S1PR4 Signaling Attenuates ILT 7 Internalization To Limit IFN-α Production by Human Plasmacytoid Dendritic Cells Free

Plasmacytoid dendritic cells (pDCs) are a rare DC subpopulation, characterized by their potent ability to produce large amounts of type I IFN (IFN-α/β; IFN-I), a plasma cell-like morphology, and a unique set of cell-surface markers. IFN-α plays a crucial role in antiviral immunity by augmenting the expression of a broad repertoire of antiviral molecules that dampen viral spread and promote apoptosis of infected cells (reviewed in Ref. 1). Furthermore, IFN-α was shown to inhibit the growth of cells undergoing malignant transformation (reviewed in Ref. 2). In pDCs, IFN-α is produced mainly via the TLR7/9–MyD88 pathway and the rapid production of high levels IFN-α is due to the constitutive expression of the transcription factor IFN regulatory factor 7 (IRF7) (3). TLR7 and -9 sense viral ssRNA and unmethylated CpG oligodeoxynucleotides (ODNs), respectively, and are located within endosomes. The subcellular localization of TLR7/9 prevents the host to respond to self-DNA, which might induce autoreactive immune responses. The uptake of physiological TLR7/9 ligands and their shuttling to the endosomal compartment is a controlled, receptor-mediated process. However, detailed mechanisms of TLR7/9 ligand uptake are currently unclear and require further investigation (4).

Controlling pDC activation is indispensable to avoid an overshooting immune response that can harm the host. Therefore, pDCs express a set of regulatory surface receptors that regulate IFN-α production. Among these, Ig-like transcript 7 (ILT7), a surface marker exclusively expressed on human pDCs, was shown to associate with the adapter protein FcεRIγ. Activation of this receptor complex causes ITAM-mediated signaling to restrict IFN-α production (5–7). The only known ligand for ILT7 is bone marrow stromal Ag 2 (BST2), which is expressed on immune cells such as pDCs, but also on cancer cells, including melanoma and breast cancer (8). The current model of ILT7/BST2 interaction suggests that BST2 expression is upregulated in an inflammatory environment to interact with ILT7, serving as a negative-feedback loop to prevent an uncontrolled and overshooting IFN-α response. In contrast, constitutive expression of BST2 by various cancer types inhibits IFN-α production of tumor-infiltrating pDCs. As IFN-α has antiangiogenic and proapoptotic effects on cancer cells, this mechanism allows tumor-associated immunosuppression. However, leukemic pDCs tend to downregulate ILT7 although the underlying reason is unclear (9). pDCs were shown to infiltrate human solid tumors without secreting adequate levels of IFN-α after TLR9 activation (10). Tumors orchestrate an immunosuppressive microenvironment, which is characterized by specific mediators (e.g., sphingosine-1-phosphate [S1P], PGE₂, IL-10, vascular endothelial growth factor, and TGF-β), released from tumor-associated immune or cancer cells, favoring regulatory immune cell phenotypes (reviewed in Ref. 11). Furthermore, pDCs, like all DCs, are able to present Ags to T cells and therefore induce, depending on the microenvironment, variable adaptive immune responses. There is evidence that pDCs promote regulatory T cell (Treg) expansion within the tumor by presentation of self-Ags to naive T cells. However, other studies show that fully activated pDCs can have a cytotoxic, tumoricidal function (12).

S1P is a bioactive sphingolipid mediator that is produced by various (immune) cell types (e.g., RBCs, platelets, macrophages, and DCs) (reviewed in Ref. 13). S1P acts as a specific ligand for five receptors (S1PR1–5), which differ in tissue distribution. For instance, S1PR1 is ubiquitously expressed, whereas S1PR4 is largely restricted to lymphoid tissue (14). Due to the fact that S1PRs couple to diverse G-proteins, S1P is able to affect multiple cellular and physiological functions (e.g., immune cell trafficking, angiogenesis, vascular maturation, cardiac development, and immune cell activation) (reviewed in Ref. 15). Currently, S1PR4 signaling is ill defined. However, its activation might alter immune cell phenotypes and activation (16–19).

In the current study, we investigated the influence of S1P on pDC activation and function. Administration of S1P decreases IFN-α production via S1PR4. Mechanistically, S1PR4 signaling preserves ILT7 surface expression that normally decreases upon pDC activation. This enables ILT7 to bind to its ligand and to decrease IFN-α production. As a consequence of reduced IFN-α production, S1PR4 signaling in pDCs shifts cytokine production of cocultured T cells from a Th1 (IFN-γ) to a regulatory (IL-10) profile. S1PR4 signaling inhibits human pDC activation and might therefore be a promising target to restrict pathogenic IFN-α production in several IFN-I–induced autoimmune diseases or to increase IFN-I production in tumors.

PBMCs were obtained from Buffy Coats (DRK-Blutspendedienst Baden-Württemberg-Hessen, Frankfurt, Germany) using Ficoll–Isopaque (PAA Laboratories, Cölbe, Germany) gradient centrifugation; RBCs were removed by KCl lysis. PBMCs were cultured in six-well plates in RPMI 1640 (PAA Laboratories) containing 2% FCS. Primary pDCs were magnetically purified from PBMCs using a pDC purification kit (negative selection), according to the manufacturer’s instructions, and the autoMACS Separator (Miltenyi Biotec, Bergisch Gladbach, Germany). pDCs were characterized as blood DC Ag 4⁺, CD123⁺, MHC class II (MHC II)^int, and CD11c⁻. The purity was ≥93%. Cells were cultured in 96-well plates in X-Vivo (Lonza, Verviers, Belgium) supplemented with 2.5% human serum containing 20 ng/ml IL-3 (Immunotools, Friesoythe, Germany). T cells were isolated from PBMCs using the Pan T cell isolation kit, according to the manufacturer’s instructions, and the autoMACS Separator (Miltenyi Biotec) and cultured in RPMI 1640 (PAA Laboratories) supplemented with 10% heat-inactivated FCS, 1% nonessential amino acids, 1% essential amino acids, 1% sodium pyruvate, 1% HEPES buffer (all from PAA Laboratories), 2-ME, and 10 ng/ml IL-2 (Immunotools).

S1P and VPC23019 (both 1 μmol; Avanti Polar Lipids, Alabaster, AL) were dissolved following the manufacturers’ instructions. JTE-013 (100 nmol; Biomol, Hamburg, Germany), Cym50358 (200 nmol), and Cym50138 (200 nmol) were dissolved in DMSO. CpG-A (2336) ODN (5 μg/ml), CpG-A (2336) FITC, CpG-B (2006) ODN (1 μmol), imiquimod (IMQ; 2.5 μg/ml) (all from Invivogen, San Diego, CA), 10% v/v FSME-Immun (Baxter, Unterschleißheim, Germany), Lyn peptide inhibitor (LPI; 10 μmol), RhoA inhibitor Rhosin (10 μmol), and Rho-associated protein kinase (ROCK) inhibitor Y-27632 (10 μmol) (Tocris Bioscience, Bristol, U.K.) were dissolved following the manufacturers’ instructions.

RNA from PBMCs was isolated using peqGold RNA Pure (Peqlab Biotechnologie, Erlangen, Germany) followed by cDNA transcription with the iScript cDNA synthesis kit (Bio-Rad, München, Germany). pDC RNA (≤1 × 10⁶ cells/sample) was isolated using the RNeasy Micro kit (Qiagen, Hilden, Germany) and quantitated using the Bioanalyzer (Agilent Technologies, Böblingen, Germany) followed by transcription with sensiscript RT kits (Qiagen). Real-time quantitative PCR (qPCR) was performed using the MyIQ real-time PCR system (Bio-Rad) and Absolute Blue qPCR SYBR Green fluorescein mix (Thermo Scientific, Karlsruhe, Germany). Primers against all S1P receptors (S1PR1–5), IFN-α2, and IRF7 were from Qiagen. Additional primers (Biomers, Ulm, Germany) were: Mx1 sense, 5′-CACCGTGACACTGGGATTC-3′ and antisense, 5′-ATAGCGAGGAGGTGCTGAAG-3′; and dsRNA-activated protein kinase sense, 5′-CTAATTTGGCTGCGGCATT-3′ and antisense, 5′-CCGTCAGAAGCAGGGAGTAG-3′. Actin expression was used for normalization. RT-qPCR results were quantified using Gene Expression Macro (Bio-Rad).

S1PR4 silencing in primary human pDCs was performed using the Accell siRNA system (Dharmacon, Lafayette, CO) essentially following the manufacturer’s instructions. Briefly, 1 × 10⁵ pDCs were incubated with 1 μmol pooled small interfering RNA (siRNA; S1PR4 or nontargeting) in 100 μl serum-free medium containing IL-3 for 48 h, followed by addition of 100 μl medium containing human serum for another 24 h.

CCL5, CXCL10, IL-2, IL-6, IL-8, IL-10, IFN-α, INF-γ, or TNF-α in cell-culture supernatants were quantified using a cytometric bead array (CBA) Th1/Th2/Th17 Kit or CBA Flex Sets (BD Biosciences, Heidelberg, Germany). Samples were acquired by flow cytometry and processed with FCAP software V1.0.1 (BD Biosciences).

For quantification of PGE₂ levels in pDC supernatants, a commercial ELISA (Biomol) was used as described (20).

For detection of surface markers, cells were harvested and washed in PBS, followed by incubation with Fc-blocking reagent (Miltenyi Biotec) in 0.5% BSA/PBS for 15 min and Ab staining for 30 min on ice. The following Abs were used: anti-CD123–allophycocyanin or anti-CD123–PE-Cy5, anti- blood DC Ag 4–allophycocyanin, anti-ILT7–biotin (all from Miltenyi Biotec); anti–MHC II–PE-Cy7, anti-ILT7–PE, anti-CD86–FITC, streptavidin-PE–CF594 (all from BD Biosciences); anti-CD11c–FITC (Immunotools), anti-CD40–FITC, anti-CD80–allophycocyanin, anti-CD83–PE, anti-OX40L–Alexa Fluor647 (all from BioLegend, San Diego, CA); and anti-CD3–eFluor 605NC (eBioscience, San Diego, CA), or anti-TLR9–PE Ab (clone 26C593.2; Novus Biologicals, Wiesbaden, Germany). To detect intracellular protein levels, pDCs were resuspended in Fix/Perm solution (BD Biosciences) and incubated for 20 min on ice. Cells were washed twice in Perm/Wash solution (BD Biosciences), followed by anti-S1PR4–FITC (Biozol, Eching, Germany), anti–IFN-α–PE (BD Biosciences), or anti-granzyme B (GrB)–PE Ab (Immunotools) staining for 30 min on ice. To analyze accumulation of pDCs in cell clusters, PBMCs were stimulated as indicated and immediately fixed in paraformaldehyde (1% in PBS), followed by Ab staining. Samples were analyzed on an LSR II/Fortessa flow cytometer (BD Biosciences).

Human primary pDCs were isolated as described above and plated in 96-well plates with IL-3 (20 ng/ml) overnight. pDCs were treated with FITC-labeled CpG-A (10 μg/ml) for 5 min and prestimulated with an S1PR4 agonist (Cym50138; 200 nmol) for 30 min. Cells were then harvested and centrifuged on glass slides using cytospin. pDCs were fixed with 1% paraformaldehyde for 30 min at room temperature followed by permeabilization in 2% Triton X-100/PBS for 10 min. For analyzing ILT7 internalization, biotinylated ILT7 Abs (Miltenyi Biotec) and streptavidin-PE–CF594 (BD Biosciences) were added for 1 h each. Nuclei were counterstained with DAPI (1 μg/ml) for 20 min at room temperature. Cover slips were mounted on microscopy slides using Vectashield H 1400 mounting medium (Vector Laboratories, Burlingame, CA) and analyzed with the AxioVert 200M fluorescence microscope and the AxioVision software (all from Carl Zeiss Microimaging, Jena, Germany).

For direct coculture assays, T cells and pDCs were isolated from human blood of the same donor. A total of 5 × 10⁵ T cells/ml was seeded in a 96-well plate and cultured with IL-2 (10 ng/ml) for 24 h. A total of 1 × 10⁵ pDCs/ml was seeded in a 96-well plate and cultured with IL-3 (20 ng/ml) and incubated with tetanus toxoid (TTX; 2 μg/ml) for 24 h. To investigate the influence of pDC S1PR4 signaling on T cell proliferation and cytokine production, pDCs were stimulated with Cym50138 (200 nml) for 30 min. Afterwards, pDCs were washed with PBS and resuspended in T cell medium containing IL-3 (20 ng/ml) and IL-2 (10 ng/ml) with or without the addition of CpG-A (5 μg/ml). T cells were stained with the proliferation dye eFluor 670 (5 mmol; eBioscience), washed twice in PBS, and mixed with pDCs at a ratio of 1:5 (pDC/T cells). For indirect coculture assays, primary human T cells were stained with eFluor670 and preactivated using ImmunoCult Human CD3/CD28/CD2 T Cell Activator (Stemcell Technologies, Cologne, Germany), seeded in 96-well plates (5 × 10⁵ T cells in 100 μl T cell medium), and daily pulsed with 50 μl primary pDC supernatants for 5 d. Cells and supernatants were harvested after 5 d of culturing. Supernatants were measured for T cell cytokine secretion as described above. Cells were stained for CD3 expression, and proliferation of CD3⁺ T cells was measured using flow cytometry by analyzing eFluor 670 dilution.

Data were analyzed using GraphPad Prism 5.0 (GraphPad Software, San Diego, CA). The p values were calculated using one-sample t test, Student t test, or ANOVA with Bonferroni correction as indicated in the figure legends.

Taking immunomodulatory effects of S1P on various cell types into account, we asked whether stimulation with S1P affects pDC activation. We stimulated human primary pDCs with the TLR9 agonist CpG-A to induce IFN-α secretion. Supplying S1P, followed by CpG-A addition for additional 16 h, concentration-dependently reduced IFN-α secretion (Fig. 1A). Next, we asked for the S1P receptor suppressing IFN-α secretion. We first analyzed the S1PR expression profile on human primary pDCs before and after activation with CpG-A by RT-qPCR. Unstimulated human pDCs dominantly expressed S1PR1 as well as S1PR4 and, to a lesser extent, S1PR5 (Fig. 1B). Interestingly, S1PR4 was strongly downregulated 16 h after pDC activation both on mRNA (Fig. 1B) and protein level (Fig. 1C), suggesting a regulatory role in pDC activation. To identify the S1PR transmitting the S1P-dependent IFN-α inhibition, we prestimulated human primary pDCs with S1P in combination with antagonists of different S1PRs for 30 min, followed by TLR9 activation with CpG-A for an additional 16 h. We used VPC23019, a selective antagonist of S1PR1/3 (21), JTE-013, a S1PR2 inhibitor when used at 100 nmol (18), and Cym50358, a potent and selective S1PR4 antagonist (22). Neither VPC23019 nor JTE-013 prevented the S1P-triggered decrease in IFN-α, indicating that S1PR1/2/3 signaling were dispensable for IFN-α suppression by S1P. However, blocking S1PR4 with Cym50358 significantly abolished the S1P-dependent IFN-α decrease (Fig. 1D). Thus, S1P inhibits IFN-α secretion exclusively via S1PR4. Because IFN-α production by pDCs is amplified in a positive feed-forward loop, we asked whether the S1P-triggered reduction was due to a blockade of IFN-α/β receptor signaling or whether it affected early IFN-α production. We used the selective pharmacological S1PR4 agonist Cym50138 (23) to mimic S1P effects. S1PR4-dependent reduction of TLR9-induced IFN-α secretion was obvious 4 h after TLR9 activation (Fig. 1E). Therefore, S1PR4 signaling influences early IFN-α production by pDCs. To validate the specificity of the S1PR4 agonist, we performed siRNA-mediated knockdown of S1PR4 in primary human pDC and analyzed IFN-α2 expression by qPCR after activation with CpG-A with or without addition of Cym50138. In control siRNA–transfected pDCs, the S1PR4 agonist reduced CpG-A–dependent IFN-α expression, which was reversed in S1PR4 knockdown pDCs (Fig. 1F, 1G). The effect of S1PR4 signaling on soluble mediator production seemed to be rather specific for IFN-α, as CpG-A–dependent expression of the immune mediators IL-6, IL-8, TNF-α, CXCL10, CCL5, and PGE₂ was not affected by preincubation with the S1PR4 agonist (Fig. 1H–J). In conclusion, S1P selectively reduced IFN-α expression by primary human pDC via S1PR4.

Next, we asked for consequences of S1PR4-dependent IFN-α suppression. PDC-derived IFN-α mediates its antiviral and growth inhibitory effects through expression of IFN-I–stimulated genes (ISGs) (reviewed in Ref. 24). To investigate the relevance of the reduced IFN-α levels after S1PR4 activation on antiviral properties of pDCs, whole PBMC cultures were prestimulated with Cym50138 and activated with CpG-A, followed by IFN-α measurement and elucidation of specific ISGs at mRNA level. As PBMCs depleted of pDCs only produce minor amounts of IFN-α (<2 pg/ml) in response to CpG-A compared with whole PBMC cultures containing pDCs (>10,000 pg/ml) (Fig. 2A), we presume that pDCs are the main source of IFN-α. TLR9-induced cytokine secretion was also significantly reduced in whole PBMC cultures upon Cym50138 addition (Fig. 2B). Next, we analyzed three ISGs for their expression after CpG-A treatment, with or without Cym50138 prestimulation. The expression of the ISGs Mx1, IFN-induced dsRNA-activated protein kinase, and IRF7 were upregulated after CpG-A addition. Preadministration of Cym50138 did not reduce their expression, rather inducing a slight but insignificant upregulation (Fig. 2C). Therefore, we conclude that reduced IFN-α levels did not translate into reduced antiviral activity in our experimental setup.

Besides conventional DCs (cDCs), also pDCs are considered as APCs that can induce T cell proliferation/activation (12, 25). The quality of the resulting T cell response depends on the pDC activation state and microenvironment (12, 25–27). We asked whether S1P alters surface expression of costimulatory molecules on pDCs and concomitant T cell activation. Activation of human primary pDCs with CpG-A for 16 h increased MHC class I (MHC I) but not MHC II surface expression (Fig. 3A). Neither MHC I nor MHC II expression was altered upon S1P treatment (Fig. 3A). Furthermore, we analyzed costimulatory molecules on pDCs (Fig. 3B). The unaltered expression of CD80, CD83, CD86, CD40, and OX40L after S1P treatment suggested that S1P did not affect APC potential of pDCs. GrB production by pDCs is downregulated upon TLR9 activation to support T cell proliferation (28, 29). Freshly isolated pDCs did not express GrB (not shown), whereas IL-3 induced strong GrB expression (Fig. 3C). Administration of the S1PR4 agonist together with IL-3 induced GrB expression slightly but not significantly. Supplying CpG-A to IL-3 and Cym50138-treated cells failed to alter intracellular GrB expression compared with the corresponding control (Fig. 3C). These data suggested that S1P, via S1PR4, did not influence the Ag-presenting capacity of pDCs upon CpG-A treatment. CpG-A strongly induces IFN-α production by pDCs, but has only minor effects on costimulatory molecule expression, which is strongly induced by CpG-B oligonucleotides (30). Therefore, we analyzed whether S1PR4 signaling would affect the APC potential of CpG-B–activated pDCs. As expected, CpG-B induced strong upregulation of CD80, CD83 and CD86 compared with CpG-A (Fig. 3D, 3E). However, pretreatment with the S1PR4 agonist Cym50138 had no impact on CpG-B–induced expression of these molecules (Fig. 3E). Taken together, S1P signaling through S1PR4 unlikely alters the APC capacity of pDCs.

Cytokines act in synergy with presented Ags and costimulatory molecules to support T cell activation and proliferation. IFN-α directly affects CD8⁺ T cell responses by maintaining their proliferation (31). Despite the fact that S1PR4 signaling had no influence on costimulatory molecule expression on pDCs, we asked if altered IFN-α levels affected T cell activation. We used a T cell activation and proliferation assay in which cocultured pDCs were pulsed with TTX. Cytokine secretion (IFN-α) of pDCs was induced by CpG-A and modulated by the addition of Cym50138. TTX served as a natural Ag, presented by pDCs via MHCs to induce T cell proliferation. As pDCs can induce T cell anergy in coculture systems, which can be reversed by adding IL-2 (32), we supplemented adequate IL-2 levels. pDCs were washed before adding them to T cells to prohibit a direct contact of T cells with the Ag or Cym50138. Proliferation of CD3⁺ T cells was measured 5 d after coculture by monitoring eFluor670 proliferation dye dilution by FACS (Fig. 4A, 4B). Compared to the negative control, activated and Ag-loaded pDCs induced moderate T cell proliferation, but Cym50138 prestimulation of pDCs failed to mediate a significant change in T cell proliferation. In general, APCs are able to prime naive cells into different effector cells. For instance, IFN-α induces Th1 responses, which are characterized by IFN-γ secretion (33–35). Moreover, priming of naive T cells into IL-10 producing Tregs by alternatively activated pDCs is discussed. Based on this discrepancy we asked, if altered IFN-α levels in pDC/T cell cocultures may alter the differentiation of cocultured T cells. To address this question, we measured cytokine levels (IL-2, IL-10, and IFN-γ) in coculture supernatants. Besides IL-2, as a marker for T cell proliferation (which we supplied), all other cytokines were induced in cocultures with activated and Ag-loaded pDCs (Fig. 4C). Interestingly, IL-10 was strongly induced in the Cym50138-treated group, whereas IFN-γ levels were diminished compared with CpG-A only–treated cells. To gain further evidence that IFN-α production rather than Ag presentation or costimulation were involved in altered T cell differentiation by S1PR4-stimulated pDCs, we employed an approach in which isolated T cells were preactivated by stimulating CD2, CD3, and CD28 and subsequently pulsed daily with pDC supernatants for 5 d. Also in this setting, supernatants of Cym50138 plus CpG-A–stimulated pDCs resulted in increased IL-10, but decreased IFN-γ production by T cells in comparison with supernatants of only CpG-A stimulated pDCs (Fig. 4D). Importantly, readdition of IFN-α largely prevented these alterations in cytokine production (Fig. 4D). Taken together, S1PR4 signaling in pDCs translated into altered T cell differentiation, which was largely dependent on reduced IFN-α production.

Next, we were interested in the mechanism by which S1P, via S1PR4, blocks IFN-α production in pDCs. To address this question, we first analyzed the binding and uptake of FITC-labeled CpG-A by human primary pDCs. The first steps of endosomal TLR7/9 activation are the recognition and uptake of the ligand, followed by intracellular trafficking to early endosomes. Because only one CpG uptake receptor (DEC-205) has been described (4), we analyzed its surface expression on human primary pDCs. As shown in Fig. 5A, human blood pDCs express DEC-205. However, activation with CpG-A for different time points did not result in receptor internalization (Fig. 5B). Recently, it was shown that CpG ODNs enter the cell rapidly, and colocalization of CpG ODNs with TLR9 within early endosomes takes place 5 min after treatment (36). Time kinetics of CpG-A-FITC binding/uptake in human primary pDCs by FACS showed that CpG-A-FITC binds to pDCs during the first 5 min of activation (Fig. 5C). However, there was no difference between the control group (Fig. 5C, black line) and Cym50138 prestimulated cells (Fig. 5C, red line). Moreover, we also failed to detect differences in intracellular accumulation of FITC-labeled CpG-A by immunofluorescence analysis (Fig. 5D). A next option was the availability of the receptor for CpG-A. Recently, it was shown that TLR9 traffics from the Golgi apparatus to the plasma membrane, followed by shuttling to the endosomes, where ligand attachment takes place (37). Thus, enhanced TLR9 accumulation at the plasma membrane might be a marker of cells responding weaker to TLR9 ligands. However, S1PR4 signaling did not induce TLR9 expression at the plasma membrane, which was observed upon CpG-A administration, indicating recycling of the activated receptor. Rather, S1PR4 activation reduced TLR9 accumulation, thus being another marker (besides IFN-α production) of delayed/reduced cell activation in the presence of an S1PR4 agonist (Supplemental Fig. 1, black bar). Primary pDCs were reported to acquire a more mature phenotype after isolation upon overnight culture in vitro, which affects IFN-α production (38). Thus, we wondered whether S1PR4 signaling affects human pDC maturation. After overnight culture indeed two pDC subpopulations could be distinguished based on their light scatter profile as reported before (Supplemental Fig. 2A) (38). Interestingly, CpG-A uptake was higher in the reported mature (forward light scatter^lowside scatter^high) pDC subpopulation (Supplemental Fig. 2B). Adding CpG-A to freshly isolated pDCs also increased the forward light scatter^lowside scatter^high pDC population compared with control cells. However, S1PR4 signaling slightly, but not significantly, decreased this population (Supplemental Fig. 2C, 2D), pointing to an S1PR4-independent in vitro maturation and arguing against maturation as the underlying principle of S1PR4-dependent IFN-α suppression.

To avoid an overshooting IFN-I response, IFN-α production by pDCs is highly regulated. Various regulatory surface receptors on pDCs signal via ITAM/ITIM pathways that involve members of the Src kinase family as downstream signaling modules (6, 39). Some of these receptors undergo internalization after ligand binding (reviewed in Ref. 1). To elucidate the mechanism behind the decreased IFN-α production downstream of S1PR4 activation, we analyzed, among others, the surface expression of the inhibitory receptor ILT7 on primary blood pDCs by FACS (Fig 6A). Unexpectedly, ILT7 surface expression rapidly (5 min) decreased after CpG-A stimulation (Fig. 6B, black line) and reappeared after additional 5 min. In contrast, pDCs prestimulated with Cym50138 and activated with CpG-A did not show any changes in ILT7 surface expression during uptake of the TLR ligand (Fig. 6B, red line, and 6C, 6D). These findings were confirmed by immunofluorescence microscopy (Fig. 6E). Although a decrease in ILT7 mRNA after pDC activation was observed before (5), long-term activation (16 h) of pDCs with CpG-A had no effect on ILT7 surface expression (Supplemental Fig. 3A). Because the ligand for ILT7, BST2, is expressed on human blood pDCs, we investigated if the expression of BST2 was affected in our experimental setup. BST2 is an IFN-I–induced gene. Hence, surface expression was upregulated after long-time CpG-A stimulation (16 h) (Supplemental Fig. 3B). Although S1PR4 signaling (30 min) slightly increased BST2 surface expression, activation of S1PR4 in combination with TLR9 activation failed to alter BST2 expression (Supplemental Fig. 3B), indicating that alterations in BST2 expression did not contribute to S1PR4-dependent IFN-α suppression. We therefore hypothesized that the internalization of ILT7 was required for CpG-A–dependent IFN-α production and that inhibition of this process was the primary mechanism by which S1PR4 attenuated IFN-α production. A requirement for such a mechanism would be that altered ILT7 expression in the observed short time frame translated into differences in ILT7 receptor engagement. We approached this question by asking whether accumulation of pDC in cell clusters, indicative of enhanced receptor/ligand interaction, was altered after short-term CpG-A treatment with or without S1PR4 activation. Human PBMCs were therefore prestimulated with Cym50138 followed by activation with CpG-A for 5 min, fixation, and flow cytometric analysis (Fig. 6F). Indeed, we observed a small but very consistent decrease in pDC accumulation in cell clusters at this early time point after CpG-A treatment, which was prevented by S1PR4 activation (Fig. 6G, 6H). These data may reflect ILT7 internalization.

Our hypothesis gained further indirect support by the observation that not all TLR ligands altered ILT7 surface expression on human blood pDCs. Activation of whole PBMCs with another TLR9 ligand, CpG-B (CpG2006), provoked ILT7 internalization on pDCs, whereas TLR7 activation with IMQ did not (Fig. 7A). Interestingly, a more physiological TLR7 ligand, the tick-borne encephalitis vaccine (FSME), promoted a rapid and prolonged ILT7 internalization (Fig. 7B, 7C). ILT7 internalization by CpG-B or FSME was reversed by S1PR4 activation, whereas ILT7 levels on IMQ-treated cells were not altered by S1PR4 signaling (Fig. 7A–C). Importantly, for all used TLR ligands, ILT7 internalization was inversely correlated with IFN-α production. Both CpG-A and CpG-B, as well as the tick-borne encephalitis vaccine FSME, induced robust amounts of extracellular IFN-α, whereas IMQ administration elicited low IFN-α secretion (Fig. 7D). IFN-α levels induced by CpG-B or FSME were decreased by Cym50138. In contrast, pDCs prestimulated with Cym50138 and activated with IMQ produced equal amounts of IFN-α compared with only IMQ-treated cells (Fig. 7E). These data suggest that unaltered ILT7 expression after IMQ stimulation might be responsible for low IFN-α production compared with TLR ligands that induce ILT7 internalization. In contrast, S1PR4 signaling only prevents IFN-α secretion triggered by ligands that induce ILT7 depletion from the cell surface. To support this hypothesis, we blocked ILT7 downstream signaling by inhibiting of the tyrosine-protein kinase Lyn. As shown in Fig. 7F, Cym50138-pretreated cells decreased CpG-A–induced IFN-α production compared with control pDCs. However, preadministration of an LPI for 30 min blocked the effect of S1PR4 signaling on IFN-α secretion. Lyn is not exclusively activated downstream of ILT7, but also other inhibitory pDC receptors such as CD303 (40). Therefore, extensive future studies will be required to link these molecules in the context of our setting. Finally, we approached signaling pathways downstream of S1PR4 preactivation (Supplemental Figure 4) that are involved in blocking ILT7 internalization following pDC activation. S1PR4 couples to Gα12/13, which triggers activation of the RhoA/ROCK pathway that is, among others, involved in vesicle trafficking and membrane dynamics. Blocking of RhoA as well as ROCK with pharmacological inhibitors in combination with Cym50138 indeed restored ILT7 internalization after pDC activation with CpG-A (Fig. 7G). Unfortunately, RhoA/ROCK inhibition prevented CpG-A–induced IFN-α secretion. Therefore, these ILT7 expression data could not be correlated with IFN-α levels.

Together, our data suggest a mechanism of S1PR4-dependent IFN-α suppression (Fig. 8). S1PR4 signaling through the RhoA/ROCK pathway preserves surface expression of the inhibitory receptor ILT7. Increased ILT7 signaling consequently limits IFN-α production.

The sphingolipid S1P signals via G-protein–coupled receptors to affect inflammation. Whereas S1PR1 and 2 mainly regulate immune cell migration, the role of the immune cell-specific S1PR4 is ill defined (14). There is evidence that S1PR4 signaling modulates immune cell phenotypes, because a lack of S1PR4 in mouse cDCs reduces Th17 and increases Th2 immune responses (17). Moreover, S1PR4 activation on human monocyte-derived DCs induced IL-27 production to increase Treg-dependent suppression of cytotoxicity against human tumor cells (19). The present study suggests that S1PR4 signaling also regulates critical functions of human pDCs, by blocking IFN-I secretion.

Upon in vitro activation, pDCs rapidly produce significant amounts of IFN-α, which initiates a feed-forward loop that further increases IFN-α production at later time points (41, 42). The S1PR4-triggered decrease in IFN-α secretion was apparent already 4 h after TLR9 activation, pointing to a regulatory impact on the initial steps following TLR ligand uptake/recognition rather than on feed-forward signaling. This notion is supported by experiments with a reversed experimental setup (i.e., TLR9 activation followed by S1PR4 agonist stimulation), in which S1PR4 signaling failed to diminish IFN-α secretion (Supplemental Fig. 4). The first step leading to endosomal TLR activation is, for a variety of agonists, receptor-dependent uptake. For instance, viruses and bacteria typically enter the cell by receptor-mediated endocytosis or phagocytic pathways, followed by their degradation and the release of immunogenic nucleic acids. The uptake of synthetic TLR ligands such as CpG is also receptor mediated, although the underlying mechanisms are largely unknown. DEC-205, an Ag delivery receptor expressed by murine CD8⁺ DCs and different human leukocytes including pDCs (43–45), was recently identified as an uptake receptor for CpG ODNs (4). However, DEC-205 was irrelevant at least for CpG-A uptake in human primary pDCs, because the receptor failed to internalize upon CpG-A administration in our experimental setup. Moreover, uptake of FITC-labeled CpG by the whole pDC population was not affected by S1PR4 signaling, indicating a regulatory impact downstream or independent of ligand uptake/recognition.

We observed that CpG was taken up at higher levels by a subpopulation of primary human pDCs, which was, to our knowledge, not reported before. This subpopulation corresponds to pDCs showing increased maturation marker expression, which are generated upon in vitro culture of primary human pDCs isolated from blood and further expanded following TLR9 ligation (38) (Supplemental Fig. 2). An S1PR4-dependent alteration of this pDC subset was not apparent (Supplemental Fig. 2C). However, we noticed an impact of S1PR4 signaling on ILT7 expression, which was proposed as a maturation marker of human pDC in vitro (38). We observed rapid ILT7 internalization after CpG-A administration, which was prevented by S1PR4 signaling and coupled to IFN-α production. By using different TLR ligands, we confirmed the impact of S1PR4 on ILT7 trafficking and concomitant IFN-α regulation. Strikingly, only stimulation with TLR ligands that are taken up by receptor-mediated endocytosis (FSME vaccine or CpG ODNs) triggered ILT7 internalization and were responsive to S1PR4-dependent IFN-α suppression. In contrast, IMQ, which enters the cell via diffusion through the plasma membrane, failed to alter ILT7 surface expression, and IMQ-dependent IFN-α production was not regulated by S1PR4. These data allow proposing a model of ILT7-dependent IFN-I regulation. ILT7, when expressed at the cell surface, binds its ligand BST2 (e.g., on bystander pDCs), thereby limiting IFN-α production. Full-blown pDC activation requires ILT7 internalization, which is coupled to endocytosis of TLR7/9 ligands via an unknown mechanism, to prevent ILT7 interaction with its ligand and downstream inhibitory signaling (Fig. 8). Interestingly, ILT7 internalization was prolonged when using an immobilized virus instead of synthetic TLR ligands, thus strengthening the physiological relevance of this mechanism during pDC activation.

S1PR4 signaling was previously coupled to receptor trafficking. We observed that production of inflammatory cytokines by tumor-associated macrophages required shuttling of the nerve growth factor receptor TrKA to the plasma membrane, which was initiated by S1P and S1PR4 (18). Membrane trafficking of cell-surface receptors involves alterations of actin dynamics. Accordingly, S1PR4 predominantly couples to G12/13 that activates Rho and regulates cell shape and stress fiber formation (46). Indeed, inhibiting Rho and downstream ROCK activity prevented the impact of S1PR4 on ILT7 trafficking. Future studies will provide further mechanistic insight as to how S1PR4 signals locally prevent ILT7 internalization and whether it involves the activation of known immunomodulatory mediator systems that are known to affect IFN-α production by pDC. Recently, it was shown that other G-protein–coupled receptor ligands, such as PGE₂, can reduce IFN-α production by pDCs (47–49). PGE₂ in immune cells signals mostly via EP2 or EP4, which couple to Gs to elevate cAMP. Interestingly, G12/13 are also capable of increasing cAMP via adenylate cyclase 7 that is highly expressed in human primary pDCs (50). It will be interesting to determine whether cAMP signaling contributes to inhibiting IFN-α production downstream of S1PR4.

The defense against viral infection is a hallmark of pDC function (51). Hence, production of IFN-I in response to viral components via TLR7/9 is indispensable to ensure upregulation of antiviral IFN-stimulated genes. In our setup, decreased IFN-α levels after S1PR4 activation were not translated into decreased mRNA levels of different ISGs, which might be due to the relatively high remaining IFN-α levels (mean >1000 pg/ml) and the high affinity of the IFN-α/β receptor 1/2 heterodimer for IFN-α. However, reduced IFN-α production was coupled to T cell polarization. The induction of adequate T cell responses requires the ability of APCs to process Ags, followed by precise presentation through the MHCs, the expression of costimulatory molecules, and, finally, a suitable milieu displayed by specific (inflammatory) cytokines (e.g., IFN-α and IL-12) (52). Human pDCs as APCs were shown to capture and cross-present viral Ags via MHC I to naive T cells, resulting in CD8⁺ T cell responses (53). Furthermore, human blood pDCs express several costimulatory and T cell–activating molecules, which are required for optimal T cell activation. S1PR4 activation failed to alter the expression of MHCs and costimulatory molecules or others regulators of T cell activation/proliferation such as GrB (29). Taken together, our data do not reveal an impact of S1PR4 on APC functions of human blood pDCs, corroborating a previous observation that S1PR4 did not affect Ag presentation by human cDCs (19). T cells require, besides Ag presentation and costimulation, a third signal to develop strong effector functions (52). IFN-α was shown to maintain proliferation of CD8⁺ T cells (54, 55). However, activation of S1PR4 signaling in pDCs and thus reducing IFN-α failed to alter T cell proliferation. Beside its role in lymphocyte activation, IFN-α modulates the outcome of T cell responses. It has been described that IFN-α directly affects human naive CD4⁺ T cells to favor their differentiation into IFN-γ–producing Th1 cells (33). Furthermore, IFN-α suppresses the expansion of regulatory T cells, while increasing IFN-γ–producing effector T cells (56). Triggering S1PR4 on Ag-pulsed and TLR-activated pDCs, followed by cocultivation with autologous T cells, shifted cytokine secretion from IFN-γ to IL-10, corroborating these previous reports. In this regard, S1PR4 signaling enables pDCs to adopt a regulatory phenotype inducing tolerogenic T cell responses.

Our findings illustrate one possible mechanism by which the tumor microenvironment, via enhanced S1P production, avoids tumor eradication by high levels of IFN-I (57, 58). In concordance with this hypothesis, tumor-associated pDCs show a partially activated phenotype and produce very low amounts of IFN-α compared with pDCs isolated from patient’s blood (59). Furthermore, high numbers of infiltrating pDCs in tumors were correlated with poor patient prognosis in ovarian cancer (60). Targeting the S1P–S1PR axis is a promising tool to overcome several diseases, including autoimmune disorders and cancer. There are many studies dealing with the modulation of S1PR1 in different autoimmune diseases (e.g., multiple sclerosis and rheumatoid arthritis), in which the therapeutic benefit is attributed to inhibition of immune cell trafficking (reviewed in Ref. 61). Interestingly, S1PR1 signaling did not affect pDC trafficking into diseased areas in a murine MS model, which proved beneficial for disease outcome (62). This was attributed to a low S1PR1 expression level in murine pDC, which selectively express S1PR4 (63). Whether the immunosuppressive effect of pDCs accumulating in the brain of experimental animals in the multiple sclerosis model is at least partially transmitted through S1PR4 remains to be determined. However, the current study and previous studies (17–19) reveal a role for S1PR4 in modulating immune cell activation that is relevant in an inflammatory context. S1PR4 might be a promising therapeutic target to modulate pathogenic IFN-α production in IFN-I–driven diseases or to relieve a block of IFN-I production in tumors.

We thank Praveen Mathoor and Margarethe Mijatovic for excellent technical assistance.

The authors have no financial conflicts of interest.