Rapamycin (RAPA), an inhibitor of cytokine responses, is under investigation in humans for graft-vs-host disease (GVHD) prevention. The mechanisms responsible for GVHD prevention are unknown. We show that RAPA is more effective in inhibiting CD8+ or TCR γδ+ than CD4+ T cell-mediated murine GVHD. To determine how RAPA inhibited GVHD, thoracic duct lymphocytes (TDL) were isolated from recipients of allogeneic donor grafts. Compared with controls, RAPA-treated recipients had a marked decrease in donor TDL T cell number between days 5 and 24 posttransplant. CD8+ T cell expansion was preferentially inhibited. RAPA inhibited Th1 or Th1 cytotoxic (Tc1) cytokines, but not Th2 or Tc2, cell generation. In situ mRNA hybridization also showed that TDL T cells from RAPA-treated mice had a lower frequency of granzyme B+ cells, indicating that RAPA inhibited the generation of CTL capable of mediating cytolysis through the release of granzyme B. In another system, RAPA was found to inhibit the GVL response of delayed donor lymphocyte infusions. Since CD8+ T cells are the primary effectors in this system, these data suggest that RAPA directly interfered with GVL effector cell expansion or function. We conclude that RAPA is effective in inhibiting Th1 or Tc1 cytokine production and CD8+ and TCRγδ+ T cell-mediated GVHD, but abrogates GVL.
Rapamycin (RAPA)3 (Sirolimus), a macrocylic lactone, is a potent pharmacologic immune suppressive agent that inhibits T cell function (1, 2, 3, 4, 5). Although RAPA is structurally similar to FK506 and binds to the same intracellular protein (FK binding proteins) as FK506, it has an entirely different mechanism of action (6, 7, 8). RAPA has a primary effect on lymphokine responses rather than lymphokine production (1, 2, 3, 4, 5). By inhibiting cytokine responses, RAPA prevents the factor-dependent growth of activated T cells, but does not prevent the autocrine production or release of growth factors from activated T cells. As a preventive agent, RAPA has been shown to significantly prolong survival in recipients of many types of solid organ allografts and is currently being studied in clinical trials (9, 10, 11). In BMT studies, we have shown that RAPA administration is highly effective in reducing murine GVHD lethality in several donor-recipient strain combinations differing at MHC loci, minor histocompatibility Ag loci, or MHC and minor histocompatibility Ag loci (12, 13). In addition, we have demonstrated that RAPA is effective in preventing graft rejection (14) and in the treatment (15) of ongoing GVHD.
In several of the GVHD model systems, RAPA-treated recipients had mean weight values identical with recipients of pan-T cell-depleted (TCD) donor grafts (12, 13). It is possible that RAPA administration prevents GVHD lethality by reducing the expansion of alloreactive donor T cells, such that a critical threshold of donor T cells is not reached. Alternatively, RAPA may modify GVHD reactivity by inducing counter-regulatory cells such as Th2 or Tc2 (IL-4- and IL-10-producing) cells or at least skewing alloreactive T cell responses away from T cell responses generally associated with acute GVHD lethality, i.e., Th1 or Tc1 (IL-2- and IFN-γ-producing) cells (16, 17, 18, 19, 20, 21). In addition, RAPA may inhibit the in vivo capacity of cytolytic donor CD8+ T cells or CD4+ T cells to mediate the tissue-destructive effects of GVHD (22, 23, 24, 25, 26). The effect of RAPA on CD4+ vs CD8+ T cell-mediated GVHD is unknown.
The present study was undertaken to determine 1) the T cell types (CD4+, CD8+, TCR γδ+) that are most susceptible to RAPA administration, 2) the effect of RAPA on Th1 or Tc1 and Th2 or Tc2 induction in heavily irradiated allogeneic BMT recipients, and 3) whether the graft-vs-leukemia (GVL) effect mediated by the delayed post-BMT donor splenocyte infusion was adversely affected by RAPA administration. Together, these studies provide the first detailed examination of how RAPA administration alters donor anti-host T cell responses in irradiated BMT recipients of allogeneic inocula.
Materials and Methods
B10.BR/SgSnJ (H2k), C.H-2bm1 (termed bm1), and C.H2bm12 (termed bm12) mice were purchased from The Jackson Laboratory (Bar Harbor, ME). B6 (H2b: CD45.2) mice and B6-CD45.1 congenic mice were purchased from the National Institutes of Health (Bethesda, MD). Adult H-2d/d TCR γδ-expressing transgenic mice (G8) were originally generated by the cross of a homozygote G8 transgenic male to BALB/c females and then backcrossed for five generations before establishing a breeding colony (27). Mice were housed at the University of Minnesota Hospital in a specific pathogen-free facility in microisolator cages. Donors and recipients were 8 to 10 wk of age at the time of BMT.
Bone marrow transplantation
For examining the effects of RAPA on inhibiting GVHD induced by either CD4+ T cells or CD8+ T cells, MHC class II (bm12) or class I (bm1) disparate recipients were sublethally irradiated on day 0 with 6.0 Gy of TBI from a 137cesium source at a dose rate of 85 cGy/min (28, 29). Four to six hours after TBI, recipients were injected with highly purified lymph node (LN) B6-CD45.2 or B6-CD45.1 CD4+ T cells (105 cells) or CD8+ T cells (0.3, 1, or 3 × 106 cells). In other experiments, the donor and recipient strain combination was reversed, and the same procedure was followed. To examine the effects of RAPA on modifying GVHD induced by TCR γδ+ T cells, B6 recipients were sublethally irradiated with 7.0 Gy of TBI on day 0 then injected with purified G8 TCR γδ+ T cells (2 × 106) (30) as described below. For TCR γδ+ GVHD experiments only, mice were given anti-CD8 and anti-NK1.1 (hybridoma PK-136, rat IgG2a, provided by Dr. Gloria Koo, Merck-Sharpe-Dohme, Rahway, NJ) mAbs (400 μg weekly from days −1 through day 28 post-BMT) to prevent host cells from rejecting the BALB/c T cell inocula. Hematocrit values were obtained at periodic intervals as an indicator of the possible bone marrow-destructive effects of infused T cells (28, 29, 30, 31). Although T cell-mediated hemopoietic destruction contributes to the lethality effect of donor T cells in these three different systems, recipients have histologic evidence of GVHD, indicating that the donor T cells have tissue-destructive capabilities as well (28, 29, 30).
To purify LN cells, single cell suspensions of axillary, mesenteric, and inguinal LN cells were obtained (as a source of GVHD-causing effector cells) by passing minced LN through a wire mesh and collecting them into PBS supplemented with 2% FCS. Cell preparations were depleted of NK cells and enriched for CD4+, CD8+, or TCR γδ+ T cells by depletions with anti-CD8 (hybridoma 2.43, rat IgG2b, provided by Dr. David Sachs, Cambridge, MA), anti-CD4 (hybridoma GK1.5, rat IgG2b, provided by Dr. Frank Fitch, Chicago, IL), or both mAb, respectively. mAb-coated T cells were passaged through a goat anti-mouse and goat anti-rat Ig-coated column (Biotex, Edmonton, Canada). The final composition of T cells in the donor graft was determined by flow cytometry and was always found to be ≥94% T cells of the desired phenotype.
Our GVL induction procedure has been described in detail previously (32). In brief, mice undergoing BMT for GVL assessment were conditioned with 800 cGy irradiation from an x-ray source on day −1 and infused with either 8 × 106 T cell-depleted (anti-Thy1.2 (clone 30-H-12, provided by Dr. David Sachs, Cambridge, MA) and complement-treated) allogeneic B10.BR BM on day 0. Donor B10.BR splenocytes were administered i.v. at a dose of 25 × 106 on day 21 post-BMT. On day 28 post-BMT, mice were challenged with an MHC class I+, class II− acute myeloid leukemia line, C1498 (32). C1498 (American Type Culture Collection, Rockville, MD), originally derived as a spontaneous tumor from a B6 mouse, was grown in RPMI (Life Technologies, Grand Island, NY) with 10% heat-inactivated FBS (HyClone, Ogden, UT), 2 mmol/l l-glutamine, MEM amino acids, 1 mmol/l sodium pyruvate, 50 mmol/l 2-ME, and penicillin/streptomycin (33, 34). Frozen stocks were used for experiments. C1498 cells (2 × 105) were given via the i.v. route. All available animals were examined for evidence of GVHD and tumor, and were extensively examined by necropsy. Representative mice were examined for histologic evidence of tumor or GVHD as previously described (32).
Thoracic duct cannulation studies
To induce lethal GVHD mediated by CD4+ and CD8+ T cells, B10.BR recipients were irradiated with 8.0 Gy of TBI delivered by x-ray at a dose rate of 0.41 cGy/min as previously described (12). Recipients were given 10 × 106 splenocytes along with 8 × 106 B6 bone marrow cells treated with anti-Thy1.2 (antibody 30-H-12, rat IgG2b, provided by Dr. David Sachs, Boston, MA) plus complement (Nieffenegger, Woodland, CA) as described previously (12). The bone marrow plus splenocyte inoculum was administered via the caudal vein in a 0.5-ml volume. For TDL isolation, cannulae were inserted in the thoracic duct of recipients at the indicated times post-BMT, and TDL were collected over a period of 18 h (28, 29).
RAPA and cyclosporin A (CsA) administration
RAPA was administered as a suspension in carboxymethylcellulose (CMC; Sigma, St. Louis, MO) (12). A stock solution of 2.5 mg/ml rapamycin in CMC was prepared. For GVHD experiments, RAPA was resuspended in CMC at a final concentration of 0.2% CMC for injections and administered at doses of 1.5 mg/kg/dose i.p. beginning on the day of or the day before BMT and continuing daily for 14 days, and then three times per week until 21 or 28 days post-BMT, as indicated. This RAPA dose and schedule have been shown to be optimal in preventing GVHD in each donor-recipient strain combination tested to date. CMC was administered to controls. CsA, provided by Dr. Arcesse (Sandoz Pharmaceuticals, Hanover, NJ), was prepared from lyophilized material (SandImmune, lot 3905 692; 50 mg/ml), resuspended in 0.2% CMC and administered at doses of 10 mg/kg (13) according to the schedule described above for RAPA. For GVL experiments, RAPA was administered daily for 14 days (days 20–34), then three times per week for an additional 2 wk (through day 50) as described for GVHD experiments.
The T cell, B cell, and granulocyte/macrophage constituency of splenocytes or TDL effluent was measured using mAb directed toward CD4 or CD8, CD45R/B220, and Mac1, respectively. All fluorochrome-labeled mAb, unless indicated, were obtained from PharMingen (San Diego, CA). To determine donor or host origin of B10.BR recipients of B6 donor grafts, single cell suspensions were costained with αH2k FITC-labeled mAb or αH2b-phycoerythrin mAb. For bm1 recipients of B6-CD45.2 CD8+ T cells, splenocyte suspensions were monitored with αCD45.2 (clone 104-2, rat IgG2a) and αCD45.1 (clone A20-1.7, rat IgG2a), both provided by Dr. U. Hammerling (New York, NY). For splenic flow cytometry studies, cells were first incubated with 2.4G2 to block Fc receptors and then incubated with an optimal concentration of fluorochrome-labeled mAb for 45 min at 4°C. Cells were washed three times and resuspended for analysis. TDL were also assessed for the expression of activation Ags (CD25 (IL-2R α-chain), CD40 ligand (gp39) CD69, CD80 (B7-1), CD86 (B7-2), CD132 (OX40), and CD152 (CTLA-4)) or Ags associated with an effector/memory cell phenotype (CD62 ligand (L-selectinlow)) by two- or three-color flow cytometry using FITC-, phycoerythrin-, or biotin (along with SA-PerCP)-conjugated mAb purchased from PharMingen or Becton Dickinson (Mountain View, CA). Irrelevant mAb control values were subtracted from values obtained with relevant mAbs. All results were obtained using a FACScan (Becton Dickinson). Forward and side scatter settings were gated to exclude red cells and debris, and 0.7 to 1 × 104 cells were analyzed for each determination.
In vitro MLR cultures
For quantifying alloantigen responses, TDL T cells from post-BMT or non-BMT B6 controls were mixed with irradiated (30 Gy) αThy1.2 plus complement-treated splenocyte stimulators from B10.BR or B6 mice (29). Cells were suspended in DMEM (BioWhittaker, Walkersville, MD), 10% FCS (HyClone, Logan, UT), 2-ME (5 × 10−5 M; Sigma), 10 mM HEPES buffer, 1 mM sodium pyruvate (Life Technologies), and amino acid supplements (1.5 mM l-glutamine, l-arginine, and l-asparagine; Sigma), antibiotics (penicillin, 100 U/ml; streptomycin, 100 μg/ml; Sigma). Cell responders (105) and irradiated stimulators (5 × 105) were plated into 96-well round-bottom (Costar, Cambridge, MA) plates and placed at 37°C and 10% CO2
Histology and in situ antisense:sense mRNA hybridization of tissues
Mice were sacrificed and autopsied, and tissues were taken for histopathologic analysis. Samples were placed in either 10% neutral buffered formalin and then embedded in paraffin or were embedded in OCT compound and snap-frozen in liquid nitrogen. Organs were scored positive for GVHD if there was single cell necrosis (skin, colon), crypt dropout (colon), periportal infiltrate with acute necrosis (liver), or endothelialitis with a lymphocytic infiltrate (lung) (35). In previous studies, these features were present only in mice with active GVHD and not in irradiated recipients of either syngeneic or αThy1.2 plus complement-treated fully allogeneic BM (35). For in situ hybridization, TDL cytospins were hybridized with riboprobes specific for IFN-γ bases (1–270), IL-10 (392–937), granzyme A (80–910), and granzyme B (239–775) mRNA. The mRNA:RNA probe hybrids were detected using an alkaline phosphatase α-digoxigenin F(ab′)2 Ab and 5-bromo-4-chloro-3-indoyl-phosphate/nitro blue tetrazolium color substrate. The in situ hybridization technique has been described previously (36, 37).
Group comparisons of continuous data were made using Student’s t test. Survival data were analyzed by lifetable methods using the Mantel-Peto-Cox summary of χ2 (38). Actuarial survival and relapse rates are shown. p < 0.05 was considered significant.
RAPA is only modestly effective in inhibiting GVHD lethality induced by CD4+ T cells given to sublethally irradiated MHC class II disparate recipients
To determine whether RAPA could inhibit CD4+ T cell-mediated GVHD, sublethally irradiated bm12 recipients (n = 8/group) were given the minimal uniformly lethal dose of purified CD4+ T cells (105). RAPA-treated recipients had a significantly higher actuarial survival rate than controls (Fig. 1⇓A). However, only 12% of RAPA-treated recipients survived the first month posttransfer, indicating that the benefits of RAPA were modest. While day 14 mean hematocrit values in RAPA-treated mice were significantly higher than those in controls (34 vs 18%, respectively), RAPA-treated recipients developed GVHD-induced aplasia, manifested by a progressive anemia (day 21 mean hematocrit values <25% in RAPA-treated mice). Therefore, despite the fact that we have previously shown that RAPA administration increases hematocrit values above those in non-BMT control mice, RAPA administration could not prevent CD4+ T cell-mediated aplasia. Mean weight curves, reflective of survival rates and hematocrit, demonstrated a 1-wk delay in the onset of a severe GVHD-induced weight loss (Fig. 1⇓B).
To determine whether the effects of RAPA on CD4+ T cell-mediated GVHD were not unique to the particular B6→bm12 system and to optimize the likelihood for finding a beneficial effect of RAPA, we changed our GVHD model to the somewhat less aggressive bm12→B6 GVHD model system. Sublethally irradiated B6 recipients were infused with bm12 CD4+ T cells (105), and a cohort of recipients was given RAPA. Two replicate experiments were performed with similar results, and the data were pooled for analysis. Although RAPA-treated recipients had a significantly (p = 0.0089) higher actuarial survival rate than the controls (25 vs 0%, respectively; n = 16/group), only a minority of RAPA-treated recipients survived long term (data not shown). Compared with controls, RAPA-treated mice had day 14 mean hematocrit values (26 vs 19%) and mean weight curves (not shown) that were only modestly superior. Experiments in both model systems are consistent with the fact that RAPA administration is unable to prevent the majority of sublethally irradiated recipients of CD4+ T cells from succumbing to GVHD-induced lethality.
In contrast to CsA, RAPA inhibits cytokine responses, leaving cytokine production unimpaired. CsA coadministered with RAPA had a dramatic effect on inhibiting CD4+ T cell-mediated GVHD-induced mortality (Fig. 2⇓). While none of the recipients given RAPA or CsA survived beyond 1 mo posttransfer, 75% of RAPA- plus CsA-treated recipients survived the 3-mo observation period. Day 14 mean hematocrit values were highest in the RAPA- plus CsA-treated group than in the recipients given RAPA, CsA, or neither agent (34 vs 28 vs 16 vs 19%, respectively). By day 28, mean hematocrit values in the RAPA- plus CsA-treated group had normalized to 45%, and only one of the mice in the other three groups (mean hematocrit value = 8%) survived to this time point for analysis. Consistent with the survival and hematocrit data, RAPA- or CsA-treated recipients experienced severe GVH-induced weight loss before death (not shown). In contrast, the mean body weight values in RAPA- plus CsA-treated mice exceeded pre-BMT body weights by 2 mo posttransfer. Splenic flow cytometry of RAPA- plus CsA-treated mice showed marked T and B cell lymphopenia (Table I⇓). Mice also had histologic evidence of GVHD predominantly involving the lung and liver and, to a far lesser extent, the colon and skin. Collectively, these data indicate that inhibition of cytokine responses by RAPA alone is insufficient to inhibit CD4+ T cell-mediated GVHD. CsA coadministration is necessary to down-regulate cytokine production. When administered together, RAPA and CsA are highly effective in inhibiting CD4+ T cell-mediated GVHD lethality.
RAPA is highly effective in preventing GVHD-induced lethality by either CD8+ or TCR γδ+ T cells infused into sublethally irradiated MHC class I disparate recipients
To determine whether CD8+ T cell-mediated GVHD lethality was susceptible to RAPA inhibition, sublethally irradiated bm1 recipients were given a uniformly lethal CD8+ T cell dose (106) from B6 donors. Two replicate experiments with pooled actuarial survival data (n = 16 mice/group) were performed with similar results, showing that RAPA is highly effective in preventing CD8+ T cell-mediated GVHD lethality (Fig. 3⇓A). RAPA-treated recipients had a significantly (p = 0.0002) higher actuarial survival rate than controls (94 vs 0%, respectively). Day 14 mean hematocrit values in RAPA-treated mice were significantly higher than control values (38 vs 10%, respectively). Although RAPA-treated recipients regained their pre-BMT body weight by day 46 posttransfer (Fig. 3⇓B), mice were not entirely GVHD free by clinical assessment, as confirmed by histologic assessment at this time (not shown). Splenic flow cytometry of RAPA-treated recipients (n = 4) analyzed ≥2 mo posttransfer showed a mean of 8.5 × 106 donor CD8+ T cells, representing at least an eightfold expansion over input cell number and a dramatic CD4+ T cell lymphopenia (Table I⇑). Thus, RAPA did not preclude donor CD8+ T cell expansion or GVHD target tissue damage.
To quantify the efficacy of RAPA in inhibiting CD8+ T cell-mediated GVHD, CD8+ T cell dose-response studies were performed in the bm1→B6 model. Sublethally irradiated B6 recipients (n = 8/group) were given 0.3, 1, or 3 × 106 purified donor bm1 CD8+ T cells. Cohorts of mice were treated with RAPA. In the bm1→B6 system, the administration of RAPA was associated with a significant increase in actuarial survival rates. There was a statistical trend (p = 0.068) toward a higher actuarial survival rate in RAPA-treated recipients of 3 × 106 CD8+ T cells compared with control recipients of 106 (75 vs 38%, respectively), indicating at least a threefold reduction in the capacity of CD8+ T cells to mediate lethal GVHD. Day 14 hematocrits were significantly (p < 0.001) higher in RAPA-treated recipients compared with their respective controls. Consistent with both the survival and mean body weight data, RAPA-treated recipients of 3 × 106 CD8+ T cells had significantly (p = 0.001) higher day 14 hematocrit values than control recipients of 106 CD8+ T cells (37 vs 28%, respectively).
In other studies, the effect of RAPA on inhibiting CD8+ T cell-mediated GVHD lethality was quantified using the more aggressive B6→bm1 system. RAPA-treated recipients had a significantly higher actuarial survival rate than their respective CD8+ T cell dose controls (n = 8/group). RAPA-treated recipients of 3 × 106 CD8+ T cells had a significantly (p = 0.018) higher actuarial survival rate than control recipients of a 10-fold lower CD8+ T cell dose (3 × 105 CD8+ T cells; 80 vs 0%, respectively). RAPA-treated recipients of 1 × 106 CD8+ T cells had lower mean weight curves and a significantly (p = 0.030) lower 6.5-mo actuarial survival rate than control recipients of 1 × 105 CD8+ T cells (40 vs 80%, respectively). Day 14 mean hematocrit values were 8 and 4% in control recipients and 36 and 26% in RAPA-treated recipients of 1 or 3 × 106 CD8+ T cells, respectively. Control recipients of 0.1 × 106 CD8+ T cells had a day 14 mean hematocrit value equivalent to that in RAPA-treated recipients of 106 CD8+ T cells (35 vs 36%, respectively). Flow cytometry results of day 159 posttransfer splenocytes were similar in the latter two groups (n = 3–4 mice/group; Table I⇑). Recipients in both groups were noted to have modest T and B cell lymphopenia and histologic evidence of GVHD. Similar flow cytometric and histologic findings were observed in both groups, suggesting an equivalent severity of GVHD. Taken together, these data suggest that RAPA administration led to an approximately 10-fold reduction in the GVHD lethality capacity of CD8+ T cells, while the lethality capacity of donor CD4+ T cells was reduced <3-fold.
To determine whether RAPA could inhibit the GVHD-induced lethality of a different type of CTL population, TCR γδ+ T cells, we used our previously described sublethal TBI model system in which the infusion of highly purified allogeneic (H2d) G8 TCR γδ+ T cells into B6 recipients results in lethal GVHD (30). Compared with control recipients of TCR γδ+ T cells, RAPA-treated recipients of G8 transgenic TCR γδ+ T cells (2 × 106) had a significantly (p = 0.00096) higher actuarial survival rate, similar (p = 0.16) to that in recipients not receiving donor T cells (0 vs 88 vs 100%, respectively; Fig. 4⇓A). Although tissues were not available for assessment of GVHD pathology, RAPA-treated mice had clinical and body weight evidence of transient GVHD (Fig. 4⇓B). Compared with control recipients of TCR γδ+ T cells, RAPA-treated recipients had significantly higher day 14 mean hematocrit values comparable to those in irradiation controls (15 vs 25 vs 27%, respectively). Flow cytometry of splenocytes from long term recipients did not reveal the presence of donor TCR γδ+ T cells (Table I⇑). In earlier studies in which high dose TBI was used to generate TCR γδ+ T cell-mediated GVHD in this strain combination, splenic flow cytometry performed on day 9 post-BMT revealed a high proportion of splenic donor-derived TCR γδ+ T cells (12%) and more total TCR γδ+ T cells that localized to the spleen at that time than initially infused (30). The current data suggest that RAPA interfered with donor TCR γδ+ expansion post-BMT. Compared with the irradiation controls, RAPA-treated recipients also had no evidence of T or B cell lymphopenia. RAPA-treated mice, in fact, had an unexplained relative increase in the number of B220+ cells. Thus, RAPA treatment markedly reduced GVHD-induced lethality, but did not prevent a transient GVH reaction.
In vivo administration of RAPA inhibits donor T cell expansion and the generation of Th1- or Tc1-associated molecules but does not inhibit either the up-regulation of T cell activation Ags or the in vitro antihost proliferative capacity of donor T cells isolated post-BMT
In previous studies involving heavily irradiated recipients, we have observed that RAPA administration significantly increased actuarial survival rates in each of four donor-recipient strain combinations differing at MHC class I and II loci, such that 50 to 100% of RAPA-treated mice vs 0 to 12% of controls survived ≥2 mo post-BMT (12, 13). Our data, described above, show that CD8+ T cell-mediated GVHD lethality is more susceptible to RAPA inhibition than is GVHD induced by CD4+ T cells. Because CD4+ and CD8+ T cells both expand during the GVHD response against MHC class I and II disparate recipients, we next asked whether RAPA administration prevented GVHD lethality in full MHC disparate recipients by inhibiting only CD8+ T cell expansion or by inhibiting both CD4+ and CD8+ T cell expansion.
To accomplish this goal, TDL were isolated by thoracic duct cannulation, which was performed at sequential time periods post-BMT. There is a high degree of recirculation of T cells through the lymphatics in normal mice (28, 29, 39, 40, 41). In BMT mice, alloreactive T cells gain access to the lymphatics and can be isolated in high numbers via overnight cannulation. TDL cells isolated early post-BMT are comprised almost exclusively of donor T cells. Since there is a continual migration of alloreactive donor T cells into the lymphatics, TDL T cell enumeration provides the most accurate estimate of T cell expansion possible post-BMT. Moreover, TDL T cell function can be directly assessed without further T cell purification, as would be necessary if splenocytes were analyzed.
To quantify donor T cell expansion, heavily irradiated RAPA or vehicle-treated control B10.BR recipients were given B6 BM plus supplemental splenocytes containing an average of 2.4 × 106 CD4+ T cells and 1.4 × 106 CD8+ T cells. At various time periods post-BMT, mice were cannulated for quantification of TDL T cell production. Two replicate experiments were performed, and data were pooled for analysis (Fig. 5⇓). RAPA administration inhibited the generation of CD4+ and CD8+ TDL T cells, which was particularly evident on days 5 and 7 post-BMT, the time of peak donor T cell expansion. For example, on day 5 post-BMT, control recipients produced about 14.8 × 106 TDL T cells/day (0.62 × 106 TDL T cells/ml/h), while RAPA-treated mice produced only 0.9 × 106 TDL T cells per day, which was a 94% reduction in TDL T cell number per day (Fig. 5⇓A). Compared with controls, CD4+ TDL T cell production was reduced by 91% (Fig. 5⇓B) and CD8+ TDL T cell production was reduced by 95% (Fig. 5⇓C) in RAPA-treated recipients on day 5 post-BMT. However, by day 7 post-BMT, CD4+ TDL T cell expansion was reduced by only 26%, while CD8+ TDL T cell content was reduced by 82% in RAPA-treated mice compared with controls. The greater degree of inhibition of donor CD8+ vs CD4+ T cell expansion in RAPA-treated mice compared with controls was evident through the first 15 days post-BMT. Thus, RAPA administration had a longer lasting effect on inhibiting donor CD8+ vs CD4+ T cell expansion.
Although RAPA administration inhibited donor T cell expansion, T cell activation, as measured by flow cytometry of TDL T cells on days 5, 7, and 9 post-BMT, was not inhibited by RAPA administration. For the purpose of illustration, selective flow cytometry data are shown (Table II⇓). Both CD4+ and CD8+ TDL from vehicle-treated controls and RAPA-treated recipients had high levels of each activation Ag examined (CD25 (IL-2), CD80 (B7-1), CD86 (B7-2), CD134 (OX40), CD152 (CTLA-4)). In addition, CD40 ligand was up-regulated on CD4+ TDL T cells, and CD69 was up-regulated on both CD4+ and CD8+ TDL T cells (data not shown). The only evident difference between the control and RAPA-treated recipients was the lack of CD62 ligand (L-selectin) down-regulation in the latter group. Each of these findings was observed on days 7 and 9 post-BMT and was entirely replicated in a second experiment (data not shown).
TDL T cells isolated from RAPA-treated mice on day 5 post-BMT (flow cytometry data are shown in Table II⇑) were primed to host alloantigens in vivo, as evidenced by the high proliferative response of TDL T cells to host splenocytes, which peaked on day 1 or 2 of culture (Fig. 6⇓). No apparent differences in the proliferative responses between RAPA-treated and vehicle-treated TDL were observed despite the fact that similar proportions of CD4+ T cells were plated per well in the two groups (56 and 49%, respectively). To determine whether TDL T cells from these two groups had different Th or Tc profiles, cytospins of TDL T cells used for the MLR studies were analyzed by in situ mRNA hybridization. The frequency of TDL T cells from RAPA-treated recipients that expressed IFN-γ (Fig. 7⇓) mRNA was lower than that in TDL T cells obtained from vehicle-treated controls. In contrast, the frequency of TDL T cells from RAPA-treated recipients that expressed IL-10 mRNA (Fig. 7⇓) was higher than that in vehicle-treated controls. Consistent with the in situ hybridization findings, TDL T cells isolated from RAPA-treated, but not vehicle-treated control, recipients were skewed away from a Th1 or Tc1 phenotype, as determined by MLR supernatant ELISA assays for IL-2 and IFN-γ (Th1) and IL-4 and IL-10 (data not shown).
In addition to the skewing of TDL T cells away from a Th1 or Tc1 phenotype in RAPA-treated compared with vehicle-treated control recipients, there also was a lower frequency of TDL T cells that expressed granzyme B, a serine protease released from CTL and NK cells that can initiate apoptosis in target cells (42). Because granzyme A is constitutively in CD8+ T cells, the greater degree of reduction in frequency of granzyme B-expressing (10 vs 33%) compared with that of granzyme A-expressing (23 vs 32%) TDL T cells in RAPA-treated recipients is indicative of a preferential inhibition of TDL T cells with induced cytolytic capacity. The lower frequency of TDL T cells expressing Th1 or Tc1 cytokine or granzyme B mRNA persisted through day 9 post-BMT. Each of these findings was confirmed in a second experiment, although TDL T cells were analyzed only on days 7 and 9 post-BMT in that study. In that experiment, the frequency of IL-10 mRNA-positive TDL T cells in the vehicle-treated control was higher than that presented in Figure 7⇑; no increases in the frequency of IL-10 mRNA-positive TDL T cells in the RAPA-treated group over the relative high frequency in the control group were noted. Collectively, the in situ mRNA hybridization and ELISA studies indicate that RAPA treatment skews TDL T cells away from a Th1 or Tc1 phenotype.
In vivo administration of RAPA inhibits the GVL effect of delayed lymphocyte infusion (DLI) against AML cells after allogeneic BMT
Because RAPA was highly effective at inhibiting GVHD lethality in a number of different donor-recipient strain combinations differing at MHC class I and II loci, we asked whether GVL also would be diminished by RAPA. For this purpose, we used a B10.BR→B6 GVL system in which donor splenocytes given on day 21 post-BMT permit the recipient to resist a supralethal dose of AML cells given 1 wk later. DLI has become a mainstay of therapy for leukemia patients relapsing post-BMT (43, 44, 45, 46).
Previously, we had shown that RAPA significantly reduces GVHD lethality in this strain combination, such that 52% of RAPA-treated recipients given 25 × 106 donor splenocytes on day 0 post-BMT survived 2 mo post-BMT vs none of the controls (13). In the present studies, B6 recipients were heavily irradiated (day −1), infused with B10.BR TCD BM (day 0), and then given B10.BR donor splenocytes (25 × 106 cells) on day 21 post-BMT with or without concurrent RAPA (1.5 mg/kg/dose daily from days 20–34 post-BMT then three times weekly through day 50 post-BMT) (Fig. 8⇓). Recipients given supralethal doses of C1498 (2 × 105) cells iv on day 28 post-BMT all died of leukemia by day 51 post-BMT (Fig. 8⇓, A and C). Compared with mice given C1498 alone, mice given DLI and C1498 cells had a significantly (p = 0.00069) higher actuarial survival rate, with a prolongation of the median survival time from 49 to 75 days. RAPA administration removed most, but not all, of the GVL effect of DLI in this model system. As control recipients of DLI and C1498, mice also given RAPA had a significantly (p = 0.0012) lower survival rate and a higher incidence of leukemia (Fig. 8⇓, A andC). A slight GVL effect was still evident, however, since recipients given DLI, C1498, and RAPA had a higher (p = 0.0044) actuarial survival rate than controls given C1498 without either DLI or RAPA. However, the difference in median survival time between these two groups was only 2 days.
The GVH-related side effect of DLI (32, 43, 44, 45, 46) is evident when examining the actuarial survival rate and mean weight curve of recipients of DLI and no C1498 compared with those of the BM control group (0 vs 90% survival, respectively). Mice receiving DLI and C1498 also had clinical and weight (Fig. 8⇑B) evidence of GVHD. Interestingly, RAPA administration prevented the GVH-induced weight loss of DLI. Collectively, these data indicate that RAPA administration reduces both the anti-AML GVL- and GVH-related side effects of DLI after allogeneic BMT.
The major findings of this study are the following 1) RAPA administration inhibits CD8+ or TCR γδ+ T cell more so than CD4+ T cell-mediated lethality in sublethally irradiated allogeneic recipients. The coadministration of CsA, required for the down-regulation of cytokine production, is necessary to inhibit CD4+ T cell-mediated GVHD. 2) RAPA administration reduces the expansion of donor CD8+ and, to a lesser extent, CD4+ T cells in vivo. 3) RAPA administration skews donor TDL T cells away from a Th1 or Tc1 phenotype, which also is associated with a reduction in the frequency of cells expressing granzyme B mRNA. Although Th2 cell generation is favored, TDL T cells exposed to RAPA are stimulated by host alloantigen-expressing cells, as evidenced by the up-regulation of T cell activation Ags and in vitro response to host-type stimulator cells. 4) RAPA administration markedly diminishes the GVL capacity of DLI.
The greater biologic efficacy of RAPA in preventing GVHD lethality mediated by CD8+ and TCR γδ+ T cells than by CD4+ T cells is most consistent with the greater dependency of the former than the latter cell types on paracrine growth factor release. Although CD8+ and TCR γδ+ T cells can produce low levels of cytokines, these cells typically mediate cytolysis and are dependent upon CD4+ T cells for help with activation and expansion. CD4+ T cells respond to triggering stimuli via the release of cytokines that drive their own activation and expansion as well as that of other cell types, including CD8+ and TCR γδ+ T cells. The inhibition of T cell cytokine responsiveness by RAPA can be overcome at high concentrations of IL-2 (47). Autocrine growth factor production by CD4+ T cells should result in higher local concentrations of cytokines to CD4+ T cells than those produced by paracrine stimulation of CD8+ and TCR γδ+ T cells. The fact that CsA coadministration markedly increased the efficacy of RAPA suggests that in situations in which cytokines are limiting, as would occur with CsA inhibition of cytokine production, the effect of RAPA on inhibiting GVHD is most pronounced. Moreover, in a BMT setting in which donor CD4+ and CD8+ T cells can expand, the inhibition of donor CD4+ T cell expansion in RAPA-treated recipients was transient, being most evident only on day 5 post-BMT. In contrast, in the same RAPA-treated recipients, donor CD8+ T cell expansion was consistently suppressed for 15 days post-BMT. These data support the hypothesis that CD8+ T cells are more readily inhibited than CD4+ T cells by RAPA due either to the greater dependence upon or the lower levels of available paracrine growth factors for CD8+ T cell responses.
It is possible that the cells capable of migrating into the thoracic duct lymphatics represent the small population of T cells that have escaped RAPA-mediated inhibition. Consistent with that hypothesis, detailed flow cytometric analysis of CD4+ and CD8+ T cells isolated between days 5 and 24 post-BMT did not show that RAPA administration interfered with the activation of either T cell subset as denoted by the up-regulation of CD25 (IL-2R), CD80 (B7-1), CD86 (B7-2), and CD134 (OX40). Consistent with the flow cytometry data, RAPA administration did not prevent the in vivo priming of donor T cells to host alloantigens, as assessed in an in vitro MLR culture. On the surface, these data differ somewhat from those of other studies demonstrating that T cells exposed to RAPA can be rendered tolerant. Recent studies by Boussiotis also show that RAPA induces tolerance in alloantigen-specific CD4+ T cell clones (48). Other studies by Chen et al. have shown that CD8+ T cells exposed to MHC class I disparate heart allografts under the cover of RAPA have compromises in intracellular signaling pathways dependent upon calcium flux after mitogen stimulation and on tyrosine phosphorylation upon TCR cross-linking (49).
Differences in the effect of RAPA on inhibiting the expansion of CD4+ vs CD8+ T cells also could not be accounted for by differences in the expression of CD152 (CTLA-4), a molecule that can dampen T cell responses. CTLA-4 expression was up-regulated on a similar proportion of CD4+ and CD8+ T cells in RAPA-treated recipients (Table II⇑). Although TDL T cells isolated from vehicle-treated controls down-regulated the expression of CD62 ligand (L-selectin) on both T cell subsets throughout the day 5 to 24 period in each experiment, RAPA-treated recipients maintained a high level of L-selectin expression. High L-selectin expression could be associated with a reduced capacity of these cells to migrate through the appropriate endothelial cell barriers to gain access to relevant GVHD target organs. The reason for the lack of L-selectin down-regulation in RAPA-treated recipients is unknown.
RAPA administration precluded the up-regulation of Th1 or Tc1 cytokine and granzyme B mRNA in TDL T cells. These findings may explain how RAPA administration reduces the GVHD response mediated by two distinct types of T cell populations typically characterized by CTL activity, CD8+ T cells and TCR γδ+ T cells. The frequency of TDL T cells isolated from RAPA-treated recipients that are capable of producing IFN-γ was reduced compared with that of vehicle-treated controls. Conversely, the frequency of TDL T cells capable of producing IL-10 was increased in RAPA-treated recipients (see Fig. 7⇑). Supernatants obtained from MLR cultures established using TDL T cells from RAPA- or vehicle-treated control recipients and irradiated, T cell-depleted, host-type stimulators showed a preferential skewing toward Th2 or Tc2 (IL-10-producing) cells in the former and Th1 or Tc1 (IFN-γ-producing) cells in the latter (not shown). Acute GVHD has been classified as a Th1 or Tc1 disease process in most studies (16, 17, 18, 19, 20, 21). The administration of donor T cells skewed ex vivo toward a Th2 or a Tc2 phenotype can reduce the GVHD capacity of nonmanipulated or Th1-containing donor T cell inocula (16, 17, 18, 19, 20). Granulocyte CSF mobilization of allogeneic peripheral blood cells is associated with the generation of donor T cells with a Th2 phenotype and a reduced in vivo GVHD capacity compared with those of nontreated cells (21).
The TDL T cell in situ hybridization data are consistent with in vivo studies in RAPA-treated recipients of solid organ grafts, demonstrating a skewing toward a Th2/Tc2 and away from a Th1/Tc1 phenotype (50, 51, 52). Th2 or Tc2 skewing may have resulted in the down-regulation of harmful T cell responses. The higher frequency of TDL T cells expressing IL-10 mRNA and the higher IL-10 protein content of MLR supernatants in RAPA compared with controls may be responsible in part for the GVHD protective effect of RAPA. Although the infusion of exogenous IL-10 in two different BMT model systems can accelerate GVHD lethality under certain conditions (53, 54), it is unknown what the operative mechanism(s) is for these effects. More compelling and direct evidence of the T cell immunosuppressive properties has been derived from studies of human (55) and murine T cells exposed to IL-10 in vitro (56). In an MLR reaction, the addition of exogenous IL-10 to a human bulk culture induced anergy that was long lasting (55). In other studies, the exposure of T cells to exogenous IL-10 has been shown to suppress the proliferation of Ag-specific CD4+ T cells in vitro and to prevent colitis induced by pathogenic CD4+ T cells in vivo (56).
Another important finding of our studies is that TDL T cells isolated from RAPA-treated allogeneic recipients post-BMT have a reduced frequency of granzyme B mRNA expression. Granzyme B, a serine protease released from CTL and NK cells, can cause apoptosis in target cells (42). The inhibition of CTL induction by RAPA has recently been shown to be associated with a reduction in granzyme B mRNA expression and enzymatic activity in mouse CTL induced in vitro with anti-CD3 mAb (57). Granzyme B expression in rat cardiac allograft recipients also is reduced by RAPA administration (58). The lack of granzyme B induction in TDL T cells isolated from RAPA-treated recipients may account at least in part for the preferential inhibitory effect of RAPA on CD8+ vs CD4+ T cell-mediated GVHD. In support of this hypothesis, Ley and coworkers have found that MHC class I-dependent GVHD is inhibited when granzyme B-deficient donor CD8+ T cells are infused, whereas MHC class II-dependent GVHD was not altered by infusing granzyme B-deficient CD4+ effectors (22, 26). Fas ligand-mediated lysis of Fas (APO-1: CD95)-expressing GVHD target cells is the primary effector cell mechanism for CD4+-mediated GVHD lethality in MHC class II disparate recipients (22, 26). Therefore, the relative inefficacy of RAPA in inhibiting CD4+ T cell-mediated GVHD compared with CD8+ T cell-mediated GVHD may be due to the differential inhibitory effects of RAPA on granzyme B vs Fas ligand induction.
The RAPA-induced suppression of the GVL effects of DLI also may be related to a shift away from a Th1/Tc1 phenotype, as manifested by the cytokine and granzyme B mRNA expression data. In the DLI model system used, the GVL effect against MHC class I+, class II− C1498 cells is mediated by CD8+ T cells (32). Therefore, GVHD preventive strategies such as RAPA administration that preclude CD8+ T cell expansion or function should have adverse effects on the GVL response of DLI. Fowler and Gress have shown that Tc1-mediated GVL is more potent than Tc2-mediated GVL against myeloid leukemia in mice (59, 60). At the same time, Tc2 recipients had essentially no histologic evidence of GVHD, whereas Tc1 recipients had mild to moderate GVHD.
In summary, we have shown that RAPA administration is highly effective in preventing GVHD mediated by CD8+ and TCR γδ+ T cells, while CD4+ T cell-mediated GVHD is more resistant to RAPA inhibition. RAPA administration inhibited the GVL effect of DLI. These findings are consistent with the potent ability of RAPA to inhibit CD8+ T cell expansion and granzyme B induction as well as to skew TDL T cells toward a Th2/Tc2 phenotype. Our data have important implications for designing human clinical trials of RAPA in the context of allogeneic BMT.
We thank Dr. Suren Sehgal (Wyeth-Ayerst) for providing rapamycin and for his continuing support of these studies.
↵1 This work was supported in part by grants from the National Institutes of Health (Grants RO1AI34495, RO1HL56067, RO1CA72669, and PO1AI35296).
↵2 Address correspondence and reprint requests to Dr. Bruce R. Blazar, University of Minnesota Hospital, Box 109, Mayo Building, 420 S.E. Delaware St., Minneapolis, MN 55455.
↵3 Abbreviations used in this paper: RAPA, rapamycin; BMT, bone marrow transplantation; GVHD, graft-vs-host disease; TCD, T cell depleted; Tc, T cytotoxic GVL, graft-vs-leukemia; TBI, total body irradiation; LN, lymph node; TDL, thoracic duct lymphocytes; CsA, cyclosporin A; CMC, carboxymethylcellulose; low, low level; DLI, delayed lymphocyte infusion; AML, acute myeloid leukemia.
- Received December 2, 1997.
- Accepted February 5, 1998.
- Copyright © 1998 by The American Association of Immunologists