Tolerance Induced by Apoptotic Antigen-Coupled Leukocytes Is Induced by PD-L1⁺ and IL-10–Producing Splenic Macrophages and Maintained by T Regulatory Cells

Autoimmune diseases, including multiple sclerosis (MS) and type 1 diabetes, rank third as a major cause of morbidity and mortality in humans. Ag-specific tolerance remains the most highly desired, yet elusive, technique for treating patients suffering from T cell-mediated autoimmune diseases. Strategies for inducing peripheral T cell tolerance, including administration of soluble peptide, altered peptide ligands, anti-CD3 Ab, and costimulation blockade (1–3), have been largely unsuccessful. The use of hematopoietic stem cell transplantation or T regulatory cell (Treg) immunotherapy has also been hindered by the inability to obtain sufficient quantities of stem cells and Tregs of sufficient specificity and stability. Another option with significant promise for inducing long-term T cell tolerance has been the i.v. infusion of peptides cross-linked to the surface of splenic leukocytes (Ag-coupled splenocytes [Ag-SP]) using ethylene carbodiimide (ECDI) (4–6). Ag-SP tolerance has been shown to both prevent and treat Th1/17-mediated autoimmune diseases (3, 6, 7) and allograft rejection (5). This promising tolerance therapy is currently the focus of a Phase I/IIA clinical trial investigating the safety and efficacy of myelin peptide-coupled PBLs in human MS.

The precise mechanism(s) that underlies Ag-SP tolerance remains to be defined; however, ECDI-induced apoptosis appears to be critical (6). Apoptosis, or programmed cell death, is an event that occurs on a regular basis in the body. Unlike necrosis, which triggers proinflammatory immune responses, apoptosis is usually associated with little to no proinflammatory immune activation (8, 9). Nonetheless, apoptotic cells are not invisible to the immune system. The cells responsible for their removal, predominantly macrophages, are capable of targeting apoptotic cells through a number of pathways, including recognizing proteins expressed by the dying cells themselves as well as detecting serum opsonins that coat apoptotic cells (8, 9). A large proportion of apoptotic debris is removed by marginal zone (MZ) macrophages expressing scavenger receptors, including lectin-like oxidized low-density lipoprotein receptor-1 (LOX) receptor and class B scavenger receptor (SRB) (8, 9). Within the germinal center, CD68⁺ tingible-body macrophages are important regulators of apoptotic B cell removal (10). Apoptotic debris can trigger IL-10 production (11–13). Notably, it was recently shown that apoptotic cell infusion can induce regulatory B cells, which, through their production of IL-10, can reduce the severity of collagen-induced arthritis (11). Overall, the data support the importance of apoptotic cell processing in the maintenance of peripheral self-tolerance. Indeed, dysfunction in these clearance pathways is hypothesized to be a major cause of Ab-mediated autoimmune diseases such as SLE (10, 14).

Numerous immune interactions, including CTLA-4–dependent T cell anergy as well as PD-L1–mediated T cell-negative costimulation, have been shown to play a role in long-term Ag-SP tolerance induction (7, 15, 16). The immediate responses to infusion of Ag-SP that ultimately lead to long-term T cell unresponsiveness have not been examined. We have previously shown that ECDI-induced apoptosis is a critical factor in Ag-SP tolerance and that indirect mechanisms involving host APC processing of Ag-SP are also required, as indicated by the ability of peptide-coupled allogeneic and MHC-deficient donor splenocytes to efficiently induce tolerance (6). Focusing on the events that occur within the first 72 h after i.v. Ag-SP tolerization, we found that Ag-SP rapidly localize to the splenic MZ and trigger IL-10 production by F4/80⁺ MZ macrophages. IL-10 production was critical for tolerance induction, and appears to regulate PD-L1 expression in possibly an autocrine fashion. PD-L1 blockade at the time of Ag-SP infusion abrogated tolerance induction, further highlighting the importance of PD-L1 in the regulation. Lastly, Ag-SP infusion was found to induce Tregs that appear to be dispensable for early tolerance induction, but essential for maintenance of long-term tolerance.

All mice were housed under specific pathogen-free conditions in the Northwestern University Center for Comparative Medicine and maintained according to protocols approved by the Northwestern University Institutional Animal Care and Use Committee. Female SJL/J mice, 5–7 wk old, were purchased from Harlan Laboratories. Five- to 7-wk-old wild-type and IL-10–deficient C57BL/6J mice were purchased from The Jackson Laboratory.

Synthetic peptides myelin oligodendrocyte glycoprotein (MOG)_35–55 (MEVGWYRSPFSRVVHLYRNGK), proteolipid protein (PLP)_139–151 (HSLGKWLGHPDKF), and OVA_323–339 (ISQAVHAAHAEINEAGR) were purchased from Genemed Synthesis. PLP_178–191 (NTWTTCQSIAFPSK) was purchased from Peptides International.

Mice were primed with an emulsion containing 1 mg/ml peptide and CFA containing 2 mg/ml Mycobacterium tuberculosis H37Ra (Difco). A 100 μl vol of emulsion was injected s.c. among three sites on the flank of each mouse.

Peptide-induced and adoptive transfer experimental autoimmune encephalomyelitis (EAE) was induced in both SJL and C57BL/6 mice, as previously reported (17, 18). Individual animals were observed daily, and clinical scores were assessed in a blinded fashion on a 0–5 scale, as follows: 0, no abnormality; 1, limp tail or hind limb weakness; 2, limp tail and hind limb weakness; 3, hind limb paralysis; 4, hind limb paralysis and forelimb weakness; and 5, moribund. The data are reported as the mean daily clinical score. Paralyzed animals were afforded easier access to food and water.

Tolerance was induced by i.v. injection of chemically treated Ag-SP, as described previously (5, 6). Briefly, spleens were removed from naive female mice, and the RBCs were lysed. The splenocytes were incubated with ECDI (150 mg/3.2 × 10⁸ cells; Calbiochem) and peptide (1 mg/ml) on ice, shaking for 1 h. The coupled cells were washed three times and filtered through a 70-μM cell strainer to remove cell clumps. The Ag-SP were resuspended at 250 × 10⁶ cells/ml in PBS. Each mouse received 50 × 10⁶ Ag-SP in 200 μl PBS given by i.v. injection at the indicated times before disease induction. This dosage represents delivery of a total of 15–20 μg cell-bound peptide per mouse.

Spleen cell membranes were stained with PKH76 (green) or with PKH26 (red) (Sigma-Aldrich) dye, according to the manufacturer’s instructions, before ECDI fixation. CFSE loading was performed, as described in the manufacturer’s instructions (Cayman Chemical). Cells were then treated as described for ECDI fixation.

Delayed-type hypersensitivity (DTH) was performed via ear challenge with 10 μg peptide, as previously reported (6). For proliferation assays, draining lymph nodes (axillary, brachial, and inguinal) were harvested from naive mice or primed mice at indicated days following disease induction, counted, and cultured in 96-well microtiter plates at a density of 5 × 10⁵ cells/well in a total volume of 200 μl HL-1 medium (BioWhittaker; 1% penicillin/streptavidin and 1% glutamine). Cells were cultured at 37°C with medium alone or with different concentrations of peptide Ag for 72 h. During the last 24 h, cultures were pulsed with 1 μCi/well [³H]TdR, and uptake was detected using a Topcount microplate scintillation counter; results are expressed as mean of triplicate cultures.

Abs used for immunohistochemistry on spleen sections included rabbit polyclonal anti–LOX-1 (Abcam), rabbit polyclonal SRBI (Abcam) and SRBII (Abnova), hamster anti-mouse CD11c (Biolegend), rat anti-mouse F4/80 (Biolegend), or rat anti-mouse IL-10 (BD Pharmigen). Polyclonal anti-rabbit, hamster IgG, rat IgG2a, or rat IgG1 Abs were used, respectively, as controls (Vector Laboratories, Biolegend, BD Pharmingen). CNS tissues were stained with biotin-conjugated Abs anti-mouse CD4 (H129.19) and anti-mouse F4/80 (Caltag Laboratories). For CNS histology, mice were anesthetized and perfused with PBS. Spinal cords were removed by dissection, and 2- to 3-mm lower lumbar spinal cord (approximately L2–L3) blocks were immediately frozen in OCT (Miles Laboratories) in liquid nitrogen. Spleens were removed from mice infused with PKH26-, PKH76-, or CFSE-labeled Ag-SP and fixed in paraformaldehyde for 30 min to 3 h at 4°C in the dark. Spleens were then frozen in OCT. The blocks were stored at −80°C in plastic bags to prevent dehydration. Six-micrometer–thick cross-sections were cut on a Reichert-Jung Cryocut CM1850 cryotome (Leica) mounted on Superfrost Plus electrostatically charged slides (Fisher), air dried, and stored at −80°C. Slides were stained using the Tyramide Signal Amplification Direct kit (NEN), according to the manufacturer’s instructions. Nonspecific staining was blocked using either anti-CD16/CD32 (FcIII/IIR, 2.4G2; BD Pharmingen) or 10% horse serum, as well as avidin/biotin blocking kit (Vector Laboratories), in addition to the blocking reagent provided by the Tyramide Signal Amplification kit (NEN). Sections were then stained with primary and secondary Ab mixtures. Sections were coverslipped with Vectashield mounting medium with DAPI (Vector Laboratories). Slides were examined and images were acquired using a Lica DM5000B fluorescent microscope and Advanced SPOT software. At least eight serial sections from each sample per group were analyzed at original magnification ×20, ×40, and ×100.

IL-10 ELISA was performed using a ready-set-go IL-10 ELISA kit (eBioscience). As previously described (19), spleens from individual mice were snap frozen, defrosted, and homogenized with a hand-held homogenizer. The resulting homogenate was ultracentrifuged before IL-10 quantitation. IL-10 neutralization was performed through i.p. injection of 100 μg/mouse anti–IL-10 Ab (eBioscience), as described for each experiment. CD20 depletion was performed using 250 μg/mouse anti-CD20 Ab (clone 5D2 gifted by Genentech). Treg inactivation was performed via injection of 500 μg/mouse anti-CD25 (clone 7D4), as described previously (20).

Tregs were isolated, as previously described (21).

Cells were isolated from the spleen, as previously described (17). Briefly, FcR blocking with CD16/32 was performed. Cells were then stained with either a mixture of Abs containing CD4 (BD Biosciences), CD11c (BD Biosciences), CD8 (BD Biosciences), B220 (BD Biosciences), F4/80 (Biolegend), and PD-L1 (BioXcell), or respective isotype controls. Samples were run on a FACSCanto flow cytometer with FACS DIVA software (BD Biosciences). PD-L1 expression was determined based on mean fluorescent intensity relative to isotype controls.

Mice were treated i.p. with 500 μg anti–PD-L1 (clone 10F.9G2) or with control rat IgG2b on day −7, and additionally with 250 μg on days −5, −3, −1, and +1 relative to immunization with PLP_139–151/CFA. Anti–PD-L1 and isotype control rat IgG2b Abs were purchased from BioXcell Fermentation and Purification Services (West Lebanon, NH).

Comparisons of DTH responses, mean clinical disease scores, proliferation, cytokine responses, or mean fluorescence intensities between any two groups of mice were analyzed by a standard two-tailed t test or one-way ANOVA depending on the precise comparisons made. The p values <0.05 were considered significant.

Intravenous administration of peptides cross-linked to the surface of syngeneic splenic leukocytes (Ag-SP) using ECDI safely and efficiently induces Ag-specific immune tolerance; is effective in prevention and treatment of Th1/Th17-mediated autoimmune diseases (3, 6, 15, 22, 23), Th2-mediated asthma, and food allergy models (C. Smarr and S.D. Miller, submitted for publication); and promotes potent alloantigen-specific transplant tolerance in the absence of immunosuppressive drugs (5, 24). A Phase I/IIA clinical trial is currently underway to examine the safety and efficacy of Ag-coupled peripheral blood leukocyte tolerance to a mixture of myelin peptides in patients with new-onset MS. As such, further understanding of the mechanism(s) underlying Ag-SP tolerance induction and the boundaries of this therapy is of critical importance. We first investigated the importance of the route of Ag-SP administration (Fig. 1A). Using PLP_139–151-coupled splenocytes (PLP_139–151-SP) to induce tolerance in PLP_139–151/CFA-induced relapsing EAE, we confirmed that i.v. administration prevented disease induction, but that i.p. administration was ineffective, as mice that received PLP_139–151-SP i.p. developed EAE clinical scores that were similar to control mice treated i.v. with Ag-SP coupled with an irrelevant OVA_323–339 peptide (sham tolerized). In contrast, s.c. administration of PLP_139–151-SP acted synergistically with immunization, in that treated mice displayed significantly higher disease scores than sham-tolerized controls. We hypothesized that the importance of i.v. administration of Ag-SP for tolerance induction was related to the requirement for i.v. delivery of Ag to organs such as the spleen and liver, which have been associated with tolerance induction (25, 26). We i.v. infused Ag-SP labeled with the green fluorescence membrane labeling agent PKH76. As shown in Fig. 1B, Ag-SP accumulated in the MZ of the spleen. Because the spleen is a major site for the removal of circulating senescing erythrocytes and apoptotic hematopoietic cells (25, 27), we examined the effects of Ag-SP administration on expression of certain scavenger receptors known to play a role in the removal of apoptotic cellular debris. Although scavenger receptors LOX-1, SRBI, and CD68 were not affected by the accumulation of Ag-SP in the spleen at the time points examined, SRBII was upregulated within 3 h after Ag-SP infusion (Fig. 1B–H).

To determine the mechanistic role of the spleen in Ag-SP tolerance, we asked whether tolerance could be induced in splenectomized mice. Splenectomized SJL/J mice responded via DTH (Fig. 1I) and proliferation (Fig. 1J) to PLP_139–151/CFA immunization similarly to sham-splenectomized control mice; however, splenectomized animals were resistant to tolerance induction with PLP_139–151-SP, as measured by both assays (Fig. 1I, 1J). These data show that i.v. administration is critical in Ag-SP tolerance induction, most likely due to the direct delivery of apoptotic Ag-SP to immature tolerogenic APCs in the splenic MZ.

To further elucidate the precise environment of Ag-SP localization within the spleen, we examined the temporal uptake and destruction of PKH-76–labeled OVA_332–339-coupled splenocytes (OVA_323–339-SP) (Fig. 2A–C). Within 60 min of infusion, PKH-76–labeled OVA_323–339-SP were found throughout the spleen, especially within the MZ (data not shown). By 3 h postinfusion, the PKH76 staining appeared to be punctate and fragmented, indicating that the Ag-SP had lost cell membrane integrity (Fig. 2B). Using PKH-76 as a marker of membrane debris removal, we found that the Ag-SP were undetectable by 18 h postinfusion (Fig. 2C). This was further supported by experiments using CFSE-labeled Ag-SP (Fig. 2D, 2E). At 30 min postinfusion, numerous CFSE–labeled OVA_323–339-SP can be seen throughout the spleen (Fig. 2D); however, by 3 h postinfusion, no evidence of CFSE-positive cells remains (Fig. 2E). Because CFSE is a cytoplasmic dye, these data suggest that within 3 h, all infused Ag-SP cells lose their plasma membrane integrity, resulting in CFSE diffusion into the extracellular matrix. In contrast, PKH-76 is a plasma membrane-bound dye, which will not dilute or leak during the apoptotic process. Therefore, these results collectively indicate that, upon i.v. infusion, the Ag-SP cells rapidly become unstable, lose their membrane integrity within 3 h of infusion, as supported by our earlier finding that ECDI-fixed cells become rapidly apoptotic (6), and are completely removed by phagocytosis within 18 h posttransfer.

The rapid clearance of i.v. administered Ag-SP from the spleen (Fig. 2A–E) suggested that the framework for tolerance induction is initiated very early after Ag-SP infusion. Because of the importance of IL-10 in immune regulation (28–31), we investigated the level of IL-10 present in whole-spleen homogenates in response to Ag-SP infusion (Fig. 2F). Examination of IL-10 protein revealed that within 10 min postinfusion of OVA_323–339-SP, IL-10 protein levels increased dramatically. Furthermore, these IL-10 levels remained significantly above the baseline level over the 3 d of testing. To determine the functional role of IL-10 secretion in Ag-SP tolerance induction, we first attempted to tolerize IL-10–deficient animals (IL-10gko) with ECDI-coupled splenocytes coupled with OVA_323–339-SP. Using DTH as an vivo measure of T cell tolerance induction in OVA_323–339/CFA-immunized mice, we found that whereas control mice were successfully tolerized to OVA_323–339, showing little to no ear inflammation, IL-10gko mice were not tolerized (Fig. 2G). Importantly, we found that donor splenocytes from both wild-type (data not shown) and IL-10gko (Fig. 2G) animals were similarly capable of inducing tolerance in wild-type animals, indicating the source of IL-10 was the recipient. These data are a strong indication of the critical nature of IL-10 for the induction of Ag-SP tolerance. However, IL-10gko mice are known to have altered immune regulation, commonly developing autoimmune conditions, including colitis (32). Therefore, we administered 100 μg neutralizing IL-10 Ab 30 min prior and 18 h post-OVA_323–339-SP infusion in wild-type B6 mice. Whereas immunized mice treated with isotype control Ab displayed a characteristic reduction of DTH responses indicative of tolerance induction, mice receiving anti–IL-10 exhibited ear swelling similar to mice tolerized with the irrelevant MOG_35–55-coupled splenocytes (MOG_35–55-SP) peptide (Fig. 2H). Collectively, these results confirm that the environment supporting Ag-SP tolerance induction is formed early and is critically dependent on IL-10 production.

Previously, it has been reported that the infusion of apoptotic cells with CFA stimulates IL-10–producing regulatory B cells, which can prevent CD4⁺ T cell activation (11). We investigated the importance of both B cells in Ag-SP tolerance induction. In contrast to this previous study, we found that mice devoid of B cells (μMT mice) can still be tolerized with Ag-SP. Specifically, treatment with MOG_35–55-SP was equally capable of preventing MOG_35–35/CFA-induced EAE in wild-type (Fig. 3A) and μMT mice (Fig. 3B), and tolerance was similarly reflected in MOG_35–55-specific DTH responses (Fig. 3C). In addition, tolerance could be induced in mice depleted of B cells with anti-CD20 (Fig. 3E). These data discount the importance of B cell-derived IL-10 production in Ag-SP tolerance.

IL-10–producing CD4⁺CD25⁺Foxp3⁺ Tregs have been implicated in immune regulation and tolerance induction in numerous models of inflammation and tolerance (32, 33). The importance of IL-10 in Ag-SP tolerance suggests that Treg may also play a role in the induction of Ag-SP tolerance. To address the role of Tregs, we initially asked whether tolerance could be actively transferred from mice treated with Ag-SP. On day −7, SJL mice were tolerized with 5 × 10⁷ syngeneic splenocytes coupled with either PLP_139–151 or OVA_323–339. Five days post–Ag-SP therapy, bulk splenocytes and purified CD4⁺ cells were isolated and transferred i.v. into naive SJL mice, which were then immunized 2 d later with either PLP_139–151/CFA or PLP_178–191/CFA, and clinical disease was monitored for 24 d. Transfer of bulk splenocytes as well as purified CD4⁺ T cells significantly suppressed clinical EAE compared with animals tolerized to the irrelevant OVA_323–339 peptide (Fig. 4A). Tolerance transfer was Ag specific, as recipients of T cells from PLP_139–151-SP–treated mice failed to suppress EAE induced by immunization with PLP_178–191 (Fig. 4B). Ag-specific regulation was supported by reductions in CNS inflammation, observed by immunofluorescent staining for CD4⁺ T cells and F4/80 (microglia/macrophages) on lumbar spinal cord sections (Fig. 4C,i–vi), as well as a lack of development of PLP_139–151 proliferative responses (Fig. 4D) in the animals receiving CD4⁺ splenocytes from PLP_139–151-SP–tolerized animals. These data support an important role for CD4⁺ T cells in disease regulation. To more specifically examine a potential role for Tregs, we purified CD25⁺ and CD25⁻ CD4⁺ splenic T cell populations 5 d post–PLP_139–151-SP or OVA_323–339-SP infusion, transferring 5 × 10⁶ of these cells independently into naive SJL mice, which were then immunized with PLP_139–151/CFA and monitored for disease. CD4⁺CD25⁺, but not CD4⁺CD25⁻ cells, transferred from PLP_139–151-SP–tolerized animals, induced significant protection from EAE (Fig. 4E). Overall, these data indicate that CD4⁺CD25⁺ Tregs are a component of tolerance induced by Ag-SP treatment. However, because we observe rapid IL-10 production almost immediately after Ag-SP infusion, and neutralization of IL-10 at the time of Ag-SP is capable of preventing complete tolerance induction (Fig. 2G, 2H), we next addressed the role of Tregs precisely at the time of tolerance induction. Using anti-CD25 Ab to deplete/inactivate Tregs (20), we found that the functional inactivation of Tregs had no measurable effect on tolerance induction, with anti-CD25–treated and isotype control-treated Ag-SP–tolerized animals both exhibiting significantly reduced clinical disease (Fig. 4F).

The data suggest that Tregs capable of down-regulating clinical disease are induced by Ag-SP treatment, but that there is a separate nonoverlapping tolerance mechanism induced. We hypothesized that whereas Treg may not be critical for tolerance induction, they may play a role in the long-term maintenance of Ag-SP tolerance. To investigate this possibility, we treated a large cohort of SJL/J mice with either control Ig or anti-CD25 Ab at days −4 and −2 relative to PLP_139–151-SP or control OVA_323–339-SP infusion (Fig. 5A). Separate groups of mice were then immunized with PLP_139–151/CFA on days 14, 35, and 63 post–Ag-SP treatment. Similarly to data shown above (Fig. 4F), we found that functional inactivation of Tregs had no effect on tolerance induction in animals immunized on either day 14 (Fig. 5B) or day 35 (Fig. 5C, 5E) posttolerance induction, as both control Ig and anti-CD25–treated mice tolerized with PLP_139–151-SP displayed significantly lower clinical disease and peptide-specific DTH. However, only control Ig-treated, not anti–CD25-treated, mice immunized on day 63 post–Ag-SP treatment were protected from disease induction and had significantly downregulated DTH responses (Fig. 5D, 5F). Overall, the data support two important conclusions. First, Tregs are not required for tolerance induction and are unlikely to be a significant source of the early IL-10 induced by Ag-SP injection. Second, and more importantly, Tregs appear to play a major role in long-term tolerance maintenance for protection from relapsing EAE.

We next investigated the APC subsets in the spleen involved in tolerance induction. Using PKH26-labeled Ag-SP (red), we examined the association of Ag-SP with dendritic cells (DCs; CD11c⁺; Fig. 6A–C) or macrophages (F4/80⁺; Fig. 6D, 6E) at 8 h post–Ag-SP infusion. Surprisingly, little PKH26-CD11c colocalization was observed in animals that received either nonfixed splenocytes (Fig. 6B) or ECDI-fixed splenocytes (Fig. 6C). However, fixed splenocytes colocalize at a much higher frequency with F4/80-expressing cells (Fig. 6F), especially in the MZ.

In addition to the colocalization of Ag-SP with F4/80⁺ macrophages, we also determined the expression profile of IL-10 in serial sections in the same experiments. Although little IL-10 was found in the control nontreated or animals receiving non-ECDI–fixed PKH-26–labeled splenocytes, IL-10 was strongly expressed in Ag-SP recipients. Whereas a small population of cells was found to express IL-10 in the germinal centers of Ag-SP–treated animals (data not shown), the most striking number of cells producing IL-10 were F4/80⁺ and in close proximity to PKH26-expressing Ag-SP (Fig. 6I).

Attempts to isolate F4/80 macrophages from the spleen failed to produce sufficient yields for functional assessment. As such, IL-10 responses of a macrophage cell line (J774), as well as primary thioglycolate-stimulated and resting peritoneal macrophages, to coculture with Ag-SP were evaluated. J774 cells phagocytized PKH-26–labeled (red) Ag-SP, but this uptake failed to upregulate production of IL-10 (Fig. 6K–M). In an attempt to move to a more physiologically relevant system, we found that thioglycolate-elicited peritoneal macrophages were also capable of ingesting Ag-SP, with a significant amount of PKH-26–labeled membrane localized inside the macrophages, but again, Ag-SP uptake failed to stimulate IL-10 production (Fig. 6N–P). The thioglycolate-stimulated macrophages were rounded up, with multiple nuclei, and exhibited a highly inflammatory phenotype. We have previously shown that LPS injection is capable of preventing Ag-SP tolerance in vivo (7). Because the J774 macrophage line and the thioglycolate-elicited peritoneal macrophages are of a type 1 phenotype, characterized by the production of proinflammatory cytokines, we hypothesized that the normal response to Ag-SP is overcome by the background activation state of these cells. We thus tested nonelicited peritoneal macrophages harvested from multiple mice. We found that these cells exhibited the greatest capacity to ingest Ag-SP as determined by the internalization of PKH26-labeled membrane material, and, importantly, that resting macrophages produced significantly upregulated levels of IL-10 upon ingestion of OVA_323–339-SP (Fig. 6Q–S). The production of IL-10 by these macrophages is consistent with our in vivo immunohistological findings (Fig. 6I) as well as observations by other investigators examining the response of macrophages to tolerogenic stimuli (33, 34). In conclusion, these results support a scenario in which resting MZ macrophages respond rapidly to Ag-SP and are likely to be the major source of the early IL-10 produced in response to i.v. Ag-SP infusion and to be critical for the induction of tolerance.

To this point, our data indicate that the long-term Ag-specific tolerance triggered by the infusion of Ag-SP is the result of more than one mechanism, with Tregs primarily required for tolerance maintenance. Numerous previous studies have implicated CD8α⁺ DCs, as well as different macrophage subpopulations, in tolerance induction secondary to uptake of apoptotic cell debris (34–39). We thus asked whether there was a role for one or both of these APC subsets, in Ag-SP uptake, as a source for IL-10 and/or as major drivers of Ag-SP tolerance.

Further examination of the ratio of APC populations in the spleen 3 h after Ag-SP infusion revealed that the ratio of the major DC subsets, including CD4⁺ DCs (CD4⁺CD11c⁺CD8α⁻), CD8α⁺ DCs (CD8α⁺CD11c⁺CD4⁻), and plasmacytoid DCs (CD8⁻CD4⁻CD11c^int), remains unchanged, and was further unaffected in mice treated with anti–IL-10 (Fig. 7A, 7B). In contrast, F4/80-expressing macrophages (F4/80⁺CD11C⁻CD4⁻CD8⁻) significantly increased in relative percentage within 3 h post–Ag-SP infusion (Fig. 7B).

To further examine the potential uptake of Ag-SP by macrophages, OVA₃₂₃-coupled B6 CD45.1 congenic splenocytes were labeled with PKH-26 and injected into CD45.1 mice, which were sacrificed 3 h after infusion. Whereas no PKH-26 colocalization was observed with recipient CD45.1 in untreated controls (Fig. 7C), 3 h after Ag-SP infusion there was a distinct population of PKH-26⁺/CD45.1⁺ cells as well as possibly intact PKH26⁺CD45.1⁻ donor cells (gate R1; Fig. 7D, 7E). The PKH-26⁺CD45.1⁺ cells were 85% CD11b⁺F4/80⁺/CD11c^−/low (gate R3; Fig. 7F), with only 11.6% expressing CD11c^high (Fig. 7F). Cells from gate R3 were 77.5% F4/80^int and 11.3% F4/80^high. The majority of the cells in gate R3 were CD11c^int, which is consistent with the phenotype of splenic MZ macrophages (Fig. 7F, 7G). Significantly, PKH-26–colocalized macrophages also expressed high levels of PD-L1 (Fig. 7H). These results show that shortly after i.v. infusion of Ag-SP, macrophages not only change in their relative percentage in the spleen, but they are also the major population taking up the apoptotic Ag-SP debris and expressing PD-L1.

Interestingly, administration of IL-10–neutralizing Ab 30 min prior to Ag-SP infusion completely abrogated the increase in F4/80-expressing macrophages (Fig. 7B), suggesting that IL-10 may play a role in the overall kinetics of cellular proliferation/migration within the splenic microenvironment. PD1/PD-L1 and IL-10 have been reported to reciprocally regulate each other (40). Therefore, we examined PD-L1 expression on APC populations after Ag-SP infusion. We found that within 3 h, macrophages displayed the greatest increase in PD-L1 expression (Fig. 7I). This was not reflected by increased expression of other molecules such as MHC-II and CD80/86, which remained unchanged on all examined APC populations (data not shown). Interestingly, CD8α⁺CD11c⁺CD4⁻ DCs also upregulated PD-L1 within the time frame examined; however, unlike the macrophages, neutralization of IL-10 did not significantly alleviate PD-L1 expression on DCs.

Finally, to address the functional role of PD-L1 in tolerance induction, we infused anti–PD-L1 Ab at the time of PLP_131–151-SP infusion and subsequently primed the mice with PLP_131–151/CFA. Animals became moribund within 14 d postimmunization (data not shown). It has been described previously that administration of anti–PD-L1 Ab early in EAE induction can significantly exacerbate disease (18, 41, 42). Because of the severe disease phenotype observed, we examined tolerance induction in anti–PD-L1–treated animals by DTH at 7 d postpriming. PD-L1 inhibition at the time of Ag-SP infusion completely ablated the induction of tolerance (Fig. 7J). Based on the disease observations, it is tempting to speculate that Ag-SP infusion in the presence of anti–PD-L1 may actually prime mice such that later immunization with PLP_139–151/CFA results in lethal EAE. We are currently investigating this possibility. Collectively, the data suggest a critical role for IL-10 and build on our and others previous findings regarding the importance of PD-L1 in the induction of Ag-SP tolerance (5, 15, 16).

Previously, we and others have shown that Ag-SP infusion is a safe and highly efficient method for the induction of T cell tolerance in models of Th1/Th17 autoimmunity and Th2-mediated allergy. The precise mechanisms driving Ag-SP–induced tolerance remain to be completely defined; however, it is clear from previous observations that direct Ag-SP–T cell interaction is responsible to some degree for tolerance induction, particularly in in vitro anergy models. This is thought to be the result of TCR stimulation in the absence of secondary costimulatory signals (4). However, this mechanism fails to explain the fact that both Ag-coupled, ECDI-fixed allogeneic and MHC class I/II-deficient donor cells also effectively induce T tolerance in vivo (6). In this study, we show that Ag-SP tolerance recapitulates how tolerance is normally maintained in the hematopoietic compartment and is the result of two parallel, but convergent pathways. First, Ag-SP trigger rapid and sustained IL-10 release from macrophages that appears to modulate PD-L1 expression on these cells, which is critical for Ag-SP tolerance induction (5, 7, 15). Second, we show that although not important for Ag-SP tolerance induction, Tregs play a critical role in the long-term maintenance of unresponsiveness.

Tolerance induction with ECDI-coupled splenocytes was found to critically depend on the route of administration. Intravenous infusion of Ag-SP resulted in complete protection from PLP_131–159-induced EAE, but neither i.p. nor s.c. infusion of Ag-SP prevented disease development. In fact, s.c. infusion of PLP_139–151-SP acted synergistically with PLP_139–151/CFA immunization, further exacerbating disease instead of tolerizing animals. Relationships between route of administration and tolerance induction have also been described for certain peptides used in high zone tolerance induction (43). For example, myelin basic protein in human patients induced tolerance only when given via the i.v. route, with intrathecal and s.c. administration having either short-lived or no effect on disease (43). The data suggest that the context of Ag delivery, including the cytokine milieu and local cell populations, is important in Ag-SP tolerance induction. Currently, it is thought that Ag-SP injection via i.p. or s.c. injection, especially the latter, is uptaken by and/or triggers activation of resident leukocyte populations, such as Langerhans cells, dermal DCs, and/or γδ T cells. In turn, the production of proinflammatory cytokines is likely to negate the upregulation of negative costimulatory molecules, such as CTLA-4 and PD-L1, which have previously been shown to play important roles in Ag-SP tolerance induction (15, 44). PD-L1, for example, has been shown to play a role with Tregs in long-term allotolerance after infusion of ECDI-fixed allogeneic splenocytes (15, 42). In addition, the processing and quantity of Ag delivered to secondary lymphoid organs are likely to play a role. Intravenous infusion of Ag-SP results in rapid accumulation of Ag-SP apoptotic debris in the spleen. Whereas both i.p. and s.c. injection of Ag-SP are likely to result in some Ag-SP–associated Ag drainage to the spleen, it is likely that most of the Ag-SP will ultimately be sequestered in local draining lymph nodes. Furthermore, the Ag is likely to be transported by APC from the skin and peritoneal cavity, thereby circumventing the involvement of macrophages in the secondary lymphoid organs.

Although apoptotic cells like Ag-SP are thought to be nonimmunogenic, they are by no means invisible to the immune system. The work of Barker et al. (9) showed that stimulation of macrophages with necrotic, but not apoptotic, cells resulted in triggering of recall responses in OVA-specific T cells. Within the spleen, which is critical for Ag-SP tolerance, there are numerous mechanisms through which apoptotic debris is recognized and rapidly removed (9, 45). It appears that these functions are mostly performed by macrophages within the spleen, which encompass a variety of subtypes depending on their anatomical localization and receptor expression. Within the germinal centers exist the tingible body macrophages, identified by the expression of the scavenger receptor CD68 (10). These macrophages have been shown to play a role in B cell tolerance through their uptake of MFGE8-coated apoptotic debris (10). Whereas B cell tolerance is compromised in MFGE8-deficient animals, a role for this pathway in peripheral T cell tolerance has not been described. However, it appears unlikely that this pathway plays a role in Ag-SP–mediated tolerance, as few coupled cells are found within germinal centers or localized with CD68-expressing cells. Another scavenger receptor that has been shown to aid in uptake of apoptotic debris is the oxidized low-density lipoprotein receptor-1 (LOX-1). Although this receptor has been shown to mediate DC uptake and response to apoptotic debris (46), we found little colocalization of LOX-1 or CD11c with Ag-SP at the early time points investigated. Similarly, the SRBI, which is a phosphatidylserine-recognizing receptor known to play a role in the uptake of apoptotic sperm (47), was also uncommonly found to colocalize with Ag-SP. In contrast, we found that SRBII, an isoform of SRBI (48, 49), colocalized with Ag-SP and was also rapidly upregulated in response to the i.v. infusion of Ag-SP. These findings suggest that there may be a role for SRBII in Ag-SP uptake and T cell tolerance induction. The lack of colocalization with SRBI and LOX-1 requires further investigation. However, it is becoming clear that scavenger receptors may have multiple ligands and functions. SRBI, for example, is known to be expressed on testicular sertoli cells (47), and has also been shown to prevent anemia through its interaction with erythropoietin and high-density lipoprotein (50). Finally, there are numerous other scavenger receptors, for example, MARCO (51), that may also play a role in Ag-SP tolerance induction and are currently under investigation.

In 1997, Voll et al. (52) showed that feeding apoptotic cells to peripheral blood-derived macrophages triggered the production of IL-10. IL-10 was originally defined as a T cell cytokine, predominately secreted from Th2 T cells (13). However, it is now known that many cells, including macrophages and regulatory B cells, as well as T cells, can produce IL-10 (13). IL-10 plays an important immune regulatory role, preventing inflammatory immune responses and the development of autoimmunity. The infusion of Ag-SP results in rapid and sustained production of IL-10. Importantly, this IL-10 response is critical for the induction of tolerance, as neither IL-10–deficient mice nor mice treated with anti–IL-10 can be tolerized with ECDI-fixed splenocytes. These observations raise the question of whether IL-10 plays a direct role in tolerance induction or whether the IL-10 detected is produced subsequent to tolerance induction, by activated Tregs and B regulatory cells. Indeed, B cell-dependent IL-10 production, in response to apoptotic cells and CFA immunization, has been shown to prevent CD4⁺ T cell activation. However, we found that animals lacking functional B cells via either genetic deletion (μMT) or following anti-CD20 treatment could still be tolerized to MOG_35–55 EAE by MOG_35–55-SP. Our data, which contrast with a previous study in which induction of tolerance using apoptotic thymocytes was B cell dependent (11), could be explained by the different methods of induction of apoptosis. In the former study, apoptosis was induced in syngeneic thymocytes using 2-ME. Furthermore, apoptosis in that study was found only to be 43% at the time of infusion, and it is likely that this mixed apoptotic/nonapoptotic thymocyte population may trigger different responses. Finally, in contrast to specific tolerance induction by Ag-SP, the infusion of apoptotic thymocytes appears to induce a nonspecific anti-inflammatory response (11). From our studies, it appears that B cells are not a critical requirement for Ag-SP tolerance induction, and as such they are not likely to be a major source of IL-10 in response to Ag-SP infusion.

Macrophages have been shown to play a critical role in T cell tolerance induced by Ag administration in the anterior chamber of the eye, in which F4/80 deficiency prevents tolerance induction (53). F4/80⁺ macrophages in this model mediate the expansion of Tregs. In this study, we show that PLP_139–151-coupled cells induced Ag-specific Tregs. These cells were capable of transferring specific unresponsiveness to naive recipients. Previously, CD4⁺CD25⁺Foxp3⁺ IL-10–secreting T cells have been implicated in the regulation of colitis and inhibition of xenogeneic T cell proliferation (11, 32, 33). The ability to transfer tolerance with CD4⁺CD25⁺ T cells strongly suggests that Ag-SP tolerance may in part result from Treg expansion. However, the functional inactivation of Tregs at the time of tolerance induction failed to prevent tolerance induction in mice primed either 14 or 35 d post–Ag-SP infusion. Therefore, although Tregs are activated by Ag-SP, they appear to be dispensable for induction of short-term tolerance. Instead, we found that Tregs play a role in the maintenance of tolerance, as indicated by the failure of mice depleted of Tregs at the time of tolerance induction to regulate disease when primed 63 d after Ag-SP treatment. This finding has significant implications for how tolerance as an immunological phenomenon is viewed. As opposed to one mechanism driving T cell unresponsiveness, tolerance induction after Ag-SP appears to be the result of several parallel mechanisms having specific temporal roles. By day 63 posttolerization, the cell-intrinsic regulatory mechanism appears to have waned, with Tregs mediating long-term tolerance. Previous findings from our laboratory, including the fact that treatment with anti–CTLA-4 can reverse established Ag-SP tolerance, further support this hypothesis (7, 44). Another plausible explanation may include regulation of T cells that have recently emigrated from the thymus. Assuming that Ag-SP infusion and the presence of s.c. Ag (i.e., the peptide/CFA depot) do not interfere with normal deletion mechanisms, these fresh T cells would be expected to respond to Ag. Further investigation into the mechanisms at the effector T cell and Treg level is required to address this. Together these mechanisms may combine to regulate T cells that have recently emigrated from the thymus and may have autoantigen specificity.

IL-10 in this model appears to play a major role in the immediate response to Ag-SP, and therefore, in the short-term tolerance induction phase. Blocking IL-10 with anti–IL-10 Ab within the first 24 h after Ag-SP abrogates tolerance. Because the data suggest a distinct role for Tregs subsequent to the tolerance induction phase, we propose that this IL-10 is secreted by splenic macrophages. Studies have shown that apoptotic debris do not need to be phagocytized by macrophages to trigger IL-10 production, with cell to cell (i.e., apoptotic cell to macrophage) contact capable of triggering p38 MAPK and subsequent IL-10 production (12). This mechanism of macrophage IL-10 production fits with our observations that IL-10 is rapidly produced/secreted prior to the clearance of Ag-SP. We are currently further investigating the potential role of other spleen macrophage subsets to produce IL-10 in response to Ag-SP infusion. Significantly, we show that IL-10 regulates the expression of PD-L1 on splenic F4/80⁺ macrophages. Previously, we have shown that PD-L1 plays a critical role in the induction of tolerance for protection of allogeneic islet cell transplants (5). In addition, PD-L1 has been shown to regulate interactions between DCs and T cells. Blocking IL-10 with neutralizing Ab prevented Ag-SP–mediated PD-L1 upregulation on F4/80⁺ macrophages, but not on CD8⁺ DC, suggesting an autocrine pathway in which macrophages regulate the self expression of PD-L1. Exactly how macrophage-expressed PD-L1 induces tolerance induction is currently under examination. Based on observations by Fife et al. (15), it is likely that PD-L1 on macrophages acts similar to that on DCs, interacting with PD-1, triggering enhanced T cell mobility, and thereby preventing T cell swarming around APCs. Furthermore, PD-L1 is known to deliver negative costimulatory signals to T cells (54–56), and MZ macrophages may also simply activate these pathways directly in autoreactive T cells, accounting for the T cell-intrinsic tolerance pathway we have described. It remains to be determined how long the IL-10–mediated PD-L1 regulation is maintained, what the contributions of other splenic APCs are in driving Ag-SP tolerance induction, and how the APC response induces and/or regulates the Treg-mediated long-term tolerance.

In conclusion, we have shown that Ag-SP tolerance induction recapitulates how tolerance is normally maintained in the hematopoietic compartment, in that MZ macrophages are responsible for the daily safe removal of millions of senescing red cells and leukocytes (25, 27), and is the result of multiple mechanisms that appear to play different roles in tolerance induction and maintenance. Whereas Tregs are dispensable at the induction phase of Ag-SP tolerance, they appear to play an essential role in long-term tolerance maintenance. In contrast, macrophages respond immediately to i.v. infused Ag-SP by rapidly secreting IL-10, which appears to regulate PD-L1 expression on these cells, inducing a second Treg-independent, cell-intrinsic tolerogenic signal. Collectively, to our knowledge, these findings show for the first time that long-term Ag-SP tolerance induction is the result of multiple temporally distinct innate and adaptive immune mechanisms. Further understanding these interactions should provide significant clues into enhancing tolerance induction in the clinical setting for treatment of immune-mediated diseases.

We thank the Northwestern Immunobiology Center Flow Cytometry Core Facility and all of the Miller laboratory members for helpful discussions.

D.G. and S.D.M. receive financial support from Tolera Therapeutics. The other authors have no financial conflicts of interest.