Ex Vivo and In Vitro Effect of Serum Amyloid A in the Induction of Macrophage M2 Markers and Efferocytosis of Apoptotic Neutrophils

Macrophages are dynamic and heterogeneous innate immune cells that are essential for the initiation and resolution of inflammation. These cells display different phenotypes and respond to environmental cues in a polarized manner (1, 2). Polarized macrophages have been broadly classified into two groups, M1 macrophages and M2 macrophages (3, 4). Macrophages stimulated with LPS and IFN-γ assume an M1 proinflammatory phenotype, which is characterized by high levels of inducible NO synthase expression, increased capacity of Ag presentation, and elevated production of proinflammatory cytokines, including TNF-α, IL-1β, and IL-12 (5). The M2 macrophages, which are induced by exposure to IL-4, IL-13, TGF-β, and glucocorticoids, produce anti-inflammatory cytokines, including IL-10, TGF-β, IL-1R antagonist (IL-1Rn), CCL17, and, in mice, are found in inflammatory zone-1 (Retnla, Fizz-1) (3), chitinase 3–like 3 (Chi3l3, Ym1), and arginase 1 (Arg1) (6). Additionally, the M2 macrophages express high levels of CD206 (mannose receptor C type 1 [MRC1]) and CD163 (1, 6, 7). The different cytokines, surface markers, and metabolic enzymes expressed by these macrophage subsets reflect the diverse functions of the M1 and M2 macrophages. The M2 macrophages are often seen in the late phase of inflammation and are thought to play a role in the resolution of inflammation, tissue repair, and remodeling (8).

Neutrophils are the first cells recruited to the site of inflammation (9). These terminally differentiated cells have a short lifespan once released from bone marrow. Neutrophil apoptosis, either spontaneous or pathogen-induced, can lead to the release of toxic contents and cause tissue damage. Therefore, proper clearance of apoptotic neutrophils by macrophages, a process termed efferocytosis (10), is essential for homeostasis, tissue repair, and remodeling (11, 12). The M2 macrophages display an enhanced efferocytosis and are known as “efferocytic-high” macrophages (13–15).

Serum amyloid A (SAA) is produced in response to severe infection and other forms of environmental insults (16, 17). A major acute-phase protein, SAA serves as a biomarker for clinical diagnosis of inflammatory diseases (18). Elevated SAA production is associated with rheumatoid arthritis, atherosclerosis, Crohn’s disease, and type 2 diabetes (19–22), bringing the possibility that SAA may play an active role in these inflammatory disorders. In ex vivo and in vitro studies, recombinant human SAA is chemotactic for phagocytes (23) and induces the transcripts of several proinflammatory cytokines, including TNF-α, IL-1β, IL-6, IL-12p40, IL-23p19, and G-CSF (24–29). SAA stimulates the expression of metalloproteinase genes in rheumatoid synovium (30) and contributes to neutrophilia in sterile inflammation (27). However, recent reports indicate that systemic induction or administration of SAA failed to increase the plasma levels of proinflammatory cytokines (31, 32), raising the possibilities that SAA produced at different phases of inflammation and present at different locations (blood versus tissue) may have different functions.

In the present study, we report that ex vivo and in vitro stimulation of macrophages with SAA induces M2 macrophage markers and promotes efferocytosis of apoptotic neutrophils. We have further investigated the underlying signaling mechanisms and identified MyD88 and IFN regulatory factor (IRF)4 as key components for SAA-induced M2-like macrophage polarization and efferocytosis of apoptotic neutrophils.

All knockout mice used in this study were obtained from The Jackson Laboratory (Bar Harbor, ME) except for the Fpr2 knockout mice (33), which were provided by Dr. Ji Ming Wang (National Cancer Institute–Frederick, National Institutes of Health). C57BL/6 mice were purchased from Shanghai Laboratory Animal Center (Shanghai, China). All animal studies were conducted at Shanghai Jiao Tong University using age- and sex-matched littermates. The procedures involving mice were approved by the Institutional Animal Care and Use Committee at Shanghai Jiao Tong University.

Recombinant human SAA (Apo-SAA, catalog no. 300-13) and SAA1 (Apo-SAA1, catalog no. 300-53) were obtained from PeproTech (Rocky Hill, NJ). The content of bacterial endotoxin was <0.1 ng/μg protein according to the manufacturer. Using Tachypleus amebocyte lysate (Zhanjiang A&C Biological, Zhanjiang, China), we found the endotoxin level to be <0.05 ng/μg SAA protein. Inhibitors for the protein kinases MEK (PD98059), p38 MAPK (SB203580), JNK (SP600125), and PI3K (LY294002) were purchased from Calbiochem (San Diego, CA). Ab to Arg1 was obtained from Santa Cruz Biotechnology (Dallas, TX). The Ym1 Ab was obtained from StemCell Technologies (Vancouver, BC, Canada). Abs to IRF4 and IRF5 were purchased from Cell Signaling Technology (Danvers, MA). Anti-CD11b (PerCP-Cy5.5, clone M1/70) and anti-CD206 (FITC, clone C068C2) were obtained from BioLegend (San Diego, CA) and used for flow cytometry analysis.

Monocytes were purified from buffy coats obtained from healthy blood donors following a protocol approved by the Institutional Review Board at Shanghai Jiao Tong University. Briefly, human PBMCs were separated by standard gradient centrifugation on a Ficoll-Paque gradient. Monocytes were then isolated from PBMCs by positive sorting using anti-CD14–conjugated magnetic microbeads (Miltenyi Biotec, Bergisch Gladbach, Germany). Monocyte purity was >90% as assessed by flow cytometry (data not shown). Mouse bone marrow–derived macrophages (BMDMs) were prepared from bone marrow cells isolated from wild-type (WT) C57BL/6 mice or knockout mice as described previously (34). Human monocytic leukemia cell line THP-1 (TIB-202) and mouse macrophage cell line RAW264.7 (TIB-71) was purchased from American Type Culture Collection (Manassas, VA). The cells were maintained in RPMI 1640 supplemented with 10% FBS, 1% streptomycin-penicillin, and 10 μM 2-ME. All cell cultures were kept at 37°C in a humidified atmosphere with 5% CO₂.

Total RNA was extracted from the cells using an RNeasy Mini kit (Qiagen China, Shanghai, China). First-strand cDNA was prepared from 1 μg total RNA using avian myeloblastosis virus reverse transcriptase (Promega, Madison, WI). Gene-specific real-time quantitative PCR was performed using SYBR Green real-time PCR master mix (Toyobo, Osaka, Japan) on a Mastercycler ep realplex system (Eppendorf, Hamburg, Germany). The results were normalized to that of GAPDH expression in the sample. Human IRF4 promoter extending from −1992 to +13 was amplified by PCR with genomic DNA. The PCR product was excised with XhoI and SacI digestion and cloned into the pGL3 luciferase vector (Promega). The ΔΔCt method was used for calculating the transcript level.

Arginase activity based on urea formation was measured as previously described (35). Briefly, MnCl₂ (25 μl, 10 mM in 50 mM Tris-HCl) was added to cell lysates and incubated at 56°C for 10 min for enzyme activation. l-Arginine was added to the tubes, incubated at 37°C for 1 h for enzyme activation, and the reaction was stopped by adding an acid mixture (H₂SO₄/H₃PO₄/H₂O in a ratio of 1:3:7). For colorimetric determination, α-isonitrosopropiophenone was added to each tube. The mixture was heated (100°C, 45 min) and kept in the dark for 10 min. Readings were taken with a spectrophotometer at an absorbance of 540 nm. Sample blank contained lysate without the addition of MnCl₂ or l-arginine to measure basal arginase activity, and these readings were subtracted from all samples. The values were divided by protein content to determine specific activity of the enzyme. The amount of the urea produced was used as an index for arginase activity.

Mouse polymorphonuclear neutrophils (PMNs) were purified from bone marrow cell suspensions as described previously (36). Cells (2 × 10⁷/ml were incubated with 10 μM 5-chloromethylfluorescein diacetate (CMFDA; Molecular Probes) at 37°C for 15 min. Spontaneous apoptosis of CMFDA-labeled PMNs was achieved by culturing PMNs in RPMI 1640 medium supplemented with 10% autologous serum at 37°C for 24 h. To monitor apoptosis, 10⁶ cells were stained with annexin V–FITC (BD Biosciences, San Jose, CA) and propidium iodide, according to the manufacturer’s protocol. Cells were analyzed without fixation by flow cytometry within 30 min of staining.

Approximately 5 × 10⁵ mouse BMDMs were seeded in a 24-well plate, exposed to experimental stimuli, and incubated with CMFDA-labeled apoptotic cells at a ratio of 1:1 at 37°C for 60 min. The cells were collected and extracellular fluorescence associated with membrane-bound but nonengulfed apoptotic cells and bodies was quenched with trypan blue (0.04% in PBS). The proportion of BMDMs that exhibited increased fluorescence (corresponding to phagocytosis of fluorescent-labeled apoptotic cells) was determined by flow cytometry on a BD LSRFortessa flow cytometer (BD Biosciences) or by fluorescent microscopy on a Leica TCS SP8 inverted confocal microscope (Leica, Wetzlar, Germany). The phagocytosis index is calculated as the percentage of macrophages containing at least one ingested neutrophil. For efferocytosis in peritoneal cavity, 1 × 10⁶ CMFDA-labeled apoptotic neutrophils were resuspended in 100 μl PBS and injected into the peritoneal cavity of mice 18 h after SAA injection. After another 2 h, the mice were sacrificed and peritoneal cells were harvested by lavage. The percentage of macrophages (CD11b⁺F4/80⁺) that ingested apoptotic neutrophils was taken as the phagocytosis index.

A second method was used to discriminate the attached neutrophils from internalized neutrophils. Apoptotic neutrophils were biotinylated with EZ-Link Sulfo-NHS-Biotin (Thermo Fisher Scientific) and then treated with allophycocyanin-streptavidin (BioLegend). Cell surface–bound apoptotic neutrophils were seen under both CMFDA (green) and allophycocyanin-streptavidin (red) optics, whereas internalized apoptotic neutrophils were visualized only under CMFDA (green) optics.

For nonspecific phagocytic uptake, phagocytosis of latex beads/rabbit IgG/FITC complex (Cayman Chemical, Ann Arbor, MI) was performed according to the manufacturer’s instruction. Phagocytosis of dextran was performed by adding FITC-dextran (molecular mass, 70,000 Da; Sigma-Aldrich) to the cell suspension at different final concentrations. The cells were incubated for 45 min at 37°C, washed with ice-cold PBS, and analyzed by FACS.

After treatment, cells were fixed, permeabilized, and stained with Alexa Fluor 563–conjugated phalloidin (Molecular Probes). The nuclei were visualized with DAPI staining. The cells were affixed on a glass coverslip by cytospin (Shandon). The coverslip was mounted on a glass slide using PermaFluor mounting medium (Thermo Fisher Scientific). Microscopic analysis was performed on a Leica TCS SP8 inverted fluorescent confocal microscope (Leica).

GST–p21 binding domain (PBD) was immobilized by adsorption to glutathione-Sepharose beads (GE Healthcare). After SAA stimulation, BMDMs (2 × 10⁶) were lysed in RIPA buffer (25 mM Tris-HCl [pH 7.6], 150 mM NaCl, 1% Nonidet P-40, 1% sodium deoxycholate, 0.1% SDS). The lysate was mixed with the immobilized GST-PBD and gently rotated at 4°C for 60 min. The beads were collected by centrifugation and washed before SDS-PAGE analysis (12% polyacrylamide). The GTP-bound Rac1 that was pulled down was detected by Western blotting using an anti-Rac1 Ab (Cell Signaling Technologies).

The mouse macrophage cell line RAW264.7 was transfected with small interfering RNA (siRNA) using a Silencer siRNA transfection II kit (Ambion, Austin, TX) according to the manufacturer’s instructions. The HiPerFect transfection reagent (Qiagen) was used for siRNA transfection of human monocyte-derived macrophages. The cells were then cultured for 48 h before stimulation. The siRNA oligonucleotides were designed and synthesized by Shanghai RiboBio (Guangzhou, China).

Chromatin immunoprecipitation (ChIP) was performed using a ChIP assay kit (Millipore, Billerica, MA) according to the manufacturer’s instructions with minor modifications. Briefly, 5 × 10⁶ PBMC or THP-1 cells were spread in a six-well cell culture cluster (Corning) and stimulated with or without SAA at specified concentrations. After stimulation, cells were cross-linked and then washed and resuspended in SDS lysis buffer. Nuclei were fragmented by sonication. Chromatin fractions were precleared with protein A–agarose beads followed by immunoprecipitation overnight with an anti-IRF4 Ab or a control IgG. Cross-linking was reversed and proteinase K digestion was performed. The input and immunoprecipitated DNA were amplified by real-time PCR. The ratio of the amount of DNA recovered relative to the input control was calculated.

For each experimental group, data are presented as means ± SEM (n ≥ 3). Statistical significance among different samples was assessed with a paired Student t test or with a one-way ANOVA with repeated measures when there were two or more treatment groups. A p value <0.05 was considered significant. Analysis and graphing were performed using the Prism software (version 5.0; GraphPad Software, San Diego, CA).

Previous studies have shown that SAA has cytokine-like activity and can stimulate phagocyte production of proinflammatory cytokines, including TNF-α, IL-1β, IL-6, IL-12p40, IL-23p19, and G-CSF (24–29). It is unclear whether SAA can induce other genes relevant to macrophage reprogramming. To test this possibility, human CD14⁺ monocytes from peripheral blood (PBMCs) were stimulated with recombinant human SAA (Apo-SAA, 0.5 μM). Real-time PCR was performed to quantify the transcripts of IL-10, IL-1Rn, MRC1, and CCL17. Time-dependent induction of these M2 macrophage markers was observed (Fig. 1A). In mice, the transcripts of these M2 markers were induced in BMDMs (Fig. 1B) and peritoneal macrophages (see Fig. 4) upon SAA stimulation. At the protein level, there was a 30-fold increase in IL-10 production that peaked at 8 h after SAA stimulation (Fig. 2A). Increased cell surface expression of Mrc1 (Fig. 2B) and upregulated Ym1 production (Fig. 2C) were also observed in SAA-stimulated BMDMs. Additionally, there was a 3-fold increase in Arg1 expression (Fig. 2D), along with increased arginase activity (13.74 ± 1.52-fold over control; Fig. 2E). SAA was compared with IL-4, a prototype M2 macrophage inducer, in the induction of arginase activity (Fig. 2F). In contrast, LPS, a classic activator of macrophages, only had a minor effect on arginase activity (Fig. 2F).

Resolution of inflammation requires timely removal of bacteria and apoptotic cells, with the latter being accomplished through a process termed efferocytosis (37). In vitro phagocytosis assays were performed to determine whether SAA could affect efferocytosis of apoptotic neutrophils. BMDMs were cultured in complete medium for 5 d, and LPS (for M1), IL-4 (for M2), fresh medium (for M0), or SAA was added to the cell culture to induce different macrophage phenotypes for another 2 d. CMFDA-labeled early apoptotic neutrophils were added at a 1:1 (macrophages/neutrophils) ratio for a 1 h incubation. After quenching nonphagocytosed cells, the macrophages were analyzed by flow cytometry. As shown in Fig. 3A, the ability of BMDMs to clear apoptotic neutrophils was markedly increased in samples containing SAA, with a phagocytic index comparable to or better than that of the IL-4–treated cells. A concentration-dependent increase in efferocytosis was seen between 0.05 and 1.0 μM SAA (Fig. 3B).

To further determine whether the increase in phagocytic index is a consequence of improved uptake efficiency or specific for the apoptotic neutrophils, we performed phagocytosis experiments using FITC-labeled dextran and fluorescently labeled latex beads. SAA treatment using the same experimental condition as in Fig. 3 did not alter phagocytosis of dextran or latex beads when compared with medium-treated cells (data not shown). Thus, the effect of SAA on efferocytosis appears to be specific for the uptake of apoptotic neutrophils.

Localized polymerization of actin filaments is associated with the formation of pseudopodia that surround the bacterial particles or apoptotic cells to be engulfed (38, 39). In SAA-stimulated macrophages, F-actin rearrangement was found at sites of apoptotic cell ingestion (Fig. 3C). Likewise, activation of the small GTPase Rac1 was observed in SAA-stimulated macrophages (Fig. 3D). These results suggest a correlation between SAA-stimulated Rac activation and increased macrophage efferocytosis (38).

The physiological functions of SAA remain an unsettled issue (31, 32) because early studies used commercially available recombinant human SAA (Apo-SAA) that contained substitution at two positions in SAA1. To address this concern, we tested the ability of recombinant human SAA1 (Apo-SAA1) to induce M2 markers in mouse peritoneal cavity. SAA1 or PBS was administered into the peritoneal cavity for 18 h; peritoneal macrophages were collected and the induction of selected M2 markers was measured. SAA1 potently induced the M2 markers at the mRNA (Fig. 4A) and protein levels (Fig. 4B). Additionally, peritoneal macrophages isolated from SAA1-treated mice displayed higher arginase activities than did those from PBS-treated mice (Fig. 4C). To investigate the effect of SAA1 on macrophage efferocytosis, CMFDA-labeled early apoptotic neutrophils from mice were injected into the peritoneal cavity of C57BL/6 mice 18 h after SAA1 administration. Peritoneal macrophages were isolated 2 h later, and the uptake of apoptotic neutrophils by CD11b⁺F4/80⁺ macrophages was quantified by flow cytometry (Fig. 4D). Macrophage efferocytosis of apoptotic neutrophils was significantly higher in SAA1-treated samples than in PBS-treated controls (Fig. 4E). Collectively, these results show that SAA1, similar to the recombinant Apo-SAA, is able to induce an M2-like macrophage phenotype.

Because classical activation of M2 macrophage is mediated through a Stat6-dependent pathway (1), we investigated the involvement of Stat6 in SAA-induced expression of the M2 markers. As shown in Supplemental Fig. 1A, induction of the M2 macrophage markers by SAA was largely intact in Stat6-deficient BMDMs. Additionally, promotion of BMDM efferocytosis by SAA was not affected by Stat6 deficiency (Supplemental Fig. 1B). These results suggest that the SAA-induced phenotypic change differs from classical activation of M2 macrophages, which requires IL-4–dependent Stat6 activation. Among other transcription factors that might be involved, IRF4 has been recently identified as an important transcription factor for M2 macrophage polarization (40). Indeed, the level of IRF4 transcripts was markedly increased following SAA stimulation in both THP-1 and human CD14⁺ monocytes, and the kinetics were consistent with the expression of the M2 markers examined (Fig. 5A, Supplemental Fig. 2A). An increase in IRF4 protein was also observed based on Western blot analysis of whole-cell lysates prepared from SAA-stimulated samples (Fig. 5B). In contrast, SAA had little effect on IRF5 expression. A human IRF4 promoter-driven luciferase reporter (pGL3-IRF4) was constructed and transiently transfected into the human monocytic cell line THP-1. Stimulation of the transfected cells with SAA (0.5 μM) led to an increase in luciferase activity, confirming that SAA could activate IRF4 gene transcription (Fig. 5C). Given the robust response of IRF4 expression to SAA stimulation, IRF4 recruitment to the promoters of selected M2 marker genes was examined by ChIP. In SAA-treated THP-1 (Fig. 5D) and PBMCs (Supplemental Fig. 2B), anti-IRF4 but not control IgG was found to precipitate products that were subsequently amplified by PCR with primers specific for the promoters of IL-10 and Arg1. These results support an increase in the DNA binding activity of IRF4 following SAA stimulation.

To determine whether IRF4 is required for SAA induction of M2 markers, siRNA was used to silence IRF4 expression (Fig. 6A). Transfection of the specific siRNA led to a significant reduction in SAA-induced expression of IL-10, IL-1Rn, and CCL17, as well as in arginase activity in RAW264.7 macrophages (Fig. 6B) and in human CD14⁺ monocytes (Supplemental Fig. 2D). Potentiation of macrophage efferocytosis by SAA was reduced by >80% in the siRNA1-transfected cells (Fig. 6C, 6D).

The above results suggest an important role for IRF4 in SAA-induced M2 marker gene expression and macrophage efferocytosis. Experiments were conducted to explore the regulatory mechanism for IRF4 expression using pharmacological inhibitors. As shown in Fig. 7A, the PI3K inhibitor LY294002 and the MEK inhibitor PD98059 significantly reduced the accumulation of IRF4 transcripts, possibly at the transcriptional level based on IRF4-driven luciferase reporter activity (Fig. 7B). The observed inhibition was not due to cytotoxicity, as most (>85%) of the cells remained viable under the experimental condition (data not shown). Furthermore, LY294002 and PD98059 also significantly decreased SAA-induced expression of IL-10, IL-1Rn, and CCL17 (Fig. 7C) and Mrc1 (CD206) expression in SAA-treated BMDMs (Fig. 7D). The SAA-induced arginase activity and efferocytosis were inhibited by PD98059 and LY294002 (Fig. 7E, 7F). No significant inhibition was seen with SP600125 (a JNK inhibitor) and SB203580 (a p38 MAPK inhibitor). These data suggested that the ERK and PI3K pathways are critical for SAA induction of M2 macrophage markers and efferocytosis.

SAA is known to activate several receptors, including formyl peptide receptor (FPR)2, TLR2, and TLR4 (28, 41, 42). The role for these receptors in SAA-induced expression of IRF4 and M2 macrophage markers was examined. As shown in Fig. 8A, deficiency in TLR2 and MyD88, but not fpr2, significantly decreased SAA-induced expression of IRF4 in BMDMs (Fig. 8A). Likewise, BMDMs from Tlr2^−/− and Myd88^−/− mice responded to SAA with significantly reduced expression of IL-10, IL-1Rn, and CCL17 (Fig. 8B). Furthermore, deletion of Myd88 and Tlr2 also significantly impacted the effect of SAA on efferocytosis (Fig. 8C, 8D). Taken together, these results suggest that the effect of SAA on the induction of the M2 marker genes and on efferocytosis may involve the MyD88/IRF4 pathway. Interestingly, TLR4 deficiency did not significantly alter the expression of the M2 markers, nor did it affect SAA potentiation of macrophage efferocytosis (Supplemental Fig. 3), suggesting that TLR4 plays a less important role in these settings.

Elevated levels of SAA have been associated with a variety of inflammatory diseases resulting from infectious and noninfectious causes (reviewed in Refs. 17, 43). However, the exact role of SAA in inflammation has not been fully understood. Prior to this study, experimental evidence favored the concept that SAA is proinflammatory, as this acute-phase protein is chemotactic and capable of inducing proinflammatory cytokines such as TNF-α, IL-1β, and IL-8 in in vitro settings. A previous attempt for in vivo expression of SAA1 in mice, through adenoviral infection that increases circulating levels of SAA (32), did not result in systemic or local inflammatory response. These findings have led us to question whether SAA is proinflammatory in vivo. In the present study, we report that SAA plays a role in macrophage reprogramming. Our results show that SAA induces M2 macrophage markers and promotes macrophage efferocytosis of apoptotic neutrophils. These actions are consistently observed ex vivo with isolated monocytes and macrophages, and in vitro with primary human CD14⁺ monocytes and cultured cell lines such as THP-1 and RAW264.7. Additionally, peritoneal administration of SAA produced similar effects. Taken together, these findings suggest a potential function of SAA to promote resolution of inflammation, tissue repair, and remodeling through induction of M2-like macrophages.

Macrophage efferocytosis of apoptotic neutrophils is critical to the resolution of inflammation (37). Our data show that SAA increases the rate of efferocytosis, comparable to that of cells stimulated with IL-4. SAA was previously reported to rescue isolated neutrophils from constitutive apoptosis, thus extending their lifespan (44). These findings suggest a possibly homeostatic function of SAA through its actions on different types of cells. The chemotactic and antiapoptotic functions of SAA may accelerate inflammation at its early stage by attracting additional neutrophils to the site of inflammation, whereas the ability of SAA to induce M2 macrophage markers and promote macrophage efferocytosis is expected to facilitate resolution of inflammation at its late stage. Our observation that peritoneal administration of SAA produces an M2-like macrophage phenotype indicates that this function may be preserved in vivo, where different cells and signaling crosstalk are present. It is likely that SAA functions differently in different experimental settings, hence requiring additional work in vivo to determine its functions during inflammation.

The functional plasticity of macrophages is the result of crosstalk between different receptors and signal cascades. The network is likely complex, including the interactions between receptors, adaptor proteins, and the expression and activation of transcription factors. Although the transcription factor Stat6 is a major driving force for the expression of M2 markers following IL-4 stimulation, it remains inactive in SAA-treated macrophages (data not shown). Additionally, our results demonstrate, to our knowledge for the first time, that SAA induces the expression and activation of IRF4 in monocytes and macrophages. IRF4 was recently identified as a key transcription factor that controls the expression of a variety of M2 markers (40). Our results show that SAA induces IRF4 transcription and promotes IRF4 binding to the promoters of IL-10 and Arg1. Likewise, loss of IRF4 abrogates SAA induction of the M2 markers and macrophage efferocytosis of apoptotic neutrophils. It has been reported that IRF4 expression requires Jmjd3, a histone demethylase that releases transcriptional repression of the Irf4 promoter (40). In a separate study completed recently, we found that SAA could induce the expression of Jmjd3, resulting in diminished histone H3 methylation at lysine 27 (45). As shown in Supplemental Fig. 4, Jmjd3 deficiency by shRNA silencing significantly reduced the expression of IRF4 and the M2 markers in SAA-stimulated RAW26.4 cells. Silencing Jmjd3 expression also reduced SAA potentiation of macrophage efferocytosis. Taken together, these results show that SAA affects macrophage phenotype at least in part through the Jmjd3/IRF4 pathway (46). Epigenetic regulation of SAA-induced gene expression was not previously known, and it could potentially produce a broad effect in the target cells.

The cytokine-like activity of SAA was first associated with its ability to stimulate phagocyte production of proinflammatory cytokines, including TNF-α, IL-1β, IL-6, IL-12p40, IL-23p19, and G-CSF (24–29). This cytokine-like activity now has been redefined by induction of several M2 macrophage markers. Given that SAA is capable of activating both M1 and M2 macrophage markers, it is likely that the “dual actions” of SAA are regulated through multiple signaling pathways. It is known that several receptors, including FPR2, TLR2, and TLR4, are activated by SAA (28, 41, 42). Our data from different gene deletion or functionally deficient mice show that the TLR2 and MyD88 signal cascades are essential for the activation of IRF4 as well as the expression of M2 markers and efferocytosis. In most published studies, activation of TLRs can skew macrophages toward an M1 phenotype. It is therefore interesting to observe that SAA-induced expression of M2 macrophage markers is significantly reduced in Tlr2^−/− macrophages. This TLR appears to play a pivotal role in SAA-induced expression of multiple cytokines, including IL-10 and IL-1Rn (28). Induction of IL-10 has also been observed in SAA-stimulated neutrophils in an invariant NKT cell–regulated manner (47). Thus, SAA may exert biological actions in a cell-specific manner. Additionally, SAA activation of IRF4, which interacts with MyD88, may initiate a negative regulatory mechanism that suppresses TLR-dependent expression of proinflammatory cytokines (48). In a preliminary study, we found that the expression of the M1 markers, such as IL-1β, TNF-α, and IL-6, was not decreased in IRF4 knockdown cells compared with control cells (data not shown). Interestingly, a more profound effect was observed with Myd88 deletion than Tlr2 deletion, suggesting that in addition to mediating signaling initiated by TLR2, MyD88 has another unidentified function relevant to SAA-induced expression of certain cytokines. Collectively, these findings suggest homeostatic regulation of the inflammatory process by SAA through crosstalk with different signaling pathways.

SAA is a group of proteins of ∼104 aa encoded by different genes (49). In humans, SAA1 and SAA2 (94% identical) are inducible, SAA3 is a pseudogene, and SAA4 is constitutively expressed. Although it is not yet possible to assign the M2-like skewing capability to a specific isoform of SAA, our results show that this function is shared by human SAA1 and Apo-SAA from PeproTech, with the latter being a hybrid of SAA1 and SAA2 (SAA1 with Asp⁶¹ and His⁷² substituted by Asn and Arg from SAA2, respectively). It was reported recently that the SAA hybrid is more potent than SAA1 in the induction of proinflammatory cytokine genes (31). However, our results from ex vivo and in vitro studies (using SAA1 and Apo-SAA) are similar, suggesting that both the SAA hybrid (Apo-SAA) and SAA1 are able to skew macrophages toward an M2-like phenotype. Endogenous SAA purified from blood of cardiac surgery patients is as active as the SAA hybrid in the induction of G-CSF production by the J774 macrophages (32). However, systemic elevation of blood SAA levels through infection with human SAA1-adenoviral vectors failed to increase G-CSF levels in the blood (32). In contrast, s.c. injection of SAA resulted in neutrophilia accompanied by an increase in G-CSF concentrations in the blood (27). These findings indicate that the microenvironment, in which SAA is present, can impact the functions of SAA. In blood circulation SAA is closely associated with high-density lipoproteins, and the high-density lipoprotein–bound SAA has diminished cytokine-inducing property (24). In contrast, i.p. administration of SAA1 skewed macrophages toward an M2-like phenotype that was comparable to the ex vivo phenotype induced by the SAA hybrid. Activated macrophages are a major source of nonhepatic SAA, thus creating a potential autocrine loop for efficient utilization of the newly secreted SAA at inflammatory tissues.

It is postulated that the acute-phase response serves the purpose of isolating pathogens, minimizing injury to the host, and promoting tissue repair (16). As a major acute-phase protein, SAA binds to several Gram-negative bacteria and facilitates their elimination by phagocytes (50). Our results, together with the report that invariant NKT cells regulate the production of IL-10 in SAA-stimulated neutrophils (47), suggest a potential role for SAA to promote clearance of apoptotic neutrophils, limit inflammation, and facilitate its resolution.

We thank Dr. Ji Ming Wang for the Fpr2 knockout mice, Dr. Mingjie Chen for technical assistance, and Dr. Runsheng Li for helpful discussions.

The authors have no financial conflicts of interest.