The hypothalamic arcuate nucleus (ARC) contains 2 key neural populations, neuropeptide Y (NPY) and proopiomelanocortin (POMC), and, together with orexin neurons in the lateral hypothalamus, plays an integral role in energy homeostasis. However, no studies have examined total neuronal number and volume after high-fat diet (HFD) exposure using sophisticated stereology. We used design-based stereology to estimate NPY and POMC neuronal number and volume, as well as glial fibrillary acidic protein (astrocyte marker) and ionized calcium-binding adapter molecule 1 (microglia marker) cell number in the ARC; as well as orexin neurons in the lateral hypothalamus. Stereological analysis indicated approximately 8000 NPY and approximately 9000 POMC neurons in the ARC, and approximately 7500 orexin neurons in the lateral hypothalamus. HFD exposure did not affect total neuronal number in any population. However, HFD significantly increased average NPY cell volume and affected NPY and POMC cell volume distribution. HFD reduced orexin cell volume but had a bimodal effect on volume distribution with increased cells at relatively small volumes and decreased cells with relatively large volumes. ARC glial fibrillary acidic protein cells increased after 2 months on a HFD, although no significant difference after 6 months on chow diet or HFD was observed. No differences in ARC ionized calcium-binding adapter molecule 1 cell number were observed in any group. Thus, HFD affects ARC NPY or POMC neuronal cell volume number not cell number. Our results demonstrate the importance of stereology to perform robust unbiased analysis of cell number and volume. These data should be an empirical baseline reference to which future studies are compared.
Control of food intake and energy metabolism relies upon a complex interplay between neural networks in the central nervous system and peripheral tissues. Within the central nervous system, energy homeostasis is largely controlled by a fine balance between orexigenic and anorexigenic neuropeptides in the hypothalamic arcuate nucleus (ARC). Neuropeptide Y (NPY) and Agouti-related protein (AgRP) are coexpressed in neurons of the ARC and are potent orexigenic peptides, whereas proopiomelanocortin (POMC) precursor protein in the ARC is cleaved into potent anorexigenic α-melanocyte-stimulating hormone peptides. NPY/AgRP and POMC neurons in the ARC are considered “first-order” sensory neurons in the control of energy homeostasis (1) and receive, coordinate, and respond to changes in nutrient and hormonal fluctuations associated with changes in metabolic status. NPY acts on Y1 and Y5 receptors in the paraventricular nucleus of the hypothalamus to stimulate feeding, whereas acetylated α-melanocyte-stimulating hormone and AgRP peptides act as agonists and antagonists, respectively, at the melanocortin 4 receptor (MC4R). The POMC and AgRP interaction at the MC4R is collectively known as the melanocortin circuit. The critical importance of the melanocortin system in food intake and energy balance is highlighted by conditional gene ablation experiments. Ablation of POMC neurons in adulthood produced an increase in food intake and body weight and caused glucose intolerance (2, 3), whereas ablation of AgRP resulted in rapid hypophagia, weight loss, and starvation (2, 4). Moreover, POMC and MC4R knockout mice develop obesity by 8 weeks of age on a normal chow diet (5, 6).
It is generally accepted that high-fat diet (HFD) suppresses hypothalamic NPY mRNA expression in mice and rats (7–10) and NPY peptide release (11, 12). However, this is not a consistent effect, because a number of older studies show this depends on length of diet exposure or show that HFD actually increases NPY mRNA expression (13–15). Studies show that HFD consumption has either no effect (16), reduces (17), or increases POMC mRNA expression (8, 10, 18). Consistent with increased POMC mRNA, melanocortin agonists produce a greater reduction in feeding in HFD-fed rats compared with chow-fed controls (19). Moreover, rats selectively bred to develop diet-induced obesity show impaired ARC projections in adulthood and reduced AgRP and POMC fiber density in the paraventricular nucleus (20). Recent studies also propose that diet-induced obesity increases apoptosis in arcuate AgRP and POMC neurons (21) and decreases the neurogenic turnover of hypothalamic arcuate neurons (22, 23). Collectively these data show that HFD affects NPY and POMC mRNA and may affect NPY or POMC cell number, although this remains to be examined.
Despite the importance of arcuate POMC and NPY/AgRP neurons to overall energy homeostasis (24, 25), no study has addressed the fundamental impact of diet on arcuate POMC and NPY/AgRP neuronal number and volume in a systematic and unbiased manner. Moreover recent evidence suggests that HFD exposure affects the number of glial fibrillary acidic protein (GFAP) and/or ARC ionized calcium-binding adapter molecule 1 (Iba1) cells in the ARC, markers of astrocytes and microglia, respectively (26, 27). The increase in GFAP cell number in HFD-fed mice correlates with altered synaptic input organization and is hypothesized to contribute to hormone resistance in diet-induced obese mice (28). The emerging importance of hypothalamic inflammation on diet-induced obesity necessitates a robust stereological analysis of cell number.
Orexin neurons located in lateral hypothalamus are also important in the control of arousal and energy homeostasis. In terms of energy homeostasis, orexin neurons control brown fat development and function (29), glucose use in skeletal muscle (30), and energy expenditure (31). There are currently no robust unbiased stereological data describing the effects of HFD on this key neuronal population. In this study, we examined the effect of chow or high-fat feeding for 2 or 6 months on NPY, POMC, GFAP and Iba1 cell number in the ARC and orexin neurons in the lateral hypothalamus, assessed using the stereologic optical fractionator probe, or NPY, POMC and orexin cell volume, assessed using the stereologic nucleator probe. We hypothesized that HFD exposure would significantly affect neuronal number and volume of ARC NPY and POMC neurons, as well as GFAP and Iba1 cells in the ARC and orexin neurons in the lateral hypothalamus.
Materials and Methods
Animals
Male NPY GFP mice (B6.FVB-Tg(Npy-hrGFP)1Lowl/J; stock number 006417, 8 wk old; The Jackson Laboratory) were group housed (2–4 per cage) under controlled conditions (21°C and a 12-hour light, 12-hour dark cycle). Mice received either a chow diet (8720610; Barastoc Stockfeeds), which contained 22% protein, 9% total fat, 3.2% crude fiber, 4.4% acid digested fiber, and 13.2 MJ/kg of digestible energy, or a HFD (SF04–001; Specialty Feeds) containing 22.6% protein, 23.50% total fat, 5.4% crude fiber, 5.4 acid digested fiber, and 19 MJ/kg of digestible energy. Mice on each diet were split into 2 groups: 2- or 6-month ad libitum access to food and water. Mice were 8 weeks old at the beginning of the experiment, therefore experimental mice on a diet for 2 months were 16 weeks of age, and experimental mice on a diet for 6 months were 32 weeks of age. This allowed us to examine not only the effect of diet but also the effect on age on NPY and POMC neuronal number and volume. There were 8 mice per group except for the 6-month HFD group (n = 7). Experiments were conducted in accordance with the Monash University Animal Ethics Committee guidelines.
Glucose tolerance test (GTT) and insulin tolerance test (ITT)
For ip GTTs and ITTs, mice were fasted from 8am in the morning and ip GTT or ITTs were performed 5 hours later at 1 pm. At this time, a tail tip baseline blood sample was taken from fasted mice (5 h). Intraperitoneal GTTs were performed at 2 months (7–8 wk) or 6 months (23–24 wk) after diet exposure. Blood glucose concentration was immediately measured with ACCU-CHEK Active (Roche Diagnostics GmbH), after which D-glucose (50% solution; 2 g/kg, ip) or insulin (0.5 U/kg, ip) was injected. Additional samples for blood glucose were taken at 15, 30, 60, and 90 minutes after injection.
Tissue collection
Animals were deeply anesthetized with isoflurane, and blood was collected immediately before perfusion via cardiac puncture. The gastrocnemius muscle, liver, and epididymal fat pads were also collected before perfusion in order to verify the effect of HFD on metabolic parameters. Plasma collected from blood samples was stored at −20°C, and samples collected were snap frozen in liquid nitrogen. Mice were then perfused with 0.05M PBS, followed by 4% paraformaldehyde. Brains were collected, postfixed in 4% paraformaldehyde overnight at 4°C, then placed in 30% sucrose.
Immunohistochemistry
Brains were cut at 30 μm on a cryostat, and every 4th section through the entire brain and brainstem was collected and stored in cryoprotectant at −20°C. Brains were analyzed for POMC and NPY GFP expression in the ARC. Sections were washed in 0.1M PB and incubated with 1% hydrogen peroxide (H2O2) for 15 minutes to prevent endogenous peroxidase activity and blocked for 1 hour with 5% normal horse serum (NHS) in 0.3% Triton 0.1M PB. Sections were incubated with POMC rabbit antibody (H02930; Phoenix Pharmaceuticals), goat antiorexin A antibody (catalog number sc-807; Santa Cruz Biotechnology, Inc), GFAP rabbit antibody (catalog number 7260; Abcam), and Iba1 (catalog number 019–19741; Wako) all at 1:1000 in diluent of 1% NHS in 0.3% Triton in 0.1M PB. For POMC staining, after incubation, the sections were washed and incubated with biotin-SP-conjugated goat antirabbit IgG (Jackson ImmunoResearch) at 1:200 in 0.1M PB. Sections were washed and incubated with avidin-biotin complex (1:200) for 90 minutes. To visualize immunoreactivity, sections were incubated with a solution containing 1% nickel ammonium sulfate, 1% diaminobenzidine, and 4 μL of 30% H2O2. Sections were then washed in 0.1M PB, mounted, and coverslipped. The nonspecific staining of our antibodies used was confirmed with no primary antibody controls (incubated in diluent that consisted of 1% NHS and 0.3% Triton in 0.1M PB) and after preabsorption with POMC peptide (Propiomelanocortin Precursor [27–52], catalog number 029–30 30; Phoenix Pharmaceuticals). For orexin, GFAP, and Iba1 staining, sections were washed after the primary antibody step and incubated with Alexa Fluor goat antirabbit 594 antibody or donkey antigoat 594. Due to the high degree of GFP photostability in the NPY GFP mice, there was no need to conduct fluorescent immunohistochemistry to visualize and analyze NPY neurons. NPY neurons were analyzed based on endogenous hrGFP fluorescence.
Stereology
A design-based unbiased stereology method was used to quantify NPY and POMC neurons in the ARC. We used StereoInvestigator software (MicroBrightField) to estimate neuronal number (using the optical fractionator probe) and neuron somal volume (using the nucleator probe). Cells were visualized using a Zeiss microscope with a motorized stage coupled with a MicroFibre digital camera to a computer. Systematic sampling of every 4th section was collected through the ARC beginning at approximately bregma −1.22 mm and finishing at approximately −2.54 mm, with the first sampled set of sections chosen at random. Sections were cut at 30 μm to allow for a 20-μm optical dissector within each section after dehydration and mounting. Cells were counted using a randomly positioned grid system controlled by StereoInvestigator in a previously defined region in all optical planes, thus creating a 3-dimensional counting area. Guard zones were set at 10% of the section thickness to account for damage during the staining procedure and to prevent overcounting. The counting frame width (X) and height (Y) was 40.2 μm producing a counting frame area (XY) of 1616 μm2. The dissector height (Z) was 20 μm creating a dissector volume (XYZ) of 32 320 μm3. With this counting frame area we discovered that we needed to sample approximately 150–200 sites throughout the entire ARC and count approximately 160–200 neurons throughout the entire ARC to give an acceptable coefficient of error (CE) (using the Gunderson method) of 0.1 using the smoothness factor m = 1. The CE provides a means to estimate sampling precision, which is independent of natural biological variance. As the value approaches 0, the uncertainty in the estimate precision reduces. CE = 0.1 is deemed acceptable within the field of stereology. Cells were only counted if they touched the inclusion border or did not touch the exclusion border of the sampling grid.
We also used an additional stereological probe, the nucleator method (32), to measure cell volume at the same time as we used the optical fractionator to estimate cell number. Thus, for every POMC and NPY cell that was counted, the volume was also measured. The nucleator uses orthogonal lines generated at the midpoint of the cell, in this instance, we used the center of the nucleus as a common point. The cell body can be identified where the lines intersect with the boundary of the cell. Using the distance measured between the nucleus and the cell boundary, total cell volume is calculated by taking the third powers of these measurements. To ensure unbiased sampling of neuronal volume, this method requires uniformly random sections or vertical uniformly random sections. In practical neuroscience it is difficult to meet these requirements as neuronal sections were cut in a coronal plane to allow for light and electron microscopy. Although this created a bias, as published elsewhere (33), we sampled from a large number of neurons in each treatment and believe we have a robust, albeit biased, estimate of NPY and POMC cell volume in the ARC. CE < 0.01 reflects the robust nature of our sampling precision.
Analysis of blood chemistry
Plasma Insulin concentration was determined through an in-house ELISA. Plasma nonesterified fatty acid concentration was measured using a Nonesterified fatty acids (NEFA) C Assay kit (Wako Pure Chemical Industries, Ltd) according to manufacturer’s instructions.
Plasma, muscle, and liver triglycerides
Triglycerides in plasma and tissue were determined with triglyceride Assay kit (Roche/Hitachi, Roche Diagnostics GmbH). Crushed Tissue was incubated with a chloroform:methanol (2:1) solution overnight. Pure chloroform and 0.9% NaCl were added and incubated for 10 minutes on an orbital shaker, followed by a 10-minute centrifugation step at 2000 rpm. The chloroform phase was transferred to fresh glass tubes and evaporated to complete dryness under N2 at 40°C. The dried extract was dissolved in absolute ethanol for the triglyceride assay.
Liver and muscle glycogen
Liver and muscle glycogen concentrations were determined with glucose oxidase assay kit (Sigma-Aldrich) after glucose extraction. Briefly, the ground frozen tissue was digested in KOH at 70°C for 20 minutes. Saturated NaSO4 and 95% ethanol were added, mixed, and centrifuged for 10 minutes. The pellet was resuspended in deionized water and digested at 70°C for 10 minutes, before adding 95% ethanol, mixing and centrifuging. The precipitate was resuspended in amyloglucosidase buffer (0.3-mg/mL amyloglucosidase in 0.25M acetate) and incubated at 37°C overnight.
Statistical analysis
All data are presented as mean ± SEM. Repeated measures two-way ANOVA or standard two-way ANOVA (2 variables were diet and time) with Tukey’s multiple comparison post hoc tests and unpaired Student’s t tests were used to determine statistical significance (set at P < .05) between groups using GraphPad Prism software.
Results
HFD exposure causes metabolic dysfunction
In order to examine the effect of diet-induced obesity on ARC NPY and POMC neuronal number and volume, we assessed a number of metabolic parameters after 2 and 6 months of chow diet or HFD exposure. Mice on a HFD for 2 months gained significantly more body weight than chow-fed controls (Figure 1A), and this was associated with hyperinsulinemia. No differences in plasma NEFAs or plasma triglycerides were observed (Figure 1, B–D). Two-month HFD-fed mice were glucose intolerant but remained insulin sensitive, when compared with chow-fed control mice (Figure 1, E–H). Finally, 2 months of HFD feeding increased hepatic triglyceride content but not muscle triglyceride content when compared with chow-fed mice. We observed no differences in liver or muscle glycogen concentrations between chow-fed and HFD-fed mice (Figure 1, K and L).
HFD consumption for 2 months causes metabolic dysfunction. A, Mice on a HFD gain significantly more body weight over the 8-week period. HFD increases plasma insulin (B) but not plasma NEFA (C) or plasma triglycerides (D). HFD mice exhibit greater glucose excursions (E) and greater area under the curve (AUC) (F) compared with chow mice, indicating impaired glucose tolerance. No difference in insulin sensitivity, as assessed by ITTs (G) and AUC analysis (H). Two-month HFD exposure increased liver triglycerides (I) but not muscle triglycerides (J) and had no effect on liver glycogen (K) or muscle glycogen (L). All data are presented as mean ± SEM (n = 8/group). We used two-way repeated measures ANOVA followed by Sidak’s post hoc test or Student’s t tests to determine statistical significance with GraphPad prism software. *, P < .05; **, P < .01; ***, P < .001; ****, P < .0001.
After 6 months on a HFD, mice gained significantly more weight and exhibited hyperinsulinemia compared with chow-fed controls (Figure 2, A and B). Plasma triglycerides and NEFAs were not significantly different between chow and HFD-fed mice (Figure 2, C and D), highlighting that these parameters are not good indices of metabolic dysfunction after HFD exposure. Glucose and ITTs show that 6-month HFD exposure causes significant glucose intolerance and insulin resistance (Figure 2, E–H). Moreover, these results show that insulin resistance develops with progressive exposure to HFD, because no insulin resistance was observed in mice fed HFD for 2 months. Liver triglycerides but not muscle triglycerides were significantly elevated in 6-month HFD-fed mice compared with chow controls (Figure 2, I and J), and we observed no difference in liver glycogen or muscle glycogen, consistent with results from mice fed HFD for 2 months. These results serve to highlight diet-induced metabolic dysfunction and strengthen our conclusions on the stereological investigation of NPY and POMC cell number and volume.
HFD consumption for 6 months causes metabolic dysfunction. A, Mice on a HFD gain significantly more body weight over the 6-month experimental period. HFD increases plasma insulin (B) but not plasma NEFA (D) or plasma triglycerides (C). HFD mice exhibit greater glucose excursions (E) and greater AUC (F) compared with chow mice, indicating impaired glucose tolerance. HFD mice exhibit relative insulin resistance compared with chow-fed mice, as assessed by ITTs (G) and AUC analysis (H). Six-month HFD exposure increased liver triglycerides (I) but not muscle triglycerides (J) and had no effect on liver glycogen (K) or muscle glycogen (L). All data are presented as mean ± SEM (n = 7 HFD group; n = 8 chow group). We used two-way repeated measures ANOVA followed by Sidak’s post hoc test or Student’s t tests to determine statistical significance with GraphPad prism software. *, P < .05; **, P < .01; ***, P < .001; ****, P < .0001.
ARC NPY cell number and cell volume
Stereological counts of total NPY neurons using the optical fractionator revealed that the ARC contains approximately 8000 NPY neurons throughout the rostral-caudal ARC axis. Total ARC NPY neurons did not differ after 2 or 6 months on a chow diet or HFD (Figure 3A). To examine the distribution of NPY neuron number through the rostro-caudal ARC, we measure the average NPY neurons number per serial section analyzed. Our data show a significant main effect of section number, indicating that NPY neuron number is significantly differently through the ARC nucleus (Figure 3B). NPY neuron number was highest at section 9, which corresponds approximately to bregma position −1.50 mm (34). However, there was no effect of diet on the distribution on NPY neuronal number throughout the ARC (Figure 3B).
The effect of HFD on ARC NPY cell number and volume. A, HFD exposure for either 2 or 6 months does not affect total ARC NPY cell number compared with chow-fed mice. B, Analysis of average NPY neuron number through the rostro-caudal ARC indicates a significant main effect of section number. C, Length of diet exposure significantly increases average ARC NPY cell volume (main effect Ptime). D, NPY cell volume distribution analysis indicates the number of NPY cells within 100-μm3 cell volume bins. The length of HFD exposure significantly reduces the number of NPY cells with low cell volumes ranging between 201 and 500 μm3. D, Representative images showing NPY GFP (green) neurons in the rostral, mid, and caudal ARC. Sections countered stained with DAPI. All data are presented as mean ± SEM (n = 8/group except HFD 6 mo, where n = 7). We used two-way ANOVA followed by Tukey’s multiple comparisons with GraphPad prism software. Significant main effects from two-way ANOVA are shown on graph. a, Significant with respect to 2-month chow. b, Significant with respect to 2-month HFD. Scale bars, 100 μm.
In order to estimate total NPY cell volume, we used the nucleator probe. We sampled on average approximately 300 NPY neurons throughout the rostral-caudal ARC axis from each mouse on a chow diet or HFD for 2 or 6 months. The average cell volume per animal was calculated from the approximately 300 sampled neurons to generate an average cell volume per animal (n = 7–8/group). Our results show a main effect of time such that total average NPY volume increased with time independent of diet (Figure 3C). There was a significant increase in total NPY cell volume after 6 months of HFD relative to 2 months on a chow diet and a trend for an increase (P = .0585) in NPY cell volume for 2 vs 6 months on a HFD (Figure 3C). In order to look more closely at the effect of HFD on NPY cell volume, we performed a distribution analysis by examining the number of NPY cells per binned 100-μm3 cell volume. We observed a large range of cell volumes from less than 100 μm3 to more than 1500 μm3, as shown by the distribution of NPY cell volume (Figure 3D). The cell volume distribution revealed that there were significantly fewer NPY neurons ranging between 200 and 500 μm3 in mice fed a chow diet or HFD for 6 months compared with 2 months (Figure 3D). Moreover, there was a trend for 6-month HFD vs chow diet exposure to reduce the number of NPY cells with cell volumes ranging from 301 to 400 μm3 and 401 to 500 μm3 (P = .15 and P = .078, respectively).
ARC POMC cell number and cell volume
Our stereological analysis of POMC neuronal number estimates that the ARC contains approximately 9000 POMC neurons in total. We did not observe any differences in total ARC POMC neuronal number mice fed a chow diet or HFD for 2 or 6 months (Figure 4A). To examine the distribution of POMC neurons through the rostro-caudal ARC, we estimated the average number of POMC neurons from each counted serial section. We observed a significant main effect of section number, indicating that the distribution of POMC neurons changes according to serial section number, with the highest number of POMC neurons per section from mid to caudal ARC. However, there was no effect of diet on the distribution on POMC neuronal number throughout the ARC (Figure 4B).
The effect of HFD on ARC POMC cell number and volume. A, HFD exposure for either 2 or 6 months does not affect total ARC POMC cell number. B, Analysis of average POMC neuron number through the rostro-caudal ARC indicates a significant main effect of section number. C, HFD exposure for either 2 or 6 months does not affect average POMC cell volume. D, POMC cell volume distribution analysis indicates the number of POMC cells within 100-μm3 cell volume bins. Both HFD and the length of HFD exposure significantly reduce the number of POMC cells with low cell volumes ranging between 201 and 700 μm3. E, Representative images of POMC immunoreactive neurons in the rostral, mid, and caudal ARC. All data are presented as mean ± SEM (n = 8/group except HFD 6 mo, where n = 7). We used two-way ANOVA followed by Tukey’s multiple comparisons with GraphPad prism software. Significant main effects from two-way ANOVA are shown on graph. a, Significant with respect to 2-month chow. Scale bars, 100 μm.
In order to estimate total POMC cell volume, we sampled on average approximately 120 POMC neurons throughout the rostral-caudal ARC axis from mice on a chow diet or HFD for 2 or 6 months. The average cell volume per animal was calculated from approximately 120 sampled neurons to generate an average cell volume per animal (n = 7–8/group). Again, there were no significant differences in average POMC cell volume between mice on a chow diet or HFD for 2 or 6 months (Figure 4C). Despite no differences in average POMC cell volume, volume distribution analysis revealed a significant reduction in POMC cell number at small cell volumes, ranging from 201 to 700 μm3. The most striking differences were observed between 301 and 500 μm3, where there was a significant reduction POMC cell number at 2-month HFD, 6-month chow diet, and 6-month HFD compared with 2-month chow diet (Figure 4D). Taken together, these results indicate that although average POMC neuronal volume does not change with the length of chow diet or HFD exposure, both HFD exposure and length of diet exposure alters the normal distribution of POMC cell volumes.
ARC GFAP and Iba1 cell number
For GFAP cell number in the ARC, we observed a significant main effect of diet, such that HFD exposure increased GFAP cell number compared with chow diet (Figure 5A) independent from the length of time on the diet. Two-month HFD exposure increased the total number of ARC GFAP cells compared with chow mice (1803 ± 204 chow diet vs 2894 ± 362 HFD) (Figure 5A), although no significant difference was observed between chow and HFD groups after a 6-month exposure. An analysis of GFAP cell distribution through the ARC nucleus revealed no significant differences (Figure 5B).
The effect of HFD on ARC GFAP and Iba1 cell number. A, Two-month HFD exposure increases ARC GFAP cell number compared with chow-fed mice, whereas 6-month HFD exposure had no significant effect on average ARC GFAP cell number compared with chow-fed mice. B, An analysis of average GFAP cell number per serial section counted, revealed no differences in GFAP cell number throughout the rostro-caudal ARC nucleus. C, No difference in ARC Iba1 after 2 or 6 months on HFD, although analysis of Iba1 cells per section revealed a significant main effect, indicating an difference in Iba1 cells across the rostro-caudal ARC nucleus. Representative images of GFAP (E) and Iba1 (F) immunoreactive cells in 2-month chow and HF-fed mice and 6-month chow-fed mice. We used two-way ANOVA followed by Tukey’s multiple comparisons with GraphPad prism software. Significant main effects from two-way ANOVA are shown on graph. a, Significant with respect to 2-month chow. Scale bars, 100 μm.
We observed no differences in diet exposure or age in total Iba1 cell exposure (Figure 5C) although the distribution of Iba1 cells in the ARC showed a significant increase in Iba1 cells per section in the caudal ARC (significant main effect) (Figure 5D).
Number of orexin neurons in the lateral hypothalamus
There were approximately 7000–8000 total orexin neurons in the lateral hypothalamus and HFD exposure trended to increase cell number (significant main effect; P = .0578) independent of length of time on the diet (Figure 6A). We also observed a significant difference in the distribution of orexin neurons through the rostro-caudal lateral hypothalamus with fewer neurons counted per section in the caudal region of the lateral hypothalamus (Figure 6B). The average orexin cell volume was significantly reduced with exposure to HFD independent of the length of time on the diet (significant main effect of diet) (Figure 6C). Cell volume distribution analysis revealed bimodal effect of HFD on cell volume, because we observed that HFD increased the number of cells with relatively small volumes (750–1750 μm3) (Figure 6D). However, HFD decreased the number of cells with relatively large volumes (2250–2750 μm3), and therefore, it is the reduced number of cells with large volumes that account for the significant reduction in average cell volume observed in Figure 6C.
The effect of HFD of orexin neurons in the lateral hypothalamic area. A, Neither 2- nor 6-month HFD exposure increased total orexin neurons in the lateral hypothalamic area, although there is a trend for diet to increase cell number irrespecific time (main effect of diet P = .0578). B, An analysis of average orexin neuron number per serial section counted revealed a significant difference throughout the rostro-caudal ARC nucleus (significant main effect of section number), with fewer orexin cells at the caudal end of the ARC nucleus. C, There was a significant main effect of diet, indicating HFD significantly reduced orexin cell volume independent of length of time on diet. D, Orexin cell volume distribution analysis indicates that HFD exposure significantly increases the number of cells within the 750- to 1750-μm3 range whereas decreasing the number of orexin neurons in the 2250- to 2750-μm3 range. We used two-way ANOVA followed by Tukey’s multiple comparisons with GraphPad prism software. Significant main effects from two-way ANOVA are shown on graph. a, Significant with respect to 2-month chow. b, Significant with respect to 2-month HFD. c, Significant with respect to 6-month chow. MCLH represents the magnocellular part of the lateral hypothalamus according to Franklin and Paxinos (34). Scale bars, 50 μm.
Discussion
Stereology is a methodology that quantifies properties of three-dimensional objects, such as neuron number or volume, in two-dimensional sections. These measurements are unbiased based on proven mathematical formulae, rigorous counting rules, geometric overlays, and section guard zones to prevent over counting (35). The use of systematic random sampling of the entire area of interest is crucial (36, 37), because it prevents conclusions based on subjective sampling measures, which may unwittingly be biased or underpowered. In this study, we have counted POMC, NPY, orexin, GFAP, and Iba1 cells and then measured the volume of NPY, POMC and orexin neurons from approximately 9 sections per animal, spaced approximately 120 μm apart, to cover the entire ARC.
Although HFD exposure can affect NPY and POMC gene expression and neuronal function (7–15, 17, 18), we show that neither HFD exposure nor length of exposure affects ARC NPY or ARC POMC neuronal number. This is in contrast with Thaler et al (27), because they showed that 8-month HFD exposure reduced POMC neuronal number in the ARC from approximately 20 (chow mice) to 15 (HFD mice) cells per section. However, this study did not perform stereology and examined POMC number based on 2–4 sections per animal. The use of random, serial systematic sections was not noted. It is possible that the length of HFD exposure may have accounted for the differences as we had mice on HFD for 24 weeks. Moreover a similar nonstereological study from the same authors showed no difference in POMC neuronal number after 20 weeks on a HFD (38), suggesting that a loss of POMC neurons must occur between 6 and 8 months. Another possible explanation could be the diet used (60% vs 43% [current study] calories from fat). Previous stereological studies also show that HFD does not affect total hypothalamic neuron number (39). Thus, our results suggest that HFD-induced dysfunction in NPY and/or POMC neurons is a product of impaired intracellular processing rather than cell number. The fact that length of diet exposure does not affect total NPY or POMC neuronal number supports previous work by Abel and Rance (40), in which no difference in total human infundibular neuronal number was observed.
Our studies do not agree with the idea that diet-induced obesity increases apoptosis in arcuate AgRP and POMC neurons (21), because we would expect increased apoptosis to decrease NPY cell number with HFD feeding. Although it should be noted that Moraes et al (21) did not measure NPY or POMC cell number in their study but rather showed decreased POMC and NPY mRNA levels in HFD-fed rats and mice. It seems most plausible that HFD-induced apoptotic markers in NPY and POMC neurons impair cell function rather than reduce total cell number.
Reports also suggest that HFD decreases the neurogenic turnover of hypothalamic arcuate neurons (22, 23), although others have observed increased neurogenesis in the median eminence (41) or no effect of HFD on hypothalamic neurogenesis (42). Regardless of whether HFD affects the neurogenic turnover of NPY or POMC neurons, our results suggest that this does not regulate the overall NPY or POMC neuronal number.
The inability to reliably and effectively label NPY neurons in the ARC has prevented previously stereological analyses of NPY neurons. To circumvent this issue, we used NPY hrGFP mice, and therefore, it should be noted that our results are based on GFP transgene expression under the control of the NPY promoter. In the present study, we estimated the ARC contains approximately 8000 NPY neurons spread across the entire rostral caudal axis of the ARC. Recent reports using transgenic mice (n = 2) estimated that the ARC contains 9965 ± 66 AgRP neurons and that 94% of genetically labeled AgRP neurons express NPY hrGFP (43). The relative similarity between genetically labeled AgRP neuronal counts and NPY hrGFP counts reinforces the accuracy of using a GFP transgenic approach to quantify ARC NPY cell populations. Although only n = 2, the AgRP counts from Betley et al are in accordance with our current data (43).
In terms of POMC cell number, we estimated the ARC contains approximately 9000 total POMC neurons spread across the entire rostral caudal axis of the ARC. Previous stereological studies in female rats estimated the ARC contained approximately 5500 ± 800 β-endorphin-positive neurons (44). Because β-endorphin is a peptide product processed from the POMC precursor peptide, it is not surprising that we observed greater number of POMC neurons compared with previous observations of β-endorphin neurons. We also noted that HFD trended to increase orexin neurons in the lateral hypothalamus, independent of time spent on HFD. This result does not support previously published work (45) but again this previous published study did not employ robust stereological methods to quantify orexin neurons. We also observed a significant reduction in orexin cell volume in the lateral hypothalamus. Orexin neurons have emerged as crucial regulators of arousal and energy homeostasis (29–31, 46), and given the role of orexin to increase energy expenditure, the trend for HFD to increase orexin neurons may be an adaptive strategy to curtail weight gain.
The most notable finding from our studies was that HFD increased cell volume or altered the distribution of NPY and POMC cell volumes in the ARC. NPY neuronal average cell volume was increased at 6 months relative to 2 months. Although neither diet nor length of diet exposure significantly increased POMC neuronal average cell volume, we specifically noted fewer POMC or NPY neurons with cell volumes ranging from 200 to 500 μm3. An increase in cell volume is significant, because it increases membrane capacitance, which leads to a state of relative membrane hyperpolarization, making it harder for cells to reach a threshold potential and depolarize. Previous studies demonstrate that aged POMC neurons exhibit higher membrane capacitance and reduced firing rates (47, 48). Our cell volume distribution analysis revealed that HFD and length of HFD exposure reduced the number of NPY and POMC neurons with relatively small cell volumes (200–500 μm3). Given the small cell volume and lower membrane capacitance, these smaller volume neurons would require less depolarization to reach an action potential threshold. Thus, the loss of small volume NPY and POMC neurons may represent novel mechanisms through which metabolic dysfunction reduces POMC and NPY neural activity in the ARC. In terms of orexin neurons, the bimodal changes in volume distribution, with HFD increasing the number of neurons with small cell volume but decreasing neurons with larger cell volumes, might preferentially promote activation of orexin neurons based on size. Indeed, these changes may underlie numerous HFD-induced metabolic and sleep disturbances.
Another important observation from our current studies is that HFD increases GFAP cell number in the ARC after a 2-month diet exposure. Interestingly, this significant increase in GFAP cell number at 6 months on a HFD was not significantly different from chow mice, because 6 months of chow diet significantly increased GFAP relative to younger 2-month chow-fed mice. Our results are largely in accordance with nonstereological cell counts from Thaler et al (27), and gene expression from Horvath et al (28), and support the literature, indicating that aging increases gliosis (49, 50). Our results suggest that there is ceiling effect for the number of GFAP cells in the ARC (∼3000), because 6 months of HFD does not further increase GFAP cell number relative 2-month HFD mice. It should be noted that although total number might not change GFAP cell number, HFD increases fiber length and number (26). Surprisingly, we did not observe any differences in ARC Iba1 cell number at 2 or 6 months on HFD. One previous study reported a significant increase at 3 and 14 days after HFD exposure (27) but did examine number at 2 or 6 months. The proinflammatory actions are regulated by activated microglia (51), and in this study, we measured total number, therefore it remains possible that HFD activates microglia independent of total cell number. In support of this, HFD regulates microglia morphology (27), consistent with that seen in activated microglia (51).
Our results provide fundamental biological information on how HFD affects the number and structure of ARC NPY and POMC neurons. The importance of understanding these neuronal populations is underscored by human genetic studies on obesity, because the most common forms of human monogenic obesity involve mutations in neural peptides and receptors, such as peptides from POMC neurons, leptin receptors, and the MC4R (52, 53). Moreover, recent studies based on genome-wide association studies suggest that human obesity is largely a heritable disorder affecting the neural control of energy balance (52–55). Although most human obesity is not caused by monogenic mutations, the studies above clearly highlight the need to understand how the brain controls energy balance to fully appreciate the polygenic nature of human obesity. It is important to acknowledge that our current study does not address the direct functional significance of the morphological changes observed. The functional significance of these morphological changes after HFD exposure remain to be determined.
In summary, we used unbiased designed based stereology to measure the effect of HFD on total ARC NPY and POMC neuronal number and volume, as well as GFAP and Iba1 cell number and orexin neuronal number and volume in the lateral hypothalamus. We observed that neither HFD nor the length of HFD exposure affected the total number of ARC NPY or POMC neurons and Iba1 cells. HFD increased GFAP cell number in the ARC. Our main finding was that HFD and the length of diet exposure significantly increased average NPY cell volume and altered NPY and POMC cell volume distribution. Similar changes were seen in orexin cell volume and not cell number (although HFD trended to increase orexin cell number). Taken together, our results suggest that metabolic perturbations, such as HFD exposure, affect NPY, POMC, and orexin cell morphology rather than total cell numbers. Similar morphological changes after HFD exposure have been observed in NPY and POMC synaptic plasticity (56) and nerve terminal projections (20). We suggest that these morphological changes are a response to the length of HFD exposure, because time spent on HFD exacerbates changes in cell volume. These studies provide the first stereological description of ARC cell number and volume and offer important and fundamental information into the structure and function of ARC NPY and POMC neurons. These data have significant implications for the regulation of energy homeostasis and should be used as an empirical baseline reference to which future studies are compared.
Acknowledgments
This work was supported by a Monash Fellowship, Monash University, Australia; the Australia Research Council Future Fellowship FT100100966; and National Health and Medical Research Council Project Grants 1011274 and 10030037 (to Z.B.A.).
Disclosure Summary: The authors have nothing to disclose.
Abbreviations
- AgRP
Agouti-related protein
- ARC
hypothalamic arcuate nucleus
- CE
coefficient of error
- GFAP
glial fibrillary acidic protein
- GFT
green fluorescent protein
- GTT
glucose tolerance test
- HFD
high-fat diet
- Iba1
ionized calcium-binding adapter molecule 1
- ITT
insulin tolerance test
- MC4R
melanocortin 4 receptor
- NEFA
nonesterified fatty acids
- NHS
normal horse serum
- NPY
neuropeptide Y
- POMC
proopiomelanocortin.
References
