MHC-mismatched DBA/2 renal allografts are spontaneously accepted by C57BL/6 mice by poorly understood mechanisms, but both immune regulation and graft acceptance develop without exogenous immune modulation. Previous studies have shown that this model of spontaneous renal allograft acceptance is associated with TGF-β-dependent immune regulation, suggesting a role for T regulatory cells. The current study shows that TGF-β immune regulation develops 30 days posttransplant, but is lost by 150 days posttransplant. Despite loss of detectable TGF-β immune regulation, renal allografts continue to function normally for >200 days posttransplantation. Because of its recently described immunoregulatory capabilities, we studied IDO expression in this model, and found that intragraft IDO gene expression progressively increases over time, and that IDO in “regulatory” dendritic cells (RDC) may contribute to regulation associated with long-term maintenance of renal allografts. Immunohistochemistry evaluation confirms the presence of both Foxp3+ T cells and IDO+ DCs in accepted renal allografts, and localization of both cell types within accepted allografts suggests the possibility of synergistic involvement in allograft acceptance. Interestingly, at the time when RDCs become detectable in spleens of allograft acceptors, ∼30% of these mice challenged with donor-matched skin allografts accept these skin grafts, demonstrating progression to “true” tolerance. Together, these data suggest that spontaneous renal allograft acceptance evolves through a series of transient mechanisms, beginning with TGF-β and T regulatory cells, which together may stimulate development of more robust regulation associated with RDC and IDO.
It has been known for many years that transplantation of fully mismatched renal allografts in nonimmunosuppressed mice results in spontaneous acceptance for some strain combinations. This curiosity was first reported by Russell et al. (1, 2) who demonstrated immune alteration during the first 30 days posttransplantation in renal allograft acceptors. Despite these initial observations from almost 30 years ago, little progress has been made to fully explain them. Many believe the initial speculation of Russell et al. (1, 2) that these models, which are not dependent upon immunosuppressive strategies, might hold important clues to development of immune tolerance. Our group has therefore studied spontaneous acceptance of DBA/2 renal allografts by C57BL/6 mice, hoping to elucidate mechanisms that lead to long-term renal allograft acceptance, and to gain further insight into development of tolerance.
We have reported that, like cardiac allografts in this strain combination, renal allograft acceptance occurs despite presence of alloreactive T and B cells. Interestingly, spontaneous renal allograft acceptors display regulation of splenocyte alloresponse to donor Ag. Presence of alloreactive T cells and regulation are demonstrated as delayed-type hypersensitivity (DTH)3 detectable immune regulation mediated by splenic T cells involving TGF-β (3, 4). In addition, these allograft acceptors also develop donor-reactive Abs (3, 4). Given its known immunoregulatory functions, and its association with T regulatory cells (Tregs), involvement of TGF-β suggested that regulation associated with spontaneous renal allograft acceptance involved Tregs.
Although we assumed involvement of Tregs, subsequent data suggested that Tregs alone are not robust enough to explain allograft acceptance and regulation in this model (5). Immunoregulatory mechanisms involving TGF-β and Tregs have been identified in rodent recipients that have spontaneously accepted renal allografts, but their association with graft acceptance is poorly understood (3, 6). Additionally, TGF-β-associated immune regulation is only transiently expressed and is not observed in late renal allograft acceptor splenocytes (>150 days posttransplant) (5). These observations led us to explore several possibilities regarding immune regulation in our model: 1) graft acceptance might not require Tregs or TGF-β-mediated regulation, which only develop as epiphenomena; 2) persistence of allograft acceptance beyond 150 days involves replacement by a subsequent regulatory mechanism or, alternatively, clonal anergy or deletion; or 3) Tregs relocate or establish themselves locally in accepted grafts, and are not maintained in the periphery. Even more fundamentally, it is unknown whether spontaneous renal allograft acceptance represents true immunologic tolerance, or merely prolonged or delayed rejection, given our previous observations that early renal allograft acceptance is associated with development of alloantibodies and mild fibrosis (3). We therefore studied the long-term fate of spontaneously accepted renal allografts to determine whether graft survival persists over time, and what mechanisms might contribute to long-term allograft acceptance and tolerance.
In the current report, we present evidence suggesting that early renal allograft acceptance is associated with TGF-β-induced DTH immune regulation, both peripherally by splenocytes as well as locally by graft-infiltrating cells (GICs). Curiously, during the late transplant period (>150 days), TGF-β-associated mechanisms no longer appear to be operative in GICs or splenocytes and, even more interestingly, some renal allograft acceptors are able to accept donor-matched skin allografts at this time. Our data suggest that early TGF-β-associated mechanisms may be involved in induction, but not long-term maintenance, of allograft acceptance. To preserve late graft function in the absence of TGF-β-associated mechanisms, presumably some other mechanism must be operational. We demonstrate that IDO expression is significantly increased in accepted renal allografts 150 days after transplantation, and further, that immune regulation at day 150 may be mediated by IDO. Perhaps most importantly, we demonstrate that regulatory dendritic cells (RDC) are one likely source of IDO in late allograft acceptors, and that these cells are capable of controlling alloresponses. Both RDCs and Tregs localize in accepted allografts, suggesting that their synergism might contribute to maintenance of tolerance. Taken together, these data suggest that spontaneous renal allograft acceptance is a complex immune response that evolves through different mechanistic steps over time, and that acceptance and tolerance may ultimately be mediated by RDCs.
Materials and Methods
C57BL/6 (H-2b), DBA/2 (H-2d), and FVBN (H-2q) mice were obtained from Taconic Farms. All mice were housed and treated in accordance with Animal Care Guidelines established by the National Institutes of Health and the Ohio State University.
Murine kidney transplantation
Murine kidney transplantation was performed as described by Zhang et al. (7). Briefly, donor left kidneys were isolated by ligating and dividing the adrenal and testicular vessels with microsuture. The aorta and inferior vena cava (IVC) were mobilized at their junction, with the left renal artery and vein. The aorta was ligated below the renal vessel. An elliptical patch of bladder containing the left ureterovesical junction was excised. Grafts were perfused in situ with 0.2–0.4 ml of cold, heparinized Ringer’s lactate. Finally, kidneys with their vascular supply and ureter attached to the bladder patch were harvested en bloc. Recipient right native kidneys were removed immediately before transplantation. The infra-renal aorta and IVC were carefully isolated and cross-clamped. An end-to-side anastomosis between donor renal vein and the recipient IVC was performed. Following successful anastomosis, the kidney graft perfused instantly. Urinary reconstruction was then performed by a bladder-to-bladder anastomosis. The left native kidney was removed 1 wk posttransplantation. Kidney graft survival was followed by daily examinations of overall animal health and weekly serum creatinine checks. All transplanted mice received a single s.c. injection of penicillin (500 U/10 g) just after surgery in addition to drinking water containing sulfatrim (100 mg/kg) until the experiment was terminated.
Serum creatinine determination
Quantitative serum creatinine levels were determined using kits from Roche Molecular Biochemicals. Creatinine reagents and a Boehringer Mannheim/Hitachi analyzer were used to perform the analysis. Conventional units (mg/dl) were converted to SI units by multiplying conventional units by 88.4. Concentrations of creatinine in the serum are expressed in micromoles per liter.
Murine skin transplantation
Skin allografts were performed using abdominal skin from donor mice. Square full-thickness grafts (∼8 × 10 mm) were placed on graft beds prepared on each recipient’s flank. Grafts were covered with protective bandages for 7 days. Rejection was considered to occur when grafts exhibited dark discoloration, scabbing, and necrotic degeneration.
Graft infiltrate isolation
Renal graft tissues were excised and suspended in 1 mg/ml collagenase (Sigma-Aldrich). Grafts were cut into small pieces in a petri dish using a scalpel and small plunger and incubated for 40 min at 37°C. Cells were washed and removed from the debris followed by RBC lysis. After three washes in PBS, cells were used in transfer DTH assays.
Transfer DTH responses
GICs or splenocytes were isolated from kidney acceptor mice at 60 and 150 days posttransplant and were tested in transfer DTH assays. For this assay, pinnae of naive B6 mice were injected using a 30-gauge insulin syringe, with 35 μl containing cells (either 1–3 × 106 GIC, 8 × 106 renal allograft acceptor unfractionated splenocytes, 8 × 106 renal allograft acceptor DC-depleted splenocytes, 8 × 106 skin allograft rejector splenocytes, or 0.25 × 106 DC+ renal allograft acceptor splenocytes) plus subcellular donor alloantigens ± 25 μg of neutralizing TGF-β or IDO Abs. Changes in ear thickness were measured both before injection and 24 h after injection using a dial-thickness gauge (Swiss Precision Instruments). For reference, changes ranging 0–30 × 10−4 inches represent background swelling due to injection trauma, changes ranging 40–60 × 10−4 inches represent moderate DTH responses, and changes ranging 70–100 × 10−4 inches represent strong DTH responses. It was also noted that in 33% of experiments performed, injections of GIC alone from day 60 and 150 renal allograft acceptors resulted in swelling responses ranging 50–90 × 10−4 inches. We have previously observed this “hot prep” phenomenon when cell isolates are contaminated with RBC or platelets (A. A. Bickerstaff and C. G. Orosz, unpublished observations). Although it might also be the nature of these cells, these “hot” results were seen equally in day 60 and 150 graft acceptors. Thus, only experiments for which an acceptable background swelling response was obtained for GIC alone (<50 × 10−4 inches) were included in data presented in this article. Renal allograft acceptor splenocytes contain normal numbers of T cells when compared with nontransplanted control mice (data not shown).
Subcellular alloantigen was prepared according to published methods of Engers et al. (8). Briefly, fresh RBC-depleted DBA/2 splenocytes suspended in PBS were subjected to three rapid freeze/thaw cycles, using liquid nitrogen, and spun at 13,000 rpm for 30 min to remove residual debris. Supernatants were adjusted to 3–5 mg of protein/ml and used as subcellular alloantigen. For DTH challenge, 25 μl (75–125 μg of protein) of this solution was injected into murine pinnae.
Polyclonal rabbit anti-TGF-β and control rabbit Ig were obtained from R&D Systems. Polyclonal rabbit anti-IDO Ig was obtained from Alexis Biochemicals. In addition to anti-IDO Ab, IDO activity was neutralized using 1-methyl-d-tryptophan (1-MT) obtained from Sigma-Aldrich.
Presence of donor-reactive IgG Ab was determined by the ability of sera to bind to DBA/2 thymocytes. Binding was detected by flow cytometry, using FITC-conjugated goat anti-mouse IgG (γ-chain specific; Pierce). Results are shown as percentage of DBA/2 thymocytes that bound detectable alloantibody. Treatment of DBA/2 thymocytes with sera from naive C57BL/6 mice resulted in <5% binding, and sera from allosensitized C57BL6 mice react with >98% of thymocytes.
Histological examination of renal tissue
Renal tissues were excised and fixed in 10% neutral-buffered formalin, dehydrated in upgraded ethanol (70, 95, and 100%), and embedded in paraffin. For histological analysis, 4-μm sections were mounted on slides and stained with H&E or trichrome.
Immunohistochemistry was performed on formalin-fixed, paraffin-embedded tissue from representative accepted renal DBA to B6 allografts (n = 12, days 28–158) and B6 to B6 isografts (n = 9, days 30–153) with a double marker technique optimized for the simultaneous identification of Foxp3 and a surface differentiation molecule CD3. Sections were baked for 30 min in oven, deparaffinized in xylene, rehydrated in absolute and 95% ethanol, incubated for 5 min in 3% H2O2 in methanol to block endogenous peroxidase, and washed in TBS/Tween 20. Ag retrieval was done with Borg Decloaker solution (pH 9.5; Biocare Medical) in a pressure cooker. Blocking normal goat serum (1/50) and avidin D (1/10) were used, followed by CD3 polyclonal Ab (DakoCytomation) diluted at 1/400 in 1% TBS/BSA and incubated overnight at 4°C. After biotin blockade, slides were incubated with biotinylated goat anti-rabbit IgG secondary Ab (Vector Laboratories) for 35 min, followed by streptavidin peroxidase (Biogenex) for 60 min, and developed with 3,3′-diaminobenzidine. All steps included washing with TBS/Tween 20 in between. Slides were then incubated in blocking normal goat serum (1/50) and avidin D (1/10) for 20 min. An affinity purified rat anti-Foxp3 (clone FJK-16s) Ab (eBioscience) diluted at 1/50 in 1% TBS/BSA was incubated overnight at 4°C. After biotin blockade, a biotinylated rabbit anti-rat (mouse ads) IgG secondary Ab (Vector Laboratories) was used for 35 min followed by avidin-biotinylated-alkaline phosphatase complex (ABC-AP; Vector Laboratories) for 60 min. All steps included washing slides in between with TBS/Tween 20 and with TBS only before developing Foxp3 staining. Tissue sections were developed with Vector Blue alkaline phosphatase substrate (Vector Laboratories) and mounted with Faramount (DakoCytomation).
For IDO, after the Decloaker and blocking steps above, polyclonal rabbit anti IDO Ab at a 1/2000 dilution (Alexis Biochemical) was added, and the sections were incubated overnight at 4°C. Sections were then rinsed in PBS and treated with biotin (10 μg/L PBS), washed three times in PBS, and incubated in the biotinylated goat anti-rabbit IgG for 35 min. Slides were washed and incubated for 1 h in streptavidin peroxidase (Biogenex). Sections were washed and developed with Romulin AEC Chromogen (Biocare Medical), counterstained with hematoxylin, dehydrated in ethanol, cleared in xylene, and coverslipped with permanent mounting medium.
Sections stained for Foxp3 and CD3 were scored for positive cells per ×10 fields in coded sections and converted to cells per millimeter-squared (×10 field = 2.14 mm2). Sections stained for IDO were scored qualitatively according to the type and frequency of positive cells.
Foxp3 and IDO RT-PCR
Total RNA was isolated from kidney tissue using TRIzol (Invitrogen Life Technologies). A total of 2 μg of total RNA was reverse transcribed using Qiagen Omniscript RT kit and amplified using Superarrays gene-specific RT2 End-point Foxp3 and IDO PCR kits. Samples were separated on a 2% ethidium bromide containing agarose gel and bands visualized through UV transillumination. Images were captured using a Kodak DC120 camera and analyzed using Kodak 1D Gel Analysis Software. Band intensity was normalized using a mouse GAPDH internal control and reported as relative units.
MACS microbead cell separations
Splenocytes were depleted of DCs using MACS microbead technology developed by Miltenyi Biotec. Cells were labeled with pan DC microbeads, and passed through a separation column, LS+/VS+, in a magnetic field. The positive and negative fractions of cells were separately collected and used for further analysis.
Long-term renal allograft acceptor function
We have previously reported that ∼70% of C57BL/6 mice spontaneously accept heterotopic DBA/2 renal allografts for at least 60 days (3). Longitudinal graft function in renal allograft acceptors was evaluated for deterioration over time. Baseline serum creatinine levels were maintained in renal allograft acceptors when measured either at day 60 (30 ± 15 μM/L), >day 150 (31 ± 10), or at day 200 (36 ± 13). Note that day 200 was following a donor skin graft challenge at day 150. These levels were essentially identical with levels that were measured in normal, nontransplanted animals and in C57BL/6→C57BL/6 renal isograft acceptors during the same time periods (data not shown). Mice driven to reject renal allografts by priming with skin allografts (14 days before renal grafting) develop much higher creatinine levels (102 ± 12 μM/L) within 14 days of kidney transplantation (data not shown).
Relocation of active immune regulation
We have reported that renal allograft acceptors gain DTH-detectable immune regulation by 60 days posttransplant, but maintain the ability to respond to third-party Ags such as tetanus toxin (3). In those studies, recipient splenocytes were tested for donor-reactive immune regulation in transfer DTH assays. Other previous studies have suggested that development of donor-reactive immune regulation in spleens of cardiac allograft acceptor and renal allograft acceptor mice is transient (5). In the present study, we extended DTH testing of splenocytes to 150 days posttransplant to determine the immune status of the recipient at this later time point. As shown in Fig. 1 , left panel, splenocytes show DTH-detectable, TGF-β-mediated, graft-reactive immune regulation on day 60, that becomes non-TGF-β-mediated by day 150 posttransplant. Consistent with our previous studies (3), IL-10 was also unable to restore DTH responses (data not shown). Despite losing peripheral immune regulation mediated by TGF-β by day 150, allografts maintained normal function. Because graft function was maintained, we hypothesized that the site of graft-reactive immune regulation migrates over time from peripheral lymphoid organs to the graft itself.
To test this hypothesis, we performed transfer DTH assays using allograft-infiltrating lymphocytes derived from renal allograft acceptor mice. Accepted DBA/2 renal allografts were obtained from C57BL/6 recipients either 60 or 150 days posttransplant. GICs were isolated by collagenase digestion and were transferred into pinnae of naive C57BL/6 mice along with subcellular donor alloantigen. Donor-reactive DTH responses were measured as ear swelling 24 h later. As shown in Fig. 1 , right panel, 60-day kidney GICs did not elicit ear swelling when challenged with donor alloantigen unless TGF-β was neutralized at DTH sites. Similarly, 150-day kidney GICs were unable to promote donor alloantigen-induced ear swelling. Interestingly, TGF-β-neutralizing Abs did not restore donor alloantigen-induced ear swelling for 150 day kidney GICs. Thus, within 60 days posttransplant, cells present within accepted renal allografts demonstrate DTH-detectable, TGF-β-mediated, graft-reactive immune regulation. By 150 days after transplantation, however, graft-reactive immune regulation no longer appears to be dependent upon TGF-β.
Together, these findings suggest that both spleen and graft serve as concurrent active sites for expression of graft-reactive immune regulation in renal allograft acceptors. Furthermore, it appears that loss of TGF-β-mediated splenic immune regulation does not reflect late migration of this regulatory center from peripheral lymphoid organs to the graft itself.
Previous studies have demonstrated that donor-reactive alloantibodies are generated by renal allograft acceptor mice at levels comparable to rejecting heart graft recipients (3). In the current study, we tested for alloantibody persistence. Donor-reactive humoral immune responses were tested in renal allograft acceptors by flow cytometric analysis of serum reactivity with donor thymocytes. All kidney allograft acceptors demonstrate donor-reactive Abs, and as shown in Fig. 2, these Abs persist 150 posttransplant. Additional analyses demonstrated that alloantibody titers were 50-fold lower in renal allograft acceptors compared with skin allograft rejectors (data not shown). These findings confirm that donor-reactive humoral immune responses in renal allograft recipients develop, and persist under conditions in which donor-reactive cellular immune responses are subject to immunoregulation.
Challenge with donor skin allografts
Renal allografts induce long-lasting, donor-reactive alloantibodies, and yet continue to function after apparent loss of DTH-detectable immune regulation, raising the possibility that immune tolerance to donor Ag develops in renal allograft acceptors. Currently, the most stringent functional test for alloantigen tolerance in murine organ transplant models is acceptance of donor-matched skin allografts. To test this hypothesis, cohorts of DBA/2→C57BL/6 renal allograft acceptor mice received DBA/2 skin allografts either 60, 150, or 180 days posttransplant. As shown in Fig. 3 A, 100% of 60-day kidney allograft acceptors reject DBA/2 skin grafts within 24 days posttransplant, and six of eight acceptors retain functional kidney allografts for >50 days postskin transplant. In surviving mice, serum creatinine levels remain normal after skin grafting, indicating that renal allograft function does not deteriorate following skin allograft rejection.
As shown in Fig. 3,B, all 150-day renal allograft acceptors retain normal renal function after DBA/2 skin allograft implantation, and 30% (3 of 10) have skin allograft acceptance for >50 days postskin transplant. These 150-day acceptors simultaneously reject all concurrently placed, third-party FVBN (H-2q) skin allografts within ∼15 days (Fig. 3,B). Thus, some renal allograft acceptors gain the ability to accept donor-matched skin grafts by day 150 posttransplant, despite loss of detectable TGF-β-mediated DTH immune regulation. These data suggest progression to true tolerance in 30% of renal allograft acceptors between 60 and 150 days posttransplant. To determine whether this progression continues to evolve over time, we similarly studied 180-day renal allograft acceptors. As shown in Fig. 3 C, 33% (2 of 6) of 180-day renal allograft acceptors have donor-matched skin graft acceptance and all retain normal renal function and their ability to reject third-party FVBN skin allografts. This indicates no further progression in development of tolerance over time.
Recipient renal function is maintained solely by allograft kidneys in this model, thus serum creatinine levels were used to monitor allograft function throughout all experiments. Because renal tissues can show major histological injury before physiologic function is impaired to a detectable degree, accepted allografts were also examined histologically after staining with H&E. As shown in Fig. 4, day 60 (Fig. 4,A) and day 150 (Fig. 4, B and C) accepted renal allografts exhibit abnormal histology. Both undergo prominent leukocytic cuffing of larger vessels, as well as mild fibrosis, while most renal parenchyma is unremarkable. Interestingly, diffuse interstitial leukocyte migration that is characteristic of acute cardiac allograft rejection was not observed in accepted renal allografts. Observed histopathology in accepted renal allografts is alloantigen dependent, as no similar leukocytic infiltration is seen in accepted, day 150 renal isografts (Fig. 4 D) which maintain normal tissue histology throughout this time period.
As suggested in Fig. 2, most renal allografts endure prolonged exposure to circulating, donor-reactive Abs. The immunologic and histologic impact of these alloantibodies remains unclear. We have confirmed by immunofluorescence that C3d is not depositing in accepted allografts (data not shown). In cardiac allografts, alloantibodies have been associated with development of neointimal formation and tissue fibrosis (9). Neointimal formation is not observed in arteries of the renal allografts, but mild interstitial fibrosis is present (Fig. 4 C). Based on these observations, both day 60 and 150 renal allografts appear to be the focus of an active, but unusual, immune process.
Allograft Foxp3 expression
Foxp3 expression is unique to Tregs (10, 11, 12, 13), thus presence of Tregs in renal allografts should be mirrored by expression of Foxp3. To determine presence of active Tregs, total RNA was isolated from accepted renal allografts at 30, 60, and 150 days posttransplant to evaluate for presence or absence of Foxp3. Controls included normal, nontransplanted DBA/2 and C57BL/6 kidneys and renal isografts collected at 30, 60, and 150 days posttransplant. These RNA were analyzed by RT-PCR for expression of Foxp3. As shown in Fig. 5, Foxp3 mRNA is significantly elevated in renal allografts by day 30 posttransplant, but decreases to control kidney levels by day 150. Foxp3 expression in renal isografts is comparable to background with only a slight increase at day 150. Increased expression of Foxp3 in 30- to 60-day renal allografts occurs concomitantly with expression of TGF-β-mediated DTH immune regulation. It is intriguing that loss of TGF-β-mediated immune regulation between days 60 and 150 (Fig. 1) occurs simultaneous to decreases in Foxp3 transcription. We therefore conclude that diminishing levels of Foxp3 suggest decreased presence or activity of Treg cells, and further, that alternate mechanisms must be involved in subsequent maintenance of renal allograft acceptance.
Allograft IDO expression
IDO, which is known to be expressed by DCs (14), is known to suppress alloreactive T cell responses (15, 16, 17, 18). Because the TGF-β and Foxp3 data together suggested a diminishing role for Tregs, we hypothesized that late tolerance might be maintained by transition from Treg mediated to DC mediated mechanisms in accepted allografts. Accepted allografts were therefore evaluated by RT-PCR for presence of IDO transcripts. As shown in Fig. 6, IDO mRNA expression increases over time in accepted renal allografts, becoming significantly elevated by day 150 posttransplant. In contrast, low IDO mRNA levels are found in normal nontransplanted kidneys, and renal isografts show only slight elevation by day 150. Thus, the increasing presence of IDO over time in accepted renal allografts suggests possible development of an alternate mechanism of allograft acceptance dependent upon immunoregulatory effects of IDO.
Evidence of IDO-mediated immune regulation of alloresponse
IDO has been shown to inhibit alloreactive responses in vitro, yet there are few in vivo data to support this (16, 17, 18, 19). As shown in Fig. 2, 150 days after transplantation, splenic donor-reactive DTH responses remain suppressed, but no longer appear to be associated with TGF-β. Because IDO transcription is elevated at this time point (Fig. 6), we hypothesized that alloreactive immune responses were suppressed by IDO. To test this, splenocytes from 60- or 150-day renal allograft acceptors were combined with donor alloantigens with or without neutralizing Abs to IDO and used in our transfer DTH assay. As shown in Fig. 7, day 150 allograft acceptor splenocytes show IDO-mediated, donor-reactive immune regulation. In contrast, IDO regulation does not appear to be active at day 60, when TGF-β appears to be active. We initially chose to neutralize IDO in our DTH assay with Abs because of our experience with other neutralizing Abs, but in retrospect it was somewhat surprising that these neutralized IDO function, because IDO is thought to be an intracellular enzyme. Accordingly, this experiment was repeated using 1-MT, a known inhibitor of IDO and tryptophan metabolism (20, 21). Addition of 1-MT at DTH challenge sites restores DTH reactivity with results identical with those observed with anti-IDO Ab (data not shown). Anti-IDO Ab or 1-MT mixed with splenocytes (without donor Ag) show DTH responses consistent with negative controls (data not shown). These results suggest that regulation of donor-reactive DTH responses transitions from TGF-β mediated early (60 days) to IDO mediated later (150 days).
Evidence of immune regulation by DCs
DCs are known to be an important source of IDO (14, 22, 23). Given the compelling IDO data from unfractionated splenocytes, we hypothesized that DCs might be the source of IDO associated with regulation in spontaneous renal allograft acceptors (Fig. 7). To test this hypothesis, DCs were isolated from splenocytes of late renal allograft acceptors and combined with alloreactive splenocytes from sensitized (alloreactive) mice. DC regulation was then tested using transvivo DTH. As shown in Fig. 8, when DCs from allograft acceptors are added to alloreactive splenocytes, tolerance to alloantigen is induced. Furthermore, blocking IDO restores alloresponses, suggesting that this conferred tolerance is mediated by IDO. DCs from naive mice were similarly isolated, and show no regulation of alloreactive splenocytes (data not shown). These results thus suggest that RDCs are present in spontaneous renal allograft acceptors, and that these cells have the capacity to regulate alloresponse in our DTH model.
Tregs and RDCs in accepted allografts
Although our RT-PCR and DTH data were suggestive that there is transition from Treg-regulated immune responses to RDC-regulated responses in allograft acceptors, it remained unclear whether Treg or RDCs were present in accepted allografts. Immunohistochemistry analyses of late accepted allografts show prominent infiltration of CD3+ cells becoming less diffuse and more nodular with time (mean infiltrate 101.9 ± 44.4 cells/mm2). Foxp3+ cells initially infiltrate diffusely in the cortex, eventually developing nodular aggregates around small arteries or tubules (35.2 ± 40.1 cells/mm2) (Fig. 9,A). Inconsistent with our quantitative RT-PCR results, allograft Foxp3 staining density seems to increase over time (Foxp3+ cells days 30–90 = 18.4 ± 11.1/mm2; days 150+ = 60.2 ± 56.1/mm2). Of all CD3+ cells, 21.9 ± 10.8% were Foxp3+, a fraction that increases over time from 17 ± 10% (days 30–90) to 28 ± 10% (days >150). Conversely, almost all (98.8%) Foxp3+ cells express CD3. Cells expressing IDO are prominent in allografts at all times studied. IDO-positive DCs (Fig. 9, B and E) are identified as larger angular cells within nodular aggregates. In addition, IDO-positive arterial endothelium and tubules are often associated with nodular infiltrates of Foxp3 cells (Fig. 9 E).
In contrast, isografts show scattered CD3+ cells (35.4 ± 18.2/mm2) and rare Foxp3+ cells (0.5 ± 0.4/mm2) (Fig. 9,C). Foxp3+ cells comprise 1.9 ± 2.0% of CD3+ cells. Density of these cells does not change appreciably between days 30 and 150 (data not shown). Rare IDO+ tubular cells are found in the cortex in early samples (day ∼30). No IDO+ DC or endothelial cells are identified in any isografts at any time studied (Fig. 9 D). These data thus confirm coexistence of both Tregs and RDCs in spontaneously accepted renal allografts.
This study demonstrates for the first time the presence of DCs with a regulatory phenotype (RDCs) in mice that spontaneously accept MHC-mismatched renal allografts. It appears that development of spontaneous long-term renal allograft acceptance in mice is associated with multiple immunoregulatory mechanisms that evolve over time. Early escape from cellular rejection is associated with regulation of graft-reactive T cells by a transient TGF-β-mediated mechanism, while later escape is associated with continued regulation of graft-reactive T cells by IDO and not TGF-β. Our data further suggest that clonal deletion is not required for spontaneous renal allograft acceptance, and that persisting alloreactive cells may be regulated by IDO secreted by RDCs. Even more intriguing, spontaneous development of true tolerance occurs in ∼30% of these graft acceptors at late time points, when RDCs can be isolated from their spleens. Although these data are far from conclusive, they support the very exciting possibility that development of immune regulation and tolerance may involve development of RDCs.
We have previously demonstrated that allosensitization occurs in renal allograft acceptors evidenced by presence of donor-reactive T cells and alloantibody (5, 24). Despite this allosensitization to donor Ag, data from the current study confirm that accepted grafts continue to function indefinitely (Fig. 3), confirming that this model is not one of delayed or slow rejection. Spontaneous renal allograft acceptors display immune regulation of cellular responses, and our data suggest that regulation of cellular responses appears to be divided into at least two functionally distinct phases: early and late (arbitrarily defined as <90 and >150 days, respectively) after transplant. Early on, donor-reactive cells may be regulated at least in part by TGF-β, as anti-TGF-β Ab restores reactivity to donor Ag in DTH studies (Fig. 1). This regulation occurs both systemically in splenocytes, and locally in GICs (Fig. 1). Because TGF-β is known to stimulate maturation of CD4+CD25+ Tregs (13, 25, 26, 27), it is perhaps not surprising that Foxp3, a marker of active Tregs (10, 28, 29, 30) is highly expressed in accepted allografts when TGF-β appears to be contributing to regulation. Because functional tolerance must depend at least in part upon regulation of cellular or humoral responses to donor Ag, we postulate that early expression of TGF-β in accepted renal allografts stimulates or activates Tregs, contributing to induction of early tolerance.
Since their first description by Sakaguchi et al. in 1995 (31), we have been enamored with the possibility that Tregs contribute to our models of immune regulation and allograft tolerance. These unique cells are now known to regulate both CD4 and CD8 T cell responses, and there are accumulating data that support the role of regulatory T cells in long-term allograft acceptance (3, 27, 32, 33, 34, 35). Recently, Waldmann and colleagues have shown that Tregs are both highly enriched in accepted allografts (36) and dependent on TGF-β (27), and that TGF-β appears to be an essential intermediary in establishing long-term tolerance (37). We too have shown a role for CD4+CD25+ regulatory cells in our heart transplant model (24), and in the current study confirm the presence of Foxp3+ cells within our accepted renal allografts (Fig. 9 A). Based upon these results and those of others, we therefore hypothesized that TGF-β and Tregs might contribute to long-term regulation.
Despite our convictions, however, Tregs alone are not robust enough to explain all of our experimental findings. We have previously observed that when Tregs are depleted from allograft acceptor splenocytes, tolerance to alloantigen still persists in remaining splenocytes, and this tolerance is transferable to alloreactive cells (5). In addition, current data shows loss of TGF-β-mediated regulation in renal allograft acceptors over time. As Fig. 1 clearly demonstrates, anti-TGF-β Ab no longer restores reactivity to alloantigen in late allograft acceptors. Three possible conclusions could initially be drawn from these observations. First, there could be a switch from TGF-β-mediated regulation, to some other subsequent mechanism that maintains regulation of alloreactive cells. Second, there could be clonal deletion or anergy of alloreactive cells. Third, transient TGF-β regulatory mechanisms might be an epiphenomenon, associated with, but not responsible for, acceptance of renal allografts.
Our data strongly suggest that there is transition from TGF-β to IDO-mediated regulation by day 150 posttransplant. Concomitant with loss of TGF-β-mediated regulation, there is increasing IDO expression in accepted allografts (Fig. 6). More importantly, IDO appears to mediate suppression of donor-reactive immune responses of day 150 renal allograft acceptor splenocytes (Fig. 7). One source of IDO in our model appears to be RDCs localized both in accepted allografts (Fig. 9) and in spleens of late allograft acceptors (Fig. 8). Perhaps most compelling is the observation that RDCs isolated from allograft acceptors are capable of regulating fully alloreactive cells (Fig. 8). Taken together, these findings confirm persistence of alloreactive T cells in our model, and suggest transition to RDC-mediated IDO immune regulation by day 150.
There is substantial and growing interest in IDO and its tolerogenic properties, which are thought to occur primarily through profound influences upon T cell-mediated responses (15, 16, 17, 38, 39). Functionally, IDO is known to prevent rejection of fetuses during pregnancy by inhibiting alloreactive T cells (20), modulate immune resistance of tumors (40), and regulate alloreactive T cell responsiveness (15, 16, 17, 18). Other investigators have recently identified more specific roles for IDO in transplantation tolerance. Using a rat model, Degauque et al. (6) have shown that accepted renal allografts accumulate Foxp3 and IDO 100 days posttransplant, and that IDO plays an important role in controlling tolerant T cell responses to donor alloantigens. Yet other studies have demonstrated prolongation of islet graft survival attributed to IDO function (41). Our current data suggest that IDO may indeed play a regulatory role in spontaneous renal allograft acceptance.
Fortunately, our Treg and RDC/IDO hypotheses are not mutually exclusive, and in fact published data and results from current experiments suggest that there is perhaps a synergistic relationship between Tregs and DCs in our model (42, 43). Mechanistically, Tregs have been demonstrated to “collaborate” with DCs to initiate IDO production (44, 45). Conversely, it has also been shown that IDO+ DCs are capable of stimulating development of Tregs (46, 47, 48). Our immunohistochemistry results (Fig. 9) demonstrate presence of lymphoid aggregates containing Foxp3+ CD3 cells in proximity to IDO+ DCs in accepted allografts, which is consistent with the hypothesis of synergism. In addition to DCs, we observe that other cells such as arterial endothelium and tubular epithelium in these lymphoid nodules also express IDO. Similar findings have been recently reported by Thebault et al. (43) in a cardiac allograft model, whose results suggest that Foxp3 cells traffic to allografts and induce endothelial expression of IDO. Although we were initially surprised that late allografts contained Foxp3+ cells, given our quantitative RT-PCR data, Tregs are indisputably present in late allografts, and actually appear to increase numerically over time. The reason for this obvious discrepancy is unclear to us, but could include changes in mRNA stability, decreased Foxp3 protein expression per positive cell, or simple methodological differences from others who report Foxp3 mRNA expression relative to numbers of CD3+ cells (27). Whatever the cause, this early erroneous finding was fortuitous because it suggested to us the possibility of an alternate mechanism of regulation. Although not direct proof, our data taken together with that of others support the hypothesis that RDCs and Tregs might play complementary roles in tolerance induction, with Tregs possibly conditioning DCs to become RDCs capable of controlling alloresponses to donor Ag.
Although the data presented most strongly support transition away from a TGF-β mechanism to an IDO-mediated mechanism of regulation, an alternate early interpretation was that our findings might represent clonal deletion of alloreactive T cells, or onset of anergy. Mechanisms such as clonal deletion and anergy have been documented in other transplant tolerance models (49, 50). Nonetheless in our model, blocking IDO restored late alloantigen responsiveness, confirming that alloreactive T cells are still present, and not completely deleted from allograft acceptors. In addition, renal allograft acceptors maintain their ability to reject third-party Ags at least 150 days posttransplant (Fig. 3), confirming specificity of this regulation, and further lack of general anergy. Although we have not ruled out the possibility that there is partial clonal deletion of alloreactive cells occurring by day 150, there are clearly alloreactive splenocytes that remain in these recipients that must be controlled by some regulatory mechanism.
In addition to regulation of cellular alloimmune responses, our data also suggest that spontaneous renal allograft acceptors might display regulation of humoral alloreactivity. We have found that cardiac (51) and renal (Fig. 2) allograft acceptors display circulating donor-reactive IgG alloantibodies, but despite these Abs, both demonstrate long-term graft acceptance and function. It is also interesting that despite persistence of preformed Ab in these mice, 30% of renal allograft acceptors at day 150 do not reject donor-matched skin allografts (Fig. 3). Although one could argue that graft-reactive Abs induced in our model are of low affinity, of the wrong isotype, or of inadequate titer to disrupt tolerance, it is curious that this same antigenic challenge (skin) in naive C57/BL6 recipients causes 50-fold higher Ab titers than those seen in our renal allograft acceptors challenged 60 days posttransplant. It has been previously shown that under certain circumstances, B cell responses are dependent upon T cell function (52, 53), so it is possible that spillover of regulation from the cellular to the humoral immune response might be occurring in our model. Although our data are inadequate to prove this hypothesis, and the argument for humoral regulation is admittedly not as strong as that for cellular regulation, our results do not dispute this possibility. Alternatively, there are growing data suggesting that under proper circumstances, B cells are tolerogenic rather than immunogenic (54), and can in fact induce Tregs (55). Analogous to our changing understanding of the importance of alloantibody responses during human transplantation, which continues to evolve (56), the importance of Ab and B cells in our model of allograft acceptance will clearly require further exploration.
Our model of renal allograft acceptance demonstrates a strong association with immunoregulatory mechanisms, but to fit the strictest definition of tolerance, animals should accept subsequent donor-matched skin grafts. Interestingly, ∼33% of late renal allograft acceptors became fully tolerant to donor Ag, compared with 0% of early renal allograft acceptors (Fig. 3,A). Additionally, no late renal allograft acceptors demonstrated renal allograft loss when challenged with donor skin, compared with 25% losses in early renal graft acceptors. Encouraged by our day 150 skin transplant acceptance, we performed day 180 donor-matched skin transplants in another cohort, but saw no further improvement in tolerance (Fig. 3 C). It is possible that given more time, this tolerance might progress to involve a larger percentage of graft acceptors, and these studies are currently ongoing. Nonetheless, it is extremely exciting that true tolerance may be developing over time, despite presence of donor-reactive T and B cells. It is highly intriguing that at the time when some animals become tolerant to donor skin, there are DCs in their spleens capable of regulating donor responses. Although not proven here, we suspect that it is not simple coincidence that true tolerance occurs concomitant with peripheral (splenic) appearance of these RDCs. Thus, persistence of alloreactive cells in the setting of true tolerance suggests that regulation may be critically important to development of tolerance.
Based on data presented in this manuscript, we thus propose a model of evolution of immune regulatory mechanisms associated with spontaneous renal allograft acceptance (Fig. 10). In this model, donor-reactive immune responses are regulated by TGF-β and Tregs during early stages of renal allograft acceptance. TGF-β is known to down-regulate immune responses, and its presence in both grafts and peripherally may suppress early donor-reactive immune responses preventing graft destruction. This transient mechanism appears operative between 30 and 120 days posttransplant, and may be responsible for induction of allograft acceptance. Tregs are capable of inducing tryptophan catabolism in APCs (44, 45, 57), and intragraft Tregs may thus facilitate development of RDCs and expression of the immunoregulatory molecule IDO. We posit that these different mechanisms are not mutually exclusive, and might in fact be complementary, as illustrated in Fig. 10.
In summary, metastable immunologic regulation involving TGF-β and Tregs might provide time and conditions that allow or promote development of immunoregulatory DCs. Presence of RDCs in both grafts and spleens of spontaneous allograft acceptors is extremely exciting, particularly given their apparent ability to modulate alloresponses. Although we feel that this evidence is compelling, we recognize that we have yet to directly prove that Tregs or RDCs are responsible for graft acceptance or tolerance in this model, and these studies are ongoing.
We thank Kelly Nye, Misty Bear, Christina Sass, Bree Heestand, Nicholas DiPaola, and Jacob Iselin for their technical assistance with these studies.
The authors have no financial conflict of interest.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
This work was supported by National Institutes of Health RO1 AI053094 (to C.H.C.) and a grant from the Roche Organ Transplant Foundation (to R.B.C.).
Abbreviations used in this paper: DTH, delayed-type hypersensitivity; Treg, regulatory T cell; GIC, graft-infiltrating cell; DC, dendritic cell; RDC, regulatory DC; IVC, inferior vena cava; 1-MT, 1-methyl-d-tryptophan.