Recruited adipose tissue macrophages contribute to chronic and low-grade inflammation causing insulin resistance in obesity. Similarly, we hypothesized here that Kupffer cells, the hepatic resident macrophages, play a pathogenic role in hepatic insulin resistance induced by a high-fat diet. Mice were fed a normal diet or high-fat diet for 3 days. Kupffer cell activation was evaluated by immunohistochemistry and quantitative RT-PCR. Insulin sensitivity was assessed in vivo by hyperinsulinemic-euglycemic clamp and insulin-activated signaling was investigated by Western blot. Liposome-encapsulated clodronate was injected intravenously to deplete macrophages prior to a short-term exposure to high-fat diet. Here, we characterized a short-term high-fat diet model in mice and demonstrated early hepatic insulin resistance and steatosis concurrent with Kupffer cell activation. We demonstrated that selective Kupffer cell depletion obtained by intravenous clodronate, without affecting adipose tissue macrophages, was sufficient to enhance insulin-dependent insulin signaling and significantly improve hepatic insulin sensitivity in vivo in this short-term high-fat diet model. Our study clearly shows that hepatic macrophage response participates to the onset of high-fat diet-induced hepatic insulin resistance and may therefore represent an attractive target for prevention and treatment of diet- and obesity-induced insulin resistance.
- high-fat diet
nonalcoholic fatty liver disease (NAFLD) becomes in our sedentary societies the most common cause of chronic liver disease (1). It is associated with obesity and insulin resistance and is now considered as the hepatic manifestation of the metabolic syndrome (50, 51). Fat accumulation in the liver is associated with higher cardiovascular mortality, in addition to liver-related cause of death, such as progression to nonalcoholic steatohepatitis and fibrosis or sensitization to other hepatic injuries (1, 49). Hepatic insulin resistance refers to inadequate insulin-dependent suppression of hepatic glucose production (HGP) and plays a pivotal role in the development of hyperglycemia and evolution toward type 2 diabetes (17, 30). Hepatic insulin resistance is the hallmark of NAFLD (31, 46). Hepatic lipid overload, lipid metabolites, and proinflammatory mediators have all been shown to alter hepatic insulin sensitivity in vivo and in vitro (38, 47). Knowing the role of diabetes in the progression of hepatic inflammation and fibrosis (7, 14, 18) and the potential benefits of improving insulin resistance in decreasing risk of cardiovascular related death, targeting hepatic insulin resistance may represent a new therapeutic approach for NAFLD and related diseases.
Several studies have linked insulin resistance to systemic inflammation (11, 45). Anti-inflammatory aspirin is long known to lower blood glucose (13). Mice lacking the proinflammatory cytokine TNF are protected from insulin resistance induced by chronic high-fat feeding (10). Conversely, activation of proinflammatory JNK or IκB kinase-β pathways have been found to promote obesity-induced insulin resistance (6, 12). Fat-overloaded adipocyte, because it produces numerous chemokines and (adipo)cytokines affecting immunity and inflammation, was seen as an important player in this process. More recently, attention has been brought to bone marrow-derived macrophages recruited to obese visceral adipose tissue (28, 56, 59) as a source for inflammatory factors altering metabolic functions of adipocytes and causing insulin resistance both locally and in distant organs. Accordingly, F4/80+CD11b+CD11c+ M1 adipose tissue macrophages (ATM) increase in number in obese adipose tissue (28, 34) and suppression of such inflammation improves insulin sensitivity and dissociates adiposity from insulin resistance (15, 37, 42, 48, 55). Genetic manipulations of IκB kinase-β or JNK signaling in myeloid lineage affect systemic metabolic regulations (48, 58, 61). Collectively, data available to date show that, without a myeloid cell-driven inflammatory component, obesity itself does not cause insulin resistance.
Recent data suggest that hepatic steatosis resulting from high-fat feeding is associated with activation of Kupffer cells (KC), the hepatic resident macrophages (2, 57). A causal link between KC activation and hepatic insulin resistance has been suggested (33, 44) but never convincingly experimentally addressed.
The aim of our study is to characterize hepatic inflammation induced by high-fat feeding and determine whether such inflammation participate to the initiation of hepatic insulin resistance. We used a short-term high-fat diet (HFD) model in mice and demonstrated early hepatic macrophage activation associated with hepatic insulin resistance and steatosis. We addressed the role of activated resident macrophages in the liver (KC) on hepatic insulin sensitivity using experimental macrophage depletion obtained by intravenous injections of liposome-encapsulated clodronate (53, 60).
MATERIALS AND METHODS
Animals, diets, and treatment.
Five-week-old male C57BL/6J (Janvier, Le Genest-Saint-Isle, France) mice were fed for 3 days a HFD in which 60% of calories are derived from fat (D12492 from Research Diets, New Brunswick) or a standard rodent chow (10% of calories from fat, Carfil Quality, Oud-Turnhout, Belgium) referred to as normal diet (ND). The animals were handled according to the guidelines for humane care for laboratory animals established by the Université catholique de Louvain in accordance with European Union Regulation. The study protocol was approved by the local ethics committee. At the time of euthanasia, under anesthesia, mice received 5 U insulin (Actrapid, Novo Nordisk, Bagsvaerd, Denmark) or an equal volume of PBS into the portal vein to analyze the intrahepatic response to insulin. The liver was removed 1 min after the injection, and epididymal and subcutaneous adipose tissues and right quadriceps were dissected and weighed. Portions of tissue were immersed in formalin 4% or frozen in optimal cutting temperature medium for histological studies; the remaining tissue was snap frozen in liquid nitrogen and kept at −80°C until analyses. To deplete macrophages, 100 μl/10 g body wt of 1 mg/ml liposome-encapsulated clodronate, or liposome-encapsulated PBS, or an equal volume of NaCl 0.9% were injected to mice intravenously (iv) via the retroorbital plexus, 24 h prior to the administration of the HFD. In a parallel experiment, mice received NaCl 0.9%, liposome-encapsulated PBS, or liposome-encapsulated clodronate by intraperitoneal injection. Clodronate (a gift from Roche Diagnostics, Mannheim, Germany) was encapsulated into liposomes as described previously (53). Clodronate and liposomes by themselves are not toxic. Liposomes, taken up by phagocytic cells (macrophages), are disrupted by lysosomal enzymes. This provokes the intracellular release of clodronate, causing apoptotic cell death of macrophage (54). Analyses were performed on eight animals per group.
Hyperinsulinemic-euglycemic clamp study.
Femoral iv catheter was implanted 7 days before clamp studies (4). HGP and whole body glucose turnover (TO) were evaluated by the hyperinsulinemic-euglycemic clamp technique (8, 16), conducted at 11:00 AM after a 5-h fast with a continuous infusion of 2.5 mU·kg−1·min−1 human insulin (Actrapid) coupled with [3-3H]glucose (0.33 μCi, PerkinElmer, Boston, MA). Tail blood glycemia was obtained at time 0 and every 10 min thereafter to adjust a variable 20% glucose infusion rate (GIR) to maintain euglycemia (90–110 mg/dl). When steady-state was obtained (±90 min), blood samples (5 μl) were collected every 10 min for 1 h for estimation of plasma glucose specific activity. At steady state, the rate of glucose appearance (Ra) measured as = [3-3H]GIR/glucose specific activity equals the rate of peripheral glucose disposal (Rd) or TO (4, 5). In basal conditions, the rate of HGP (mg·kg−1·min−1) equals Ra. In insulin stimulated conditions (2,5 mU·kg−1·min−1), HGP was obtained by subtracting the GIR (mg·kg−1·min−1) to the Rd previously calculated: HGP (Ra) = Rd − GIR. Analyses were performed on five to eight animals per group.
Blood glucose concentrations were determined with a glucose monitor (Accu-chek, Aviva, Roche) and plasma insulin levels by using an enzyme-linked immunoassay (Mercodia, Uppsala, Sweden). Total liver lipids were extracted with methanol and chloroform and quantified by the vanillin-phosphoric acid reaction (22).
Histology, immunohistochemistry, and immunofluorescence.
Hematoxylin and eosin staining were used for routine histological evaluation. Immunohistochemical detection of F4/80 was performed on paraffin-embedded sections treated with proteinase K by using a primary rat anti-mouse F4/80 monoclonal Ab (1/200, Serotec, Oxford, UK), a rabbit anti-rat immunoglobulin (1/200, Dako, Glostrup, Denmark), and then a goat anti-rabbit streptavidin horseradish peroxidase-conjugated Ab (En Vision, Dako). The peroxidase activity was revealed with diaminobenzidine and slides were counterstained with hematoxylin. Morphometrical quantification of the surface stained was performed on microphotographs (×20) taken from two random sections per liver (n = 6 per group) by use of the Axiovision 4.6 image analysis software (Carl Zeiss, Goentingen, Germany) and expressed as percentage of total area as previously described (52). CD68, C11b, and CD11c-positive cells were detected on cryosections (5 μm thick) by using a monoclonal rat anti-mouse CD68 (1/300, Serotec), a monoclonal rat anti-mouse CD11b (1/100, Developmental Studies Hybridoma Bank, Iowa City, IA) or a monoclonal hamster anti-mouse CD11c Ab (1/50, Serotec). For double immunofluorescence, the same primary Abs were used as well as a polyclonal F4/80 rabbit anti-mouse Ab (1/500, Santa Cruz, CA) when indicated. Secondary Abs were anti-rat/AlexaFluor 594, anti-rabbit/AlexaFluor 488, or anti-hamster/AlexaFluor 488 (dilution 1/1,000, Invitrogen, Merelbeke, Belgium). Hoechst (dilution 1/10.000) was used to reveal the nuclei.
Protein studies: Western blotting.
Liver and muscle homogenates were prepared as described elsewhere (23). Proteins (100 μg) were assayed by Western blotting using the Abs and conditions listed in Supplemental Table S1. The immunoreactivity was detected with a horseradish peroxidase-conjugated secondary Ab (1/10,000) and chemiluminescence (PerkinElmer). One membrane was sequentially probed with the Ab against the total protein, the phosphoprotein, and HSP90 to control for protein loading. The quantification of immune-reactive proteins was obtained by densitometry using the Gel DocTM XR System 170–8,170 device and software (Bio-Rad). The levels of immunoreactivity relative to the invariant control are reported as arbitrary densitometry units.
RNA extraction, reverse transcription, and quantitative PCR.
Total RNA was extracted from frozen liver, subcutaneous or epididymal fat samples by using TRIpure Isolation Reagent (Roche Diagnostics Belgium, Vilvoorde, Belgium). cDNA synthesis and real-time PCR analysis were carried out as previously described (23). Primer pairs for transcripts of interest were designed using the Primer Express software (Applied Biosystems, Lennik, Belgium) and listed in Supplemental Table S2. RPL19 mRNA was chosen as an invariant standard. All experimental tissues and standard curve samples were run in duplicate in a 96-well reaction plate (MicroAmp Optical, Applied Biosystems). Results are expressed as fold expression relative to expression in the control group following the ΔΔCt method (26).
All the data are presented as means ± SD. Statistical analysis was performed by ANOVA for multiple groups comparison and Student's t-test for comparing two groups. When P was less than 0.05, differences were considered significant.
Short-term high-fat feeding induces obesity, adiposity, and liver steatosis.
Compared with ND, a 3-day course of HFD induced body weight gain, increased adiposity, and increased relative liver weight (Table 1). On histological examination, lipid vacuoles were seen predominantly in the intermediate and periportal zones (Fig. 1B). Steatosis in HFD-fed animals was confirmed by a fourfold increase in hepatic lipid content (Table 1).
Short-term high-fat fed mice are insulin resistant, with the evidence of a specific hepatic insulin resistance.
HFD-fed animals had increased fasting glucose levels with normal insulinemia (Table 1). Insulin sensitivity, assessed by the gold standard hyperinsulinemic euglycemic clamp technique, was reduced in the HFD group as demonstrated by a lower GIR required to maintain euglycemia (Fig. 1C). This was owed to a marked impairment of insulin-dependent inhibition of HGP, a parameter completely inhibited in controls. In contrast, whole body glucose uptake (or TO) was not different in HFD and ND (Fig. 1C).
In the liver, HFD decreased the expression of insulin receptor (IR) but did not induce significant changes in Akt protein levels (Fig. 1D). Insulin-stimulated phosphorylation of IR and Akt (on Ser473 and Thr308) was significantly lower in HFD compared with ND livers (Fig. 1D). These changes were not observed in the muscle, where expression of IR, Akt and insulin-stimulated p-IR and p-Akt were similar in HFD and ND groups (Fig. 1E).
Collectively, those data demonstrate a marked hepatic insulin resistance but normal peripheral insulin sensitivity in mice fed the HFD for 3 days.
HFD induces an early activation of hepatic and ATM.
Compared with ND livers (Fig. 2A), enlarged F4/80-positive mature tissue resident macrophages were preferentially found in the vicinity of lipid-loaded hepatocytes in HFD livers (Fig. 2B). This was confirmed by morphometrical analysis (F4/80-stained surface of 2.31 ± 0.85% in HFD livers vs. 0.85 ± 0.21% in controls, P = 0.005) and by the upregulation of F4/80 mRNA (Fig. 2C). An increased number of cells expressing the CD68/macrosialin, a transmembrane protein expressed by activated tissue macrophages (39), was found in the HFD livers compared with controls (Fig. 2, D and E) and confirmed by upregulation of hepatic CD68 mRNA (Fig. 2F). Immunofluorescence double staining revealed that, in control livers, a small proportion of F4/80-positive macrophages also expressed CD68 (Fig. 2G). Under HFD, the proportion of F4/80 to CD68 double-positive cells increased (Fig. 2H), supporting that hepatic macrophages are activated. CD11b/MAC-1 is classically expressed by granulocytes, monocytes, NK cells, and tissue macrophages (24). Two distinct liver populations of CD11b+ cells were detected: elongated and faintly stained cells morphologically and topologically reminiscent of tissue macrophages, and a population of round cells compatible with neutrophils or monocytes (Fig. 2I, arrows and arrowheads, respectively). Under HFD, the population of CD11b-positive cells appeared increased (Fig. 2J). The mRNA analysis showed an approximately fourfold upregulation of CD11b (Fig. 2K). CD11c is an integrin mainly expressed by dendritic cells and by some macrophages (27). CD11c-positive cells were rare and preferentially located around the portal tract (Fig. 2L) in the normal liver. Upon HFD, slightly more numerous CD11c-positive cells, remaining located in the periportal zone (Fig. 2M), were seen together with an approximately twofold increased in CD11c mRNA (Fig. 2N). Double immunostaining demonstrated that only a minority of F4/80-positive macrophages were CD11c positive in HFD livers (Fig. 2P). These were apparently absent in controls (Fig. 2O). In addition to phenotypic changes in hepatic macrophages, HFD also significantly induced TNF, TLR2, TLR4, and CD14 mRNA expression, considered as markers of inflammation (Table 2).
In the adipose tissue, whether epididymal or subcutaneous, no apparent inflammation was detected by (immuno)histochemical techniques (Fig. 3, A and B), contrasting with adipose tissue inflammation classically described in chronic HFD and obesity models (28, 56, 59). Nevertheless, a 3-day HFD increased F4/80, CD68, and CD11b mRNA levels by twofold (Fig. 3, C–E) in the epididymal adipose tissue, whereas CD11c mRNA remained unchanged (Fig. 3F). Those changes were not seen in the subcutaneous fat pads (not shown). Our data thus support the activation of ATM in intra-abdominal depots in our model despite the apparent absence of macrophage recruitment.
Selective KC depletion ameliorates hepatic insulin sensitivity.
The injection of PBS-loaded liposomes to HFD-fed mice had no effect on macrophage populations in the liver (Fig. 4, H–J). By contrast, the injection of liposome-encapsulated clodronate markedly and significantly depleted the liver of its F4/80, CD68, and CD11b-positive macrophages (Fig. 4, A–F). This was also assessed by a blunting in F4/80 and CD68 transcript expression (Fig. 4, H–I) whereas CD11b mRNA levels remained stable (Fig. 4J), reflecting the increased in nonmacrophage round CD11b-positive cells (Fig. 4F, arrowheads), consistent with the recruitment of blood monocyte precursors described by others (60). Hepatic CD11c mRNA remained stable in the clodronate-treated HFD group (not shown). Paralleling macrophage depletion, clodronate injection was associated with a significant decrease in hepatic TNF, TLR4, and CD14 mRNA levels (Table 2).
PCR analysis demonstrated that clodronate liposomes did not affect inflammatory cell activity in the adipose tissue (Fig. 4K). Indeed, such liposomes, being unable to cross vascular barriers, are readily taken up by phagocytic cells in the reticuloendothelial system (53).
Fasting blood glucose, serum insulin levels, hepatic lipid content and body weight gain were not different whether HFD animals received or not PBS or clodronate whereas epididymal fat weight was lower in the clodronate treated group compared with others HFD groups (Table 3).
During clamp experiments, the GIR needed to maintain euglycemia significantly increased and insulin-dependent inhibition of HGP was significantly ameliorated in clodronate-treated animals compared with both PBS-treated and untreated HFD animals (Fig. 5A). At the level of insulin signaling, the specific depletion of hepatic macrophages did not improve IR phosphorylation whereas it significantly increased the phosphorylation of Akt in HFD livers (Fig. 5B).
In this study, we characterized a short-term HFD regimen in mice and used simultaneous ablation of KC through injection of liposome-encapsulated clodronate to demonstrate the role of hepatic macrophages in modulating hepatic insulin sensitivity.
Consistent with previous reports in rats (17, 43), feeding mice a HFD for 3 days provoked a significant body weight gain, steatosis, and selective hepatic insulin resistance, proved in vivo by hyperinsulinemic-euglycemic clamp technique and confirmed by defects in hepatic insulin signaling. Conversely, whole body glucose uptake as well as insulin signaling in skeletal muscle were unaltered, supporting that peripheral tissues remained insulin sensitive. In the liver, steatosis and insulin resistance were associated with local inflammatory changes. In control livers, as previously reported by others (27), few F4/80-positive hepatic macrophages coexpressed CD68 but all were CD11c negative. After 3 days of HFD, F4/80-positive cells appeared enlarged and the proportion of F4/80 and CD68 double positive cells increased together with upregulation of mRNA expression of macrophage-specific markers (TLR4, CD14, and TNF), whereas markers of alternative activation (35) remained unchanged (not shown). This supports the proinflammatory activation of resident hepatic macrophages (6, 57).
In contrast with long-term exposure to high-fat diet in which hepatic steatosis is variable, if not absent (20, 21, 32), short-term high-fat diet yields reproducible steatosis (19, 43). Moreover, whereas KC activation is obvious in our 3-day model as also reported in human nonalcoholic steatohepatitis (25, 29, 36, 40), this is no longer found in the longer term regimen in rodents (9, 56, 59). By contrast, peripheral insulin resistance and recruitment of inflammatory cell in the adipose tissue, the hallmark of chronic obesity (28, 59), failed to be observed in our model. This implies that, in response to high-fat feeding, metabolic and inflammatory alterations occur first in the liver and secondarily in muscles and adipose tissues (peripheral tissues). We demonstrated here that selective ablation of KC significantly improved HFD-induced hepatic insulin resistance and alterations of hepatic insulin signaling, confirming a major role of KC in the initiating mechanism of HFD-induced hepatic insulin resistance, besides or irrespectively of inflammatory changes occurring in the adipose tissue.
Our results contrast with a recent publication (9) showing that, in mice exposed to a HFD for 16 wk, a single intraperitoneal injection of liposome-encapsulated clodronate decreased hepatic insulin sensitivity. Besides depleting the liver of F4/80-positive cells, intraperitoneal clodronate also induced significantly F4/80 mRNA in the adipose tissue (9). Such adipose tissue inflammation might be a confounding factor. We also performed intraperitoneal injections of clodronate in our short-term model. Contrasting with the selective depletion of KC via intravenous injections (liposomes do not cross vascular barriers), intraperitoneal clodronate affected ATM, as also reported by others (3), and upregulated fat MCP-1 and TNF implicated in macrophage recruitment (Supplemental Figure S1). Therefore, the significant upregulation of F4/80 in an already inflamed adipose tissue under chronic high-fat feeding reported by Clementi et al. (9) is consistent with a recruitment of macrophages following the clodronate-induced depletion. The effect of selective KC depletion on hepatic insulin sensitivity in a long-term HFD model remains to be evaluated.
Tissue fat accumulation is a cause of insulin resistance (31, 47). A dissociation between adiposity induced by high-fat feeding and insulin sensitivity, however, occurs when inflammatory pathways are inhibited (42, 48). Similarly, hepatic macrophage depletion prevented hepatic insulin resistance and alteration of hepatic insulin signaling without reducing hepatic steatosis elicited by high-fat feeding. Thus KC depletion also dissociates fat accumulation and hepatic insulin resistance. Therefore, hepatic accumulation of lipids or lipid metabolites known to alter insulin signaling in hepatocyte cultures (41) cannot be the sole cause of insulin resistance in vivo. Our results support that signals from local inflammatory cells acting downstream of IR concur to this effect, whether acting independently or amplifying those produced by the fat-loaded hepatocytes.
Intriguingly, KC depletion decreased adipose tissue weight while having no effect on body weight gain, liver weight, or degree of steatosis. Decreased adipose tissue could be the consequence of amelioration of hepatic inflammation and/or insulin sensitivity. Alternatively, although not associated with decreased inflammatory signals, decreased adiposity may possibly contribute to amelioration of hepatic insulin sensitivity. Such phenomenon might also participate in the amelioration of insulin resistance observed upon gadolinium chloride treatment, which prevented HFD-induced weight gain (33). Further experiments are needed to address those issues.
Results presented here firmly establish that KC activation is a key mechanism in the induction of hepatic insulin resistance that develops upon high-fat diet feeding. Targeting KC to block this deleterious activation might therefore represent an effective strategy to prevent the development of hepatic insulin resistance. Further assessment of the role of KC in chronic obesity-induced metabolic alteration and deciphering the priming for KC activation, which could be triggered either by dietary factors or by products, whether cytokines or fatty acids, released by the gut, the adipose tissue or the fat-loaded hepatocytes, will be of major relevance for the understanding of diet-induced insulin resistance and for the development of new therapeutic approaches for the metabolic syndrome.
This work was supported by Fonds National de la Recherche Scientifique (FNRS) grants (no. 3.4578.07 and F.45.08.08), la Direction de la Recherche Scientifique de la Communauté Française de Belgique (Action de Recherche Concertée grant, 05/10-328), University FSR grants 2007 and 2008 to I. A. Leclercq, grant “liver, virus and diabetes” from the Fondation St Luc to I. A. Leclercq and Y. Horsmans. I. A. Leclercq and P. D. Cani are FNRS research associates.
The authors have no conflict of interest in relation to this work to disclose. The foundation Clodronate Liposomes (N. van Rooijen) is a nonprofit foundation related to the Department of Molecular Cell Biology of the Free University in Amsterdam, The Netherlands.
The authors are grateful to Martine Petit, Valérie Lebrun, and Jorge Abarca-Quinones for expert technical assistance and to Prof. Hubert Piessevaux for help with statistical analysis. The monoclonal CD11b Ab developed by Timothy A. Springer was obtained from the Developmental Studies Hybridoma Bank developed under the auspices of the National Institute of Child Health and Human Development and maintained by the University of Iowa, Department of Biological Sciences, Iowa City, IA 52242.
- Copyright © 2010 the American Physiological Society