Surprisingly, antitumor responses can occur in patients who reject donor grafts following nonmyeloablative hemopoietic cell transplantation. In murine mixed chimeras prepared with nonmyeloablative conditioning, we previously showed that recipient leukocyte infusions (RLI) induced loss of donor chimerism, IFN-γ production, and antitumor responses against host-type tumors. However, the mechanisms behind this phenomenon remain to be determined. We now demonstrate that the effects of RLI are mediated by distinct and complex mechanisms. Donor marrow rejection is induced by RLI-derived alloactivated T cells, which activate non-RLI-derived, recipient IFN-γ-producing cells. RLI-derived CD8 T cells induce the production of IFN-γ by both RLI and non-RLI-derived recipient cells. The antitumor responses of RLI involve mainly RLI-derived IFN-γ-producing CD8 T cells and recipient-derived CD4 T cells and do not involve donor T cells. The pathways of donor marrow and tumor rejection lead to the development of tumor-specific cell-mediated cytotoxic responses that are not due to bystander killing by alloreactive T cells.
The clinical efficacy of allogeneic hemopoietic cell transplantation (HCT)3 is limited by treatment-related toxicities, especially graft-vs-host disease (GVHD). Progress has been made toward minimizing transplant-associated complications by using less toxic, nonmyeloablative doses of chemotherapy and/or irradiation before allogeneic HCT (1, 2, 3, 4, 5). Based on a murine model that separates GVHD and graft-vs-leukemia (6, 7, 8), we have recently developed a nonmyeloablative HCT regimen that includes in vivo T cell depletion with pre- and posttransplant antithymocyte globulin, thymic irradiation, pretransplant cyclophosphamide, and a short course of cyclosporine, which is discontinued by 5 wk and followed by donor leukocyte infusion in the absence of GVHD (4, 5, 9). The regimen was well tolerated and led to the development of initial mixed chimerism in all patients. However, a fraction (∼30%) of these patients subsequently lost chimerism (4, 9). Although sustained antitumor responses occurred with the highest frequency among patients who achieved full donor chimerism following donor leukocyte infusion (4, 5, 9, 10, 11), a fraction (∼20%) of the patients who lost donor chimerism enjoyed sustained remissions of advanced hematologic malignancies, without developing GVHD (12, 13, 14, 15, 16). Similar results were observed in another nonmyeloablative HCT study (17).
In our study, patients who rejected their grafts had significantly fewer circulating donor T cells within the first 50 days post-bone marrow transplantation (BMT), but increased numbers of recipient-derived CD8+ T cells by day +35 post-BMT, and recipient-derived anti-donor IL-2-producing and cytotoxic T cell responses were detected in this group by day +35 post-BMT (18). In these refractory lymphoma patients, complete responses occurred concomitantly with and following the loss of donor T cells, the expansion of recipient CD8 T cells, and the development of anti-donor recipient-derived CTL responses, suggesting a role for recipient T cells in the antitumor effects (14, 16, 18). Thus, we hypothesized that in these patients, the immune process associated with donor marrow rejection may induce the generation of antitumor effector cells.
An understanding of the mechanisms involved in the antitumor effects associated with donor marrow rejection could lead to the development of a new strategy for separating GVHD from graft-vs-leukemia. For this purpose, we previously developed a mouse model. In mixed chimeras prepared with nonmyeloablative conditioning, we showed that recipient leukocyte infusions (RLI), administered after T cell-depleting Abs were cleared, induce loss of donor chimerism within 3 wk. Antitumor effects against host-type tumors were seen against tumors injected 1 wk post-RLI, mimicking minimal residual disease. In this model, the injection of the tumor could not be performed before or at the time of BMT because of the prolonged duration of T cell depletion induced by the conditioning regimen (4–5 wk), which would lead to death of the animals before the injection of RLI. Similar tumor protection has been seen when tumor-bearing mice were used as RLI donors and the RLI was purged of tumor cells (T. I. Saito and M. Sykes, unpublished data), suggesting that purged RLI harvested before conditioning and HCT might have the potential to promote antitumor responses in cancer patients.
We previously demonstrated that the antitumor effects of RLI against the BALB/c B cell lymphoma A20 were dependent on prior allogeneic BMT. RLI did not induce antitumor effects in conditioned, non-BMT recipients, confirming the role for the rejection process in the development of the antitumor responses. Although tumor protection was dependent on recipient-derived IFN-γ produced in response to RLI (19), mouse rIFN-γ did not directly affect the proliferation or death of the tumor cells in this model. Thus, we hypothesized that the antitumor effects observed after loss of donor chimerism induced by RLI were mediated by recipient and/or RLI-derived effector cells generated as a consequence of the rejection process. To test this hypothesis, we have evaluated the respective roles of donor-derived, RLI-derived and non-RLI, recipient-derived T cell subsets in the effects induced by RLI. We have also analyzed the origin of the IFN-γ involved in these effects. We demonstrate that donor marrow and tumor rejection leads to the development of antitumor cytotoxic responses, and that these responses are not due to bystander killing of tumor cells by alloreactive T cells.
Materials and Methods
Female B10.BR (H2k) and C.129S7(B6)-Ifngtm1Ts (common name: BALB/c-Ifngtm1Ts) (H2d) (BALB/c IFN-γ-deficient) mice were purchased from The Jackson Laboratory. C.129S7(B6)-Rag1tm1Mom (common name: BALB/c-Rag1tm1Mom) (H2d) (BALB/c RAG1-deficient) mice were expanded in our animal facility from breeder pairs purchased from The Jackson Laboratory. Female BALB/c wild-type (WT) (H2d) mice were purchased from Frederick Cancer Research Facility, National Cancer Institute. Female mice were used in experiments at 8–12 wk of age.
A20 cell line
Bone marrow transplantation
Mixed chimerism was induced in female BALB/c WT, or BALB/c IFN-γ-deficient mice using a nonmyeloablative regimen, as described (7). The regimen consists of in vivo T cell-depleting anti-CD4 (GK1.5) (1.76 mg/mouse) and anti-CD8 (2.43) (1.4 mg/mouse) mAbs (bioreactor culture supernatant produced and purified at the National Cell Culture Center, (NCCC)) administered i.p. on day −5; 200 mg/kg cyclophosphamide (Cytoxan; Mead Johnson) i.p. on day −1; and 7 Gy thymic irradiation from a 60Co source on day 0. Because BALB/c RAG1-deficient mice are deficient in T and B cells, but have increased numbers of NK cells (22) that might induce graft rejection (23, 24), they received 0.44 mg of GK 1.5 mAb and 0.35 mg of 2.43 mAbs (to deplete donor T cells) plus 50 μg of anti-asialo-GM1 mAb (Wako), which inhibited >90% of ex vivo NK cell activity 5 days after administration. Donor bone marrow cells (BMC) were prepared from B10.BR (H2k) mice, as described (7), and 25 × 106 BMC were administered i.v. on day 0.
In vivo T cell depletion of mixed chimeras
In some experiments, BALB/c WT mixed chimeras were randomly assigned to receive or not receive in vivo T cell depletion with anti-CD4 (GK1.5) (1.76 mg/mouse) and/or anti-CD8 mAb (2.43) (1.4 mg/mouse) (both prepared at NCCC) i.p. on day 41 post-BMT, i.e., 1 day before recipient lymphocyte infusions.
RLI and ex vivo T cell depletion
Spleens were prepared, as described (7), before their injection i.v. (30 × 106 splenocytes (SPC) per recipient) on day 42 post-BMT. When RLI were from BALB/c RAG1 knockout (KO) mice, numbers were adjusted to the number of non-T and non-B SPC contained in 30 × 106 SPC from BALB/c WT RLI donors used in the same experiment. In some experiments, T cells were depleted from RLI ex vivo using Ab and complement, as described (25). T cell depletion was analyzed by flow cytometry, and <1% residual cells of the depleted subsets remained. Numbers of injected SPC were adjusted to include the same numbers of CD4 and/or CD8 T cells as those contained in 30 × 106 untreated SPC. In all experiments, animals were randomly assigned to treatment groups and mixed between cages to avoid cage-related bias.
Flow cytometric assessment of chimerism and ex vivo RLI T cell subset depletion
Chimerism in various white blood cell lineages and T cell recovery were analyzed by flow cytometry, as described previously (7). To reduce nonspecific Ab binding, 10 μl of culture supernatant containing mAb 2.4G2 (anti-FcγRII, CDw32) (26) was added to all tubes. Recipients’ cells were labeled with 34-2-12 FITC (anti-H2Dd mAb) (prepared in our laboratory), and chimerism in various cell lineages as well as T cell recovery were analyzed with the following Abs: anti-CD4 PE, anti-CD8β PE, anti-B220 PE (all purchased from BD Pharmingen), and anti-Mac-1 PE (Caltag Laboratories). For analysis of the purity of ex vivo T cell-depleted RLI, SPC were stained with 10 μl of 2.4G2 mAb, non-GK1.5 anti-CD4 FITC, and anti-CD8β PE (both purchased from BD Pharmingen). Nonreactive mAb HOPC FITC and rat IgG2a PE were used as negative controls. Percentages of donor cells were calculated as described (19).
ELISA for the determination of serum IFN-γ levels
Serum was collected 7 days post-RLI injection in groups of treated mice. IFN-γ concentrations were measured using a specific ELISA kit for mouse IFN-γ purchased from R&D Systems, according to the manufacturer’s instructions.
Spleen cells were isolated from mice at various times, and cytotoxicity assays were performed as described (27). Triplicate wells containing 8 × 105 responder SPC with 8 × 105 donor SPC irradiated with 30 Gy, or 105 A20 cells irradiated with 250 Gy, were cultured for 5 days in medium that consisted of RPMI 1640 (Mediatech Cellgro) supplemented with 10% FCS (Sigma-Aldrich), nonessential amino acids (Invitrogen Life Technologies), l-glutamine, sodium pyruvate, penicillin/streptomycin (Invitrogen Life Technologies), HEPES buffer (Fisher Biotech), and 2-ME (Sigma-Aldrich) at 37°C in 7% CO2. Lymphoblasts were generated by 48-h stimulation of SPC with LPS (10 μg/ml) or Con A (2 μg/ml). Cytotoxic activity of SPC was measured against donor or recipient lymphoblasts, or against untreated vs IFN-γ-pretreated A20 cell targets in a standard 4-h 51Cr release assay, as described (27). Briefly, 15 × 106 target cells were labeled with 250 μCi of 51Cr for 1 h, washed, and incubated for 4 h, in triplicate, with responder cells at different responder/target ratios. Spontaneous and maximum release were determined by incubating target cells without effectors in medium alone and in 5% Nonidet P-40 (Sigma-Aldrich), respectively. Spontaneous release was always <25% of maximum. Radioactivity was counted in a gamma counter, and the percent specific lysis was calculated as: ((experimental release − spontaneous release)/(maximal release − spontaneous release)) × 100%.
Survival data were analyzed using Prism software (GraphPad) and the log rank test. Differences between group means were tested using Student’s t test by Microsoft Excel software. A p value < 0.05 was considered to be significant.
The antitumor effects of RLI against the B cell leukemia/lymphoma A20 are mediated by RLI or recipient-derived T cells, but not by donor T cells
To evaluate the role of T cells in the antitumor responses of RLI against the BALB/c (H2d) A20 B cell leukemia/lymphoma, we performed in vivo T cell depletion, using anti-CD4 and anti-CD8mAbs 1 day before the injection of RLI. These mAbs depleted T cell subsets present in chimeras before the administration of RLI, and those supplied by the RLI. As shown in Fig. 1 A, BALB/c mice received the conditioning described above before transplantation of 25 × 106 B10.BR (H2k) BMC. Controls received conditioning alone. Some groups received a second injection of depleting mAbs 1 day before injection of RLI consisting of 30 × 106 BALB/c SPC on day 42 and A20 cells (5 × 105) on day 49.
As shown in Fig. 1 B, tumor survival of T cell-depleted (TCD) chimeras receiving RLI (median survival time (MST) 28.5 days) was significantly shorter than that of mice receiving conditioning without BMT and of non-TCD chimeras that did not receive RLI (MST 40 and 40.5 days, respectively) (p < 0.02), or of non-TCD chimeric recipients of RLI (MST 66 days) (p < 0.0001). These results show the involvement of T cells in the antitumor effects of RLI.
Percentages of recipient and donor CD4+ and CD8+ lymphocytes in non-TCD chimeras, analyzed before and after the time of RLI, are shown in Fig. 1, C and D. In contrast to non-TCD chimeras that did not receive RLI, chimeric recipients of RLI never acquired a measurable donor-derived CD8 T cell population and, by 7 days following RLI (the time of tumor injection), lost the small population of donor CD4 T cells that had been detectable before RLI. These results make it highly improbable that donor T cells could play a role in the antitumor effect of RLI, suggesting that these are mediated by recipient and/or RLI-derived T cells.
Optimal tumor protection requires both CD4+ and CD8+ T cells
In the same experiments as those shown in Fig. 1, we analyzed the role of CD4+ and CD8+ T cell subsets in the antitumor responses of RLI by administering only anti-CD4- or anti-CD8-depleting mAb 1 day before the injection of RLI.
As shown in Fig. 1 E, tumor survival of CD4 TCD chimeric recipients of RLI was similar to that of CD8-TCD chimeric recipients of RLI (MST 49 and 48 days, respectively, NS), which was significantly prolonged compared with survival of conditioned (non-BMT) control mice and non-TCD chimeras not receiving RLI (p < 0.05), but significantly shorter than that of non-TCD chimeric recipients of RLI (p < 0.04). Mice depleted only of CD4 or CD8 cells showed significant tumor protection compared with recipients depleted of both subsets (p < 0.001). Thus, both CD4 and CD8 T cell subsets are required for optimal tumor protection, but each subset induces some protective effect in the absence of the other.
CD8+ T cells mediate graft rejection and IFN-γ production induced by RLI; CD4+ T cells down-regulate IFN-γ production
The antitumor effects of RLI are associated with measurable serum IFN-γ levels and donor marrow graft rejection (19). We evaluated the role of CD4 and CD8 T cell subsets in these effects of RLI by comparing the evolution of peripheral donor chimerism, T cell recovery, and serum IFN-γ levels in mixed chimeras that received CD4 and/or CD8 or no T cell-depleting mAbs 1 day before RLI.
Percentages of peripheral recipient or RLI-derived (H-2Dd+) CD4 and CD8 T cells at various times are shown in Fig. 2, A and B. Chimeras that received CD4 and/or CD8 T cell-depleting mAbs showed complete CD4 and CD8 depletion 7 days post-RLI and reduced percentages of both subsets until 35 days post-RLI. Of note, CD4 TCD chimeras showed significantly increased percentages of peripheral recipient-type CD8+ T cells by day 21 post-RLI compared with all other groups (p < 0.01).
Donor granulocyte chimerism, which, at steady state, is representative of chimerism in all lineages in this model (7) and reflected chimerism in all lineages following RLI (CD4 and CD8 T cells, B cells, and monocytes were also evaluated at each time point; data not shown), is presented in Fig. 2 C. Chimeras that did not receive RLI, CD4 and CD8 TCD chimeric recipients of RLI and CD8 TCD chimeric recipients of RLI maintained similar donor chimerism up to 50 days post-RLI, all decreasing over time. Among those groups, levels of donor chimerism were significantly lower in CD8 TCD chimeric recipients of RLI after day 50 post-RLI (p < 0.05). In contrast, both CD4 TCD and non-TCD chimeric recipients of RLI began to lose donor chimerism by 7 days post-RLI, to a greater extent in non-TCD recipients of RLI (p < 0.01). Complete loss of donor chimerism was observed by 21 days post-RLI in both groups. These results demonstrate a requirement for CD8 cells for RLI to induce rapid loss of chimerism, and suggest that CD4 cells accelerate the loss of chimerism. Additionally, CD4 cells alone may cause a gradual decline in donor chimerism.
We analyzed serum IFN-γ levels 7 days post-RLI. Compared with non-TCD chimeras that did not receive RLI, no significant increase in serum IFN-γ was observed in CD4 and CD8 TCD or in CD8 TCD chimeric recipients of RLI (Fig. 2 D). Non-TCD chimeric recipients of RLI showed significantly increased serum IFN-γ levels compared with chimeras not receiving RLI (p < 0.001), confirming the role of RLI in inducing the IFN-γ production. Surprisingly, however, serum levels of IFN-γ were significantly higher in CD4 TCD chimeric recipients of RLI compared with non-TCD chimeric recipients of RLI (p < 0.01). Thus, following RLI, donor graft rejection was associated with IFN-γ production, and both phenomena were mediated mainly by CD8 T cells. CD4 T cells inhibited the production of IFN-γ by CD8 T cells and controlled the expansion of CD8 T cells.
RLI-derived T cells play a critical role in the host-vs-graft (HVG) reactions and the antitumor effects induced by RLI
We used RAG1 KO mice to explore the role of T cells present in chimeric recipients before the administration of RLI (referred to henceforth as non-RLI recipient-derived T cells) vs that of T cells within the RLI (RLI-derived T cells). More than 10% donor CD4+ cells and >1% donor CD8+ T cells were detected in white blood cells of RAG1-deficient recipients of B10.BR BMT by day 35, and these T cell levels were similar to the total levels of recipient plus donor T cells found in WT recipients at the same time (data not shown). Some groups received WT BALB/c spleen cells (WT RLI) or adjusted numbers of BALB/c RAG1 KO spleen cells (RAG1 KO RLI) on day 42 and A20 cells on day 49, as summarized in Fig. 3 A.
Survivals of nontumor controls and tumor-injected mice are shown in Fig. 3, B and C. A complete absence of tumor protection was observed in WT chimeric recipients of RAG1 KO RLI, which displayed similar survival to WT recipients of conditioning without BMT and to WT chimeras that did not receive RLI (MST 39, 37, and 38 days, respectively, NS). In contrast, WT chimeric recipients of WT RLI, as usual, showed significantly prolonged tumor survival (MST 53 days, p < 0.0001) (Fig. 3,B). RAG1 KO chimeric recipients of RAG1 KO RLI did not show antitumor responses (MST 33 days) (Fig. 3,C). In contrast, tumor survival in RAG1 KO chimeric recipients of WT RLI was significantly prolonged (MST 60 days, p < 0.0001) and was comparable to that of WT chimeric recipients of WT RLI (Fig. 3 C).
Compared with WT chimeras, the level of donor chimerism was significantly higher in RAG1 KO chimeras at all time points analyzed (Fig. 3,D, p < 0.05). Although levels of donor chimerism remained higher in RAG1 KO chimeric recipients of WT RLI than in WT chimeric recipients of RLI by 7 days post-RLI, all mice (WT or RAG1 KO) that received WT RLI lost donor chimerism within 3 wk following RLI (Fig. 3,D). Moreover, T cells in RLI were required to induce loss of chimerism within 21 days, as no loss of chimerism was detected in WT or KO recipients of RAG1 KO RLI (Fig. 3 D).
Fig. 3 E shows that elevated levels of serum IFN-γ 7 days following RLI were detected only in chimeras that received WT RLI, and that the presence or absence of recipient T cells did not influence the level of IFN-γ following WT RLI.
Together, our data demonstrate that RLI-derived T cells are not only required, but are sufficient for the induction of the HVG reactions associated with loss of chimerism, elevated serum IFN-γ levels, and tumor protection induced by RLI.
Differential roles of RLI-derived CD8+ and CD4+ T cells in the antitumor effects, IFN-γ production, and donor graft rejection induced by RLI
To analyze the respective roles of CD4 and CD8 T cell subsets of RLI in the HVG and antitumor responses observed in our model, mice were conditioned and transplanted and received RLI that had been ex vivo treated with either complement alone (non-TCD RLI), or with anti-CD4- and/or anti-CD8-depleting mAbs plus complement (CD4 and/or CD8 TCD RLI) before the injection of A20 tumor cells, as outlined in Fig. 4 A.
The absence of tumor protection in chimeric recipients of CD4 and CD8 TCD RLI (Fig. 4,B) confirms the critical requirement for T cells in RLI for the achievement of antitumor effects. Tumor survival of chimeric recipients of CD8 TCD RLI was significantly prolonged compared with that of non-BMT mice or non-RLI chimeras (MST 42, 35, and 37 days, respectively, p < 0.01), but was significantly shorter than that of chimeric recipients of non-TCD RLI (MST 56.5 days, p < 0.004) (Fig. 4,C), demonstrating a major role for RLI-derived CD8 T cells in the antitumor effect. Surprisingly, however, tumor protection of chimeric recipients of CD4 TCD RLI (MST 52.5 days) was comparable to that of chimeric recipients of non-TCD RLI (Fig. 4 D). Thus, in the presence of non-RLI recipient-derived T cells and RLI-derived CD8 cells, RLI-derived CD4 T cells may not contribute to the antitumor effects of RLI.
CD4- and CD8-depleted RLI neither promoted the production of IFN-γ 1 wk later, nor induced loss of donor chimerism by 3 wk post-RLI (Fig. 4, E and F). However, compared with non-RLI recipients, chimeric recipients of CD4-depleted RLI showed significantly increased serum levels of IFN-γ (Fig. 4 E) (p < 0.002). Moreover, serum IFN-γ levels were not increased in chimeric recipients of CD8-depleted RLI compared with non-RLI recipients or chimeric recipients of TCD RLI. These results demonstrate a requirement for RLI-derived CD8 cells in inducing IFN-γ production. However, at 7 days following RLI, the highest levels of serum IFN-γ were observed in the presence of both RLI-derived T cell subsets (p < 0.05 compared with recipients of CD4-depleted RLI). Therefore, in the presence of RLI-derived CD8 cells, RLI-derived CD4 cells appear to play a role in promoting the production of IFN-γ.
Analysis of chimerism (Fig. 4 F) showed that both CD4 and CD8 RLI T cell subsets were required for donor marrow rejection within the first week following RLI, and that either subset alone in the RLI could induce rejection by 3 wk after RLI.
Differential roles of RLI-derived and recipient-derived IFN-γ in the effects of RLI
We next explored the respective roles of recipient-derived and RLI-derived IFN-γ in the effects mediated by RLI in our model. We compared serum levels of IFN-γ 7 days post-RLI, tumor survival, and chimerism before and after RLI in BALB/c WT or IFN-γ-deficient mixed chimeras that received either WT or IFN-γ-deficient RLI before the injection of A20 tumor cells, as outlined in Fig. 5 A.
As shown in Fig. 5 B, serum levels of IFN-γ 7 days post-RLI were higher in IFN-γ KO chimeric recipients of WT RLI than in WT chimeric recipients of IFN-γ KO RLI (p < 0.04), and these levels in both groups were lower than those in WT chimeric recipients of WT RLI (p < 0.001). Thus, both RLI-derived and recipient-derived IFN-γ are required for maximal production of IFN-γ within the first week following RLI.
Tumor survival of WT chimeric recipients of IFN-γ KO RLI was slightly, but significantly prolonged compared with that of non-BMT WT conditioned mice and of WT chimeras that did not receive RLI (MST 37, 34.5, and 34 days, respectively, p < 0.03) (Fig. 5,C). Tumor survival of WT chimeric recipients of IFN-γ KO RLI was, however, significantly less prolonged than that of WT chimeric recipients of WT RLI (MST 54 days, p < 0.001). Compared with nonprotected IFN-γ KO chimeras that did not receive RLI or IFN-γ KO chimeric recipients of IFN-γ KO RLI (MST 34.5 and 35 days, respectively), tumor survival of IFN-γ KO chimeric recipients of WT RLI was significantly prolonged (MST 43 days, p < 0.001) and was not significantly different from that of WT chimeric recipients of WT RLI (MST 54 days, NS) (Fig. 5 D). Together, these results argue for a predominant role for RLI-derived IFN-γ in the antitumor responses induced by RLI.
Analyses of chimerism in this experiment indicated that both RLI-derived and non-RLI-derived IFN-γ participated in the loss of chimerism (data not shown).
Generation of tumor-specific cell-mediated cytotoxic responses in chimeras that rejected donor marrow grafts and received A20
To determine whether the antitumor effects of RLI were associated with the development of measurable tumor-specific responses in vitro, SPC from mice prepared as described in Fig. 6 A were stimulated by A20 cells or recipient BALB/c SPC, and their proliferation and cytotoxic responses were tested. Naive third party (C57BL/6) SPC were used as positive controls.
At all time points analyzed, antitumor cytotoxic activity was not observed for SPC from chimeras that did not receive RLI (similar to SPC from naive BALB/c and non-BMT control mice), regardless of whether or not tumor was injected (Fig. 6,B, and data not shown). Although not detectable by 1 wk post-RLI (data not shown), significant antitumor cytotoxic responses were detected for 40% of chimeric recipients of RLI and A20 by 3 wk post-RLI and for all RLI recipients surviving the tumor by 12 wk post-RLI (Fig. 6,B). These results contrasted with the absence of detectable antitumor cytotoxicity by SPC from chimeric recipients of RLI that did not receive the tumor (Fig. 6 B).
Cytotoxic responses to A20 were specific for the tumor cells, as no anti-recipient BALB/c LPS blast cytotoxic responses were detected in chimeric recipients of RLI and tumor at 3 or 12 wk post-RLI (data not shown, and Fig. 6 C). Of note, tumor-specific cytotoxic responses were not detectable for SPC from chimeras that received both RLI and tumor in direct anti-A20 target CTL assays (data not shown); in vitro incubation with A20 stimulators was necessary for the detection of these responses.
The antitumor effects of RLI are not mediated by alloactivated recipient cells
A20 cells are known to express Fas and retinoic acid early inducible-1, the ligand of the NK-activating receptor NKG2D (28, 29). We explored the possibility that the antitumor effect induced by RLI could be mediated by alloactivated recipient CTL that could kill tumor cells by a bystander mechanism through such pathways, as described in other tumor models (28, 30, 31, 32).
For this purpose, we asked whether anti-donor-activated CTLs could kill A20 cells in our model. Thus, we compared the antitumor cytotoxic activity of anti-donor-activated SPC from chimeric recipients of RLI and A20, which showed anti-donor cytotoxic responses by 1–12 wk post-RLI (data not shown), with that of anti-donor-activated SPC from non-BMT controls and from naive third party and naive BALB/c mice. A20-stimulated third party SPC were used as positive controls.
As shown in Fig. 7, neither alloactivated SPC from non-BMT mice nor those from chimeric recipients of RLI showed significant antitumor cytotoxic responses following incubation with B10.BR stimulators at any time analyzed. In contrast, weak antitumor cytotoxicity was seen for similarly alloactivated third party and naive BALB/c SPC at all time points (p < 0.05). These data argue against a role for bystander killing of tumor cells by alloactivated cytotoxic cells in our model.
RLI induce donor marrow graft rejection and antitumor effects against host-type tumors in mixed chimeras prepared with nonmyeloablative conditioning. Although IFN-γ was critical for the antitumor effect of RLI, it did not affect the survival and proliferation of A20 tumor cells in vitro (19). Furthermore, studies presented in this work show that recipient CD4 depletion increases IFN-γ levels while decreasing antitumor effects. Thus, the antitumor effect of RLI involves donor and/or recipient effector cells rather than a simple, direct effect of IFN-γ on the tumor cells.
Data presented in this work show that donor T cells were rejected by the time of tumor injection and could not be the effectors of the observed antitumor responses. In contrast, recipient and RLI-derived T cells are now shown to play important roles in the antitumor effects of RLI, and a role for specific antitumor cytotoxic responses is suggested.
The critical role of recipient T cells in the HVG and antitumor effects of RLI was clearly confirmed by two approaches presented in this study. Neither in vivo CD4 and CD8 T cell depleted chimeras that received RLI, nor T cell-deficient (RAG 1 KO) chimeras receiving RAG1 KO RLI showed tumor protection, donor marrow rejection, or IFN-γ production in response to RLI. Similar results were obtained following the administration of T cell-depleted RLI, either by ex vivo CD4 and CD8 depletion or by the use of RAG1 KO RLI donors. Thus, T cells in the RLI are absolutely required for HVG responses and antitumor effects. The natural history of mixed chimerism induced with this regimen is the development of donor-specific tolerance (27), thus explaining the requirement for RLI to break this tolerance and induce HVG responses. In studies not presented in this work, we observed that the strength of the anti-donor alloresponse induced by RLI, as measured by the rapidity of loss of chimerism and IFN-γ levels, correlates with the strength of the antitumor response.
Experiments involving in vivo depletion of CD4 or CD8 T cells demonstrated that CD8 T cells were the major mediators of rejection, which could be accelerated by CD4 T cells. Experiments using RAG1-deficient mice or ex vivo T cell depletion of RLI revealed that donor graft rejection can be induced, albeit more slowly, either by RLI-derived T cells alone or by either one of the RLI-derived T cell subsets in the presence of non-RLI recipient-derived T cells. The observation that chimerism disappeared more quickly following WT RLI in WT than RAG1 KO recipients might implicate a contribution by non-RLI recipient T cells to this rejection. However, rejection may have been delayed because of the increased levels of donor chimerism before RLI in RAG1 KO compared with WT recipients.
Consistent with a contribution from non-RLI recipient T cells, in all of our experiments the strongest anti-donor responses, leading to donor marrow graft rejection within the first week post-RLI, were observed in the simultaneous presence of non-RLI-derived and RLI-derived T cells. We also observed a requirement for non-RLI, recipient-derived IFN-γ for rapid donor marrow rejection (data not shown). Thus, the most efficient rejection of donor marrow might involve collaboration of RLI-derived CD8 T cells that recognize allogeneic donor Ags with non-RLI, recipient-derived CD4 cells and IFN-γ-producing cells.
Experiments involving depletion of both non-RLI, recipient-derived and RLI-derived CD4 or CD8 T cell subsets clearly demonstrated that IFN-γ production was mediated by CD8 T cells, while CD4 T cells down-regulated IFN-γ production and controlled the expansion of CD8+ T cells. These results suggest the presence of CD4 T cells with regulatory function, which may be CD4+CD25+ T regulatory cells (33, 34, 35). Further studies are needed to explore the potential to enhance antitumor effects by depleting this putative regulatory population.
Our studies showed further that RLI-derived CD8 T cells were absolutely required for IFN-γ production within the first week following RLI, but that both non-RLI and RLI sources contributed to IFN-γ levels. Thus, RLI-derived CD8 T cells may produce IFN-γ themselves or may induce its production by both RLI-derived and non-RLI-derived cells. In contrast to mice that received RLI, but were fully depleted of CD4 T cells, non-T cell-depleted chimeras that received CD4-depleted RLI did not show increased levels of serum IFN-γ or CD8 T cell expansion (data not shown), suggesting that the regulatory T cells are non-RLI, recipient CD4 T cells. This hypothesis could also explain the similar levels of IFN-γ in T cell-deficient and WT recipients of RLI. The absence of non-RLI, recipient-derived IFN-γ in the first group might be counterbalanced by the presence of recipient regulatory CD4 T cells in the second group.
Altogether, these data suggest that following RLI, RLI-derived CD8 T cells induced the production of IFN-γ by both non-RLI, recipient-derived and RLI-derived cells, and that non-RLI recipient-derived CD4 regulatory T cells controlled this phenomenon. Although CD8 T cells are a major source or inducer of IFN-γ production, the possible contribution of other cell types (e.g., NK cells) remains to be explored. Our results suggest a helper effect of CD8 cells in RLI, and the nature and target cell population of this help remain to be determined.
Both CD4 and CD8 T cell subsets were required for optimal antitumor responses. The diminished antitumor responses and HVG reactions in the absence of CD8 T cells in the RLI ± the recipient demonstrate an important role for CD8 T cells in RLI in these effects. In contrast, despite donor graft rejection, the presence of high levels of IFN-γ, and the expansion of CD8+ T cells, CD4 T cell-depleted recipients of RLI also showed weak antitumor responses. It is possible that regulatory CD4 T cells down-modulate antitumor effects by controlling the production of IFN-γ, while another population of CD4 cells contributes strongly to antitumor responses, more than counterbalancing the effect of regulatory cells. If this is correct, selective depletion of regulatory cell populations might lead to further enhancement of antitumor effects. Surprisingly, depletion of CD4 cells from the RLI did not diminish their antitumor effects, suggesting that non-RLI recipient-derived CD4 cells play the relevant role in the antitumor effect of RLI. This conclusion must, however, be reconciled with the observation of similar tumor protection from WT RLI in WT and T cell-deficient (RAG1 KO) recipient chimeras, which argues against any role for non-RLI recipient T cells in the antitumor effect when all RLI T cell subsets are intact. We are investigating the possibility that the increased number of NK cells in RAG1 KO mice compensates for the absence of non-RLI, recipient-derived CD4 T cells by contributing to antitumor effects.
Experiments with IFN-γ-deficient mice showed that only RLI-derived IFN-γ, and not non-RLI, recipient-derived IFN-γ, plays a role in the antitumor effect of RLI. Together with the probable role for recipient-derived CD4 T cells, these results suggest that the antitumor effects of RLI are mediated mainly by RLI-derived CD8+ IFN-γ-producing T cells with the help of recipient-derived CD4 T cells. However, the tumor protection observed in the absence of recipient-derived or RLI-derived CD8 T cells suggests that non-RLI, recipient-derived CD4 T cells may mediate significant antitumor responses either by direct killing of A20 tumor cells (20), or via the activation of non-T cells, such as NK cells, which can kill A20 cells (36, 37).
Animals that had received tumor and had rejected donor marrow developed measurable tumor-specific cytotoxic activity. Such responses were seen by 3 wk post-RLI in 40% of animals that received RLI and tumor, corresponding with the typical percentage of animals that showed long-term survival after tumor injection (19), and these responses were seen in all animals that survived their tumors by 12 wk post-RLI. Such responses were not detected in recipients of either RLI or tumor alone, suggesting that priming with both alloantigens and tumor Ags is essential for the generation of this response, whose development correlates with the occurrence of antitumor responses in vivo. Thus, it is likely that these antitumor cytotoxic responses are relevant to tumor protection in vivo. The requirement for ex vivo restimulation with intact tumor cells to be able to detect tumor-specific cytotoxic responses in chimeric recipients of both RLI and tumor suggests that the tumor-specific cytotoxicity may be mediated by central memory rather than effector memory cells (38).
We first hypothesized that alloactivated recipient T cells could kill tumor cells by a bystander mechanism, as described in other lymphoid tumor models (30, 31). Consistent with this hypothesis, A20 expresses Fas and retinoic acid early inducible-1, which have been involved in antitumor responses in mice (28, 32). However, alloactivated T cells from chimeric recipients of RLI, which might express Fas ligand and NKG2D (39, 40), could not kill A20 in vitro in our model. The phenotype of the cytotoxic cells and targets of antitumor killing are currently under investigation. Complex pathways with several possible effector cell types appear to be involved.
These studies advance our understanding of the paradoxical antitumor effects observed after bone marrow rejection. We have now shown that the previously described effects of RLI, i.e., donor marrow rejection, IFN-γ production, and antitumor effects, are mediated by distinct and complex mechanisms involving recipient-derived and RLI-derived T and possibly other effector cells. They suggest that donor marrow rejection induced by RLI is mediated by alloactivated RLI-derived CD4 and CD8 T cells and that these latter break the tolerance of non-RLI, recipient-derived T cells against the donor. The donor rejection process leads to the production of IFN-γ, mainly induced by RLI-derived CD8+ T cells, but also involving additional RLI and non-RLI, recipient cell populations, and down-modulated by non-RLI, recipient-derived CD4+ regulatory T cells. We demonstrate that antitumor responses are dependent on non-RLI, recipient-derived CD4 T cells and IFN-γ-producing, RLI-derived CD8+ and/or possibly other effector cells such as NK cells. These interactions lead to the generation of antitumor cytotoxic responses that are not due to bystander killing of tumor cells by alloreactive recipient effector cells.
Complete elucidation of these mechanisms could lead to ways of enhancing the antitumor effect of this approach, leading to a new strategy for achieving antitumor responses without GVHD. In addition, because RLI have been reported to reverse ongoing GVHD (41, 42, 43), this approach might potentially reverse GVHD while achieving antitumor responses.
We thank Orlando Moreno for expert animal care, Drs. Bimal Dey and Ronjon Chakraverty for helpful review of the manuscript, and Luisa Raleza for expert assistance in manuscript preparation.
The authors have no financial conflict of interest.
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.
This work was supported by National Institutes of Health Grant RO1 CA79989 (to M.S.) and La Fondation de France and La Federation Nationale des Centres de Lutte Contre le Cancer (to M.-T.R.).
Abbreviations used in this paper: HCT, hemopoietic cell transplantation; BMC, bone marrow cell; BMT, bone marrow transplantation; GVHD, graft-vs-host disease; HVG, host vs graft; KO, knockout; MST, median survival time; RLI, recipient leukocyte infusion; SPC, splenocyte; TCD, T cell depleted; WT, wild type.