The Role of Dendritic Cells in Graft-versus-host Disease

Allogeneic hematopoietic cell transplantation (alloHCT) represents a definitive therapy for a number of otherwise fatal conditions. Graft-versus-host disease (GVHD) is the major cause of morbidity after transplant and limits the extended use of this critical therapeutic modality. Dendritic cells (DCs), a population of professional antigen-presenting cells (APCs), are thought to play a critical role in the initiation of this devastating disease. In this article, we will discuss studies that have established the role of DCs in GVH reactions, as well as the therapeutic implications of these studies.

Dendritic Cell Subsets in Lymphoid and Non-lymphoid Tissue

DCs belong to a population of hematopoietic cells called APCs that also include B cells and macrophages. The two main DC subsets include conventional DCs and plasmacytoid DCs (PDCs). DCs take up antigens through a wide range of mechanisms ¹ and are very well equipped to form peptide-major histocompatibility complex (MHC) class II and peptide-MHC class I complexes, migrate to the draining lymph nodes (LNs), and present tissue-derived MHC-peptide complexes to T lymphocytes.¹ Presentation of peptide-MHC class II complexes to CD4 T cells is called direct presentation and can be mediated by all APCs, including DCs, macrophages, and B cells.¹ The generation of peptide-MHC class I from soluble antigens² is called cross-presentation and appears to be specific to DCs.³ Similarly to DCs, PDCs express MHC class II molecules constitutively. Freshly isolated human and mouse PDCs are very poor inducers of T-cell proliferation. However, upon activation PDCs can differentiate into mature DCs with a high level of MHC class II and co-stimulatory molecules and T-cell stimulatory activity.⁴

Dendritic Cell Homeostasis After Allogeneic Hematopoietic Cell Transplantation

AlloHCT has been an experimental tool in mice and a therapeutic modality in humans for almost half a century.^6,7 Engraftment of allogeneic stem cells is facilitated by myelosuppressive and immunosuppressive conditioning given just prior to the infusion of stem cells. Mice, after a single dose of total body irradiation (TBI), are able to accommodate completely mismatched transplants without post-transplant immunosuppression. In humans, successful transplantation usually requires five to seven days of conditioning, an MHC geno-identical donor, and post-transplant immunosuppression.⁸ Even with these measures, GVHD, manifest as inflammation of the skin, bowel, and liver, occurs in 10–50% of human transplants and leads to death in 10–20%.^9,10 Many parameters, such as the cellular composition of the graft (human bone marrow (BM) harvest contains many more leukocytes (LCs)), greater minor antigenic diversity, prior exposure to infection, and loss of thymic function, may explain the greater burden of alloreactivity in humans. However, murine GVHD induced by additional infusion of splenocytes or T cells across major or minor histocompatibility mismatches shows many parallels with human disease.¹¹ Recent advances in clinical transplantation have led to the use of dramatically reduced or ‘non-myeloablative’ conditioning regimens.^12,13 These are yet to be fully exploited in mouse models but have important implications for DC homeostasis, as described below.
As a result of accessibility issues, studies on DC turnover in human patients after alloHCT have been restricted mostly to circulating blood DCs. Less than 20% of the pre-transplant recipient DC pool persists in the blood of patients after full myeloablative conditioning.¹⁴ Donor DCs and PDCs are detected in patient blood as early as one day post-transplant, which suggests that graft-derived DCs can successfully engraft in patients and participate in immune responses. By day 14 after myeloablative or reduced non-myeloablative transplant, the majority of circulating DCs are of donor origin, even in the presence of mixed chimerism in other lineages including T cells.^14,15 Because post-transplant immune responses occur mostly in the tissue, it is critical to examine the chimerism of tissue DCs. Surprisingly, the entire recipient LC pool and a subset of DC in the dermis survive lethal doses of X-ray irradiation (9.5 Gray) and persist in the recipient skin after transplantation of T-cell-depleted grafts, despite complete donor-derived chimerism of circulating LCs.^16-18 In contrast, in recipients of T-cell-replete allografts epidermal LCs are eliminated once donor alloreactive T cells infiltrate the skin.¹⁷ These results clearly establish that residual host LCs are present in the skin and skin-draining LNs when donor alloreactive T cells are primed to induce GVHD, but they are also the targets of this process. Below we discuss the immunological consequences of this finding.
It is likely that patient-related factors such as age, type of disease, and preparative regimen affect DC and PDC recovery after transplant. For example, the recovery of both DC populations seems faster after a reduced-intensity regimen compared with historical controls treated with a myeloablative regimen.¹⁹ In addition, occurrence of immune reactions such as GVHD has been consistently associated with reduced numbers of circulating DCs and epidermal LCs, even prior to corticosteroid administration.²⁰ The exact mechanisms of DC decrease remain to be carefully examined and may be the result of excess DC migration induced by inflammatory cytokines, impaired DC production, or excess cell death.

Dendritic Cells and Post-hematopoietic Cell Transplantation Immunity

It is now well established that the immunological activity of donor T cells in alloHCT grafts not only causes GVHD but is also a critical factor in eradicating residual recipient hematopoiesis and malignancy.^21–23 Thus, donor T-cell depletion abolishes GVHD but also leads to increased relapse risk.²⁴ The potential of donor T cells to secure remission is demonstrated most dramatically by the use of donor lymphocyte infusion (DLI) to treat post-transplant relapse.^25–27 Several minor histocompatibility antigens that drive GVH reactions have now been identified, allowing cellular responses to be monitored directly.^28,29
The importance of recipient APCs in initiating GVH responses, although long suspected, has only recently been proved in animal models.³⁰ Proof of a critical role for recipient APCs in GVHD has been achieved using tandem transplant experiments. Chimeric animals with defective APCs are generated by a preparatory transplant, and the effect on GVHD is tested with a second experimental transplant. These maneuvers show attenuation of acute CD8-mediated GVHD if MHC class I or co-stimulatory molecules are deleted from recipient APCs or if recipient APCs are first replaced by donor cells.^31–33 Donor APCs play a supplementary role and are able to augment acute GVHD through cross-presentation.³⁴ Further work confirms the importance of DCs compared with other APCs and demonstrates that recipient DC ‘add-back’ is sufficient to induce GVHD.^35,36 It is conceivable that the radio-resistant, self-renewable DC population could play some part in the predilection of GVHD for certain target organs. In mice transplanted with T-cell-depleted BM, persistent recipient cutaneous DCs are sufficient to generate cutaneous GVHD upon DLI.¹⁷ Both dermal DCs and LCs may persist under these experimental conditions,^18,37 and further dissection of their relative importance in cutaneous GVHD will soon be tractable using conditional LC knockout mice models.^38,39
Comparable results have been obtained in humans. In contrast to blood DCs and monocytes, human LCs survive conditioning therapy in significant numbers^40,41 and may persist after transplantation.⁴² The extent of recipient LC survival is linked to the intensity of conditioning and to GVHD.⁴⁰ Patients given the lowest-intensity non-myeloablative transplantation maintain 100% LCs and 75% dermal DCs of the host origin, despite complete donor-derived LC chimerism in the blood (see Figure 1).¹⁸ In contrast, in patients with advanced cutaneous GVHD lesions, residual host DCs are absent and are replaced by donor-derived cells. These data underline the similarity between cutaneous DC homeostatic properties in mice and humans and also suggest that residual host DCs are likely to play a key role in cutaneous GVHD in both species. As cutaneous DCs are themselves a target of GVHD, it is very difficult to prove a causal link between recipient DCs and GVHD levels in humans. A more promising approach is to examine correlations between DC chimerism and GVHD kinetics in cohorts of patients transplanted with different regimens.
In contrast to skin, other GVHD target organs have not yet been closely scrutinized. The gut and associated Peyer’s patches may harbor radio-resistant DC precursors, even though rapid turnover of mesenteric LN DCs is apparent.⁴³ In the liver, the bulk of parenchymal DCs rapidly equilibrate with the blood,¹⁷ but this does not exclude a niche population of cycling DCs in focal GVHD targets such as the portal triad of the liver. It would be contentious to claim that self-renewing populations of DCs are the sole initiators of GVH responses. Even though they may persist longer after conditioning therapy, there is ample evidence that the bulk of DC populations in the viscera or hematopoietic compartment easily survive for a sufficient duration to prime donor T cells. Recipient splenic DCs persist for only five days after irradiation, but alloreactive T cells express activation antigens after six hours and begin to proliferate within 24 hours.³² However, the self-renewable DC population is likely to be critical in recipients of DLI, as DCs will be the most likely to survive and initiate T-cell immune reactions.
The role of PDCs in transplant immunity has been studied less extensively than that of conventional DCs. In mouse models, expansion of immature DCs and PDCs using Flt3 ligand injection correlates with a diminished risk of acute GVHD after alloHCT,⁴⁴ but the role mediated by each DC subset in this protective effect was not carefully dissected. Administration of anti-CD52 antibodies has been shown to deplete circulating DCs and to protect from GVHD.¹⁴ PDCs also express CD52 and it is possible that elimination of PDCs can also play a role in this process.¹⁴ The presence of a high number of donor PDCs in BM grafts correlates with a decreased risk of chronic GVHD but not acute GVHD.⁴⁵ Peripheral blood stem cell grafts are associated with a lower risk of acute GVHD compared with BM transplants, despite the presence of a higher number of allogeneic T cells. This protective effect has been attributed to PDCs, because G-CSF-mobilized peripheral blood stem cell transplants are enriched in PDCs compared with BM grafts. Interestingly, donor PDCs have also been shown to facilitate engraftment of HSC and prolong mice survival after alloHCT.⁴⁶ The exact mechanism by which PDCs enhance engraftment remains to be examined.

Therapeutic Implications

The critical role of DCs in post-transplant immunity raises the opportunity to harness DC function to improve transplantation outcome. Murine experiments have raised the profile of ultraviolet (UV) irradiation of the skin as a means to accelerate the depletion of recipient LCs and prevent cutaneous GVHD.¹⁷ A general immunosuppressive effect of UV irradiation is well known,⁴⁷ and use of UV is encouraging as a regional therapy. However, it is critical to understand the effect of UV light on LC homeostasis in order to take advantage of this therapy clinically. The adverse effects of the peri-transplant use of UV light on GVHD that has been reported recently could have been anticipated.⁴⁸ In this study, patients were exposed to UV light prior to transplant at a time when host monocytes were still present in the blood. Based on our studies, exposure to UV light at this time-point would lead to the elimination of host LCs from the skin, but also to the recruitment of host monocyte-derived LCs to the skin, increasing the risk of GVHD at the time of transplant. The results of this study underline what we have already established in our mice studies, which is the need to expose patients to UV light only when host monocytes are absent or strongly reduced in the blood. In general terms, targeting of DCs may have the potential to control alloresponses beyond the means currently available. Targeting strategies can be used to eliminate recipient DCs to inhibit GVHD. Such a strategy will become attractive only when the control of GVHD is optimal. ■

Acknowledgments

I would like to acknowledge all members of my laboratory for their work, critical discussion, and challenging ideas. This work was supported by grants from the National Institutes of Health (RO1-CA112100).