Protocols including manipulation of animals (registration number 853.01) were reviewed and approved by our local Institutional Animal Care and Use Committee (“Comité d'éthique en expérimentation animale n°105”), and were in strict accordance with the European Community guidelines (directive 86/906). Experiments were carried out with 2 months-old male C57Bl/6JOla mice from Harlan Laboratories. Mice were housed individually, fed either a standard diet (catalog number A04; Safe Laboratories; Augy, France; 3.3 kcal/kg; energy from carbohydrate/fat/protein: 72.4/8.4/19.3) or a customized highly palatable high-fat diet (catalog number U8954P V.7; Safe Laboratories; 4.4 kcal/kg; energy from carbohydrate/fat/protein: 40.8/42.9/16.3). Free access was given for food and water. Recordings of food intake and body weight were done manually at 9:00 h. For tissue collection, mice were killed between 14:00 and 16:00 h after a 6-h fast.

After sacrifice, brains were quickly removed and immersed in cold PBS solution. A 2 mm coronal slice was cut with a brain matrix. The slice was laid on a 6% agarose block and the MBH was dissected under stereomicroscope and cold-light illumination using a scalpel. Tissues were immediately frozen in liquid nitrogen and stored at −80°C until experiments.

Tissues were lysed and homogenized in RIPA lysis buffer using the TissueLyser system (Qiagen; Courtaboeuf, France) and 5 mm stainless steel beads (Qiagen). The homogenates were centrifuged 5 min at 5,000 g and supernatants were collected for PSA-NCAM assay using an enzyme-linked immunosorbent assay kit (PSA NCAM ELISA kit; Eurobio; Courtaboeuf, France) and for protein assay (Bio-Rad Protein Assay Kit ll; Biorad; Marnes-la-Coquette, France).

All data are expressed as means. Error bars indicate standard errors of the mean (SEM). Multiple comparisons between groups were carried out by one- or two-way ANOVA using Prism 5.0 software (GraphPad Software; San Diego, CA, United States). Post-hoc analyses were done when main effects reached significance. Before comparison, Bartlett's and Shapiro–Wilk's tests were applied to check equality of variances and to evaluate the normality of the distribution, respectively.

For a 2-month old mice fed with standard diet, the typical daily energy intake is 0.5 kcal per gram of body weight (The Jackson Laboratory, Mouse Phenome Database, http://phenome.jax.org/). It is well-known that high-energy foods cause transient overeating in most of the mice, which corresponds to an acute increase in energy intake during a few days (14–16), as illustrated in Figure S1A. In this study, the period of overconsumption was variable between mice and normalization of energy intake occurred after 2 days or more (Figure S1B; in kcal/gram of body weight). Consequently, the individual cumulative energy intake during 1-week HFD, hereinafter referred to as the “feeding response,” ranged from 3.96 to 6.92 kcal/g (Figure S1C). According to this response, we identified by a median split HFD-intolerant mice with high feeding response due to slow normalization, and HFD-tolerant mice with low feeding response due to fast normalization (Figures 1A,B). The two groups of mice were initially undistinguishable with respect to their energy intake on standard diet or to their initial body weight (Figures 1C,D). On HFD, tolerant mice normalized their energy intake in only 2 days after HFD introduction, whereas intolerant mice normalized it after 10 days (Figure 1E; in kcal/gram of body weight). After 2 weeks on HFD, energy intake was normalized for all mice. Nevertheless, intolerant mice kept on HFD for a long term had further episodic increases in raw energy intake which appeared after 5 weeks on HFD (Figure 1E; in kcal). Therefore, the cumulative energy intake over 3 months with HFD was significantly higher in intolerant mice than in tolerant mice (Figure 1F). Moreover, the longitudinal follow-up showed that body weight of intolerant mice maintained on HFD increased strongly after 8 weeks in comparison to tolerant mice (Figure 1G). Hence, the body weight gain between the two groups of mice was significantly different after 3 months on HFD (Figure 1H). Importantly, the terminal body weight gain was positively correlated with the feeding response to HFD (Figure 1I). By contrast, it was not correlated with the initial energy intake on standard diet or with the terminal energy intake on HFD (Figures 1J,K). These results indicated that the homeostatic feeding response to an acute HFD challenge in mice is a latent trait predictive of the propensity to gain weight under persistent caloric pressure.

Identification of individuals prone to obesity before the onset of the disease allows investigation of the biological underpinnings of susceptibility to obesity and makes possible the discovery of biological risk factors. Pre-existing differences in brain circuits controlling appetite and energy homeostasis might account for the vulnerability to common obesity (17, 18). Recent genetic studies formally implicated the brain in obesity pathology and pinpointed neuronal genes regulating synaptic plasticity (12, 19, 20). Moreover, rodent data consistently evidenced alterations of neural function and plasticity in brain feeding circuits in animals genetically predisposed to obesity (21, 22). Thus, we hypothesized that innate alteration in molecules involved in brain neuroplasticity would confer risk for obesity.

The polysialic acid-neural cell adhesion molecule (PSA-NCAM) is a cell-surface molecule that promotes various plasticity-related changes in brain circuits (23). We recently identified PSA-NCAM in the hypothalamus as a permissive factor for the brain control of energy balance (16, 24, 25). We therefore postulated that low level of hypothalamic PSA-NCAM would incite vulnerability to obesity. To test this hypothesis, we identified in a second cohort of mice (n = 14), HFD-tolerant and HFD-intolerant individuals according to their feeding response to 1-week HFD and then we examined the concentration of PSA-NCAM in the medio-basal hypothalamus (MBH) collected after this oral fat tolerance test. Intolerant mice showed significant lower levels of hypothalamic PSA-NCAM in comparison to tolerant mice (Figure 2; tolerant: 684.7 ± 59.4 ng/mg; intolerant: 505.9 ± 27.1 ng/mg; p = 0.038). As a reference, the constitutive level of PSA-NCAM in the MBH of control mice fed a standard diet was 508.5 ± 31 ng/mg protein (n = 16). The level of PSA-NCAM in the MBH of intolerant mice did not differ from the constitutive value. These data suggest that tolerant mice might have higher ability to mobilize PSA-NCAM to trigger synaptic plasticity in the hypothalamus than intolerant mice.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.