Macrophage Migration Inhibitory Factor Deficiency Ameliorates High-Fat Diet Induced Insulin Resistance in Mice with Reduced Adipose Inflammation and Hepatic Steatosis

Macrophage infiltration is a critical determinant of high-fat diet induced adipose tissue inflammation and insulin resistance. The precise mechanisms underpinning the initiation of macrophage recruitment and activation are unclear. Macrophage migration inhibitory factor (MIF), a pro-inflammatory cytokine, displays chemokine-like properties. Circulating MIF levels are elevated during obesity however its role in high-fat diet induced adipose inflammation and insulin resistance remains elusive. Wildtype and MIF−/− C57Bl\6J mice were fed chow or high-fat diet. Body weight and food intake was assessed. Glucose homeostasis was monitored by glucose and insulin tolerance tests. Adipose tissue macrophage recruitment and adipose tissue insulin sensitivity was evaluated. Cytokine secretion from stromal vascular fraction, adipose explants and bone marrow macrophages was measured. Inflammatory signature and insulin sensitivity of 3T3-L1-adipocytes co-cultured with wildtype and MIF−/− macrophage was quantified. Hepatic triacylglyceride levels were assessed. MIF−/− exhibited reduced weight gain. Age and weight-matched obese MIF−/− mice exhibited improved glucose homeostasis coincident with reduced adipose tissue M1 macrophage infiltration. Obese MIF−/− stromal vascular fraction secreted less TNFα and greater IL-10 compared to wildtype. Activation of JNK was impaired in obese MIF−/−adipose, concomitant with pAKT expression. 3T3-L1-adipocytes cultured with MIF−/− macrophages had reduced pro-inflammatory cytokine secretion and improved insulin sensitivity, effects which were also attained with MIF inhibitor ISO-1. MIF−/− liver exhibited reduced hepatic triacyglyceride accumulation, enhanced pAKT expression and reduced NFκB activation. MIF deficiency partially protects from high-fat diet induced insulin resistance by attenuating macrophage infiltration, ameliorating adipose inflammation, which improved adipocyte insulin resistance ex vivo. MIF represents a potential therapeutic target for treatment of high-fat diet induced insulin resistance.

MIF is a pleiotropic cytokine, ubiquitously expressed and paramount in regulating inflammatory responses [21,22]. Clinical studies suggest a link between MIF and adiposity. Circulating plasma MIF and mononuclear cell MIF mRNA expression are associated with increasing BMI and fatty acid concentration, impaired glucose tolerance and T2D [23][24][25], whereas reducing body weight or metformin treatment decreases serum MIF levels [26,27]. Human and murine adipocytes express and secrete MIF [24,28,29]. Moreover glucose and insulin regulate MIF expression in adipocytes [30]. Recently MIF was identified as a non-cognate ligand of CXC chemokine receptors CXCR2 and CXCR4 in macrophages and T-cells respectively [31]. More recently in an atherosclerotic mouse model (LDLR 2/2 MIF 2/2 ) maintained on a standard chow diet, absence of MIF reduced monocyte adhesion, macrophage lesion content, and atherosclerotic lesion size; coincident with improved glucose homeostasis [32].
Our study was undertaken to further define the impact of the obesogenic environment induced by high-saturated fat feeding on insulin sensitivity in a MIF-deficient setting. We hypothesized that lack of MIF protein would protect mice from the adverse effects of high-fat feeding by reducing ATM recruitment and improving adipose tissue immunophenotype. Additionally we examined the functional consequences of MIF 2/2 macrophages on adipocyte biology and explored whether inhibition of exogenous MIF could block macrophage-induced adipocyte IR in vitro. This study highlights MIF as a critical mediator of ATM recruitment and regulator of adipose tissue inflammation during HFD-induced obesity.

Ethics statement
Ethical approval was obtained from UCD Ethics Committee and mice were maintained according to European Union and Irish Department of Health regulations. Body weight was monitored prior to and after all metabolic procedures to ensure full recovery of the animals. Any animals found to be exhibiting symptoms of pain or distress during or after procedure were euthanized immediately.

Body mass composition
Body mass composition was measured using Bruker's minispec LF50 body composition analyzer (Bruker Optik GmbH, Germany). Lean tissue, fat and fluid was calculated based on body weight of mouse (University College Cork/Alimentary Pharmabiotic Centre).

Insulin secretory response
Overnight fasted mice were subjected to i.p. glucose challenge (1.5 g glucose/kg bw). Blood samples were collected by tail vein bleed prior to and 30, 60 minutes post glucose challenge. Plasma insulin secretory response was determined using an ultra-sensitive insulin ELISA kit (Crystal Chem Inc., USA).

Isolation of stromal vascular fraction (SVF) and flow cytometry
To separate the SVF from adipocytes, epididymal adipose tissue (EAT) was minced then collagenase (2 mg/ml) digested for 70 minutes. Adipocytes were removed and digested EAT suspension was filtered and centrifuged for 5 minutes at 1,700 rpm. The SVF was re-suspended and blocked in 2% BSA/PBS. Cells were stained with fluorescently labelled antibodies; F4/80-FITC, CD11B-AF647/PE, CD11C-RPE, CD3-APC, CD4-FITC, CD8-PE (AbD Serotec, Kidlington, UK). Unstained, single stained and fluorescence minus one controls were used for setting compensation and gates. Flow cytometry was performed on Dako CyAn ADP platform and analyzed using Summit v4.3 software (Beckman-Coulter Ltd., UK). Adipocytes and remaining SVF were seeded at 1610 6 cells/ml and cultured in serum rich media (DMEM, 10% FBS and 1% penicillin/streptomycin) for 24 h. Media was harvested for cytokine profiling. Cells were resuspended in TRI ReagentH for gene expression analysis.
Ex vivo adipose tissue culture EAT explants harvested from lean and obese mice (50 mg/well) were cultured in serum rich media for 24 h. Tissue was homogenized in 1 ml radioimmunoprecipitation assay (RIPA) buffer using a tissue lyser (Qiagen, West Sussex, UK).

Insulin-stimulated glucose uptake into adipose explants
Adipose explants (50 mg) were placed in PBS+0.2% BSA prior to stimulation 6 insulin (100 nM) for 15 min. 3 H-glucose (0.1 mM 2-deoxyglucose+0.5 mCi/ml 3 H-deoxyglucose) was added for 45 min. Tissue was washed, lysed in RIPA buffer and homogenized using a tissue lyser (Qiagen, West Sussex, UK). Glucose uptake was measured by liquid scintillation counting. Fold increase in glucose uptake over basal was calculated for each individual mouse.

Cell culture
Bone marrow derived cells were isolated from femurs and tibias of 8-11 week old lean WT and MIF 2/2 mice. Cells were cultured in serum rich media supplemented with 30% L929 conditioned medium for 7 days at 37uC to differentiate to bone marrow macrophage (BMM). On day 7 of differentiation macrophages were treated 6 lipopolysaccharide (LPS) (10 ng/ml) for 30 minutes; cells were harvested for protein or incubated with fresh media for 6-24 h to measure cytokine secretion. 3T3L1-fibroblasts (ECACC, Salisbury, UK) were exposed to adipogenic medium containing 10% FBS, 0.5 mM isobutylmetylxanthine, 1 mM dexamethasone and 10 mg/ml insulin to induce differentiation.

Co-culture assay
WT and MIF 2/2 BMM were seeded on 0.4 mm Corning transwell filters, LPS-stimulated for 30 minutes and washed with PBS. Fresh media was added; transwells were transferred onto 3T3-L1 adipocytes (day 7 differentiation) and incubated for 72 h. Media was collected for cytokine analysis and 3T3-L1 adipocytes were harvested for gene expression analysis. On an identically treated 3T3-L1 adipocytes culture plate insulin-stimulated 3 Hglucose transport assay was conducted.

Statistical analysis
Data are reported as mean 6 SEM. For GTT/ITT studies with multiple time-points an ANOVA was used to test for differences in means between WT and MIF 2/2 groups. When ANOVA was significant post-hoc Bonferroni corrected t-tests were applied. AUC analysis was performed on GTT and ITT curves using Graphpad Prism 5 software (GraphPad Software Inc., San Diego, CA) calculated using the trapezoidal method. For comparison of data between two groups at a single time-point unpaired t-tests were performed. Statistical significance is presented as *p,0.05, **p,0.01 and ***p,0.001 in all figures.

General laboratory techniques
Detailed description of plasma, gene expression, immunohistochemistry and immunoblot analysis are supplied in File S1.

MIF deficiency partially protects mice from HFD-induced obesity and insulin resistance
The severity of HFD-induced IR was less in obese MIF 2/2 mice, with significantly lower GTT and ITT compared to obese WT mice ( Figure 1A-D). Fasting plasma insulin levels increased following the HFD irrespective of genotype, however obese MIF 2/2 mice secreted significantly less insulin in response to glucose compared to obese WT mice ( Figure 1E). Baseline GTT and ITT were not different between WT and MIF 2/2 mice. Agematched chow-fed WT and MIF 2/2 mice GTTs and ITTs were equivalent and significantly lower than obese WT and MIF 2/2 mice (Figure S1A-D). Despite equivalent body weight at baseline and comparable food intake during the intervention, MIF 2/2 mice gained significantly less weight than WT mice in response to HFD ( Figure 1F-G). Body mass composition confirmed the lower body weight was due to reduced fat mass ( Figure 1H), however we could not determine the body regions of fat distribution. Nevertheless, we measured the weight of various organs and observed both liver and epididymal weights were significantly greater in obese WT compared to MIF 2/2 mice (Table 1). Histological analysis confirmed obese MIF 2/2 mice display a hyperplasic morphology ( Figure S2A). Since weight is a key determinant of insulin sensitivity, we sought to distinguish between direct effects of MIF deficiency on insulin sensitivity from secondary effects of reduced weight. Weight-matched (45-47 g) obese MIF 2/2 mice had significantly lower GTT and ITT, compared to equivalently obese WT mice, indicating that improved glucose homeostasis in obese MIF 2/2 mice is independent of body weight ( Figure S2B-E). Fasting plasma triacylglycerol (TAG), NEFA, cholesterol, IL-6 IL-10, IL-12p70 and MCP-1 levels increased equivalently in both genotypes following HFD compared to lean counterparts. Plasma leptin and keratinocyte chemoattractant (KC) levels were lower in HFD MIF 2/2 mice compared WT mice following feeding of a HFD (Table 1).
Adipose tissue MIF expression is elevated in obese WT mice while lack of MIF reduced ATM recruitment following HFD To confirm the positive relationship between obesity and heightened levels of adipose tissue MIF, we profiled the expression of MIF and its corresponding receptors. Obese EAT expressed significantly higher protein MIF levels compared to lean EAT (Figure 2A-B). Intriguingly, obese EAT and visceral adipose tissue have exhibited reduced Mif mRNA expression compared to lean adipose tissue ( Figure 2C). In terms of defining the primary cellular source (adipocyte or SVF) of the enhanced adipose tissue MIF, we demonstrated that Mif mRNA expression was markedly increased in the adipose SVF but not adipocyte fraction of the obese mice compared to lean mice ( Figure 2C). In contrast, Mif expression was not altered in liver in response to HFD ( Figure 2C). Further gene expression analysis revealed expression of known MIF receptors Cxcr2, Cxcr4 and Cd74 was significantly upregulated in the EAT in response to HFD ( Figure 2D). Protein levels of CXCR2, CXCR4 and CD74 could not be detected by immunoblot analysis. This may be attributed to reduced levels of stromal vascular cells in whole adipose tissue sections. Next, we speculated that a reduction in adipose tissue immune cell infiltration would explain the attenuated IR phenotype observed in MIF 2/2 mice. Both SVF and whole adipose tissue F4/80 mRNA expression were decreased in MIF 2/2 mice ( Figure  3SA-B). Concomitant, the recruitment of M1 ATM in response to HFD was markedly reduced in obese MIF 2/2 mice compared to obese WT ( Figure 2E). M2 ATM number decreased in response to the HFD, irrespective of genotype ( Figure 2E). Correspondingly, M2 ATM marker Cd206 was equivalently expressed in adipose of WT and MIF 2/2 mice ( Figure S3B). No difference in either CD8 + or CD4 + T-cell number was evident between genotypes ( Figure 2F). The inflammatory signature of SVF displayed a reduction in TNFa ( Figure 2G) and IL-1b ( Figure S3c) secretion compared to obese WT. Conversely IL-10 ( Figure 2H) and MCP-1 secretion ( Figure S3D) was increased in obese MIF 2/2 SVF; while IL-6 secretion was equivalent between genotypes ( Figure  S3E). No difference in cytokine secretion was evident between genotypes from lean or obese adipocytes.

MIF deficiency attenuates adipose tissue inflammation and improves insulin sensitivity ex vivo
Adipose tissue inflammation was assessed in explants from lean and obese WT and MIF 2/2 mice TNFa and IL-1b secretion was significantly lower from obese MIF 2/2 explants compared to obese WT explants ( Figure 3A-B), with reduced Tnfa and Il-1b mRNA expression in obese MIF 2/2 adipose tissue ( Figure 3C-D). Furthermore, phosphorylated JNK, but not NFkBp65, p38 and ERK, was significantly reduced in obese MIF 2/2 versus WT adipose tissue ( Figure 3E; Figure S4A-C). Subsequently, we investigated if reduced adipose tissue inflammation translated to improved adipose tissue insulin sensitivity in MIF 2/2 mice. Insulin-stimulated 3 H-glucose uptake into adipose explants was reduced in WT obese adipose compared to MIF 2/2 obese adipose ( Figure 3F). Coincident with HFD, Glut-4 mRNA expression decreased in both genotypes; however Glut-4 mRNA expression remained higher in obese MIF 2/2 mice compared to obese WT ( Figure 3G). A comparable reduction in adipose Irs-1 mRNA expression was observed in obese WT and MIF 2/2 mice following HFD ( Figure S4D). To investigate the effect of MIF on insulin signaling lean and obese WT and MIF 2/2 animals were injected with or without insulin. Obese MIF 2/2 adipose tissue displayed markedly increased levels of phosphorylated AKT compared to corresponding obese WT adipose tissue ( Figure 3H). MIF 2/2 BMM have improved adipocyte-macrophage crosstalk, while ISO-1 ameliorates the adverse effects of MIF on adipocyte insulin sensitivity Given the improved immunophenotype observed in obese MIF 2/2 SVF we hypothesized that MIF 2/2 macrophages would have an attenuated pro-inflammatory phenotype and improve cross-talk with adipocytes in vitro. We determined the effect of MIF 2/2 BMM on adipocyte biology as a surrogate for ATM; as the number of ATM were limited for mechanistic studies. MIF 2/2 BMM secreted significantly less IL-6, IL-1b and MCP-1 and exhibited reduced Il-6 and iNos mRNA expression compared to WT (Figure 4A-D; Figure S3E). While anti-inflammatory IL-10 demonstrates a marked increase in MIF 2/2 BMM compared to WT ( Figure S3F). Furthermore, activation of MAPKs; p38, JNK, ERK, and NFkB pathways was significantly impaired in MIF 2/2 BMM compared to WT ( Figure 4E). Exogenous rMIF directly impaired insulin-stimulated glucose transport in 3T3-L1-adipocytes ( Figure 5A) which corroborates previous work [33]. Furthermore we demonstrated that 3T3-L1-adipocytes, co-cultured with WT BMM, but not with MIF 2/2 BMM, had reduced adipocyte insulin-stimulated glucose uptake, with decreased Glut-4 and Irs-1 mRNA expression ( Figure 5B-D). Co-culture of adipocytes with WT BMM significantly reduced insulin-stimulated phosphorylation of AKT, an effect which was not observed when co-cultured with MIF 2/2 BMM ( Figure 5E). Also, adipocytes cocultured with MIF 2/2 BMM secreted less TNFa and IL-6 compared to adipocytes co-cultured with WT BMM (Figure 5F-G). Lastly we examined if ISO-1, a MIF inhibitor which targets MIFs D-dopachrome tautomerase enzymatic activity would inhibit MIF inflammatory effects in macrophages and improve macrophage-adipocyte crosstalk. Pre-treatment with ISO-1 blocked MIF-induced TNFa cytokine secretion from J774.2 obese; black squares = MIF2/2 obese). Lean n = 9, obese n = 9. (f) Accumulative weight of WT (white circles) and MIF 2/2 (black circles) mice. (g) Accumulative food intake of WT (white circles) and MIF 2/2 (black circles). (h) Body mass composition (white bars = lean fat mass; black bars = fat fat mass). Lean n = 3-4, obese n = 8-9. Data are mean 6 SEM, *p,0.05, **p,0.01 and ***p,0.001 w.r.t obese WT. doi:10.1371/journal.pone.0113369.g001   macrophages ( Figure S5A). Subsequently, 3T3-L1 adipocytes were incubated with MIF treated CM in the presence or absence of ISO-1. CM from unstimulated macrophages had no effect on insulin-stimulated glucose uptake into adipocytes while pretreatment with ISO-1 reversed the insulin desensitizing effects of MIF in 3T3-L1 adipocytes ( Figure 5H).

Short-term ISO-1 treatment does not impede HFDinduced IR or alter immune cell recruitment
We next assessed if short-term treatment with ISO-1 would mirror our results observed in MIF 2/2 mice. As obesity is a low grade chronic inflammatory state we first speculated a low dose of 10 mg/kg over 14 consecutive days would have therapeutic benefits [34,35]. ISO-1 treated animals exhibited similar glucose tolerance to vehicle control treated animals ( Figure S5B). As such a higher dose was proposed. A subset of WT mice were treated with  35 mg/kg ISO-1 daily for 3 days. Insulin sensitivity was comparable between ISO-1 treated and saline treated groups ( Figure S5C). ISO-1 treated and saline treated animal's recruited equivalent numbers of M1 and M2 ATM following a HFD ( Figure  S5D). Overnight cultured SVF harvested from ISO-1 treated and vehicle control groups secreted equivalent levels of TNFa, and IL-1b ( Figure S5E-F).

MIF deficiency alleviates hepatic steatosis and improves hepatic insulin sensitivity in response to HFD
Adipose tissue expansion exacerbates lipolysis and elevates NEFA influx into liver. MIF 2/2 mice had reduced liver weight compared to WT mice following HFD indicating partial protection from HFD-induced hepatomegaly ( Figure 6A) Furthermore, fasting plasma alanine aminotransaminase (ALT) levels were considerably lower in obese MIF 2/2 mice compared to obese WT, indicative of reduced liver tissue injury ( Figure 6B). Hepatic TAG content was significantly lower in obese MIF 2/2 mice compared to obese WT mice ( Figure 6C). Hematoxylin and eosin (H&E) studies confirmed that obese MIF 2/2 mice have reduced lipid accumulation compared to obese WT after HFD ( Figure 6D). Also lipogenic genes Cd36, Dgat-1, Fasn, Srebp-1c, Pgc-1a, Lpl and Pparc mRNA expression was significantly lower in obese MIF 2/2 mice compared to obese WT ( Figure 6E). Lastly, we investigated the effect of MIF on hepatic insulin signaling and inflammation. Insulin-stimulated hepatic phosphorylated AKT expression was significantly greater in both lean and obese MIF 2/2 mice compared to WT controls ( Figure 6F). Furthermore phosphorylated NFkBp65 levels were markedly lower in obese MIF 2/2 mice compared to obese WT mice ( Figure 6G)

Discussion
Expanding adipose tissue mass in response to HFD is accompanied by a progressive infiltration of macrophages and T-cells which contribute significantly to inflammation and IR [4]. The mediators which initiate ATM recruitment remain to be fully deciphered. Here we have shown that HFD augments local adipose tissue MIF expression, which is primarily attributed to the adipose SVF. We demonstrate that MIF is pivotal for HFDinduced ATM recruitment and adipose tissue inflammation. Impeding MIF with ISO-1 attenuates MIFs pro-inflammatory, insulin desensitizing effects in a macrophage-adipocyte conditioned media system. However the beneficial effects of ISO-1 were not evident in vivo. Furthermore MIF deficiency partially protects from HFD-induced obesity, IR and hepatic steatosis.
MIF affects glucose metabolism at several levels, including pancreatic insulin secretion [36], TNFa mediated adipocyte IR and glucose uptake and glycolysis in skeletal muscle [37]. Recent reports present conflicting data regarding its role in glucose homeostasis in vivo. Deletion of MIF in LDLR 2/2 mice impeded the development of atherosclerosis, glucose intolerance and IR without altering adiposity [32]. We extend this knowledge by demonstrating that obese MIF 2/2 mice are partially protected from HFD-induced IR, are less hyperinsulinemic in response to a glucose load and exhibit improved glucose homeostasis. Increased levels of MIF can enhance insulin secretion in a glucose dependent manner [36] and this may in turn explain the reduced hyperinsulinemia observed in obese MIF 2/2 mice. Consistent with Verschuren et al., findings MIF 2/2 mice display adipocyte hyperplasia. However unlike Verschuren et al we illustrate MIF 2/2 mice exhibit reduced weight gain attributable to lower fat mass, without altered feeding behavior in response to HFD. Plasma insulin levels in chow fed animals are equivalent, suggesting that absence of MIF does not impair insulin production/secretion but reduction may be reflective of reduced weight gain and improved insulin sensitivity in HFD-fed MIF 2/2 mice. Improved glucose tolerance was still evident in obese MIF 2/2 mice compared to weight-matched obese WT, thus we speculate that MIF has additional beneficial mechanisms beyond reduced weight-gain. Conversely, Serre-Beiner et al. showed that absence of MIF led to impaired glucose tolerance, hyperinsulinemia and increased body fat mass in mature mice compared to WT mice in response to chow diet [38], however interactions with HFD were not examined. These contradictory findings underscore the complexity of MIF inflammatory signals within glucose and energy metabolism depending on diet and age. Further, increases in plasma leptin levels observed in the obese state are usually directly proportional to increases in plasma insulin levels and expanding adipose tissue mass [39]. It is therefore likely that the reduced adipose tissue mass in MIF 2/2 mice accounts for differences in plasma leptin levels.
ATM infiltration and inflammation are crucial to the pathogenesis of HFD-induced IR. Ablation of pro-inflammatory CD11c + (M1) macrophages ameliorates IR coincident with diminished systemic inflammation in obese mice [40]. Furthermore mice lacking PPARc in myeloid cells exhibit reduced numbers of M2 macrophages and are susceptible to HFD-induced IR [41]. We postulated that given the potent chemotactic properties of MIF [31], reduced ATM accumulation would explain the partial protection from HFD-induced IR observed in MIF 2/2 mice. In support of our hypothesis, M1 ATM number was lower in obese MIF 2/2 adipose tissue compared to WT coincident with reduced adipose F4/80, while M2 ATM number was equivalent. Our study demonstrated reduced circulating levels of keratinocytederived chemokine (KC), a mouse ortholog for human IL-8, in MIF 2/2 mice after HFD compared to WT mice. KC signals via a known MIF receptor CXCR2 which is involved in chemotaxis. It is plausible that the combination of reduced KC and lack of MIF contributed to reduced inflammatory cell recruitment into MIF 2/2 adipose tissue and preservation of insulin sensitivity. Furthermore, a previous study by Lin et al., demonstrated that administration of exogenous MIF increased circulating KC in mice indicative of a direct capacity of MIF to regulate KC levels [42].
Obese MIF 2/2 SVF secreted less TNFa and IL-1b compared to WT SVF. IL-10 secretion was increased from MIF 2/2 SVF compared to WT SVF. In contrast lack of MIF did not alter adipose T-cell CD8 + or CD4 + populations as illustrated in Figure 2F. Recent studies have implicated the immunogenic phenotype of ATM is of critical importance in dictating the severity of adipose tissue inflammation and IR [43]. We speculated that the improved inflammatory profile was macrophage derived, however given that we are limited to interpreting the specific cellular source of cytokines while working within the mixed cell pool of the SVF, we investigated the response of a more purified N = 4/group, data are mean 6 SEM, *p,0.05, **p,0.01 and ***p,0.001 w.r.t adipocytes co-cultured with WT BMM. (h) The effect of unstimulated J774.2 CM, MIF (100 ng/ml)-stimulated (CM+M), ISO-1 (50 mMl) treated (CM+I) and cells pretreated with ISO-1 (50 mM) 1 hour prior MIF-stimulation (CM+M+I) on insulin (100 nM)-stimulated 3 H-glucose transport into 3T3-L1adipocytes was evaluated and expressed in DPM. N = 4/group, data are mean 6 SEM, *p,0.05, **p,0.01 and ***p,0.001 w.r.t adipocytes treated with (CM+M), #p,0.05, ##p,0.01 and ###p,0.001 w.r.t control. doi:10.1371/journal.pone.0113369.g005 macrophage population from WT and MIF 2/2 mice to an LPS stimulus by utilizing BMM. Furthermore, MIF 2/2 macrophages have previously displayed an attenuated inflammatory phenotype in response to the TLR4 ligand LPS in vitro [44,45].
Correspondingly MIF 2/2 BMM demonstrated improved immunogenic phenotype with reduced expression of M1 marker iNOS and heightened IL-10 secretion. Furthermore co-culture of MIF 2/2 BMM with adipocytes resulted in reduced inflammation and preservation of adipocyte insulin sensitivity compared to coculture with WT BMM. This data highlights the significance of MIF in the pathophysiology of HFD-induced IR not only as controller of M1 ATM macrophage recruitment but also as a modulator of macrophage activation status and subsequent insulin de-sensitizing capacity. Recently, Saksida and co-workers showed inhibition of MIF in vitro utilizing MIF inhibitor ISO-1 reduced palmitic acid-induced pancreatic beta cell dysfunction [46]. We also have highlighted the potential therapeutic potential of MIF inhibitors by demonstrating that ISO-1 successfully blocked MIFinduced macrophage inflammation and reversed the detrimental effects of MIF-stimulated macrophages on adipocyte function in vitro. On the contrary, short term treatment with ISO-1 failed to attenuate HFD-induced IR and or impede immune cell infiltration in vivo. It is probable that the timing and dosing of the ISO-1 inhibitor accounted for lack of efficacy in the obese phenotype. An alternative MIF antibody with greater specificity may provide greater efficacy.
Several models of inflammation have indicated MIF is a central controller of systemic inflammation [21,47,48]. This study extends our current understanding of MIF as an inducer of local inflammation within the context of HFD-induced adipocyte dysfunction. We hypothesized that reduction of M1 ATM number and diminished immunogenic phenotype of adipose tissue SVF in obese MIF 2/2 mice may attenuate whole adipose tissue inflammation and further dissipate adipocyte dysfunction. Obese MIF 2/2 whole adipose tissue cultured ex vivo exhibited a marked reduction in TNFa and IL-1b secretion accompanied by reduced Tnfa and Il-1b mRNA expression compared with obese WT. Moreover JNK activation was impaired in obese MIF 2/2 adipose tissue, while no difference in pNFkBp65 or pp38 activity was observed. Attenuated TNFa and IL-1b secretion may account for the observed increase in JNK activity. Interestingly, deletion of JNK-1 in obese mice reduced adiposity and alleviated IR [49], we could therefore speculate that diminished JNK activation in obese MIF 2/2 adipose tissue may contribute to reduced adiposity. Concomitant with attenuated local adipose tissue inflammation and corroborating our hypothesis we observed improved insulin sensitivity in obese MIF 2/2 adipose tissue, whereby in insulinstimulated 3 H-glucose uptake into adipose explants was increased in obese MIF 2/2 compared to WT mice.
The contribution of MIF to HFD-induced hepatic steatosis has been relatively unexplored. We demonstrate that obese MIF 2/2 mice have reduced hepatomegaly, lower systemic ALT levels and are partially protected from HFD-induced hepatic steatosis; coincident with reduced Pparc and Srebp-1c mRNA expression compared to WT. Pparc regulates expression of the fatty acid translocase protein Cd36, expression of which is also reduced in obese MIF 2/2 mice compared to WT. Previous studies have reported that lean mice lacking MIF have improved hepatic insulin sensitivity [38]. In addition, we demonstrated that obese MIF 2/2 liver tissue has heightened insulin sensitivity compared to obese WT. Overflow of NEFA from obese, insulin resistant adipose tissue to liver contributes to hepatic steatosis and resultant IR, an effect which is markedly improved upon disruption of ATM recruitment [50,51] [12]. Whether MIF directly induces hepatic lipid accumulation or indirectly via ATM recruitment is yet to be established.
In conclusion, this study illustrates a direct role for MIF signaling in HFD-induced ATM recruitment, adipose dysfunction and glucose homeostasis. However, there were limitations to our study. These studies were performed in whole-body MIF deficient mice, whether lack of MIF within the immune or non-immune system is primarily governing these protective effects remains to be addressed. Furthermore the significant reduction in body weight warrants further investigation to decipher whether this is an intrinsic characteristic of this mouse model or if MIF regulates energy homeostasis and/or gut assimilation. Nevertheless, given the body of evidence presented, we speculate that small molecule MIF inhibitors may have therapeutic potential to ameliorate obesity-induced IR, halting the progression to T2D and nonalcoholic fatty liver disease.