Serotonin drives the acquisition of a profibrotic and anti-inflammatory gene profile through the 5-HT7R-PKA signaling axis

Peripheral serotonin (5-hydroxytryptamine, 5-HT) regulates cell growth and differentiation in numerous cell types through engagement of seven types of cell surface receptors (HTR1–7). Deregulated 5-HT/HTR levels contribute to pathology in chronic inflammatory diseases, with macrophages being relevant targets for the physio-pathological effects of 5-HT. In fact, 5-HT skews human macrophage polarization through engagement of 5-HT2BR and 5-HT7R receptors. We now report that 5-HT primes macrophages for reduced pro-inflammatory cytokine production and IFN type I-mediated signaling, and promotes an anti-inflammatory and pro-fibrotic gene signature in human macrophages. The acquisition of the 5-HT-dependent gene profile primarily depends on the 5-HT7R receptor and 5-HT7R-initiated PKA-dependent signaling. In line with the transcriptional results, 5-HT upregulates TGFβ1 production by human macrophages in an HTR7- and PKA-dependent manner, whereas the absence of Htr7 in vivo results in diminished macrophage infiltration and collagen deposition in a mouse model of skin fibrosis. Our results indicate that the anti-inflammatory and pro-fibrotic activity of 5-HT is primarily mediated through the 5-HT7R-PKA axis, and that 5-HT7R contributes to pathology in fibrotic diseases.

Serotonin (5-hydroxytryptamine, 5-HT) is a monoamine neurotransmitter derived from L-tryptophan via a rate-limiting reaction catalyzed by tryptophan hydroxylases (TPH1 in periphery, TPH2 in brain) 1,2 . Brain-derived 5-HT controls mood, behavior, sleep, blood pressure and thermoregulation 3 whereas peripheral 5-HT regulates vascular and heart functions 4 and gastrointestinal mobility 5 . Enterochromaffin cells in the gastrointestinal tract produce 90% of the human body's 5-HT 6 , which is actively taken up by blood platelets and stored in dense granules. Upon platelet activation, released 5-HT modifies vascular smooth muscle tone, promotes proliferation of smooth muscle cells 7 , hepatocytes 8 and endothelial cells 9 , and critically contributes to wound healing. All 5-HT actions are exerted through engagement of seven types of receptors (5-HT1-7R) which, except for 5-HT3R, belong to the G protein-coupled superfamily of receptors 10 . 5-HT also functions as a regulator of immune and inflammatory responses 11 . 5-HT modulates T-cell activation, proliferation and differentiation 12 and modifies cytokine production in a cell type-dependent manner [13][14][15][16] . The regulatory role of 5-HT in inflammation is illustrated by the pathological consequences of its altered production or absence in chronic inflammatory diseases. 5-HT contributes to Pulmonary Arterial Hypertension (PAH) 17 , atopic dermatitis 18 and systemic sclerosis 19 , and modifies the outcome of inflammatory gut disorders [20][21][22][23][24][25] . 5-HT also favors colon cancer angiogenesis 26 and neuroendocrine neoplasms proliferation 27 , and its absence increases pathologic scores in collagen-induced arthritis 28 . The close link between 5-HT and chronic inflammatory pathologies 29 is in line with the anti-inflammatory actions of selective 5-HT reuptake inhibitors (SSRI) like fluoxetine 30,31 . Further supporting the 5-HT/inflammation link, 5-HT2BR has been shown to mediate the effects of 5-HT on tissue fibrosis 19 and PAH 17 , 5-HT7R mediates the 5-HT contribution to gut inflammation in IBD models 20,22,23 , and 5-HT3R or 5-HT4R ligands reduce inflammatory reactions during postoperative ileus 32,33 .
Macrophages are critical for maintaining tissue homeostasis and promoting the initiation and resolution of inflammatory processes. The balance between pro-and anti-inflammatory (resolving) macrophages is required for restoring tissue homeostasis [34][35][36] , and its deregulation leads to chronic inflammatory diseases 37,38 . Since macrophages rapidly adapt their functions to micro-environmental stimuli (e.g., cytokines, growth factors, pathogenand damage-associated molecular patterns) [34][35][36][37][38] , targeting macrophages is currently proposed as a therapeutic approach for chronic inflammatory diseases 37 . Not surprisingly, some of the effects of 5-HT on inflammation are mediated through direct and indirect actions on myeloid cells 17,26,32,33 . Further supporting this relationship, bone marrow-derived cells are responsible for the contribution of 5-HT and 5-HT7R to intestinal inflammation 22 , and macrophages mediate the anti-inflammatory action of SSRI 30,39 .
We have previously demonstrated that human pro-inflammatory and anti-inflammatory macrophages 40-43 exhibit a distinct profile of 5-HT receptors, and that 5-HT2BR and 5-HT7R shape macrophage effector functions towards the anti-inflammatory side 44 . To dissect the molecular mechanisms underlying the inflammation-modulating action of 5-HT, we undertook the determination of the 5-HT-dependent transcriptome of human macrophages. 5-HT rapidly altered the human macrophage transcriptome towards a growth-promoting, anti-inflammatory and pro-fibrotic gene profile, whose acquisition was dependent on the 5-HT7R -PKA signaling axis. Moreover, and in line with these findings, Htr7 −/− mice exhibited significantly reduced macrophage accumulation and collagen deposition in a bleomycin-induced model of skin fibrosis.

5-HT promotes the expression of an anti-inflammatory gene profile and inhibits pro-inflammatory cytokine production.
To determine the 5HT-dependent transcriptome of human macrophages, a global gene expression analysis was performed on human monocyte-derived macrophages (M-MØ) exposed to 5-HT for 6 hours (Fig. 1A), a time at which exposure to 5-HT modifies the LPS-induced production of inflammatory cytokines (see below). Transcriptional profiling revealed that 5-HT increases the expression of 170 annotated genes (p < 0.01; log 2 ratio 5-HT/untreated ≥0.6) and downregulated the expression of 41 genes (p < 0.01; log 2 ratio 5-HT/untreated ≤ −0.6) (Supplementary Table I). Further filtering (normalized expression levels higher than 100 in untreated or 5-HT-treated M-MØ, and p < 0.0017 for the M-MØ vs. M-MØ + 5-HT comparison) identified 74 genes upregulated and 14 genes downregulated upon 5-HT exposure (Table 1) (Fig. 1B). Analysis of independent 5-HT-treated M-MØ samples confirmed the microarray data and revealed distinct kinetics for the Results are expressed as the mRNA level of each gene relative to the GAPDH mRNA level in the same sample (n = 3; *p < 0.05). (D) Gene ontology analysis of the set of genes upregulated by 5-HT in M-MØ, as determined using the ENRICHR tool and the indicated databases [combined score = log(p-value) x z-score]. (E) GSEA on the "t statistic-ranked" list of genes obtained from the 5-HT-treated M-MØ versus M-MØ limma analysis, using the proinflammatory (left panel) and anti-inflammatory (right panel) gene sets previously defined 48 . Vertical black lines indicate the position of each of the genes comprising the "Pro-inflammatory" and "Anti-inflammatory" gene sets. (F) Production of IL-12p40, TNFα and Activin A by non-treated (-) or 5-HT-pretreated (6 h) LPSstimulated (18 h) M-MØ, as determined by ELISA (n = 12; *p < 0.05; **p < 0.01). 5-HT-upregulated genes. As shown in Fig. 1C, EREG expression was maximally upregulated only two hours after exposure to 5-HT, while other genes (TREM1, MET, THBS1) exhibited maximal level of up-regulation 4-6 hours after 5-HT treatment. Therefore, 5-HT modifies the gene signature of human macrophages and its effects can be detected as early as 2 hours after exposure to the neurotransmitter.
Since 5-HT7R engagement leads to increased intracellular levels of cAMP 53 , whose effectors include PKA and "Exchange factor directly activated by cAMP" (Epac), we next assessed the effect of cAMP analogs (BrcAMP, dBrcAMP) and modifiers of cAMP-initiated signaling (6Bnz, 8cPT, RP8) on the acquisition of the 5-HT-dependent gene signature in M-MØ. The cAMP analogs BrcAMP and dBrcAMP greatly enhanced the expression of genes upregulated by 5-HT (PDE2A, TREM1, THBS1, MET) (Fig. 3A). Similar changes in the expression of these genes were seen in M-MØ exposed to the PKA-specific activator 6-Bnz-cAMP (Fig. 3A), while the Epac activator 8-pCPT had no effect (Fig. 3A). Moreover, the positive effect of 5-HT on the expression of 5-HT-upregulated genes was significantly blunted or inhibited in the presence of the PKA inhibitor RP8 (Fig. 3B). Therefore, 5-HT shapes M-MØ gene expression primarily via engagement of 5-HT7R and activation of PKA.
Since the 5-HT7R-PKA axis mediates the expression of the 5-HT-dependent M-MØ transcriptome, we next evaluated whether this signaling axis contributes to the inhibitory effect of 5-HT on the LPS-induced production of inflammatory cytokines by M-MØ, which produce undetectable levels of TNFα and IL-12p40 in the absence of stimulation 44 . Both 5-HT7R antagonists (SB269970 and SB258719) dose-dependently reversed the inhibitory action of 5-HT on the production of TNFα and IL-12p40 induced by LPS (Fig. 4A), while the 5-HT7R agonist AS19 (1 μM) mimicked the effect of 5-HT on the LPS-induced expression of TNFα and IL-12p40 (Fig. 4B). Furthermore, the inhibitory effect of 5-HT on the LPS-induced production of TNFα and IL-12p40 was significantly reduced in the presence of the PKA inhibitor RP8 (Fig. 4C). Thus, 5-HT conditions macrophages for a diminished production of pro-inflammatory cytokines primarily via engagement of 5-HT7R and activation of PKA. As a whole, this set of results demonstrates that the 5-HT7R-PKA axis mediates the acquisition of the 5-HT-dependent gene and cytokine profile in human macrophages.  (Fig. 5A). In agreement with the GSEA data, the mRNA levels of type I IFN-dependent genes (CXCL10, CXCL11, IDO1, RSAD2, IL27 and IFIT2) (Fig. 5B) and the production of the type I IFN-dependent chemokine CXCL10 (Fig. 5C) were significantly reduced in LPS-treated M-MØ that had been pre-treated for 6 hours with 5-HT. Moreover, M-MØ also produced reduced levels of CXCL10 in response to IFNβ if previously exposed to 5-HT (Fig. 5D). The decrease in LPS-or IFNβ-induced CXCL10 production caused by 5-HT exposure correlated with a reduced activation of STAT1 in response to either LPS or IFNβ (Fig. 5E,F), further supporting the 5-HT´s ability to limit macrophage responses to type I IFN. Importantly, the ability of 5-HT to limit the expression of type I IFN-responsive genes is also mediated by the 5-HT7R-PKA axis, as it was abolished by either 5-HT7R antagonists (SB269970 and SB258719) or the PKA inhibitor RP8 (Fig. 5G).

5-HT also promotes pro-fibrotic gene expression in human macrophages. Additional GSEA
results evidenced that a short-term (6 h) exposure to 5-HT causes a global upregulation of the "Angiogenesis" gene sets, as well as a very significant downregulation of the "Cholesterol homeostasis" and "Fatty Acid Metabolism" genes sets (Supplementary Figure 2). Furthermore, 5-HT treatment led to a significant upregulation of the "TGFβ signaling" gene set (FDR q-val = 0.126) (Fig. 6A). In line with the GSEA data, 5-HT was found to induce a significant increase in TGFB1 mRNA (Fig. 6B). Like most 5-HT-upregulated genes, the enhanced expression of TGFB1 mRNA was prevented by 5-HT7R antagonists (Fig. 6B) and did not occur in the presence of the PKA inhibitor RP8 (Fig. 6C). More importantly, 5-HT treatment of M-MØ resulted in a significantly increase production of TGFβ1 (Fig. 6D). Therefore, engagement of 5-HT7R by 5-HT promotes the production of TGFβ1 as well as the acquisition of a pro-fibrotic gene signature in human macrophages.
Lack of Htr7 results in diminished macrophage infiltration and pathology in a mouse model of skin fibrosis. Macrophages exert critical functions during tissue repair after injury but their deregulated polarization can also result in excessive scarring and chronic fibrosis 54,55 . In fact, macrophages are important cells for the onset of scleroderma 56,57 and pulmonary fibrosis 58,59 , and their deregulated polarization results in fibrosis in muscle 60 . Given the 5-HT/5-HT7R-upregulated expression of TGFβ1 in human macrophages, and to analyze the contribution of 5-HT7R to skin fibrosis development, we assessed the effect of Htr7 gene ablation in the  To this end we first determined Htr7 expression of in mouse macrophages in vitro and in vivo. Htr7 mRNA was readily detected in liver Kupffer cells and peritoneal F4/80 + macrophages, with Htr7a being the predominant splicing isoform in both cases (Supplementary Figure 3A). Unlike human monocyte-derived macrophages, Htr7 mRNA was only detected in murine pro-inflammatory GM-MØ, where also Htr7a was the predominant isoform (Supplementary Figure 3B). Also in marked contrast with human macrophages, where HTR7 mRNA is greatly reduced in response to LPS 44 , macrophage Htr7 mRNA was greatly upregulated after LPS stimulation (Supplementary Figure 3C).
Once the presence of Htr7 mRNA had been demonstrated in murine macrophages, the effect of Htr7 gene deletion in a mouse model of fibrosis was assessed. Whereas no histological differences in collagen stained area were observed between saline-treated Htr7 −/− and WT mice, a significant increase in skin collagen content was observed after bleomycin injection in WT mice (Fig. 6E). Conversely, Htr7 −/− mice appeared protected from bleomycin-induced fibrosis since significantly reduced collagen area content was observed in bleomycin-treated Htr7 −/− skin compared to WT skin (Fig. 6E). In addition, myofibroblast differentiation evaluated as α-SMA expression, was also significantly reduced in Htr7 −/− compared to WT mice (Fig. 6E). Likewise, a significantly lower infiltration of F4/80 + cells was found in bleomycin-treated Htr7 −/− mice (Fig. 6E). In addition, bleomycin treatment significantly enhanced α-SMA expression in WT mice but not in Htr7 −/− mice (Fig. 6E). Therefore, and in line with the transcriptional results in human macrophages, 5-HT7R expression contributes to macrophage accumulation and fibrosis in the bleomycin model of skin fibrosis.

Discussion
Macrophages exhibit a huge phenotypic and functional heterogeneity, and their effector functions ("polarization state") are determined by the integration of the intracellular signals initiated by the surrounding extracellular cues and stimuli. Elimination of inflammatory insults requires the balanced and sequential dominance of pro-inflammatory and anti-inflammatory/resolving macrophages 62 whose deregulation leads to chronic inflammatory diseases [63][64][65] . Given their critical role in the initiation and resolution of inflammation, modulation of the macrophage polarization state has been proposed as a therapeutic approach for numerous chronic inflammatory pathologies 37 . While the physiological processes regulated by 5-HT (cell proliferation, tissue repair, inflammation) are also critically modulated by macrophages 66 , the influence of 5-HT on macrophage plasticity is not yet completely understood 11 . Based on the ability of 5-HT to modulate the macrophage cytokine profile 44 , we undertook the determination of the 5-HT-dependent human macrophage transcriptome. Our results indicate that 5-HT conditions macrophages for impaired production of pro-inflammatory cytokines and type I IFN-inducible cytokines, and also shapes the macrophage gene signature towards the acquisition of an anti-inflammatory and pro-fibrotic gene profile, with all these effects being primarily mediated by the 5-HT7R-PKA axis.
The link between 5-HT and fibroblast proliferation/fibrosis has been known to be primarily mediated by the 5-HT2BR receptor, which induces extra-cellular matrix synthesis in fibroblasts 19 and whose over-activation leads to excessive proliferation of cardiac valves fibroblasts and severe cardiac pathologies [67][68][69] . Our results reveal that 5-HT7R is a primary mediator of the pro-fibrotic action of 5-HT on human macrophages because 5-HT7R antagonists and inhibitors of 5-HT7R-initiated signaling block the acquisition of the pro-fibrotic gene signature as well as the 5-HT-upregulated production of TGFβ1 in human macrophages. The pro-fibrotic action of 5-HT7R is further supported by in vivo results, as Htr7 −/− mice exhibit diminished pathology (lower collagen deposition and F4/80 + cell infiltration) in the bleomycin-induced model of skin fibrosis. Macrophages play a critical role in fibrotic processes 54 . In fact, elimination of macrophages expressing Folr2, a myeloid-specific protein exclusively expressed by anti-inflammatory M-MØ 70 , greatly diminishes pathology in the bleomycin-induced experimental skin fibrosis 71 . Therefore, it can be speculated that the absence of 5-HT7R in mouse macrophages contributes to the diminished skin pathology we have observed. Alternatively, other macrophage 5-HT7R-dependent functions might contribute to the reduced pathology seen in Htr7 −/− mice. Specifically, impaired migration of myeloid cells to the bleomycin-treated tissue might explain the reduced accumulation of F4/80 + cells in the damaged skin of Htr7 −/− mice. This explanation would fit with the impaired migration of mouse bone marrow-derived dendritic cells from Htr7 −/− mice 72 and is supported by the significant 5-HT-dependent upregulation of genes involved in cell chemotaxis (Fig. 1D). However, these interpretations should be taken cautiously, because expression of 5-HT7R appears to be differentially regulated in mouse and human macrophages: 5-HT7R in human macrophages is greatly downregulated by pathogenic stimuli like LPS 44 whereas Ht7r expression in mouse myeloid cells is greatly upregulated by LPS 72 . In line with our findings, the 5-HT7R agonist LP-44 has been reported to reduce pro-inflammatory cytokine production in vivo in a carbon tetrachloride-induced rat model of liver fibrosis 73 , where the agonist was, however, also capable of inhibiting Tgfb1 mRNA 73 . The latter discrepancy between these results and ours might be explained by the use of different 5-HT7R agonists (5HT versus a chemical agonist) and animal models of fibrosis (bleomycin-induced mouse skin fibrosis versus carbon tetrachloride-induced rat liver fibrosis), but indicate a significant involvement of 5-HT7R in fibrotic responses.
In peripheral tissues, 5-HT7R is expressed in smooth muscle cells of blood vessels and the gastrointestinal tract, as well as in kidney, liver, pancreas and spleen [74][75][76] . Within the immune system, the functional consequences of 5-HT7R ligation are far from clear. 5-HT7R on mouse naive T cells contributes to early T-cell activation 12 , whereas 5-HT7R on human monocytes has been reported to inhibit 77 or enhance 78 LPS-induced pro-inflammatory cytokine release. A similar controversy appears to exist regarding the in vivo role of Htr7, whose deletion has been reported to improve 22 or exacerbate 23 mouse gut inflammation. Our transcriptional results clarify this issue and demonstrate that 5-HT7R, acting through PKA, conditions macrophages towards the acquisition of a more anti-inflammatory and pro-fibrotic polarization state and for an impaired production of pro-inflammatory functions. As a consequence, our results place 5-HT7R as a potentially relevant molecule for modulation of macrophage effector functions under physiological and pathological settings.
The anti-inflammatory gene profile promoted by 5-HT/5-HT7R is compatible with the known signaling ability of 5-HT7R. Although 5-HT7R has been shown to activate NFκB in monocytes 12,78 , and ERK1/2 79 , Akt 80 , p38MAPK and protein kinase Cε 81 , or the Cdc42-Gα12-SRF axis 82 in various cell types, 5-HT7R couples positively to adenylate cyclase through activating Gαs, leading to increased cAMP levels and activation of PKA and Epac1/2 74 . These cell-specific differences in 5-HT7R signaling might derive from the presence of distinct splicing isoforms or heterodimerization with other receptors 83,84 . In the case of human macrophages, where the three splicing isoforms can be detected at the mRNA level (data not shown), our results clearly establish PKA, and not Epac1/2, as a major effector of 5-HT7R, as most of the transcriptional actions of 5-HT7R engagement by 5-HT can be abolished by PKA inhibitors and mimicked by PKA activators. Furthermore, the connexion between 5-HT7R and PKA activation fits well with the global anti-inflammatory skewing induced by 5-HT via 5-HT7R because PKA leads to CREB activation, which favours the acquisition of a "M2 polarization state" 62 . In addition, cAMP-initiated signaling limits the effector functions of pro-inflammatory stimuli 85,86 , which is consistent with the reduced production of proinflammatory cytokines seen with 5-HT7R activation.
The ability of 5-HT to promote the acquisition of a pro-fibrotic and anti-inflammatory signature in human macrophages has relevant pathophysiological implications. While normal peripheral blood levels of 5-HT range between 0.7 and 2.5 μM 87-91 , 5-HT concentrations at the neuronal synapse have been estimated to reach the millimolar range 92 , and the available platelet serotonin is close to 20 μM 93 . Since platelets release serotonin during inflammation as a means to activate endothelial cells and promote leukocyte adhesion and recruitment 93 , our findings on the macrophage polarizing effects of 10 μM 5-HT are physiologically relevant, especially at the initial stages of inflammatory responses. Besides, and regarding pathology, the serum levels of 5-HT detected in metastatic carcinoid tumors exceed 30 μM 91 , thus pointing to 5-HT as a factor that contributes to polarization of macrophages in serotonin-producing neuroendocrine tumors. Therefore, 5HT7R-regulated genes should be considered as potential targets to modify macrophage polarization under pathological settings.

Materials and Methods
Ethical statement. Ethical approvals for all blood sources and processes used in this study have been approved by the Centro de Investigaciones Biológicas Ethics Committees. Subjects gave written informed consent in accordance to the Declaration of Helsinki. All experiments were carried out in accordance with the approved guidelines and regulations.
All experiments on mice were conducted according to the Spanish and European regulations on care and protection of laboratory animals and were approved by the Consejo Superior de Investigaciones Científicas ethics committee.
Quantitative real-time RT-PCR. Oligonucleotides for selected genes were designed according to the Universal Probe Roche library system (Roche Diagnostics) for quantitative real time PCR (qRT-PCR). Total RNA was extracted using the RNeasy kit (Qiagen), retrotranscribed, and amplified in triplicates. Results were expressed relative to the expression level of TBP RNA. When indicated, results were expressed relative to the mean of the expression level of endogenous reference genes HPRT1, TBP and RPLP0. In all cases the results were expressed using the ΔΔCT method for quantitation.
Microarray analysis. Global gene expression analysis was performed on RNA obtained from untreated or 5-HT-treated (10 μM, 6 h) M-MØ from three independent healthy donors. RNA was isolated using the RNeasy Mini kit (Qiagen, Germantown, MD) and labeled RNA used as a hybridization probe on Whole Human Genome Microarrays (Agilent Technologies, Palo Alto, CA). Only probes with signal values > 60% quantile in at least one condition were considered for the differential expression and statistical analysis. Statistical analysis for differential gene expression was carried out using empirical Bayes moderated t test implemented in the limma package (http://www.bioconductor.org). The p values were further adjusted for multiple hypotheses testing using the Benjamini-Hochberg method to control the false discovery rate 96 . All the above procedures were coded in R (http://www.r-project.org). Microarray data were deposited in the Gene Expression Omnibus (http://www.ncbi. nlm.nih.gov/geo/) under accession GSE94608. Differentially expressed genes were analysed for annotated gene sets enrichment using the online tool ENRICHR (http://amp.pharm.mssm.edu/Enrichr/) 97,98 . Enrichment terms were considered significant when they had a Benjamini-Hochberg-adjusted p value < 0.05. For gene set enrichment analysis (GSEA) 99 , the gene sets contained in the Molecular Signature databases available at the GSEA website, and the previously defined "Pro-inflammatory gene set" and "Anti-inflammatory gene set" 48 , which contain the top and bottom 150 probes from the GM-MØ versus M-MØ limma analysis of the microarray data in GSE68061 (ranked on the basis of the value of the t statistic), were used. Western blot. Cell lysates were obtained in RIPA buffer (10 mM Tris-HCl pH 8, 150 mM NaCl, 1% NP-40, 2 mM Pefabloc, 2 mg/ml aprotinin/antipain/leupeptin/pepstatin, 10 mM NaF, and 1 mM Na 3 VO 4 ). 20 µg of cell lysate was subjected to SDS-PAGE and transferred onto an Immobilon polyvinylidene difluoride membrane (Millipore). Protein detection was carried out using antibodies against phosphorylated STAT1 (BD Biosciences, CA, USA) and a monoclonal antibody against GAPDH (sc-32233, Santa Cruz, CA, USA).

Statistical analysis.
For comparison of means, and unless otherwise indicated, statistical significance of the generated data was evaluated using the Student t test. In all cases, p < 0.05 was considered as statistically significant.
Data Availability. All data generated or analysed during this study are included in this published article.