The Endogenous Dual Retinoid Receptor Agonist Alitretinoin Exhibits Immunoregulatory Functions on Antigen-Presenting Cells

Retinoids are a frequently used class of drugs in the treatment of inflammatory as well as malignant skin diseases. Retinoids have differential affinity for the retinoic acid receptor (RAR) and/or the retinoid X receptor (RXR). The endogenous dual RAR and RXR agonist alitretinoin (9-cis retinoic acid) demonstrated remarkable efficacy in the treatment of chronic hand eczema (CHE) patients; however, detailed information on the mechanisms of action remains elusive. Here, we used CHE as a model disease to unravel immunomodulatory pathways following retinoid receptor signaling. Transcriptome analyses of skin specimens from alitretinoin-responder CHE patients identified 231 significantly regulated genes. Bioinformatic analyses indicated keratinocytes as well as antigen presenting cells as cellular targets of alitretinoin. In keratinocytes, alitretinoin interfered with inflammation-associated barrier gene dysregulation as well as antimicrobial peptide induction while markedly inducing hyaluronan synthases without affecting hyaluronidase expression. In monocyte-derived dendritic cells, alitretinoin induced distinct morphological and phenotypic characteristics with low co-stimulatory molecule expression (CD80 and CD86), the increased secretion of IL-10 and the upregulation of the ecto-5′-nucleotidase CD73 mimicking immunomodulatory or tolerogenic dendritic cells. Indeed, alitretinoin-treated dendritic cells demonstrated a significantly reduced capacity to activate T cells in mixed leukocyte reactions. In a direct comparison, alitretinoin-mediated effects were significantly stronger than those observed for the RAR agonist acitretin. Moreover, longitudinal monitoring of alitretinoin-responder CHE patients could confirm in vitro findings. Taken together, we demonstrate that the dual RAR and RXR agonist alitretinoin targets epidermal dysregulation and demonstrates strong immunomodulatory effects on antigen presenting cell functions.


Introduction
In the early 20th century, vitamin A and its derivatives (retinoids) were discovered to be essential for mammalian health and survival [1]. The active form of vitamin A, retinol, is oxidized by alcohol dehydrogenases and retinal dehydrogenases in vivo to yield retinoic

CHE as a Model Disease to Investigate the Effects of Alitretinoin (9-cis RA) on Skin Homeostasis and Inflammation
In patients treated with alitretinoin, we used the modified Total Lesion Symptom Score (mTLSS) to assess the therapeutic efficacy on CHE. Twenty-three of twenty-nine CHE patients (79%) of the alitretinoin-treated patients showed a response of ≥50% in the mTLSS and were considered responders ( Figure 1). Characteristics of the responder group are listed in supplementary Table S6. Among the selected alitretinoin-responders, a significant reduction of the mTLSS from 10.48 ± 3.37 before to 2.23 ± 1.82 after 12 weeks of treatment was observed (p < 0.001). This corresponded to an average reduction of the mTLSS by 78.63% (Figure 1).

Effects of Alitretinoin (9-cis RA) on Gene Expression in CHE and Cultured Human Primary Keratinocytes
As a first step, we selected skin specimens of alitretinoin-responder CHE patients and performed transcriptome analyses. Using cDNA microarray analysis of skin biopsies of lesional CHE skin before and week 12 of alitretinoin treatment (n = 3), we detected a significant regulation of 231 genes with adjusted p-value < 0.05, log fold-change >1 in skin biopsies (Supplementary Table S1). Interestingly, gene expression patterns not only highlight down-and upregulated clusters (n = 95 and n = 136, respectively) but specific gene groups spanning a wide range of inflammatory, tissue remodeling, as well as barrierrelated pathways. Downregulated genes include general inflammatory markers such as serum amyloid A1 (SAA1) and indoleamine 2,3-dioxygenase (IDO1) alongside chemokines CXCL9 and CXCL10 but also disease-specific markers such as CCL18 and CCL13. Levels of antimicrobial peptides (AMPs) typically upregulated during inflammation, such as peptidase inhibitor 3 (PI3/SKALP) or late cornified envelope genes LCE3A and LCE3C, are

Effects of Alitretinoin (9-cis RA) on Gene Expression in CHE and Cultured Human Primary Keratinocytes
As a first step, we selected skin specimens of alitretinoin-responder CHE patients and performed transcriptome analyses. Using cDNA microarray analysis of skin biopsies of lesional CHE skin before and week 12 of alitretinoin treatment (n = 3), we detected a significant regulation of 231 genes with adjusted p-value < 0.05, log fold-change >1 in skin biopsies (Supplementary Table S1). Interestingly, gene expression patterns not only highlight down-and upregulated clusters (n = 95 and n = 136, respectively) but specific gene groups spanning a wide range of inflammatory, tissue remodeling, as well as barrier-related pathways. Downregulated genes include general inflammatory markers such as serum amyloid A1 (SAA1) and indoleamine 2,3-dioxygenase (IDO1) alongside chemokines CXCL9 and CXCL10 but also disease-specific markers such as CCL18 and CCL13. Levels of antimicrobial peptides (AMPs) typically upregulated during inflammation, such as peptidase inhibitor 3 (PI3/SKALP) or late cornified envelope genes LCE3A and LCE3C, are also normalized during treatment, similarly to matrix metalloproteinase 12 (MMP12). On the other hand, barrier function-related genes such as filaggrin 2 (FLG2), betacellulin (BTC) or Wnt inhibitory factor 1 (WIF1) are upregulated. Dermcidin, an AMP often sup- To gain further insights through the gene expression dataset, the FANTOM5 database was mined to extract cell types that are major sources of the above-mentioned genes (Supplementary Table S2). As expected, a remarkable number of the regulated genes are expressed in keratinocytes, identifying this cell subset as a major target for alitretinoin. In barrier defect-associated inflammation, IL-1β and TNF-α play a major role [17,18]. Thus, we simulated barrier defect-associated inflammation by stimulating cultured human primary keratinocytes with IL-1β and TNF-α in the presence or absence of alitretinoin or acitretin. The RAR agonist acitretin was selected as a comparator for the biological effects of the endogenous dual (RAR/RXR) agonist alitretinoin (9-cis RA). In addition, we also used key mediators of type 2 (IL-4) or type 1 (IFN-γ) inflammation to investigate the retinoid signaling driven effects on these conditions. A selection of genes based on both microarray data and previous findings was analyzed via qPCR, representing the main categories of AMPs, barrier function and tissue regeneration-related genes ( Figure 3, Supplementary Figures S1-S3). Interestingly, under proinflammatory conditions with TNFα and IL-1β, alitretinoin reduces the expression of top regulated genes small proline rich protein 3 (SPRR3) and serine protease inhibitor of Kazal type 6 (SPINK6). In turn, the opposite effect is observed at low doses of the sole RAR agonist acitretin. Furthermore, significant upregulation occurs in betacellulin (BTC) and hyaluronic acid synthase 2 (HAS2) under both retinoid treatments; however, the effect on HAS2 is found to be highly specific to alitretinoin across all conditions (Supplementary Figure S2). Regarding AMPs, defensin beta 4A and defensin beta 103B, which are often co-regulated with PI3 in inflammation, display extreme fold upregulation in stimulated keratinocytes under inflammatory conditions. However, even low doses of retinoids decrease the expression of these genes, and this effect is stronger with high dosages (Figure 3). On the other hand, RNASE7, S100A7 or hornerin (HRNR) do not show marked regulation under induced inflammatory conditions (Supplementary Figure S2). Drawing from the proinflammatory signature observed in vivo, a selected set of chemokines was also investigated in primary human keratinocytes under the above-mentioned conditions. The significant downregulation of CCL5, CCL20 and CXCL14 further supports the notion that 9-cis RA (alitretinoin) directly and indirectly regulates chemokine expression and thereby leukocyte recruitment (Supplementary Figure S5).  To gain further insights through the gene expression dataset, the FANTOM5 database was mined to extract cell types that are major sources of the above-mentioned genes (Supplementary Table S2). As expected, a remarkable number of the regulated genes are expressed in keratinocytes, identifying this cell subset as a major target for alitretinoin. In barrier defect-associated inflammation, IL-1β and TNF-α play a major role [17,18]. Thus, we simulated barrier defect-associated inflammation by stimulating cultured human primary keratinocytes with IL-1β and TNF-α in the presence or absence of alitretinoin or acitretin. The RAR agonist acitretin was selected as a comparator for the biological effects genes, and this effect is stronger with high dosages (Figure 3). On the other hand, RNASE7, S100A7 or hornerin (HRNR) do not show marked regulation under induced inflammatory conditions (Supplementary Figure S2). Drawing from the proinflammatory signature observed in vivo, a selected set of chemokines was also investigated in primary human keratinocytes under the above-mentioned conditions. The significant downregulation of CCL5, CCL20 and CXCL14 further supports the notion that 9-cis RA (alitretinoin) directly and indirectly regulates chemokine expression and thereby leukocyte recruitment (Supplementary Figure S5).

Morphology, Phenotype and Function of 'Retinoid-DCs' Generated from Isolated CD14 + Monocytes
The FANTOM5 analysis also indicated that monocyte-derived cells are another prominent cell type that is targeted by alitretinoin (9-cis RA) (Supplementary Table S2). This is in good concordance with data previously published on all-trans retinoic acid (ATRA) [19][20][21]. Hence, we directed our focus on antigen presenting cells and dendritic cell functions. For this purpose, CD14 + monocytes were isolated from healthy donors and stimulated with GM-CSF and IL-4 to generate monocyte-derived dendritic cells (MoDC). After 6 days, these cells were matured with TNF-α (mMoDC) or left untreated to maintain an immature phenotype (iMoDC). These procedures were performed either with the cytokines alone or in the presence of 9-cis RA (alitretinoin) ('9-cis RA-DCs') or acitretin ('Acitretin-DCs'). Upon treatment with 9-cis RA, a marked change in the morphology of dendritic cells was observed. Compared to untreated MoDC, '9-cis RA-DC' exhibited an elongated shape with dense cell-to-cell connections ( Figure 4).

Morphology, Phenotype and Function of 'Retinoid-DCs' Generated from Isolated CD14 + Monocytes
The FANTOM5 analysis also indicated that monocyte-derived cells are another prominent cell type that is targeted by alitretinoin (9-cis RA) (Supplementary Table S2). This is in good concordance with data previously published on all-trans retinoic acid (ATRA) [19][20][21]. Hence, we directed our focus on antigen presenting cells and dendritic cell functions. For this purpose, CD14 + monocytes were isolated from healthy donors and stimulated with GM-CSF and IL-4 to generate monocyte-derived dendritic cells (MoDC). After 6 days, these cells were matured with TNF-α (mMoDC) or left untreated to maintain an immature phenotype (iMoDC). These procedures were performed either with the cytokines alone or in the presence of 9-cis RA (alitretinoin) ('9-cis RA-DCs') or acitretin ('Acitretin-DCs'). Upon treatment with 9-cis RA, a marked change in the morphology of dendritic cells was observed. Compared to untreated MoDC, '9-cis RA-DC' exhibited an elongated shape with dense cell-to-cell connections ( Figure 4).
Given the profound morphological changes in '9-cis RA-DC', we performed a transcriptomic screening comparing pooled RNA of six independent donors each of iMoDC, mMoDC and '9-cis RA DCs' (Supplementary Figure S8, Supplementary Table S7). Considering the pooling strategy, only the top most robustly regulated genes were taken into account in downstream analysis. Of these, functional network clustering according to GO terms and fold regulation resulted in clusters enriched for: 1, antigen processing presentation (including CD80, CD83, TRAF and CTSD), 2, chemokine mediated signaling (CXCR5, CCL7 and CCL17) 3, negative regulation of cell proliferation (IL12B, OSM and EGR1) and 4, extracellular matrix regulation (including collagen and integrin family members as well as MMP12). Furthermore, qPCR analyses of a selected panel of cell typerelevant genes also confirmed 9-cis RA-specific transcriptomic changes in dendritic cells (Supplementary Figure S6). Since pathway analyses of transcriptomic changes induced by 9-cis RA in dendritic cells pointed towards antigen processing and presentation, we further investigated the phenotype of '9-cis RA-DC' using flow cytometry. To assess the maturation status of '9-cis RA-DC', we analyzed the surface expression of co-stimulatory molecules such as CD80, CD83 and CD86. Following maturation with TNF-α, an increased expression of CD80, CD83 and CD86 was observed ( Figure 5, Supplementary Table S3). In contrast, cells matured in the presence of either 9-cis RA or acitretin showed decreased ex-pression of these maturation markers. At equimolar doses, 9-cis RA was significantly more potent to maintain an immature phenotype and inhibit the cytokine-driven maturation process ( Figure 5).  ing isotype controls (light gray) on immature (iMoDC, dark gray), TNF-a maturated (mMoDC, black) and "9-cis RA-DCs" (blue) after 9 days of culture. 9-cis RA strongly downregulates CD83 and CD86 compared to mMoDCs. (C) Percentage of gated, positive cells for the maturation marker CD83 (n = 11-28 independent donors), co-stimulatory molecules CD80 (n = 11-12 independent donors) and CD86 (n = 11-25 independent donors). Fold changes are shown as mean + SEM of the percentage of gated cells positive for CD83, CD80 and CD86. Unpaired two-tailed Mann-Whitney U-test (* p < 0.05, ** p < 0.01, *** p < 0.001).

'9-cis RA-DC' Show Functionally Active Expression of the Ecto-5'-Nucleotidase CD73 In Vitro
Immature dendritic cells are known to promote tolerance through the generation of regulatory T cells and a lack of T cell activation. Here, we show for the first time that retinoids upregulate the expression of the ecto-5′-nucleotidase CD73 in vitro and demonstrate that it is enzymatically active ( Figure 5). CD73 produces adenosine from adenosine monophosphate, which can exert immunomodulatory functions on T cells via adenosine receptors, suggesting a possible mechanism of action of 9-cis RA [22][23][24][25][26]. Maturation had no effect on mMoDCs as opposed to iMoDCs. The addition of 100 and 1000 nM 9-cis RA

'9-cis RA-DC' Show Functionally Active Expression of the Ecto-5'-Nucleotidase CD73 In Vitro
Immature dendritic cells are known to promote tolerance through the generation of regulatory T cells and a lack of T cell activation. Here, we show for the first time that retinoids upregulate the expression of the ecto-5 -nucleotidase CD73 in vitro and demonstrate that it is enzymatically active ( Figure 5). CD73 produces adenosine from adenosine monophosphate, which can exert immunomodulatory functions on T cells via adenosine receptors, suggesting a possible mechanism of action of 9-cis RA [22][23][24][25][26]. Maturation had no effect on mMoDCs as opposed to iMoDCs. The addition of 100 and 1000 nM 9-cis RA or 100 nM acitretin significantly upregulated the expression of CD73 (n = 7-22 independent donors) on the cell surface of MoDCs ( Figure 5). To test the enzyme activity of CD73, etheno-AMP degradation to etheno-adenosine was measured. CD73 was not only expressed on the cell surface but also active with enzymatic activity increasing with addition of 100 nM 9-cis RA by 1.75-fold and with 1000 nM 9-cis RA (p = 0.0286) by 4.3-fold compared to mMoDCs ( Figure 5). Using a cytokine array, we also examined the expression of various cytokines and chemokines in the cell supernatant (Supplementary Figure S7) and verified this by ELISA for selected cytokines. After stimulation with alitretinoin, dendritic cells showed a marked secretion of IL-10, an immunoinhibitory cytokine, and an increased secretion of CCL5 and CCL20 was shown. The effect of alitretinoin is more pronounced than for acitretin ( Figure 5).

Reduction of Proliferative Response of Allogenic Stimulated CD3 + T Cells by Retinoids
In view of the altered maturation markers, the higher expression of CD73 as well as the release of IL-10, the question arose whether retinoid-treated dendritic cells have an immunomodulatory effect on other cell types such as T cells. iMoDCs, mMoDCs and concentration-dependent matured "Alitretinoin-and Acitretin-DCs" were co-cultured with allogenic T cells at a ratio of 1:200. Proliferation was measured by BrdU incorporation after 6-7 days ( Figure 6). Interestingly, after maturation in the presence of retinoids, DCs showed a significantly lower potential to lead to T cell proliferation, which also underlines the immunomodulatory functional significance of retinoids. In equimolar doses, alitretinoin is significantly more effective than acitretin ( Figure 6). To confirm the inhibitory effect on T cell proliferation of 9-cis RA treated immune cells in vivo, PBMCs from patients suffering from CHE before and after week four, eight and twelve of treatment with 9-cis RA were isolated and cultivated with allogenic T cells isolated from healthy donors. At day six, the proliferation rate of leukocytes was measured by BrdU-incorporation. BrdU incorporation was analyzed with one-tailed paired Student's t-test. 9-cis RA treatment significantly decreased T cell proliferation following allogenic stimulation after week eight (p = 0.0285) and week twelve (p = 0.0346) when compared with T cell proliferation rates before the treatment ( Figure 6). In addition, the effect of alitretinoin treatment was assessed on the phenotype of CD14 + monocytes in patient samples, where the percentage of CD14 + CD86 + monocytes was significantly reduced after 12 weeks of treatment with oral 9-cis to baseline (p = 0.027). Moreover, CD14 + CD73 + cells were upregulated during treatment at week four (p = 0.0043) and week twelve (p = 0.016) (Supplementary Figure S9).

Discussion
Here, we present a detailed characterization of immunomodulatory effects of the dual retinoid receptor agonist alitretinoin in the context of CHE. Microarray data collected during the course of 12-week treatment provided a wide perspective on the potential mechanisms and affected pathways, highlighting targets for downstream in vitro analyses. Although the effects of retinoids on tissue remodeling pathways and metalloproteinases are well established, among the top regulated genes we also find hallmark chemokines CCL18 and CCL13, alongside inflammatory markers SAA1 or IDO1. The normalization in expression levels of these genes by week 12 is a strong indicator of reduced inflammation, while also outlining closely co-regulated genes with less clear roles in the pathology. Notably, AMPs such as PI3 or LCE3A are also downregulated, while DCD, an antimicrobial component of sweat typically downregulated in eczema [27], shows the opposite direction. Members of the defensin family were not found among significantly regulated genes in vivo; however, follow-up experiments on stimulated keratinocytes showed very high upregulation of DEFB4A and DEFB103A, both well-known players in inflammatory skin conditions, with remarkable downregulation upon alitretinoin treatment. Similar patterns were observed for RNASE7 but not hornerin (HRNR), while DCD was even upregulated upon alitretinoin treatment. This suggests that homeostatic levels of antimicrobial activity are restored, which aligns well with the clinical observation that alitretinoin treated patients do not present a significantly increased frequency of infections. Regarding skin structure and barrier constituents, SPRR family genes also show high levels of regulation and are increasingly investigated in skin disease [28,29]. Here, the eczema-associated gene SPRR3 shows prominent fold regulation, indicating barrier function-related alterations alongside FLG2, AQP9 and several keratins in vivo. On the other hand, investigating the expression of key hyaluronan synthase and hyaluronidase genes, alongside betacellulin, significant upregulation of synthases, particularly HAS2, is observed under alitretinoin treatment. Hyaluronidase genes are however not affected by alitretinoin, suggesting a positive effect towards matrix remodeling and cutaneous regeneration. Moreover, a decrease in the amount of hyaluronan fragments due to the imbalance of hyaluronan production and lower degradation may also lead to impaired inflammatory responses and a thickening of the epidermis, as previously observed [30].
In the analysis of significantly regulated genes using the FANTOM5 database, antigen presenting cells were identified to be the main cellular targets of alitretinoin, besides keratinocytes. The presented findings indicate that dendritic cells treated with 9-cis RA (alitretinoin) remain in an immature state with low co-stimulatory molecule expression (CD80 and CD86). The mMoDC phenotype presented here resembles the "TNF-α/CD40L" phenotype with an upregulation of key indicator genes previously described [31,32]. '9-cis RA-DCs' demonstrated a decreased expression of maturation-associated genes CD83, CD80, CD40, CCL17 and CCL22 in comparison to mMoDCs. Markers such as CD14 or CD163 were upregulated compared to mMoDCs. Altogether, on the transcriptional level, "9-cis RA-DCs" demonstrated a more iMoDC-like expression pattern and supported the hypothesis that RA-treatment shapes monocytes/DCs into immunomodulatory DCs, which have the ability to down-modulate T cell responses and control inflammatory processes. These observations were further supported by results obtained by flow cytometry of MoDCs which showed that 9-cis RA (alitretinoin) as well as acitretin have an inhibitory effect on the differentiation and maturation of DCs. An inhibitory effect on the maturation and differentiation with ATRA has been previously observed (1 µM and 10 nM). CD80 and CD86 were downregulated in both concentrations upon GM-CSF + IL-4 stimulation and additional TNF-α maturing [31]. Additionally, ATRA reduced the percentage of CD83 + HLA-DR + mature dendritic cells [32][33][34]. Overall, these results confirmed findings on other RAs and showed that 9-cis RA inhibited the maturation of dendritic cells by reducing co-stimulatory molecules CD80 and CD86. Impairment of CD86 and CD80 expression on macrophages and dendritic cells leads to a reduced activation and a lower cytokine production of T cells [35,36]. As a result of 9-cis RA (alitretinoin), 'retinoid-DCs' have a lower activation potential and might function as immunomodulatory cells. Further, the endogenous retinoid alitretinoin (9-cis RA) is not only effective in reducing maturation of TNF-α stimulated DCs in vitro, but it is more effective than the synthetic RAR agonist acitretin. To evaluate whether soluble factors may also support immunoregulation, we performed a cytokine array of the supernatants of mMoDCs and '9-cis RA-DCs'. We observed a significantly increased secretion of the anti-inflammatory cytokine IL-10 in the supernatants of '9-cis RA-DCs'. These results were confirmed by ELISA showing a higher efficacy to release IL-10 in response to alitretinoin when compared with equimolar doses of acitretin. Hence, transcriptomic, flow cytometric and secretory analyses pointed towards an immunomodulatory phenotype of 9-cis RA-DC. To further explore this notion, we investigated the adenosinergic pathway. ENTPD1 (CD39) and the downstream CD73 break down adenosine triphosphate into adenosine monophosphate and finally into adenosine [37,38]. Enzyme activity was tested, demonstrating not only the presence of CD73 but also the increased activity of the molecule on the surface of '9-cis RA-DCs'. Findings of the present study demonstrate for the first time that 9-cis RA and acitretin can induce CD73 on the surface of human moDCs and that the molecule is functionally active. These findings support the role of CD73 in the regulation of immune functions of DCs, and findings indicate a new role for retinoid receptor signaling by inducing CD73 and resulting in an immunomodulatory response [39][40][41]. To show the direct impact of '9-cis RA-DC', we demonstrated a significantly reduced allogeneic T-cell response in vitro and confirmed this inhibitory effect on T cell activation and proliferation using PBMCs of alitretinoin-treated CHE patient ex vivo. Hence, these findings strongly support that alitretinoin (9-cis RA) is inducing an immunomodulatory phenotype in dendritic cells in vitro and in vivo. Taken together, we demonstrate that the dual RAR and RXR agonist alitretinoin targets epidermal dysregulation and demonstrates strong immunomodulatory effects on antigen presenting cell functions.

Flow Cytometry Analysis
The lymphocyte subsets were analyzed by using specific antibodies and FACS Calibur flow cytometer (BD Biosciences, San Jose, CA, USA). At least 50,000 events were counted in every tube. The specific antibodies are listed in Supplementary Table S4

RNA Extraction
RNA extraction of the cells was performed using TRIzol ® reagent (Invitrogen, Carlsbad, CA, USA) and addition of 0.2 volume chloroform (Merck, Darmstadt, Germany). The cell suspension was vortexed and centrifuged (15 min, 12,000 rpm, 4 • C), and the supernatant was transferred to a new tube. RNA precipitation was performed using 0.5 volume isopropanol (Merck, Darmstadt, Germany). This was followed by incubation at −20 • C overnight. The supernatant was discarded. The RNA pellet was washed in 80% ethanol (Merck, Darmstadt, Germany) and centrifuged (12,000 rpm, 30 min, 4 • C) and dissolved in H 2 O. The RNeasy Mini Kit (QIAgen, Hilden, Germany) was used for RNA extraction from the tissue samples according to the manufacturer's protocol.

cDNA Microarray Analysis
Extracted RNA was hybridized to Agilent G3 Human Gene Expression Microarrays (G4851C) according to manufacturer protocols. Quality metrics were generated using ArrayQualityMetrics. Differential gene expression analyses were conducted on RMAnormalized probe intensity values summarized at the gene level in the R package limma, considering as cutoffs absolute log-fold change >0.5 and adjusted p-value < 0.05, with further narrowing down of gene lists for visualization. Functional enrichment analysis has been carried out using Panther v15. Functional gene networks were visualized in Cytoscape v3.6.

Quantitative Real Time (qPCR) Analysis
For qPCR, 18S RNA gene expression was used as an internal standard. Quantitative PCR was performed using the Applied Biosystems 7000 System/Quantstudio 6 and Power SYBR ® Green PCR Master Mix or TaqMan ® Universal PCR Master Mix (Applied Biosystems Inc., Foster City, CA, USA) according to the manufacturer's instructions.

Primer Design
Oligonucleotides for qPCR were either designed on mRNA sequences deduced from GenBank or obtained as TaqMan ® Gene Expression Assays by Applied Biosystems (Supplementary Table S5).

ELISA
Cytokine and chemokine concentrations in the supernatants were measured by enzymelinked immunosorbent assay (ELISA DuoSet, R&D Systems, Minneapolis, MN, USA) according to the manufacturer's protocols. The optical density of each well was determined using an ELISA reader (Multiscan Ascent, Thermo Fisher Scientific, Wilmington, DE, USA). Sample concentrations were calculated against standard curves and with the Graphpad Prism software (version 5.03, GraphPad Software, La Jolla, CA, USA).

Cytokine Array
We analyzed the supernatants of monocytic dendritic cells matured with TNF-α with or without alitretinoin [100 nM] and examined them with the Human Cytokine Array C5 (Ray Biotech, Norcross, GA, USA) according to the manufacturer's protocol. We performed the calculations using ImageJ software v1.42 (National Institutes of Health, Bethesda, MD, USA) against internal standards as a percentage of positive control. We considered a fold log 2 difference of more than 1.5 to be relevant.

Isolation of T Cells
Isolated PBMCs were resuspended in RPMI 1640. Depletion of monocytes and macrophages was carried out in 175 cm 2 cell culture flasks (Greiner Bio-One GmbH, Kremsmünster, Austria) at standard cell culture conditions as monocytes and macrophages will adhere to the wall of the plastic flask. Therefore, the non-adherent cells were suspended in 20 mL of RPMI 1640, centrifugated as above and resuspended in 4 mL of 1 × column wash buffer (R&D Systems, Inc., Minneapolis, MN, USA). T cell isolation was performed by the use of human T cell enrichment columns (R&D Systems, Inc., Minneapolis, MN, USA) according to the manufacturer's instructions. Finally, T cells were resuspended in appropriate medium.