Abstract
Mouse models of minor histocompatibility Ag-mismatched bone marrow transplantation were used to study donor dendritic cell (DC) reconstitution after conditioning, variables influencing the persistence of residual host DCs in different compartments, their phenotype, and their role in governing donor lymphocyte infusion (DLI)-mediated alloresponses. Reconstitution of all splenic DC subsets occurred rapidly after bone marrow transplantation and before T cell reconstitution. However, in contrast to MHC-mismatched chimeras, residual host-derived DCs persisted in the cutaneous lymph nodes (CLNs) of MHC-matched chimeras despite the presence or addition of donor T cells to the graft. The phenotype of these residual host-derived DCs in CLNs was consistent with Langerhans’ cells (LCs). We confirmed their skin origin and found near-complete preservation of host-derived LCs in the skin. Host-derived LCs retained their ability to continuously traffic to the CLNs, expressed homogeneously increased levels of costimulatory molecules, and could capture and carry epicutaneously applied Ags. To determine the role of residual host LCs in governing DLI-mediated alloresponses, we administered DLI alone or after topical application of the TLR7 ligand imiquimod, which is known to enhance the LC emigration from the skin. DLI administration resulted in a decrease in host-derived DCs in the CLNs and increased recruitment of donor-derived DCs to the skin, whereas imiquimod augmented their alloreactivity. These results suggest uniqueness of the MHC-matched setting in relation to the persistence of host-derived DCs in the skin and points to a previously unrecognized role of host-derived LCs in the induction of DLI-mediated graft-vs-host alloresponses.
Allogeneic T cell-mediated responses, like any other acquired immune responses, are initiated in the secondary lymphoid organs by the encounter of lymphocytes and APCs (1, 2). An understanding of T cell-dendritic cell (DC)4 interactions after allogeneic bone marrow transplantation (alloBMT) is critical for augmenting antitumor immunity and limiting graft-vs-host disease (GVHD). Early studies of the cellular and molecular mechanisms of GVHD induction focused on identifying the type of T cell (CD4 vs CD8) mediating GVHD across MHC class I, class II, or minor histocompatibility barriers (3, 4, 5) or on the cytotoxic effector molecules, such as perforin, granzyme, and FasL, that are involved in GVHD-associated tissue destruction (6, 7, 8, 9). Recently, attention has been given to the differential capacity of host vs donor APCs to initiate and/or sustain GVHD and the graft-vs-leukemia effect (10, 11, 12, 13, 14, 15). To some extent, these studies have produced conflicting results. Using the C3H.SW→C57BL/6 (B6) model, in which donors and recipients differ only in the expression of multiple minor histocompatibility Ags (miHAs), Shlomchik et al. (15) have established that the development of lethal GVHD after alloBMT depends on T cell recognition of alloantigens presented by functional host-derived professional APCs. In this study, purified CD8+ T cells from C3H.SW donors did not initiate acute GVHD following infusion into reirradiated C3H.SW→B6 chimeras, suggesting that residual host APCs do not persist after MHC-matched miHA disparate allografting. In an earlier study by Korngold and Sprent (3), however, CBA lymph node-derived T cells infused into reirradiated CBA→B10.BR chimeras, or B10.BR T cells infused into reirradiated B10.BR→CBA chimeras did cause acute GVHD, prompting the authors to conclude that residual host APCs remained or that donor-derived APCs were capable of cross-presenting host miHAs for induction of GVHD. These studies were recently extended by showing that B6→BALB.B MHC-matched miHA-mismatched chimeras upon reirradiation and transfer of purified B6 CD4+ T cells do develop lethal GVHD (16). A recent study in a model of MHC-mismatched bone marrow transplantation (BMT) showed that host Langerhans cells (LCs) persist in the skin after conditioning and transplantation of T cell-depleted but not T cell-replete bone marrow (BM) and that the extent of residual host chimerism correlated with dose of T cells injected (12). However, these studies did not examine host DC chimerism after the more clinically relevant scenario of MHC-matched miHA-mismatched alloBMT, and therefore do not provide an explanation for the susceptibility of reirradiated MHC-matched chimeras to develop GVHD. They also do not explain the frequent occurrence of GVHD after delayed administration of donor lymphocyte infusions (DLI) in humans with apparent full donor chimerism in blood and marrow. In light of the inconsistent and possibly incompatible outcomes of alloBMT or DLI using distinct mouse models, we have directly compared DC reconstitution and the effect of DLI after syngeneic, MHC-matched, and MHC-mismatched alloBMT.
We hypothesized that, when donor and recipient are MHC-matched, residual host DCs persist after T cell-replete BMT and influence the potency of graft-vs-host (GVH) reactions induced by delayed DLI. Our hypothesis was based on the known existence of the increased precursor frequency of T cells recognizing allogeneic MHC molecules in comparison to miHAs expressed on host APC and findings that histoincompatibility between the donor and host is the primary determinant of donor T cell activation when transferred into an allogeneic environment (17, 18). Accordingly, using mouse models of alloBMT, we show a critical difference between the outcome of MHC-matched vs -mismatched transplants in the level of the residual host DC persistence in skin, thus implicating the residual host skin-derived DCs in guiding alloresponses of adoptively transferred naive donor T cells posttransplant.
Materials and Methods
Animals
C57BL/6 (B6; H-2b, CD45.2+) and B6.SJL (H-2b; CD45.1+) mice were obtained from the National Cancer Institute (Frederick, MD). C3H.SW-H2b/SnJ (H-2b; CD45.2+) and B6.PL-Thy1a/CyJ (H-2b; Thy1.1+) mice were obtained from The Jackson Laboratory. Mice were maintained in microisolator cages and fed ad libitum with autoclaved laboratory chow and acidified water. All mice were ∼8–12 wk of age at the time of transplantation. All experiments involving mice were performed in accordance with protocols approved by the Animal Care and Use Committee of the Johns Hopkins University.
Hemopoietic cell transplantation, UV irradiation, DLI, and imiquimod administration
Animals were irradiated by a dual source 137Cs irradiator (Gammacell 40; Atomic Energy of Canada) at an exposure rate of ∼73 cGy/min. Conditioning consisted of 900 cGy of total body irradiation (TBI) on the day before marrow transplantation unless otherwise indicated. A total of 107 BM cells was injected into a tail vein in a final volume of 0.5 ml of Eagle Hanks amino acid medium (Biofluids). T cell depletion (TCD) of BM was performed as described previously (19). Exposure to UVB light was done on days 13 and 15 after transplantation with a dose of 100 mJ/cm2 per day. DLI comprised an i.v. infusion of 12 million donor lymph node (LN) cells. In designated experiments, a purified population of donor T cells was obtained by passage of cells through nylon wool columns or by using T cell isolation kits (Miltenyi Biotec) according to the manufacturer’s instructions. Cells were resuspended and injected in 0.5 ml of Eagle Hanks’ amino acid medium. Imiquimod in a 5% cream formulation (Aldara; 3M Pharmaceuticals) or vehicle was topically administered on shaven skin of mice under anesthesia. A total of 6.25 mg of imiquimod was applied daily on the chest or abdomen on days −3 to −1 before DLI and on the back of the mouse on the day of DLI administration.
Isolation of DCs from spleen, LNs, and epidermal sheets
DCs from spleen and LNs were isolated as previously described (20, 21, 22). Briefly, harvested spleens and LNs were incubated with collagenase D (400 U/ml; Roche) for 60 and 30 min, respectively. In our experiments, we either first obtained a low-density splenocyte fraction on a BSA gradient or immediately incubated the cell suspension with CD11c MicroBeads (Miltenyi Biotec). The LN cell suspension was immediately incubated with MACS CD11c MicroBeads. After 30-min incubation at 4°C, beads were washed, and the CD11c+ fraction was isolated using LS separation columns and MidiMACS magnets (Miltenyi Biotec). Epidermal sheets were obtained from mouse ears that were first mechanically split with forceps and then incubated in 0.5% trypsin (Invitrogen Life Technologies) and EDTA (5 mM) in PBS for 60 min at 37°C (20, 23, 24). Peeled epidermal sheets were cultured in the presence of 10 ng/ml GM-CSF and 10 ng/ml TNF-α (both from PeproTech) (23, 25). After 48-h incubation at 37°C, the emigrant cells were collected and stained for flow cytometry analysis.
Flow cytometry analysis
At designated time points, animals were sacrificed, and spleen, cutaneous LNs (CLNs; pooled axillary, brachial, cervical, and submandibular nodes), and mesenteric LNs were collected. Single-cell suspensions were prepared, enriched for CD11c+ cells in corresponding experiments, labeled with fluorochrome-conjugated mAbs, and analyzed by flow cytometry as previously described (19). The mAbs used were FITC, PE, PE-Cy5, or allophycocyanin-conjugated anti-CD45.1, CD45.2, CD11c, CD4, and CD8α (all from BD Biosciences). In some of the experiments, biotinylated anti-CD11c mAb was detected with streptavidin-PE or -allophycocyanin (BD Biosciences). FITC-conjugated anti-DEC205 was obtained from Cedarlane Laboratories. In the experiments using this Ab, staining was performed at 4°C after surface staining on fixed and permeabilized cells, because the Ag is primarily localized intracellularly (26). LC-specific gp40 (Ly74) Ab was obtained from Developmental Studies Hybridoma Bank developed under the auspices of the National Institute of Child Health and Human Development and maintained by the University of Iowa Department of Biological Sciences. For in vivo trafficking studies, three to five chimeras were painted on the abdomen with 200 μl of 1% tetramethylrhodamine isothiocyanate (TRITC) (Molecular Probes) dissolved in a solvent of acetone-dibutylphtalate (1:1 ratio). Two and 4 days after TRITC painting, the draining CLNs were removed, enriched for DCs, stained, and analyzed as described above. FACSCalibur (BD Biosciences) was used for flow cytometry event collection and events were analyzed using CellQuest software (BD Biosciences). Viable cells were selected for analysis based on the forward- and side-scatter analysis. In designated experiments, propidium iodide (PI) was added immediately before analysis. To determine IFN-γ production, cells were stimulated with PMA (3 ng/ml final concentration) and ionomycin (500 ng/ml; both from Sigma-Aldrich) for 5 h in culture in the presence of GolgiStop (BD Biosciences). After stimulation and surface staining, cells were fixed and permabilized (CytoFix/CytoPerm kit; BD Biosciences) according to the manufacturer’s instructions, and subsequently stained with anti-IFN-γ for 30 min at 4°C.
In vivo and histopathological analysis of GVHD
To quantify clinical GVHD, we used a previously described scoring system that sums changes in five clinical parameters: weight loss, posture, activity, fur texture, and skin integrity (maximum index, 10) (27, 28). In designated experiments, we also assessed acute GVHD by detailed histopathological analysis of ear skin, tongue, liver, and intestine. All samples were placed in 10% neutral buffered formalin and prepared for routine histological analysis, as previously described (16, 29). An experienced pathologist (G. F. Murphy) made all observations in a double-blind manner. Apoptotic epidermal basal cells were counted per linear millimeters (Lmm), with a minimum of 6 mm per animal evaluated. Data are reported as the dyskeratotic index (mean ± SEM number) of apoptotic keratinocytes per Lmm (16, 30). Gut GVHD was scored on the basis of crypt apoptosis (31). Apoptotic intestinal crypts cells were counted for at least 25 crypts. Data are expressed as the number of apoptotic cells per crypt. Histologic assessment of liver GVHD was performed by assessing portal lymphocytic triaditis (32, 33). At least 10 portal triads per specimen were evaluated, and data are expressed as number of triads involved per 10 triads examined.
Statistical analysis
Data were analyzed using PRISM software (version 4.0; GraphPad). Values are presented as the mean ± SEM. The statistical differences were calculated using Student’s two-tailed t test. The ANOVA analysis was used for statistical analysis of clinical scores. A value of p < 0.05 was considered statistically significant.
Results
Persistence of host DCs after alloBMT is influenced by MHC disparity
We hypothesized that increased MHC disparity speeds donor DC reconstitution and eliminates residual host blood-derived DCs after alloBMT. To test this hypothesis, recipient mice were lethally irradiated (900 cGy) and reconstituted with 10 million non-TCD BM cells from syngeneic donors or from donors differing in the expression of multiple miHAs vs both MHC and multiple miHAs. Six weeks later, chimerism was determined in the spleen and CLNs. To distinguish the origin of DCs, donors and recipients were chosen based upon differential expression of alleles of CD45, the common leukocyte Ag. In all three sets of chimeras, full reconstitution with donor-derived CD11c+ DCs was found in the spleen (Fig. 1,A). However, we repeatedly found host-derived CD11c+ DCs in the CLNs of syngeneic and MHC-matched but not mismatched chimeras (Fig. 1 B). Residual host DCs represented 12 ± 1.2 and 10.6 ± 1.1% of CD11c+ DCs in CLNs of syngeneic and MHC-matched allogeneic chimeras, respectively, compared with 0.3 ± 0.05% in MHC-mismatched chimeras. Because the transplant regimen was identical for all groups, the results demonstrate that the degree of donor/host histoincompability influences residual host DC chimerism after alloBMT.
We next characterized the kinetics of host-to-donor DC turnover in the spleen of MHC-matched chimeras. In the spleen, 99.5 ± 0.2 and 99.3 ± 0.4% of CD11c+ DCs were of donor origin at 3 and 8 wk after BMT, respectively (Fig. 1,C, upper panel). In contrast, CLNs contained 8.5 ± 0.43 and 11.6 ± 1.7% residual host DCs at 3 and 8 wk after BMT, respectively (Fig. 1,C, middle panel). Residual host CD11c+ DCs persisted in similar percentages in the CLNs up to 6 mo posttransplant (latest time point tested; data not shown). However, residual host CD11c+ DCs were not detected at either time point in the mesenteric LNs (Fig. 1 C, lower panel). Overall, these results demonstrate that mature T cells in the donor graft, which are sufficient to eliminate host DCs after MHC-mismatched BMT, are insufficient to eliminate residual host DCs that persist exclusively in the CLNs after MHC-matched alloBMT.
DC reconstitution after MHC-matched alloBMT is rapid and precedes T cell reconstitution
Because there were no major phenotypic differences in DC reconstitution of MHC-matched chimeras at 3 and 8 wk after transplantation, we next examined turnover of DCs early after BMT. In the spleen, we found a rapid conversion to nearly complete donor DC chimerism with <5% of residual host DCs remaining as early as 8 days after the transplantation (Fig. 2,A). In the CLNs, the kinetics of host-to-donor DC reconstitution was slower, with 28% of CD11c+ cells being still of host origin on day 8 and ∼13.2% of host DCs persisting on day 28 after BMT. Next, we examined the reconstitution of splenic DC subsets after alloBMT. Splenic CD11c+ DCs can be divided into three subsets that can be distinguished based on their expression of CD4 and CD8α: CD11c+CD4−CD8α+, CD11c+CD4+CD8α−, and CD11c+CD4−CD8α− (21, 34). Although the subsets differ in their surface markers, all three populations are efficient in activating allogeneic T cells in conventional MLC (21, 34, 35, 36). The transfer of donor BM resulted in a rapid reconstitution of CD11c+ DCs with the dominance of the CD4−CD8− subset 8 days after transplantation (Fig. 2,B). The CD4−CD8α+ subset also peaked at 8 days posttransplant but then drifted toward their normal distribution pattern. The CD4+CD8α− subset showed a slower rate of reconstitution, but by day 28 became the dominant population. The distribution of CD11c+ subsets in the spleen of chimeras 8 wk posttransplant was similar to that of the normal donor and recipient mice. The use of the MHC-matched donor→recipient pair differing at the CD45.1 locus also allowed us to characterize T cell reconstitution. Contrary to the rapid conversion to full donor DC chimerism in the spleen, conversion of donor T cell chimerism was slower. By day 28, 94 and 97% of the CD8+, and 83 and 84% of the CD4+ T cells in the spleen and CLNs were of donor origin, respectively (Fig. 2,C). The absolute numbers of T cells at 3 and 8 wk after transplantation are presented in Fig. 2,D. Compared with normal, nontransplanted donor mice, the splenic content of CD4+ and CD8+ T cells was ≈15% at 3 wk in comparison with total CD11c+ content, which was ≈50% at the same time point (Fig. 2,D). Although the total number of splenic CD11c+ cells was equivalent in control mice and in transplanted mice 8 wk after BMT (Fig. 2 E), the number of donor T cells in transplanted mice lagged behind. Thus, donor DCs reconstitute more rapidly than donor T cells after lethal conditioning and alloBMT.
TBI dose and addition of mature donor T cells to the MHC-matched BM influence donor DC reconstitution and the level of residual host DC persistence in the CLNs
After demonstrating that there is a rapid and complete turnover of host-to-donor DCs in the spleen but not in the CLNs of syngeneic and MHC-matched chimeras after myeloablative conditioning and transplantation of non-TCD BM, we wanted to better characterize the variables that determine the extent of their turnover and persistence. To confirm that our observations are not restricted to one donor→recipient pair, the next set of experiments was performed in the reverse direction, i.e., with C3H.SW donors and B6.SJL recipients (15, 22). We hypothesized that intensity of conditioning and presence of donor T cells independently influence DC turnover and the amount of residual host DCs after alloBMT. To determine the role of conditioning independently of donor/recipient histocompatibility, we titrated a dose of TBI in the context of syngeneic BMT and measured donor DC chimerism in spleen and CLNs 6 wk after transplantation. In the spleen, donor DC chimerism increased proportionally to the TBI dose and was close to 100% at a 700-cGy dose (Fig. 3,A). Because there is no immunologic barrier in this donor→recipient pair, we concluded that the extent of host-to-donor DC chimerism turnover is directly affected by the conditioning. In the CLNs, the increase in donor DC chimerism was slower and never exceeded 90%. To address the role of conditioning and donor T cell-mediated alloreactivity, B6.SJL mice were conditioned with varying doses of TBI (400–900 cGy) and transplanted with MHC-matched C3H.SW BM, with or without 107 added donor T cells. Consistent with our previous studies, 700 cGy of TBI was required to achieve successful engraftment of MHC-matched allogeneic marrow (Fig. 3, B and C) (37). Presence of additional mature T cells in the donor marrow reduced the minimum dose of TBI required for engraftment from 700 to 500 cGy (p < 0.05). The presence of donor T cells in the graft clearly decreased the level of residual host DCs in the CLN (12.7 vs 4.4% at 900 cGy) (Fig. 3,C), although they were never completely eliminated. The persistence of host DCs was also influenced by the dose of TBI; however, they were still detectable even if the dose was escalated to 1100 cGy, given as a single dose or in two 550-cGy fractions (Fig. 3,D). The lower percentage of residual host DCs in the CLNs of chimeras that received additional T cells in the graft demonstrates that the level of their persistence is determined by the potency of the donor GVH response. The role of the donor T cell-mediated GVH response on host cells was also reflected by significantly different levels of residual host CD4+ and CD8+ T cell chimerism in the spleen (Fig. 3, E and F) and CLNs (data not shown).
Skin-derived DCs are the dominant residual host DC population in LNs after MHC-matched alloBMT
We next characterized the phenotype of the DCs persisting in the CLNs after myeloablative conditioning and transplantation of non-TCD MHC-matched miHA mismatched BM. It is well established that the DC content of mouse CLNs and spleen differs (20, 34). The DCs in CLNs comprise the three populations also present in the spleen (discussed above) as well as two additional populations, characterized by relatively low expression of CD8α and low or high expression of DEC205. The two latter populations represent tissue-derived DCs that migrate out of the skin into CLNs and originate from epidermis (CD11c+CD8α−DEC205high) and dermis (CD11c+CD8α−DEC205low) (20, 38). The epidermally derived CD11c+CD8α−DEC205high population represents LCs based on their selective staining with langerin, a marker of the Ag-processing system of LCs (20, 39). In our experiments, we have not attempted to distinguish dermally derived from epidermally derived DCs, and thus, consistent with literature, we will be referring to them hereafter as “epithelium-derived DCs” (epDCs) (38). Both B6.SJL→C3H.SW and C3H.SW→B6.SJL chimeras contained all of the described DC subpopulations in their CLNs (data not shown). Next, differential expression of the CD45.1 allele was used to distinguish donor BM-derived DCs from host DCs in the CLNs (Fig. 4,A). In the host fraction, CD11c+CD8α−DEC205high epDCs were the predominant population of DCs (Fig. 4,A, lower panel). In contrast, the donor fraction contained all CLN DC subpopulations. epDCs of host origin could be detected reproducibly as late as 6 mo after transplantation, although we were able to detect low levels of the other host-derived DC subpopulations at late time points after alloBMT, particularly CD11c+CD4−CD8α− and CD11c+CD4−CD8α+ in the CLNs. However, host-derived CD11c+CD8α−DEC205high DCs always dominated (data not shown). The origin of the host-derived fraction in the CLNs was also confirmed by staining with an Ab to gp40, a murine homolog of the human epithelial cell adhesion molecule (Fig. 4 B), and E cadherin (data not shown), both expressed specifically on mouse LCs (40, 41, 42).
The origin of CD11c+ cells that migrate out of ex vivo cultured epidermal sheets (20, 23, 39) was analyzed next. Epidermal sheets isolated from the B6.SJL→C3H.SW chimeras were cultured in the presence of GM-CSF and TNF-α for 48 h to elicit skin DC migration into the medium. In contrast to the predominant donor origin of DCs in CLNs, DCs from epidermis were mostly from the host (Fig. 4,C). Because these studies have shown that persistence of residual host skin-derived DCs is influenced by MHC disparity, we extended the analysis of epDCs to histocompatible and to MHC-mismatched chimeras. In the histocompatible B6→B6.SJL chimeras, the host CD45.1+ fraction was found to contain all CLN DC subpopulations (Fig. 4,D and data not shown), whereas in the MHC-mismatched B6→BALB/c chimeras constructed after myeloablative conditioning and transplantation of non-TCD BM, the host fraction was absent and all CD11c+CD8α−DEC205high cells in the CLNs were exclusively donor derived (Fig. 4 E). These results with histocompatible and MHC-mismatched chimeras are consistent with other studies using predominantly skin to analyze the presence of residual host LCs (12, 23, 43). However, our data in the clinically relevant MHC-matched model show that the majority of epDCs are of host origin despite complete donor DC chimerism in the spleen and mesenteric LNs.
Persistence of host-derived epDCs in the CLNs of MHC-matched chimeras in the steady state is result of their continuous migration from the skin
In normal mice, epDCs continuously migrate from skin to the CLNs and express high levels of costimulatory molecules in the steady state, even in the absence of inflammatory signals (39, 44, 45, 46). Consistent with these and other studies (20, 39, 44, 47), we found that CD11c+DEC205+ but not CD11c+DEC205− DCs in nontransplanted C3H.SW mice expressed cell surface markers (MHC class II, CD86, and CD40) associated with a mature phenotype (Fig. 5,A, upper panel, and data not shown). In the B6.SJL→C3H.SW chimeras, residual host-derived CD11c+DEC205+ DCs also expressed homogeneously high levels of MHC class II (Fig. 5,A, lower panel) and other activation markers (data not shown). To confirm in a more direct way that residual host DCs can capture epicutaneously applied Ags and traffic from skin to the CLNs, mouse abdominal skin was painted with TRITC (39, 41, 48). Two or 4 days later, the presence of TRITC-bearing CD11c+ cells in the draining CLNs and their origin was analyzed by flow cytometry. As expected, the CD11c+TRITC+ DCs were only present among those CD11c+ DCs expressing the highest levels of class II MHC and represented ∼20 and 26% of all CD11c+, MHC class IIhigh cells in CLNs on both days, respectively. Interestingly, on day 2 after TRITC application, a large majority of the cells were of donor origin, whereas at day 4, host-derived CD11c+TRITC+class IIhigh DCs represented a significant fraction (Fig. 5 B). This delayed appearance of host-derived CD11c+class IIhighTRITC+ cells is consistent with previous observations that the labeled epidermal CD11c+ DC are the slowest to migrate to the CLN in this model (21, 41, 49). The CD11c+, MHC class IIhigh, TRITC+ DCs at day 2 most likely represent interstitial donor-derived cells or, less likely, DCs recruited from circulating precursors having taken up TRITC in the CLNs.
The dose of T cells differentially influences donor LC chimerism after MHC-matched vs -mismatched alloBMT
Titrations of donor T cells were performed to examine more comprehensively the relationship between the dose of donor T cells and residual host chimerism after MHC-matched or -mismatched alloBMT. To allow direct comparison between the chimeras, all animals were transplanted on the same day and received the TCD BM and donor T cells prepared from the same source. After transplantation, all chimeras were monitored for signs of GVHD (27, 28). As presented in Table I and Fig. 6,A, administration of ≥1.5 × 106 T cells with TCD BM to the MHC-mismatched B6.SJL→BALB/c chimeras resulted in severe GVHD and death of the majority of animals. MHC-mismatched chimeras that received 0.5 × 106 donor T cells also developed signs of moderate GVHD by 6 wk posttransplant; therefore, the experiment was terminated 2 wk later. None of the MHC-matched B6.SJL→C3H.SW chimeras that received TCD BM alone or with added 0.5 × 106 donor T cells showed clinical signs of GVHD (Fig. 6,B). In contrast, MHC-matched chimeras that received 1.5 × 106 and in particular 3.0 × 106 T cells with TCD BM showed mild signs of GVHD (weight loss; Fig. 6,B). No signs of cutaneous GVHD were noted in MHC-matched chimeras during the 8 wk of observation. At the time of analysis, comprehensive assessment of donor T cell chimerism in lymphoid tissues and LC chimerism in skin was performed. As expected, in both sets of chimeras, the administration of TCD BM alone after myeloablative conditioning and especially with added donor T cells resulted in high levels of donor T cell chimerism. Consistent with the study by Merad et al. (12), we found a significant difference in donor LC chimerism after MHC-mismatched BMT between mice that did or did not receive 0.5 million mature donor T cells in the graft (93.64 ± 5.61 vs 1.03 ± 0.71%, respectively; p < 0.0001). Interestingly, despite a graded increase in donor T cells added to the graft, the changes in donor LC chimerism in the MHC-matched chimeras were minimal (0.75 ± 0.33 vs 2.96 ± 2.48). To determine whether these findings apply to the MHC-matched setting, we performed a second experiment in the C3H.SW→B6.SJL model. As shown in Table I and consistent with the previous experiment, addition of 5 × 106 mature donor T cells to the TCD BM resulted in only a slight increase in donor LC chimerism (2.51 ± 0.43 vs 7.67 ± 2.76). To assess the ability of donor-derived hemopoietic cells in MHC-matched chimeras to repopulate the skin and to rule out selective effects of in vivo culture in assessing donor-host CD11c+ DC chimerism in the skin, an experiment in which the C3H.SW→B6.SJL chimeras were treated with UV light was performed. As expected, the UV treatment after transplant resulted in nearly full replacement of skin LCs by donor BM-derived CD11c+ DCs (2.51 ± 0.43 vs 87.30 ± 3.72; p < 0.001). These findings confirm the validity of the in vitro analysis and demonstrate that the ability of donor T cells to eliminate host LCs after alloBMT is profoundly influenced by the degree of donor-host histoincompatibility.
Group . | Donor (H2) . | Recipient (H2) . | 800-cGy TBI BM . | Purified Donor T Cells i.v. day 0 . | Nb . | CLN (106) Cellularity (SEM) . | CLN Mean % Donor CD4+ Chimerism (SEM) . | CLN Mean % Donor CD8+ Chimerism (SEM) . | Mean % Donor CD11c+ Chimerism (SEM) Earsc . |
---|---|---|---|---|---|---|---|---|---|
Expt. 1 | B6.SJL (H2b)d | BALB/c (H2d) | 107 | — | 4/4 | 2.58 ± 10.72 | 67.81 ± 10.29 | 76.34 ± 14.23 | 1.03 ± 0.71g |
B6.SJL (H2b)d | BALB/c (H2d) | 107 | 0.5 × 106 T cells | 4/4 | 0.67 ± 0.29 | 98.61 ± 0.30 | 94.38 ± 2.78 | 93.64 ± 5.61g | |
B6.SJL (H2b)d | BALB/c (H2d) | 107 | 1.5 × 106 T cells | 1/4 | 0.13 | 93.23 | 98.58 | 92.60 | |
B6.SJL (H2b)d | BALB/c (H2d) | 107 | 3.0 × 106 T cells | 0/5 | — | — | — | — | |
B6.SJL (H2b)e | C3H.SW (H2b) | 107 | — | 4/4 | 24.45 ± 4.41 | 80.54 ± 3.08 | 92.85 ± 0.64 | 0.75 ± 0.33 | |
B6.SJL (H2b)e | C3H.SW (H2b) | 107 | 0.5 × 106 T cells | 5/5 | 15.52 ± 16.83 | 86.88 ± 6.68 | 95.53 ± 2.06 | 1.32 ± 0.86 | |
B6.SJL (H2b)e | C3H.SW (H2b) | 107 | 1.5 × 106 T cells | 5/5 | 0.60 ± 0.42 | 99.49 ± 0.39 | 98.75 ± 1.09 | 1.68 ± 0.39 | |
B6.SJL (H2b)e | C3H.SW (H2b) | 107 | 3.0 × 106 T cells | 5/5 | 0.46 ± 0.26 | 99.68 ± 0.06 | 93.56 ± 4.34 | 2.96 ± 2.48 | |
Expt. 2 | C3H.SW (H2b)e | B6.SJL (H2b) | 107 | — | 3/3 | 21.21 ± 3.43 | 87.27 ± 0.61 | 72.23 ± 2.33 | 2.51 ± 0.43♦ |
C3H.SW (H2b)e | B6.SJL (H2b) | 107 | 5 × 106 T cells | 3/3 | 0.93 ± 0.76 | 98.39 ± 1.44 | 95.11 ± 3.53 | 7.67 ± 2.76 | |
Expt. 3 | C3H.SW (H2b)ef | B6.SJL (H2b) | 107 | — | 3/3 | 11.58 ± 1.08 | — | — | 87.30 ± 3.72♦ |
Group . | Donor (H2) . | Recipient (H2) . | 800-cGy TBI BM . | Purified Donor T Cells i.v. day 0 . | Nb . | CLN (106) Cellularity (SEM) . | CLN Mean % Donor CD4+ Chimerism (SEM) . | CLN Mean % Donor CD8+ Chimerism (SEM) . | Mean % Donor CD11c+ Chimerism (SEM) Earsc . |
---|---|---|---|---|---|---|---|---|---|
Expt. 1 | B6.SJL (H2b)d | BALB/c (H2d) | 107 | — | 4/4 | 2.58 ± 10.72 | 67.81 ± 10.29 | 76.34 ± 14.23 | 1.03 ± 0.71g |
B6.SJL (H2b)d | BALB/c (H2d) | 107 | 0.5 × 106 T cells | 4/4 | 0.67 ± 0.29 | 98.61 ± 0.30 | 94.38 ± 2.78 | 93.64 ± 5.61g | |
B6.SJL (H2b)d | BALB/c (H2d) | 107 | 1.5 × 106 T cells | 1/4 | 0.13 | 93.23 | 98.58 | 92.60 | |
B6.SJL (H2b)d | BALB/c (H2d) | 107 | 3.0 × 106 T cells | 0/5 | — | — | — | — | |
B6.SJL (H2b)e | C3H.SW (H2b) | 107 | — | 4/4 | 24.45 ± 4.41 | 80.54 ± 3.08 | 92.85 ± 0.64 | 0.75 ± 0.33 | |
B6.SJL (H2b)e | C3H.SW (H2b) | 107 | 0.5 × 106 T cells | 5/5 | 15.52 ± 16.83 | 86.88 ± 6.68 | 95.53 ± 2.06 | 1.32 ± 0.86 | |
B6.SJL (H2b)e | C3H.SW (H2b) | 107 | 1.5 × 106 T cells | 5/5 | 0.60 ± 0.42 | 99.49 ± 0.39 | 98.75 ± 1.09 | 1.68 ± 0.39 | |
B6.SJL (H2b)e | C3H.SW (H2b) | 107 | 3.0 × 106 T cells | 5/5 | 0.46 ± 0.26 | 99.68 ± 0.06 | 93.56 ± 4.34 | 2.96 ± 2.48 | |
Expt. 2 | C3H.SW (H2b)e | B6.SJL (H2b) | 107 | — | 3/3 | 21.21 ± 3.43 | 87.27 ± 0.61 | 72.23 ± 2.33 | 2.51 ± 0.43♦ |
C3H.SW (H2b)e | B6.SJL (H2b) | 107 | 5 × 106 T cells | 3/3 | 0.93 ± 0.76 | 98.39 ± 1.44 | 95.11 ± 3.53 | 7.67 ± 2.76 | |
Expt. 3 | C3H.SW (H2b)ef | B6.SJL (H2b) | 107 | — | 3/3 | 11.58 ± 1.08 | — | — | 87.30 ± 3.72♦ |
—, Not applicable.
Number of animals transplanted/number alive at time of analysis.
Chimerism was measured in epidermal sheets.
Donor and recipient differ in the expression of major plus minor histocompatibility Ags.
Donor and recipient differ in the expression of miHA.
Chimeras were treated with UV (2 × 100 mJ/cm2).
, Statistically significant: p < 0.0001; ♦, p < 0.001.
Although MHC-matched chimeras that received TCD BM with added donor T cells exhibited only mild clinical signs of GVHD, we found significant differences in the cellularity of their CLNs (Table I). Because it is well established that GVH reactivity is associated with atrophy of peripheral lymphoid tissues (50), we concluded that their cellularity could be used as an objective marker for assessing subclinical lymphohemopoietic GVH (LH-GVH) reactivity.
Residual host epDCs are the targets of DLI-mediated GVH response
The observation that residual host DCs persist after MHC-matched alloBMT and express high levels of costimulatory molecules prompted us to investigate whether they serve as targets of DLI-mediated alloreactivity. Three weeks after transplantation, B6.SJL→C3H.SW chimeras received nylon wool-purified B6.SJL T cells, to reduce the transfer of donor DCs, and DC chimerism was analyzed 2 wk later. Residual host CD11c+ cells represented 3.4 and 8.6% of the total CD11c+ DC population in the CLNs of the B6.SJL→C3H.SW chimeras that did or did not get DLI, respectively (Fig. 6,C, upper panel). The effect of the DLI-mediated GVH reaction was reflected in a lower proportion of host-derived CD11c+CD8−DEC205high DCs and a reduced number of DCs in the CLNs (Fig. 6,C, lower panel, and data not shown). To assess whether donor alloreactivity influences the donor to host ratio of DCs in the skin, we analyzed the ratio of CD11c+ cells that migrate out of ex vivo-cultured epidermal sheets. We found that, on days 13 and 48 after DLI administration, there was a higher proportion of donor-derived CD45.1+CD11c+ cells in the group of animals that received adoptively transferred T cells (Fig. 6,D). The observed replacement of host with donor-derived LCs demonstrates that host-derived LCs are the targets of alloreactive donor T cells. Compared with uninjected controls, chimeras that received DLI also showed decreased cellularity in CLNs starting on day 14 after DLI administration (Fig. 6,E). The reduction of CLN cellularity was progressive such that by time of termination of the experiment on day 48, the CLNs were difficult to find. In the spleen, total cell numbers were decreased, although not as much as in the CLNs (p < 0.05; Fig. 6 F). These results indicate that residual host LCs in the CLNs and the skin are the targets of DLI-mediated GVH response and that this response is associated with hypoplasia of CLNs and spleen.
Increased emigration of host LCs to the CLNs augments GVH response of DLI without causing GVHD
In the next set of experiments, we sought to determine directly the role of residual host-derived LCs in the DLI-mediated GVH response. We hypothesized that depletion of host LCs should decrease the GVH reactivity of DLI, whereas their increased emigration to the CLNs caused by the inflammation should augment alloreactivity. To test the first part of our hypothesis, chimeras were exposed to UV light. Because turnover of LCs in the skin after UV treatment takes ∼60 days (12, 23), the model as used above had to be modified. Thy 1-congenic donor T cells were administered 3 or 8 wk after BMT and their number in the spleen was monitored. DLI-derived CD4+ and CD8+ T cells expanded significantly less in vivo when administered 8 as opposed to 3 wk after BMT (Fig. 7 A). Based on the poor expansion of DLI-derived T cells observed in 8-wk-old chimeras, we conclude that it will be difficult to examine the effect of residual host LCs elimination by UV on their in vivo GVH response in this model system.
We then tested the second part of our hypothesis, that increased emigration of LCs from skin to the CLNs should augment DLI-mediated alloresponses. To induce inflammation in the skin, we applied imiquimod, a synthetic ligand for TLR7 (51). In both mice and humans, imiquimod induces production of inflammatory cytokines, maturation, and reversible emigration of LCs from the epidermis to the CLNs (52, 53, 54). To determine the effect of imiquimod on the GVH response of DLI, we used the model described above. Eight weeks after BMT, B6.SJL→C3H.SW chimeras received imiquimod topically administered to shaved skin (∼1 sq. in.) for 3 days before and on the day of the DLI administration. Expansion of both CD4+ and CD8+ DLI-derived T cells on days 14, 28, and 48 in the spleen (Fig. 7,B) and CLNs (C) of the chimeras treated with imiquimod was superior to that in animals receiving DLI alone. Next, we assessed whether imiquimod augments the function of adoptively transferred T cells. First, we found that pretreatment of animals with imiquimod, in contrast to vehicle, increases IFN-γ secretion in DLI-derived CD8+ T cells in the spleen (Fig. 7,D). Second, administration of imiquimod augmented DLI-mediated LH-GVH reactivity as measured by an increase in peripheral blood donor CD8+ T cell chimerism (Fig. 7 E) and an increase in donor CD11c+ DC chimerism in the skin (F). Taken together, these results suggest that the imiquimod-induced activation and emigration of residual chimeric host LCs augments the DLI-mediated GVH response.
To examine whether imiquimod promotes GVHD after DLI, we performed a series of histologic analyses in two MHC-matched models. In the C3H.SW→B6 model, we administered 3 × 106 CD8+ purified donor T cells at the time of BMT or to imiquimod- or vehicle-treated chimeras 8 wk after BMT. As previously published (15, 22, 55), administration of donor CD8+ T cells to the lethally irradiated mice resulted in GVHD-associated changes in the skin, liver, and gut (Fig. 8, A–D). In contrast, delayed administration of 3 million (Fig. 8, A–D) donor CD8+ T cells or 12 million LN T cells (data not shown) to vehicle- or imiquimiod-treated chimeras did not cause any GVHD-induced changes in the same tissues. We therefore conclude that imiquimod, which induces LC activation and emigration, promotes a LH-GVH reaction without inducing histologic or clinical GVHD.
Discussion
We explored DC reconstitution after alloBMT and determined the factors that influence the kinetics of DC turnover and the persistence of host DCs in different compartments. To determine the effect of donor-host histoincompatibility on DC reconstitution, we constructed syngeneic, MHC-matched, and -mismatched chimeras after myeloablative conditioning and transplantation of non-TCD BM. In all three sets of chimeras, we found full reconstitution with donor-derived CD11c+ DCs in the spleen. However, we repeatedly found host-derived CD11c+ DCs in the CLNs of syngeneic and MHC-matched but not -mismatched chimeras. Because all chimeras received the same transplantation conditioning and identical grafts but the percentage of residual host DCs in the CLNs was different, we concluded that the level of residual host DCs after alloBMT is directly influenced by MHC disparity.
This is the first study to compare LC chimerism after MHC-matched vs -mismatched alloBMT and, as such, illustrates the cardinal influence that donor-host histoincompatibility has on the persistence of host LCs. In addition, comparisons of the results reported here with those of previously published studies may help to reveal important aspects of GVH reactions after alloBMT. For example, it has been suggested that host APCs are both necessary and sufficient for the induction of acute GVHD (11, 15). Although this may be true for MHC-mismatched donor recipient pairs (Table I; Ref. 12), it is clearly not the case after MHC-matched alloBMT (Table I) or DLI (Fig. 7). The results of the experiments using imiquimod may yield some insights into the differences between the studies. We found that imiquimod augmented the capacity of DLI to induce a LH-GVH reaction that increased donor hemopoietic chimerism without producing GVHD. It is therefore likely that induction of GVHD after alloBMT requires not only donor T cells and host APCs, but also two additional factors: 1) activation of host APCs by TLR ligands; and 2) tissue damage, which may facilitate emigration of donor T cells into GVHD target organs. In the absence of tissue damage, activation of host APCs by a TLR ligand may be sufficient to promote an LH-GVH reaction but not acute GVHD. Perhaps the best illustration of the role of exogenous TLR ligands in the pathogenesis of GVHD is the prevention of acute GVHD after alloBMT by prior decontamination of the bowel microflora (56), which removes the source of bacterial LPS, a TLR4 agonist. Thus, it is possible that differences in the outcome of MHC-matched alloBMT in different studies (Korngold and Sprent (3); Shlomchik et al. (15); this study) may be explained by differences in the availability of exogenous TLR ligands in different animal colonies.
Nonetheless, an important question remains: why do residual host skin-derived DCs persist in the MHC-matched but not the mismatched setting? One possibility is that the precursor frequency of T cells reactive to MHC Ags is several orders higher in magnitude than the precursor frequency of T cells reactive to miHAs (57). Another possibility is that the MHC-alloreactive repertoire contains a higher number of effector memory T cells by virtue of priming to cross-reactive Ags that are capable of emigrating to nonhemopoietic tissues and eliminating host LCs in the skin (58). Compared with naive T cells, alloreactive memory T cells may have lower activation thresholds or may be better at activating APCs (e.g., through CD40L), which may explain why MHC-mismatched DLI are able to induce an LH-GVH reaction without a requirement for exogenous TLR ligands. Consistent with this explanation, differences in APC activation or T cell activation thresholds may also explain the differential effect of MHC-matched vs -mismatched BMT on donor LC chimerism and acute GVHD. Finally, it has recently been reported that host LCs persist after MHC-matched BMT in humans (59). However, in contrast to our study, in which host LC chimerism persisted indefinitely, in that study patients eventually converted to full donor LC chimerism in the absence of DLI. We propose that the differences may be explained by differences in the proportion of alloreactive memory T cells in the miHA reactive repertoires of mice vs humans, or in the availability of exogenous TLR ligands.
Our studies also addressed the kinetics of donor DC and T cell reconstitution. We found that donor DCs reconstituted rapidly, with <5% of residual host DCs remaining at 1 wk after transplantation. In the CLNs, the kinetics of host-to-donor DC reconstitution was slower, with a notable persistence of host DCs. The decreasing percentage of residual host DCs in the CLNs of chimeras that received escalating doses of TBI clearly suggests that level of their persistence is influenced by the intensity of conditioning. Importantly, our data also demonstrate that donor DCs approach normal numbers faster than do donor T cells. Three weeks after BMT, the total number of CD11c+ cells in the spleen reached almost half of their final numbers, a time at which donor T cells are at only 15% of their final number. Two months after alloBMT, the splenic CD11c+ cellularity is almost normal but the number of donor T cells still lags. These observations suggest that DCs may be more susceptible than T cells to therapeutic manipulation after alloBMT.
Although LCs migrate to the draining LNs following activation, there is also a steady-state migration of these cells in the absence of inflammation (45, 46, 60). Additionally, it is becoming more clear that under steady-state LCs present in the CLNs express maturation markers such as CD86, CD40, and high levels of MHC II, whereas all other DC subsets appear immature (38, 39, 47). Thus, the finding of residual host LCs with an activated phenotype after MHC-matched alloBMT raised questions about their role in the activation of DLI. Several lines of evidence suggest that these cells play a role in stimulating the DLI-mediated GVH response. First, DLI promptly reduced the percentage and total number of residual host CD11c+ DCs, including CD11c+DEC205highCD8α− DCs, in the CLNs of the chimeras. This response was followed by an increase in donor LC chimerism in the skin and profound changes in the CLNs and spleen of the chimeras consistent with a DLI-mediated GVH response. Second, we found that the imiquimod-induced emigration and activation of LCs (52, 53, 54) promoted a DLI-mediated LH-GVH reaction at a time when DLI alone was inert. This finding may have important implications for the treatment of relapsed leukemia after alloBMT, because the major cause of failure of DLI is lack of efficacy, not toxicity. Nonetheless, DLI administration after imiquimod pretreatment did not induce GVHD. As previously discussed, the lack of GVHD after DLI administration despite imiquimod pretreatment may represent the outcome of the TLR ligand-mediated host APC activation without tissue damage. As such, the critical difference that can explain inability of residual host LCs to trigger skin GVHD after adoptive transfer of donor T cells in our experiments, in contrast to those by Merad et al. (12), is lack of conditioning-induced tissue damage.
In conclusion, our observations highlight the potential clinical significance of host LC persistence after MHC-matched alloBMT. They provide a potential mechanism for the efficacy of UV light in the treatment of acute GVHD because host-derived LCs may be present in the skin despite full donor chimerism in the blood (61, 62). Our findings support the strategy of preemptive depletion of host LCs with pretransplant UV with a goal of limiting GVHD in patients undergoing alloBMT for primary BM failure or sickle cell anemia. Finally, they point to residual host LCs as the stimulus for GHVD and graft-vs-leukemia reactivity after DLI. A more direct characterization of residual host LC migration, maturation, and Ag processing, and their effects on the induction of allogeneic and Ag-specific responses is the focus of future studies.
Acknowledgments
We thank Dr. D. P. Pardoll for critical reading of the manuscript and Z. Bajic and K. Vidosusic for statistical analysis.
Disclosures
The authors have no financial conflict of interest.
Footnotes
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 grant from the Amy Strelzer Manasevit Scholars Program funded by the Marrow Foundation in cooperation with the National Marrow Donor Program, National Cancer Institute (K08 CA89546, P01 CA15396, and RO1 CA40358) and the Multiple Myeloma Research Foundation.
Abbreviations used in this paper: DC, dendritic cell; BM, bone marrow; BMT, bone marrow transplantation; alloBMT, allogeneic bone marrow transplantation; GVH, graft-vs-host; GVHD, graft-vs-host disease; miHA, minor histocompatibility Ag; LC, Langerhans cell; DLI, donor lymphocyte infusion; TBI, total body irradiation; TCD, T cell depletion; LN, lymph node; CLN, cutaneous lymph node; TRITC, tetramethylrhodamine isothiocyanate; PI, propidium iodide; Lmm, linear millimeter; epDC, epithelium-derived DC; LH-GVH, lymphohemopoietic GVH.