Expression of SAA1, SAA2 and SAA4 genes in human primary monocytes and monocyte-derived macrophages

Circulating serum amyloid A (SAA) is increased in various inflammatory conditions. The human SAA protein family comprises the acute phase SAA1/SAA2, known to activate a large set of innate and adaptive immune cells, and the constitutive SAA4. The liver synthesis of SAA1/SAA2 is well-established but there is still an open debate on extrahepatic SAA expression especially in macrophages. We aimed to investigate the ability of human primary monocytes and monocyte-derived macrophages to express SAA1, SAA2 and SAA4 at both the transcriptional and protein levels, as previous studies almost exclusively dealt with monocytic cell lines. Monocytes and derived macrophages from healthy donors were stimulated under various conditions. In parallel with SAA, pro-inflammatory IL1A, IL1B and IL6 cytokine expression was assessed. While LPS alone was non-effective, a combined LPS/dexamethasone treatment induced SAA1 and to a lesser extent SAA2 transcription in human monocytes and macrophages. In contrast, as expected, pro-inflammatory cytokine expression was strongly induced following stimulation with LPS, an effect which was dampened in the presence of dexamethasone. Furthermore, in monocytes polarized towards a pro-inflammatory M1 phenotype, SAA expression in response to LPS/dexamethasone was potentiated; a result mainly seen for SAA1. However, a major discrepancy was observed between SAA mRNA and intracellular protein levels under the experimental conditions used. Our results demonstrate that human monocytes and macrophages can express SAA genes, mainly SAA1 in response to an inflammatory environment. While SAA is considered as a member of a large cytokine network, its expression in the monocytes-macrophages in response to LPS-dexamethasone is strikingly different from that observed for classic pro-inflammatory cytokines. As monocytes-macrophages are major players in chronic inflammatory diseases, it may be hypothesized that SAA production from macrophages may contribute to the local inflammatory microenvironment, especially when macrophages are compactly organized in granulomas as in sarcoidosis.

The corresponding SAA proteins are highly conserved among species, indicating important biological functions [9]. However, whether SAA1 and SAA2 have distinct roles in physiology and in the pathophysiology of human disorders is still unknown [10]. In response to tissue injury or infection, a dramatic increase in SAA plasma levels (SAA1 and SAA2) is observed; consequently, SAA is considered as a marker of inflammation [11]. During acute phase, high levels of circulating SAA1 and SAA2 displace apolipoprotein A1 from high density lipoprotein (HDL), becoming an HDL-associated apolipoprotein [12][13][14][15]. This leads to a loss of the antiinflammatory properties of HDL [16]. SAA proteins are also the precursors of Amyloid A (AA) fibril formation, responsible for tissue and organ amyloidosis [17,18]. Recombinant human SAA (rhSAA) has been shown to induce chemotactic activity in neutrophils, dendritic cells, monocytes and T lymphocytes [8,11,[19][20][21] and secretion of cytokines [22,23]. These pleiotropic functions of SAA are thought to be mediated through signalling pathways following engagement of cell-surface receptors like the Toll-like and scavenger receptors [24][25][26]. Overall, based on these effects, SAA are therefore considered as pro-inflammatory proteins. However, SAA is not only an acute inflammation response mediator, but also plays a significant role in the pathogenesis of various chronic diseases at the crossroad of autoimmunity and autoinflammation such as Behcet's disease [27,28] or sarcoidosis [29,30].
The secretion of SAA by the liver as for all the acute phase proteins is well established. Indeed, hepatic cells, such as the HepG2 cell line secrete SAA following stimulation with proinflammatory cytokines like interleukin (IL) IL1β, IL6 and tumor necrosis factor alpha (TNFα) [31,32]. However, the possibility that extrahepatic sites of SAA synthesis exist is supported by expression of SAA in epithelial components of a variety of normal tissues [33]. Particularly challenging is the potential extrahepatic synthesis of SAA by immune cells. Supportive to this hypothesis is the detection of SAA mRNA by in situ hybridization (ISH) in macrophage-derived foam cells present in atherosclerotic lesions [34][35][36]. Furthermore, SAA proteins have been detected by immunohistochemistry (IHC) in macrophage accumulation areas in rheumatoid arthritis joints [37,38] and in macrophage-derived cells constituting the main components of epithelioid and giant cell granulomas in sarcoidosis [29]. It has been suggested that a potential local production of SAA by macrophage-derived epithelioid cells can lead to granuloma formation and maintenance [39][40][41]. However, although the localization of SAA in macrophages from various inflammatory conditions [29,33,[42][43][44] suggested a possible local production of SAA by macrophages at sites of inflammation, it is still unclear whether the intracellular SAA observed by IHC in macrophages is a result of internalization of the circulating liver-produced SAA or is due to expression by tissue macrophages. Studies on SAA synthesis by macrophages have almost exclusively dealt with human or rodent cell lines. SAA gene transcription has been demonstrated in the human macrophagelike THP1 and U-937 cells treated with LPS-dexamethasone and/or D3 vitamin or PMA (Phorbol Myristate Acetate) [43,45] and in the rodent J-774 cell line following activation by LPS [46]. To our knowledge, there are no extensive studies carried out in naïve human monocytes and macrophages. Therefore, after stimulation by various pro-inflammatory conditions, changes in SAA1, SAA2 and SAA4 gene expression at both the transcriptional and protein levels were evaluated during maturation of freshly collected human monocytes into macrophages.

Isolation of human peripheral blood monocytes and monocyte-derived macrophages
Peripheral blood mononuclear cells (PBMCs) were isolated from buffy coats, from apparently healthy blood donors, provided by the "Etablissement Français du Sang, ((EFS); convention N1 5EFS012 between "INSERM" and EFS; informed consent was obtained from all subjects). PBMCs were isolated using Ficoll density gradient as described previously [47]. Monocytes were selected from PBMCs by 1 h adherence at 37˚C and cultured for 24 h in the presence of RPMI 1640 (Cat No. 21875-091, Fischer, Paisley, Scotland, UK) complemented with 10% pooled human serum (HS) (Cat No. C15-021, PAA), L-glutamine (Cat No. 25030-024, Gibco), penicillin and streptomycin (Cat No. 11548876, Gibco) (complete medium). Monocyte-derived macrophages were obtained after 7, 14 and 21 days in culture of adherent monocytes in the presence of complete medium (named respectively 7-day, 14-day and 21-day culture macrophages from here after). Cell viability was tested using the Cell Counting 8 kit (Cat No. 277CK04-05, Tebu Bio).
Alternative to isolation by adherence, monocytes from 2 donors were isolated from PBMCs using CD14 positive selection (130-050-201, Miltenyi Biotec, Bergisch Gladbach, Germany) following the manufacturer's instructions. Cells were treated with LPS and/or dexamethasone for 24 h in the absence of HS as described below.

Monocyte polarization
In selected experiments, monocytes were polarized in order to obtain M1 or M2 phenotypes as described before [48]. Briefly, adherent monocytes were treated with either 100 ng/ml interferon (IFN)-γ (M1 phenotype) or 10 ng/ml of IL4 and IL13 (M2 phenotype) for 24 h. Cells were additionally treated with LPS and dexamethasone as described below.

Monocyte and macrophage treatments
Monocytes immediately after adherence, or 7-day, 14-day, or 21-day culture macrophages were treated with 0.1 or 2 μg/mL LPS or 0.01, 1, 10 μM dexamethasone (Mylan, Saint Priest, France) for 24 h in the absence of HS unless otherwise stated. Results of time-and concentration-dependent experiments are presented in the supplementary data. Based on these results (S1 and S2 Figs), most of the treatments were done, unless otherwise specified, using 100 ng/ mL LPS and 1μM dexamethasone. Methylprednisolone (Mylan 20 mg, Saint Priest France), another synthetic glucocorticosteroid was used on selected experiments performed on 7-day culture macrophages from 4 donors.
Following treatment, cell culture supernatants were collected and stored at -20˚C for cytokine measurement by ELISA and cell lysates were used for RNA isolation, immunoprecipitation or western blot analysis.

Concentration of supernatants
In selected experiments, cell culture supernatants (4 ml) of human monocytes or HepG2 cells were concentrated by centrifugation (2100 g for at least 90 min at room temperature) using Amicon filters with a molecular cut off of 3 kDa (Merck Milipore) until a final volume of 100 μL.

Reverse transcription and quantitative-PCR (RT-qPCR)
Total RNA from monocytes or macrophages was isolated after treatments using the RNAeasy mini kit (Qiagen) including a DNase step (Qiagen). RNA was then reverse transcribed in the presence of 2.5 mM oligo-dT using the Transcriptor High Fidelity cDNA Synthesis Kit (Roche), following the manufacturer's instructions. Five ng of cDNA was amplified using the Mesa Blue qPCR MasterMix Plus for SYBR Assay (Eurogentec) in the Light Cycler LC480 (Roche). mRNA expression was normalized to the levels of ribosomal protein L13a, RPL13A (NM_012423.2), which was used as housekeeping gene. The glucocorticoid-induced leucine zipper (GILZ) gene expression was used as a marker of glucocorticoid induction [49]. The relative level of expression of a gene between sample 1 (treated) and sample 2 (control) was calculated using the ΔΔCt formula: 2 −(Ct1−Ct RPL13A 1)−(Ct2−Ct RPL13A 2) . Normalized Ct values (mean of controls) were used to calculate the gene expression ratio between two samples. Primers (S1 Table) were designed either using the Probe Finder software (http://qpcr.probefinder.com/ organism.jsp; Roche Life sciences) or manually for specificity reasons.

Western blot analysis
Monocytes or macrophages were lysed in 2.5× Laemmli SDS buffer supplemented with 4% beta-mercaptoethanol, 0.02% bromophenol blue and a protease inhibitor cocktail (Roche). Equal amounts of protein were analyzed by 15% SDS-PAGE. Proteins were transferred to PVDF membrane and incubated overnight with a primary mouse anti-human serum amyloid A1 (capture antibody from DuoSet ELISA (DY3019-05 kit R&D, Biotechne, Abingdon, UK) used at 1 μg/ml in 5% BSA-PBS-0.1% Tween) and a secondary anti mouse IgG HRP (1:1000) (Cat No. 7076, Cell Signaling, Danvers, MA, USA). Following wash steps, the blots were revealed using the SuperSignal West Dura Extended Duration substrate (Thermo Scientific) and visualized using a BIORAD camera (Universal Hood II).

Immunoprecipitation
Immunoprecipitation was done according to the manufacturer's instructions using our own lysis buffer. Briefly monocytes or 7-day culture macrophages were incubated for 1 hour on ice with 1M Tris pH 7.5, 5M NaCl, 0.5M EDTA buffer, 10% Triton, supplemented with protease (Complete protease inhibitor cocktail, Cat No. 11836145001, Roche, Germany) and phosphatase (from the universal magnetic co-IP kit Cat No. 54002, Active Motif) inhibitors. Lysates were collected and 5 μg of anti-SAA antibody (mouse anti-human serum amyloid A1 capture antibody from DuoSet ELISA DY3019-05 kit R&D, Biotechne) was then added to the sample and incubated overnight at 4˚C. 25 μL of magnetic beads were added to each sample the following day, incubated for 1 hour. Five washes were done with the lysis buffer. After the last wash, 20 μL of 2.5x Laemmli-SDS buffer was added to each sample and eluates were used for western blotting.

ELISA
Cell culture supernatants from human monocytes or macrophages were collected after treatment, centrifuged for 10 minutes at 300 g and kept at -20˚C for cytokine measurements. In selected experiments concentrated supernatants were also tested. SAA1 and IL1β were quantified using the DuoSet ELISA kits (Cat No DY3019-05, and DY201-05 respectively, R&D, Biotechne) following manufacturers' instructions.

Statistics
Results are presented as the mean ± standard error of the mean (SEM) from experiments performed in cells from at least 4 independent donors unless otherwise specified. The Mann Whitney test was used to evaluate the statistical significance of differences between 2 groups. A value of p<0.05 was considered significant (indicated as � p < 0.05; �� p < 0.01, ��� p< 0.0001 or # p < 0.05; ## p < 0.01, ### p< 0.0001, depending on the comparisons made). Graphpad Prism was used to design all figures and statistical tests.

Expression of SAA1, SAA2, SAA4 and pro-inflammatory cytokines in human adherent monocytes
Basal SAA1, SAA2 and SAA4 mRNA levels were very low (Ct>37) and not significantly modified following stimulation of adherent monocytes with 100 ng/ml LPS (Fig 1). Treatment with 1μM dexamethasone (Dex) alone significantly increased SAA1 gene expression (Ct from 37 to 32), while no significant change was observed for SAA2 and SAA4. The concentration and incubation periods used for LPS and Dex were defined in separate experiments as described in the methods and presented in the supplement (S1 Appendix, S1 and S2 Figs). Surprisingly, a combined treatment of 100 ng/mL LPS with 1μM Dex induced a major increase in SAA1 expression (Ct from 37 to 27-28) (p<0.0001) and to a lower extent in SAA2 expression (Ct from 37 to 33) (p = 0.0003) (Fig 1, upper panel). In contrast, no significant change was observed in SAA4 expression (Fig 1, upper panel). Under all conditions used SAA3 expression was not detected.
Expression of the pro-inflammatory cytokines IL1A, IL1B and IL6 was used as a positive control of the LPS treatment. As expected, in the presence of LPS alone, mRNA expression of IL1A, IL1B and IL6 was significantly induced (p<0.0001), whereas treatment with Dex alone had no effect on their basal expression levels (Fig 1, lower panel). The combined treatment LPS-Dex significantly downregulated IL1B gene expression (p = 0.03) as compared to treatment with LPS alone (Fig 1, lower panel).

Expression of SAA1, SAA2, SAA4 and pro-inflammatory cytokines in CD14 positively-selected human monocytes
In order to determine whether the technique used for monocyte selection from PBMCs may have an impact on SAA gene expression, we isolated monocytes using CD14 positive selection instead of adherence. Basal expression of SAA genes in CD14+ monocytes was very low (Ct>37), similar to what was observed in adherent monocytes (S3 Fig). Treatment with Dex alone had no significant impact, while the combination of LPS-Dex induced an increase in SAA1 expression (Ct from 37 to 28), comparable to that observed in monocytes selected by adherence ( S3 Fig). The expression of SAA2 was also significantly up-regulated (Ct from 38 to 32) in the presence of LPS-Dex although to a lesser extent. SAA4 expression was variable but under all treatments very low (Ct>35). In the presence of LPS alone, IL1B and IL11A mRNA expression was induced as expected, whereas treatment with Dex alone had no effect on their basal expression levels. The combined treatment of LPS-Dex down-regulated the LPS-induced gene expression of both IL1B and IL1A in the 2 donors tested (S3 Fig, lower panel).

SAA and IL1β secretion by human monocytes
In contrast to mRNA expression, secreted SAA1 levels were always below the ELISA detection threshold (~1.5 ng/mL), in supernatants of cells treated with LPS or LPS-Dex or in supernatants of non-treated cells.
However, LPS-induced IL1β secretion followed gene expression. Dexamethasone alone had no effect in IL1β secretion, whereas treatment with LPS-Dex induced a milder secretion of IL1β as compared to LPS alone (Fig 2).
We were unable to detect SAA1 even after supernatant concentration, although under the same conditions IL1β was detected.
Expression of SAA1, SAA2, SAA4 and pro-inflammatory cytokines in human 7-, 14-and 21-day culture macrophages SAA gene expression was studied in 7-, 14-and 21-day cultures of human monocyte-derived macrophages. Similarly to monocytes, the basal relative mRNA level of SAA1, SAA2, SAA4 was very low (Ct>35) (upper panels of Figs 3-5 respectively). Treatment with LPS alone induced a minor increase in SAA1 and SAA2 expression, significant only for SAA1 in 7-day culture macrophages (p = 0.01). SAA4 expression was not impacted by LPS treatment. Dex   4). A non-significant effect of LPS-Dex on SAA2 expression was observed in 21-day culture macrophages (Fig 5). A minor, not significant, impact on SAA4 expression was observed in 14-and 21-day culture macrophages (Figs 4 and 5).
As expected, IL1B and IL6 mRNA expression was significantly induced in the presence of LPS alone in 14-and 21-day culture macrophages. IL1A expression was significantly increased in 7-day but to a lesser extent in 14-day culture macrophages. Dexamethasone alone had no effect on cytokine gene expression as compared to their basal levels. The combined treatment LPS-Dex significantly down-regulated the LPS-induced IL1B and IL6 expression in 7-day (p = 0.04 and p = 0.004, respectively) and IL6 expression in 14-day culture macrophages (p = 0.04) (Figs 3-5 lower panel), while the expression of IL1A, at all-time points, IL1B in 14and 21-day and IL6 in 21-day culture macrophages was reduced, although did not reach statistical significance.

Effect of methylprednisolone on the expression of SAA1, SAA2, SAA4 and pro-inflammatory cytokines in human 7-day culture macrophages
To confirm the impact of glucocorticoids combined to LPS on the expression of SAA genes, we comparatively tested methylprednisolone and dexamethasone on 7-day culture macrophages from 4 donors (S5 Fig). Methylprednisolone alone increased the SAA1 mRNA expression (Ct from 39 to 32, p = 0.02) but had no impact on the expression of SAA2 and SAA4. The combination with LPS and methylprednisolone induced a major SAA1 expression (Ct from 39 to 25). LPS and methylprednisolone also induced the expression of SAA2 (Ct from 35 to 30) but the variability among donors was important. SAA4 expression remained very low (Ct>31).
Similar to LPS-Dex, treatment with LPS-methylprednisolone significantly decreased the LPS-induced mRNA expression of IL6 and decreased IL1B and IL1A expression even if the difference between the two conditions did not reach statistical significance due to the variability among the 4 donors. The combined treatment with IL1β and IL6 greatly increased the SAA1 expression (Ct from 37 to 24). This effect was more pronounced when HepG2 were treated with Dex concomitantly to IL1β and IL6 (Ct from 37 to 23). Similar effects were observed for SAA2, whereas the constitutively expressed SAA4 was not impacted by the treatment (Ct range from . SAA1 secretion was measurable (S7A Fig) in the supernatants of HepG2 cells treated with IL1β-IL6 (~17 ng/mL). When cells were treated with IL1β-IL6-Dex (S7A Fig), the secretion was higher (~120 ng/mL) as compared to IL1β-IL6 alone, a result that correlated with the mRNA expression.

Expression of SAA1, SAA2, SAA4 in human adherent monocytes in the presence of cytokines
In selected experiments, monocytes were treated with IL1β and IL6 to mimic conditions that induce the SAA expression in HepG2 cells. The combined treatment of IL1β-IL6 had no impact on basal SAA gene expression (Ct~38, either for SAA1, SAA2, or SAA4). The combined treatment of IL1β-IL6-Dex induced SAA1 and SAA2 gene expression (from Ct~38 tõ 28 and 34 respectively). However, no SAA1 secretion was measurable under IL1β-IL6-Dex treatment.

SAA1 protein expression by human monocytes
Immunoprecipitation experiments were performed in cell lysates of human monocytes treated or not with LPS-Dex (S7B and S8 Figs) or INF-γ-LPS-Dex (Fig 6 and S9 Fig). HepG2 cells treated with IL1β-IL6-Dex were used as a positive control. As shown in Fig 6 and S7B Fig, fol-lowing immunoprecipitation, the SAA1 protein was detected in HepG2 cells but not in lysates of monocytes treated with LPS-Dex (S7B Fig). However, SAA1 protein was evidenced by immunoprecipitation in monocytes of two out of six healthy donors (Fig 6) after combining INF-γ (M1 phenotype) with LPS-Dex.

Discussion
Human SAA proteins are important acute phase reactants and precursors of AA amyloidosis and partners of the "cytokine-serum amyloid A-chemokine network" [11]. Despite evidence for extrahepatic SAA production [33], the potential role of human monocytes and macrophages -major players in inflammation-in local SAA expression is still unknown. A difficulty for this evaluation is that human resident macrophages, with the exception of alveolar macrophages, are not easily accessible [50]. In this study, we used human monocytes from peripheral blood, selected by adherence or by CD14-positive selection, and macrophages that were differentiated from monocytes after 7, 14 or 21 days in culture; during this period, cells acquire the morphological and functional characteristics of tissue macrophages, which after 14 days in culture, present an epithelioid cell pattern similar to that of macrophages observed in granulomas [51,52].
In the present study, we clearly show an induction of SAA1 and SAA2 gene transcription in monocytes and monocyte-derived macrophages in response to a combined stimulation with LPS and glucocorticoids. As previously reported by others [5,8], no expression of SAA3 was detected whereas SAA4 was constitutively expressed at very low levels whatever the conditions used. Our results are in line with those previously reported on human macrophage cell lines showing a SAA gene transcription in response to LPS-Dex stimulation [43,45]. However in those studies, detection of SAA mRNA was performed using primers that cannot distinguish between SAA1 and SAA2 due to the high homology between the two genes [45,53]. For this reason, in the present study, we designed primers that recognize specifically SAA1 or SAA2 mRNA. We were therefore able to show that SAA1 is the major SAA acute phase gene induced in human monocytes and derived macrophages in the presence of LPS combined to glucocorticoids. Similar results were obtained either for monocytes selected after 1h adherence or using CD14-positive selection, suggesting that adherence per se does not have a significant effect on SAA gene expression. In addition, when a pro-inflammatory M1 phenotype was induced by IFN-γ, a potentiation of LPS-Dex effect in inducing SAA1 gene expression was observed, suggesting that the inflammatory environment modulates SAA gene expression.
The acute SAA gene induction in the presence of LPS-dexamethasone/methylprednisolone illustrates the pleiotropic roles of glucocorticoids in an inflammatory context [54]. The antiinflammatory actions of glucocorticoids are regulated mainly through glucocorticoid receptor signaling [55]. However, pro-inflammatory effects of glucocorticoids have also been demonstrated, particularly during the initial phase of activation of the immune system in response to stress signals. For example, it has been shown that in the presence of pro-inflammatory cytokines such as TNFα, glucocorticoids enhance the expression of both TLR2 mRNA and protein in A549 epithelial cell line [56,57]. In addition, whole genome microarray analysis in the A549 epithelial cell line highlighted 311 genes coregulated by glucocorticoids and TNFα [58]. Expression of the SAA genes in human monocytes and macrophages Among them, SAA1 and SAA2 were found to be up-regulated by the synergistic treatment [58]. In such cases, the action of glucocorticoids is considered to prepare the innate immune system for acute response [54].
In the present study, LPS alone was not sufficient to induce SAA1 and SAA2 gene expression although it up-regulated the expression of IL1B, IL1A and IL6. LPS is a potent activator of pro-inflammatory gene expression in immune cells, especially in monocytes [59], and although its exact mode of action is not clearly established, transcription factors like nuclear factor-κB (NFκB), nuclear factor-IL-6 (NFIL6) and activator protein-1 (AP-1) have been shown to be involved [60][61][62]. Despite the presence of NFκB, NFIL6 and AP-1 binding sites on both the SAA1 and SAA2 promoters [31,63,64], a combined treatment of LPS with glucocorticoids (dexamethasone or methylprednisolone) was necessary for SAA gene induction, as also shown by Yamada et al [45] in the THP1 cell line. Similar results were obtained when, instead of LPS, a combination of glucocorticoids with pro-inflammatory cytokines were used. This supports the hypothesis that LPS-induced pro-inflammatory cytokines may be the necessary co-activators for glucocorticoid actions in SAA gene expression. Understanding the role of glucocorticoids in the induction of acute SAA gene expression is particularly challenging. It is known that glucocorticoids, after binding to specific cytosolic receptors (GR), act mainly via direct binding to glucocorticoid responsive elements (GRE) present in target gene promoters, or through GR interactions with NFκB and AP-1 [65,66]. A putative GRE has been identified in the SAA1 promoter whereas the corresponding region in the SAA2 gene is disrupted by a nine-base insertion [31]. In our studies, glucocorticoids alone were not sufficient to lead to a major increase in SAA gene expression in human monocytes or macrophages. Therefore, in agreement with Rao et al., [67], a crosstalk between GR and NFκB pathways may be hypothesized. However, in line with the classical anti-inflammatory role of glucocorticoids [68,69], dexamethasone reduces LPS-induced cytokine expression.
While a combined treatment of LPS-Dex induced the mRNA expression of SAA1 and to a lesser extent of SAA2, we had great difficulty to detect the protein in monocyte-or macrophage-lysates even after immunoprecipitation experiments. Finally, after combining IFN-γ (M1 phenotype) with LPS-Dex, we succeeded in identifying SAA in monocyte lysates. Yet, we were unable to detect SAA secretion even in concentrated monocyte-or macrophage supernatants. However, IL1β secretion measured in parallel was present before and after concentration. The possibility that the levels of SAA were below the detection limit of the ELISA kit (below 1.5 ng/ml) must be considered. Yamada et al. were the only authors reporting SAA secretion by human monocytes-macrophages; it is noteworthy to remind that in order to detect SAA concentrations between 0 to 1.5 ng/mL in monocyte supernatants, they have used an in-house ELISA [45]. One hypothesis is that, even after monocyte-macrophage stimulation, the protein is unstable and quickly degraded. A second hypothesis is that the SAA secretion by monocytes or macrophages is very low and undetectable with the methods used. A third hypothesis is that stimulatory factors present in the human serum may be required to reach a measurable level of SAA; our experiments have been performed in serum-free medium to avoid the effect of SAA internalized from human serum [70]. Yamada et al. described a low SAA secretion related to incubation with elution fractions from fetal calf serum containing IgG and or albumin [45]. However, adding fetal calf serum in the culture medium did not modify our results. The dependence on serum may be specific to some cell types, as in similar experiments with HepG2 cells, we were able to detect SAA secretion in serum-free conditions. However, one should bear in mind that a discrepancy between a low transcriptional activity and high SAA mRNA accumulation was previously reported in Hep3B hepatic cell lines [30,71]. Indeed, besides transcriptional and post-transcriptional control of SAA gene transcription, there is evidence for a translational regulation in SAA transcripts while it is not yet well understood [10,30].
The present study demonstrates that stimulated human monocytes and macrophages can express the SAA1 and SAA2 genes, mainly SAA1, a result which until now was only shown in macrophage-like cell lines. Therefore, it may be hypothesized that SAA can be locally produced by activated monocytes-macrophages in an inflammatory milieu, even at a very low level. In that context, SAA may activate through autocrine/paracrine stimulation various cells especially monocytes-macrophages via receptors like TLR2/4 or FPR2 [8,11] inducing a SAA-chemokine-cytokine cascade [11]. A low concentration of SAA may favour the production of chemokines as CCL2, CCL3, CXCL8 recruiting more monocytes-macrophages and playing pleiomorphic roles in the immune reaction [8,11]. SAA-stimulated monocytes from patients with Behcet's disease were capable of promoting Th17 from peripheral CD4+ T cells [28].
Macrophages in inflammatory lesions usually form packed aggregates as in the characteristic granulomas of sarcoidosis [72,73]. SAA has been detected by immunohistochemistry in sarcoidosis granulomas and is suggested to play a role in their maintenance [29,40]. It may be hypothesized that secreted SAA from macrophage-derived cells may reach a significant concentration in the confined intercellular milieu of granulomas and play an autocrine/paracrine role towards the neighbouring cells. This type of stimulation may also lead to a fibrous remodelling of tissues [29,41,74].
In addition, recombinant SAA was shown to have mixed effect on macrophages, inducing both pro-and anti-inflammatory cytokine expression and M2 differentiation markers [75]. It has been suggested that the persistence of this M1/M2 population may contribute to the deleterious remodelling of tissues and persistence of Th17-mediated inflammatory mechanisms [28,75,76].
In conclusion, while previous studies on this topic almost exclusively dealt with cell lines, this study clearly demonstrates that human monocytes and monocyte-derived macrophage express SAA1 and SAA2 genes, mainly SAA1, when in an inflammatory milieu, especially after M1 differentiation. This encourages us to consider SAA not only as an acute-phase inflammatory mediator secreted by the liver with a general action on innate immune system but also as a cytokine-like mediator secreted by extra-hepatic cells as macrophages and acting locally. The requirement of glucocorticoids (dexamethasone or methylprednisolone) for in vitro monocyte-macrophage activation raises questions about the actions of these widely used therapeutic agents in autoinflammatory/autoimmune chronic diseases.
Supporting information S1 Appendix. LPS and dexamethasone time and concentration dependent experiments in human monocytes and 7-day culture macrophages. (TIF) S1 Fig. Dose-dependent expression of SAA1, SAA2 and SAA4 in human monocytes and 7-day culture macrophages. mRNA gene expression was measured by RT-qPCR and presented as relative fold change of non-treated monocytes (upper panel) and 7-days macrophages (lower panel). Data represent the mean ± standard error of the mean (SEM) of 4 experiments performed in cells isolated from 4 independent donors. Asterisks indicate significant differences as compared to non-treated cells (Mann Whitney test: � p < 0.05). # indicates significant differences between the two delimited conditions (Mann Whitney test: # p < 0.05). (TIF) S2 Fig. Kinetics of SAA1, SAA2, SAA4 and pro-inflammatory cytokine expression in human monocytes. mRNA gene expression was measured by RT-qPCR and presented as relative fold change of non-treated monocytes. Data represent the mean ± standard error of the mean (SEM) of 4 experiments performed in cells isolated from 4 independent donors. Asterisks indicate significant differences as compared to non-treated cells (Mann Whitney test: � p < 0.05). (TIF) S3 Fig. SAA1, SAA2, SAA4 and pro-inflammatory cytokine expression in human nonadherent monocytes (CD14+ selected). mRNA gene expression was measured by RT-qPCR and presented as relative fold change of non-treated monocytes. Data represent the mean ± standard error of the mean (SEM) of 2 experiments performed in cells isolated from 2 independent donors. (TIF) S4 Fig. SAA1, SAA2, SAA4 and IL1B expression in human polarized monocytes. mRNA gene expression was measured by RT-qPCR and presented as relative fold change of nontreated monocytes. Data represent the mean ± standard error of the mean (SEM) of 5 experiments performed in cells isolated from 5 independent donors except for M2 conditions (n = 3). Asterisks indicate significant differences as compared to non-treated cells (Mann Whitney test: � p < 0.05, �� p < 0.01. # points out significant differences between the indicated groups (Mann Whitney test: # p < 0.05, ## p <0.01).