Funding agencies: This study was supported in part by research grants from the Ministry of Health and Welfare of Japan.
Disclosure: The authors declared no conflict of interest.
Author contributions: AF conceived and carried out the experiments. HN, YS, EU, and ST contributed to data interpretation. TO, JK, AF, and TH contributed to data analyses and JN and YO were involved in manuscript writing. YH contributed to the study design, data interpretation, and manuscript writing. All authors gave final approval to the submitted and published versions.
The S100 calcium binding protein B (S100B) implicated in brain inflammation acts via the receptor of advanced glycation end products (RAGE) and is also secreted from adipocytes. We investigated the role of S100B in the interaction between adipocytes and macrophages using a cell-culture model.
Design and Methods
RAW264.7 macrophages (RAW) were stimulated by recombinant S100B to observe alterations in TNF-α and M1 markers; 3T3-L1 adipocytes (L1) were stimulated by TNF-α to examine S100B secretion. RAW and L1 were then mutually stimulated with conditioned media of each other, or co-cultured. The effects of S100B silencing or a RAGE-neutralizing antibody were also investigated.
S100B upregulated TNF-α and M1 markers in RAW, and TNF-α augmented S100B secretion from L1. L1 conditioned media stimulated TNF-α secretion from RAW, and RAW conditioned media increased S100B secretion from L1. The co-culture of RAW and L1 increased TNF-α, S100B, and the expression of M1 markers and the MCP-1 receptor CCR2. The silencing of S100B or RAGE neutralization significantly ameliorated TNF-α hypersecretion from RAW that were stimulated with L1 conditioned media.
Thus, S100B as an adipokine may play a role in the interaction between adipocytes and macrophages to establish a vicious paracrine loop.
Chronic, low-grade inflammation in visceral fat has been implicated in the pathogenesis of obesity, insulin resistance, metabolic syndrome, and type-2 diabetes [1-3]. The adipose tissue inflammation is characterized by augmented macrophage infiltration . Adipose tissue macrophages are known to be the main sources of TNF-α . TNF-α likely plays a central role in insulin resistance, since it has been reported to interfere with insulin action [1, 3]. Additionally, mice lacking the TNF-α or the p55 TNF-α receptor were shown to be partially protected from insulin resistance in obesity [4, 5].
TNF-α secretion from macrophages is known to be stimulated by the interaction of the macrophages with adipocytes. Previous investigations have revealed that saturated free fatty acids (FFAs) released from adipocytes enhance TNF-α secretion from macrophages and that TNF-α reciprocally stimulates lipolysis and FFA release from adipocytes, establishing a vicious cycle . The chemokine, monocyte chemotactic protein-1 (MCP-1), may also be involved in the paracrine loop since it is upregulated by TNF-α, and in turn recruits macrophages [7, 8]. In addition, there are numerous cytokines secreted from adipocytes . However, the involvement of cytokines other than TNF-α and MCP-1 in macrophage activation has not been fully clarified.
Among the cytokines originating from adipocytes, we focused on the S100 calcium binding protein B (S100B). S100B—a member of the S100 protein family—is a 21 kDa calcium-binding dimer protein that is increasingly being investigated for its role in the central nervous system (CNS). Intracellular S100B functions to maintain calcium levels . S100B is also known to be secreted from astrocytes, neurons, and glia [11, 12]. Extracellular S100B acts as a cytokine and exhibits concentration-dependent effects on nerve tissues, with high picomolar to lower nanomolar levels stimulating neurite outgrowth and enhancing survival of neurons during development, and higher concentrations stimulating the expression of proinflammatory cytokines . The proinflammatory effect of S100B has been attributed to its binding affinity to the receptor for advanced glycation endproducts (RAGE) in microglia, astrocytes, and neurons in the CNS [11-14]. The S100B level in the spinal fluid or plasma has been reported to be a marker of brain damage .
S100B is also expressed or secreted by a variety of cells including lymphocytes  and adipocytes . Among them, adipocytes are known to express and secrete S100B as abundantly as the brain [17, 18]. It is interesting to note that a higher concentration of S100B has been observed in the serum of obese patients . Additionally, a significant correlation was reported between body mass index and S100B concentrations .
Based on these observations, we hypothesized that S100B from adipocytes may play a role in chronic inflammation in adipose tissues, as it does in nerve tissues. The aim of this study was to clarify whether adipocyte-derived S100B contributes to the inflammatory interaction of adipocytes with macrophages. We suggest that S100B may be a novel mediator in the vicious cycle between adipocytes and macrophages.
The murine RAW264.7 macrophages (RAW) were provided by the RIKEN BioResource Center through the National BioResource Project of the NEXT, Japan, and the 3T3-L1 pre-adipocytes were purchased from American Type Culture Collection (Manassas, VA). Cells were maintained in Dulbecco's modified eagle medium (DMEM; Sigma-Aldrich, St. Louis, MO) containing 10% fetal bovine serum (FBS; Thermo Scientific, Waltham, MA), penicillin, and streptomycin (Sigma-Aldrich) at 37°C in a humidified 5% CO2/95% air atmosphere. Two days after reaching full confluency, the pre-adipocytes were differentiated into mature adipocytes by culture with 10% FBS-supplemented DMEM containing 0.5 mM 3-isobutyl-1-methly-xanthine (IBMX, Sigma-Aldrich), 0.25 μM dexamethasone (Sigma-Aldrich), and 5 μg/mL insulin (Sigma-Aldrich) for 2 days, and subsequently with medium containing insulin for 2 days. The 3T3-L1 adipocytes (L1) were used at approximately day 7 of differentiation.
To examine the effects of S100B protein on macrophages, RAW were treated with recombinant mouse S100B (ATGen, Seongnam, South Korea) for the indicated time, and assessed for cytokines, proinflammatory M1 markers, and chemokine (C–C motif) receptor 2 (CCR2). L1 were also stimulated with TNF-α, epinephrine, and insulin to clarify the regulation of S100B in adipocytes.
L1 and RAW were then mutually stimulated with the conditioned medium of the other, and also co-cultured on the bottom and inner wells of Cell Culture Insert® (BD Falcon, Franklin Lakes, NJ) that permitted the passage of humoral factors between the chamber wells, in order to further investigate the interaction between L1 and RAW.
Measurement of protein levels of TNF-α and MCP-1
Protein levels of TNF-α and MCP-1 were determined using ELISA Kits (R&D Systems, Minneapolis, MN) as per the manufacturer's instructions.
Quantification of S100B
S100B levels were measured using sandwich ELISA. Briefly, a 96-well microplate was coated with 10 ng/well of mouse monoclonal anti-mouse S100B antibody (Abcam, Cambridge, UK) in a 50 mM carbonate buffer, pH 9.7 at 4°C overnight. The plate was then blocked with 3% (W/V) BSA in a 5 mM carbonate buffer for 2 h. Fifty microliters of the samples or a standard solution with 50 μL of PBS containing 0.5% BSA were poured into the wells and incubated for 3 h at 37°C. After washing the wells, rabbit polyclonal antibodies against anti-mouse S100B (Dako, Carpinteria, CA) were added into the wells and incubated for 1.5 h at room temperature. The wells were washed and incubated with anti-rabbit IgG and avidin-biotinylated horseradish peroxidase (HRP) complex (Vectastain ABC kits®, Vector, Burlingame, CA) for 30 min. Color development with tetramethylbenzidine (TMB) was measured at 490 nm.
Measurement of palmitate
The concentration of palmitate in cultured media was measured using an acyl-coenzyme A oxidase-based colorimetric assay kit (BioVision, Mountain View, CA) as per the manufacturer's instructions.
Quantitative real-time PCR
Total RNA was extracted from L1 and RAW using the High Pure RNA® Kit (Roche, Indianapolis, IN). cDNA was synthesized from the total RNA using the ReverTraAce® qPCR RT Kit (Toyobo, Osaka, Japan). The mRNA expression of S100B, TNF-α, MCP-1, leptin, adiponectin, lipoprotein lipase, perilipin, F4/80, CD11b, CD11c, IL-1β, IL-6, CD86, CD163, CD206, IL-10, Mannnose receptor, Ym-1, and CCR2 was quantified by quantitative real-time PCR using specific primers and the qPCR MasterMix plus for MESA Green® Low ROX (Eurogentec, Seraing, Belgium) on a sequence detector (MX 3000P, Stratagene, La Jolla, CA). Levels of mRNA were normalized to those of GAPDH. The sequences of the primers used are described in the Supporting Information Table S1.
S100B in L1 was detected by immunocytochemistry using mouse monoclonal anti-S100B antibody (1:100, Abcam) and Alexa Fluor® 594-coupled anti-mouse IgG antibody (1:500; Invitrogen, Carlsbad, CA), as per the manufacturer's protocol. Nuclei were stained by DAPI (1:10000).
Western blot analysis of S100B protein
In brief, L1 cells were lysed with RIPA buffer containing cocktails of protease inhibitors and phosphatase inhibitors (Roche). Fifteen μg of proteins were loaded and fractionated on a SDS-PAGE gel, and transferred onto the PVDF membrane using iBlot® system (Invitrogen). The membrane was blocked with 5% milk powder in TBS-Tween 20, washed and then incubated with primary antibodies against murine S100B (1:200; Abcam) or β-actin (1:200; AnaSpec, San Jose, CA). After washing, the membrane was incubated with HRP-conjugated secondary antibodies, and washed again. The protein bands were detected with a ECL Plus® detection kit (GE Healthcare, Tokyo, Japan) on ChemiDoc® XRS (BioRad).
Silencing of S100B in 3T3-L1 adipocytes
L1 adipocytes were transfected with S100B specific siRNA or a scramble siRNA (Applied Biosystems, Carlsbad, CA) using N-TER transfection reagent (Sigma-Aldrich). In brief, the adipocytes were detached from the wells by trypsinization and suspended in serum-free DMEM containing 40 nM siRNA and the transfection reagent. After 2 h, 10% FBS was added to the wells and the cells were incubated for an additional 2 h. The medium was then replaced with fresh DMEM containing 10% FBS. Successful transfection was determined by evaluating the mRNA expression and protein levels of S100B 72 h after the transfection.
Blockade of RAGE
RAW were pre-treated for 2 h with polyclonal anti-RAGE neutralizing antibody (10 μg/ml, Millipore #MAB5328) and then stimulated with glyceraldehyde-originated AGEs (GA-AGEs), S100B protein, and L1 conditioned media. The cultured media and cells were harvested 3 h after stimulation.
All statistical analyses were performed with the Prism 4.0 program (GraphPad software, San Diego, CA). For comparison among groups, the Student's t-test and one-way analysis of variance (ANOVA) were applied. P values < 0.05 were considered significant.
S100B is expressed in cultured adipocytes
Immunochemical staining with monoclonal antibody against mouse S100B showed that the S100B protein is abundantly expressed in the cytosol of L1 adipocytes (Figure 1A). Significant expression of S100B mRNA in L1 was also demonstrated at the similar level to that of MCP-1 (Figure 1B). The similarity in the expression levels of S100B and MCP-1 was also observed in primary adipocytes from epididymal fat pads of healthy C57/BL6 mice (Supporting Information Figure S1D). The S100B mRNA levels diminished soon after induction of differentiation, and the S100B protein level in the cultured media was mildly decreased during differentiation, yet a significant amount of S100B was still secreted from mature L1 adipocytes (Figure 1C).
Effects of Recombinant S100B protein on macrophages
Recombinant S100B stimulated TNF-α secretion from RAW, as determined by its mRNA expression, in a dose-dependent manner (Figure 2A). The S100B protein also significantly increased the gene expression of the M1 markers CD11b, CD11c, IL-1β, and IL-6 as well as of CCR2, a known receptor of MCP-1 (Figure 2B). Conversely, the gene expression of macrophage M2 markers CD206, CD163, IL-10, mannose receptor, and Ym-1 was not affected by the treatment with S100B. The stimulatory effects of S100B on TNF-α secretion and M1 marker expression were also observed in monocytes collected from the blood of healthy C57/BL6 mice (Supporting Information Figure S1A, B).
Regulation of S100B secretion from 3T3-L1 by TNF-α, epinephrine, and insulin
Recombinant mouse TNF-α increased S100B concentration in cultured media in a dose-dependent manner and also augmented the protein levels in L1 cells (Figure 3A). Western blot analysis of L1 cell lysate showed a comparable result (Figure 3B), indicating that TNF-α stimulates not only S100B release but also its synthesis. The TNF-α-induced S100B increase was also observed in primary adipocytes from C57/BL6 mice (Supporting Information Figure S1C). In contrast, the S100B mRNA level in L1 was decreased upon stimulation with TNF-α (Figure 3C). Epinephrine also enhanced S100B secretion from L1, while insulin decreased its secretion in a dose-dependent manner (Figure 3D). In addition, insulin partially ameliorated the hypersecretion of S100B by TNF-α and epinephrine (Figure 3E).
In these experiments, TNF-α and epinephrine also elevated palmitate concentrations in the L1 cultured media (Supporting Information Figure S2A, B), and insulin significantly reduced the stimulatory effects of TNF-α and epinephrine (Supporting Information Figure S2C). Unlike S100B, MCP-1 secretion from L1 was elevated by insulin and not altered by epinephrine, although TNF-α enhanced MCP-1 secretion (Supporting Information Figure S3).
Effects of mutual stimulation by conditioned media on cytokine release
The stimulation of L1 with RAW conditioned media markedly increased S100B secretion (Figure 4A). Conversely, the treatment of RAW with L1 conditioned media significantly amplified TNF-α secretion from RAW (Figure 4B). Pre-incubation of L1 with epinephrine enhanced the stimulatory effect on TNF-α secretion, whereas epinephrine alone did not show a direct effect on the TNF-α level. In contrast, pre-treatment of L1 with insulin eliminated the L1 conditioned media-induced increase in TNF-α secretion from RAW, with or without pre-incubation of epinephrine (Figure 4B).
The interaction between adipocytes and macrophages in a co-culture system
The S100B protein concentration in the co-cultured media was increased as compared to that in the media of L1 cultured alone (Figure 5A). In contrast, co-culture with RAW downregulated S100B mRNA expression in L1. TNF-α concentration in the co-cultured media was lower than that in the RAW-alone cultured media, but the co-culture significantly enhanced TNF-α mRNA expression in RAW (Figure 5B). The gene expression of F4/80, CD11c, IL-1β, IL-6, and CCR2 was also increased by the co-culture, whereas CD206 in RAW was downregulated and CD163 along with IL-10 tended to be lowered (Figure 5C). The co-culture also elevated MCP-1 levels in the media as well as its mRNA expression in L1 (Supporting Information Figure S4).
The effects of adipocyte S100B silencing and RAGE blockade on macrophage TNF-α secretion
To determine the role of adipocyte-derived S100B on RAW, we knocked down endogenous S100B in L1 using S100B-specific siRNA. The siRNA transfection successfully reduced the expression of S100B mRNA by approximately 70% and of S100B secretion by about 55% as compared with cells transfected with scrambled siRNA (Figure 6A). The levels of MCP-1 and palmitate that likely influence the interaction between adipocytes and macrophages were not changed by the S100B silencing (Figure 6B). The exposure of RAW to the cultured media after the siRNA transfection resulted in a significant attenuation in the TNF-α hypersecretion from the macrophages (Figure 6C).
We also pre-treated RAW with RAGE-neutralizing antibody and exposed the cells to GA-AGEs, recombinant mouse S100B protein, and L1 conditioned media. The binding of the antibody to RAGE was confirmed by immunostaining (Supporting Information Figure S5). As expected, TNF-α secretion from RAW was stimulated by GA-AGEs, S100B protein, and the L1 conditioned media. The effect of AGEs was completely abrogated by the RAGE antibody, whereas the stimulation by S100B or the L1 conditioned media was partially ameliorated (Figure 6D).
In accordance with previous investigations [17, 18], this study showed that 3T3-L1 adipocytes copiously secrete S100B protein. The mRNA expression level of S100B in mature L1 adipocytes and primary adipocytes was close to that of MCP-1, an important cytokine in adipose tissue inflammation. Although the S100B expression was decreased during adipocyte differentiation similar to MCP-1 , a substantial amount of S100B was still secreted from mature adipocytes.
The recombinant S100B protein stimulated RAW macrophages and primary monocytes from murine blood to secrete TNF-α and upregulated the M1 markers CD11b, CD11c, IL-1β, and IL-6, indicating a proinflammatory macrophage phenotype. The downregulation of CD206, a representative M2 marker, appear to support the S100B-induced proinflammatory alterations in macrophages. The alterations in mRNA expression of TNF-α and the M1/M2 markers were reproducible when macrophages were stimulated with adipocyte conditioned media or co-cultured with adipocytes. The silencing of adipocyte S100B or the blockade of RAGE on macrophages ameliorated the TNF-α hypersecretion. These results indicate that S100B may play a role in the adipocyte-macrophage interaction.
Macrophage infiltration into adipose tissues is thought to be integral to the inflammatory changes in adipose tissues seen in obese subjects [1, 2, 22]. Most of the infiltrated macrophages are known to show the “classically activated” feature with proinflammatory M1 markers on their membranes [22, 23], possibly exacerbating inflammation and subsequent insulin resistance. The switching of macrophage polarization from “alternatively activated” or anti-inflammatory M2 to the M1 phenotype is induced by interaction with adipocytes , and it has been suggested that TNF-α, MCP-1, and saturated FFAs such as palmitate are important mediators between both cells . Our results indicate that S100B acts as a novel adipokine that plays a tangible role in the inflammatory interaction of macrophages.
Although the detailed mechanisms underlying the S100B stimulation of macrophages remain to be clarified, RAGE, a receptor to which S100B binds as an agonist  is likely involved, considering previous reports of RAGE-mediated macrophage classical activation . The downstream signals of S100B/RAGE have been shown to include transcription factors NFkB and AP-1 that affect gene expression of proinflammatory cytokines .
Our data also revealed that TNF-α reciprocally stimulated S100B secretion from L1 adipocytes. The increase of S100B by TNF-α was also observed in primary adipocytes from murine epididymal fat. These results suggest that TNF-α and S100B may form a paracrine loop between macrophages and adipocytes, exaggerating the secretion of each other. The effects of mutual stimulation by conditioned media and of co-culture indicate the establishment of a vicious cycle between both types of cells.
While S100B production from 3T3-L1 adipocytes was increased by TNF-α, the S100B mRNA level was significantly decreased. This paradoxical discrepancy in S100B is consistent with previous investigations that demonstrated a lack of protein-mRNA correlation  and a complex, cell-specific mechanism of S100B expression . Although the underlying mechanism of this known disconnect between S100B protein and mRNA expression remains to be clarified, it should be noted that cytokine production is controlled not only by transcription, but also by post-transcriptional mechanisms . It may also be of significance that TNF-α has been reported to modify the translation rate of proteins [30, 31]. These findings may help explain the discrepancy in the S100B gene expression and protein synthesis.
In our experimental model, MCP-1 and FFAs possibly contribute to the adipocyte–macrophage interaction, given the increase in their levels by TNF-α. This is consistent with previously reports . In addition, other adipokines such as adipolin  or macrophage-inducible C-type lectin (Mincle)  may also participate in the interaction. Nevertheless, the observed amelioration of macrophage TNF-α secretion by the suppression of adipocyte S100B without changing MCP-1 and palmitate levels likely indicates that S100B makes a significant contribution to the interaction between both cells. The present results do not exclude the involvement of MCP-1 and other cytokines, because S100B inhibition demonstrated not total, but partial suppression of TNF-α in our experiments. Notably, S100B upregulated CCR2 in macrophages. This may boost MCP-1 effects on macrophages even if the cytokine level is not altered, presumably enhancing chemotactic activity and cytokine release.
In concurrence with previous reports [34, 35], S100B secretion was augmented by epinephrine and suppressed by insulin in our study. S100B has also been shown to be increased by glucagon, isoproterenol, and dibutyryl-cAMP, a cell membrane-permeable cAMP analogue , indicating the crucial role of cAMP in the regulation of S100B. This hormonal regulation may be important from a clinical point of view because catecholamines and glucagon are increased in obesity, metabolic syndrome, and in type 2 diabetes [36-38], whereas insulin action may be impaired in these diseases. Our data may provide an explanation, at least in part, for the reported elevation of S100B levels in adipose tissues in streptozotocin-induced diabetic rats  and in the serum of obese subjects , although there is conflicting data showing decreased serum S100B levels in type 2 diabetic patients . We speculate from our present results that the increased S100B along with elevated FFAs and TNF-α in these disorders may contribute to adipose tissue inflammation as hypothetically depicted in Supporting Information Figure S6, yet in vivo evidence remains to be clarified. It should be noted that FFAs were also shown to increase S100B secretion from rat epididymal fad pads . However, we were unable to demonstrate the effect of FFAs on S100B, and recombinant S100B did not increase FFA release from adipocytes cultured alone (data not shown). Hence it is likely that there is no direct relationship between S100B and lipolysis, although the S100B-induced TNF-α hypersecretion from macrophages may indirectly enhance adipocyte lipolysis as shown in Supporting Information Figure S2.
A limitation of this study is the lack of in vivo data. Although our results demonstrate that S100B influences the interaction of adipocytes with macrophages, its importance relative to other cytokines or FFAs in vivo remains to be clarified. It also needs to be determined whether blood S100B secreted from nerves or other tissues is involved in adipose inflammation. These factors limit the interpretation of the present data in the in vivo role of S100B in adipose tissue inflammation and further in the pathogenesis of obesity and related diseases.
In conclusion, our results demonstrate that adipocyte S100B may play a significant role in the interaction between adipocytes and macrophages. This report identifies S100B as an inflammatory adipokine and adds a novel aspect to its immune-modulating properties observed in other tissues. Future in vivo studies and clinical investigations will help to further understand the importance of this protein in the pathogenesis of obesity, insulin resistance, and the metabolic syndrome.
The authors express sincere thanks to Ms. Michiko Yamada for excellent technical assistance.