipients.
Anti-CD3-conditioned long-term complete chimeras maintained GVL activity without clinical signs of GVHD
It was reported that administration of delayed donor lymphocyte infusion on day 35 post-HCT led to conversion from mixed to complete donor chimerism and mediated a GVL activity (41). However, long-term donor lymphocyte infusion recipients lost in vivo GVL activity and in vitro CTL responses, although the recipients had anti-host proliferative response in MLR (42). Thus, we tested whether the long-term complete chimeras induced with anti-CD3-conditioning and injections of donor BM and CD8+ T cells could mediate GVL activity. Accordingly, 120 days after anti-CD3-conditioning, the complete chimeras and the control mice given conditioning only were injected i.p. with 2000 BCL1/luc cells. We found that BCL1/luc cells grew rapidly in the control mice and killed all (eight of eight) of them 42 days after injection. In contrast, the BCL1/luc cells in the chimeras were killed gradually and eliminated by day 26 after injection and all (eight of eight) of the complete chimeras survived without clinical signs of GVHD for >50 days after tumor injection (Fig. 2⇓, A and B). These results indicate that GVL activity is retained in the anti-CD3- conditioned long-term complete chimeras even without prior exposure to tumor cells.
Furthermore, spleen and LN cells from the complete chimeras were stimulated with host-type BALB/c and third-party B10.A spleen cells in an MLR culture. Although the recipient cells showed no proliferation to host-type stimulators, they proliferated vigorously in response to third-party cell stimulation (Fig. 2⇑C), indicating that there is no detectable CD4+ T mediated anti-host immune response in the anti-CD3-conditioned long-term chimeras. Although the recipient cells did not proliferate to stimulation by host-type spleen cells, they killed host-type BCL1 cells in a cell dose-dependent manner after culture in vitro with BCL1 cells (Fig. 2⇑D). They also showed weak cytotoxic activity against host-type LPS-blast cells (dada not shown). These results indicate that there are host-reactive CTLs in the long-term chimeras with no clinical signs of GVHD and that these CTLs should be from the injected donor CD8+ T cells, because de novo developed host-reactive CD4+ and CD8+ donor T cells were reported to be deleted by negative selection in the thymus of GVHD-free chimeras (31, 43).
Next, we studied the tissue distribution and clonal deletion of the injected donor CD8+ T cells in the long-term chimeras. Accordingly, donor CD8+ T cells (CD45.2+) were coinjected with congenic BM cells (CD45.1+) into anti-CD3-conditioned recipients to induce complete chimeras as described above. A total of 120 days after HCT, the thymocytes, spleen, and BM cells of the chimeras were analyzed with multicolor flow cytometry and we found that the injected donor CD8+ T cells (CD45.1−) among the total donor CD8+ T cells was <0.1% in the thymus, 10–14% in the spleen, and 42–51% in the BM (Fig. 2⇑E). The injected donor CD8+ T cells showed predominantly (>80%) effector/memory phenotype (CD62LlowCD44high, data not shown). These results are consistent with a previous report that memory T cells are enriched in BM (44).
Endogenous superantigen-mediated deletion of TCR Vβ subunits was previously used as an indication of clonal deletion of alloreactive T cells (31, 45, 46). Vβ subunits of donor CD8+ T cells in the long-term chimeric BALB/c recipients were measured with flow cytometry. As shown in Table I⇓, Vβ5+, but not Vβ6+, CD8+ T cells were deleted in host BALB/c mice. Neither Vβ5+ nor Vβ6+ CD8+ T cells were deleted in donor C57BL/6 mice. Compared with C57BL/6 donor mice, the Vβ5+ cells among the injected donor CD8+ T cells and among the de novo-developed donor CD8+ T cells from the chimeras were reduced 2- and 4-fold (p < 0.05), respectively, but no reduction of Vβ6+ cells among the injected or de novo-developed donor CD8+ T cells was observed. However, the percentage of Vβ5+ cells among the injected donor CD8+ T cells was still ∼2-fold higher than that among the de novo-developed donor CD8+ T cells (p < 0.05). These results indicate that host-reactive donor CD8+ T cells among the injected ones are only partially deleted.
Taken together, the GVL activity in the long-term anti-CD3-conditioned chimeras is most likely mediated by the residual memory host-reactive donor CD8+ T cells from the injected donor CD8+ T cells.
Expansion of donor CD8+ T cells in anti-CD3-conditioned recipients was limited and confined to host lymphohematological tissues
Next, we explored the GVHD prevention mechanisms in anti-CD3-conditioned recipients. First, we used in vivo BLI to visualize the kinetics of migration and expansion of donor CD8+ T cells in TBI- or anti-CD3-conditioned recipients. Accordingly, TBI- or anti-CD3-conditioned recipients were injected with donor CD8+ T cells from luciferase transgenic (luc+) FVB/N donors (20 × 106) in combination with BM cells (100 × 106) from nontransgenic donors on day 0. luc+CD8+ T cells were injected again on days 5 and 10 after HCT. The recipients were checked with BLI every other day for the first 2 wk and then every 5 days for up to 40 days after HCT. The expansion of donor luc+CD8+ T cells was reflected by the intensity of BLI signals.
As shown in Fig. 3⇓, in TBI-conditioned recipients, the luc+ CD8+ T cells expanded rapidly, and only 5 days after HCT, their BLI signals reached the first peak, and almost covered the whole body of the recipient. Thereafter, the intensity of BLI signals fluctuated and gradually subsided, reaching the lowest level around day 15 after HCT. However, from then on, the luc+CD8+ T cells expanded again, and their BLI signals reached the second peak around day 40 after HCT (Fig. 3⇓, A and B). The recipients all developed GVHD and the expansion levels of luc+CD8+ T cells were correlated with the severity of signs of GVHD (data not shown). Forty days after HCT, BLI was not continued with the TBI-conditioned recipients due to the severe GVHD and anesthesia-related death.
In contrast, compared with TBI-conditioned recipients, expansion of luc+CD8+ donor T cells in the anti-CD3 conditioned recipients was much weaker and slower (Fig. 3⇑, A and B). The BLI signals of luc+CD8+ T cells in these recipients reached a small peak around day 13 after HCT, covering only part of the abdominal area of the recipients, and the peak intensity was >2-fold lower than the peak of TBI-conditioned recipients (p < 0.001). Thereafter, the BLI signals gradually subsided and only weak signals were seen over the area of lower abdomen by day 40 after HCT (Fig. 3⇑, A and B), which was further reduced in long-term chimeras but still detectable 120 days after HCT (data not shown). The recipients showed no signs of GVHD during the whole observation period of >120 days (data not shown). Although the BLI signals of the injected luc+CD8+ T cells in anti-CD3-conditioned recipients faded by day 40 after HCT, donor-type CD8+ T as well as CD4+ T, and B cells, and granulocytes/macrophages were abundant (Fig. 3⇑D), indicating that donor hemopoietic progenitors have engrafted.
We then used ex vivo BLI to identify the tissue location of luc+CD8+ T cells in the TBI- or anti-CD3-conditioned recipients. As shown in Fig. 3⇑C, 5 days after HCT, strong signals from luc+CD8+ T cells were observed in lymphohematological tissues such as spleen, inguinal LN, mesenteric LN, and Peyer’s patches of TBI-conditioned recipients, and also in all GVHD target tissues, including gut, lung, liver, and thymus. However, in the anti-CD3-conditioned recipients, BLI signals from luc+CD8+ T cells were mainly detected in mesenteric LN, and weakly detected in thymus, Peyer’s patches, and inguinal LN, but were nearly undetectable in GVHD target tissues such as gut, lung, or liver. Even at the peak time point, day 13 after HCT, the BLI signals of donor luc+CD8+ T cells were limited to mesenteric LN and spleen and were not detected in GVHD target organ tissues. Taken together, the results indicate that expansion of donor CD8+ T cells in anti-CD3-conditioned recipients is limited and confined to host lymphohematological tissues.
Delayed conversion to complete donor T cell chimerism and lack of up-regulation of tissue homing receptors by donor CD8+ T cells in anti-CD3-conditioned recipients
Next, we tested whether the different tissue distribution of donor CD8+ T cells early after HCT in TBI- or anti-CD3-conditioned recipients was related to the differential status of T cell chimerism and differential expression of α4β7 and chemokine receptors. Due to the limited availability of good anti-mouse chemokine receptor mAbs, we used only anti-CCR7 and anti-CXCR3 for the current study. Accordingly, the percentage of donor T cells in blood, spleen, and LNs of TBI- or anti-CD3-conditioned recipients was measured, and the expression of CD62L, CD44, α4β7, CCR7, and CXCR3 by donor CD8+ T cells from spleen and LNs of the recipients at the peak expansion time points (days 5 and 13) after HCT were analyzed. As shown in Fig. 4⇓A (top row), 5 days after HCT, T cells in the blood of TBI-conditioned recipients were almost all donor-type already, but T cells in the blood of anti-CD3-conditioned recipients contained both donor and host type, and donor-type T cells were only ∼50%. In addition, the percentage of donor-type T cells among blood mononuclear cells of the TBI-conditioned recipients was ∼10-fold higher than that of the anti-CD3-conditioned recipients (p < 0.01). However, day 13 after HCT, 99% of T cells in the anti-CD3-conditioned recipients were also donor type, and the difference in donor T cell percentage among blood mononuclear cells between the TBI- and the anti-CD3-conditioned recipients was only ∼2-fold, although it was still significant (p < 0.01). Similar results were observed with spleen and LN cells (data not shown). These results are consistent with the above BLI results that show a rapid expansion of donor T cells in TBI-conditioned recipients and a slow expansion in anti-CD3-conditioned recipients. These results also indicate that there is a delayed conversion to complete donor T cell chimerism in anti-CD3-conditined as compared with TBI-conditioned recipients.
Furthermore, a majority (∼75%) of donor spleen CD8+ T cells before HCT showed CD62LhighCD44low naive T cell phenotype. However, 5 days after HCT, the majority (∼84%) of donor CD8+ T cells in the spleen of TBI-conditioned recipients showed CD62lowCD44high effector/memory phenotype. In contrast, only a small portion (∼33%) of the donor CD8+ T cells in the spleen of anti-CD3-conditioned recipients showed CD62Llow phenotype, although most (∼96%) of them showed CD44high activation marker. Interestingly, by day 13 after HCT, most (∼91%) of the donor CD8+ T cells in anti-CD3-conditioned spleen showed CD62lowCD44high effector/memory phenotype (Fig. 4⇑A, second row). It was reported that down-regulation of CD62L was associated with cell division cycles (47). Thus, the delayed down-regulation of CD62L was consistent with the above observation that donor CD8+ T cells in anti-CD3-conditioned recipients expanded slower than those in TBI-conditioned recipients.
Expression of α4β7 integrin by donor spleen CD8+ T cells was not detectable before HCT. However, 5 days after HCT, >35% of donor CD8+ T cells in the spleen, mesenteric LN, Peyer’s patches, and inguinal LN of TBI-conditioned recipients became α4β7+. In contrast, at the same time, α4β7+ cells were almost undetectable among CD8+ T cells in those tissues of anti-CD3-conditioned recipients (Fig. 4⇑, A, third row, and B). Even at the peak expansion time point day 13 when ∼90% of donor CD8+ T cells had become CD62LlowCD44high effector/memory cells in anti-CD3-conditioned recipients, there were only ∼3% of α4β7+ cells (Fig. 4⇑A, third row). We should point out that α4β7+ cells among donor CD8+ T cells declined to ∼10% in the spleen and LNs of TBI-conditioned recipients by 13 days after HCT (Fig. 4⇑A and data not shown). This reduction might be due to the migration of α4β7+CD8+ T cells to gut nonlymphoid tissues.
Expression of CXCR3 was observed on a portion (∼23%) of donor spleen CD8+ T cells before HCT, which were found to be CD62LlowCD44high memory cells (data not shown). However, 5 days after HCT, 45–80% of donor CD8+ T cells in the spleen and LNs of TBI-conditioned recipients became CXCR3+. In contrast, at the same time, there was no up-regulation of CXCR3 by donor CD8+ T cells in anti-CD3-conditioned recipients as compared with pre-HCT (Fig. 4⇑, A, bottom row, and B). Similar to α4β7 expression, 13 days after HCT, there was still no up-regulation of CXCR3 expression on donor CD8+ T cells in anti-CD3-conditioned recipients, and there was a decline of CXCR3 expression in TBI-conditioned recipients (Fig. 4⇑A). In addition, a majority of donor CD8+ T cells before HCT were CCR7+, but 5 days after HCT, the majority of donor CD8+ T cells in both anti-CD3- and TBI-conditioned recipients were CCR7−, and no significant difference was observed (data not shown).
Taken together, although donor CD8+ T cells in anti-CD3- conditioned recipients were all fully activated by day 13 after HCT, they did not up-regulate α4β7 integrin or chemokine receptor CXCR3.
The injected donor CD8+ T cells in anti-CD3-conditioned recipients were rendered unresponsive, anergic, FoxP3+, or Tc2 phenotype
As we mentioned above, donor CD8+ T cells in TBI-, but not in anti-CD3-conditoned recipients, showed the second wave of expansion 20 days after HCT (Fig. 3⇑, A and B). The lack of second expansion wave in anti-CD3-conditioned recipients was unlikely due to the blocking effect of residual anti-CD3, because in vitro blocking assay showed that there was little anti-CD3 in the mouse serum 5–7 days after anti-CD3 injection (data not shown). In addition, we found that donor and host CD8+ T cells showed TCRαβ staining already 5 days after HCT (Fig. 4⇑A). T cells would not show TCRαβ staining if sufficient anti-CD3 exists in the serum due to the anti-CD3 activation-induced down-regulation of TCRαβ. Therefore, we compared the in vitro polyclonal proliferation capacity of C57BL/6 donor CD8+ T cells from the two kinds of recipients. Twenty days after HCT, almost all spleen T cells in both recipients were donor type, and were predominantly (>94%) TCRαβ+CD8+ T cells (Fig. 5⇓A), and all the TCRαβ+ cells were CD3+ (data not shown). The donor CD8+ T cells were sorted and stimulated with anti-CD3 and CD28. We found that the proliferation of donor CD8+ T cells from TBI-conditioned recipients was 10-fold higher than those from anti-CD3-conditioned recipients (Fig. 5⇓B, p < 0.001). Addition of IL-2 to the culture increased the proliferation of CD8+ T cells from anti-CD3-conditioned recipients by 3-fold (p < 0.01), although it was still 5-fold lower than that of donor CD8+ T cells from TBI-conditioned recipients (p < 0.01).
In addition, the polyclonal proliferation of the injected donor CD8+ T cells (CD45.1−) from anti-CD3-conditioned long-term (>120 days after HCT) chimeras were compared with that of the de novo-developed donor-type CD8+ T cells (CD45.1+). We found that the proliferation of the injected donor CD8+ T cells was 10-fold lower than that of de novo-developed ones, and addition of IL-2 increased the proliferation of both by 10-fold (p < 0.001, Fig. 5⇑C).
Furthermore, we compared the percentage of apoptotic annexin V+CD8+ T cells among the injected donor CD8+ T cells from anti-CD3- or TBI-conditioned recipients 20 days after HCT, and no significant difference was found (data not shown). Taken together, these results indicate that some donor CD8+ T cells in anti-CD3-conditioned recipients have become unresponsive or anergic.
Because anergy and apoptosis could not adequately account for the unresponsiveness of donor CD8+ T cells from anti-CD3- conditioned recipients to anti-CD3/CD28 stimulation, we tested whether suppressor cells existed among them. Thus, donor CD8+ T cells from anti-CD3-conditioned recipients were added to the culture of donor CD8+ T cells from TBI-conditioned recipients, which were stimulated with anti-CD3 and CD28. We found that the CD8+ T cells from anti-CD3-conditioned recipients significantly suppressed the proliferation of those from TBI-conditioned recipients (p < 0.05, Fig. 5⇑D). The C57BL/6 donor CD8+ T cells from anti-CD3- or TBI-conditioned recipients were also added to an MLR culture of donor C57BL/6 responders and host BALB/c stimulators, and we found while donor CD8+ T cells from anti-CD3-conditioned recipients suppressed the MLR in a dose-dependent manner, those from TBI-conditioned recipients augmented MLR (Fig. 5⇑E). No suppression was observed when donor CD8+ T cells from anti-CD3-conditioned recipients were added to an MLR culture with third-party B10.A stimulators (data not shown). These results indicate that there are donor-specific suppressor cells among donor CD8+ T cells from anti-CD3- conditioned recipients.
Because expression of FoxP3 is one of the characteristic features of T suppressor cells (48), and FoxP3+CD8+ T suppressor cells were observed in tolerant organ transplantation recipients (49), we also measured the donor FoxP3+CD8+ T cells in the recipients, using anti-FoxP3 mAb. We found that there was a 4- or 7-fold increase in percentage and a 10- or 25-fold increase in yield of FoxP3+ cells among donor CD8+ T cells from the spleen of anti-CD3-conditioned recipients as compared with that of donors before HCT or TBI-conditioned recipients (p < 0.01, Fig. 5⇑F and Table II⇓). These results indicate that donor FoxP3+CD8+ T suppressor cells may have been expanded or generated in anti-CD3-conditioned recipients.
Finally, we compared the cytokine secretion profile of donor CD8+ T cells from TBI- or anti-CD3-conditoned recipients. We found that donor CD8+ T cells from anti-CD3-conditioned recipients secreted 2-fold less of IFN-γ, but 4-fold more of IL-4, as compared with donor CD8+ T cells from TBI-conditioned recipients (p < 0.01, Fig. 5⇑G), indicating that some donor CD8+ T cells in anti-CD3-conditioned recipients have become Tc2 phenotype.
The injected donor CD8+ T cells from anti-CD3-conditioned recipients mediated GVL activity without clinical signs of GVHD in TBI-conditioned secondary recipients
Next, we tested whether the suppressive donor CD8+ T cells from anti-CD3-conditioned recipients mediated GVL activity in the secondary recipients. Sorted CD8+ T cells (1 × 106) from anti-CD3- or TBI-conditioned recipients 20 days after HCT with C57BL/6 donor TCD-BM cells (5 × 106) were respectively injected i.v. into lethally irradiated (800 rad) BALB/c recipients that were injected i.p. with 2000 BCL1/luc cells at the same time. The control BALB/c recipients were injected with TCD-BM and BCL1/luc cells only. BCL1/luc tumor cells grew rapidly in control recipients as revealed by BLI and killed all the mice by day 55 after injection (Fig. 6⇓). In contrast, BCL1/luc tumor cells were eliminated in recipients injected with donor CD8+ T cells from anti-CD3- conditioned recipients, and the recipients survived for >100 days without clinical signs of GVHD. Although donor CD8+ T cells from TBI-conditioned recipients also eliminated BCL1/luc tumor cells as revealed by in vivo BLI, they also caused GVHD and killed ∼50% of the recipients (p < 0.01, Fig. 6⇓). These results indicate that donor CD8+ T cells from anti-CD3-conditioned recipients mediate GVL activity without clinical signs of GVHD in irradiated secondary recipients.
We have demonstrated here that donor CD8+ T cells mediated GVL activity without clinical signs of GVHD in anti-CD3 conditioned recipients and that the residual memory host-reactive donor CD8+ T cells persisted in the long-term complete chimeras and also mediated GVL activity without clinical signs of GVHD. This persistent GVL activity should be helpful in preventing leukemia or lymphoma relapse.
GVL is initiated by host APCs and alloantigens that also initiate GVHD (4, 29, 30). Thus, separation of GVL effect from GVHD is a very delicate and complicated process. We observed that separation of GVL effect from GVHD in anti-CD3-conditioned recipients involved multiple mechanisms. First, donor CD8+ T cells in anti-CD3-conditioned recipients showed markedly reduced expansion compared with that in TBI-conditioned recipients, and multiple factors may account for the difference: 1) activated host APCs were reported to play a critical role in induction of donor T cell activation and expansion in TBI-conditioned recipients (5). We observed that, early after HCT, DCs in TBI- but not in anti-CD3-conditioned recipients up-regulated expression of costimulatory molecules such as B7-1, B7-2, and CD40 (N. Li and D. Zeng, unpublished data). Resting APCs or APCs without up-regulation of costimulatory molecules induced T cell unresponsiveness and anergy instead of proliferation in vitro (50, 51, 52). Thus, APCs in anti-CD3-conditioned recipients may be less efficient in induction of donor CD8+ T cell expansion. 2) We observed a delayed conversion to complete donor T cell chimerism in anti-CD3- conditioned recipients, so that there were more residual host T cells (including NKT and CD25+CD4+ T regulatory cells) in the anti-CD3 conditioned recipients early after HCT. The host-vs-graft reaction and the host regulatory T cells could inhibit donor T cell proliferation or expansion (53, 54, 55). 3) It was reported that lymphopenia in TBI-conditioned recipients led to homeostatic expansion of donor T cells (56, 57). We observed that, right before HCT, the spleen cell number of anti-CD3-conditioned recipients was at least 2-fold higher than that of TBI-conditioned recipients (C. Zhang and D. Zeng, unpublished data), thus, there was less lymphopenia in anti-CD3-conditioned than in TBI-conditioned recipients. 4) Residual anti-CD3 in the host may partially block the interaction between donor CD8+ T and host APCs early after HCT.
Second, donor CD8+ T cells in anti-CD3-conditioned recipients were confined to host lymphohemological tissues and the confinement was associated with lack of up-regulation of α4β7 integrin and chemokine receptors (i.e., CXCR3) by donor CD8+ T cells. We should point out that another important factor contributing to the donor T cell confinement should be lack of tissue inflammation in anti-CD3-conditioned recipients. It has been reported that tissue inflammation caused by TBI conditioning play an important role in attracting donor T cells to GVHD target organ tissues (58).
The mechanisms for the lack of up-regulation of homing and chemokine receptors by donor CD8+ T cells in anti-CD3- conditioned recipients are not yet clear. It was proposed that DCs in anatomic lymph nodes play a critical role in the induction of homing phenotypes on the resident lymphocytes (6). For example, DCs in Peyer’s patch and mesenteric LN, but not spleen, supported expression of CCR9 and α4β7 integrin by activated CD8+ T cells (47, 59, 60); DCs from GALT, but not from peripheral LNs, specifically possessed enzymes capable of metabolizing retinol (vitamin A) to retinoic acid, which enhanced the expression of α4β7 and CCR9 concurrently on T cells upon activation, but suppressed the expression of skin homing receptors such as CLA and CCR4 (61). In addition, homing receptor expression on T cells was also regulated by some cytokines and chemokines. For example, CLA expression on T cells was induced by IL-2 and IL-12 and inhibited by IL-4 (8, 14). CXCR3 expression was up-regulated by chemokine IFN-γ-inducible protein 10 and cytokine IFN-γ (62, 63). CCR5 and CXCR4 expression on T cells was up-regulated by IFN-γ, but down-regulated by IL-4 and IL-10 (64, 65). Therefore, we hypothesize that lack of up-regulation of tissue homing and chemokine receptors on donor CD8+ T cells in anti-CD3- conditioned recipients is due to the activation status of host DCs in the LNs and the cytokine environment early after HCT, and residual host NKT cells may regulate the host DCs and donor T cells via their cytokines. Anti-CD3-conditioned recipients had a high percentage of NKT cells that secreted high levels of IFN-γ, IL-4, and IL-10 (31).
Third, the injected host-reactive donor CD8+ T cells were partially deleted and the residual ones were rendered heterogeneous: some unresponsive or anergic cells, some Tc2 cells, some FoxP3+ suppressor cells, and some alloreactive effector cells. We hypothesize that, in the anti-CD3-conditioned long-term chimeras, the residual host-reactive CD8+ T effectors mediate GVL activity, and the regulatory T cells prevent the alloreactive effectors from damaging host tissues. Alloreactive Tc2 cells were reported to mediate GVL activity without clinical signs of GVHD (66, 67). Both donor and host FoxP3+CD4+ regulatory T cells were reported to suppress GVHD but retain alloreactive CD8+ T-mediated GVL activity (37, 68, 69).
The mechanisms of generation or expansion of different subsets of donor CD8+ T cells in anti-CD3-conditioned recipients are not yet clear. It was previously reported that alloreactive-transgenic anti-H-2Ld CD8+ T cells were rendered anergic and suppressive in nonirradiated H-2d/b F1 recipients (70). Our previous study showed that donor CD4+ T cells were rendered Th2 cells in recipients that had high percentages of NKT cells among residual host T cells after TLI and ATG conditioning (53, 54). NKT cells have also been shown to induce CD8+ T suppressor cells in an anterior chamber-associated immune deviation model (71, 72). Thus, it is conceivable that donor CD8+ T cells could be rendered unresponsive, anergic, FoxP3+, or Tc2 phenotype in the recipients conditioned with a radiation-free anti-CD3 regimen, because there were also high percentages of residual host NKT cells in the anti-CD3-conditioned recipients early after HCT (31). In addition, anti-CD3 treatment was reported to induce regulatory CD4+ T cells (73, 74).
In summary, the radiation-free anti-CD3-conditioning regimen not only avoids host-tissue inflammation caused by TBI conditioning, but also modifies the homing and chemokine receptor expression and function of the alloreactive CD8+ T cells, so that the host-reactive donor CD8+ T cells mediate GVL activity without causing clinical signs of GVHD in anti-CD3-conditioned recipients.
We thank Stephen Scott and Tammy Huang in our department, Lucy Brown and Claudio Spalla at the City of Hope (COH) Flow Cytometry Facility, and Sofia Loera at the COH Anatomic Pathology Laboratory for their excellent technical assistance; we thank Dr. Richard Ermel and his staff at the COH Research Animal Facility for providing excellent animal care; and we thank Autumn Tate for assistance in manuscript preparation.
C. H. Contag is a scientific founder of Xenogen, and is currently a consultant and chairman of the scientific advisory board of Xenogen.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
↵1 This work was supported by National Institutes of Health (NIH) Grant R21-DK71002 (to D.Z.), a pilot grant from Lymphoma Special Program Grant of Research Excellence (to S.F.), NIH Grant KO1-DK071716 (to Y.-A.C.), and NIH Grant R24-CA92862 (to C.H.C.).
↵2 C.Z. and J.L. contributed equally.
↵3 Address correspondence and reprint requests to Dr. Defu Zeng, Beckman Research Institute, City of Hope National Medical Center, Gonda Building, R2017, 1500 East Duarte Road, Duarte, CA 91010. E-mail address:
↵4 Abbreviations used in this paper: HCT, hemopoietic cell transplantation; GVHD, graft-vs-host disease; TBI, total body irradiation; Tc, cytotoxic T; LN, lymph node; GALT, gut-associated lymphoid tissue; CLA, cutaneous lymphocyte Ag; GVL, graft vs leukemia; BLI, bioluminescent imaging; TCD, T cell depleted; BM, bone marrow.
- Received July 11, 2006.
- Accepted October 27, 2006.
- Copyright © 2007 by The American Association of Immunologists