Introduction

Growing evidence shows that endoplasmic reticulum (ER) stress is an important mechanism linking obesity, insulin resistance and glucose intolerance [1, 2]. Several reports have demonstrated that ER stress markers are upregulated in the liver and fat of obese mouse models [2] and that ER stress in the hypothalamus is associated with central insulin and leptin resistance, hyperphagia and weight gain [3].

C/EBP homologous protein (CHOP) is a 29 kDa protein regulating ER-stress-mediated apoptosis [4]. Downstream of Chop, different apoptotic effectors have been reported [5], but overall, Chop-induced signalling is not well understood. Overexpression of Chop decreases B cell CLL/lymphoma 2 (BCL-2) protein levels and causes translocation of BCL2-associated X protein (BAX) protein from the cytosol to the mitochondria, transmitting the death signal to the mitochondria [6]. Progressive hyperglycaemia in the Akita mouse, a model spontaneously developing hyperglycaemia with reduced beta cell mass [7], is accompanied by Chop induction and beta cell apoptosis, but deletion of Chop delayed onset of diabetes in heterozygous Akita mice [8]. CHOP is also involved in the development of obesity [9] and regulation of inflammation [1013]. Lung damage induced by lipopolysaccharide (LPS) treatment is attenuated in Chop / mice, suggesting that the ER-stress pathway involving CHOP is activated and displays a role in the pathogenesis of septic shock lung. Attenuation of IL-1β activity in broncho-alveolar lavage fluid of Chop −/− mice and suppression of caspase-11, which is needed for the activation of procaspase-1 and pro-IL-1β, have been demonstrated [10, 11]. It has been suggested that LPS, a Toll-like receptor 4 (TLR4) ligand, induces a marked increase in protein synthesis, leading to prolonged ER stress during LPS–TLR4 signalling [14]. In addition, LPS caused accelerated upregulation of ER-stress-response genes such as Chop and pro-inflammatory cytokine genes [15], but the underlying molecular mechanisms of LPS-induced ER stress are not well understood. Chop −/− mice also show low expression levels of caspase-1 and -11 and Il-1β (also known as Il1b) in chemically induced pancreatitis, whereas these mediators are strongly upregulated when pancreatitis is induced in control mice. As a result, Chop deletion results in suppressed apoptosis of acinar cells and lower infiltration of inflammatory cells in the pancreas [13]. A role for Chop was also described in experimental colitis: when colitis was induced, the elevation of mRNA expression levels of Mac-1 (also known as Itgam), Ero1α (also known as Ero1l) and caspase-11 was suppressed in Chop −/− compared with wild-type mice. As a result, it was suggested that CHOP activity involves various stimulatory mechanisms, such as macrophage infiltration via induction of macrophage-1 antigen (MAC-1), reactive oxygen species (ROS) production via ER oxidase 1α (ERO1α) and IL-1β production via caspase-11, all resulting in mucosal cell apoptosis [12]. Finally, it has been demonstrated that CHOP is an inhibitor of the wingless-Int (WNT) signalling pathway and is necessary for normal bone development as Chop overexpression induced osteopenia [16].

The aim of the present study was to define the in vivo role of CHOP in obesity and insulin resistance by studying the effects of systemic deletion of this transcription factor in mice.

Methods

Details of the islets and peritoneal macrophage isolation and culture, transmission electron microscopy quantitative RT-PCR and western blot analysis are available in the electronic supplementary material (ESM).

Animals

Female C57Bl/6 (wild type [WT]) (Harlan, Horst, the Netherlands) and Chop −/− mice (provided by S. Akira [Osaka University, Japan] and M. Mori [Kumamoto University, Japan]) [17], backcrossed ten generations on a C57Bl/6 background, were maintained on regular chow containing 70% energy from carbohydrate, 20% from protein and 10% from fat (Research Diets #D12450B; Research Diets, New Brunswick, NJ, USA), and housed on a 12 h light/dark cycle. For experiments in the fasted state, mice were fasted overnight and kept in a separate cage for the indicated period of time. All animal experimental procedures were approved by the Ethics Commission of KULeuven.

Whole body dual-energy X-ray absorptiometry

Fat percentage, bone mass content (BMC) and bone mass density (BMD) were measured in vivo by dual-energy X-ray absorptiometry (DEXA; PIXImus densitometer; Lunar, Madison, WI, USA), using an ultra-high resolution (0.18 × 0.18 pixels, resolution of 1.6 line pairs/mm) and software version 1.45.

Histology and immunohistochemistry

Abdominal adipose tissue, liver, skeletal muscle and pancreas were fixed in 10% (vol./vol.) formaldehyde and embedded in paraffin. Tissues were sectioned in 5 μm slices and stained with haematoxylin and eosin. Photomicrographs were captured with a Zeiss Axiovert microscope (Carl Zeiss, Oberkochen, Germany) and analysed with Axiovision Rel.4.6 software. Adipocyte cell size was measured by determining the size of at least 50 adipocytes in four microscopic fields in fat pads from each mouse (×20 magnification). Islet size was measured by acquiring images from eight to ten distal, random, non-overlapping images of haematoxylin and eosin-stained pancreatic sections (×10 magnification).

Oil red O was used to stain neutral lipids in frozen sections of muscle, liver and pancreas. For immunohistochemistry, paraffin sections were stained for T cells using rabbit anti-CD3 (Dako, Glostrup, Denmark) and for macrophages using rat anti-F4/80 (Serotec, Düsseldorf, Germany).

Metabolic studies

Blood glucose levels were determined from tail vein blood samples using an Accu-Chek Aviva glucose meter (Roche Diagnostics, Vilvoorde, Belgium).

For the glucose tolerance test, mice were fasted overnight and injected intraperitoneally with glucose solution (2 g/kg). Blood glucose concentrations were determined before and at 15, 30, 60, 90 and 120 min after injection. Insulin, leptin and adiponectin levels were measured in plasma by ELISA using insulin standards (Mercodia, Uppsala, Sweden), leptin standards (Crystal Chem, Chicago, IL, USA) and adiponectin standards (Millipore, Billerica, MA, USA), respectively.

Plasma triacylglycerol levels were determined by means of a coupled enzymatic assay previously described [18] but adapted to 96-well plate readers and omitting the lipid extraction step (final volume 250 μl). Analysis of plasma NEFA was done by a coupled enzymatic assay [19], using a commercial kit (NEFA C kit; Wako, Neuss, Germany) adapted to 96-well plates (final volume 240 μl).

Plasma levels of alanine aminotransferase, aspartate aminotransferase and alkaline phosphatase were determined by enzymatic assays using commercial kits (Roche/Hitachi Modular Analytics, Laval, Canada).

Plasma levels of T3 and T4 were measured by radioimmunoassay as described previously [20].

Euglycaemic–hyperinsulinaemic clamp studies

Euglycaemic–hyperinsulinaemic clamp studies were performed on 24-week-old female Chop −/− and WT mice as previously described [21]. Insulin was infused at a rate of 18 mU kg1 min1 for 3 h, whereas euglycaemia was maintained by periodically adjusting the 15% (wt/vol.) glucose infusion rate. Glucose turnover and hepatic glucose production rates were calculated as previously described [21].

For measuring specifically insulin signalling in liver and skeletal muscle, overnight-fasted (16 h) 24-week-old female Chop −/− and WT mice were anaesthetised by intraperitoneal injection of avertin (0.02 ml/g body weight; Fluka Chemica, Bornem, Belgium). Human insulin, 5 U, (Actrapid; Novo Nordisk, Bagsvaerd, Denmark) or NaCl solution was injected through the vena cava inferior. After 5 min, liver, abdominal fat tissue and hindlimb muscles were removed and lysates were subjected to western blotting [22].

Indirect calorimetric and ambulatory activity

At 24 weeks of age, the energy expenditure of WT and female Chop −/− mice was measured for 24 h by indirect calorimetry (Oxylet; Panlab-Bioseb, Chaville, France). Mice were scored for oxygen consumption (\( \dot{V}{{\text{O}}_{{2}}} \)), carbon dioxide production (\( \dot{V}{\text{C}}{{\text{O}}_{{2}}} \)), ambulatory activity and energy expenditure (calculated according to the following formula:

$$ 1.44 \times \dot{V}{{\text{O}}_2} \times \left( {3.815 + 1.232 \times {\text{RQ}}} \right) $$

\( \dot{V}{{\text{O}}_{{2}}} \) and \( \dot{V}{\text{C}}{{\text{O}}_{{2}}} \) were measured over a 24 h period. Ambulatory activities of the mice were monitored by an infrared photocell-beam-interruption method (Sedacom; Panlab-Bioseb).

Statistical analysis

For all experiments, comparisons were performed by a two-sided unpaired t test. p < 0.05 was considered statistically significant. The data are presented as means ± SEM.

Results

Chop deletion induces obesity

At 8 weeks of age, body weight was higher in Chop −/− mice and they gained progressively more weight compared with WT mice (Fig. 1a). Body composition revealed no difference in percentage body fat up to 16 weeks of age. At 20 weeks of age, body fat tended to increase in Chop −/− compared with WT mice (20.5% ± 1.2 vs 16.2% ± 2.0; p = 0.08) and it was significantly higher from week 24 onwards (Fig. 1b). Chop −/− mice showed mainly an increase in abdominal fat (Fig. 1c). The adipocyte size in abdominal fat pads of Chop −/− mice was 34% and 41% larger compared with WT mice at 24 and 40 weeks of age, respectively (Fig. 2a,b). Compared with WT mice, the hypertrophic adipocytes of Chop −/− mice showed increased protein levels of the typical ER-stress markers glucose-regulated protein 78 (GRP78), X-box binding protein 1 spliced (XBP-1s) and activating transcription factor (ATF) 4 and 6 (Fig. 2c,d), whereas no differences could be observed in mRNA expression levels of these markers in Chop −/− compared with WT mice (ESM Fig. 1a).

Fig. 1
figure 1

Body weight and body fat percentage of female WT and Chop −/− mice. Body weight (a), percentage of body fat (b) of female WT (white bars) and Chop −/− (black bars) mice and representative WT and Chop −/− mice photographed at 24 weeks of age (c). Body weight was measured weekly, whereas body fat was measured at 8, 12, 16, 24 and 34 weeks of age using dual X-ray absorptiometry. Data are shown as the means ± SEM (n = 8). ** p < 0.01 and *** p < 0.001 vs WT mice

Fig. 2
figure 2

Adipocyte size, ER-stress markers and leptin levels in WT and Chop −/− mice. a Adipocyte size of WT (white bars) and Chop −/− (black bars) mice at 24 and 40 weeks of age. Adipocyte cell size was measured by determining the diameter of at least 50 cells in four different microscopic fields in fat pads of two different mice (×20 magnification). *** p < 0.001 vs WT mice. b H&E staining of representative adipocyte sections of 24- and 40-week-old mice of the genotypes indicated. Photomicrographs (×20 magnification) were captured with a Zeiss Axiovert microscope and analysed with Axiovison Rel. 4.6 software. The pictures are representative of observations made in two different mice. c Western blotting for the ER-stress markers ATF6, ATF4, GRP78, XBP-1 s and stable housekeeping protein β-actin on total abdominal fat extracts of female WT and Chop −/− mice at 24 weeks of age (n = 4). d Relative protein level normalised to the stable housekeeping protein β-actin in total abdominal fat extracts of female WT (white bars) and Chop −/− mice (black bars) at 24 weeks of age. Results are means ± SEM (n = 4 independent experiments). * p < 0.05; ** p < 0.01; *** p < 0.001 vs WT mice. e Plasma leptin levels of WT (white bars) and Chop −/− mice (black bars) at 8, 24 and 40 weeks of age. Data are shown as the means ± SEM (n = 8). ** p < 0.01 and *** p < 0.001 vs WT mice

Levels of mRNA expression of C/ebpα (also known as Cebpa) (2.0 ± 0.5 vs 1.7 ± 0.5; NS) and Pparγ (also known as Pparg) (1.2 ± 0.1 vs 1.2 ± 0.1; NS) in abdominal fat tissue were comparable between Chop −/− and WT mice, whereas C/ebpβ (also known as Cebpb) expression was lower in Chop −/− mice compared with WT mice (0.2 ± 0.03 vs 0.3 ± 0.03; p < 0.05). Expression levels of mRNA encoding carbohydrate responsive element binding protein (ChREBP) and sterol regulatory element binding protein (SREBP), as well as of hexokinase 2 (HK2), acetyl-CoA carboxylase (ACC) and fatty acid synthase (FAS) were not different in the abdominal fat tissue of Chop −/− and WT mice (ESM Fig. 1c).

The increase in fat mass in Chop −/− mice was accompanied by an increase in BMC and BMD (ESM Fig. 2a,b), as well as by increased plasma leptin levels (Fig. 2e). Chop −/− mice showed a trend to higher cumulative food intake compared with WT mice (3.56 ± 0.74 vs 1.91 ± 0.39 g/day; p = 0.09). Energy expenditure was similar in Chop −/− and WT mice, both during daylight and night time (Fig. 3a,b), whereas the RQ of Chop −/− mice was lower compared with WT mice (Fig. 3c). Plasma levels of thyroid hormones T3 and T4, mRNA expression levels of Ucp1 in brown adipose tissue and cage movements were similar between Chop −/− and WT mice (ESM Fig. 3).

Fig. 3
figure 3

Energy expenditure and RQ of WT and Chop −/− mice at 24 weeks of age. Energy expenditure, expressed as kJ day1 metabolic weight1, of WT (white bars) and Chop −/− mice (black bars) at 24 weeks of age during daylight (a) and night time (b). Data are shown as the means ± SEM (n = 4). c Recording over 24 h of RQ of WT (white squares) and Chop −/− mice (black squares) at 24 weeks of age. Data are shown as the means ± SEM (n = 4)

Chop deletion induces hepatic fat accumulation

At 24 weeks of age, Chop −/− mice showed increased fat deposition in liver compared with WT mice (Fig. 4a,b,g), whereas no difference in fat deposition could be detected in skeletal muscle (Fig. 4c,d) and pancreas (Fig. 4e,f). Liver enzyme levels were lower in Chop −/− mice (Fig. 4h) and mRNA (ESM Fig. 1b) as well as protein levels of the major ER-stress markers GRP78 and XBP-1s were not altered in Chop −/− compared with WT mice (ESM Fig. 4a,b).

Fig. 4
figure 4

a–f H&E and Oil Red O staining of liver, muscle and pancreas and liver enzymes of WT and Chop −/− mice. H&E (a,c,e) and Oil Red O (b,d,f) staining of representative (of two independent observations) liver (a,b), skeletal muscle (c,d) and pancreas (e,f) sections from mice of the indicated genotypes at 24 and 40 weeks of age. Photomicrographs were captured with a Zeiss Axiovert microscope and analysed with the Axiovison Rel. 4.6 software. g Representative liver of female WT and Chop −/− mice photographed at 24 weeks of age. h Levels of the transaminases alkaline phosphatase, aspartate aminotransferase and alanine aminotransferase in WT (white bars) and Chop −/− mice (black bars) at 24 weeks of age. Data are shown as the means ± SEM (n = 6). * p < 0.05 and ** p < 0.01 vs WT mice. ALT, alanine aminotransferase; AP, alkaline phosphatase; AST, aspartate aminotransferase

No differences were detected in plasma NEFA and triacylglycerol levels between WT and Chop −/− mice at 24 and 40 weeks of age (ESM Fig. 4c,d).

Levels of mRNA expression of Chrebp (also known as Mlxipl), Srebp (also known as Srebf1), Hk2, Acc (also known as Acaca) and Fas (also known as Fasn) were not different in the livers of Chop −/− and WT mice (ESM Fig. 5a). In addition, nuclear protein levels of ChREBP and SREBP, as well as total ACC and FAS protein levels were unchanged in liver of Chop −/− compared with WT mice at 24 weeks of age (ESM Fig. 5b).

Chop−/− mice maintain a normal glycaemia, but Chop deletion does not protect islets

Chop −/− mice showed normal fasting glycaemia (2.82 ± 0.27 vs 3.13 ± 0.27 mmol/l in WT mice; NS) and glucose tolerance remained comparable between Chop −/− and WT mice (ESM Fig. 6). Plasma insulin levels were higher in Chop −/− mice at 24 weeks of age (149.18 ± 16.53 vs 99.35 ± 3.83 pmol/l in WT mice; p < 0.05), which was associated with increased total islet size (17,256 ± 1443 vs 8,954 ± 719 μm2 in WT mice; p < 0.001). Electron microscopy of beta cells did not reveal any changes in ER structure between Chop −/− and WT islets (ESM Fig. 7a). No signs of apoptosis were observed in islets of Chop −/− and WT mice after exposure to thapsigargin for 48 h (92.4% ± 1.4 and 93.3% ± 2.3% of living islet cells in Chop −/− and WT mice, respectively, vs 95.6% ± 1.1 and 96.4% ± 0.2 of living islet cells in control conditions in Chop −/− and WT mice, respectively). Exposure of islets to unsaturated or saturated NEFA, thapsigargin or the combination of IL-1 and IFN-γ for 5 days caused similar apoptotic rates in Chop −/− and WT islets (ESM Fig. 7b).

Chop−/− mice maintain normal insulin sensitivity

The glucose turnover rate, glycolysis and glycogen synthesis rates, as measured by euglycaemic–hyperinsulinaemic clamps at 24 weeks of age, were similar in WT and Chop −/− mice (Fig. 5a–c). Hepatic glucose production (−5.87 ± 3.81 vs −4.79 ± 4.83 mg kg1 min1 in Chop −/− and WT mice, respectively; NS), as well as the level of serine and threonine phosphorylated thymoma viral proto-oncogene 1 (AKT) levels, were similar in liver of Chop −/− and WT mice (Fig. 5d,e), all confirming sustained hepatic insulin signalling. Hepatic insulin signalling was further corroborated as p(Ser)AKT phosphorylation levels were similarly increased on insulin injection (about 1.5-fold) in Chop −/− and WT mice, and as total IRS1 and IRS2 levels were unaffected in Chop −/− mice compared with WT mice. The level of serine phosphorylation of IRS1 and IRS2, a negative regulator of insulin signalling, was similar in Chop −/− and WT mice. In addition, Chop −/− mice showed a trend (p = 0.1) for increased protein level of suppressor of cytokine signalling 3 (SOCS3) compared with WT mice, whereas no changes in SOCS3 level could be observed on insulin stimulation (Fig. 6a–g).

Fig. 5
figure 5

Glucose turnover rate, glycolysis and glycogen synthesis during euglycaemic–hyperinsulinaemic clamps and p(Ser)AKT and p(Thr)AKT protein levels in Chop −/− and WT mice. Glucose turnover rate (a), glycolysis (b) and glycogen synthesis rate (c) during euglycaemic–hyperinsulinaemic clamps of female WT (white bars) and Chop −/− (black bars) mice at 24 weeks of age. Data are the means ± SEM (n = 6–8). d,e Protein phosphorylation levels of p(Ser)AKT (d) and p(Thr)AKT (e), measured in total liver extract of non-fasted female WT (white bars) and Chop −/− (black bars) mice at 24 weeks of age. Data are shown as means ± SEM (n = 4). A representative image of both WT and Chop −/− mice is shown

Fig. 6
figure 6

Protein levels of different markers of insulin sensitivity in liver and skeletal muscle. a Western blotting for the indicated markers of insulin sensitivity and stable housekeeping protein β-actin on total liver extracts of female WT and Chop −/− mice, injected with either NaCl solution or 5 U insulin in the vena cava inferior, after an overnight fast at 24 weeks of age (n = 4). A representative image of two mice/experimental group is shown. b-g Relative protein level normalised to the stable housekeeping protein β-actin in total liver extracts of female WT and Chop −/− mice, injected with either NaCl solution (white bars) or 5 U insulin (black bars) in the vena cava inferior, at 24 weeks of age. Results are means ± SEM (n = 4 independent experiments). h Western blotting for p(Ser)AKT and stable housekeeping protein α-tubulin on total hind-limb muscle extracts of female WT and Chop −/− mice, injected with either NaCl solution or 5 U insulin in the vena cava inferior at 24 weeks of age (n = 4). A representative image of two mice/experimental group is shown. In a and h, Chop = Chop −/−

Finally, insulin sensitivity was sustained in the skeletal muscles of Chop −/− mice, as the level of AKT serine phosphorylation was elevated to the same extent in the hind limbs of Chop −/− and WT mice on insulin stimulation (Fig. 6h).

Reduced inflammation in Chop−/− mice

Immunohistochemical analysis of livers from 24-week-old Chop −/− and WT mice revealed a decreased presence of macrophages (7.9% ± 0.9 vs 15.6% ± 0.6 F4/80+-stained macrophages per microscopic field in Chop −/− vs WT mice, respectively; p < 0.001; Fig. 7a) and T lymphocytes (3.9% ± 0.2 vs 10.3% ± 0.6 CD3+-stained T lymphocytes per microscopic field in Chop −/− vs WT mice, respectively; p < 0.001; Fig. 7b) in Chop −/− mice compared with WT. The lower inflammatory status of liver was confirmed at the mRNA level, with lower expression of mRNA encoding the chemokines macrophage-derived chemokine (MDC) and interferon-inducible protein 10 (IP10) in the livers of 8-week-old Chop −/− mice (Fig. 7c). These differences persisted at 24 weeks when, in addition, lower levels of mRNA encoding the chemokines monocyte chemotactic protein 1 (MCP1) and the pro-inflammatory cytokine TNFα were detected in Chop −/− mice compared with WT (p < 0.05; Fig. 7d).

Fig. 7
figure 7

Macrophage and lymphocyte infiltration and cytokine/chemokine mRNA expression in liver of WT and Chop −/− mice. Anti-F4/80 (a) and anti-CD3 (b) staining of representative liver sections of WT and Chop −/− mice at 24 weeks of age. Representative photomicrographs (×20 magnification) in two mice were captured with a Zeiss Axiovert microscope and analysed with the Axiovison Rel. 4.6 software. Levels of mRNA expression, normalised to the stable housekeeping gene Rpl27, of the indicated cytokines and chemokines in liver of female WT (white bars) and Chop −/− (black bars) mice at of 8 (c) and 24 (d) weeks of age. Data are shown as means ± SEM (n = 8). * p < 0.05 vs WT mice. (Ctack is also known as Ccl27a; Itac is also known as Cxcl11; Mcp1 is also known as Ccl2; Mdc is also known as Ccl22; Mip3a is also known as Ccl20)

In adipose tissue, a lower number of T lymphocytes (CD3+) were observed in 24-week-old Chop −/− mice (6.05 × 10−5 ± 1.2 × 10−5 vs 2.0 × 10−4 ± 2.6 × 10−5 CD3+ cells per μm2 in Chop −/− vs WT mice, respectively; p < 0.001), but there was no difference in macrophage (F4/80+) infiltration (ESM Fig. 8a) or cytokine/chemokine mRNA expression (ESM Fig. 8b). In addition, Chop −/− mice were characterised by lower plasma adiponectin levels compared with WT mice p < 0.01; ESM Fig. 8c).

Peritoneal macrophages from Chop −/− mice showed lower expression levels of the pro-inflammatory cytokines Il1, Il6 and Tnfα (also known as Tnf), the chemokine Ip10 (also known as Cxcl10), iNos (also known as Nos2) and caspase-1 when stimulated in vitro with LPS (Fig. 8).

Fig. 8
figure 8

Cytokine and chemokine mRNA expression in intraperitoneal macrophages of female WT and Chop −/− mice. Levels of mRNA expression, normalised to the stable housekeeping gene Rpl27, of Il1 (a), Il6 (b), Tnfα (c), Ip10 (d), iNos (e) and caspase-1 (f) in isolated intraperitoneal macrophages of 8-week-old female WT (white bars) and Chop −/− (black bars) mice, stimulated with LPS or not (control) for 6 h. Data are shown as means ± SEM (n = 8). * p < 0.05 vs WT mice. Con, control

Discussion

Deletion of Chop disclosed an interesting phenotypic trait, presenting as increased body weight and fat mass, liver steatosis and absence of insulin resistance. The observed weight gain and increased body mass in female Chop −/− mice are in line with previous results [9, 23]. Chop −/− mice had a similar energy expenditure, but showed a trend for increased food uptake. This observation is in contrast with a previous report [9]. However, it should be stated that our measurement of food intake is more reliable, as it was determined using extremely sensitive metabolism cages, whereas Ariyama et al. did not use this gold standard. The lower RQ of Chop −/− compared with WT mice, demonstrating a higher lipid over carbohydrate oxidation, indicates qualitative changes in energy expenditure. It is unlikely that Chop deletion induces obesity by adipogenesis as similar expression levels of C/ebpα and Pparγ and a decreased expression level of C/ebpβ were observed in the abdominal fat of Chop −/− mice. These results are in contrast with previous findings showing increased protein levels of C/EBPβ and of the 30 kDa truncated form of C/EBPα in perimetrial fat tissue of female Chop −/− mice, whereas expression of the 42 kDa form of C/EBPα was reduced [9]. Whereas we measured mRNA levels, Ariyama et al. determined protein levels of the C/EBPs. As discussed below, mRNA and protein levels do not always correlate. As both forms of C/EBPα are produced by alternative initiation of translation and as one isoform was found to be increased whereas the other was found to be decreased, it is plausible that total C/ebpα mRNA levels, as measured in our study, are unchanged. Furthermore, we cannot rule out differences in levels between abdominal fat tissue (our study) and perimetrial fat tissue [9]. Finally, Ariyama et al. performed their estimations of C/EBP level at 12 months of age, whereas we performed them at 24 weeks of age. In general, it should be stated that in the paper buy Ariyama et al., little information is given about the age of the mice at time of the analyses and the diet used. Chop −/− mice were also characterised by increased BMC and BMD, even at 4 weeks of age, suggesting a regulatory role for CHOP in bone development.

Altered thermogenesis and physical activity do not play a predominant role in the obese phenotype of Chop −/− mice. The decreased serum oestrogen levels previously found [9] could contribute to increased fat mass, but as a similar decrease in oestrogen levels without any effect on abdominal fat mass was observed in male Chop −/− mice in the same study, differences in oestrogen levels probably cannot explain the abdominal obesity. Thus, to date, the cause of the observed obesity remains unclear, as of the processes that were tested only nutrient consumption showed a trend to be abnormal.

As targeted disruption of Chop delays the onset of diabetes in heterozygous Akita mice [8], we hypothesised that deletion of Chop could protect beta cells from apoptosis and delay the onset of type 2 diabetes. This hypothesis was strengthened by recent observations that knockdown of Chop by small interfering RNA protects adult beta cells against NEFA and chemical ER stressors [24]. We observed increased islet size in Chop −/− mice, confirming previous observations [23], but Chop −/− islets were not protected against beta cell death induced by NEFA or thapsigargin in vitro. These observations are in line with previous observations that islets isolated from WT and Chop −/− mice did not reveal any differences in acute toxicity after streptozotocin treatment [23]. As Chop −/− mice showed milder glucose intolerance under high-fat-diet conditions and after streptozotocin administration, they suggested that Chop deletion rather preserves beta cell function and influences cell recovery and/or function [23].

Electron microscopy of beta cells did not reveal differences between Chop −/− and WT mice. This absence of a clear effect on beta cells may be due to the multiple effects of CHOP in vivo and to putative adaptation to long-term lack of CHOP in beta cells, starting in embryonic life.

The increase in protein levels of ER-stress markers in Chop −/− mice is in line with previous reports, demonstrating that excessive fat deposition leads to hypertrophy of adipocytes, activating an ER stress response [25]. Of note, there is a discrepancy between mRNA and protein levels. In this view, we have previously shown that changes in mRNA levels initiating the unfolded protein response are not translated into protein changes. Moreover, we stated an important role for post-translational modifications in the regulation of ER stress [26, 27]. Activation of ER stress induces an increased release of NEFA, adipokines and inflammatory mediators [28, 29], triggering infiltration and activation of macrophages in adipose tissue, contributing to insulin resistance [30, 31]. These infiltrating immune cells further stimulate secretion of adipokines from adipocytes [31] and secrete pro-inflammatory cytokines of which TNFα and IL-6 are implicated in the induction of insulin resistance [32, 33]. The absence of inflammation around the hypertrophic adipocytes of Chop −/− mice contrasts with the increased numbers of resident macrophages in adipose tissue of obese individuals [34] and suggests that CHOP provides a crucial pro-inflammatory signal in fat. In contrast to a previous finding [9], adiponectin levels were lower in plasma of Chop −/− compared with WT mice. This controversy could be explained by differences in, for example, age or diet used, which are not stated in the previous report. It could be speculated that CHOP is an important link between the activated ER-stress response and the induction of inflammation in adipocytes, an effect independent of adiponectin. It has been demonstrated extensively that the presence of ectopic fat in insulin sensitive tissues contributes to insulin resistance [35, 36]. Chop −/− mice are characterised by massive liver steatosis, but remain perfectly insulin sensitive and glucose tolerant and have normal circulating NEFA and triacylglycerol levels.

The expression level of major regulatory enzymes of lipogenesis was unchanged in Chop −/− mice, both at the mRNA and protein level, ruling out increased production of triacylglycerol in the liver. A reduction in VLDL secretion by the liver of Chop −/− mice could be a possible explanation for the observed steatosis, as blocking VLDL secretion causes hepatic steatosis without affecting hepatic glucose production, hepatic insulin signalling, peripheral lipid stores or insulin sensitivity in mice [37]. Despite massive steatosis, no signs of steatohepatitis were present. On the contrary, there was less infiltration of macrophages and T lymphocytes and lower transaminase levels. Others have previously suggested that chronic steatosis-associated inflammation leads to insulin resistance [38]. The lower levels of immune cells in liver were paralleled by lower mRNA expression of chemokines, already noticeable in 8-week-old Chop −/− mice. Some chemokines, such as MCP1, are induced in the adipose tissue of obese mice and may contribute to the development of insulin resistance and de-differentiation of adipocytes [3941]. Blocking the MCP1 receptor ameliorates insulin resistance and steatosis in db/db mice [42], so decreased hepatic Mcp1 expression in Chop −/− mice could contribute to decreased macrophage recruitment and amelioration of insulin sensitivity. Additionally, inflammation in Chop −/− mice could also be suppressed by SOCS3, as increases in SOCS3 protein levels, although minor, were observed in liver. It has been demonstrated that SOCS3 suppresses cytokine-mediated signal transduction and liver apoptosis induced by several pathogen-derived agonists, leading to increased survival rates [43]. Nevertheless, other reports demonstrated that hepatic SOCS3 is a negative regulator of insulin sensitivity by interfering with correct IRS1 and IRS2 phosphorylation, inhibiting downstream insulin signalling. Deletion of hepatic Socs3 increases liver insulin sensitivity in mice fed a control diet, but promotes development of obesity and systemic insulin resistance by mimicking chronic inflammation [44, 45]. However, insulin signalling was sustained in the livers of Chop −/− mice on insulin stimulation, indicating that in our model, SOCS3 has a role in maintaining insulin sensitivity by suppressing inflammation.

As steatosis is associated with ER stress [46] and as activation of ER stress is related to the induction of inflammatory pathways, absence of ER stress in Chop −/− mice, both at the mRNA and protein level, may also contribute to the absence of inflammation in liver and to the maintenance of insulin sensitivity.

Finally, Chop −/− mice have no ectopic fat deposition in muscle and a preserved AKT serine phosphorylation on insulin stimulation, which could contribute to preserving peripheral insulin sensitivity in these mice [47].

Recent evidence points to a role of CHOP in various inflammatory diseases, such as LPS-induced lung damage [10, 11], chemically induced pancreatitis [13] and experimental colitis [12]. Here, we demonstrate that CHOP is indeed central in the inflammatory response of liver, fat and macrophages, as peritoneal macrophages from Chop −/− mice exhibit a blunted inflammatory response to LPS. The exact downstream targets of CHOP involved in inflammation are not well understood, but CHOP could induce inflammation via several pathways. As such, CHOP induces activation of ERO1α, leading to exaggerated production of ROS [46, 48], which promotes the production of pro-inflammatory cytokines [49]. In addition, CHOP directly regulates the IL-8 promoter [50] and can induce IL-1β production through activation of caspase-11 [11].

Based on the present observations we suggest that Chop −/− mice remain insulin sensitive in spite of severe obesity and ectopic fat deposition because Chop deletion prevents the inflammatory reaction. This indicates that it is not the fat mass per se, but rather the associated inflammatory status that determines the onset of insulin resistance. We propose that CHOP is a crucial mediator in the induction of insulin resistance, serving as a link between fat accumulation and insulin resistance.