Graft-vs-host disease (GVHD) is a devastating, frequently fatal, pathological condition associated with lesions in specific target organs, including the intestine, liver, lung, and skin, as well as pancytopenia and alopecia. Bone marrow (BM) atrophy is observed in acutely diseased animals, but the underlying mechanisms of hemopoietic stem cell depletion remained to be established. We used an experimental mouse model of acute GVHD in which parental cells were injected into F1 hosts preconditioned by sublethal irradiation. The resulting graft-vs-host response was kinetically consistent, resulting in lethality within 3 wk. We observed disease pathology in the liver and small intestine, and consistent with previous observations, we found BM atrophy to be a factor in the onset of acute disease. The product of the protooncogene, p53, is known to be a key player in many physiological examples of apoptosis. We investigated the role of p53 in the apoptosis of BM cells (BMC) during the development of acute disease and found that at least one copy of the p53 gene is necessary for depletion of BM and subsequent lethality in host animals. BM depletion was preceded by induction of the death receptor, Fas, on the surface of host stem cells, and induction of Fas was coincidental with the sensitization of BMC to Fas-mediated apoptosis. Our data indicate that BM depletion in acute GVHD is mediated by p53-dependent up-regulation of Fas on BMC, which leads to Fas-dependent depletion and subsequent disease.
Apoptotic death of host cells is the underlying cause of disease leading to death from acute graft-vs-host disease (GVHD)3. Death is induced primarily by cytotoxic effector cells from the transplanted graft population. Killing can be attributed to two predominant mechanisms. One is dependent on surface expression of the death ligand, Fas (CD95) ligand, and the corresponding death receptor, Fas, by graft cytotoxic lymphocytes and host target cells, respectively. The other is mediated by components of cytolytic granules in conjunction with its principal protein component, perforin. Although disease progression in acute GVHD involves tissue destruction mediated by cytotoxic donor T cells involving perforin and/or Fas/Fas ligand interactions (1, 2, 3, 4), Fas ligand-mediated cytotoxicity has been reported to be the principal mechanism in lethally irradiated hosts (5). Consistent with the model in which Fas ligand induced apoptosis of host bone marrow cells (BMC), transfer of Fas ligand-defective gld splenocytes showed reduced anti-host cellular destruction (6).
Pathological effects during acute GVHD are commonly observed in the intestine, liver, and skin (7, 8, 9, 10, 11, 12), but bone marrow (BM) depletion is one of the major symptoms in acute GVHD, and marrow atrophy appears to be an underlying cause of lethality (13). Host animals that received low numbers of CD4+ donor lymphocytes along with preconditioning irradiation showed BM atrophy and concomitant hemopoietic failure as a likely cause of death (13). In yet another model, host animals that received a greater dose of donor lymphocytes developed disease due to graft-vs-host reaction (GVHR), albeit inconsistently, without an absolute necessity for additional experimental treatment such as radiation preconditioning (14).
The product of the protooncogene, p53, plays a pivotal role in many forms of apoptosis. p53 is an extensively studied tumor suppressor protein that regulates the cell cycle and is capable of inducing apoptosis (15, 16, 17). Patients with germline mutation in p53 and mice genetically engineered to be deficient in p53 (p53−/−) spontaneously develop tumors with increased frequency (18). Cells derived from mice deficient in p53 have been shown, in vivo and in vitro, to be resistant to apoptosis induced by numerous DNA-damaging agents such as UV light, UV radiation, and chemotherapeutic agents (19, 20). p53 has also been demonstrated to influence the display of the death receptor, Fas, on the cell surface by increased expression (21, 22) or by trafficking of Fas to the plasma membrane from existing intracellular stores (15).
We used an experimental model of acute GVHD to investigate the role of p53 in apoptosis leading to BM depletion and lethality from acute disease. We found that animals with at least one copy of the p53 gene have a high susceptibility to development of lethal disease, whereas p53−/− recipient animals were significantly less affected. Previous studies have shown that, in some models, pretreatment with radiation can sensitize recipient animals to GVHD. In this study, we characterized a mechanism by which radiation sensitized BMC to receptor-mediated apoptosis and demonstrated that this process is dependent on BMC p53 expression. Furthermore, the lethal effects of acute GVHD in our model were completely abrogated if BMC did not possess at least one allele of the p53 gene. In addition to uncovering a molecular mechanism underlying the depletion of BMC in GVHD, these data also support the assertion that BM depletion is a major causative factor in disease progression that ultimately culminates in lethality.
Materials and Methods
Mice and induction of acute GVHD
C57BL/6 (H-2b), B6C3F1/J (C57BL/6 × C3H/He, H-2b×k), C57BL/6 perforin−/−, C57BL/6lpr, and C3Hlpr mice were obtained from The Jackson Laboratory. B6.SJL-PtprcaBoAiTac (B6.SJL) mice were obtained from Taconic. B6C3F1lpr mice were obtained by mating C57BL/6lpr with C3Hlpr mice. Mice were housed under pathogen-free conditions at Lakeshore Veterans Affairs Hospital or La Jolla Institute for Allergy and Immunology. DBA/1-p53−/− mice were generated as previously described (23). Acute GVHD was induced by sublethally irradiating F1 hosts, followed by the injection of C57BL/6 spleen cells (2 × 107 unless noted otherwise) into the peritoneal cavity of recipient mice 4 h later (9, 24). We found that i.p. injection resulted in similar disease responses (compared with i.v.) in our experimental system and provided a more consistent means of introducing cells into large numbers of recipient animals while minimizing handling time. Irradiation of F1 hosts was performed with a cesium source gamma irradiator at a rate of 0.8 Gy/min. Unless otherwise stated, irradiated mice received 4 Gy and all F1 mice, except those that received radiation alone as a control, were injected with 2 × 107 F1 spleen cells. All numerical data were analyzed for statistical differences using the Student t test. Survival data were analyzed by the log-rank test.
Reagents and cells
FITC-conjugated anti-Fas (Jo-2), FITC-conjugated anti-Gr1, FITC- and PE-conjugated anti-CD3ε, FITC-conjugated anti-CD11b (Mac-1), FITC-conjugated anti-B220, anti-CD45.1 and anti-CD45.2, PE-conjugated, anti-c-Kit, and biotin-conjugated anti-Ter 119 mAb, and isotype control IgG for Fas were purchased from BD Pharmingen.
Cell isolation and flow cytometry
Isolation of BMC, intestinal epithelial lymphocytes, and liver monocytes has been previously described (25, 26, 27). Splenocytes were isolated as previously described (9). Blood was isolated from heart after euthanasia. BMC were flushed out from the femur with a fine needle and syringe and filtered through a 70-μm cell strainer. Cells were blocked with appropriate hamster and goat serum to avoid nonspecific binding and subsequently stained with indicated Abs. Background fluorescence was established based on staining with relevant, fluorescently conjugated isotype control Abs. Two- and three-color flow analyses were performed with a FACSCalibur flow cytometer, and data were analyzed with the CellQuest program (BD Pharmingen).
Sensitivity of host BMC to Fas-mediated cytotoxicity
Sensitivity of host-derived BMC to Fas-mediated death was analyzed by isolating the BMC of F1 mice 2 wk after the induction of acute GVHD. BMC were positively sorted for host-derived cells (CD45.2+) with 99% purity using a Coulter EPICS Cell Sorter. A total of 2 × 105 purified host-derived cells was subsequently cultured in 200 μl of RPMI 1640 medium in the presence or absence of soluble Fas ligand (Alexis Chemical) and a cross-linking anti-FLAG mAb as described previously (28). Fas ligand-induced cytotoxicity of host BM-derived cells was analyzed in a double-blind fashion by trypan blue staining. This method was used due to the diminishingly small cell numbers available for analysis. Specificity of Fas ligand-mediated killing was confirmed by inhibition with soluble Fas-Fc fusion protein as described previously (9).
Host p53 is required for lethality from acute GVHD
To study the underlying mechanisms leading to lethality from acute GVHD, we used an experimental model of GVHD wherein lethality occurs consistently within a specific time period. To do so, we introduced parental spleen cells into mismatched F1 animals that had been subjected to a sublethal regimen of total body irradiation. Induction of disease in this manner was consistently and predictably lethal in 100% of animals within 3 wk (n = 50; Fig. 1,A). Mice that were irradiated or received parental splenocytes alone showed no overt signs of disease or mortality. Necropsy analysis following death in acutely diseased animals showed destruction to the liver and small intestine. Interestingly, very low numbers of splenocytes were required to induce lethal effects in irradiated animals, whereas in the absence of radiation pretreatment, vast numbers of splenocytes were required to elicit GVHR, which was often weak and unpredictable and often resolved without significant mortality (Fig. 1 A).
Low-dose radiation is required for lethality in this model, and radiation has been shown to mediate programmed cell death through p53, so we investigated the possibility that host p53 is involved in the GVHR. This prompted us to investigate the role of p53 in acute GVHD. To do so, we crossed DBA/1p53−/− with C57BL/6 (p53−/− and p53+/+) to yield (C57BL/6 × DBA/1)F1 recipients that were p53−/− or p53+/−. We then transferred parental C57BL/6 splenocytes into the mismatched F1 recipients to induce acute GVHD as described above. In stark contrast to the lethality we observed when we transferred wild-type cells into wild-type hosts, p53−/− hosts were refractory to devastating disease and most recovered from treatment without succumbing to subsequent lethality (Fig. 1 B).
BM depletion is an underlying cause of lethality in acute GVHD
It is clear from our observations that p53 plays a critical role in acute GVHD. Therefore, we set out to determine the mechanism by which p53 is involved in disease progression. Sprent et al. (13) demonstrated that BM atrophy is a critical event leading to death in a similar model of acute GVHD using sublethal irradiation; therefore, we assayed BMC apoptosis and BM atrophy in our model system. GVHD was induced in irradiated animals by introducing splenocytes from C57BL/6 mice as described above. To establish the kinetics of BM depletion, we analyzed BM cellularity at 0, 7, and 14 days after induction of disease. As shown in Fig. 2,A, BM depletion occurred only in host animals that had been preconditioned with sublethal radiation and received parental splenocytes, whereas neither of these treatments alone showed any sustained effects on BM cellularity. The atrophy observed in animals that were induced to undergo acute disease (see Fig. 1) occurred 14 days postinduction, which was before the appearance of gross pathology, visible indications of wasting, or lethality. To further characterize the BM depletion observed during acute GVHD, we assayed the surviving pool of BMC for the presence of early hemopoietic precursor cells. There was a marked depletion of the c-kit+lin− population, which is believed to represent pluripotent hemopoietic stem cells (data not shown). F1 animals that received only radiation or parental cells did not show a drop in the c-kit+lin− population.
We verified these previous observations in our model and also found that injection of parental BMC before the induction of GVHD rescued host animals from lethality (Fig. 2,B). F1 mice that were scheduled to undergo acute GVHD were treated 24 h prior by injecting BMC from either parental or F1 animals. Both groups showed an increased survival relative to animals that received no additional BMC, but only parental BMC conferred a complete rescue from lethality (Fig. 2 B). Because each of these treatments involved introducing BMC (or media control) into a similar cytokine environment, it appears that the effects from cytokine storm alone were not sufficient to result in lethal disease. Additionally, BMC appear unique in their ability to prevent lethality from GVHD because injection with either F1 or parental lymph nodes before the induction of GVHD failed to prevent death in any recipient animals (data not shown).
Fourteen days after induction of disease, mice with acute GVHD displayed severe BM atrophy (Fig. 2,A); therefore, we chose this time point at which to investigate the role of host p53 in BM depletion during acute GVHD. We transferred parental cells into radiation pretreated p53−/− and p53+/− littermates and analyzed BM 14 days following the induction. At this time, wild-type animals displayed signs of severe disease but had yet to succumb to lethal effects (see Fig. 1,B). As shown in Fig. 2 C, we observed reduction of BMC in p53−/− animals, whereas there was a severe reduction in the number of cells in the marrow of littermate control animals possessing one copy of the p53 gene.
To exclude the possibility that the number of surviving BMC in p53−/− undergoing GVHD was not simply a replacement of host BMC by donor BMC, we investigated the infiltration of graft lymphocytes into the BM of host animals using congenic B6.SJL splenocytes as the parental (graft) population. These splenocytes bear the CD45.1 isoform, which allowed us to differentiate parental (CD45.1+) from B6DBAF1 (CD45.2+) recipients. Disease was induced in B6DBAF1 host mice as described above. Graft and host cells were quantified 14 days after the induction of disease. p53−/− host animals showed less reduction in marrow cellularity than mice with at least one copy of the p53 gene, and p53−/− mice also displayed a lower percentage of donor-derived cells in the marrow. Consistent with the notion that p53 is involved in the depletion of BM, survival of BMC was predominantly limited to those of the p53−/− origin (Fig. 2 D).
As a means to control for the possibility that radiation was merely causing suppression of BM in p53-competent mice that allowed for increased susceptibility to graft-induced damage, we assessed the level of BM cellularity in irradiated p53+ and p53−/− mice. Mice were subjected to total body irradiation as above and rested for 14 days, a time at which significant BM atrophy was observed in GVHD-treated animals (Fig. 2,A). Total cellularity of the marrow was not significantly altered in either p53+ or p53−/− animals relative to the mock-treated group (Fig. 2,E). Additionally, there were no significant changes in the percentage of CD34+lin− cells in the BM (Fig. 2 F) or in the c-kit+lin− population (data not shown).
The use of B6.SJL graft splenocytes described above also allowed us to assess the degrees of infiltration into other tissues. At 14 days postinduction, significant infiltration into peripheral tissues had begun to occur in disease-destined animals, but there were no visible signs of disease in the liver, small intestine, or spleen (data not shown). Transfer of host cells in the absence of radiation pretreatment, a condition that was consistently resolved by the host, did not result in the presence of graft lymphocytes in the target organs analyzed. Although the absolute number of BMC declined more dramatically in p53+/− hosts, which also showed a higher percentage of parental cells in the BM, absolute numbers of graft cells were similar in hosts of each genotype, suggesting that engraftment was not was adversely affected in p53−/− hosts.
Transfer of p53−/− BM before induction of disease reduces lethality in acute GVHD
p53−/− mice, which are not susceptible to lethal disease, displayed less of a reduction in the number of BM cells compared with p53+/+ or p53+/− mice. To establish further the role of BMC p53 in the development of acute GVHD, we tested the hypothesis that p53−/− BMC might rescue host animals from the lethal effects of disease. We introduced p53+/− and p53−/− F1 BM into p53+/− host animals that were then treated to undergo acute GVHD. Transfer of p53−/− BM 24 h before the induction of acute disease (i.e., by radiation pretreatment and transfer of parental splenocytes) delayed onset of disease as measured by lethality and resulted in significantly increased long-term survival (Fig. 3). p53+/− F1 animals that received p53+/− BM displayed similar sensitivity to the lethal effects of induced GVHD as observed for animals that were induced to undergo GVHD without additional treatment (Fig. 1,A), although there was a modest increase in survival (Fig. 3).
p53 is required for the induction of BMC Fas
How p53 is involved in the lethality in acute GVHD is not clear. We and others have shown that apoptosis induced by the Fas ligand/Fas system is a major contributor to cell death and disease in similar acute GVHD models (5, 6, 9, 14). In agreement with these previous studies and as shown in Fig. 4,A, both Fas and perforin are involved in lethality in this acute GVHD model, but previous studies suggest that Fas ligand-mediated killing predominates in this experimental model (5). p53 has been shown to increase sensitivity to Fas-mediated killing via transcriptional regulation of the Fas gene and/or by stimulating display of Fas protein on the cell surface by releasing cellular stores (15, 21, 22). Because BM depletion preceded lethality and recent evidence suggests that p53 can regulate Fas, we investigated the involvement of p53 in the induction of Fas-mediated cytotoxicity in BMC during GVHD. As a first approach, we assayed the sensitivity of BMC to Fas-induced apoptosis. BMC were isolated and incubated in the presence of soluble Fas ligand. We found that cells from control animals (i.e., those not induced to undergo GVHD or GVHR) showed no appreciable sensitivity to Fas-mediated apoptosis (Fig. 4,B, □). However, BMC from GVHD animals (14 days) showed an exquisite sensitivity (Fig. 4 B, ▪).
To test the possibility that the lack of sensitivity of p53−/− BMC to Fas-mediated apoptosis was a function of Fas surface expression (or rather a lack thereof), we measured the surface expression of Fas on BMC from GVHD and control animals. Consistent with the lack of sensitivity to soluble Fas ligand, we observed little or no expression of Fas on the surface of BMC from animals that had not received radiation preconditioning (Fig. 4,C). Because we observed that Fas surface expression (and concomitant sensitivity to Fas-mediated apoptosis) was increased on BMC during conditions leading to lethality from acute GVHD, we compared the induction of surface Fas between p53+ and p53−/− hosts animals 14 days after they were induced to develop acute GVHD as described above. In keeping with our previous observations, p53-null animals did not display the same induction of surface Fas relative to that observed on BMC possessing at least one copy of the p53 gene (Fig. 4 D). Surface Fas on p53−/− BMC remained at a level similar to isotype control staining (data not shown).
BM depletion is one of the precipitating factors leading to lethality in this murine model of acute GVHD. Our data provide insights into the underlying mechanism by which BMC become susceptible to cytotoxic attack by graft-derived lymphocytes. We used a model of acute GVHD in which parental spleen cells were injected into radiation-conditioned mismatched F1 mice. Radiation increases the susceptibility of host BMC to Fas-mediated apoptosis. This susceptibility is mediated either directly or indirectly by a p53-dependent increase in surface expression of Fas on BMC, and this results in an increased sensitivity to induction of death mediated by the expression of Fas on the surface of host BMC.
In addition to low Fas levels, hemopoietic progenitor cells have also been reported to express the Fas-signaling inhibitor, c-Flip (29), and Bcl-2, an inhibitor of mitochondrial apoptosis pathways (30). p53 has been implicated in the sensitization of BMC after chemotherapy, which involved slight down-regulation of Bcl-2 expression and induction of Bax (31), but it was indicated that other mechanisms were also involved. It is evident that Fas up-regulation alone might not be sufficient to impart sensitivity, and it is likely that regulation of other components of the receptor-mediated death pathway may also need to be coordinated to impart sensitivity to Fas-mediated apoptosis.
Our observations support the earlier finding that a nonspecific host response is mediated by radiosensitive cells (32), which are eliminated (along with the response) by sublethal irradiation. It is interesting to note that injection of parental spleen cells into nonirradiated F1 hosts primarily induces an expansion of host-derived spleen cells, presumably due to a MLR (data not shown). Although our results suggest that radiation enhances lethality due to acute GVHD by increasing the infiltration of cytotoxic donor-derived cells into the host marrow, this may not be the only mechanism by which radiation increases susceptibility to lethality. Our results with gld host mice suggest that Fas-mediated cytotoxicity plays an important role (although clearly not the only mechanism at play) in this model of acute GVHD (Fig. 4,A). We found that host BMC normally express very low levels of Fas and are resistant to Fas-mediated death (Fig. 4, B and C). Interestingly, however, the combination of radiation and parental spleen cells resulted in increased infiltration of cytotoxic parental cells into the F1 BM and host marrow cells expressed markedly increased levels of Fas and increased susceptibility to Fas-mediated apoptosis.
Our studies clearly indicate a role for p53 in acute GVHD; however, exactly how host p53 contributes to the pathogenesis of disease is not exactly clear. It is unlikely that p53 contributes to the severity of acute GVHD through its direct ability to mediate apoptotic death of host cells found in organs specifically targeted during acute GVHD. p53 has been shown to mediate radiation-induced apoptosis of thymocytes, BMC, and crypt epithelial cells in the small intestine (intestinal epithelial cells (IEC)) (33, 34, 35, 36); however, the effects of p53-dependent apoptosis disappear within 1 wk after exposure to radiation. Irradiation of mice results in about a 25% decrease in the total number of BMC within 3 days after irradiation, but the total number of BMC returns to normal within 1 wk (Fig. 4,E and data not shown). Similarly, Merritt et al. (35, 36) reported that exposure to 4 Gy induced crypt IEC apoptosis within 4 h in a p53-dependent manner, but apoptotic crypt IEC disappear within 1 wk after exposure to radiation. In contrast to this limited response to radiation, most of the adverse outcomes seen in acutely diseased animals occur significantly later, notably BM depletion and lethality, which occur 2–3 wk after induction of acute GVHD (Fig. 1). Overall, our results suggest that radiation, acting via p53, sensitizes BMC to Fas-mediated apoptosis, which subsequently leads to BM depletion and death of host animals, and this occurs in a manner that is distinct from the immediate and transient apoptotic effect of p53.
Radiation increases the susceptibility of host cells to Fas-mediated death (37) and p53 appears to either directly or indirectly control the up-regulation of Fas on host cells during acute GVHD. Host BMC from p53+/− F1 mice show significant up-regulation of surface levels of Fas 2 wk after the induction of acute GVHD. In contrast, host BMC from p53−/− F1 mice induced to undergo acute GVHD show only a modest increase in surface expression of Fas and do not become as susceptible to Fas-mediated death (Fig. 4). Overall, these results suggest that host p53 mediates the up-regulation of Fas and the subsequent increased sensitivity to Fas-mediated death of host cells that normally occurs during acute GVHD. The conclusions based on these results are consistent with the observation that p53 can increase the susceptibility of cells to Fas-mediated death by regulating Fas expression or by translocating Fas from the Golgi apparatus to the cell membrane (15, 21, 22).
However, it would be naive to imply that p53-mediated translocation of Fas to the cell surface is the only mechanism that could explain our findings. Emerging studies strongly suggest that host APC are important in initiating GVHD in both MHC class I and II disparate models (38, 39). Hence, it is possible that sublethal irradiation may enhance the effect of GVHD through its effect on host APC, a population that is relatively resistant to radiation-induced apoptosis. We have shown that sublethal radiation induces the systemic production of numerous proinflammatory cytokines such as TNF (40), which is primarily produced by APC, and this induction is attenuated in the p53−/− mice. Hence, it is plausible that radiation induces an initial critical inflammatory response through its effect on p53, and this is further propagated by the injection of parental cells recognizing alloantigen in a concept commonly regarded as cytokine storm (41). The inflammation may then induce the up-regulation of Fas. If cytokine storm alone were responsible for lethality in this model, we would not expect the injection of F1 BMC (which were not irradiated) into GVHD mice to be effective in preventing lethality. Although our data do not argue against the occurrence of cytokine storm, they demonstrate that the environment in which it occurs does not affect the induction of Fas on BMC or lethality in this model system. In agreement, we do not see any effect in this model if parental spleen cells are injected 1 wk after sublethal irradiation. It may prove informative to study GVHD in wild-type chimeric animals reconstituted with p53−/− bone marrow. This would provide a system to elucidate the nature of the cytokine storm and establish a potential causal contribution of BM depletion in the progression to lethality due to acute GVHD.
In summary, we have shown in a GVHD model that host animals undergo Fas-mediated cytotoxicity of BMC and lethality. More importantly, we have linked Fas to p53 and have shown that p53 status is important in determining BMC cytotoxicity and mortality in GVHD. These results have relevance to the numerous murine GVHD studies that have used sublethal radiation and use lethality as an end point. BM depletion is not a common problem in clinical GVHD as donor stem cells often replace and recolonize the host BM. However, graft-vs-host is a one-sided alloreaction, and the BMC studied here give us a convenient method of studying this mechanism. Therefore, these results may be relevant in clinical GVHD as well as in clinical cancer treatments in which p53-inducing agents such as radiation or chemotherapy are used in conjunction with immunostimulatory regiments or possibly in cases where graft-vs-tumor (host) occurs to prevent tumor recurrence.
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 Grants DK56601-01 and DK02445-04 and a grant from the Northwestern Medical School Howard Hughes Institute for new investigators (to T.L.), National Institutes of Health Grants AI44828-02 and GM52735 (to D.R.G.), and Swiss National Science Foundation Awards 31-65021.01 and Oncosuisse OCS-01161-09-2001 (to T.B.).
Abbreviations used in this paper: GVHD, graft-vs-host disease; BMC, bone marrow cell; BM, bone marrow; GVHR, graft-vs-host reaction; IEC, intestinal epithelial cell.