Retinoic acid (RA) is the most prevalent and metabolically active vitamin A derivative in the human body that regulates a wide range of biological processes including cell proliferation and differentiation, embryogenesis and development, furthermore it supports the homeostasis of various tissues (1). The pleiotropic effect of RA are mediated by nuclear receptors either via genomic mechanisms by regulating transcription processes; or via non-genomic mechanisms by directly modifying the activity of various signal transduction pathways (2). Importantly, RA has also been shown to play a crucial role in the modulation of various immune functions. These include the differentiation and maturation of immune cells of both the innate- and adaptive immune system, such as lymphocytes, dendritic cells and macrophages (3, 4). Specifically in macrophages, RA has been implicated in various functions including phagocytosis, efferocytosis as well as balancing inflammatory responses through shaping cytokine production (3, 5, 6). Importantly, the modulatory effect of RA on inflammatory responses is highly context dependent as it is affected by the cell type, the stimuli and the tissue microenvironment.

Macrophages (MФs) are indispensable effector cells to modulate inflammatory responses. They form a diverse population of innate immune cells that are present in various tissues and organs where they either provide homeostatic functions as resident MФs, or support protective mechanism as infiltrating MФs by inducing or resolving inflammatory responses (7). Importantly, MФ inflammatory activity involves a complex molecular network including the cooperative crosstalk of pattern recognition receptors, signaling cascades and cytokine release. It is widely recognized that NOD-like receptors (NLRs), a family of cytosolic pattern recognition receptors that regulate various innate immune response, are highly involved in MФs functions (8).

NOD1 and NOD2 are prominent members of the NLR family as they were the earliest to be identified and characterized among the mammalian NLRs, but their roles in the inflammatory responses of MФ subsets are unknown. NOD1 and NOD2 were initially described to sense conserved, but distinct motifs of bacterial peptidoglycan (PGN), however, they can be activated by non-bacterial microorganisms (such as viruses, parasites and fungi) as well as danger-associated molecules (such as sphingosine-1-phosphate released during ER stress (9). Following activation, NOD1 and NOD2 participate in the regulation of cellular processes including metabolism, autophagy, cell death (10, 11) and the secretion of various cytokines (12–14).

Accumulating evidence shows that cytokine secretion by activated myeloid cells is highly modulated by RA (1). Accordingly, ATRA has been reported to enhance anti-inflammatory IL-10 secretion, but suppress IL-12 and TNF-α secretion in LPS-activated THP-1 cell line and splenic MФs (15, 16); while in human alveolar MФs and THP-1 cell line, LPS-induced IL-β expression is enhanced by ATRA (17, 18). Furthermore, we reported that in M-MФs, the LPS–induced IL-6 and IL-1β secretion is enhanced, while IL-10 is down-regulated by ATRA (19). While both NOD1 and NOD2 are important pattern recognition receptors that ignite inflammatory responses, including cytokine secretion in MФs, and the local microenvironment of MФs in several tissues may contain substantial amount of RA, the effect of RA on NOD1- and NOD2-mediated inflammatory cytokine and chemokine release has not been addressed before. Thus, we hypothesized that ATRA may modulate MФ inflammatory responses by interfering with NOD1/NOD2-induced cytokine responses in different MФ subsets.

To this end, we carried out a comparative analysis of typical inflammatory cytokine secretion induced by NOD1 and NOD2 in two distinct subpopulations of human monocyte-derived MФs that are differentiated in the presence of either GM-CSF or M-CSF, and that are commonly used in vitro models of pro-inflammatory- (GM-MФs) or pro-resolving cells (M- MФs), respectively. We focused on a few cytokines that are well-established to mediate the functional responses of different MФ populations as typical pro-inflammatory cytokines (IL-1β/IL-18, IL-6, TNF-α), or anti-inflammatory and pro-resolving cytokines (IL-10, IFNβ) (20, 21). IL-8 was specifically selected as a well-known downstream target of NOD1/NOD2 activation (22, 23). We show that the activation of NOD1 or NOD2 results in the release of different cytokine patterns by the two MФ subpopulations. Furthermore, we report for the first time that ATRA highly modulates cytokine secretion triggered by NOD1 and NOD2. Importantly however, the ATRA-induced changes of cytokine secretion by the two MФ subpopulations follow opposite tendency.

Macrophages are remarkably plastic immune cells, whose functional characteristics are highly determined by their origin, and shaped by the tissue microenvironment of their location, as well as the activating stimuli that trigger their effector functions (25, 26). In vitro, differentiation of macrophages from human primary monocytes using GM-CSF or M-CSF are widely used methods to study macrophage polarization and functions, as the generated GM-MФs or M-MФs possess inflammation inducing or inflammation resolving macrophage characteristics, respectively (27, 28). Here we show that the activation of these primary monocyte-derived MФ subpopulations via NOD1 and NOD2 cytosolic receptors results in characteristic profile of secreted cytokines that are typically released during macrophage inflammatory responses, and contribute to pro-inflammatory or pro-resolving/tissue regenerating mechanisms. We detected significantly higher release of the inflammatory IL-6, TNF-α, IL-8 and IL-1β cytokines from the pro-inflammatory GM-MФs compared to the pro-resolving M-MФs. However, the secretion of IL-18 pro-inflammatory cytokine was comparable between the two cell types. Furthermore, although the level of anti-inflammatory IL-10 was more pronounced in the M-MФs compared to the GM-MФs, the level of IL-10 secretion was also in a comparable range between the two subtypes. These results highlight that depending on the type of MФs affected, NOD1/NOD2-mediated cytokine secretion may drive potentially different induction and polarization of cells which is also likely to shape their interactions with cells of adaptive immunity.

NOD1 and NOD2 are the first described members of the cytosolic pattern recognition NOD-like receptor family. Upon recognition of respective ligands by NOD1 or NOD2 induces recruitment of RIPK2 which then mediates the activation of NF-kB and MAPK (ERK, p38, JNK) signaling pathways and drives various cellular processes including cytokine/chemokine secretion (29). RIPK2 activity may also induce the nuclear translocation of IRF7 transcription factor to promote type I IFN release (29, 30) which is an efficient upstream signaling cytokine of IL-10. Furthermore, RIPK2 may also recruit and directly bind caspase-1, which is one of the major regulatory enzymes of IL-1β and IL-18 secretion (10). Besides, NOD1/NOD2 may interact with other, inflammasome forming Nod-like receptors to induce IL-1β and IL-18 secretion (12). However, our comparative analysis also reveals substantial differences in the amount of the secreted typical cytokines between the MФ subtypes, indicating that activation of NOD1/NOD2 results in a subpopulation characteristic pattern of cytokine secretion.

Importantly, these results are in line with our previous findings on LPS-activated MФ subtypes (31, 32). However, while LPS is a harsh activator that is detected by TLR4 and drives high level of IL-10 secretion from M-MФs and high level of pro-inflammatory cytokine from GM-MФs, NOD1/NOD2 induce milder effects and results in a less pronounced secretion of the corresponding cytokines. Importantly, however, we show now that the most highly induced cytokine by NOD1/NOD2 activation is the pro-inflammatory interleukin-8 (IL-8). Surprisingly, NOD1/NOD2-induced IL-8 levels are similar to what we reported following LPS stimulation from these types of cells (26, 27). IL-8 is a crucial regulator of angiogenesis and a potent neutrophil chemotactic factor, hence plays an important role in physiological and pathological conditions (33). It was shown that at low concentrations, IL-8 gradient attracts neutrophils to the inflammatory foci, while at high, receptor-saturating concentration, it triggers neutrophils to release granule proteins and chromatin to form neutrophil extracellular traps (NETs) (33, 34). Hence, we speculate that while both MФ populations possess high capability of neutrophil attraction, NOD1/NOD2-activated GM-MФs may even have good potency to induce NET formation. RA, the most common, physiologically active metabolite of vitamin A, appears as an important tissue-derived signal that contributes to MФ polarization. However, the presence and distribution of RA shows high variability between the organs and even within a tissue, providing different microenvironments to the MФs. While NOD1 and NOD2 are one of the main cytosolic pathogen recognition receptors of MФs, the potential modulatory effect of ATRA on NOD1/NOD2-mediated responses by the different MФ subpopulations has not been addressed before. Here, we show that ATRA significantly modifies cytokine and chemokine release induced by NOD1 and NOD2 activation, and importantly, ATRA treatment results in various effects in the cytokine secretion ( Figure 4A ). In M-MФs, we observed that the NOD1-induced pro-inflammatory TNF-α, IL-1β and IL-18 were upregulated, while IL-6 and IL-8 were down-regulated by ATRA. Furthermore, the NOD1-induced anti-inflammatory IL-10 was also decreased by ATRA. In GM-MФs, however, under the same conditions, we observed completely the opposite effect of ATRA on cytokine secretion, with the exception of IL-6. Indeed, IL-6 was the only cytokine that, in each condition, was affected the same way by ATRA, as it was down-regulated in both MФ subtypes following both NOD1 and NOD2 activation. Importantly, NOD2-induced cytokine secretions were affected with the same tendency by ATRA as in the case of NOD1 activation, except for IL-8, as the secretion of this cytokine was enhanced both in M-MФs and GM-MФs. Cytokines released by MФs may act in an autocrine or paracrine manner to regulate the secretion of other cytokines. Like IFNβ is an upstream regulator of IL-10 (20), and both IFNβ and IL-10 are negative regulators of caspase-1-mediated IL-1β secretion (35, 36), which may explain the negative correlation between IFNβ/IL-10 and IL-1β/IL-18 secretion by the MФs in the presence of ATRA in our results. Of note, it was reported that ATRA induces the expression of IRF1 transcription factor in human myeloid cell lines and MФs (37). On the other hand, IRF1 was identified as a transcription regulator of Interferon-regulated genes (ISG) by facilitating chromatin accessibility for IRF3 binding which is a major driver of IFNβ expression (38). Nevertheless, ATRA-induced engagement of IRF1 in IFNβ secretion requires further comparative analysis in these MФ subpopulations.

ATRA may be recognized by various nuclear receptors (mainly by RAR, but also by PPAR or ROR) that may directly regulate gene expression in the nucleus (39). However, non-genomic effects of nuclear receptors have also been described, as in the cytosol, they can directly interact and modulate proteins of various signal transduction pathways (such as, AKT, NF-κB and ERK) (40). Importantly, many of these pathways overlap with those that are also affected by NOD1/NOD2 activation, rising the potential possibility of potential cross -talk mechanisms. Previous studies have reported context-dependent effect of ATRA on the expression of Toll-like receptors (TLRs) and RIG-like helicases (RLHs). Like ATRA downregulated the expression of TLR2 in monocytes (41) and TLR4 in the inflamed lung (6), while it upregulated TLR5 in THP-1 cells (42) and TLR4 expression in CaCo2 cells (43). ATRA alone upregulated TLR3, TLR4 and RIG-I intestinal epithelial cells, however it suppressed TLR3, TLR7 and RIG-I expression following viral infections (44, 45). Despite these observations, the mechanism of ATRA action has not been revealed. Regarding NLRs, the effect of ATRA has only been studied on the NLRP3 inflammasome activation. It has been reported that ATRA supplementation reduces the expression of NLRP3 inflammasome and caspase-1 in the brain tissue of a chronic alcoholic rat model (46). While, in our previous studies using in silico method, we found putative RAR-response elements in the promoter of NLRP3 and we detected increased NLRP3 protein level following ATRA treatment alone in human M-MФs. Furthermore, we found that ATRA shifts M-MФs metabolism toward glycolysis (19), which can highly effect MФs polarization. In this current study, we didn’t detect changes in the mRNA expression of NOD1/NOD2 (data not included) nor in the cytokine secretion following ATRA treatment alone, which suggests that ATRA acts as a modulator of NOD1/NOD2-induced cytokine production via downstream actions (and/or overall metabolic status of the cell) rather than inducing cytokine production alone.

It has been reported that ATRA treatment modulates several major intracellular signaling pathways, including MAPK signaling (such as ERK, JNK and p38). In human scleral fibroblasts, ATRA decreased ERK1/2 activation, while it increased JNK activation (47), however, in human MФs, ATRA enhanced the phosphorylation of ERK and attenuated that of p38 (19). The ATRA effect on NF-κB is also controversial, as it exhibits an inhibitory effect in THP-1, alveolar MФs and epithelial cells (6, 48), while in human keratinocytes, it induces the activation of both the NF-κB and p38 pathways (49). ATRA can also inhibit the PI3K/AKT signaling in human fibroblasts (50), and induces IRF-1 expression and its nuclear localization in human MФs (37, 51). Importantly, many of these pathways overlap with those that are also affected by NOD1/NOD2 activation, rising the possibility of cross-talk mechanisms ( Figure 4B ).

In summary, our results show that NOD1/NOD2-induced cytokine secretion by macrophages (MФ) is cell type dependent and it is highly modified by the nuclear receptor agonist ATRA. The implications of these observations are far reaching. NOD1 and NOD2, as sensors of microorganisms and cellular stress, have been identified as potential therapeutic targets in a range of autoimmune and autoinflammatory disorders (52) and in diabetes, asthma, atherosclerosis, various forms of CNS diseases and even cancer (29). In line with this, while it is well established that vitamin A and retinoic acid play important homeostatic and regulatory roles in multiple tissues, based on our data, MФs activated by NOD1/NOD2-dependent or independent stimuli are likely to be profoundly modulated by ATRA. Major immune organs also require a steady dietary intake of vitamin A to maintain their functions and vitamin A deficiency has been associated with dysbiosis, frequent infectious diseases and reduced immune response to vaccines, although the underlying mechanisms have remained unclear to date. Our data shed light on this issue, by showing the potent and diverse modulatory role of ATRA on NOD1/NOD2-induced cytokine secretion. Based on previous reports, we suggest that this modulation is mediated, in part, via interrelated signaling pathways and altered metabolic status of the cell. Hence, we speculate that ATRA may function as a general regulator of cytokine secretion induced by various PRRs. While understanding the precise mechanism of ATRA action on NOD1/NOD2-induced mechanisms in different tissues and MФ populations requires further studies, our results highlight the therapeutic relevance of these pathways in a number of inflammatory conditions, which represent an increasing burden to society.