TLR-Induced IL-12 and CCL2 Production by Myeloid Cells Is Dependent on Adenosine A₃ Receptor–Mediated Signaling

Toll-like receptors are evolutionary conserved pattern recognition receptors expressed by different cell lineages of the immune system (1). In myeloid cells, agonist recognition by TLRs initiates a cascade of intracellular signaling events that ultimately culminate in the activation of transcription factors such as NF-κB and AP-1 (2–4). These transcription factors induce the expression of inflammatory soluble mediators like TNF-α, IL-12, and CCL2, as well as of molecules involved in Ag presentation (1, 5).

The purine nucleoside adenosine is a known regulator of inflammatory responses (6, 7). During inflammatory conditions, intracellular and extracellular adenosine levels rapidly rise. Extracellular adenosine interacts with four G protein–coupled adenosine receptor (ADORA) subtypes. Adenosine A₁ and A₃ receptors (A₁R and A₃R) are coupled to Gi proteins and mediate opposite effects to adenosine A_2A and A_2B receptors (A_2AR and A_2BR) that are coupled to Gs proteins. The ADORA expression pattern, which is highly dynamic, therefore orchestrates the cellular response to adenosine (8, 9).

ADORA-mediated signaling also modulates innate immune responses induced by myeloid cells. A_2AR-mediated signaling inhibits TLR-induced NF-κB activation and concomitant production of cytokines in macrophages and microglia (10–13). Recently, our group described that A₃R-mediated signaling counteracts A_2AR-mediated inhibitory signaling (14). Data from the same study suggested that A₃R-mediated signaling was also directly involved in TLR-induced cytokine responses.

In this study, we show that A₃R-mediated signaling contributes to IL-12 and CCL2 production as induced by a broad range of TLRs and pathogens and has downstream effects on T cell responses to whole pathogens. Dendritic cell (DC)–CD4⁺ T cell coincubation experiments demonstrate that bacterial and fungal-induced IFN-γ and IL-17 responses are significantly inhibited when they were evoked in the presence of an A₃R antagonist. Next-generation sequencing of mRNA expression profiles revealed that A₃R-mediated signaling controls the expression levels of methallothioneins, which in turn can inhibit the phosphorylation of the transcription factor STAT1. These results demonstrate that endogenously produced adenosine provokes A₃R-dependent signaling, which essentially contributes to the TLR-induced cytokine response profile.

Adenosine A₁R antagonist 8-cyclopentyl-1,3-dipropylxanthine (DPCPX), adenosine A_2AR antagonist 7-(2-phenylethyl)-5-amino-2-(2-furyl)-pyrazolo-[4,3-e]-1,2,4-triazolo[1,5-c]pyrimidine (SCH58261), adenosine A_2BR antagonist 8-(4-[([4-cyanophenyl]carbamoylmethyl)oxy]phenyl)-1,3-di(n-propyl)xanthine hydrate (MRS1754), adenosine A₃R antagonists 9-chloro-2-(2-furanyl)-5-([phenylacetyl]amino)-[1,2,4]triazolo[1,5-c]quinazoline (MRS1220), and 3-ethyl-5-benzyl-2-methyl-4-phenylethynyl-6-phenyl-1,4-(±)-dihydropyridine-3,5-dicarboxylate (MRS1191; all Sigma-Aldrich, Saint Louis, MO) were diluted in DMSO, aliquoted, and stored at −20°C. The 4-methoxy-N-(3-[2-pyridinyl]-1-isoquinolinyl)benzamide (VUF8504) was synthesized in-house. DMSO controls were included in all functional assays.

TLR agonists used were Pam₃CSK₄ (TLR2), poly(I:C) (TLR3), LPS (TLR4), flagellin (TLR5), and CL075 (TLR8; all Invivogen, San Diego, CA). Heat-killed Candida albicans and heat-killed Staphylococcus aureus were obtained from Invivogen; Haemophilus influenzae strain 86-028NP [nontypeable (15)] was cultured in brain heart infusion medium at 37°C, washed, and heat-killed for 30 min at 56°C. Recombinant human IFN-γ was from Peprotech (London, U.K.), and PMA was from Sigma-Aldrich.

Human PBMCs were isolated from human buffy coats obtained from anonymized healthy donors (Sanquin, the Netherlands) using lymphocyte separation medium gradient centrifugation (Lymphoprep; Axis-Shield, Norway). Monocytes were isolated with anti-CD14, and T cells with anti-CD4 mAb-coated Microbeads using MACS single-use separation columns from Miltenyi Biotec (Bergisch Gladbach, Germany) as described by the manufacturer. CD4⁺ T cells were resuspended in 10% DMSO (v/v) with heat-inactivated FCS and frozen in liquid nitrogen until further use.

Purified CD14⁺ cells were resuspended in RPMI 1640 (Life Technologies, Thermo Fisher Scientific, Waltham, MA) containing 10% (v/v) FCS (Life Technologies) and penicillin 100 U/ml and streptomycin 0.1 mg/ml (Life Technologies) supplemented with ≥4 U recombinant human M-CSF/ml or ≥40 U recombinant human GM-CSF and 200 ng of recombinant human IL-4/ml (all Peprotech) to yield macrophages or DC, respectively. Half of the medium was replaced by fresh medium containing new growth factors every 3–4 d. After 6–7 d in culture, cells were used for functional assays. For the Ag presentation assays, donor-matched CD4⁺ T cells were thawed and added to DC in a ratio of 5:1 together with either VUF8504 or vehicle control. Heat-killed Candida albicans, heat-killed Staphylococcus aureus, or heat-killed Haemophilus influenzae strain 86-028NP (1.25 × 10⁶ CFU/well) or an equivalent volume of medium was added subsequently. Following stimulation, supernatant was isolated 24 h and 5 d later and stored at −20°C until further analysis.

THP-1 cells were cultured in RPMI 1640 supplemented with 10% (v/v) FCS, 2 mM GlutaMAX, 1 mM nonessential amino acids, 1 mM sodium pyruvate, 100 U penicillin/ml, and 0.1 mg streptavidin/ml (all Life Technologies).

THP-1 cells were seeded in a 96-well plate at a concentration of 1 × 10⁴ cells per well and incubated for 24 h before lentiviral short hairpin RNA (shRNA) transduction particles (SHCLNV MISSION pLKO.1-puro nonmammalian shRNA; Sigma-Aldrich) were added to the cells at different multiplicities of infection in the presence of 80 μg polybrene/ml for overnight incubation. Medium containing lentiviral particles was replaced by fresh cell culture medium and incubated for 48 h. At day 3, fresh medium containing 5 μg puromycin/ml (Invivogen) was added, and medium was refreshed 1:1 v/v every 3–4 d. After outgrowth, clones were selected and characterized.

The Ultra Directional RNA Library Prep Kit for Illumina (New England Biolabs, Ipswich, MA) was used to prepare and process the samples. Briefly, mRNA was isolated from total RNA using oligo(dT) magnetic beads. After fragmentation of the mRNA, cDNA synthesis was performed followed by ligation of sequencing adapters and PCR amplification. The quality and yield after sample preparation was measured with a fragment analyzer (Advanced Analytical, Heidelberg, Germany). The size of the resulting products was consistent with the expected size distribution (a broad peak between 300 and 500 bp, data not shown). Clustering and DNA sequencing using the Illumina NextSeq 500 was performed according to the manufacturer’s protocols. A total of 1.6 pM of DNA was loaded and analyzed using NextSeq control software 2.0.2. Image analysis, base calling, and quality verification were performed with Illumina data analysis pipeline RTA v2.4.11 and Bcl2fastq v2.17. The reads were mapped to reference sequence Homo_sapiens.GCRh37.75 using a short-read aligner based on Burrows-Wheeler transform (mismatch rate of maximum 2%). For the 16 samples, the number of reads obtained was between 22,537,666 and 32,979,730. The % reads of reads aligned to the reference was between 95.4 and 95.7%. Reads that aligned to multiple locations were between 7.5 and 9.4%. Reads that aligned to genes were between 60.0 and 87.8%. SAMtools 0.1.18 was used to sort and index the BAM files. The number of times a read aligned to a gene and normalized read counts in fragments per kb million were determined using in-house–developed scripts. The R package DESeq 1.10.1 was used to calculate false discovery rate–corrected p values for the genes differentially expressed in the tested samples.

Total cellular RNA was isolated using TriReagent (Sigma-Aldrich) according to the manufacturer’s protocol. Subsequently, mRNA was reverse transcribed into cDNA using the Fermentas kit according to the manufacturer’s protocol (Qiagen Benelux, Venlo, the Netherlands) using 1 μg of mRNA as template and 0.25 μg of oligo(dT)15 primers supplied with the kit. RT-PCRs were performed on the CFX96 Thermal Cycler (Bio-Rad, Hercules, CA) using primer (Life Technologies, Thermo Fisher Scientific) and probe (human Exiqon ProbeLibrary, Roche, Woerden, the Netherlands) combinations listed in Table I. Gene of interest mRNA expression levels were standardized to GAPDH expression levels using the Pfaffl method (16).

IL-12p40 and TNF-α levels were determined by ELISA according to the manufacturer’s protocol (U-Cytech, Utrecht, the Netherlands). IL-17 and IFN-γ levels were determined by ELISA kits (IL-17; R&D Systems, IFN-γ; Sanquin, the Netherlands) according to the manufacturer’s protocol. Multiplex assays were performed using a customized Non-Human Primate MILLIPLEX MAP kit (Millipore, Billerica, MA). CCL2, IL-12, IL-6, IL-8, MIP1α, MIP1β, and TNF-α levels were determined according to the manufacturer’s protocol and analyzed on the Luminex 200 System (Bio-Rad).

NF-κBp65, p–NF-κBp65 (Ser536), p–stress-activated protein kinase (SAPK)/JNK (Thr183/Tyr185), p-p38 (Thr180/Tyr182), and p-ERK1/2 (Thr202/Tyr204) levels were measured in cell lysates using PathScan ELISA Kit (Cell Signaling Technology, Beverly, MA). Cell lysates were generated according to protocol in lysis buffer provided with the kit, supplemented with protease inhibitors (Roche). Samples were stored at −80°C before use and diluted 1:1 in dilution buffer provided with the kit before incubation for 16 h at 4°C. All following steps were performed according to the manufacturer’s protocol. Nonphosphorylated NF-κBp65 levels were used as reference values for each sample.

NF-κBp65, AP-1, and ATF2 DNA binding activity was analyzed by TransAM ELISA according to the manufacturer’s protocol (Active Motif). Bound transcription factors were visualized by addition of a primary Ab directed against NF-κBp65, AP-1, or ATF2 elements, followed by detection with a HRP-conjugated secondary Ab. Absorbances were read at 450 nm with a reference wavelength of 655 nm. For the NF-κB and AP-1 assays, a protein standard curve was performed using recombinant NF-κBp65 or c-Jun recombinant protein to quantitate transcription factor levels in the protein samples.

Cells were lysed in lysis buffer (Cell Signaling) supplemented with protease inhibitors (Roche), and protein concentrations were determined using the Bradford Protein Assay (Pierce, Rockford, IL) according to the manufacturer’s protocol. Samples were stored at −20°C before use. Proteins (2.5 mg/ml per lane) were separated on 4–12% Bis-Tris gels (Invitrogen) and transferred onto Hybond-ECL nitrocellulose membranes (Amersham, Arlington Heights, IL) via a semidry blotting system (Thermo Fisher Scientific). The membranes were blocked and probed with primary Abs directed against STAT1 phosphorylated at tyrosine 701 (clone: D4A7), STAT1 phosphorylated at serine 727 (polyclonal), or total STAT1 (clone: 42H3; all Cell Signaling), followed by anti-rabbit–IgG-HRP AffiniPure Ab (Jackson ImmunoResearch Laboratories, Suffolk, U.K.) developed using Femto ECL substrate (Pierce) and quantitated with the ChemiDoc MP system (Bio-Rad).

GraphPad Prism 4.0b (GraphPad Software, San Diego, CA) was used for statistical analysis. Principal component and redundancy analysis was performed using Canoco 5.02 (Microcomputer Power, Ithaca, NY) at default settings.

Recently, we published on the complex interplay between A_2AR and A₃R as modulators of innate immune responses (14). Our data also suggested that A₃R-mediated signaling was directly involved in TLR-induced IL-12 production by myeloid cells. To investigate this, human monocytes, monocyte-derived macrophages, and monocyte-derived DC were exposed to the TLR4 agonist LPS in the absence or presence of specific ADORA antagonists. The presence of ADORA antagonists alone did not induce measurable levels of IL-12p40 in any of the cell types, whereas exposure to LPS induced the production of variable amounts of IL-12p40 in different cell types (Supplemental Table I). The presence of an A₃R antagonist significantly inhibited LPS-induced IL-12p40 production in monocytes, macrophages, and DC, whereas antagonizing A₁R, A_2AR, or A_2BR did not significantly affect the production of IL-12p40 in any of the cell types (Fig. 1A). An important role for A₃R-mediated signaling in LPS-induced IL-12p40 was confirmed by using combinations of different antagonists (Fig. 1A).

To validate that the observed effects were attributable to A₃R-mediated signaling, DC were stimulated with LPS in the presence of different concentrations of three structurally unrelated A₃R antagonists (Table II). All A₃R antagonists significantly inhibited LPS-induced IL-12p40 production (Fig. 1B). Finally, we silenced A₃R expression in THP-1 cells (THP-1 cells transduced with A₃R-targeting shRNA [THP-1^A3]) with lentiviral shRNA and measured LPS-induced IL-12p40 production. A₃R knockdown was >90% as confirmed by quantitative RT-PCR (Supplemental Fig. 1). LPS-induced IL-12p40 levels in THP-1^A3 cells were decreased compared with those in THP-1 cells or in THP-1 cells transduced with nontargeting control (ntc) shRNA (THP-1^ntc; Fig. 1C). In contrast to THP-1 and THP-1^ntc cells, exposure of THP-1^A3 cells to an A₃R antagonist had no effects on LPS-induced IL-12p40 production (Fig. 1C), confirming the specific involvement of A₃R-mediated signaling.

We next characterized the involvement of A₃R-mediated signaling on IL-12p40 production as induced by other TLRs that are expressed by DC. Engagement of TLR2, 3, 4, 5, and 8 on DC from four donors induced the production of variable levels of IL-12p40 (Table III). Exposure of DC to TLR ligands in the presence of an A₃R antagonist significantly reduced TLR2, TLR3, TLR4, TLR5, and TLR8-induced IL-12p40 levels (Fig. 2A), demonstrating that A₃R-mediated signaling is broadly involved in TLR-induced IL-12p40 production. It should, however, be noted that although the qualitative aspect of the contribution of A₃R-mediated signaling to TLR responses is clear, effect sizes vary considerably between different TLR as well as between different donors.

To expand our analyses to other TLR-induced soluble factors, we choose to evaluate the effects of A₃R-mediated signaling on the robust cytokine responses induced by TLR4 and TLR8 ligands on DC from three different donors using Luminex technology. Exposure to TLR4 and TLR8 ligands triggered the production of multiple soluble factors (Fig. 2B). When the inhibition of A₃R-mediated signaling was analyzed against normalized TLR-induced cytokine production from different donors, only IL-12p40 and CCL2 were significantly affected (Fig. 2B, right panels, Table IV). Exposure of DC to TLR2, -3, and -5 ligands in the presence of an A₃R antagonist also reduced CCL2 levels (data not shown), demonstrating that the effects were not limited to TLR4- and TLR8-induced responses.

Subsequently, we investigated the contribution of A₃R-mediated signaling to responses induced by whole pathogens that contain multiple ligands for innate immune receptors. We exposed monocytes, monocyte-derived macrophages, and monocyte-derived DC of four different donors to the fungus C. albicans as well as to Gram-negative H. influenzae and Gram-positive S. aureus bacteria in the absence or presence of an A₃R antagonist and measured IL-12p40 production. Monocytes produced relatively low levels (Supplemental Table II) of IL-12p40 in response to C. albicans and even lower levels when A₃R-mediated signaling was antagonized (Fig. 2C). Although exposure to H. influenzae potently triggered the production of IL-12p40 in all four donors, and S. aureus triggered detectable production of IL-12p40 in three donors, these responses were not significantly affected when A₃R-mediated signaling was inhibited. Macrophages of all four donors produced detectable levels of IL-12p40 in response to H. influenzae only (Supplemental Table II), which were significantly lower when A₃R-mediated signaling was inhibited. In DC, C. albicans and H. influenzae potently induced IL-12p40 production, whereas exposure to S. aureus did not induce detectable levels of IL-12p40 (Supplemental Table II). Both C. albicans– and H. influenzae–induced IL-12p40 responses were significantly inhibited when A₃R-signaling was antagonized (Fig. 2C).

The IL-12p40 subunit can heterodimerize with p35 or p19 to form bioactive IL-12 or IL-23 respectively. IL-12 is an important factor in the polarization of T cells to Th1 cells and the consequent production of IFN-γ, whereas IL-23 is involved in the polarization to Th17 cells and the consequent production of IL-17 (17). To examine whether blocking A₃R-mediated signaling could affect downstream T cell cytokine production, DC from five different donors were exposed to C. albicans and H. influenzae in the presence or absence of A₃R antagonist and coincubated with autologous CD4⁺ T cells for 5 d. Analysis of cell culture supernatants reveals that exposure to H. influenzae potently induced IFN-γ, whereas exposure to C. albicans mainly induced the production of IL-17 (Fig. 2D). There was considerable donor-donor variation, and it is interesting to note that the two donors that were poor IFN-γ producers were the ones that produced the highest levels of IL-17. When A₃R signaling was antagonized, both IFN-γ and IL-17 responses were significantly lower with larger effect sizes on IL-17 responses (Fig. 2D). Future experiments will further focus on the implications of these differences.

A₃R-mediated signaling affected IL-12 and CCL2 production at the transcription level, as LPS-induced IL-12p40– and CCL2-encoding mRNA levels in DC in the presence of an A₃R antagonist were 5- and 39-fold lower, respectively (Fig. 3A, Table I). In addition, LPS-induced mRNA levels of IL-12p35 and IL-23p19 subunits were both ∼5-fold lower when A₃R-mediated signaling was antagonized. Interestingly, exposure of cells to A₃R antagonist alone already decreased basal mRNA expression levels of the IL-12p40 (to a mean of 23%) and p35 (to a mean of 65%) subunits as well as of CCL2 (to a mean of 14%), suggesting a role for homeostatic A₃R-mediated signaling in DC. To delineate how A₃R-mediated signaling affects TLR-induced intracellular signaling cascades, we first assessed LPS-induced phosphorylation levels of the NF-κB at serine 536 by PathScan ELISA, an important TLR-induced modulator of transcriptional activity (18). This was not affected by antagonizing A₃R-mediated signaling, and in accordance, NF-κB binding to its consensus site as assessed by TransAM ELISA was similar in nuclear extracts of DC exposed to LPS both in the absence or presence of an A₃R antagonist (Fig. 3B, 3C). In addition to activation of NF-κB (19), TLR-induced signaling activates several MAPKs. We therefore used PathScan ELISA to evaluate phosphorylation levels of SAPK/JNK, ERK1/2, and p38 in cellular lysates of DC exposed to LPS in the absence or presence of an A₃R antagonist. Whereas LPS-induced levels of p-ERK1/2 and p-p38 were unaffected, LPS-induced p-SAPK/JNK levels were decreased at early time points in the presence of an A₃R antagonist (Fig. 3D). SAPK/JNK induces activation of the transcription factor AP-1. However, LPS-induced c-Jun, c-Fos, and ATF2 binding levels were not significantly different in nuclear extracts of DC exposed to LPS in the presence of an A₃R antagonist (Fig. 3E), rendering it unlikely that differences in SAPK/JNK phosphorylation underlie the observed effects on IL-12 and CCL2 production.

Next to binding sites for NF-κB and AP-1 elements, the promoter regions of IL-12p40 and CCL2 also share binding sites for STAT1. Full activation of STAT1 requires phosphorylation at serine 727 and tyrosine 701. Whereas LPS-induced phosphorylation of STAT1 at S727 was unaffected in the presence of an A₃R antagonist, LPS-induced phosphorylation of STAT1 at Y701 was strongly decreased after 60 min of stimulation (Fig. 3F). To confirm the involvement of STAT1 tyrosine 701 phosphorylation, we tested whether the reduced levels of LPS-induced IL-12p40 in the presence of an A₃R antagonist could be rescued by the addition of IFN-γ, a well-known inducer of STAT1 tyrosine 701 phosphorylation (20). Confirming our earlier results, IFN-γ partially restored LPS-induced IL-12p40 levels in the presence of an A₃R antagonist (Fig. 3G).

Finally, we submitted mRNA samples of DC exposed to LPS in the presence or absence of an A₃R antagonist to total transcriptome analysis by RNA sequencing (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE125747). Principal component analysis of the samples (Fig. 4A) demonstrates a clear separation of the different sample classes according to stimulation conditions, which was confirmed by redundancy analysis (p < 0.002; data not shown). Venn diagrams illustrate the overlap between different sample groups (Fig. 4B). Differential gene expression analysis demonstrates that exposure to LPS significantly (adjusted p < 0.01) altered the expression levels of 4,893 gene transcripts (GSE125747), whereas exposure to an A₃R antagonist altered the expression levels of 381 gene transcripts only (Table V). Expression levels of 146 transcripts were significantly different between cells exposed to LPS alone or in combination with an A₃R antagonist (GSE125747). Analysis of the individual transcripts reveals that exposure to an A₃R antagonist, even in the absence of LPS activation, leads to a strong increase of metallothionein mRNA levels (the top 13 genes with the lowest p values are all metallothionein genes). These findings were confirmed for four metallothionein isoforms by quantitative RT-PCR (Fig. 4C, Table I). In addition, the data confirm that LPS-induced CCL2 and IL-12 (NS) mRNA levels were lower in the presence of an A₃R antagonist (GSE125747).

In this study, we demonstrate that A₃R-mediated signaling is critically involved in LPS-induced IL-12 production by myeloid cells. Further analysis of DC demonstrates that also CCL2 production as induced by a broad range of TLR ligands or whole pathogens was affected. A₃R-mediated signaling suppresses metallothionein expression levels and amplifies TLR-induced phosphorylation of the transcription factor STAT1 at tyrosine 701, demonstrating a complex and novel layer of regulation of innate immune responses. To our knowledge, we are the first to describe a role for endogenously produced extracellular adenosine and A₃R-mediated signaling in the initiation of TLR-induced cytokine responses.

Extracellular adenosine can interact with four different ADORA subtypes. Interestingly, A₃R expression levels are rapidly downregulated after TLR-mediated activation of myeloid cells, whereas expression levels of A_2AR are strongly enhanced (14). Previously, we have hypothesized that this shift in ADORA expression pattern underlies the inhibitory effects of extracellular adenosine on TLR-mediated cytokine production by unleashing the full capacity of A_2AR-mediated anti-inflammatory signaling. This study suggests that downregulation of A₃R also has direct anti-inflammatory effects by depriving TLR-activated myeloid cells of their capacity to produce IL-12 and CCL2. We did not find evidence for direct effects on Ag presentation capacity, as the LPS-induced increase of cell surface expression levels of receptors involved in Ag presentation by DC was unaffected by exposure to an A₃R antagonist (Supplemental Fig. 2).

Our transcriptome data revealed that A₃R-mediated effects not only modulated TLR-induced signaling but also affected otherwise unstimulated cells. Most striking was the strong induction of metallothioneins when homeostatic A₃R-mediated signaling was inhibited. Metallothioneins are low m.w. cysteine-rich proteins that regulate intracellular zinc homeostasis, detoxify heavy metals, and counteract superoxide stress (21). Their expression is potently induced by metal exposure and oxidative stress but also by cytokine signaling and microbial challenge (22–25). In myeloid cells, there is ample evidence for an important role of metallothioneins in the maintenance of intracellular zinc homoeostasis (26–28). Manipulation of intracellular and extracellular zinc levels can be used to modulate the phenotype of myeloid cells (29–31), and different studies have identified a role for zinc signaling in kinase and phosphatase functions (32, 33). Interestingly, a recent study demonstrated that metallothioneins inhibit the phosphorylation of STAT1 and STAT3 (34). It has been postulated that metallothionein 1 and 2 sequester intracellular zinc, thereby sustaining protein tyrosine phosphatase 1B activity that in turn dephosphorylates STATs (32, 33).

The promoter regions of IL-12p40 and CCL2 share binding sites for STAT1, and we have identified that TLR-induced phosphorylation of STAT1 at tyrosine 701 was strongly reduced when A₃R-mediated signaling was inhibited. More pathways may be involved, as IFN-γ–induced phosphorylation of STAT1 tyrosine 701 only partially restored TLR-induced IL-12p40 levels in the presence of an A₃R antagonist (Fig. 3G). Although the complete elucidation of intracellular signaling cascades remains to be established, the involvement of the STAT1 pathway is evident. Together, this leads us to propose a novel molecular mechanism that links adenosine-induced signaling via the metallothionein–zinc axis to modulation of innate and adaptive immune responses.

Extracellular adenosine is best known as an anti-inflammatory modulator, and A₃R-mediated signaling has previously been associated with the inhibition rather than with the initiation of LPS-induced IL-12 and/or TNF-α in different murine models and cell lines, as well as in human monocytes and U937 cells (35–40). Several factors may explain the discrepancies with our findings. First, the described inhibitory effects of A₃R-mediated signaling have primarily been observed in murine models and/or cells. Whereas the sequences of A₁R, A_2AR, and A_2BR are well conserved across species, A₃R sequences differ considerably. Sequence comparisons between rodent and human show 70–75% gene sequence homology and 85–87% protein sequence homology (6, 14). Second, there is evidence that G protein coupling of A₃R is different between rodents and humans. In A₃R-humanized mice, bone marrow–derived mast cells were unable to initiate conventional murine A₃R-mediated signaling pathways. This was attributed to sequence differences in the intracellular region of human A₃R, resulting in the uncoupling of murine G proteins (41). Third, discrepancies with observations in U937 cells might be attributable to the differentiation/transformed status of these cells.

In conclusion, we describe a new mechanism that is important for TLR-induced IL-12 and CCL2 responses. Because all experiments were performed in the absence of exogenous adenosine or other ADORA agonists, the observed effects were attributable to endogenously produced extracellular adenosine contributing to TLR-induced responses in an autocrine or paracrine manner. Interestingly, this mechanism bears resemblance to the mechanism regulating the production of inflammasome-induced IL-1β in which endogenous ATP acts as a similar signal through P2X7 receptors (42). Our data suggest that antagonizing A₃R signaling could selectively affect TLR-induced production of IL-12 and CCL2, thereby representing a novel immune modulatory strategy. However, given the widespread distribution of A₃R throughout the body (43), specific delivery to the site of inflammation represents an important challenge when considering such an approach.

We thank E. Remarque for expert assistance with the statistical analyses, S.B. Geutskens, M.E. Hoonakker, and B. ‘t Hart for critically reading the manuscript.

The authors have no financial conflicts of interest.