Inhibition of hypothalamic leukemia inhibitory factor exacerbates diet-induced obesity phenotype

Male Swiss mice were provided by the University of Campinas Breeding Center and the study was approved by the Ethics Committee of the University of Campinas (Project #: CEUA 2926–1). For all experiments, 5-week-old mice were kept in individual cages in a silent environment with controlled temperature using a 12 h light/12 h dark cycle (6/18 h). Mice had free access to water and food ad libitum. For the experiments, mice were randomly divided into groups fed either chow or a HFD (composition of the diets is presented in Additional file 1: Table S1). To identify OP and OR mice, we used a previously described protocol [9]. In short, 8-week-old male Swiss mice were fed a HFD for 24 h, and total caloric intake was recorded. Mice on the upper quartile of caloric intake were defined as OP, whereas mice on the lower quartile of caloric intake were defined as OR (Fig. 1). Thereafter, mice were fed chow for 15 days and the inhibition of LIF was performed. On the next day, HFD was offered for up to 15 days. Body mass and caloric intake were determined every second day. At the end of each experiment, mice were sacrificed by decapitation. For this, they were anesthetized with sodium thiopental (25 mg/kg). Tissue specimens were extracted and stored at −80 °C until analysis.

The body composition was assessed by dual energy X-ray absorptiometry (DEXA, Discovery Wi QDR Series; Hologic Apex Software, Hologic Inc.) in mice fed a HFD for 14 days.

The PCR array was performed after 1 day of HFD in OP or OR mice. The kit used for the determination of gene expression was the Chemokines and Receptors PCR Array (PAMM-022Z, Qiagen). This is a 96-gene panel, with 84 genes for chemokine/chemokine receptors and the remaining for endogenous and positive controls of the PCR reaction (Additional file 1: Table S2).

OR mice were randomly selected for the treatment with either an immunoneutralizing antibody against LIF or a non-immune IgG. OP mice were always treated with the non-immune IgG. For this, mice were anesthetized with a mixture of ketamine (100 mg/kg) and xylazine (10 mg/kg) and stereotaxically instrumented using a Stoelting stereotaxic apparatus. The OR anti-LIF group received an intracerebroventricular microinjection of 1.0 μl of a solution of an anti-LIF antibody (40 ng/μl) (sc 1336; Santa Cruz Biotechnology, Santa Cruz, CA) in the third ventricle close to the hypothalamus. In the same way, OR IgG or OP IgG mice received 1.0 μl of non-immune IgG. The coordinates were Bregma 0.0 mm; depth 4.5 mm; lateral 0.0 mm to reach the third ventricle. Thereafter, the animals were fed an HFD ad libitum for 1 or 15 days.

The glucose tolerance test (GTT) was performed after 12 days on the HFD. In the morning, mice were fasted for 4 h and the blood collection was carried out from the tail to determine glucose levels. A 25% glucose solution was administrated via intraperitoneal injection, and determinations occurred immediately before injection and after 15, 30, 60, 90, and 120 min. The measurements were performed using a portable glucose meter (Optium Xceed - Abbott®). Then, the area under the glucose curve (AUC) was calculated.

The insulin-tolerance test (iTT) was performed after 13 days on the HFD. In the morning, mice were fasted for 4 h and the blood collection was carried out from the tail to determine glucose levels. After 4 h of fasting, the iTT test was performed. The blood collection was carried out from the tail to obtain glucose levels. Insulin (1.5 IU/kg body mass) was administrated via intraperitoneal injection and blood glucose was measured immediately before injection and after 5, 10, 15, 20, 25, and 30 min.

For acclimation, mice were placed in the respirometric chamber 1 day prior to the beginning of the measurements. The O₂ consumption, CO₂ production, respiratory quotient (RQ), and spontaneous activity were determined for a period of 24 h in the Gas Analyzer LE405 Gas Analyzer (Panlab - Harvard Appliance, Holliston, MA, USA). The airflow was maintained by Air Supply and Switching (Airlab - Harvard Apparatus, Holliston, MA, USA) and gas analyzer calibrated with known concentrations of O₂ and CO₂ (Air Liquid, São Paulo, Brazil). The calculations were performed using the Metabolism 2.2v software.

Reactions were performed using the TaqMan™ System (Applied Biosystems). The gene glyceraldehyde-3-phosphate dehydrogenase (GAPDH) was chosen as the endogenous control for the reaction. The expression of tumor necrosis factor alpha (TNF-α, Mm00443258_m1), interleukin 1 beta (IL-1β, Mm00434228_m1), interleukin 6 (IL-6, Mm00446190_m1), interleukin 10 (IL-10, Mm01288386_m1), Cx3cl1 (Mm00436454_m1), Ccl20 (Mm01268754_m1), Cxcl1 (Mm.PT.5842076891), and LIF (Mm.PT.58.13926050) was quantified in the hypothalamus. Hypothalami were extracted 1 day after the introduction of the HFD in anti-LIF or IgG-treated mice. For the determination of relative transcript expression, real-time PCR reactions were performed in duplicate as follows: 3.0 μl TaqMan Universal PCR Master Mix 2X, 0.25 μl of the primers and probe solution, 2.75 μl water, and 4.0 μl cDNA. The values of relative gene expression were obtained by analyzing the results using 7500 System SDS software (Applied Biosystems).

Six-week-old male Swiss mice fed on chow were anesthetized with a mixture of ketamine (100 mg/kg) and xylazine (10 mg/kg) perfused with saline and 4% paraformaldehyde. The brain was totally removed and kept in a 4% paraformaldehyde solution for 24 h, followed by 48 h in a 30% sucrose solution for cryoprotection. Coronal sectioning at a thickness of 20 μm was performed using a cryostat (LEICA Microsystems, CM1860, Buffalo Grove, IL, USA). Sections were rinsed with PBS and blocked in a solution containing 5% normal serum and 0.2% Tween in phosphate buffered saline for 1 h at room temperature followed by incubation at 4 °C overnight with antibodies against LIF (ab113262, rabbit polyclonal, 1:500, ABCAM, Cambridge, UK) or LIF receptor (sc659, rabbit polyclonal, 1:200, Santa Cruz Biotechnology, Inc.) with ionized calcium binding adaptor molecule 1 (IBA-1, sc28530, goat polyclonal, 1:200, Santa Cruz Biotechnology, Inc.) or neuropeptide Y (NPY, sc133080, mouse monoclonal, 1:200, Santa Cruz Biotechnology, Inc.) or POMC (ab32893, goat polyclonal, 1:500, Abcam, Cambridge, UK) or glial fibrillary acidic protein (GFAP, ab4648, mouse monoclonal, 1:500, Abcam, Cambridge, UK) or alpha-MSH (AB5087, sheep polyclonal, 1:500, Millipore, Billerica, Massachusetts, USA) in a blocking buffer (1% bovine serum albumin in PBS-Tween). Next, sections were incubated for 2 h with goat anti-rabbit Cy3 (ab6941, 1:500, ABCAM, Cambridge, UK) or donkey anti-rabbit Alexa 546 and goat anti-mouse FITC or donkey anti-goat FITC. Nuclear staining was obtained using TO-PRO®-3 Iodide ((642/661) T3605, 1:1000, Life Technologies, Carlsbad, CA, EUA) in PBS. Analysis and documentation of the results were performed using a Leica TCS SP5 II confocal fluorescence microscope.

LIF and ^Tyr705phosphor-STAT3 protein expressions were determined in total protein extract samples by immunoblotting using a previously described method [13].

Statistics analysis. Results are presented as the mean ± standard error of the mean (SEM). For the comparison of means between two groups, we used Student’s t test for independent samples. Linear regression test was utilized to calculate kITT (based on the ITT test). The significance level was set at p < 0.05. Graph Pad Prism® was used to analyze the data.

cDNA obtained from the hypothalamus was employed to evaluate the expression of 84 chemokine-related transcripts in OP and OR mice fed on a HFD for 1 day. The complete list of transcripts evaluated in this study is shown in Additional file 1: Table S2. As a whole, ten transcripts were differentially expressed (Fig. 2a). However, most of them were regulated in the same direction in OP and OR mice. Thus, Ccbp2, Ccr1, Ccr4, Ccrl2, Ccr10, Il18, and Cxcr1 underwent an increase both in OP and OR, as compared to mice fed regular chow (Fig. 2a). Only three transcripts, Ccl20, Cxcl1, and LIF, were differentially regulated in OP and OR mice (Fig. 2a). For all those three transcripts, 1 day of feeding with the HFD was accompanied by an increase in OR and a reduction in OP mice, as compared to mice fed on chow (Fig. 2a). Because of the known involvement of LIF in the control of feeding and regulation of POMC expression in the hypothalamus [12], we decided to further explore the involvement of this protein in the different body compositions, metabolic, and inflammatory phenotypes observed in OP and OR mice fed a HFD. In order to confirm the findings of the real-time PCR array, LIF transcript expression was determined in the hypothalamus of control mice fed chow and in OP and OR mice fed the HFD for 1 day. As depicted in Fig. 2b, LIF transcripts were reduced in the hypothalamus of OP as compared to OR mice.

Immunofluorescence staining was employed to determine the anatomical and cellular distribution of LIF and LIF receptor (LIFR) in the hypothalamus of mice. As shown in Fig. 3a–d, most immunostaining of LIF and LIFR co-localize with NPY (Fig. 3a, b) and POMC (Fig. 3c, d). The LIF/LIFR-POMC co-localization was further confirmed by employing anti-alpha-MSH antibody (Additional file 1: Figure S1). Some cells expressing GFAP (astrocytes) also express LIF (Fig. 3e). Virtually no co-localization was detected for LIFR and GFAP (Fig. 3f). In addition, no localization was detected for either LIF or LIFR in cells immunostained with IBA1 (microglia) (Fig. 3g, h).

Because LIF expression was rapidly induced in the hypothalamus of OR mice fed a HFD, we hypothesized that it could play a protective role against the progression to diet-induced obesity. In order to test this hypothesis, mice were submitted to a protocol as depicted in Fig. 4a, which was designed to successfully immunoneutralize hypothalamic LIF, as shown by immunoblot (Fig. 4b); thereafter, a number of inflammatory and metabolic parameters were evaluated. The immunoneutralization of hypothalamic LIF resulted in an early increase in the expression of transcripts of inflammatory proteins, TNF-α (Fig. 4c), IL1β (Fig. 4d), Cx3cl1 (Fig. 4e), and Cxcl1 (Fig. 4f). This was accompanied by a trend to reduction of the transcripts of the anti-inflammatory proteins, IL6 (Fig. 4g), IL10 (Fig. 4h), and Ccl20 (Fig. 4i). There were no changes in the expression of transcript encoding for NPY and POMC (data not shown). The immunoneutralization of LIF in the hypothalamus resulted in reduced STAT3 Tyr705 phosphorylation in the hypothalamus but not in the hippocampus [14] (Additional file 1: Figure S2), suggesting that the method was appropriate for a site-specific action of the antibody.

The immunoneutralization of hypothalamic LIF in OR mice resulted in increased body mass gain over a period of 15 days (Fig. 5a, b). The change in body mass was mostly due to an increase in fat mass (Fig. 5c, d). To explore the mechanisms involved in the phenotypic change in OR mice, we evaluated caloric intake, energy expenditure and spontaneous physical activity. As shown in Fig. 5e, daily caloric intake was similar during most of the experimental period; only on the first day was there a significant increase in the caloric consumption of OR mice treated with the anti-LIF antibody. However, during the 15-day experimental period, there was a significant increase in cumulative caloric intake in OR mice treated with the anti-LIF antibody (Fig. 5f), which resulted in caloric intake similar to OP mice. The immunoneutralization of hypothalamic LIF promoted no changes in O₂ consumption, CO₂ production, or the respiratory quotient (Fig. 6a–f). However, OR mice treated with the anti-LIF antibody presented significantly reduced spontaneous physical activity during both the light and dark cycles (Fig. 6g).

We have previously shown that, in contrast to OP mice, glucose tolerance is preserved in OR mice fed an HFD [9]. Here, we evaluated glucose tolerance in OR mice treated with the anti-LIF antibody in the hypothalamus. As shown in Fig. 7a, the immunoneutralization of hypothalamic LIF resulted in higher blood glucose levels during a glucose tolerance test, leading to an increased area under the glucose curve (Fig. 7b). At least in part, the deterioration in glucose tolerance in OR mice treated with the anti-LIF antibody was due to the development of insulin resistance, as determined by an insulin-tolerance test (Fig. 7c) and the determination of the constant for glucose decay during the insulin tolerance test (kITT) (Fig. 7d).

In order to test the hypothesis that the OP-like phenotype developed by OR mice treated with the anti-LIF antibody was due to increased caloric intake (as shown in Fig. 5f), we performed an experiment in which OR mice treated with the anti-LIF antibody were fed a similar caloric amount consumed spontaneously by OR mice treated with IgG. As shown in Fig. 8a, pair feeding was sufficient to normalize body mass gain. This was accompanied by the restoration of glucose tolerance (Fig. 8b, c) and also by the restoration of insulin activity (Fig. 8d, e).