Introduction

The development of obesity in western societies is becoming a major health problem [1]. Obesity is a major risk factor for numerous pathologies, including the development of insulin resistance and type 2 diabetes [2]. Insulin resistance is strongly associated with the development of an inflammatory state in adipose tissue and with ectopic fat accumulation in muscles and liver [3]. Inflammatory cytokines, NEFA and toxic lipid metabolites, such as ceramides and diacylglycerol, activate several signalling pathways that could inhibit insulin signalling [4]. Among these, the pathway involving extracellular signal-regulated kinase (ERK), a mitogen-activated protein kinase (MAPK), is deregulated in obesity and could play a major role in insulin resistance. Indeed, the activity of ERK is abnormally increased in human and rodent adipose tissue in diabetic states [5–7]. Diabetogenic factors, including pro-inflammatory cytokines and lipid metabolites, use the ERK pathway to alter insulin signalling [4, 8]. Further immunological studies have demonstrated that MAPK pathways are involved in the production of inflammatory cytokines by immune cells [9], suggesting a potential role of ERK in the development of inflammation linked to obesity and insulin resistance. Moreover, inflammatory cytokines could increase lipolysis through the ERK pathway, which could be involved in the increase in NEFA responsible, at least in part, for the lipotoxicity [10–13].

ERK1 and ERK2 are the two main proteins of the ERK family, encoded by Erk1 (also known as Mapk3) and Erk2 (also known as Mapk1), respectively. They share 75% overall identity at the amino acid level and are activated by the same stimuli [14]. However, unlike Erk1 −/− mice, Erk2 −/− mice are not viable, suggesting that these kinases have non-redundant functions [15–17]. We have previously reported that ERK1 rather than ERK2 was involved in adipocyte differentiation and in adipogenesis in vivo [18, 19]. Indeed, Erk1 −/− mice have reduced fat content and remain lean when exposed to a high-fat diet. The leanness of the mice on a high-fat diet could be explained, at least in part, by the reduced adipogenesis, but these mice also have an increase in their postprandial metabolic rate that could contribute to the observed phenotype [18]. Erk1 −/− mice are also protected against insulin resistance when exposed to a high-fat diet [18]. However, due to the leanness of the mouse, it is not clear whether the invalidation of ERK1 per se is responsible for the improved insulin sensitivity of the Erk1 −/− mice fed a high-fat diet.

To address the importance of the ERK1 pathway in the development of obesity-induced insulin resistance, we investigated the impact of Erk1 deficiency in the context of severe obesity induced by the lack of leptin. Indeed, it is well known that leptin regulates both weight gain and energy expenditure and it could also modulate insulin sensitivity [20]. For this investigation, we intercrossed Erk1 −/− mice and leptin-deficient ob/ob mice and examined the metabolic responses in the resulting animals. In this setting, despite developing as severe obesity as the ob/ob controls, ob/ob–Erk1 −/− mice were partially protected against systemic insulin resistance and hepatic steatosis. This metabolic phenotype could be explained by an increase in glucose uptake by muscles and a decrease in adipose tissue inflammation.

Methods

Generation of ob/ob–Erk1 +/+ and ob/ob–Erk1 −/− mice

Mice deficient in Erk1 (Erk1 −/− mice) were generated from the C57BL/6J genetic background as previously described by Pages et al. [21]. The Erk1 −/− mice were then intercrossed with heterozygote ob/+ mice to generate double heterozygote mice (ob/+–Erk1 +/−). These mice were then intercrossed to generate ob/+–Erk1 −/− and ob/+–Erk1 +/+ mice, which subsequently served as parents to lean and obese (ob/+ and ob/ob, respectively) animals, either wild type (Erk1 +/+ ) or null (Erk1 −/− ) in the Erk1 locus. Mice were exposed to a 12 h light/dark schedule and had free access to water and standard chow diet. Mice were killed by cervical dislocation and epididymal and subcutaneous fat pads, liver and muscles were removed, freeze-clamped in liquid nitrogen and stored at −80°C until used. The Principles of Laboratory Animal Care (NIH publication no. 85–23, revised 1985; http://grants1.nih.gov/grants/olaw/references/phspol.htm) were followed, as well the European Union guidelines on animal laboratory care (http://ec.europa.eu/environment/chemicals/lab_animals/legislation_en.htm). All procedures were approved by the Animal Care Committee of the Faculty of Medicine of the Nice-Sophia Antipolis University, Nice, France.

Biochemical assays

Plasma insulin level was measured using ELISA (Mercodia, Uppsala, Sweden). Quantification of NEFA was performed using a colorimetric diagnostic kit (NEFA-C; Wako Chemicals, Neuss, Germany).

Glucose tolerance test

A glucose tolerance test (GTT) was performed on 12 and 18 week old animals after an overnight fast (~16 h). Glucose (1 g and 0.5 g d-glucose/kg body weight for 12 and 18 week old mice, respectively) was administered by intraperitoneal injection in awake mice. Blood was collected via the tail vein at different time points, and glucose levels were measured using a glucometer (Medisens Optimum XCD; Abbott, Rungis, France).

Hyperinsulinaemic–euglycaemic clamp studies

These experiments were performed on 12-week-old animals as described by Burcelin et al. [22, 23]. Under anaesthesia (fluothane), an indwelling catheter was introduced into the femoral vein of the mice, sealed under the back skin, and glued on the top of the skull. The mice were allowed to recover for 4–6 days, and showed normal body weight and feeding behaviour. The clamp studies were conducted with a continuous infusion of insulin (18 mU kg−1 min−1) and a variable infusion of glucose (15% [wt/vol.]). To determine the insulin-stimulated glucose utilisation in individual tissues, a rapid intravenous injection of 2-deoxy-d-[3H]glucose (1.85 MBq per mouse; PerkinElmer, Boston, MA, USA) was performed through the femoral vein 60 min before the end of the clamp. Plasma 2-deoxy-d-[3H]glucose disappearance and glucose concentration were determined in 5 μl samples of blood from the tip of the tail vein at 0, 5, 10, 15, 20, 25, 30, 45 and 60 min after injection. Different tissues were dissected for biochemical analysis [24].

For glucose turnover measurement, the level of 2-deoxy-d-[3H]glucose was determined from total blood after deproteinisation by a Zn(OH)2 precipitation as previously described by Perrin et al. [24]. Individual tissue glucose uptake measurement was determined as previously described by Kamohara et al. [25].

Liver triacylglycerol and glycogen content

Triacylglycerol extraction was performed on 50 mg of liver homogenised in methanol and chloroform (1:2 [vol./vol.]) over 16 h at 4°C. Then CaCl2 0.05% (1:5 [vol./vol.]) was added and the samples were centrifuged at 2,000×g for 20 min at 4°C. The chloroformic phase was recovered and evaporated and PBS with 5% BSA was added. Triacylglycerol content was determined using a commercial kit (DiaSys, Hozheim, Germany). For histological studies, 10 μm sections were cut from liver samples embedded in paraffin, and were stained with Oil Red O.

To measure glycogen content, 50 mg of liver was homogenised into 1 ml of KOH 0.5 mol/l for 1 h at 60°C. Glycogen was precipitated overnight with 6% Na2SO4 (wt/vol.) and 66% ethanol (vol./vol.). The pellet of glycogen was washed three times with 66% ethanol. The glycogen content of the tissue was then measured by an enzymatic method. Briefly, after digestion of the glycogen with 0.25 mg/ml of amyloglucosidase (Sigma, St Louis, MO, USA) for 2 h at 37°C, the glucose concentration of each sample was measured using the Glucose GOD FS kit (DiaSys, Hozheim, Germany).

Insulin-stimulated phosphorylation of Akt and western blot analysis

Phosphorylation of thymoma viral proto-oncogene (Akt) (Ser473) was determined in muscle and adipose tissue from ob/+, ob / ob and ob/ob–Erk1 −/− mice after an intraperitoneal injection of insulin (1 U/kg, 10 min). Thereafter, muscles and epididymal fat pads were frozen and stored at −80°C for subsequent analysis.

Muscles or fat pads were solubilised and proteins from lysates were separated by SDS-PAGE and transferred to polyvinylidene difluoride membranes as previously described by our group [8]. The membranes were incubated with the indicated antibody, horseradish-peroxidase-coupled anti-species antibodies were added and chemiluminescence was detected using a Fuji film Las-3000 apparatus (Fujifilm Life Science, FSVT Courbevoie, France). The membranes were subsequently reprobed with the indicated antibody as a loading control. Quantification was performed using MultiGauge software (Fujifilm Life Science).

Antibodies

Antibodies against acetyl-CoA carboxylase (ACC), Akt, phospho-Akt (Ser473) and ERK1/2 were purchased from Cell Signaling Technology (Beverly, MA, USA). Horseradish-peroxidase-conjugated secondary antibodies were obtained from Jackson Immunoresearch Laboratories (West Grove, PA, USA).

Real-time RT-PCR

Total RNA samples were prepared using the RNeasy total RNA kit (Qiagen, Courteboeuf, France), treated with DNAse (Applied Biosystems, Austin, TX, USA) and used to synthesise cDNAs using Transcriptor First Strand cDNA Synthesis kit (Roche, France). Real-time quantitative PCR was performed with sequence detection systems (ABI PRISM 7500; Applied Biosystems) and SYBR Green dye as previously described. Levels of mRNA were expressed relative to mouse Rplp0. The relative amount of mRNA between two groups was determined by using the second derivative maximum method. The results were expressed relative to the mean of the group of controls, which was arbitrarily assigned a value of 1. The primers used (list available on request) were designed using Primer Express software (Applied Biosystems) and synthesised by Eurogentec (Seraing, Belgium).

Statistical analysis

All calculations were performed using MINITAB software. Statistical significance between two groups was tested using the Mann–Whitney test. Comparisons among several groups were performed by ANOVA and when the results passed the ANOVA test, Bonferroni’s multiple comparison post test was used to calculate the relevant p values. A p value   < 0.05 was considered significant. All the data are reported as mean ± SEM.

Results

Erk1 deficiency does not affect the development of obesity in ob/ob mice

In the present study, we investigated the consequences of Erk1 invalidation in obese mice by crossing Erk1 −/− mice with ob/ob mice, a model of genetic obesity resulting from leptin deficiency. To determine whether Erk1 invalidation had consequences for adiposity in mice, the growth curves of ob/ob–Erk1 −/−, ob/ob and ob/+ mice were compared. The lack of ERK1 did not modify the weight of the ob / ob mice (Fig. 1a) and no statistically significant differences were observed in the weight of epididymal and subcutaneous adipose tissues between the two genotypes (Fig. 1b). However, a trend towards a decrease in epididymal adipose tissue and an increase in subcutaneous adipose tissue was noticed in ob/ob–Erk1 −/− compared with ob/ob mice (Fig. 1b). Hence, the prevention of adiposity development observed in Erk1-deficient mice on a high-fat diet [18] was lost in the presence of the ob/ob background.

Fig. 1
figure 1

The ob/ob–Erk1 −/− mice have the same body weight and adiposity as ob/ob mice. a Weight curves of male control ob/+, ob/ob and ob/ob–Erk1 −/− mice (n = 8 per genotype). White squares, ob/+; black circles, ob/ob; and white circles, ob/ob–Erk1 −/−. **p < 0.01 ob/+ vs ob/ob or ob/ob–Ek1 −/− mice. b Epididymal and subcutaneous fat pad mass of ob/ob and ob/ob–Erk1 −/− mice (n = 6 per genotype) at 12 weeks of age. Results are expressed as percentage of body weight. Black bars, ob/ob; white bars, ob/ob–Erk1 −/−

Full size image

ob/ob–Erk1 −/− mice exhibit improved glucose tolerance compared with ob/ob mice

To address the role of ERK1 on whole-body glucose metabolism in obese mice, we next determined blood glucose and serum insulin concentrations in ob/+, ob/ob and ob/ob-Erk1 −/− mice. As expected, fed ob/ob mice developed hyperglycaemia with severe hyperinsulinaemia compared with lean ob/+ mice (Fig. 2a, b). In contrast, ob/ob–Erk1 −/− developed only mild hyperglycaemia but remained hyperinsulinaemic (Fig. 2a, b). Consistent with these results, at 12 weeks of age, obese ob/ob mice were markedly glucose intolerant compared with control ob/+ mice (Fig. 2c). Remarkably, ob/ob–Erk1 −/− mice had an improved glucose tolerance compared with ob/ob mice (GTT is shown in Fig. 2c and quantification of the response by integrating the AUC is shown in Fig. 2d).

Fig. 2
figure 2

The ob/ob-Erk1 −/− mice have reduced fed glycaemia and an improved glucose tolerance compared with ob/ob mice. Fed blood glucose concentration a and serum insulin concentration b of control ob/+, ob/ob and ob/ob–Erk1 −/− mice (n = 8–10 mice per genotype) at 12 weeks of age; **p < 0.01 and ***p < 0.001. c GTT (1 g d-glucose/kg body weight) was performed with control ob/+, ob/ob and ob/ob–Erk1 −/− mice (n = 13 per genotype) at 12 weeks of age, after 16 h fasting. White squares, ob/+; black circles, ob/ob; white circles, ob/ob–Erk1 −/−; *p < 0.05. d Integrated area under the glucose disposal curves for ob/ob and ob/ob–Erk1 −/− relative to control ob/+ mice; *p < 0.05 and **p < 0.01. e Fed blood glucose concentration of ob/ob and ob/ob–Erk1 −/− mice (n = 4 per genotype) at 16–18 weeks of age; *p < 0.05. f IGTT (0.5 g d-glucose/kg body weight) was performed with ob/ob and ob/ob–Erk1 −/− mice (n = 4 per genotype) at 16–18 weeks of age, after 16 h fasting; *p < 0.05. Black circles, ob/ob; white circles, ob/ob–Erk1 −/−

Full size image

We then tested whether this improved metabolic phenotype was also found in older mice. Analysis of a subgroup of 16–18-week-old mice revealed that older ob/ob–Erk1 −/− retained a lower fed glycaemia (Fig. 2e) and an improved glucose tolerance (GTT in Fig. 2f) without any significant differences in their body weight compared with ob/ob mice (data not shown).

ob/ob–Erk1 −/− mice exhibit improved whole-body insulin sensitivity and increased insulin action in skeletal muscle

We then performed hyperinsulinaemic–euglycaemic clamp studies in order to quantify whole-body insulin sensitivity and to delineate tissue-specific sites responsible for the improved glucose homeostasis of the ob/ob–Erk1 −/−. The steady-state glucose infusion rate (GIR) was increased by 40% in the ob/ob–Erk1 −/− mice compared with ob/ob mice, demonstrating an improvement in whole-body insulin sensitivity (Fig. 3a). Measurement of glucose utilisation of individual tissues during the clamp revealed that insulin-stimulated glucose disposal rate in vastus lateralis (VL) and the extensor digitorum longus (EDL) skeletal muscles from ob/ob–ERK1 −/− mice was increased by nearly twofold compared with ob/ob mice (Fig. 3b). As insulin-induced Akt activation is critical for glucose transport, we next investigated the activation of this kinase in EDL muscles following intra-peritoneal insulin injection. Consistent with the increased glucose transport, Akt phosphorylation was higher in insulin-stimulated EDL muscles from ob/ob–Erk1 −/− mice compared with ob/ob mice (Fig. 3c). However, the level of phosphorylation remained lower than the level observed in control mice (Fig. 3c).

Fig. 3
figure 3

The ob/ob–Erk1 −/− mice have improved whole-body insulin sensitivity and an increased glucose utilisation in skeletal muscles. a Steady-state GIR during the hyperinsulinaemic–euglycaemic clamp performed with ob/ob and ob/ob–Erk1 −/− mice (n = 7–11 per genotype) at 12 weeks of age after 5 h fasting. b Glucose uptake rate in VL and EDL muscles during the clamp studies under steady state conditions (n = 5–7 per genotype). c Representative western blot analysis of the phosphorylation of Akt on Ser473 (upper blot) and of total Akt (lower blot) in EDL muscles of ob/+, ob/ob–Erk1 −/− and ob/ob mice injected with insulin (1 U/kg, i. p.) for 10 min. *p < 0.05. Black bars, ob/ob; white bars, ob/ob–Erk1 −/−

Full size image

Decreased steatosis in livers of ob/ob–Erk1 −/− mice

Insulin resistance is associated with the accumulation of the ectopic fat in liver that participates in the abnormal regulation of glucose homeostasis. We thus investigated whether the improvement in glucose tolerance and whole-body insulin sensitivity of the ob/ob–Erk1 −/− mice was associated with reduced liver fat content.

Despite the presence of a similar degree of obesity, the liver mass (Fig. 4a) and hepatic triacylglycerol (Fig. 4b) content were reduced by 20% and 35% respectively in ob/ob–Erk1 −/− mice compared with ob/ob mice. Histological examination of liver sections revealed a substantial reduction in the size and number of the lipid droplets in the liver of the ob/ob–Erk1 −/− mice compared with the ob/ob mice (Fig. 4c). The decrease in triacylglycerol content could be due to reduced de novo lipogenesis because the protein level of ACC, a rate-limiting enzyme in de novo lipogenesis, was reduced by 40% in the liver of the ob/ob–Erk1 −/− (Fig. 4d).

Fig. 4
figure 4

Reduced hepatic steatosis in the livers of ob/ob–Erk1 −/− mice. a Liver weights of ob/+, ob/ob and ob/ob–Erk1 −/− mice (n = 6 per genotype) at 12 weeks of age. Results are expressed as percentage of body weight. b Total liver triacylglycerol content was determined for ob/+, ob/ob and ob/ob–Erk1 −/− mice (n = 7) at 12 weeks of age. c Oil red O staining of liver sections of ob/+, ob/ob and ob/ob–Erk1 −/− fed mice at 12 weeks of age. d Protein lysates from liver were subjected to western blotting with antibodies against ACC1 and ACC2 and against total ERK1/2. Representative immunoblots and quantifications are shown. Data are expressed as percentage of ACC1+ACC2 production in the livers of ob/+ mice and are the mean ± SEM of 12–15 mice per genotype. e Liver glycogen content in ob/ob and ob/ob–Erk1 −/− mice (n = 7–10) at the end of the hyperinsulinaemic–euglycaemic clamp. *p < 0.05 and **p < 0.01

Full size image

We then investigated whether the reduced amount of hepatic triacylglycerol improved insulin action in the liver of the ob/ob–Erk1 −/− mice. We measured the amount of glycogen in the liver of ob/ob and ob/ob–Erk1 −/− mice at the end of the 3 h clamp study and interpreted the result as an indication of insulin action. The glycogen content was twofold higher in the liver of ob/ob–Erk1 −/− mice compared with ob/ob mice at the end of the clamp study (Fig. 4e), suggesting a better insulin response.

Decreased obesity-associated inflammation of adipose tissue of the ob/ob–Erk1 −/− mice

Inflammation of adipose tissue as well as an increase in fatty acid release by the tissue has been causally linked to the development of hepatic steatosis and muscle insulin resistance. We thus investigated whether the lack of ERK1 improved the function of the adipose tissue of the ob/ob mice. Compared with ob/ob mice, the mRNA levels of several inflammatory markers, such as chemokine (C-C motif) ligand 2 (CCL2), serine protease inhibitor 1 (SERPIN-1), IL1β and TNF, were decreased in epididymal adipose tissue of ob/ob–Erk1 −/− mice (Fig. 5a). In contrast, Il6 and Il10 mRNA levels were not statistically modified (Fig. 5a). We have recently reported that inflammatory cytokines positively regulate the production of the inflammatory MAP3 kinase tumour progression locus 2 (TPL2; also known as mitogen-activated protein kinase kinase kinase 8 [MAP3K8]) and that its level is upregulated in adipose tissue of obese mice and patients [10]. In agreement with reduced inflammatory cytokine level, we observed a decrease in the Map3k8 mRNA level in adipose tissue of ob/ob–Erk1 −/− mice (Fig. 5a). The expression of Emr1 (previously known as F4/80), which encodes the constitutive macrophage marker EGF-like module containing mucin-like, hormone receptor like sequence 1, was not statistically modified, suggesting that the number of macrophages infiltrating the tissue was not markedly modified by Erk1 invalidation (Fig. 5a).

Fig. 5
figure 5

The ob/ob–Erk1 −/− mice have a reduced inflammatory profile in epididymal adipose tissue and an improvement in adipose tissue insulin sensitivity. Relative expression of genes encoding proteins involved in inflammation a or T lymphocyte markers b in epididymal fat pads from ob/ob and ob/ob–Erk1 −/− mice (n = 5–6 per genotype). c Representative western blot analysis of the phosphorylation of Akt on Ser473 and total Akt in epididymal fat pads of ob/+, ob/ob–Erk1 −/− and ob/ob mice injected with insulin (1 U/kg, i.p.) for 10 min. d, e Adipose tissue insulin sensitivity was assessed by measuring the decrease in plasma NEFA concentration during the clamp study. Results show the circulating NEFA concentration in ob/ob and ob/ob–Erk1 −/− mice (n = 6–8 per genotype at 12 weeks of age) following a 6 h fasting period (d) and at the end of the hyperinsulinaemic–euglycaemic clamp e. *p < 0.05 and **p < 0.01. Black bars, ob/ob mice; white bars, ob/ob–Erk1 −/− mice

Full size image

As not only macrophages but also T lymphocytes contribute to the development of inflammation in the obese adipose tissue [26], we assessed the mRNA expression of different markers of T lymphocyte populations. The expression of the Cd3e mRNA, which codes for the T lymphocyte marker CD3, was reduced suggesting a decrease in T lymphocyte infiltration in the adipose tissue of ob/ob–Erk1 −/− mice (Fig. 5b). Recent studies demonstrated that CD8+ and CD4+ T helper 1 (T[h]1) cells dominate in adipose tissue of obese mice [27, 28]. Interestingly, we observed a decrease in Cd4, Ifng (coding IFNγ a TH1 cytokine) and Tbx21 (coding for T-box 21, a marker of T[h]1 cells) mRNA levels in the adipose tissue of ob/ob–Erk1 −/− mice. The mRNA expression of Cd8 was also decreased, but this reduction was not statistically significant (Fig. 5b).

We then studied whether this change in the inflammatory profile of adipose tissue of ob/ob–Erk1 −/− was associated with an improvement in insulin signalling and metabolic action. The Akt phosphorylation was higher in insulin-stimulated adipose tissue from ob/ob-Erk1 −/− mice compared with ob/ob mice (Fig. 5c). However, the level of phosphorylation remained lower than the level observed in control mice (Fig. 5c). The effect of insulin in adipose tissue was assessed by measuring circulating NEFA levels at the end of the clamp following a 6 h fasting period. After this fasting period, the plasma NEFA levels were not significantly different between ob/ob–Erk1 −/− and ob/ob mice (Fig. 5d). Importantly, after insulin infusion for 3 h, plasma NEFA levels were significantly lower in ob/ob–Erk1 −/− mice compared with ob/ob mice (Fig. 5e), suggesting an amelioration of the anti-lipolytic effect of insulin in adipose tissue.

Discussion

In this study, we demonstrate that although ob/ob mice lacking ERK1 develop severe obesity, they are partially protected against insulin resistance. The improved glucose homeostasis observed in the ob/ob–Erk1 −/− mice is associated with increased glucose transport in muscles, reduced liver fat content and a better ability of insulin to suppress NEFA release by adipose tissue. These metabolic improvements were linked to reduced production of inflammatory cytokines and T lymphocyte markers in the adipose tissue.

Our finding that invalidation of Erk1 in the C57Bl6/J ob/ob mice did not prevent the development of obesity differs from the observation made of lean mice on a C57Bl6/J background. Indeed, deletion of Erk1 in lean mice protected them against high-fat-diet-induced obesity [18]. This protection is due to a relative impairment in adipogenesis and to an increase in postprandial energy expenditure [18]. Leptin is known to regulate postprandial thermogenesis [29]. Hence, the lack of leptin in the ob/ob–Erk1 −/− mice could prevent the increase in thermogenesis observed in the high-fat-feeding model and favour the development of fat mass. Alternatively, it is possible that the relative impairment in adipogenesis that we observed in high fat fed Erk1 −/− mice was dependent on intact leptin signalling.

Importantly, we found that at 12 weeks of age, ob/ob–Erk1 −/− mice were partially protected against insulin resistance despite massive obesity, demonstrating that Erk1 deficiency can influence insulin resistance independently of an effect on the development of obesity. To exclude the possibility that ob/ob–Erk1 −/− mice may only have delayed onset of obesity-associated metabolic diseases, we studied a small group of older mice. We found that the improvement in fed glycaemia and glucose tolerance was retained. However, a deeper investigation with more mice is necessary to answer this question conclusively.

The improvement in insulin sensitivity is associated with a decrease in the expression of genes encoding inflammatory markers in epididymal adipose tissue. Adipose tissue macrophages have been identified as the primary source of inflammatory cytokine production in adipose tissue [30] but it is unlikely that the reduced production of inflammatory markers is due to a decrease in macrophage content in the adipose tissue of ob/ob–Erk1 −/− mice. Indeed, we did not find any significant modification in the expression of the macrophage marker Emr1 mRNA between ob/ob and ob/ob–Erk1 −/− mice. Adipose tissue macrophages consist of, at the minimum, classically activated M1 macrophages and alternatively activated M2 macrophages [31–33]. We found that the mRNA levels of several M1 inflammatory genes were decreased in epididymal adipose tissue while the expression of genes encoding inflammation-suppressive factors such as IL-10 was not modified. This result suggests that although Erk1 invalidation did not markedly affect macrophage numbers, it could modify the ratio of M1 to M2 macrophages. It has been reported that steadily increasing T(h)1 and CD8+ cells numbers could be responsible for a shift to M1 macrophages in obese adipose tissue [27, 28]. This is relevant to our studies because we found that the expression of mRNA of markers of T(h)1 and CD8+ cells was reduced in adipose tissue of ob/ob–Erk1 −/− mice. As a consequence, Ifng expression was also reduced, and this could have contributed to a reduction in fat inflammation [34, 35]. Another possible explanation could be an impairment in the production of inflammatory cytokines by M1 macrophage that lack ERK1. Indeed, several cellular studies have reported that the production of some cytokines, including TNF-α and plasminogen activator inhibitor type 1 (PAI-1), depends on ERK activity and that ERK1 rather than ERK2 is involved [36–38]. Nevertheless, our study indicates that the protection from obesity-induced insulin resistance appears to be paralleled by reduced adipose tissue inflammation in the ob/ob–Erk1 −/− mice and this finding underlines a role of ERK1 in fat inflammation.

Our study also revealed a greater ability of insulin to suppress circulating NEFA levels in ob/ob–Erk1 −/− mice compared with ob/ob mice. This result is indicative of improved insulin action in adipose tissue and is consistent with reduced inflammatory cytokine production. Indeed, the production of phosphodiesterase 3B, a main regulator of the anti-lipolytic effect of insulin, is negatively regulated by TNF-α through an ERK pathway [12]. Further, we have recently shown that the inflammatory cytokines increase ERK activity and lipolysis in rodent and human adipocytes through the activation of TPL2 [10]. Here, we showed that the expression of Map3k8 mRNA was decreased in the adipose tissue of the ob/ob–Erk1 −/− mice. In adipose tissue, the ERK pathway is also involved in the downregulation of insulin signalling through the negative regulation of IRS production [8]. Among the different IRS proteins, IRS-2, through Akt activation, seems to play a major role in the anti-lipolytic effect of insulin [39]. It is therefore possible that invalidation of Erk1 prevents the downregulation of IRS-2 in adipose tissue. Consistent with this hypothesis, we found improved Akt activation in the adipose tissue of the ob/ob–Erk1 −/− mice following insulin injection.

It is well known that NEFA have a lipotoxic effect in muscles that leads to an alteration in insulin signalling and action [4, 40]. Interestingly, we found that glucose transport was increased in muscles of ob/ob–Erk1 −/− mice compared with ob/ob mice and there was also a higher level of insulin-induced Akt activation. Thus, reduced NEFA flux in ob/ob–Erk1 −/− mice could protect muscles against insulin resistance, contributing to the improvement in glucose homeostasis in those mice.

We also found that the amount of triacylglycerol in the liver of the ob/ob–Erk1 −/− mice was reduced by 35% and the size and number of lipid droplets were reduced. This effect could be due to several mechanisms. The reduced NEFA flux from adipose tissue could result in lower ectopic fat deposition in the livers of the ob/ob–Erk1 −/− mice. Further, we found a reduced level of ACC that could limit liver fatty acid synthesis and could also increase fatty acid oxidation through reduced malonyl CoA formation [41]. An increase in the level of ACC and expression of other lipogenic genes is found in the livers of several models of obese mice and could result from an increase in endoplasmic reticulum stress [42, 43]. Endoplasmic reticulum stress could thus be reduced in the liver of the ob/ob–Erk1 −/− mice because of the improvement in glucose and lipid homeostasis. Hepatic accumulation of fatty acid derivatives is involved in the alteration of insulin signalling and action [44]. Thus, the decrease in fat liver content suggests that insulin action could be improved in the liver of the ob/ob–Erk1 −/− mice. In agreement with this hypothesis, we found that the glycogen content of the livers of ob/ob–Erk1 −/− mice was increased following insulin stimulation in the hyperinsulinaemic–euglycaemic clamp studies.

Taken together, our results demonstrate that the lack of ERK1 could partially protect obese mice against insulin resistance and liver steatosis by decreasing adipose tissue inflammation and by increasing muscle glucose uptake. These results link ERK1 activity to the development of insulin resistance independently of its effect on obesity and indicate that deregulation of the ERK1 pathway could be an important component in obesity-associated metabolic disorders. However, because the animal model is a global knockout, we cannot totally exclude that our observations reflect developmental events. Tissue-specific inactivation of Erk1 is now needed to answer this important question.