Signaling between pancreatic β-cells and macrophages via S100 calcium-binding protein A8 exacerbates β-cell apoptosis and islet inflammation

Chronic low-grade inflammation in the pancreatic islets is observed in individuals with type 2 diabetes, and macrophage levels are elevated in the islets of these individuals. However, the molecular mechanisms underlying the interactions between the pancreatic β-cells and macrophages neutralization in the pancreatic islets. Of note, both glucotoxicity and lipotoxicity triggerred S100A8 secretion from the pancreatic islets, which in turn promoted macrophage infiltration of the islets. Taken together, a positive feedback loop between islet-derived S100A8 and macrophages drives β-cell apoptosis and pancreatic islet inflammation. We conclude that developing therapeutic approaches to inhibit S100A8 may serve to prevent β-cell loss in patients with diabetes.

glucotoxicity) synergistically increased the expression and release of islet-produced S100A8 in a Toll-like receptor 4 (TLR4)-independent manner. Consistently, a significant increase in the expression of the S100a8 gene was observed in the islets of diabetic db/db mice. Furthermore, the isletderived S100A8 induced TLR4-mediated inflammatory cytokine production by migrating macrophages. When human islet cells were cocultured with U937 human monocyte cells, the palmitate treatment upregulated S100A8 expression. This S100A8-mediated interaction between islets and macrophages evoked β-cell apoptosis, which was ameliorated by TLR4 inhibition in the macrophages or S100A8

Introduction
Activation of the innate immune system and circulating levels of acute-phase inflammatory proteins play important roles in the onset and development of type 2 diabetes (1-3). Evidence of chronic inflammation has been demonstrated in the adipose tissue, liver, vascular endothelial cells, circulating leukocytes as well as pancreatic islets in obese and/or diabetic humans (4)(5)(6)(7)(8). Chronic islet inflammation evokes a decline in the β-cell mass by promoting β-cell apoptosis, which is a feature of hallmark of type 2 diabetes (9,10).
It has been reported that macrophages are elevated in the pancreatic islets in patients with type 2 diabetes (11). Chronic hyperglycemia promotes amyloid formation in the islets by inducing the secretion of islet amyloid polypeptide (IAPP) (12), production of reactive oxygen species (ROS) in βcells (13), and formation of advanced glycation end products (AGE) (14,15). These conditions lead to activation of the NLRP3 inflammasomes, IL-1β secretion, macrophage infiltration of the β-cells and pro-apoptotic processes (12,16). Thus, islet inflammation is closely related to β-cell failure and apoptosis in diabetes. A previous study showed that the co-culture of MIN6 insulinoma cells with RAW264.7 macrophage cells in the presence of palmitate increased the expression of inflammatory genes in the MIN6 cells and decreased insulin secretion (17). However, the precise mechanisms involved in the mutual interaction between the pancreatic β-cells and macrophages in diabetes remain unclear.
In this study, we identified S100a8 as an upregulated gene after chronic glucose stimulation, which reflects a state of sustained hyperglycemia, in the pancreatic islets. S100A8 is a small calciumbinding protein that is found at high levels in the extracellular milieu under inflammatory conditions. Furthermore, the S100A8 protein is known to be associated with various chronic inflammatory diseases and both type 1 and type 2 diabetes (18,19). S100A8 is thought to be a member of damageassociated molecular pattern molecules (DAMPs), and stimulates macrophages (20-23). Consequently, to test the hypothesis that S100A8 contributes to islet inflammation we established a co-culture system with freshly isolated primary pancreatic islets and resident peritoneal macrophages to investigate the role(s) of S100A8 in the sustenance of islet inflammation.

S100A8/A9 expression in the islets was upregulated by chronic glucose stimulation
Chronic hyperglycemia induces β-cell apoptosis, in part, through continuous glucokinase activation (24). We previously identified the target genes of glucokinase by examining the gene expression profiles of glucokinase activator (GKA)-treated isolated islets (NCBI GEO database GSE41248) (25). Among them, S100a8 and S100a9 (S100a8/a9) expression showed the greatest increase following chronic glucokinase activation of the islets (90-fold and 254-fold increase, respectively). We validated the gene expression changes in isolated islets stimulated with a glucokinase activator (Fig. 1A) and confirmed that glucose stimulation increased the expression of S100a8/a9 in the islets in a concentrationdependent manner (Fig. 1B). Thus, S100A8 expression was induced by high glucose in the islets without macrophages.

S100A8/A9 expression in the islets was enhanced by co-culture with macrophages in the presence of palmitate
We co-cultured islets with macrophages using coculture inserts ( Fig. 2A) and observed that the islet expression of S100a8, S100a9, Il-1b, Tnf-a, Il-6 and Ccl2 was increased in the presence of macrophages (Fig. 2B). The absence of elevated expression of the macrophage markers, Cd11b and F4/80, suggesting it was unlikely there was contamination of the co-cultured islet samples with macrophages (Fig. 2C). The expression of S100a8 and S100a9, but not of Il-1b, Tnf-a, Il-6, or Ccl2, was enhanced in the islets co-cultured with macrophages in the presence of the saturated fatty acid palmitate (16:0) (Fig. 2B). We confirmed the secretion of S100A8, but not of S100A9, by ELISA assays in the supernatant of islets co-cultured with macrophages in the presence of palmitate (Fig. 2D). Notably, secretion of S100A8 from the macrophages was not affected by the concentration of glucose or palmitate (Fig. 2E). S100A8 proteins were predominantly expressed in the mouse pancreatic islets, but not in acinar cells (Fig. 2F).
To test the possibility that the adipocyte-derived fatty acids contributed to the macrophage-mediated islet inflammation in vivo, we added isolated white adipocytes from the epididymal fat to the co-culture of islets with macrophages and examined islet gene expression. As expected, the adipocytes and macrophages synergistically increased the expression of S100a8 and S100a9 in the islets, and this was not associated with elevation of the expression of macrophage or adipocyte markers (Fig. 3A). Palmitate has been reported to induce islet inflammation through the TLR4/MyD88 pathway (17). Islets obtained from TLR4-knockout mice, and co-cultured with wild-type macrophages in the presence of palmitate showed a significant increase in the expression of S100a8/a9 (Fig. 3B). These results suggest that TLR4-mediated signaling was not required for the S100A8 production induced by macrophage-derived factors and palmitate in the co-cultured islets.

Glucotoxicity further enhanced the induction of S100A8/A9 in co-cultured islets
Chronic high ambient glucose concentration has been shown to accelerate inflammation in various tissues in diabetes (26,27). We undertook experiments under normal glucose (5.6 mmol/L) conditions and under high glucose (11.1 or 22.2 mmol/L) conditions to mimic the environment in diabetes. The protein expression of S100A8/A9 in the co-cultured islets was enhanced following culture in the presence of 11.1 mmol/L glucose (high concentration) (Fig. 4A). S100A8 secretion from islets co-cultured with macrophages in the presence of palmitate was also enhanced by glucose stimulation (Fig. 4B). High glucose enhanced palmitate-induced S100a8 and S100a9 gene expression, while expression of other inflammatory or macrophage markers in the co-cultured islets were not influenced by the glucose concentration (Fig. 4C). In addition, inflammatory gene expression in the macrophages were upregulated by the ambient glucose level (Fig. 4D).
We next examined the expression of S100a8/a9 in the islets of the db/db mouse, an established model of diabetes. Six-week-old and 12-week-old db/db mice exhibited morbid obesity, severe hyperglycemia, and irregular α/β-cell distribution within the islets; however, the ratio of β-to α-cells and the proportion of apoptotic β-cells were not altered in the db/db mice compared to control db/+ mice (Supplemental Table S1 and Fig. 5A and 5B). Isolated pancreatic islets from db/db mice, at both 6 and 12 weeks of age, showed higher levels of expression of S100a8/a9, of the inflammatory cytokine gene Il-6, and of the macrophage marker F4/80 compared to db/+ islets (Fig. 5C).

Factors from the islets, but not from macrophages, activated the macrophages in co-culture via TLR4
In macrophages co-cultured with islets in the presence of palmitate, expression of Tnf-a, Ccl2, Il-1b, Il-6, Il-12, Il-22, Il-23, and Il-24 genes were elevated (Fig. 6A). IL-23 and IL-24 are potent inducers of oxidative and ER stress in β-cells (28).
No increase in the expression of S100a8/a9 genes was observed in the macrophages co-cultured with the islets in the presence of palmitate or in the presence of high ambient glucose (Fig. 6B). This implies that the production of S100A8 induced by co-culture with macrophages was predominantly derived from the islets. The aforementioned increase in the cytokine gene expression was blunted by the TLR4 inhibitory peptide VIPER (Fig.  6C). These results suggest that a humoral factor derived from the co-cultured islets stimulated the macrophages via TLR4.
Because S100A8 is reported as a ligand of TLR4, we examined whether S100A8 induces inflammation in the co-cultured macrophages. Treatment with the recombinant S100A8-GST peptide increased the expression of the Tnf-a, Ccl2, Il-1b, Il-6, Il-12, Il-22, and Il-23 genes in the macrophages and prompted macrophage migration ( Fig. 7A and 7B). Neutralization of S100A8 using an antibody significantly reduced the migration of the macrophages induced by co-culture with islets ( Fig. 7C). Neutralization of S100A8 with an antibody also reduced the cytokine expression of the macrophages induced by co-culture of the macrophages with islets in the presence of palmitate (Fig. 7D). Inhibition of TLR4 with VIPER or TAK242 attenuated the migration of the macrophages and the cytokine expression in these cells induced by purified S100A8-GST peptide ( Fig. 7E and 7F). TLR4 -/macrophages exhibited significant decrease of cytokine expression following co-culture with islets in the presence of palmitate as compared to TLR4 +/+ macrophages (Fig. 7G). Taken together, S100A8 leads to the upregulation of inflammation mediators in macrophages via TLR4.

Co-cultivation of islets with macrophages in the presence of palmitate coordinately promoted βcell apoptosis via islet-derived S100A8 and macrophages
Exogenous S100a8/a9 expression induced by adenoviral transduction was also influenced by the ambient glucose levels (Fig. 8A), suggesting that glucose stimulation possibly enhanced translation or stabilized the S100A8/A9 mRNA or protein.
Overexpression of S100A8/S100A9 exerted no effect on glucose-induced insulin secretion from the islets (Fig. 8B) or on the degree of apoptosis in the islets (Fig. 8C). S100A8-overexpression in the MIN6K8 β-cells slightly decreased insulin secretion from the cells, and conversely, S100a8knockdown was capable of restoring the insulin secretion, even in the absence of macrophages (Fig.  8D). In the presence of macrophages, however, S100A8-overexpression increased insulin secretion, while S100a8-knockdown tended to decrease insulin secretion in S100A8-overexpressed MIN6K8 β cells (Fig. 8E).
To assess the effects of the co-cultivation on the islets, β-cell apoptosis was evaluated. Co-culturing with macrophages increased the number of apoptotic β-cells, and palmitate enhanced macrophage-induced β-cell apoptosis (Fig. 9A).
The combination of glucose stimulation further induced apoptosis of β-cells in the presence of macrophages (Fig. 9A). The apoptosis-associated Bax protein expression level, but not that of the necrosis-associated HMGB1 protein, increased in the islets co-cultured with macrophages in the presence of palmitate and ambient high glucose levels (Fig. 9B). The TLR4 inhibitory peptides VIPER and TAK-242 showed a tendency to reduce β-cell apoptosis caused by co-culturing of the islets with macrophages ( Fig. 9C and 9D). Furthermore, neutralization of S100A8 with an antibody specific to S100A8 significantly reduced the degree of βcell apoptosis in the islets co-cultured with macrophages ( Fig. 9C and 9D).

Expression of S100A8 in human islets
Next, investigation of human islets revealed that expression of S100A8 was significantly enhanced after stimulation with GKA in both non-diabetes and type 2 diabetes donors (Fig. 10A). Coexistence of human monocyte U937 cell line and palmitate significantly upregulated the expression of S100A8 in human islets (Fig. 10B). We also explored the localization of S100A8 in human islets. Immunohistochemical staining for S100A8 was predominantly detected in β-cells in the islets (Fig.  10C). The degree of β-cell apoptosis in the human islets co-cultured with U937 cells and palmitate tended to be decreased by the S100A8-specific neutralizing antibody (Fig. 10D). Further study is warranted to clarify the pathological significance of S100A8 in human islet inflammation.

Discussion
The results of the present study identified S100A8 as an endogenous islet-derived secretory peptide that is induced by a combination of infiltrating macrophages, palmitate (lipotoxicity) and high glucose (glucotoxicity), resulting in the activation of macrophages and potentiation of islet inflammation and β-cell death through a positive feedback loop (Fig. 11). The current results are consistent with a recent report suggesting that the serum level of the S100A8/A9 complex is a sensitive marker of acute inflammation associated with islet transplant rejection (29).
Several studies have shown that TLR4-signaling and MyD88-signaling in β-cells play important roles in the development of islet inflammation (17,30,31). However, TLR4 stimulation did not induce S100A8/A9 production in the islets, while S100A8 protein from the islets acted as a ligand for the TLR4 expressed in macrophages. Because TLR4 and RAGE receptors are also expressed in the pancreatic β-cells (17), it was suggested that S100A8 also exerts direct effects on the β-cells. TLR4 stimulation by S100A8 triggered the release of the inflammatory cytokines Il-12, Il-23 and Il-24 from the macrophages, which resulted in β-cell apoptosis. Further production of S100A8 may evoke inflammation of the surrounding islets, neighboring tissues or feeding vessels. TNF-α, CCL2 and IL-1β are known to induce islet inflammation through the NF-κB pathway (3), and IL-23 and IL-24 have been shown to be potent inducers of oxidative and ER stress in β-cells (28). The expression of these cytokines was also induced in the macrophages following stimulation with S100A8 in the present study.
Overexpression of S100A8/A9 had no inhibitory effect on insulin secretion from isolated islets (Fig.  8B). We also report that S100A8 per se impaired insulin secretion from MIN6K8 β-cells in the absence of macrophages (Fig. 8D). However, overexpression of S100A8 enhanced insulin secretion from MIN6K8 β-cells in the presence of macrophages (Fig. 8E). Overexpression of S100A8 has been reported in ductal adenocarcinoma of the pancreas, and it has been suggested that a peptide metabolite of S100A8 released from pancreatic cancer possibly suppresses insulin secretion to induce diabetes (32). Further research is required to clarify the effects of S100A8 on insulin secretion under similar pathophysiological conditions.
Our results suggested that S100A8 did not act directly to induce β-cell apoptosis, but via a mutual interaction with macrophages. S100A8 in β-cells likely triggers destruction of the β-cell membrane, since S100A8 and S100A9 have the ability to form oligodimers and induce amyloid deposition (33). Since S100A9 protein was detected in the islets but not in the culture media, the contribution of S100A9 in our experimental conditions remains unclear. S100A9 protein, a heterodimerization partner of S100A8, possibly plays an essential role in the protein expression of S100A8 protein, as indicated by previous studies that S100A8 was not detectable in S100A9-KO peripheral tissues (20,34).
Glucokinase-mediated induction of S100A8 production seems specific to pancreatic β-cells, as deduced from the results of previous studies carried out in other organs (35-37). Glucokinase activation enhances adaptive β-cell proliferation and prevents β-cell apoptosis induced by glucotoxicity or ER stress (25,38,39). However, chronic glucokinase activation and high glucose act to trigger β-cell apoptosis (24,39). There is a considerable debate about the role of glucokinase in the protection of βcells from the S100A8-mediated positive feedback loop of islet inflammation under glucolipotoxicity. Additional studies are necessary to confirm the production of S100A8 in human β-cells from obese or diabetes subjects.
In summary, our studies support the identification of S100A8 as a secretory protein to promote β-cell apoptosis and constitute an important step in the development of approaches to protect β-cells in patients with diabetes.

Experimental procedures Animals and animal care
Animal handling procedures were in accordance with institutional animal care and use committee protocols approved by Yokohama City University. Animals were housed in rooms maintained at a constant room temperature (25°C) under a 12-h light (07:00 h)/12-h dark (19:00 h) cycle, and the animals were given free access to food and water. C57BL/6J mice were purchased from CLEA Japan (Tokyo, Japan). All the mice used in this study belonged to the C57BL/6J background. Db/db mice and db/+ mice (BKS.Cg-Lepr db /Lepr db and BKS.Cg-Lepr db /Dock7 m ) were purchased from Charles River Laboratories Japan (Yokohama, Japan). TLR4 -/mice were purchased from Oriental Bio Service (Kyoto, Japan).

Reagents, viruses and cells
The S100A8 and S100A9 ELISA kits were purchased from Cloud-Clone Corporation (Houston, TX). GKA Cpd A was purchased from Merck (Darmstadt, Germany). Collagenase L was purchased from Nitta-Gelatin (Osaka, Japan). Thapsigargin was purchased from SIGMA. Collagenase XI and LPS were purchased from Sigma-Aldrich (St. Louis, MO). The TLR4 peptide inhibitor VIPER (40) was purchased from Novus Biologicals, LLC (Littleton, CO). TAK-242 was purchased from Chemscene, LLC (Monmouth Junction, NJ). D-Mannoheptulose was purchased from Carbosynth Limited (Compton, Berkshire UK). Adenoviruses containing S100A8, S100A9 or LacZ were generated using Virapower adenoviral expression system (Invitrogen). Five micrograms of adenoviral constructs were digested with Pac1, and the linearized DNA was transfected into HEK293A cells. The adenovirus produced by these cells was then collected and subjected to three cycles of freezing and thawing to release the adenovirus. The resulting adenovirus was stored at -80°C for later use. Viral titers were determined by plaque assays using cultured HEK293A cell, according to the manufacturer's instructions.
Monoclonal rat IgG2B anti-mouse S100A8 antibody and monoclonal rat IgG2B Isotype control were purchased from R&D Systems (Abingdon, U.K.). Calcein-AM was purchased from DOJINDO LABORATORIES (Kumamoto, Japan). The MIN6K8 cell line was provided by Dr. Susumu Seino (Kobe University). Lentivirus particles expressing sh-RNA for S100A8 were purchased from Santa Cruz.

Isolation and co-culture of islets, resident peritoneal macrophages and white adipocytes
Islets and peritoneal macrophages were isolated as described elsewhere (41,42). The proportion of F4/80-and CD11b-positive macrophages was more than 90%, as confirmed by flow cytometry (Fig.  2A). Adipocytes were prepared by collagenase digestion (Nitta Gelatin) of epididymal fat tissue, as previously described (43). Glucose-stimulated insulin secretion from the islets was induced as described previously (44). The co-culture was performed at 37°C in a Krebs-Ringer-bicarbonate (KRB) HEPES buffer, pH 7.4, containing 0.2% bovine serum albumin (BSA). Isolated pancreatic islets and peritoneal macrophages were plated in a netwell insert with a 74-µm mesh size polyester membrane (Corning, NY) and in the bottom wells, respectively, and the cultures were incubated for 24 or 48 hours in RPMI1640 containing fatal bovine serum (FBS), KRB containing BSA, or 500 μmol/L palmitate( Fig. 2A). Adipocytes (from 25 mg of epididymal fat) were co-cultured above the coculture netwell insert.

Real-time PCR
Total RNA isolation from pancreatic islets, cDNA synthesis and quantitative PCR were performed as previously described (25,45). Data were normalized according to the expression level of βactin, 18s rRNA, or GAPDH. The primers used for the real-time PCR are listed in Supplemental  Table S2.

GST-fused proteins
GST-fused constructs comprising the mouse S100A8 and S100A9 proteins were generated in pGEX4T-1, which were received as kind gifts from Dr. Sachie Hiratsuka and Dr. Yoshihiro Maru (Department of Pharmacology, Tokyo Women's Medical University) (46). The proteins were expressed in Escherichia coli BL21 and purified on a glutathione-sepharose column.

Macrophage migration assay
The migrated macrophages were labeled with Calcein-AM (WAKO, Osaka, Japan) and measured using ARVO™ MX plate reader (PerkinElmer, MA) at an excitation wavelength of 485 nm and emission filter of 535 nm. The migration of macrophages was evaluated using 8-μm pore Falcon BD FluoroBlok™ inserts and plates (Becton Dickinson, Franklin Lakes, NJ) with isolated pancreatic islets (50 islets) or S100A8-GST in the presence or absence of anti-S100A8 antibody, isotype control IgG2b, VIPER/CP7 or TAK242. Macrophages were seeded on to the insert mesh and incubated for 24 or 48 hours at 37°C under 5% CO2.

Fig. 11 An illustrative model of islet-inflammation-induced β-cell apoptosis via S100A8 in diabetes.
A combination of infiltrating macrophages, saturated fatty acids (palmitate) and hyperglycemia augmented the production of S100A8. S100A8 secreted from the islets induced further macrophage migration and inflammation through TLR4. This positive feedback loop potentiates islet inflammation and β-cell death.