Therapeutic effect of exosomes from indoleamine 2,3-dioxygenase–positive dendritic cells in collagen-induced arthritis and delayed-type hypersensitivity disease models

Authors


Abstract

Objective

We have demonstrated previously that dendritic cells (DCs) modified with immunosuppressive cytokines, and exosomes derived from DCs can suppress the onset of murine collagen-induced arthritis (CIA) and reduce the severity of established arthritis. Indoleamine 2,3-dioxygenase (IDO) is a tryptophan-degrading enzyme that is important for immune regulation and tolerance maintenance. DCs expressing functional IDO can inhibit T cells by depleting them of essential tryptophan and/or by producing toxic metabolites, as well as by generating Treg cells. This study was undertaken to examine the immunosuppressive effects of bone marrow (BM)–derived DCs genetically modified to express IDO, and of exosomes derived from IDO-positive DCs.

Methods

BM-derived DCs were adenovirally transduced with IDO or CTLA-4Ig (an inducer of IDO), and the resulting DCs and exosomes were tested for their immunosuppressive ability in the CIA and delayed-type hypersensitivity (DTH) murine models.

Results

Both DCs and exosomes derived from DCs overexpressing IDO had an antiinflammatory effect in CIA and DTH murine models. The suppressive effects were partially dependent on B7 costimulatory molecules. In addition, gene transfer of CTLA-4Ig to DCs resulted in induction of IDO in the DCs and in exosomes able to reduce inflammation in an IDO-dependent manner.

Conclusion

These results demonstrate that both IDO-expressing DCs and DC-derived exosomes are immunosuppressive and antiinflammatory, and are able to reverse established arthritis. Therefore, exosomes from IDO-positive DCs may represent a novel therapy for rheumatoid arthritis.

Exosomes are small-membrane vesicles, ∼50–100 nm in diameter, that are produced within the multivesicular bodies of the late endocytic compartment of hematopoietic and nonhematopoietic cells. Exosomes are then secreted into the extracellular space by fusion of the limiting membrane of multivesicular bodies with the plasma membrane. Exosomes originating from B cells, T cells, dendritic cells (DCs), and mast cells can confer immunoregulatory signals between cells in either an immunostimulatory or an immunosuppressive manner. Indeed, exosomes derived from antigen-presenting cells (APCs) contain many of the important regulatory molecules needed to carry out this function, such as class I major histocompatibility complex (MHC), class II MHC, CD80 (B7-1), and CD86 (B7-2), as well as various adhesion molecules that may target exosomes to their acceptor cells (1).

Previously, we demonstrated that exosomes derived from immature DCs treated with interleukin-10 (IL-10) produce antiinflammatory exosomes that suppress the onset of murine collagen-induced arthritis (CIA) and reduce the severity of established arthritis (2). Moreover, DCs transduced with an adenoviral vector expressing FasL or IL-4 produce exosomes that suppress inflammation in a murine model of delayed-type hypersensitivity (DTH) and partially reverse established CIA through a class II MHC–dependent, but class I MHC–independent, mechanism (3, 4). Interestingly, the DC-derived exosomes are as or more immunosuppressive than their parental DCs.

Indoleamine 2,3-dioxygenase (IDO) is a tryptophan-degrading enzyme that is also important in host defense and immunosuppression. Only certain subsets of cells appear capable of producing functional IDO, including CD19+ plasmacytoid DCs (5) and CD8α+B220+CD19+ splenic DCs (6). IDO is transcriptionally induced by a variety of inflammatory stimuli, such as interferon-γ (IFNγ), IFNα, CD40L, glucocorticoid-induced tumor necrosis factor receptor (GITR), and tumor necrosis factor α (7). Cellular infection with viruses and microbes can also induce IDO (7). In the cases of CD40L and GITR, downstream signaling appears to involve a noncanonical NF-κB–mediated induction of IDO (8, 9). Interestingly, CTLA-4 on Treg cells or soluble CTLA-4Ig induces functional IDO in DCs by binding to and inducing signaling though B7 molecules (10).

The immunosuppressive ability of IDO was initially described as being important for maternal tolerance of the fetus, since mice treated with the IDO inhibitor 1-methyl-D-tryptophan (1-MT) underwent spontaneous abortion (11, 12). More recently, it has been demonstrated that endogenous IDO is involved in maintaining tolerance in a number of settings (7), including rheumatoid arthritis (RA) (13, 14), cancer (15), transplantation (16), diabetes (16, 17), and experimental autoimmune encephalomyelitis (18, 19). The mechanism by which IDO inhibits the T cell response is currently being investigated. One possible mechanism is that IDO depletes T cells of essential tryptophan, causing activation of GCN2 kinase and rendering the T cells anergic (20). IDO also produces metabolites of tryptophan, collectively termed kynurenines, that regulate T cells through a poorly understood mechanism (21). However, these 2 possible mechanisms are not mutually exclusive, and recent data suggest that a combination of the 2 mechanisms may work together to suppress CD8+ effector T cells and to activate Treg cells (22, 23).

IDO appears to have an immunosuppressive role in arthritis, based on studies showing that inhibition of IDO accelerates CIA (13), and that orally tolerizing mice to collagen induces an IDO-positive suppressor DC population (24). Also, treatment with an agonistic monoclonal antibody to anti–4-1BB, a T cell costimulatory receptor, inhibits CIA through an IDO-dependent pathway (14). Finally, CTLA-4Ig up-regulates IDO in certain populations of cells and has recently been approved for the treatment of RA (25–28). Although tryptophan degradation is enhanced in RA (29), it is not enough to suppress disease (30).

Due to the immunosuppressive activity of IDO-positive DCs, we have examined whether IDO-positive DCs and exosomes derived from IDO-expressing DCs are effective in treating established CIA and blocking inflammation in a footpad DTH model of antigen-specific inflammation. In the present study, we show that both IDO-expressing DCs and DC-derived exosomes are antiinflammatory and therapeutic in both CIA and DTH. The suppressive effects in the DTH model were partially dependent on B7-1 and B7-2 costimulatory molecules, evidenced by the fact that exosomes from B7-1 and B7-2–double-knockout mice were less able to confer the therapeutic effect. Finally, exosomes from CTLA-4Ig–expressing DCs also reduced inflammation in an IDO-dependent manner. These results suggest that exogenous expression of IDO in bone marrow (BM)–derived DCs or induction of endogenous IDO renders them immunosuppressive. Moreover, these findings demonstrate that IDO expression in DCs results in the generation of immunosuppressive exosomes.

MATERIALS AND METHODS

Mice.

Female C57BL/6 (H-2Kb) mice and male DBA/1LacJ (H-2q) mice, all 7–8 weeks of age, were purchased from The Jackson Laboratory (Bar Harbor, ME). B7-knockout mice, double-knockout mice (B6.129S4-Cd80tm1Shr Cd86tm2Shr/J), B7-1 mice (B6.129S4-Cd80tm1Shr/J), and B7-2 mice (B6.129S4-Cd86tm1Shr/J), were also purchased from The Jackson Laboratory. Animals were maintained in a pathogen-free animal facility at the University of Pittsburgh Biotechnology Center.

Vector construction and adenovirus generation.

Region ΔE1,E3 first-generation adenoviruses expressing murine IDO (AdIDO) and CTLA-4Ig (AdCTLA-4Ig) under the regulation of the cytomegalovirus (CMV) promoter were constructed by Cre-Lox recombination, propagated, and titered according to standard protocols as previously described (31). Briefly, the recombinant adenoviruses were generated by homologous recombination in 293 cells expressing Cre recombinase (CRE8 cells), after cotransfection of DNA, an adenovirus 5–derived, E1- and E3-deleted adenoviral backbone (Ψ5), and pAdlox, the adenoviral shuttle vector. The inserted complementary DNA sequences are expressed under the human CMV promoter. The recombinant adenoviruses were purified by CsCl gradient ultracentrifugation, dialyzed in sterile virus storage buffer, divided into aliquots, and stored at −80°C until used. The CRE8 cells were grown and maintained in Dulbecco's modified Eagle's medium (Life Technologies, Gaithersburg, MD) supplemented with 10% fetal bovine serum (FBS).

DC generation.

Murine BM-derived DCs were prepared using an adaptation of the bulk culture method of Son et al (32). Briefly, BM was collected from the tibias and femurs of 6–7–week-old mice. Contaminating erythrocytes were lysed with ACK cell lysing buffer (Mediatech, Herndon, VA). Monocytes were collected from the interface after centrifuging on Nycoprep (Nycomed, Roskilde, Denmark) at 600g for 20 minutes at room temperature. Cells were then cultured for 24 hours in complete media (RPMI 1640 containing 10% FBS), 50 μM 2-mercaptoethanol, 2 mM glutamine, 0.1 mM nonessential amino acids, 100 μg/ml of streptomycin, and 100 IU/ml of penicillin) to remove adherent macrophages. The nonadherent cells were then placed in fresh complete media containing 1,000 units/ml of murine granulocyte–macrophage colony-stimulating factor and murine IL-4 to generate DCs. Cells were cultured for 4 days and harvested for adenovirus transduction. For adenovirus infection, 1 × 106 DCs were mixed with 5 × 107 plaque-forming units of the viruses in a total volume of 1 ml of serum-free media. After a 2-hour incubation with the virus, 10 ml of complete media was added to the cells. For some experiments, the cells were also treated with 1-MT (200 μM; Sigma, St. Louis, MO) or L-tryptophan (245 μM) at this time. After incubation for 24 hours, DCs were washed intensively 3 times and cultured for a further 48 hours. On day 8, culture supernatant was collected for exosome purification and recovery of the adenovirus-transduced DCs. This infection method routinely results in ∼70–80% transfection efficiency using AdeGFP as a control (33). There was no toxic effect of 1-MT or L-tryptophan on the cells as observed by trypan blue exclusion and overall cell count.

Exosome isolation.

Exosomes were isolated using differential centrifugation as previously described (2). Collected culture supernatants were centrifuged at 300g for 10 minutes, 1,200g for 20 minutes, and 10,000g for 30 minutes. The supernatant from the final centrifugation was ultracentrifuged at 100,000g for 1 hour. The exosome pellet was washed in saline, centrifuged again at 100,000g for 1 hour, and resuspended in 120 μl of phosphate buffered saline (PBS) for further studies. The exosomal protein content was quantified using a Bradford micro protein assay (Bio-Rad, Hercules, CA). Each batch was standardized by protein content, and 1 μg was suspended in 20 μl of PBS for in vivo mouse studies. This method of exosome isolation routinely yields a relatively pure population of nanovesicles that are <100 nm (as visualized by electron microscopy), and enriched in exosomal proteins (as determined by Western blotting and fluorescence-activated cell sorting [FACS]), such as Hsc70, CD81, CD80/86, class I MHC, and class II MHC.

Exosome administration in the DTH model.

C57BL/6 mice were sensitized by subcutaneous injection of 100 μg of keyhole limpet hemocyanin (KLH) antigen emulsified 1:1 in Freund's complete adjuvant (CFA) at a single dorsal site. Ten days later, 1 hind footpad of each immunized mouse was injected intradermally with 106 DCs or 1 μg of DC-derived exosomes plus KLH antigen (20 μg) in 50 μl total volume. The contralateral footpad was injected with an equal volume of saline plus antigen (20 μg in 50 μl) without DCs or exosomes. Footpad swelling was measured using a spring-loaded caliper (Dyer, Lancaster, PA). Results were expressed as the difference in swelling (×0.01 mm) before and after antigen boost injection. The in vivo experiments were performed with 5 mice per group and repeated at least twice to ensure reproducibility.

Murine CIA model.

Male DBA/1LacJ (H-2q) mice (7–8 weeks of age) were purchased from The Jackson Laboratory and maintained in a pathogen-free animal facility at the University of Pittsburgh Biotechnology Center. Bovine type II collagen (Chondrex, Redmond, WA) in 0.05M acetic acid at a concentration of 2 mg/ml was emulsified in an equal volume of CFA, and 150 μg was injected into the base of the tail. For treatment of established arthritis, mice were injected with 20 μg of lipopolysaccharide (LPS) intraperitoneally on day 28 to induce synchronous disease onset. Four days after LPS injection (on day 32), DCs or exosomes from AdIDO-transduced or nontransduced DCs were intravenously injected into the mice with evidence of disease. The in vivo experiments were performed with 12 mice per group and repeated twice to ensure reproducibility.

Mice were scored using an established macroscopic system with a scale of 0–4, where 0 = normal, 1 = detectable arthritis with erythema, 2 = significant swelling and redness, 3 = severe swelling and redness from joint to digit, and 4 = maximal swelling and deformity with ankylosis. The macroscopic score was expressed as a cumulative value for all paws, with a maximum possible score of 16 per mouse.

Western blot analysis.

For Western blotting, exosomal proteins (3–10 μg) were separated by 12% or 15% sodium dodecyl sulfate–polyacrylamide gel electrophoresis, semi-dry transferred onto polyvinylidene difluoride, and detected by Western blotting using an enhanced chemiluminescence detection kit. Primary antibodies used for Western blotting were rabbit polyclonal anti–green fluorescent protein (anti-GFP; Santa Cruz Biotechnology, Santa Cruz, CA), mouse monoclonal anti-Hsc70 (Santa Cruz Biotechnology), rabbit polyclonal anti–β-actin (Abcam, Cambridge, MA), mouse monoclonal anti-IDO (Upstate Biotechnology, Lake Placid, NY), and rat monoclonal anti-IDO (Santa Cruz Biotechnology). Semiquantitative analysis of the protein density bands was performed using the ImageJ program (NIH Image, National Institutes of Health, Bethesda, MD; online at: http://rsbweb.nih.gov/ij/).

Measurement of kynurenine concentrations.

Kynurenine was measured in exosomes (1 μg) or exosome-free supernatants (60 μl) using a spectrophotometric assay. Samples were mixed 2:1 with 30% trichloroacetic acid, vortexed, and centrifuged at 3,716g for 10 minutes. Seventy-five microliters of supernatant was then added to an equal volume of Ehrlich's reagent (2% 4-[dimethylamino]benzaldehyde in glacial acetic acid) in a 96-well microtiter plate. Samples were analyzed in triplicate against a standard curve of L-kynurenine (0–5 mM; Sigma). Absorbance was measured at 490 nm using a microplate reader.

Flow cytometry.

For phenotype analysis of DCs, the cells were blocked with normal goat serum and then stained with phycoerythrin- or fluorescein isothiocyanate–conjugated monoclonal antibodies (BD PharMingen, San Diego, CA) against murine CD11b, CD11c, CD80, CD86, H-2Kb, and I-Ab. DCs were analyzed by FACSCalibur (Becton Dickinson, San Jose, CA) and the CellQuest software program. Isotype-matched irrelevant monoclonal antibodies were used as negative controls.

Statistical analysis.

Results were compared by analysis of variance with Fisher's post hoc analysis of least significant difference. When appropriate, the Kruskal-Wallis nonparametric test was used to compare means between groups. P values less than 0.05 were considered statistically significant, and all tests were conducted using SPSS statistical software (SPSS, Chicago, IL).

RESULTS

Reduction in the severity of murine CIA after administration of DCs modified to overexpress IDO or of exosomes from DCs modified to overexpress IDO.

Since overexpression of IDO can result in immunosuppression and tolerance in vitro and in vivo in certain animal models, such as models of transplant tolerance (10), we examined the ability of DCs genetically modified to overexpress IDO and the exosomes derived from the modified DCs to treat CIA. BM-derived DCs were infected with AdIDO (a recombinant adenoviral vector expressing IDO) or were left uninfected, and the exosomes and cells were harvested 3 days later. The DCs (106 cells) or exosomes (1 μg) were injected intravenously, 32 days after immunization, into DBA/1 mice with early-stage arthritis. A single injection of either DCs overexpressing IDO or exosomes from DCs overexpressing IDO reversed the progression of arthritis, while disease progressed normally in the control groups injected with saline or uninfected DCs (Figure 1).

Figure 1.

Therapeutic effect of indoleamine 2,3-dioxygenase (IDO)–positive dendritic cells (DC/IDO) and DC-derived exosomes in murine collagen-induced arthritis. Exosomes were isolated from bone marrow–derived DCs from DBA/1 mice that were previously infected with AdIDO (EXO/IDO) or were left uninfected (EXO/NON). DBA/1 mice were immunized with bovine type II collagen and received lipopolysaccharide 28 days after immunization to synchronize disease onset. The purified exosomes or DCs were injected intravenously into mice 32 days after immunization. Mice were monitored periodically, and each paw was scored using an established macroscopic scoring system, as described in Materials and Methods. Bars show the mean and SD macroscopic score (expressed as a cumulative value for all paws, with a maximum possible score of 16) of 12 mice per group. The arrow indicates the day of treatment. ∗ = P ≤ 0.001 versus uninfected control DCs (DC/NON) and controls injected with saline; P < 0.05 versus exosomes from uninfected controls.

The uninfected exosomes were also found to be slightly therapeutic, as previously observed (4). This is likely due to the fact that there may be a significant number of still immature and thus antiinflammatory and tolerogenic BM-derived DCs in the preparation of mock-infected or uninfected DCs, as shown by our previous analysis (33) (Table 1). Western blot analysis confirmed the expression of IDO in the infected lysate of DCs overexpressing IDO, as well as in the exosomes from DCs overexpressing IDO. No IDO was detectable in the uninfected cells or exosomes by Western blotting. Also, FACS analysis showed that there was no significant change in the maturation status of the IDO-infected cells (Table 1). Since exosomes were as or more therapeutic than the DCs in the CIA model, and were likely more stable (34–36), most of the following experiments were carried out using exosomes alone.

Table 1. FACS analysis of DCs used in the present study*
 CD80CD86Class I MHCClass II MHC
  • *

    Dendritic cells (DCs) were stained with phycoerythrin (PE)–conjugated monoclonal antibodies against murine surface molecules (CD80, CD86, H-2Kb, I-Ab, and appropriate isotype controls). Isotype-matched irrelevant monoclonal antibodies were used as negative controls (1.12% PE-labeled). Values are the percentage of cells that were PE-labeled. FACS = fluorescence-activated cell sorting; MHC = major histocompatibility complex; IDO = indoleamine 2,3-dioxygenase; 1-MT = 1-methyl-D-tryptophan.

Uninfected DCs23.6527.8974.8434.87
ψ5-positive DCs29.8645.2854.6645.38
IDO-positive DCs24.2636.6061.1835.10
CTLA-4Ig–positive DCs31.5549.3953.6148.90
CTLA-4Ig–positive DCs treated with 1-MT30.8548.6252.5648.31
CTLA-4Ig–positive DCs treated with L-tryptophan27.8142.1551.8142.10

Therapeutic effect of exosomes from DCs expressing CTLA-4Ig in a murine CIA model.

CTLA-4Ig is a synthetic fusion protein that binds with high affinity to B7-1 and B7-2, resulting in up-regulation of IDO in certain DC subsets and in immunosuppression (25–28). It has previously been shown that DCs genetically engineered to express CTLA-4Ig plus NF-κB oligodeoxyribonucleotide decoys, to prevent DC maturation, could significantly prolong heart allograft survival (37). To determine whether exosomes from DCs expressing CTLA-4Ig would be immunosuppressive as well, BM-derived DCs were infected with AdCTLA-4Ig or left uninfected, and the exosomes were collected 3 days later. Western blot analysis confirmed an increased expression of IDO in DCs following AdCTLA-4Ig infection (Figure 2A).

Figure 2.

IDO-dependent therapeutic effect of CTLA-4Ig–positive DCs and DC-derived exosomes (EXO/CTLA-4Ig) in murine collagen-induced arthritis. Exosomes were isolated from DBA/1 bone marrow–derived DCs that were previously infected with AdCTLA-4Ig and treated with 1-methyl-D-tryptophan (EXO/CTLA-4Ig+MT1) or L-tryptophan (EXO/CTLA-4Ig+L-Trp) or were left uninfected. A, DC extracts (2 μg) from uninfected AdCTLA-4Ig cells were subjected to 12% sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE) and immunoblotted for IDO or β-actin. Values are the relative semiquantitative expression of IDO after normalization to β-actin using the National Institutes of Health software program ImageJ. B, DBA/1 mice were immunized with bovine type II collagen and received lipopolysaccharide 28 days after immunization to synchronize disease onset. The purified exosomes or DCs were injected intravenously into mice 32 days after immunization. Mice were monitored periodically, and each paw was scored using an established macroscopic scoring system, as described in Materials and Methods. Bars show the mean and SD macroscopic score (expressed as a cumulative value for all paws, with a maximum possible score of 16) of 12 mice per group. The arrow indicates the day of treatment. ∗ = P ≤ 0.05 versus exosomes from DCs expressing CTLA-4Ig. C, Five micrograms of total exosome protein was subjected to 15% SDS-PAGE and immunoblotted for IDO, CTLA-4Ig, or β-actin. See Figure 1 for other definitions.

As with exosomes from the DCs infected with AdIDO, a single injection of exosomes from DCs overexpressing CTLA-4Ig reversed the progression of arthritis, while disease progressed normally in the control group injected with saline (Figure 2B). This immunosuppression was dependent on IDO-mediated deprivation of tryptophan, evidenced by the fact that addition of the competitive IDO inhibitor, 1-MT, or excess L-tryptophan to the DCs reduced the effect. It is important to note that the IDO inhibitors were added to the DCs and removed prior to the exosome isolation procedure. Although CTLA-4Ig is present in the exosomes from the AdCTLA-4Ig–infected DCs (Figure 2C), the ability of 1-MT and L-tryptophan to abolish the therapeutic effect of the exosomes suggests that their suppressive effects are dependent upon IDO activity in the DC.

Inhibition of DTH response by local administration of DCs overexpressing IDO or exosomes from DCs overexpressing IDO.

To determine if the exosomes from DCs genetically modified to express IDO were therapeutic in an antigen-specific model of inflammation that is more amenable to analysis of mechanism, a mouse model of DTH was used. In this model, mice were immunized to a specific antigen, KLH, and a Th1-mediated inflammatory response was induced 2 weeks after immunization by intradermal injection of the specific antigen into the hind footpads. We have previously used this model to demonstrate that DCs and DC-derived exosomes transduced with AdvIL-10, AdIL-4, and AdFasL were antiinflammatory (2–4). Either 106 DCs or 1 μg of exosomes were injected into 1 hind paw of KLH-immunized mice at the same time as a KLH boost injection into both hind paws. Local injection of DCs overexpressing IDO and exosomes from DCs overexpressing IDO significantly suppressed paw swelling (Figure 3A), not only in the treated paw, but also in the untreated contralateral paw 24 hours, 48 hours, and 72 hours after injection of antigen. In contrast, injection of DCs expressing Ψ5 or exosomes derived from control DCs did not inhibit the DTH response.

Figure 3.

Suppression of delayed-type hypersensitivity (DTH) in a murine model by IDO-positive DCs and DC-derived exosomes. A, Suppressive effect of exosomes from IDO-positive DCs in the DTH model. Exosomes were isolated from bone marrow–derived DCs that were previously infected with AdΨ5 (Exo/Psi5; control) or AdIDO (Exo/IDO). The purified exosomes or DCs were injected into the right hind footpad of keyhole limpet hemocyanin (KLH)–immunized C57BL/6 mice concurrently with injection of KLH into both hind footpads, and the extent of swelling was measured at 48 hours. Bars show the mean and SD (n = 5 mice per group). ∗ = P ≤ 0.001 versus Ψ5-infected DCs, exosomes from Ψ5-infected DCs, and saline controls for both treated and contralateral paws. B and C, IDO-dependent suppressive effect of exosomes from CTLA-4Ig–positive DCs in the DTH model. Exosomes were isolated from bone marrow–derived DCs that were infected with either AdΨ5, enhanced green fluorescent protein (EGFP; controls), or AdCTLA-4Ig and left untreated or treated with 1-methyl-D-tryptophan (1-MT) or L-tryptophan. The purified exosomes or DCs were injected into the right hind footpad of KLH-immunized C57BL/6 mice concurrently with injection of KLH into both hind footpads, and the extent of swelling was measured at 48 hours (B) or 24 hours (C). Note that C shows exosomes only. Bars show the mean and SD (n = 5 mice per group). In B, ∗ = P ≤ 0.001 versus EGFP-infected DCs, exosomes from EGFP-infected DCs, and saline-treated controls. In C, ∗ = P ≤ 0.05 for CTLA-4Ig versus Ψ5-infected controls, CTLA-4Ig–infected DCs treated with 1-MT, and uninfected controls for the treated paws; P ≤ 0.05 for IDO versus all groups except CTLA-4Ig and CTLA-4Ig–infected DCs treated with L-tryptophan for both treated and contralateral paws. See Figure 1 for other definitions.

These results demonstrate that a single, local footpad injection of genetically modified DCs expressing IDO as well as exosomes derived from the DCs overexpressing IDO suppresses the DTH response not only in the treated paw, but also in the untreated contralateral paw, suggesting a systemic effect following local injection. This is not due to systemic injection of DCs or exosomes, since they were injected intradermally into the hind footpad. Exosomes from DCs overexpressing CTLA-4Ig also suppressed inflammation in both the treated and contralateral paws, and were as effective as the DCs overexpressing CTLA-4Ig (Figure 3B). The addition of 1-MT, but not L-tryptophan, to the DCs inhibited the antiinflammatory effect of CTLA-4 (Figure 3C). These results show that exosomes from DCs overexpressing IDO and exosomes from DCs overexpressing CTLA-Ig can both be used therapeutically to suppress inflammation in the DTH model.

Antiinflammatory effect of exosomes from DCs overexpressing IDO is dependent on B7 molecules.

To determine whether the antiinflammatory effect of exosomes from DCs overexpressing IDO was dependent on the IDO present in the exosomes or due to modification of the immunosuppressive factors on the exosomes, vesicles deficient in factors important for conferring the suppressive effects of exosomes in other experiments were used. We have previously shown that the immunosuppressive activity of exosomes in the DTH model is dependent on class II MHC, FasL, B7-1, and B7-2 (refs.3 and4 and Ruffner MA, et al: unpublished observations). Thus, we examined the ability of exosomes from B7-deficient, IDO-positive DCs to suppress the DTH response. BM-derived DCs from wild-type, B7-1−/−, B7-2−/−, and double-knockout mice were transduced with AdIDO or a control vector, and exosomes were isolated.

Interestingly, loss of either or both of the B7 costimulatory molecules significantly reduced the antiinflammatory effect of exosomes from DCs overexpressing IDO (Figure 4A). This effect was not due to differing levels of IDO in the exosomes (Figure 4B). Therefore, the immunosuppressive effect of exosomes from DCs overexpressing IDO depends, at least partially, on the B7-1 and B7-2 molecules, consistent with our previous results obtained using DCs treated with IL-10 as well as exosomes derived from the DCs treated with IL-10 (Ruffner MA, et al: unpublished observations).

Figure 4.

Suppression of delayed-type hypersensitivity (DTH) in a murine model by exosomes is dependent on B7-1 and B7-2. Exosomes were isolated from bone marrow–derived DCs from wild-type (WT) C57BL/6 mice or B7-1–knockout (B7-1KO), B7-2–knockout, or double-knockout (DoKO) mice that were previously transduced with AdIDO or were uninfected (Non). A, The exosomes from the genetically modified DCs were tested in wild-type recipient mice. The purified exosomes were injected into the right hind footpad of keyhole limpet hemocyanin (KLH)–immunized mice concurrently with injection of KLH antigen into both hind footpads, and the extent of swelling was measured at 24 hours. Bars show the mean and SD (n = 5 mice per group). ∗ = P ≤ 0.05 versus all other groups for both treated and contralateral paws. B, Five micrograms of total DC or exosome protein extract was subjected to 15% sodium dodecyl sulfate–polyacrylamide gel electrophoresis and immunoblotted for IDO or β-actin. See Figure 1 for other definitions.

Lack of tryptophan metabolites in exosomes from DCs overexpressing IDO.

IDO depletes T cells of tryptophan, and also produces cytotoxic metabolites of tryptophan, collectively termed kynurenines, that regulate T cells (21). To determine whether exosomes from DCs overexpressing IDO could confer their immunosuppressive effects through delivery of cytotoxic kynurenines to T cells, we assayed for the presence of L-kynurenine in exosomes and in the exosome-free supernatants. For this experiment, we used TA3 Hauschka cells, a mouse mammary carcinoma line, because of their ease of infection and their ability to produce large quantities of exosomes without the use of limited primary cells. The cells were infected with AdIDO or with AdGFP as a control or were left uninfected. Western blot analysis confirmed the expression of IDO or GFP in the cells and exosomes (Figure 5A). The exosomes or exosome-free supernatants were then analyzed for L-kynurenine. Only the exosome-free supernatants from exosomes from DCs overexpressing IDO contained L-kynurenine (Figure 5B). Thus, exosomes appear not to carry detectable levels of the cytotoxic tryptophan metabolites.

Figure 5.

Lack of tryptophan metabolites in exosomes from indoleamine 2,3-dioxygenase (IDO)–positive cells. A, Overexpression of IDO in TA3 cells results in the release of exosomes containing IDO. TA3 cells were either uninfected (Non) or infected with recombinant adenovirus containing expression cassettes for either green fluorescent protein (GFP) or IDO. Exosomes were collected from the culture supernatant, and analyzed by Western blotting along with cell lysates for levels of IDO, GFP, and Hsc70. Seven micrograms of total protein extract from both exosomes and cells was loaded onto the gel. B, Uninfected, GFP-infected, and IDO-infected exosomes (1 μg) or exosome-free supernatants (60 μl) were assayed for levels of L-kynurenine. Bars show the mean and SD (n = 5 mice per group).

DISCUSSION

IDO is an immunomodulatory protein that has gained significant research interest in the last decade due to its ability to induce or maintain peripheral tolerance and immunosuppression in pregnancy, autoimmune disease, cancer, asthma, and transplantation (13, 15). IDO-expressing DCs can suppress the T effector response and activate Treg cells by either depleting the local area of tryptophan or producing toxic metabolites or both (23). In this study we examined the immunosuppressive activity of DCs genetically modified to express IDO and exosomes derived from the DCs in mouse models of CIA and DTH. Our results demonstrate that both DCs overexpressing IDO and exosomes from DCs overexpressing IDO can reverse established CIA and reduce inflammation in the DTH footpad model.

The mechanism of IDO-mediated immunosuppression in general is still poorly understood. It has been reported that both IDO-mediated local deprivation of essential tryptophan, and cytotoxic tryptophan metabolites may work together to suppress CD8+ effector T cells and to activate Treg cells (22, 23). Surprisingly, the exosomes from DCs overexpressing IDO were as suppressive in the CIA and DTH models as were the DCs overexpressing IDO. Since the exosomes contain exogenous IDO protein, they may function by delivering functional IDO to IDO-negative DCs or T cells, rendering the DCs tolerogenic and/or causing T cell anergy. We did not detect any L-kynurenine metabolic product in the samples of exosomes from DCs overexpressing IDO. L-kynurenine was detected only in the exosome-free supernatants, suggesting that delivery of toxic metabolites is not the mechanism of IDO-mediated immunosuppression. There was also no significant change in the maturation status of DCs overexpressing IDO, suggesting that the therapeutic effect was not due to a change in DC maturity.

We hypothesize that IDO expression in the DC modifies the DC-derived exosomes in some other way(s) to render them tolerogenic. Indeed, we demonstrated a role for components of exosomes in conferring the suppressive effects of exosomes from DCs overexpressing IDO. In particular, we demonstrated that the costimulatory molecules B7-1 and B7-2, which are required for the suppressive effects of DCs and exosomes, are partially required for the suppressive effects of exosomes from DCs overexpressing IDO. This result suggests that the exosomes could be directly interacting with T cells. However, it is also possible that the exosomes interact with endogenous APCs to alter their function through a B7-dependent mechanism. Consistent with this hypothesis, we have previously demonstrated the necessity of class II MHC in both the exosomes and in the recipient mice for exosomes to regulate T cell responses in vivo (3).

We have also demonstrated that exosomes derived from DCs overexpressing CTLA-4Ig are immunosuppressive in the CIA and DTH models. CTLA-4 on Treg cells or soluble CTLA-4Ig can induce functional IDO in DCs by binding to B7 molecules (10). In our BM-derived DCs, we found, by semiquantitative measurement, an increase of ∼3 fold in IDO expression after CTLA-4Ig infection, but this increase was not observed in the exosomes. We demonstrated that the suppressive activity of the exosomes derived from DCs overexpressing CTLA-4Ig was dependent upon IDO activity in the DC by using 1-MT and L-tryptophan in the CIA model. In the DTH model, only 1-MT was able to block CTLA-4–induced suppression, while L-tryptophan had no effect. Thus, the mechanism may be slightly different in the 2 models. We do not believe the difference is due to any toxic effect of 1-MT, since there was never any increase in cell death during treatment (data not shown).

Overall, this study highlights the potential therapeutic use of exosomes from DCs genetically engineered to overexpress IDO. Moreover, the results demonstrate that IDO activity can modify the activity of DC-derived exosomes, rendering them more immunosuppressive. The use of exosomes, instead of DCs, allows for a more stable delivery method, without loss of activity (34–36). While this study focuses on arthritis, it is also likely that DCs overexpressing IDO and/or exosomes from DCs overexpressing IDO may have therapeutic effects in other models of autoimmunity in which IDO has been shown to have immunosuppressive effects.

AUTHOR CONTRIBUTIONS

Dr. Robbins had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study design. Bianco, Kim, Robbins.

Acquisition of data. Bianco, Kim.

Analysis and interpretation of data. Bianco, Kim, Ruffner, Robbins.

Manuscript preparation. Bianco, Ruffner, Robbins.

Statistical analysis. Ruffner.

Acknowledgements

We would like to thank Ms Joan Nash for technical assistance and Dr. Maliha Zahid for help with statistical analysis.

Ancillary