The West Virginia School of Osteopathic Medicine, Lewisburg, WV, USA
Correspondence to: B. L. Dickinson, Department of Biomedical Science, The Western Michigan University Homer Stryker M.D. School of Medicine, 1000 Oakland Drive, Kalamazoo, MI 49008-8010, USA. E-mail: Bonny.Dickinson@med.wmich.edu
Dendritic cells (DCs) initiate adaptive immune responses to pathogens and tumours and maintain tolerance to self and innocuous antigens. These functions occur in organs and tissues exhibiting wide variations in nutrients, growth factors, redox and oxygen tension. Understanding how these microenvironmental factors influence DCs to affect immunological outcomes is of increasing relevance with the emerging success of DC-based cellular vaccines. In a previous study, we examined whether redox, an important environmental cue, could influence DC expression of the immunosuppressive enzyme indoleamine 2,3-dioxygenase (IDO). IDO-competent DCs promote long-term immune homoeostasis by limiting exaggerated inflammatory responses and directing regulatory T-cell effector function. To alter redox, we manipulated the activity of the cystine/glutamate antiporter, which functions to maintain intracellular and extracellular redox. The results of that study showed that redox perturbation strongly induced IDO expression and activity in DCs. While this study was performed using standard cell culture techniques with DCs cultured under 5% CO2 and 20% O2, it is clear that DCs capture and present antigens in inflamed tissues and secondary lymphoid organs which exhibit low oxygen tension (1–5% O2). Therefore, here we investigated whether oxygen tension influences DC expression of IDO in the context of homoeostatic and altered redox.
Dendritic cells (DCs) link innate and adaptive immunity by translating innate immune responses into antigen-specific T- and B-cell responses . Importantly, these adaptive responses can be silenced by tolerogenic DCs . While significant heterogeneity is observed within tolerogenic DCs , those expressing the immunosuppressive enzyme indoleamine 2,3-dioxygenase (IDO) are well characterized [4-6]. IDO catalyses the initial and rate-limiting oxidative conversion of the essential amino acid tryptophan into signalling metabolites termed kynurenines [7, 8]. Both tryptophan depletion and kynurenine accumulation create a highly immunosuppressive milieu that suppresses T-cell responses by inhibiting T-cell proliferation, inducing T-cell apoptosis and expanding and stabilizing regulatory T cells [4, 9-15]. Because of its potent immunosuppressive effects, IDO expression and enzymatic activity are tightly regulated at the transcriptional, translational and post-translational level . This multi-pronged control mechanism has made it difficult to characterize IDO regulation within the complex microenvironments that DCs encounter in vivo.
Dendritic cells perform their effector function in the context of a barrage of signals that include microbial pathogen-associated molecular patterns, tumour cell-induced alterations in redox and oxygen tension, apoptotic and necrotic cell release of damage-associated molecular patterns, and the accumulation of reactive oxygen radicals and inflammatory mediators elaborated by innate and adaptive immune cells [17-21]. The mechanisms by which DCs integrate these environmental signals to either initiate or dampen immune responses in an antigen-specific manner are incompletely known. While several studies have shown that IDO can be induced by IFN-γ [22, 23], the combination of LPS and TNF-α  and PGE2 , it remains to be examined whether additional signals that substantially modify organ and tissue microenvironments also influence IDO expression.
To better understand the complex regulation of IDO, we recently examined whether redox, an important environmental cue, could regulate IDO expression and activity in human monocyte-derived DCs. To alter cellular redox, we manipulated the activity of the cystine/glutamate antiporter. The antiporter maintains intracellular levels of glutathione by taking up extracellular cystine in exchange for intracellular glutamate [25-27]. Once inside the cell, cystine is reduced to cysteine, the rate-limiting precursor for biosynthesis of the major cellular antioxidant glutathione (GSH) [28-30]. Cysteine and glutathione are also exported from the cell where they generate a reducing extracellular milieu [31-33]. Our data showed that blocking antiporter uptake of cystine significantly increased IDO mRNA, protein and enzymatic activity and that this correlated with impaired DC presentation of exogenous antigen to T cells via MHC class II and the cross-presentation pathways. These data suggest that redox regulates the functional expression of IDO.
Another major environmental cue is oxygen tension. Although DCs encounter wide variations in oxygen tension in vivo, DCs studied in vitro are cultured in nutrient-replete media with a balanced pH, an optimized temperature of 37 °C, 5% CO2 and 20% O2, an oxygen concentration significantly higher than that encountered in vivo [34-37]. Indeed, oxygen levels in blood are estimated at 10–12.5%  and approximately 3–6% in healthy tissues . Oxygen tension is even lower (1–3%) in inflamed tissues and lymph nodes . Thus, culturing cells in 20% O2 is inconsistent with the in vivo environment, and studies show that this clearly impacts immune cell function and behaviour [36, 41-44]. For example, studies show that oxygen tension significantly alters the DC transcriptome, interferes with DC maturation, and regulates chemokine receptors, chemotaxis and cytokine expression [42, 45-47]. For this reason, here we test the hypothesis that physiologic hypoxia (1% O2) alone or in combination with altered redox balance regulates IDO expression and activity in DCs.
Materials and methods
Escherichia coli 026:B6 lipopolysaccharide (LPS; γ-irradiated; total impurities <5% protein), NaCl, Tris base, NP-40, SDS, sodium deoxycholate and goat anti-rabbit IgG-HRP were from Sigma-Aldrich (St. Louis, MI, USA). RPMI 1640, FBS, penicillin, streptomycin sulphate and amphotericin B were from Invitrogen (Carlsbad, CA, USA). Cystine/cysteine-free medium was from MP Biomedicals (Solon, OH, USA). Mouse anti-IDO monoclonal antibody was from EMD Millipore (Billerica, MA, USA). Complete miniprotease inhibitor tablets were from Roche (Indianapolis, IN, USA). Goat anti-mouse IgG-HRP antibody was from Thermo Fisher Scientific, Inc. (Waltham, MA, USA), and rabbit anti-GAPDH monoclonal antibody was from Cell Signaling (Danvers, MA, USA). Oxygen tension studies were performed using a hypoxia chamber from Billups-Rothenberg, Inc. (San Diego, CA, USA). The BCA assay was from Pierce Biotechnology (Rockford, IL, USA). Trypan blue was from Life Technologies (Grand Island, NY, USA).
Human monocyte-derived dendritic cells
As described previously, DCs were derived from monocytes purified from normal human buffy coats . DCs were cultured in complete medium: RPMI 1640 containing 25 mm HEPES and supplemented with 10% heat-inactivated FBS, 10 U/ml penicillin, 10 μg/ml streptomycin sulphate, 25 ng/ml amphotericin B, 10 ng/ml (290 U/ml) IL-4 and 100 ng/ml (560 U/ml) GM-CSF.
As described previously, redox balance was disrupted by culturing DCs in cystine/cysteine-free medium: RPMI 1640 medium without L-glutamine, L-cystine, L-cysteine and L-methionine and supplemented with 25 mm HEPES, 10% heat-inactivated FBS, 10 U/ml penicillin, 10 μg/ml streptomycin sulphate, 25 ng/ml amphotericin B, 10 ng/ml (290 U/ml) IL-4 and 100 ng/ml (560 U/ml) GM-CSF .
Oxygen tension studies
To mimic physiologic hypoxia, DCs were cultured in a hypoxia chamber flushed with 1% O2, 5% CO2 and 94% N2. To mimic hyperoxia, DCs were cultured using a standard CO2 incubator that maintained an environment of 20% O2, 5% CO2 and 75% N2.
Cystine/glutamate antiporter assay
The activity of the cystine/glutamate antiporter was quantified as described previously . Briefly, DCs were incubated for 24 h in complete medium under atmospheric oxygen (20%) with LPS for 24 h to induce maturation. DCs were then collected, washed and incubated for an additional 24 h in complete medium under atmospheric oxygen (20%) or physiologic oxygen (1%) tension. Afterwards, DCs were harvested, washed and seeded into 96-well trays at 500,000 cells/well in uptake medium (137 mm NaCl, 0.7 mm K2HPO4, 1 mm CaCl2, 1 mm MgCl2, 5 mm glucose, 10 mm HEPES, pH 7.4). Then, uptake medium containing [3H] glutamate or [14C] cystine (final concentration of 50 or 200 μm) was added, and the cells were incubated for 5 min at 37 °C. After the incubation, trays were immediately placed in an ice-water bath, and the cells were washed twice with ice-cold uptake medium. Cells were lysed with 100 mm NaOH, and radioactivity in the lysate was determined by liquid scintillation counting and normalized to total protein in the lysate as determined using the BCA assay.
Quantification of IDO enzymatic activity
IDO enzymatic activity was examined by quantifying the accumulation of kynurenine in cell-free culture supernatants by the colorimetric method as described previously [49, 50].
IDO and GAPDH immunoblots
IDO and GAPDH were detected in DC lysates by immunoblot as described previously .
Dendritic cell viability was assessed by trypan blue exclusion as described previously . None of the treatments result in significant cell death. Culturing DCs under 20% oxygen in the presence or absence of cystine/cysteine had no significant effect on cell viability (% viable cells AVE ± SEM: complete medium = 87 ± 3 versus Cys/s-free medium = 87 ± 2; n = 5). Similarly, DCs cultured with 1% oxygen in the presence or absence of cystine/cysteine had no significant effect on cell viability (complete medium = 91 ± 1 versus Cys/s-free medium = 90 ± 1; n = 5).
TGF-β, IL-23 (p19) and MCP-1 (CCL2) were quantified in cell-free culture supernatants by ELISA as per the manufacturer's instructions. (Novus Biologicals, Littleton, CO, USA).
The Student's t-test was used to determine statistical differences between treatment groups (* P <0.05; ** P <0.01).
Oxygen tension suppresses the activity of the cystine/glutamate antiporter in human DCs
To examine whether oxygen tension influences IDO expression in DCs, we first addressed the possibility that oxygen tension could regulate the activity of the cystine/glutamate antiporter. This is based on our previous findings that blocking antiporter activity strongly induced IDO expression in DCs [49, 50]. To examine the effect of oxygen tension on antiporter activity, DCs were cultured with LPS in complete medium for 24 h to induce maturation under standard cell culture conditions (20% O2, 5% CO2 and 75% N2). Mature DCs were then collected and incubated for an additional 24 h in complete medium under 20% O2 (hyperoxic/atmospheric oxygen tension) or 1% O2 (physiologic hypoxia). Antiporter activity was then measured by quantifying the uptake of radiolabelled glutamate and cystine. We found that DCs cultured under 1% O2 exhibited significantly reduced uptake of both glutamate and cystine (Fig. 1A and B, respectively). DCs cultured under 1% O2 in medium containing 50 μm glutamate transported ~ 11-fold less glutamate (8 pmol/mg) than did DCs cultured under 20% O2 (86 pmol/mg). Similarly, when DCs were incubated with 200 μm glutamate under 1% oxygen, DCs transported ~5-fold less glutamate (29 pmol/mg) than did DCs cultured under 20% O2 (133 pmol/mg). Similar findings were observed when cystine uptake was measured. In a preliminary experiment, we found that 1% O2 also decreased the GSH-to-GSSG ratio, a key determinant of oxidative stress (data not shown). This finding is consistent with the results of our previous study showing that blocking antiporter activity induces oxidative stress . Taken together, these data suggest that low oxygen tension suppresses the activity of the cystine/glutamate antiporter and induces oxidative stress in human DCs.
Oxygen tension does not significantly influence IDO enzymatic activity
We next examined the effect of oxygen tension on IDO enzymatic activity in DCs. Based on our previous observation that IDO enzymatic activity was significantly increased by blocking the activity of the cystine/glutamate antiporter, we reasoned that low oxygen tension would upregulate IDO enzymatic activity through its suppressive effect on antiporter activity. To test this, DCs were treated with LPS for 4 h and then washed and incubated with LPS in complete medium or in cystine/cysteine-free medium under 1% or 20% O2. After 24 h in culture, kynurenine levels were quantified in cell-free supernatants. Consistent with our previous findings, blocking antiporter activity by culturing DCs in cystine/cysteine-free medium under 20% O2 significantly increased IDO enzymatic activity by ~11-fold when compared with DCs cultured in complete medium under 20% O2 (black bars, Fig. 2). Importantly, this was also true when cells were cultured in 1% O2, suggesting that redox control of IDO activity is not influenced by oxygen tension (compare the white Cys/s bar with the white control bar, Fig. 2). Interestingly, although culturing DCs in complete medium under 1% O2 significantly decreased antiporter activity, IDO enzymatic activity remained at control levels, similar to those measured in DCs cultured in complete medium under 20% O2 (compare the white control bar with the black control bar, Fig. 2). Furthermore, the combination of low oxygen tension and redox imbalance (cystine/cysteine-free medium) did not increase IDO enzymatic activity, as the levels were similar to that observed in DCs cultured under 20% O2 in cystine/cysteine-free medium (compare the white Cys/s bar with the black Cys/s bar, Fig. 2). Taken together, these data suggest that changes in oxygen tension neither contribute to nor influence redox control of IDO enzymatic activity in DCs.
Oxygen tension does not significantly alter IDO protein levels
Based on the above results, we anticipated that oxygen tension would not alter IDO protein levels. However, IDO protein levels do not necessarily correlate with enzymatic activity, making it possible that oxygen tension could regulate IDO protein levels without altering IDO enzymatic activity. To examine this, DCs were treated as described above and IDO protein levels were quantified by immunoblot. Consistent with our previous findings, blocking antiporter activity significantly increased IDO protein levels in DCs cultured under 20% O2 in complete medium (~1.5-fold increase over control; compare black bars, Fig. 3). This was also true for DCs cultured under 1% O2 in complete medium (~1.3-fold increase over control; compare white bars, Fig. 3). These data suggest that redox, but not oxygen tension, influences IDO protein levels. Taken together, this set of data shows that redox influences IDO activity by regulating IDO protein levels in a manner that is independent of oxygen tension.
Effect of redox and oxygen tension on DC cytokine and chemokine production
DCs elaborate cytokines to influence immunological outcomes. Tolerogenic IDO-competent DCs have been shown to produce high levels of IL-10 and TGF-β and low levels of the Th17-enhancing cytokine IL-23 [51, 52]. TGF-β has also been shown to be a potent inducer of sustained IDO production in DCs [9, 53]. Therefore, we next tested whether oxygen tension, in the context of homoeostatic or disrupted redox, could influence cytokine production by DCs. DCs were treated with LPS for 4 h, washed and incubated with LPS in complete medium or in cystine/cysteine-free medium under 1% or 20% O2 for 24 h, after which TGF-β and IL-23 were measured in cell-free culture supernatants by ELISA. Neither TGF-β nor IL-23 were detected in any of the conditions examined (data not shown). This set of data suggests that changes in oxygen tension and redox status do not influence DC production of TGF-β and IL-23 to sustain tolerance or promote Th17 responses, respectively.
Chemotactic cytokines can also influence immunological outcomes by recruiting relevant immune cells. Therefore, we next examined the effect of oxygen tension and redox on DC production of the chemokine MCP-1 (CCL2). MCP-1 displays chemotactic activity for monocytes, memory T cells and DCs, recruiting these cells into sites of inflammation and tissue damage, where a balance between an appropriate adaptive immune response and tolerance to self must be achieved [54-57]. When antiporter activity was suppressed, MCP-1 levels were reduced 1.4-fold under 20% O2 and 1.6-fold under 1% O2 relative to DCs cultured in complete medium (Fig. 4). Interestingly, whether DCs were cultured in complete medium or cystine/cysteine-free medium, MCP-1 production was increased 1.5-fold and 1.3-fold under 1% O2 when compared with DCs cultured under 20% O2 (Fig. 4, compare black bars with white bars). Thus, oxygen tension and redox modulate DC production of MCP-1 in a manner opposite to that observed for their control of IDO enzymatic activity (Fig. 5). Taken together, this set of data shows that oxygen tension and redox may be important environmental factors influencing the balance between immunity and tolerance through the control of IDO and chemokine expression.
The question addressed in the present study was whether variations in oxygen tension could influence IDO expression and activity in DCs in the context of homoeostatic or altered redox. This is an important question because DCs migrate within solid tumours, sites of infection, areas of severe inflammation and naturally oxygen-rich tissues as well as normal tissue microenvironments located far from the ends of capillary beds where oxygen is limited [17, 47]. How redox, oxygen tension and other environmental factors influence the tight control of IDO expression and enzymatic activity to affect immunological outcomes is incompletely known. The main finding of this study is that even in the settings of physiologic hyperoxia and hypoxia, DCs maintain the ability to sense changes in redox to control IDO expression.
We previously observed that blocking antiporter activity strongly induced IDO expression in DCs by inducing oxidative stress and altering cellular redox . This prompted us to address the possibility that oxygen tension could regulate the activity of the cystine/glutamate antiporter and in this way affect IDO expression and activity. Our data show that exposure to low oxygen levels significantly decreased the activity of the antiporter. These data are consistent with those of other studies showing that low oxygen concentration decreased antiporter activity in cultured human fibroblasts  and mouse peritoneal macrophages . Based on this observation, we anticipated that low oxygen tension, through its suppressive effects on the antiporter, would increase IDO expression in DCs. Our results, however, show that low oxygen tension did not affect IDO enzymatic activity or IDO protein levels, as both remained at baseline levels similar to that observed when DCs were cultured under 20% O2 in complete medium. Furthermore, the combination of low oxygen tension and altered redox failed to increase IDO protein levels and enzymatic activity in DCs beyond that observed in control cells (20% O2 and cystine/cysteine-free medium). Together, these results suggest that redox controls IDO expression irrespective of oxygen concentration. This may reflect a control mechanism to maintain self tolerance at sites of inflammation where local hypoxic regions are created by increased immune cell consumption of oxygen .
One possibility to explain the seemingly paradoxical finding that low oxygen tension reduced antiporter activity, but did not induce IDO expression, is that redox, but not oxygen tension, regulates transcription factors that control IDO gene expression. This is consistent with our previous observation that altered redox increases IDO mRNA . It is also possible that redox, but not oxygen tension, influences epigenetic control of IDO gene expression, although evidence to support a role for epigenetics in the control of IDO transcription is currently lacking. Another consideration is that the enzymatic activity of IDO is controlled by redox and oxygen tension. IDO is a haeme-containing dioxygenase that transitions from an inactive oxidized state (ferrous-IDO) to an active reduced form (ferric-IDO) . A major effect of interfering with antiporter activity is that the cytosol becomes less reducing due to decreased supply of cystine for reduction to cysteine to support glutathione synthesis. Thus, blocking antiporter activity is predicted to increase the abundance of the inactive IDO form and decrease the concentration of the inactive form. Whether the ~30% decrease in IDO enzymatic activity we observed when DCs were cultured in 1% O2 versus 20% O2 in cystine/cysteine-free medium (Fig. 2) relates to the effects of redox and oxygen tension on IDO binding to tryptophan and oxygen, and how this may impact IDO enzymatic activity remains to be examined. Finally, IDO activity is regulated post-translationally through nitration, N-terminal acetylation and phosphorylation [53, 61-63]. Based on our results, the potential effect of oxygen tension on IDO post-translational modification is unlikely to play a role, as oxygen tension did not regulate IDO protein levels or enzymatic activity.
Another finding of the study was that neither alteration of oxygen nor redox induced DC production of TGF-β and IL-23, two cytokines that collaborate to drive Th17 responses . Furthermore, although TGF-β has been shown to induce IDO production in DCs [9, 53], our data show that under the conditions examined, DCs do not regulate IDO expression via autocrine production of this cytokine. Interestingly, both redox and oxygen tension did regulate DC production of MCP-1, a pro-inflammatory cytokine that recruits innate immune cells, including DCs, to sites of inflammation . The effect of MCP-1 on DCs (and other cells) is highly context dependent. Depending upon the environment, MCP-1 recruits monocytes for differentiation into inflammatory DCs or, alternatively, attracts tolerogenic DCs; the latter is consistent with a protective role for MCP-1 in autoimmunity [66, 67]. Interestingly, when the effects of oxygen tension and redox on IDO enzymatic activity and MCP-1 production were compared, we found that DCs cultured under 20% O2 in cystine/cysteine-free medium, conditions supporting maximal IDO activity, produced the lowest concentration MCP-1. Conversely, DCs cultured under 1% O2 in complete medium, conditions supporting maximal MCP-1 production, correlated with the lowest level of IDO enzymatic activity. These data suggests that IDO and MCP-1 may exert their effects on opposite spectrums of immunity.
In summary, there is a growing appreciation that DCs, and other highly migratory immune cells, are exposed to wide variations in microenvironments characterized by imbalances in nutrients, redox and oxygen content. The results of our study show that redox, irrespective of oxygen tension, controls IDO expression and activity in DCs. In addition to widening our understanding of how tolerance is maintained, these data may in part explain how IDO-competent DCs arise in the settings of cancer, obesity, HIV, diabetes and aging, diseases characterized by both imbalances in redox and oxygen and immune dysregulation and decline [68-75].
We thank the following WVSOM students who contributed to this work: Evan S. Stern, Matthew J. Copeland and Zachary M. Grimes. This work was supported with funds from The Research Institute for Children and The West Virginia School of Osteopathic Medicine.
Conflict of interest
The authors declare that they have no conflict of interest.