Orexin/hypocretin immunoreactivity in the lateral hypothalamus is reduced in genetically obese but not in diet-induced obese mice

The mechanisms that link diet and body weight are not fully understood. A diet high in fat often leads to obesity, and this in part is the consequence of diet-induced injury to specific hypothalamic nuclei. It has been suggested that a diet high in fat leads to cell loss in the lateral hypothalamus, which contains specific populations of neurones that are essential for regulating energy homoeostasis; however, we do not know which cell types are affected by the diet. We studied the possibility that high-fat diet leads to a reduction in orexin/hypocretin (O/H) and/or melanin-concentrating hormone (MCH) immunoreactivity in the lateral hypothalamus. We quantified immuno-labelled O/H and MCH cells in brain sections of mice fed a diet high in fat for up to 12 weeks starting at 4 weeks of age and found that this diet did not modify the number of O/H- or MCH-immunoreactive neurones. By contrast, there were fewer O/H- (but not MCH-) immunoreactive cells in geneticallyobese db/db mice compared to wild-type mice. Non-obese, heterozygous db/+ mice also had fewer O/H-immunoreactive cells. Differences in the number of O/H-immunoreactive cells were only a function of the db genotype but not of diet or body weight. Our findings show that the lateral hypothalamus is affected differently in genetic and in diet-induced obesity, and support the idea that hypothalamic neurones involved in energy balance regulation are not equally sensitive to the effects of diet.


Introduction
A diet high in fat often leads to obesity and type 2 diabetes in both humans and animals (Astrup et al., 2008) but the mechanisms that link diet and metabolic disease are not fully understood. One consequence of high-fat diet intake is damage to the hypothalamus, which is an essential component of the network that regulates appetite and metabolism in the body (Waterson & Horvath, 2015). For example, inflammation and leptin resistance in the arcuate nucleus (Münzberg et al., 2004;Thaler et al., 2012), and reactive astrogliosis in several hypothalamic nuclei (Buckman et al., 2013) have been observed in response to high-fat diet. In addition, Moraes et al. (2009) observed loss of neurones in the arcuate and in the lateral hypothalamus of animals fed a diet high in fat (although others have failed to see this effect; see Lemus et al., 2015;Namavar et al., 2012). Thus, it is possible that obesity and metabolic disease are the result of fat-induced hypothalamic injury that may include inflammation and neuronal loss (reviewed in Jais & Brüning, 2017).
Interestingly, much of the damage that is seen in the arcuate following high-fat diet seems to preferentially affect neurones involved in appetite regulation, namely orexigenic neurones that express neuropeptide Y and anorexigenic neurones that express the peptide pro-opiomelanocortin (Moraes et al., 2009;Horvath et al., 2010). In the lateral hypothalamus two neuronal populations that are modulated by nutrients and hormones (e.g. leptin and insulin) and that in turn regulate appetite, peripheral glucose, and adiposity express the peptides melanin-concentrating hormone (MCH)  Two animal models of obesity were used for these experiments: genetically obese db/db mice and dietinduced obese mice. Diet-induced obesity was set up with wild-type male mice (C57BL/6JOlaHsd, 3 weeks-old, = 24) from Harlan Laboratories (Blackthorn, UK), housed in 6 cages of 4 animals each with free access to water and standard chow (Teklad 2018, 18% cal from fat). After one week (i.e. at 4 weeks of age) food in 3 of these cages was replaced by a diet high in fat (Research Diets D12492, 60% cal from fat), thus creating 3 control (standard diet) and 3 experimental (high-fat diet, HFD) groups. Eight animals ( = 4 control and = 4 HFD) were taken at weeks 4, 8, or 12 after introducing the diet high in fat for the experiments described here.
Genetically-obese db/db mice develop severe obesity and diabetes (Hummel et al., 1966) because of a mutation in the leptin receptor which renders leptin signalling ineffective (Chen et al., 1996). Heterozygous (db/+) mice are not different physiologically or morphologically from wild-type mice (Hummel et al., 1966), and thus were used as controls. Male db/db (BKS.Cg-+Lepr db /+Lepr db /OlaHsd; = 8) and db/+ (BKS.Cg-Dock7 m +/+ Lepr db /OlaHsd; = 8) mice were obtained from Harlan Laboratories, UK, at 7 weeks of age and were housed with free access to water and standard diet (Teklad 2018, as above). They were taken at 8 ( = 4 from each group) or 14 ( = 4 from each group) weeks of age for the experiments described below.

Immunohistochemistry
Animals were deeply anaesthetised with pentobarbital sodium (20% solution, 0.1 ml i.p.) and then perfused transcardially with 20 ml phosphate-buffered saline (PBS) followed by 20 ml of paraformaldehyde (4% in PBS, pH 7.4). The brain was immediately removed and placed in paraformaldehyde (as above) for about two hours, then kept in sucrose solution (30% in PBS) for 48-72 h at 4°C until the brain precipitated to the bottom of the container. Afterwards, the brain was frozen and stored at −80°C until further processing.   Cell counts in the lateral hypothalamus. (A) Each brain was sliced rostro-caudally into 30 µm-thick coronal sections. (B) The first section collected for immunofluorescence was the one at the caudal end of the optic chiasm, at the point where the optic tract begins its dorso-lateral trajectory into the brain (B1;~0.94 mm caudal from bregma in Paxinos & Franklin, 2001). Then, every 6 th section (and its adjacent section) was collected in rostro-caudal sequence for a total of 2 sets of 6 sections per brain; thus, each set of brain sections were 180 µm apart, and the region studied spanned 1.1 mm, which covers the area of orexin/hypocretin (O/H) expression in the lateral hypothalamus. The area delimited in (B1) is the same as that in (C1, O/H). (C) Brain sections in (B) were processed for immunofluorescence to reveal O/H-like immunoreactivity (left column), and the sections immediately adjacent to each of these were processed for melanin-concentrating hormone (MCH)-like immunoreactivity (middle column). Pictures from both sides of the brain were obtained; here, each image consists of two micrographs per brain section stitched together to show the full extent of the area studied at each coronal plane. The diagrams (right column) highlight the main anatomical landmarks (3V, third ventricle; f, fornix; ic, internal capsule; LH, lateral hypothalamus; opt, optic tract; PVN, paraventricular nucleus; zi, zona incerta). (D) Immunoreactive cells in each brain section were counted and plotted, as shown here for one animal. Notice the correspondence between section number ( axis) and brain sections as depicted in ( was as reported previously (e.g. Broberger et al., 1998;Hahn, 2010). Also, in preliminary experiments we performed double immunofluorescence for both MCH and O/H and confirmed that there was no overlapping of these peptides, which was as expected because orexin and MCH are two different cell populations (Broberger et al., 1998). The findings reported below were the result of single immunofluorescence for O/H or MCH in adjacent brain sections.

Data collection and analysis
Images were obtained with an epifluorescence microscope (Leica DM4000B; Leica Microsystems, Wetzlar, Germany) fitted with a 5×/0.15 objective for counting ir cells (as in Fig. 1C) or a 1.25×/0.04 objective for full-section images (as in Fig. 1B). The latter were used to confirm that all sections were anatomically equivalent across brains at each of the six rostrocaudal levels per brain studied. Recall that the resolving power of a lens is = /2NA, where is the wavelength of light and NA is the numerical aperture. Thus, with e.g. red light ( ≈ 600 nm) the 5× objective (NA = 0.15) that was used for counting cells can resolve up to 2 µm, which is more than enough to allow for the identification of individual O/H-or MCH-ir neurones (Fig. 1E). For counting, two images per brain section were captured, one for each side of the brain (Fig. 1C); thus, the number of immunoreactive cells per section reported here represent the total found in the full coronal plane, and it was unnecessary to align the brains along the sagittal plane because we did not analyse the medio-lateral distribution of these cells. We used ImageJ (Schneider et al., 2012) with the plugin PointPicker (http: //bigwww.epfl.ch/thevenaz/pointpicker/) to identify and count cells, and during this process the investigator was blind to the experimental condition of the images.
Statistical analyses were performed using R version 3.3.2 (R Core Team, 2016). Body weight in wildtype animals ( Fig. 2A) was analysed independently of that in db/db mice (Fig. 2B)

Animal models of obesity
To

MCH-and O/H-ir cell counts in diet-induced obese mice
We counted MCH-and O/H-ir cells in the hypothalamus of wild-type animals that received standard diet or a diet high in fat for 4, 8, or 12 weeks starting at 4 weeks of age. Cell counts were from 6 coronal sections (numbered 1 to 6 in rostro-caudal sequence; see time the animals were exposed to the diet. Because the diet was introduced at 4 weeks of age in all experimental animals, the time these animals were exposed to high-fat diet was considered to be represented by their age. We found that diet did not affect the number of MCH-ir cells, but there was a significant and unexpected difference in the number of MCH cells with age (diet, (1, 20) = 2.34, = 0.142; age, (2, 20) =

Figure 2
Body weight in animal models of obesity. (A) Wild-type mice were fed a diet high in fat starting at 4 weeks of age. Four weeks later, at age 8 weeks, their body weight was not significantly different from that seen in animals eating a standard diet. However, the animals that received the diet high in fat for 8 or 12 weeks (i.e. until 12 or 16 weeks of age) gained significantly more weight than control animals. (B) Genetically-obese db/db mice were significantly heavier than lean heterozygous (db/+) mice. n.s., not significant; * = 0.039; ** ≪ 0.001. Each bar represents mean ± sem of = 4 animals.

To investigate if obesity affected MCH and/or O/H cells independently of diet we counted MCH-and
O/H-ir neurones in obese db/db mice and in lean db/+ control mice 8 or 14 weeks old (Fig. 4A) following the same procedure described above (Fig. 1).
We did not observe any differences in MCH-ir cell numbers in these mice at the ages studied (strain, (1, 13) = 1.107, = 0.312; age, (1, 13) = 1.587, = 0.230, two-way ANOVA; Fig. 4B (González et al., 2016) which shows that the number of orexin cells directly affects body weight, but it is not known if such relationship holds true for less-extreme cases. Here, strain was necessarily included as a co-factor because of the strong link between body weight and strain (Fig. 2).

O/H-ir cell counts in db/db vs wild-type mice
As illustrated in Fig. 5F, there was a marked effect of strain on O/H cell numbers (which was already de-

Discussion
Diet has direct implications for body weight, yet the links between the two are still not fully understood.
Recent studies have demonstrated that a diet high in fat triggers histopathological changes in the arcuate nucleus of the hypothalamus that include inflammation, gliosis, apoptosis, and synaptic reorganization (Moraes et al., 2009;Thaler et al., 2012;Buckman et al., 2013;Horvath et al., 2010). These changes seem to predominantly affect specific cell types directly involved in energy homoeostasis and are thought to be a leading cause of metabolic imbalance and obesity (see Jais & Brüning, 2017, for a recent review).
Apoptosis was also observed in the lateral hypothalamus in response to high-fat diet (Moraes et al., 2009), but we do not know which cell types are af- fected. Two cell populations in the lateral hypothalamus, orexin/hypocretin and melanin-concentrating hormone neurones, are essential for appetite and metabolism regulation (Burdakov et al., 2013 (Fig. 3) despite the marked difference in body weight between these groups (Fig. 2). These results are in agreement with those by Lemus et al. (2015), who also found that high-fat diet did not affect the number of O/H cells in mice (they did not quantify MCH cells), and more generally with those by Namavar et al. (2012), who did not see a change in total neurones in the hypothalamus after high-fat diet. By contrast, we found that genetically-obese db/db mice had significantly fewer O/H-ir (but not MCH-ir) cells (Fig. 5). Lean heterozygous db/+ animals also had fewer O/H-ir cells than wild-type mice but the effect was less marked.
We counted immuno-labelled cells in the lateral hypothalamus ensuring that brain sections were anatomically equivalent between subjects (Fig. 1).
Whereas within-subject brain sections were far enough apart from each other to avoid double counting, at the same time our experimental design guaranteed that a comprehensive set of MCH-and O/Hir cell populations was sampled along the full rostrocaudal range. We also ensured that we followed a careful quantitative analysis of our data. We note that Yamamoto et al. (1999)  tor that makes cells insensitive to leptin (Lee et al., 1996), whereas in the latter leptin resistance is presumably the result of high-fat diet-induced hypothalamic injury (Münzberg et al., 2004).  (Beck et al., 2006;Chang et al., 2008), suggesting that these peptidergic systems are sensitive to changes in diet before birth.
Differences between diet-induced and geneticallyinduced obese mice may also be explained by the extent of the leptin resistance because it has been suggested that leptin resistance in animals that are fed a diet high in fat develops only in the arcuate nucleus and not in other hypothalamic areas (Münzberg et al., 2004), whereas leptin-signalling defects in db/db mice are not region specific. It is also possible that the observed differences in O/H-ir between animal models were due to a more general effect on brain development because the brains of leptin-resistant db/db mice and of leptin-deficient ob/ob mice are signifi-cantly smaller than expected (Vannucci et al., 1997;Steppan & Swick, 1999). Leptin replacement in the latter seems to directly increase the number of brain cells (Steppan & Swick, 1999). However, two obser- anatomically from wild-type mice (Hummel et al., 1966), yet in our studies db/+ mice also had fewer O/H-ir neurones than wild-type mice (Fig. 5).

Conclusions
We found fewer hypothalamic O/H-ir cells in db/db and db/+ mice than in wild-type mice, and this effect was only explained by the db genotype and not by high-fat diet intake, age, or body weight. The number of MCH-ir cells was not affected by genotype or diet, but we saw a moderate and transient increase in these at 12 weeks of age. Our findings show that lateral hypothalamic neurones that are crucial for the mainte-nance of energy homoeostasis are affected differently in diet-induced obese animals and in mice that are genetically obese, and support the notion that highfat diet intake does not affect equally all components of the network that keeps metabolism in tune.

Author contributions
All authors had full access to all the data in the study.