Dendritic cells (DCs) can induce tolerance or immunity. We identified and characterized an IDO-expressing and an IDO-negative human DC population after stimulation by various proinflammatory stimuli. IDO expression was strongly dependent on the maturation status of the cells (CD83-positive cells only). The two DC subpopulations remained IDO positive and IDO negative, respectively, over a time period of at least 48 h. IDO enzyme activity of human DCs was highest during stimulation by strongly maturation-inducing TLR ligands such as highly purified LPS (TLR4 ligand) or polyriboinosinic-polyribocytidilic acid (TLR3 ligand); factors of the adaptive immune system such as IFN-γ, a mixture of cytokines, and IFN-α had lesser stimulatory capacity for IDO induction and activity. After stimulation with CD40L, IDO-positive DCs expressed significantly increased levels of B7 family molecules such as CD40, CD80, CD86, ICOS ligand, as well as PD-L1 (B7-H1) and PD-L2 (B7-DC) compared with the IDO-negative DC subset. At the same time, the inhibitory receptors Ig-like transcripts 3 and 4 were significantly downregulated on IDO-positive cells. Functionally, IDO-positive DCs produced significantly more IL-1β and IL-15 and less IL-10 and IL-6 than the IDO-negative subset after CD40L stimulation. These results show that IDO expression is associated with a distinctive phenotype and functional capacity in mature DCs. It seems likely that the IDO-positive DC subset possesses a regulatory function and might skew a T cell response toward tolerance.
Immature dendritic cells (iDCs) express receptors for inflammatory chemokines and migrate to sites of inflammation. Inflammatory cytokines as well as classes of microbial molecules (pathogen-associated molecular patterns) trigger maturation and drive migration of dendritic cells (DCs) into the T cell areas of the draining lymph nodes. Stimuli derived from activated T cells such as IFN-γ and CD40L further enhance DC activation and survival. The outcome of the DC–T cell interaction can be productive priming or tolerance. The long-held view that iDCs lead to tolerance and mature DCs (mDCs) lead to immunity has been contested in a number of reports (1–3). It is now clear that DCs initiate and regulate immune responses in a manner that depends on the signals they receive from microbes and their cellular environment (1). Accumulating evidence suggests that the capacities to stimulate or to inhibit T cell responses can be vested in one cell type, the mDC. In humans, several DC subsets have been described to date, the best-investigated being the myeloid DC subset and the plasmacytoid DC subset. Most studies with human myeloid DCs have been performed with in vitro generated DCs. These generated DCs have also been used in clinical trials to stimulate T cell responses against, for example, melanoma Ags, although these strategies have not always been successful (4).
IDO (number 12) is an enzyme that degrades the essential amino acid tryptophan (trp) inside the IDO-expressing cell and from the environment. Because T cells are especially sensitive to low trp levels, IDO activity blocks T cell proliferation (5). Furthermore, in vitro and in vivo the activation of IDO, followed by the degradation of trp, has been shown to have a wide range of biological consequences. In vitro, it was shown that naive T cells differentiate into T regulatory cells (Tregs) when cultured at low trp and high metabolite concentrations, and monocytes in these conditions differentiate into regulatory DCs with increased inhibitory receptors such as inhibitory receptor Ig-like transcript (ILT) 3 and ILT4 (6, 7). In vivo, high IDO activity prevents the rejection of allogeneic fetuses or grafts, and may play a role in the prevention of allergy (8–10).
IDO is not constitutively expressed in APCs, epithelial cells, fibroblast, or tumor cells, but can be induced so by a variety of stimuli. The best-established IDO-inducing protocol in vitro in human DCs is with IFN-γ, but other soluble factors such as TNF-α or IL-1 also strongly increase IDO expression (11, 12). In human APCs, the degradation of trp can also be initiated by ligation of CD40 by CD40L, of CD80/86 by CTLA-4 on activated T cells, or by ligation of the high-affinity receptor for IgE, FcεRI, by human IgE (8, 13–15). Strongly maturation-inducing TLR ligands such as highly purified LPS or polyriboinosinic-polyribocytidilic acid (poly I:C) also induce IDO expression in APCs, although this is more controversial (11, 16). It is not known whether subsets of IDO-expressing DCs exist or whether all in vitro generated DCs express IDO because most of these investigations studied only whole populations by Western blotting.
In the current study, using flow cytometric analysis, we explore the expression of IDO protein in human monocyte-derived DCs. Surprisingly, the expression of IDO was stably confined to a subgroup of mDCs that expresses significantly increased levels of a number of surface molecules and had a different cytokine profile compared with the IDO-negative DC subgroup. Importantly, the two subgroups of IDO-positive and IDO-negative DCs persisted after 48 h of stimulation, suggesting that IDO expression is associated with a distinct differentiation mode of human DCs.
Materials and Methods
The following mAbs were used for flow cytometric analysis: RD1-labeled T6RD1 against CD1a (IgG1; Beckman Coulter, Krefeld, Germany); PE-labeled mAb to CD14, ICOSL, PDL-1, and PDL-2 (IgG2a; BD Biosciences, Heidelberg, Germany); and unlabeled mAbs to CD40, CD80, CD86, CD11a, CD11b, CD11c, and CD123, and mAbs to ILT2 (all IgG2b) (BD Biosciences). Mouse anti-human IDO mAb was provided by Dr. O. Takikawa (Aichi, Japan), IgG1 Ab, and used 1:150 for flow cytometry, as described (17). Purified mAb HB15a detecting CD83a, levo-1-methyl tryptophan (L-1MT), and dextro-1-methyl tryptophan (D-1MT) were from Sigma-Aldrich (IgG2b; Taufkirchen, Germany). mAbs to ILT3, ILT4 (IgG2a), and CCR6-PE were from R&D Systems (Wiesbaden, Germany). FITC-labeled F(ab′)2 of goat anti-mouse Ab and mouse serum were from Dianova (Hamburg, Germany). Mouse mAb L243 against anti-MHC class II and anti-MHC class I mouse mAb W6/32 were provided by N. Koch (Division of Immunobiology, University Institute of Genetics, Bonn, Germany). The 7-aminoactinomycin D was from Sigma-Aldrich. RPMI 1640 medium and FCS were from Cambrex (Verviers, Belgium). l-glutamine and antibiotics/antimycotics were from Invitrogen (Karlsruhe, Germany). GM-CSF was from Novartis Pharma (Nürnberg, Germany), and human rIL-4 was from Strathmann Biotech (Hannover, Germany). Ultrapure LPS and poly I:C were from Sigma-Aldrich; IFN-α (rIFN α-2b) was from INTRON A, Essex Pharma (Munich, Germany); IFN-γ and CD40L were from R&D Systems.
Isolation of monocytes
Human monocytes were obtained from healthy donors. Written informed consent was obtained from all patients, and the protocol was approved by the local ethic commitee. Monocytes were isolated from peripheral blood with a density gradient using Nycoprep (Axis-Shield, Oslo, Norway), according to the manufacturer’s protocol. Monocyte isolation was confirmed by CD14 expression and was >90%. PBMCs were isolated from heparin blood by density gradient centrifugation with Lymphoprep (Axis-Shield) for 30 min at 1000 × g. CD14+ monocytes were separated by negative selection via magnetic beads using CD14 isolation kits (Miltenyi Biotec, Bergisch Gladbach, Germany) and an autoMACS Separator (Miltenyi Biotec).
Generation of monocyte-derived DCs
Monocytes (1 × 106/ml) were cultured in the presence of 500 IU/ml human rGM-CSF (Novartis Pharma) and 500 IU/ml human rIL-4 (Strathman Biotech) in RPMI 1640 supplemented with 10% FCS, 1% l-glutamine, and 1% antibiotics/antimycotics in 24-well plastic plates. On day 6 of culture, nonadherent iDCs were collected by moderate aspiration.
In vitro stimulation of iDCs
For stimulation and maturation, iDCs were treated for 24 h with either 1000 U/ml IFN-γ, 1000 U/ml IFN-α, 25 μg/ml poly I:C, 100 ng/ml LPS, 1 μg/ml CD40L, or a cytokine mixture comprising 10 ng/ml TNF-α (Sigma-Aldrich), 10 ng/ml IL-1β, and 5 ng/ml IL-6 (R&D Systems). Time course of DC subpopulation analysis was done at 12, 24, and 48 h after stimulation with CD40L (1 μg/ml); at 24 h, DCs were restimulated with CD40L. Unstimulated controls were included. The cells were washed twice before staining.
Immunostaining and flow cytometric analysis
Cell analysis was performed, as previously described (18). Intracellular staining with anti-IDO Ab was done as follows: after fixation and permeabilization with 4% formaldehyde and saponine buffer, respectively, cells were stained with anti-IDO Ab (2.5 μg/ml) for 20 min, followed by FITC-labeled F(ab′)2 of goat anti-mouse Ab from Jackson ImmunoResearch Laboratories (West Grove, PA). If cytokines were stained intracellularly in addition to IDO, cells were washed with saponine buffer, and anti–TGF-β PE Ab and anti–IL-1β PE Ab (both from R&D Systems), anti–IL-10 PE Ab, anti–IL-12p70 PE Ab, anti–IL-15 PE Ab, and anti–IL-6 PE (all from BD Biosciences), respectively, were added for another 20 min. After washing with saponine buffer and PBS/BSA/Na-Azid, the extracellular staining was performed with anti-CD83 mAb or anti-CD1a mAb, or another surface Ag. Analysis of cells was performed with a FACS Canto cytometer (BD Biosciences).
The vital and CD1a+ DC population was gated by a combination of forward and side scatter and CD1a/7-aminoactinomycin D gate sets. To determine surface Ags or intracellular cytokines of IDO+ and IDO− DCs, additional gates were set around IDO+CD1a+CD83+ and IDO−CD1a+CD83+ DCs, respectively. The median fluorescence intensity for each Ag of the vital CD1a+ DC population was determined.
Total RNA was extracted from DCs using TRIzol (Invitrogen), according to the manufacturer’s instructions. Reverse transcription was done with 1 μg total RNA. The resulting cDNA was amplified by PCR-specific primer sequences for the genes as follows: human β-actin, forward, 5′-GAGCGGGAAATCGTGCGTGACATT-3′ and reverse, 5′-GATGGAGTTGAAGGTAGTTTCGTG-3′ (240 bp); human IDO-1, forward, 5′-CTTCCTGGTCTCTCTATTGG-3′ and reverse, 5′-GAAGTTCCTGTGAGCTGGTG-3′ (430 bp). Amplification was performed on a Perkin-Elmer GeneAmp PCR System 9600 thermocycler (Applied Biosystems, Weiterstadt, Germany). β-Actin was used to normalize specific PCR amplifications. PCR fragments were separated on 1% agarose gels, visualized using ethidium bromide staining, and analyzed by digital image analysis using the WinCam system (Cybertech, Berlin, Germany).
To determine IDO enzymatic activity, the samples were deproteinized by the addition of 250 μl 5% sulfosalicylic acid to 250 μl cell culture supernatant. After incubation for 10 min at room temperature, samples were centrifuged at 13,000 × g for 10 min at 8°C. l-trytophan and l-kynurenine were assayed by HPLC with UV detection (UV detector SPD-10A at 245 nm; Shimadzu, Duisburg, Germany).
Analyses were performed with SPSS 12.0 statistical software. Statistical analysis was done by using parametric Student t test and nonparametric Wilcoxon test. The p values <0.05 were considered as statistically significant. Results are expressed as mean ± SEM.
Activation of iDCs by DC-maturing stimuli leads to IDO gene expression
There is no full consensus in the literature as to whether IDO can be detected constitutively in human DCs, either mature or immature (19). Therefore, we first determined IDO gene transcription in immature (iDCs) and mature (mDCs) human monocyte-derived DCs.
Monocyte-derived iDCs were matured on day 6 of differentiation by the addition of various stimuli for 24 h. Maturity was determined as the expression of cell surface CD83. Cells were collected and analyzed by flow cytometry. As shown in Fig. 1, the cells expressed the characteristic maturation Ags after stimulation.
To test whether these DCs express IDO, IDO gene transcription was studied in iDCs and mDCs after stimulation of iDCs with IFN-γ (1000 U/ml), cytokine mixture (10 ng/ml TNF-α, 10 ng/ml IL-1β, and 5 ng/ml IL-6), poly I:C (25 μg/ml), or 10 ng/ml LPS by RT-PCR using IDO-specific primers (Fig. 2). The data indicated that iDCs did not constitutively transcribe IDO. However, after maturation for 24 h, IDO transcripts were clearly detectable.
Flow cytometric analysis reveals different proportions of IDO-positive and IDO-negative mDCs after stimulation of DCs
Although a number of studies have demonstrated the induction of IDO in DCs by these stimuli on the level of the total population, we were interested to know whether all cells evenly express the enzyme. To test this, iDCs and mDCs were analyzed for CD83 and intracellular IDO protein expression by flow cytometry after 24 h of stimulation (Fig. 3A, 3B). As can be seen from the data and expected from Fig. 2, iDCs did not express IDO protein. In contrast, when DCs were stimulated by IFN-α, IFN-γ, cytokine mixture, CD40L, or TLR ligands such as poly I:C and LPS, detectable levels of IDO protein could be seen, yet not in all CD1a+ DCs. IDO protein expression was always associated with CD83 surface expression, and therefore maturation of DCs. However, unexpectedly, except for the cell population stimulated with LPS, not all mDCs expressed IDO. There was always a substantial proportion of CD83+ DCs that did not express IDO after stimulation. When stimulation was performed with LPS, DCs derived from 2 of 12 healthy donors responded with IDO induction in all CD83+ DCs. Fig. 3B shows the histogram after 24 h of CD40L stimulation.
To test whether IDO-negative and IDO-positive DC populations remain stable over a longer period of stimulation, three-color flow cytometry analysis for CD83, CD1a, and IDO was done at 12, 24, and 48 h of stimulation with CD40L (Fig. 4A). Maturation of the cells (as judged by CD83 expression) continued over this area until virtually all cells were positive for this marker. However, as can be seen from the stainings, two clear CD83+ DC populations, one IDO+CD83+ and one IDO−CD83+ DC population, were seen as early as after 12 h of stimulation, and this phenotype was stable at least until 48 h after stimulation. Other stimulation protocols verified these results of IDO+ DCs and IDO− DCs until 48 h after maturation (Fig. 4B).
IDO enzyme activity correlates with IDO protein expression in CD83+ DCs and is more effectively abrogated by L-1MT than by D-1MT
From the literature, it is not entirely clear whether the presence of IDO protein always means IDO enzyme activity. IDO enzyme degrades trp, which generates the metabolite kynurenine (kyn). Therefore, IDO enzyme activity leads to low trp concentration and an increased kyn concentration in culture. The concentration of trp and kyn was measured in the supernatants of 48-h–stimulated DCs in 100 μM trp culture medium to estimate IDO activity (Fig. 5A). HPLC analysis showed the highest kyn concentrations and lowest trp values with DCs that had been stimulated with TLR ligands such as LPS and poly I:C, although there was a high variability among donors. This correlated with the highest IDO protein expression in LPS/poly I:C–stimulated DCs as determined by flow cytometry (Fig. 3). CD40L and cytokine mixture stimulation showed a substantial degradation of trp and a moderate and low kyn production, respectively. IFN-γ and IFN-α induced less and variable trp degradation activity and kyn production by human myeloid DCs. In all cases, IDO activity correlated with the levels of protein expression as estimated by flow cytometry (Fig. 3), suggesting that the presence of IDO protein is the rate-limiting factor in the degradation of trp and the immunosuppressive activity. IDO enzymatic activity in DCs after stimulation with CD40L was more effectively inhibited by the stereoisomer L-1MT than by D-1MT at a concentration of 500 μM L- and D-1MT (Fig. 5B).
IDO+ DCs are CD80highCD86highICOSLhighPDL-1highPDL-2high
We next analyzed and compared the expression of surface proteins in IDO-positive and IDO-negative mDC subsets after CD40L stimulation for 24 h (Fig. 6A, 6B). Unstimulated iDCs served as control (Fig. 6C, 6D). We analyzed in particular surface Ags whose expression was described to be functionally relevant for IDO immune regulation, such as CD80, CD86, PDL-1, PDL-2, ILT3, and ILT4. The expression of ICOSL has been determined because the ICOSL/ICOS pathway has a broad role in regulating Tregs, T cell tolerance, and autoimmunity.
As can be seen in Fig. 6A and 6B, the expression of costimulatory molecules CD40, CD80, and CD86 as well as the expression of ICOSL, PDL-1, and PDL-2 was significantly increased on IDO+ DCs compared with the stimulated IDO-negative subset. This was in contrast to the inhibitory receptors ILT3 and ILT4, whose expression was significantly decreased expressed on IDO-expressing DCs compared with stimulated IDO-negative DCs. The markers CD123 and CCR6 did not coassociate with IDO (Fig. 6B). Rather, the IDO-positive DCs had a tendency toward less CD123 and CCR6 expression than IDO-negative DCs. Overall, these results suggest that CD83+IDO+ DCs express not only IDO, but also a number of other molecules that may be connected to its IDO-mediated regulatory functions.
IDO+ DCs produce increased amounts of IL-1β and IL-15 and decreased amounts of IL-6 and IL-10 compared with IDO− DCs
Some cytokines, such as IL-6, IL-1β, and TGF-β, have been described to be strong inducers of IDO. We therefore analyzed the expression of these cytokines by intracellular staining upon CD40L stimulation for 24 h along with the induction of IDO (Fig. 7A, 7B). As control, unstimulated iDCs have been explored (Fig. 7C, 7D). IDO-expressing cells produced significantly less IL-6 and IL-10 after stimulation than IDO-negative DCs. The production of TGF-β and IL-12 was variable within each population. The levels of the cytokines IL-1β and IL-15 were significantly increased in the IDO-positive DC subset compared with the IDO-negative DC subset. The differential expression of cytokines suggests functional differences between the two populations.
The population of DCs is increasingly heterogeneous. One aspect that has become clear is that DCs may both stimulate and tolerize T cells, and evidence suggests that the latter is in part due to the expression of IDO (20). As DCs have a number of molecules and systems that may determine their regulatory or stimulatory capacity, we undertook this study to understand whether IDO expression is homogenous in a population or defines DC subpopulations.
In this study, we found by flow cytometry that: 1) only a proportion of activated mDCs expresses IDO protein, and the IDO-positive and IDO-negative DC populations remain stable even after a longer period (48 h) of stimulation; 2) IDO expression in these DCs is always associated with the expression of maturity markers like CD83 on the cell surface; 3) IDO-positive DCs show higher expression of distinct surface markers such as CD80, CD86, PDL-1, and PDL-2 compared with IDO-negative DCs; and 4) IDO-positive DCs express a pattern of cytokine that is different from IDO-negative DCs.
The role of phenotypically mature DCs has lately been revisited, and there is not yet a consensus in the literature (19, 21–23). When used for studies in animals or human trials, in vitro generated monocyte-derived DCs do not always generate a sufficient immune response, which may reflect inactive or even tolerizing properties in the population (24, 25). Minor alterations in cell culture conditions can lead to different outcomes of the used DCs also with respect to IDO production (11), which is all suggestive of the presence of a variety of different DC populations. It has long been speculated that a specialized subset of mDCs might actively divert T cell responses toward tolerance (26). However, until now no such subset has been defined. Although it has been shown that IDO-competent DCs are capable of inhibiting T cell proliferation in vitro (27) and might induce tolerance in vivo (9, 28), we are not able to select for these IDO-positive DCs, as we were unable to find surface markers that are uniquely expressed on IDO-positive DCs. Studies have shown a putative segregation for CD123 (27) and/or CCR6 (26) with IDO; however, in our study as well as in others (29), CD123 and CCR6 did not distinguish IDO-positive and IDO-negative DC expression. In our study, the surface markers such as CD80, CD86, or PDL-1/2 were associated with IDO expression, but the difference was rather gradual, making it difficult to use these markers for isolation of the cells. The gradation of quantitative expression of surface markers points also to the possibility that IDO+ DCs represent an “activation state” to which any DCs potentially could have access. The various proportions of IDO+ and IDO− DCs (the sizes depending on the stimulus) may reflect the heterogeneity of the starting population. DCs may be characterized by different activation status, different degree of maturity, different receptor expression, and therefore different sensitivities for the various stimuli (30). The relative size of the populations may be a quantitative effect (for instance, the signal through IFN-α may not be strong enough) or have qualitative elements (i.e., a graded expression of some receptors in a population of DCs). Intriguingly, the DC populations were stable over at least 48 h of continued stimulation, suggesting that there are indeed two different populations in each case, which do or do not respond to a given stimulus by IDO expression. The stimuli applied use different signaling pathways, but most likely merge in noncanonical NF-κB activation for IDO induction (31). Another possibility is that the induction of IDO occurs indirectly through a mixture of autocrine signals, which are induced by these primary stimuli (8).
Tregs have been reported to upregulate IDO expression in human DCs via binding of cell surface CTLA-4 to CD80/CD8 on DCs, resulting in B7-mediated induction of IDO (8, 14). It is conceivable that, once IDO-gene transcription and IDO-protein expression in tolerogenic DCs have been triggered, these DCs increase expression of their receptors CD80 and CD86 to reinforce reverse signaling through CTLA-4 on Tregs to sustain and mediate a tolerogenic, IDO-dependent milieu.
Interestingly, a recent study reports that IDO-activated Tregs (in this case, IDO-producing plasmacytoid DCs had activated resting Tregs) suppress target T cells in a PD-L–dependent manner, that is, activated Tregs increased PD-L on DCs that together with PD-1 expression on target T cells leads to suppression (32). It is conceivable that like B7.1 and B7.2, the expression of the PD ligands is increased in IDO-expressing tolerogenic DCs and mediates downstream effects of IDO, that is, suppression of T cells.
Unexpectedly, in our study, the inhibitory receptors ILT3 and ILT4 were significantly decreased in expression on IDO-positive DCs compared with IDO-negative DCs. The increased expression of ILT3 and ILT4 on APCs has been reported to be associated with tolerogenic properties of these cells (33, 34). In addition, we have shown that if DCs are generated under low trp conditions, these DCs acquire high ILT3 and ILT4 surface expression and lead to the induction of Tregs with suppressive activity (7). These differences probably reflect subtle functional differences or responses to changes in milieu in vivo.
Proinflammatory cytokines can, by autocrine or paracrine effects, regulate transcriptional expression of the IDO gene (35). Our data show an increased expression of the proinflammatory cytokine IL-1β in IDO-expressing DCs. However, the proinflammatory cytokine IL-6, known to restrain IDO activity by combined effects with suppressor of cytokine signaling 3 (36), is downregulated on these cells. In addition, the anti-inflammatory cytokine IL-10, brought together with the IDO pathway (26), the induction of Tregs, or tolerogenic DCs (37), is also significantly downregulated on IDO-positive DCs. To comprehensively link IDO expression and cytokine production clearly more detailed studies are needed in the future.
Our findings in this study might be significant for the design of DCs as therapeutic tools in vitro. In practice, the next required step is to find expression markers on DCs that are unequivocally linked to IDO competence, and then to select for these IDO+ DCs. We show an essential requirement to conduct flow cytometric analyses for the assignment of IDO competence if in vitro generated DC protocols are used for vaccination studies.
We thank Georg Häcker, Institute of Medical Microbiology and Hygiene (University of Freiburg), for helpful discussions and critical reading of the manuscript.
This work was supported by Grant SFB 704/TP A15 from the Deutsche Forschungsgemeinschaft and Bonn Forschungsförderprogramm.
Abbreviations used in this article:
- poly I:C
T regulatory cell
The authors have no financial conflicts of interest.