Mannose-Binding Lectin Is Required for the Effective Clearance of Apoptotic Cells by Adipose Tissue Macrophages During Obesity

Growing evidence links obesity-induced chronic inflammation in white adipose tissue (WAT) to the development of insulin resistance (1–3). Both the adaptive and innate immune systems are critically involved in modulating inflammatory response, as illustrated by the infiltration of WAT by various immune cells, including macrophages, B cells, and T cells (4–6). Various triggers are thought to contribute to macrophage infiltration in WAT, although the molecular mechanisms mediating the influx are not yet fully understood. It has been suggested that adipocyte hypertrophy results in enhanced secretion of proinflammatory and chemoattractant-like adipokines, triggering influx of macrophages (7). In addition, enhanced release of free fatty acids (FFAs) as well as development of hypoxia in the expanding adipose tissue may increase chemokine expression and promote infiltration of immune cells into adipose tissue (8). Last, adipocyte death may promote macrophage abundance in adipose tissue, with a large proportion of macrophages directed to crown-like structures surrounding necrotic or apoptotic cells (9–11).

Evidence indicating that, in addition to quantitative changes, macrophages in WAT also undergo qualitative changes during obesity, leading to the presence of distinct subclasses of macrophages, abounds (12,13). This functional heterogeneity among macrophages is broadly categorized following the M1/M2 paradigm, in which M1 macrophages represent classically activated macrophages characterized by enhanced inflammatory cytokine production, whereas M2-polarized macrophages generally attenuate inflammation and are involved in restoring tissue homeostasis (14,15).

Mannose-binding lectin (MBL) plays a role in the first-line host defense as part of the innate immune system (16,17). Primarily synthesized in the liver, MBL circulates in serum and has been suggested to function as an acute phase reactant with rapid sequestration toward sites of inflammation (18–20). MBL possesses a dual though complementary function. First, MBL is capable of binding to specific carbohydrate patterns present on microbial cell surfaces and enhancing opsonophagocytosis by leukocytes. MBL also binds endogenous ligands, including the cell surface protein calreticulin, present on apoptotic and necrotic cells (21,22). During opsonophagocytosis, MBL directly interacts with macrophage receptors (e.g., MD2 and CD14) and alters macrophage cytokine expression (23–25). Second, MBL can activate the lectin branch of the complement system. Upon its activation, MBL interacts with the so-called mannose-binding associated proteases to induce the complement cascade (26).

Interestingly, circulating concentrations of MBL vary greatly between individuals; these differences are related to several common variant alleles in the MBL gene (27,28). MBL deficiency has been linked to several inflammatory diseases such as rheumatoid arthritis (29) and coronary artery disease (30). It was recently suggested that the concentration of MBL in serum might influence the etiology of obesity and type 2 diabetes. Indeed, women with genetically determined lower concentrations of MBL were at greater risk of developing gestational diabetes (28). In addition, concentrations of MBL negatively correlated with obesity and insulin resistance in a mixed population and in women with polycystic ovary syndrome (31,32). In contrast, a positive correlation was reported between MBL serum concentrations and the development of type 2 diabetes (33). Conceivably, identification of its mode of action may help resolve the apparent conflicting data on MBL and insulin resistance/type 2 diabetes. Insofar as adipocyte death is a key feature of obese adipose tissue, we hypothesized that MBL, in line with its ability to bind endogenous ligands present on apoptotic cells, may play a role in recognizing dying adipocytes and thereby modulate the development of obesity-induced inflammation in WAT. To test this hypothesis, wild-type (WT) and MBL-deficient mice were fed a high-fat diet (HFD) to induce obesity, and the phenotypes of both strains were carefully characterized.

Adipose tissue samples were obtained from healthy, overweight (BMI 27–35 kg/m²), and lean subjects (BMI 20–25 kg/m²) between 40 and 70 years old. Individuals with signs of a current infection and a history of recurrent infections were excluded from the study. Furthermore, autoimmune diseases, immunodeficiency, or immunosuppressive treatment (including tumor necrosis factor [TNF]-α blocking agents and corticosteroids) also were reasons to exclude subjects from this study. Subcutaneous adipose tissue biopsies were obtained using local anesthesia and performing needle biopsies 6–10 cm lateral to the umbilicus after an overnight fast. The study protocol was approved by the University of Nijmegen Ethics Committee, and all participants gave written informed consent. Immunohistochemical detection of CD68 in adipose tissue was performed using a CD68 monoclonal antibody (AbD Serotec, Oxford, UK). The percentage of macrophages was expressed as the total number of CD68-positive cells divided by the total number of adipocytes counted in 20 random microscopic fields. The HOMA-IR was calculated as the product of the fasting insulin concentration (microunits per milliliter) and the fasting glucose concentration (millimoles per liter) divided by 22.5.

Three breeding pairs of MBL^−/− mice (lacking both MBL-A and MBL-C) and corresponding WT mice on a C57BL/6 background were obtained from The Jackson Laboratory and further expanded in our local animal facility. Male mice (10–12 weeks old) were fed a semipurified low-fat diet (LFD) for 2 weeks. Thereafter, mice either continued on the LFD or switched to an HFD for 20 weeks, providing 10% or 45% energy from fat (D12450B or D12451; Research Diets). In both diets, lard was replaced by palm oil. At the end of the diet intervention, blood was collected and liver and epididymal WATs were dissected, weighed, and immediately frozen in liquid nitrogen. To deplete macrophages, animals were intraperitoneally injected with clodronate liposomes. Control animals received liposomes containing PBS. All experiments were approved by the Animal Ethics Committee of Wageningen University or Leiden University.

Animals were subjected to indirect calorimetry analysis (PhenoMaster; TSE Systems, Bad Homburg, Germany). A period of 48 h of acclimatization was included before starting the experiment. Oxygen consumption (Vo₂) and carbon dioxide production (Vco₂) were determined at 20-min intervals. Respiratory exchange ratio was calculated as the ratio between Vco₂ and Vo₂. Energy expenditure, fat oxidation rate, and carbohydrate oxidation rate were calculated as previously described (34). Data from the light and dark phase were averaged and tested separately to distinguish periods of high and low physical activity.

Plasma concentrations of glucose, triglycerides, cholesterol (INstruchemie), and FFAs (Wako Chemicals) were determined following the manufacturer’s instructions. Insulin (Alpco Diagnostics), leptin (R&D Systems), and MBL-A and -C (HK209 and HK208; Hycult Biotech) concentrations were determined by means of ELISA according to the manufacturer’s instructions.

For the insulin tolerance test, mice fasted for 4 h were injected intraperitoneally with insulin (0.75 unit/kg body weight). For the glucose tolerance test, mice fasted for 4 h were injected intraperitoneally with glucose (1 g/kg body weight). Blood was collected from the tail at specific time intervals, and glucose was measured using Accu-Chek Compact Plus.

Liver triglycerides were determined in 10% liver homogenates prepared in buffer containing 250 mmol/L sucrose, 1 mmol/L EDTA, and 10 mmol/L Tris-HCl at pH 7.5 (INstruchemie).

TRIzol reagent (Invitrogen) was used to isolate RNA from various tissues. RNA was reverse transcribed (iScript cDNA Synthesis Kit; Bio-Rad) and real-time quantitative PCR was performed using a SensiMix SYBR kit (Bioline) in a CFX384 Bio-Rad real-time quantitative PCR system. Expression values were normalized using 36B4 as a housekeeping gene. Primer sequences were derived from the Harvard Primer Database.

Bone marrow cells were isolated from the femur of WT mice and cultured using standard protocol. Briefly, freshly isolated cells were cultured and differentiated into bone marrow–derived macrophages in DMEM (Gibco) containing 10% FCS, 1% penicillin/streptomycin, and 30% L929 conditioned medium during 5–8 days.

Using a Vybrant assay according to the manufacturer’s instructions (Invitrogen), the phagocytic capacity was tested. Cells were preincubated for 1 h with recombinant MBL-A and MBL-C (R&D Systems) before the phagocytosis assay was performed. The phagocytosis assay using apoptotic 3T3-L1 adipocytes was performed with RAW264.7 cells. 3T3-L1 differentiated adipocytes were rendered apoptotic using staurosporine treatment at 1 μmol/L for 18 h. Adipocytes were trypsinized and pelleted at 400 g for 5 min. After resuspension, cells were added to PKH26-stained RAW264.7 cells. The RAW264.7 cells were stained with PKH26 (Sigma-Aldrich, St. Louis, MO) following the manufacturer’s instructions.

Epididymal WAT was dissected and kept in PBS. The tissue was minced, digested for 1 h at 37°C (0.5 g/L collagenase [type 1] in HEPES buffer [pH 7.4] with 20 g/L of dialyzed BSA [fraction V; Sigma]) and filtered through a nylon mesh (236-μm pore). The stromal vascular cells were fixed using 0.5% paraformaldehyde, stored in FACS buffer (PBS, 0.02% natrium azide, 0.5% FCS) in the dark at 4°C and analyzed using flow cytometry within 1 week. Alternatively, the stromal vascular fraction (SVF) was plated in black 96-well culture plates (Corning). After the cells adhered to the cell culture plate, the Vybrant phagocytosis assay was performed.

The isolated adipocytes were incubated for 2 h at 37°C in DMEM/F12 with 2% BSA with or without 8-bromoadenosine cAMP (10⁻³ mol/L; Sigma) and/or insulin (10⁻⁹ mol/L) to determine the antilipolytic effect of insulin. Glycerol concentrations were determined using a free glycerol kit (Sigma) with the inclusion of the hydrogen peroxide–sensitive fluorescence dye Amplex UltraRed (35).

Stromal vascular cells were analyzed using flow cytometry. Cells were stained with fluorescently labeled antibodies for CD45.2-FITC (104; BioLegend), F4/80-PE (BM8; eBioscience), CD11B-PB (M1/70; BioLegend), and CD11C-APC Cy7 (N418; eBioscience). Cells were measured on an LSR II flow cytometer (BD Biosciences). The M1-to-M2 ratio was determined using the presence of either M1 markers (F4/80+CD11B+CD11C+) or M2 markers (F4/80+CD11B+CD11C−) on the membrane of cells that are part of the SVF of the adipose tissue. Data were analyzed using FlowJo software.

Hematoxylin and eosin (H-E) staining of liver and WAT was performed using standard protocols. To detect macrophages/monocytes, paraffin-embedded sections of WAT and liver were stained with an F4/80 antibody (AbD Serotec), which was visualized by applying 3,3-diaminobenzidene for 5 min.

The quantity of apoptotic cells in WAT was determined by means of a TUNEL assay (Roche Applied Science). Paraffin-embedded, 8-µm sections of WAT were deparaffinized and incubated with 20 µg/mL proteinase K for 10 min. Thereafter, sections were incubated with TUNEL reaction mixture for 60 min at 37°C in a humidified chamber. The sections were analyzed by means of an Olympus CKX41 microscope by analyzing 70 magnified photos (original magnification ×200) with TUNEL staining of cells in both HFD conditions. Values are expressed as the percentage of apoptotic cells among total adipocytes.

Results are shown as mean ± SEM. Statistically significant differences were calculated using a Student t test or a two-way ANOVA. The cutoff value for statistical significance was set at P = 0.05.

First we determined the effect of HFD-induced obesity on circulating MBL concentrations. Remarkably, obese mice fed an HFD for 20 weeks displayed significantly higher concentrations of circulating MBL-A and MBL-C, the two forms of MBL that exist in mice, compared with mice fed the LFD (Fig. 1A and B). In contrast, short-term feeding for 4 days using a diet consisting of 60% energy derived from fat only led to upregulation of MBL-A in the circulation, whereas MBL-C concentrations were not affected (Supplementary Fig. 1A).

To evaluate the role of MBL in the etiology of diet-induced obesity, we studied the effect of MBL ablation in mice fed an LFD or HFD for 20 weeks. Although both WT and MBL^−/− animals became obese eating the HFD, animals lacking MBL gained significantly more body weight (WT 43.2 g, MBL^−/− 49.2 g; P = 0.002) (Fig. 1C). Food intake, however, was similar between both groups (Supplementary Fig. 1B). Using indirect calorimetry, no differences in total energy expenditure were observed between WT and MBL^−/− animals during both LFD and HFD conditions (Fig. 1D). Plasma concentrations of FFAs and leptin were significantly higher in HFD-fed MBL^−/− mice compared with WT mice, whereas no significant differences were observed between the groups fed an LFD (Table 1). However, while HFD induced fasting hyperglycemia and hyperinsulinemia in both MBL^−/− and WT mice, no differences in plasma glucose and insulin concentrations were observed between the genotypes (Table 1).

Because the liver is the main site of production of MBL, we set out to characterize hepatic changes in WT and MBL^−/− mice. Interestingly, hepatic MBL production seemed to primarily take place in hepatocytes; deletion of Kupffer cells from the liver did not significantly alter expression levels of MBL-A or MBL-C (Supplementary Fig. 2). Liver weight was slightly but not significantly increased in HFD-fed MBL^−/− mice compared with HFD-fed WT mice (Supplementary Fig. 3). H-E staining and measurements of hepatic fat content unveiled a small, nonsignificant increase in liver fat content in MBL-deficient mice (Fig. 1E and G). Furthermore, no substantial differences in the numbers of Kupffer cell were observed between WT and MBL^−/− mice, as visualized by F4/80 staining (Fig. 1F), nor were any differences in collagen deposition seen using Sirius Red staining (Supplementary Fig. 4A). Real-time quantitative PCR analysis of the liver revealed that several pro- and anti-inflammatory cytokines were upregulated in the absence of MBL. Hepatic expression levels of the transcription factor peroxisome proliferator–activated receptor α, known to be activated by fatty acids, were similarly regulated between WT and MBL^−/− mice (Supplementary Fig. 4B).

Interestingly, the weight of the various adipose tissue depots was altered between WT and MBL^−/− mice fed the HFD (Supplementary Fig. 5). Adipocyte cell size was significantly enhanced in MBL^−/− mice compared with WT mice after the HFD intervention (Fig. 1H and J). Strikingly, the abundance of macrophages in adipose tissue was markedly elevated in MBL^−/− mice, as shown by immunohistochemical staining (Fig. 1I) and quantification of the number of crown-like structures (Fig. 1K).

Consistent with the enhanced influx of macrophages into adipose tissue as determined by an immunohistochemical approach, gene expression levels of general macrophage markers F4/80 and CD68 were increased three- to fourfold in MBL^−/− mice compared with WT mice (Fig. 2A). Noticeably, gene expression levels of the chemoattractant MCP-1 were increased by the HFD intervention yet no differences between both genotypes were displayed (Fig. 2A). To further investigate the degree of WAT inflammation, expression levels of several pro- and anti-inflammatory genes were measured. Gene expression levels of the proinflammatory marker interleukin (IL)-1β were not altered in MBL^−/− mice (Supplementary Fig. 6). In contrast, expression of TNF-α, a proinflammatory cytokine involved in adipocyte apoptosis, was threefold higher in MBL^−/− mice fed an HFD compared with WT mice fed an HFD (Supplementary Fig. 6). A similar result was found for transforming growth factor-β, an immunosuppressive cytokine capable of polarizing macrophages toward an alternative phenotype (Supplementary Fig. 6). Expression of various M2 markers including IL-1 receptor antagonist, IL-10, and CD206, were markedly elevated in adipose tissue in HFD-fed MBL^−/− mice, whereas FIZZ1 levels were only mildly elevated in animals lacking MBL (Fig. 2B). In contrast, the typical M1 marker genes IL-6, IL-12, and osteopontin were only mildly higher in HFD-fed MBL^−/− mice compared with WT mice (Fig. 2B). Noticeably, high levels of CD11C also suggest the presence of M1-polarized cells. However, CD11C-positive M1-like adipose tissue macrophages do possess a high degree of phenotypic plasticity (13). FACS analysis showed more M2 macrophages compared with M1 macrophages present in the adipose tissue of HFD-fed MBL^−/− mice (Fig. 2C).

Apoptosis is increasingly considered a potent initiator of macrophage influx into adipose tissue (10,11). MBL is capable of binding to apoptotic cells and has been shown to be involved in the clearance of apoptotic cells in vivo, with a delayed clearance of apoptotic cells observed in MBL^−/− mice (19,22). Consistent with this notion, TUNEL staining revealed a modest but significantly higher percentage of apoptotic cells in WAT of HFD-fed MBL^−/− mice compared with WT mice (P = 0.01) (Fig. 3A and B). In line with these results, cleaved concentrations of caspase-3, a marker for apoptosis, were increased in adipose tissue of HFD-fed MBL^−/− mice compared with WT mice (Fig. 3C). Noticeably, although caspase-3 concentrations were elevated in adipose tissue of MBL^−/− mice versus WT mice, differences did not reach statistical significance (upper band: WT 0.035 ± 0.001, MBL^−/− 0.077 ± 0.025 [P = 0.12]; lower band: WT 0.025 ± 0.005, MBL^−/− 0.035 ± 0.009 [P = 0.36]). In line with the enhanced presence of apoptotic cells, gene expression levels of apoptotic cell–recognizing mediators including milk fat globule-EGF factor 8 protein (MFGE8), growth arrest-specific 6 (GAS6), c-mer proto-oncogene tyrosine kinase (MERTK), and low-density lipoprotein receptor-related protein 1 (LRP1) were enhanced in adipose tissue in the absence of MBL (Fig. 3D). Because MBL is known to function as an opsonin affecting phagocytosis, we hypothesized that the absence of MBL may affect the clearance of apoptotic cells from adipose tissue (21,22). Therefore, we set out to quantify the phagocytic capacity of the SVF known to carry out phagocytic functions in the adipose tissue of chow-fed animals (36). This approach allowed us to compare SVFs with a similar cellular composition between WT and MBL^−/− animals. As shown in Fig. 3E, the phagocytic capacity of the SVF was lower in cells isolated from epididymal adipose tissue of MBL^−/− animals, whereas the addition of recombinant MBL greatly enhanced the phagocytic capacity of the SVF. Interestingly, results of insulin and glucose tolerance tests were unaltered by MBL deficiency (Fig. 3F). We also performed ex vivo measurements of adipocytes isolated from the epididymal adipose tissue. No differences ex vivo were observed for responsiveness of the adipocytes to insulin after 20 weeks of HFD feeding (Supplementary Fig. 7A). In addition, we also tested intracellular signal routes activated by insulin within adipose tissue of LFD- and HFD-fed WT and MBL^−/− animals. To that end, isolated and cultured adipose tissue explants were stimulated with insulin for 20 min. Insulin treatment of the adipose tissue explants led to phosphorylation of AKT. The HFD intervention reduced sensitivity to insulin, as visualized by a reduction in pAKT levels. However, no differences were observed between both genotypes (Supplementary Fig. 7B).

To monitor the effect of MBL on the phagocytosis of apoptotic adipocytes by macrophages, we developed an assay to monitor uptake using flow cytometry. Fully differentiated 3T3 adipocytes were rendered apoptotic. As shown in Fig. 4A, based on amounts of cleaved caspase-3 within 3T3 adipocytes, cells were treated with staurosporine (1 μmol/L) to induce apoptosis. After overnight treatment, cells were harvested and fed to RAW264 macrophages. After 4 h of incubation of apoptotic adipocytes with PKH26-labeled RAW264 cells, the macrophages were stained with Bodipy 493/502 to quantify the lipid content of the macrophage to determine the presence/uptake of lipids. As shown in Fig. 4B, we were able to monitor the interaction between both cells using fluorescent microscopy. In addition, Bodipy signals were measured in PKH-positive cells using flow cytometry (Fig. 4C). The addition of recombinant MBL modulated the Bodipy signals in RAW264 cells. Surprisingly, lower Bodipy signals were observed in cells in the presence of recombinant MBL (Fig. 4D), suggesting that MBL not only functions as an opsonin but also modulates macrophage function to improve handling of apoptotic cells that contain large amounts of lipids. However, analysis using RAW264 macrophages revealed that recombinant MBL was unable to affect the fatty acid oxidation rate. In contrast, lipopolysaccharide did lower the fatty acid oxidation rate in these cells (Supplementary Fig. 8).

To confirm the effects of MBL on systemic insulin sensitivity and macrophage influx in humans, we quantified insulin sensitivity (HOMA), the number of macrophages present in subcutaneous adipose tissue, and circulating MBL concentrations in lean and obese subjects. Interestingly, we were able to detect the presence of MBL in human adipose tissue using an immunohistochemical approach (Fig. 5A). Circulating MBL values were not affected by body weight nor by insulin sensitivity (similar concentrations in subjects with high [>1] or low [<1] HOMA and with high [>25 kg/m²] or low [<25 kg/m²] BMI) (Fig. 5B). However, in line with results obtained from the animal study, subjects characterized by a high influx of macrophages into subcutaneous adipose tissue (>3 macrophages/100 adipocytes) displayed significantly lower concentrations of circulating MBL (Fig. 5B). No correlation between age and circulating MBL concentrations existed (r = 0.037; P = 0.75). In addition, no significant differences in HOMA-IR, FFA levels, total cholesterol, and LDL cholesterol existed between young and old subjects who were part of the study (independent Student t test between the 25th and 75th percentiles of age).

MBL is an important mediator in the clearance of apoptotic cells, and clustering of MBL on the surface of apoptotic cells has been shown to trigger their phagocytosis by macrophages (19–22). We found that the percentage of apoptotic cells is significantly higher in WAT of MBL^−/− mice fed an HFD. Hence, for the first time our data point to a role for MBL in the effective recognition and removal of apoptotic cells from adipose tissue during the development of obesity. Moreover, analysis of the stromal vascular cells isolated from adipose tissue revealed that MBL profoundly affects the phagocytic capacity of these cells. Furthermore, our data indicate that MBL specifically affects the phagocytosis of apoptotic adipocytes.

MBL^−/− mice exhibit increased macrophage infiltration in WAT after eating an HFD, which may be directly linked to defective apoptotic cell clearance. Indeed, phagocytosis of apoptotic cells by macrophages was previously shown to suppress infiltration of monocytes (37). Apoptotic cells have been identified as key initial triggers of macrophage infiltration in WAT (10,11). Intriguingly, MBL deficiency not only increased the number of macrophages in WAT upon HF feeding but also changed the phenotype of macrophages present in WAT. Adipose tissue macrophages in obese MBL^−/− mice were primarily polarized toward a M2 phenotype, which is generally considered to be of a more anti-inflammatory nature and involved in tissue homeostasis and remodeling. Our data are in close agreement with a recently published study that showed that macrophage markers including CD206, IL-10, MCP-1, F4/80, and TNF-α are elevated upon adipocyte apoptosis (11), corroborating the link between apoptotic cell clearance, alternatively activated macrophage influx, and MBL.

In line with its role in apoptotic cell clearance and M2 macrophage infiltration, MBL also has been shown to directly modulate the function and cytokine production of monocytes, macrophages, and dendritic cells. The vast majority of studies indicate that MBL inhibits proinflammatory cytokine production (24,25,38–40). Contradicting results, however, depending on the level of recombinant MBL or the experimental setup used, have been published (41–44). Modulation of the macrophage response in the absence of MBL may conceivably be caused by the loss of the well-known interaction of MBL with Toll-like receptor 4 (23). The absence of MBL might therefore alter several signaling pathways that may subsequently influence the uptake of cellular debris and the influx of macrophages in WAT. Indeed, the presence of MBL seems to enhance the phagocytic capacity of the nonadipocyte fraction of the adipose tissue. Notably, recombinant MBL has been shown to enhance fatty acid oxidation in muscle cells in vitro (31). These results suggest that MBL not only functions as an opsonin but also modulates macrophage function that may improve handling of apoptotic cells that contain large amounts of lipids. Indeed, results from the in vitro assay using recombinant MBL demonstrate a role in lipid handling within the macrophage, although a direct effect of MBL on fatty acid oxidation in macrophages seems to be unlikely.

While macrophage infiltration and body weight were significantly enhanced in HFD-fed MBL^−/− mice compared with HFD-fed WT mice, systemic insulin sensitivity and glucose tolerance were not altered. The disconnection between weight gain and insulin sensitivity/glucose tolerance may be related to the preferential presence of M2 macrophages. Inasmuch as M2 macrophages are generally considered to be more anti-inflammatory compared with the insulin-desensitizing M1 macrophages (14,15), the abundance of M2 macrophages may confer protection against the development of insulin resistance.

In our study, low circulating concentrations of MBL in humans were linked to enhanced presence of macrophages in adipose tissue, whereas plasma concentrations of MBL were not associated with HOMA or BMI. Observational studies have generated conflicting results concerning MBL plasma concentrations and obesity and insulin resistance. Fernández-Real et al. (31) found that serum MBL concentrations were lower in obese subjects than in lean subjects. Furthermore, serum MBL concentrations correlated positively with insulin sensitivity. In contrast, Muller et al. (33) showed that carriers of a mutation leading to higher serum MBL concentrations have an increased risk for developing type 2 diabetes. The reason for these discrepant findings remains elusive. Clearly, additional research is necessary to clarify the relation between serum MBL and obesity and insulin resistance in humans.

Although more work is needed to further characterize the exact molecular mechanisms controlled by MBL in adipose tissue, our results convincingly demonstrate that MBL deficiency is associated with a higher percentage of apoptotic adipocytes, reduces the phagocytic rate in adipose tissue, and promotes the presence of M2-polarized macrophages. Moreover, MBL directly affects the phagocytic capacity of the stromal vascular cells in the adipose tissue and alters the response of macrophages toward apoptotic adipocytes.

We conclude that MBL is a vital component in the recognition and removal of apoptotic cells in WAT during obesity. However, the defective recognition of apoptotic cells induces a compensatory response that leads to a massive influx of Th2-polarized macrophages aimed at removing dying cells. The alternatively activated macrophages limit excessive Th1-dependent inflammation and seem to uncouple obesity and adipocyte enlargement from the development of systemic insulin resistance in the absence of MBL.

Funding. This research was funded by the European Foundation for the Study of Diabetes (EFSD) and the Netherlands Organisation for Scientific Research (VIDI grant to R.S. [no. 016.136.311]). Additional support was obtained through grants from the Centre for Medical Systems Biology; the Netherlands Consortium for Systems Biology (NCSB), established by the Netherlands Genomics Initiative/Netherlands Organisation for Scientific Research (NGI/NOW); and the Leiden University Medical Center. R.H.H. is supported by a VENI grant from ZonMw (no. 91613050).

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. R.S. and S.K. wrote the manuscript and researched data. W.D., L.v.B., H.J., M.H., R.H.H., S.D., and V.v.H. researched data. W.D., L.v.B., H.J., M.H., V.v.H., K.W.v.D., and C.J.T. reviewed/edited the manuscript. R.S. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.