Systematic cytokine receptor profiling reveals GM-CSF as a novel TLR-independent activator of human plasmacytoid predendritic cells

Plasmacytoid predendritic cells (pDCs) play a major role in innate immunity by producing large amounts of type 1 interferon¹  after Toll-like receptor (TLR) engagement by microbes or microbial products.²  They are involved in antiviral immunity in human and mouse²  and can also respond to CpG-enriched bacterial DNA.³ 

Besides infection, there is growing evidence that pDCs are also involved in nonmicrobial inflammatory diseases. In some autoimmune settings, such as in psoriasis⁴  and lupus erythematosus,⁵  a TLR-dependent pathway of pDC activation has been identified. In other settings, such as allergy,⁶  rheumatoid arthritis,⁷  multiple sclerosis,⁸  Sjögren syndrome,⁹  and cancer,¹⁰  there is no evidence for a TLR-dependent pDC activation, suggesting a prominent role for TLR-independent pathways, through inflammatory mediators and cytokines. However, these pathways remain poorly characterized. Interleukin-3 (IL-3) is the only cytokine known to potently activate pDCs,¹¹  and its physiopathologic significance remains unclear.² 

We hypothesized that other cytokines may be able to activate pDCs within a complex inflammatory environment. In this study, we performed a systematic analysis of cytokine receptor expression, which identified granulocyte-macrophage colony-stimulating factor (GM-CSF) as a novel activating factor for human pDCs.

Buffy coats were obtained from healthy adult donors at the Saint-Antoine Crozatier Blood Bank. pDCs were isolated by flow cytometry as Lin⁻CD11c⁻CD4⁺ as previously described.¹²  Purity was more than 98%. The study was approved by the Institut Curie Institutional Review Board, and all blood donors gave informed consent for research use of the biologic material in accordance with the Declaration of Helsinki.

Total RNA was extracted from pDCs freshly isolated from the blood of healthy donors, double-amplified, and labeled according to the protocol recommended by Affymetrix for hybridization to Human Genome U133 Plus 2.0 arrays. We extracted the expression levels for cytokine receptor chains from IL-1R to IL-23R, and receptors for other cytokines and growth factors that are known to be important for hematopoietic cells, such as interferon-γ (IFN-γ), tumor necrosis factor (TNF), granulocyte colony stimulating-factor (G-CSF), monocyte colony stimulating factor (M-CSF), stem cell factor (SCF), platelet-derived growth factor (PDGF), and GM-CSF.

Freshly sorted pDCs were stained with specific antibodies: phycoerythrin (PE)–interleukin-3 receptor (IL-3R; BD Biosciences), PE–IL-4R (BD Biosciences), PE–IL-6R (BD Biosciences), fluorescein isothiocyanate (FITC)–IL-7R (eBioscience), allophycocyanin–IL-8R (R&D Systems), PE–IL-10R (BD Biosciences), PE–IL-18R (BD Biosciences), PE–TNFR (BD Biosciences), PE–IFN-γR (R&D Systems), PE–M-CSFR (R&D Systems), and PE–GM-CSFR (BD Biosciences). Isotype-matched antibodies were used as control. The analysis was performed on an LSRII (BD Biosciences).

pDCs were cultured 48 hours in RPMI 10% FCS in the presence of recombinant IL-6 (Miltenyi Biotec), IL-3, IL-10, IL-18, and GM-CSF at 10 ng/mL, IFN-γ at 100 U/mL (all from R&D Systems) and CpG-C (a kind gift from Dr F. Barrat) at 5 μg/mL.

pDC viability was assessed by measuring the percentage of 4,6-diamidino-2-phenylindole-negative pDCs by fluorescence-activated cell sorter after 48 hours of culture with recombinant cytokines.

pDC maturation was analyzed after 48 hours of stimulation measuring the expression of costimulatory molecules and human leukocyte antigen (HLA)–DR levels by fluorescence-activated cell sorter. We used the following antibodies: FITC-CD80 (BD Biosciences), PECy7-CD86 (eBioscience), PE-ICOS-ligand (eBioscience), FITC-OX40L (Ancell), and Alexa 700-HLA-DR (BioLegend). Isotype-matched antibodies were used as control. pDCs were cytospun in parallel and colored with Giemsa staining. Pictures were taken with a CFW-1308C color digital camera (Scion Corporation) on a Leica DM 4000 B microscope, with a 100×/1.30 oil objective.

pDC supernatants were collected after 48 hours of culture in medium alone, IL-3, GM-CSF, and CpG-C, and frozen at −80°C until assayed. IFN-α was measured by enzyme-linked immunosorbent assay (PBL Biomedical Laboratories). TNF and IL-6 were measured by Cytometric Bead Array Flex Set (BD Biosciences).

CD4 naive T cells were sorted as CD4⁺CD45RA⁺CD25⁻CD45RO⁻ by flow cytometry as previously described.¹³  Purity was higher than 98%. Naive CD4 T cells were cultured for 6 days with allogeneic pDCs at a 5:1 ratio as previously described.¹⁴  Supernatants were collected after 24 hours of anti-CD3/CD28 microbeads (Dynal) restimulation. Cytokine measurement was performed by Cytometric Bead Array Flex Set (BD Biosciences). mRNA transcripts were quantified by real-time quantitative polymerase chain reaction on an ABI Prism 7900 sequence detector (Applied Biosystems) with Applied Biosystem predesigned TaqMan Gene Expression Assays and Absolute QPCR ROX mix (Thermo Fisher Scientific).

Comparisons between different conditions were performed using the Wilcoxon test. Statistical significance was retained for P values less than .05.

Primary human pDCs were used throughout this study. In a first screening, we used Affymetrix gene expression profiling to assess the expression of 44 cytokine receptor chains corresponding to 26 different cytokines and growth factors (Figure 1A). Eleven cytokine receptors were expressed as heterodimers: IL-3R, IL-4R, IL-6R, IL-7R, IL-8R, IL-10R, IL-18R, TNFR, IFN-γR, M-CSFR, and GM-CSFR (Figure 1A). To confirm surface expression at the protein level, we used flow cytometry. Six receptors were positive on freshly sorted pDCs: IL-3R, IL-6R, IL-10R, IL-18R, IFN-γR, and GM-CSFR; and 5 receptors were negative: IL-4R, IL-7R, IL-8R, TNFR, and M-CSFR (Figure 1B). This discrepancy between mRNA and protein level may be the result of a lower sensitivity of flow cytometry compared with mRNA expression. Thus, in addition to IL-3, our systematic receptor analysis identified 5 cytokines, IL-6, IL-10, IL-18, IFN-γ, and GM-CSF, as potential candidates for activating pDCs.

Next, we studied the effect of these cytokines on 2 major functional outcomes of pDCs: survival and differentiation into DCs. After 48 hours of culture, only IL-3 and GM-CSF were able to maintain pDC viability (Figure 1C) and to induce the up-regulation of costimulatory molecules, such as CD80, CD86, and ICOSL (inducible T-cell co-stimulator ligand) (Figure 1C; supplemental Figure 1, available on the Blood Web site; see the Supplemental Materials link at the top of the online article). After Giemsa staining, the morphology of GM-CSF–activated pDCs was compatible with a differentiation into DCs, with an increase in cell size, formation of dendrites, vacuolar cytoplasm, and a smaller nucleus (Figure 1D). Overall, our data show that, among a variety of cytokines, only IL-3 and GM-CSF maintain pDC viability and induce their differentiation into DCs. Interestingly, GM-CSF was shown to inhibit pDC development from CD34 hematopoietic precursors.²  Thus, GM-CSF may have a dual role that depends on the site of its production: negative role on pDC development but positive role on peripheral pDC activation. The fact that IL-6, IL-10, IL-18, and IFN-γ were not sufficient to induce pDC activation suggests that their function may be to regulate the response of pDCs to other stimuli. This was already described for IL-10, which was shown to decrease the IL-3-induced survival of pDCs¹⁴  and to inhibit type 1 IFN production by TLR-activated pDCs.¹⁵ 

To determine the effect of GM-CSF on pDCs, we performed an extensive phenotypic analysis compared with IL-3 and CpG-C pDCs. We observed an up-regulation of CD86, ICOSL, OX40L, and HLA-DR in GM-CSF pDCs compared with ex vivo pDCs and nonstimulated pDCs (Med). The levels of OX40L and HLA-DR were similar to that induced by IL-3. ICOSL was found to be less expressed in GM-CSF pDCs. The expression of costimulatory molecules was higher in CpG-C pDCs, in accordance with a more mature phenotype induced by TLR activation (Figure 2A-B). We did not detect significant levels of IFN-α, TNF, and IL-6 in supernatants of IL-3- and GM-CSF pDCs compared with CpG-C pDCs (supplemental Figure 2).

Because GM-CSF pDC efficiently differentiated into DCs, we tested their ability to prime and polarize allogeneic naive CD4 T cells. GM-CSF pDCs induced higher IFN-γ and TNF and less IL-4 and IL-10, compared with IL-3 pDCs and CpG-C pDCs, both at mRNA and protein level (Figure 2C-D). This represents a previously unrecognized property of GM-CSF in modulating Th differentiation. Interestingly, the reduced IL-10 production by T cells correlated with a reduced expression of ICOSL on GM-CSF pDCs, illustrating an important role of ICOSL in priming T cells for IL-10.¹³  We did not detect any differences in the levels of CD4 T cell-derived IL-5, transforming growth factor-β (TGF-β; Figure 2D), IL-17, and IL-22 (supplemental Figure 3) between GM-CSF, IL-3, and CpG-C pDCs. Neither did we observe differences in the mRNA levels of the transcription factors T-bet, GATA3, FOXP3, and SOCS3 (supplemental Figure 4).

Our data identified GM-CSF as a novel activator of pDCs. Because GM-CSF is produced in various types of inflammations, including asthma,¹⁶  rheumatoid arthritis,¹⁷  Sjögren syndrome,¹⁸  and psoriasis,¹⁹  it is a strong candidate for inducing pDC activation in these settings because the presence of pDCs has been documented.^4,7,9,20  GM-CSF is produced by a large variety of cell types, including fibroblasts, endothelial cells, T cells, macrophages, mesothelial cells, epithelial cells, and tumor cells²¹  and could potentially mediate a cross-talk between pDCs and any of these cell types. Thus, GM-CSF may serve as a link between a pathologic inflammatory environment, pDC activation, and the modulation of CD4 T-cell responses.

Recent studies suggested that GM-CSF can promote tolerance in models of diabetes,²²  autoimmune thyroiditis,²³  and tumor,²⁴  but the mechanism remains unknown. Although our data show that GM-CSF pDCs prime CD4 T cells to produce increased IFN-γ and decreased IL-10 compared with IL-3 pDCs, we cannot exclude that this IL-10 may play a tolerogenic role in some models. Moreover, pDCs were described to be involved in the generation of regulatory T cells after both TLR-dependent and -independent stimulation¹³  (C.G., unpublished data, July 2009). Given the well-described environmental plasticity of pDCs,²  further studies will be important to address their ability to promote tolerance or immunity in the context of endogenous or pharmacologic²⁵  GM-CSF activation.

The authors thank Zofia Maciorowsky and Annick Viguier for the cytofluorimetric cell sorting, David Gentien for the Affymetrix arrays hybridization, and Claire Hivroz and Philippe Benaroch for helpful suggestions and critical reading of the manuscript.

This work was supported by the Leonardo da Vinci Unipharma Graduates Program and Association pour la Recherche contre le Cancer (fellowship; C.G.), European Marie Curie Excellence Grant (R.Z., V.S.), and Fondation pour la Recherche Médicale (fellowship; R.Z.).

Contribution: C.G. designed research, performed experiments, and analyzed data; R.Z. performed experiments and analyzed data; and V.S. designed the study and wrote the paper.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Vassili Soumelis, Institut Curie, 26 rue d'Ulm, 75005 Paris, France; e-mail: vassili.soumelis@curie.net.