POMC-specific knockdown of MeCP2 leads to adverse phenotypes in mice chronically exposed to high fat diet

Abstract

Methyl-CpG binding protein 2 (MeCP2) is an epigenetic factor associated with the neurodevelopmental disorders Rett Syndrome and MECP2 duplication syndrome. Previous studies have demonstrated that knocking out MeCP2 globally in the central nervous system leads to an obese phenotype and hyperphagia, however it is not clear if the hyperphagia is the result of an increased preference for food reward or due to an increase in motivation to obtain food reward. We show that mice deficient in MeCP2 specifically in pro-opiomelanocortin (POMC) neurons have an increased preference for high fat diet as measured by conditioned place preference but do not have a greater motivation to obtain food reward using a progressive ratio task, relative to wildtype littermate controls. We also demonstrate that POMC-Cre MeCP2 knockout (KO) mice have increased body weight after long-term high fat diet exposure as well as elevated plasma leptin and corticosterone levels compared to wildtype mice. Taken together, these results are the first to show that POMC-specific loss-of-function Mecp2 mutations leads to dissociable effects on the rewarding/motivational properties of food as well as changes to hormones associated with body weight homeostasis and stress.

Graphical abstract

1. Introduction

Methyl-CpG binding protein 2 (MeCP2) is an epigenetic factor that is abundantly expressed in the brain and functions to repress or activate transcription by either selectively binding to heavily methylated-CpG residues of DNA to repress transcription [1] or by binding to lightly methylated CpG islands to activate transcription [2; 3]. Mutations in Mecp2 have been implicated in the neurodevelopmental disorders MECP2 duplication syndrome and Rett syndrome (RTT), rare genetic disorders that are defined by gain- and loss-of-function mutations, respectively, and are clinically characterized by cognitive and motor impairments, mood disorders such as anxiety, seizures, hypotonia and respiratory issues [2; 3]. Past studies have consistently demonstrated that mutations in Mecp2 lead to anxiety-like and depression-like behaviors in transgenic mouse models of RTT and MECP2 duplication syndrome [4; 5]. Moreover, mutations in MECP2 may underlie similar behavioral phenotypes in autism spectrum disorder [5; 6; 7].

While MeCP2 has been clearly implicated in the etiology of RTT and MECP2 duplication syndrome, elucidating its role in other disease states is just now being appreciated with recent evidence demonstrating that disruptions in MeCP2 function lead to obesity, anxiety-like behaviors and cognitive deficits in transgenic mouse models [8; 9; 10; 11]. We have recently shown that exposure to high fat diet perinatally decreases MeCP2 expression in a region of the hypothalamus critical for the regulation of feeding behaviors, the arcuate nucleus (ARC) [12]. Interestingly, clinical data also support a role for MeCP2 in body energy homeostasis as a subset of female RTT patients are more likely to develop obesity or become overweight [13]. Males who survive typically embryonic lethal MECP2 mutations develop severe obesity as well [14]. Given that obesity is on a precipitous rise with prevalence rates of approximately 42% among adults and 20% in children and adolescents [15], identifying mechanisms that exacerbate obesity is necessary in order to combat its deleterious effects. Monogenic forms of obesity are extremely rare [16] and constitute less than 5% of severe obesity cases, therefore other underlying mechanisms may be responsible for its development. Altered function of epigenetic factors such as MeCP2 may be a candidate mechanism for obesity as regulation of MeCP2 can occur through external factors such as diet [12; 17]. Recently, it has been demonstrated that neuron-specific knockdown of MeCP2 in either pro-opiomelanocortin (POMC) or Sim1-expressing neurons of the hypothalamus results in an overweight phenotype that is accompanied by elevated food intake and adiposity [8; 10; 11]. As Mecp2 mutations have been associated with obesity in mouse models [8; 9; 10; 11] and in clinical cases [13; 14], it is possible that MeCP2 contributes to its pathophysiology.

The aforementioned data indicate that Mecp2 mutations produce increases in food intake which ultimately may lead to the overweight phenotype observed in transgenic mouse models. Indeed, previous research has demonstrated that mice with a knockdown of MeCP2 have increased food intake when given ad libitum access to normal chow [10]. Thus, MeCP2 may directly influence feeding behavior or some aspect of feeding behavior. One way in which MeCP2 mutations may alter body weight is by affecting the rewarding/hedonic properties of food reward. Several studies have shown that MeCP2 contributes to the rewarding/addictive properties of drugs of abuse [18] and it has been hypothesized that mutations in Mecp2 enhance these properties in order to increase the incentive salience or “craving” for the drug and/or the hedonic value or “liking” of drugs of abuse. Viral-mediated knockdown or overexpression of Mecp2 in the nucleus accumbens, a neural site that integrates signals related to reward and motivation, produces bidirectional effects on amphetamine-induced conditioned place preference (CPP) [18]. Other studies have shown that exposure to synthetic rewards such as cocaine change DNA methylation patterns which have downstream effects on MeCP2 binding [19; 20]. These data implicate MeCP2 in mediating certain psychological aspects of synthetic rewards but its role in facilitating natural rewards has yet to be elucidated.

Previous work has demonstrated that RTT mice in which Mecp2 has either been overexpressed or truncated produces anxiety-like behaviors as well as maladaptive stress responses in the form of increased levels of corticosterone and corticotropin-releasing hormone (CRH) [21; 22]. CRH is a neuropeptide manufactured by the paraventricular nucleus of the hypothalamus (PVH). Increases in CRH lead to release of adrenocorticotropin releasing hormone (ACTH) from the pituitary gland after cleavage of POMC, a precursor peptide of ACTH found in the pituitary gland among other areas of the central nervous system (i.e., ARC and nucleus of the solitary tract) [23]. CRH-stimulated ACTH release provides a signal to the adrenal glands to release cortisol/corticosterone (CORT), all of which comprise the hypothalamic-pituitary-adrenal (HPA) axis, the physiological system that regulates the endocrine branch of stress responses. It has been well established that chronic elevated levels of CORT can lead to depression-like behaviors in rodent models [24; 25; 26; 27; 28], therefore, it is possible that mice with Mecp2 mutations may also have a depression-like phenotype. Past studies have shown that stress-induced phosphorylation of MeCP2 at S421 as well as mutations in Mecp2 lead to a number of different effects on Pomc and Crh transcription, demonstrating that MeCP2 may influence different aspects of the HPA axis [29] which could eventually culminate in adverse states such as anxiety and/or depression in these mouse models of RTT. The effects of neuron-specific knockdown of Mecp2 has not yet been characterized in the context of aberrant pathophysiological HPA axis function but based on these previously published data, it is possible that POMC-specific mutations in Mecp2 may lead to abnormal HPA axis function.

Collectively, these data demonstrate a potential relationship between mutations in Mecp2, obesity and their combined effects on the HPA axis and the precipitation of depression-like and/or anxiety-like phenotypes. Mutations in Mecp2 may contribute to or exacerbate these conditions therefore understanding its function in the context of obesity and depression/anxiety can provide insight into how these symptoms coalesce to produce adverse phenotypes. The current study examines the effects of MeCP2 knockdown in POMC-neurons on physiological markers associated with stress (i.e., CORT), body weight regulation, as well as anxiety-like and depression-like behaviors. More specifically, does POMC-specific MeCP2 knockdown change plasma CORT levels and if so, does an overactive HPA axis produce depression-like and/or anxiety-like phenotypes in these transgenic mice? The current study also sought to determine if the increased body weight was the result of an increase in the rewarding properties of food as assessed by CPP and/or if perhaps mice with POMC-specific MeCP2 knockdown may be more motivated to obtain food reward as determined by operant conditioning.

2. Materials & Methods

2.1. Mice

To generate knockdown of MeCP2 specifically in POMC neurons of the arcuate nucleus, POMC-Cre (Jackson Laboratory, Stock #: 005965; background strain FVB/N) mice were mated with heterozygous fMeCP2 mice (129/BALBC background backcrossed to C57BL/6) [9; 30]. Littermates from these breeding pairs were genotyped at time of weaning based on the following primers: for Cre: CB159: AC TGT AGA ATC CAT GGG CTC; CB160: ACA GCA TAA GTG AGA CAC TCA; for fMeCP2: Nsi-5: CAC CAC AGA AGT ACT ATG ATC, 2lox-3’: CTA GGT AAG AGC TCT TGT TGA; Gdf F: AAG CCC TCA GTC AGT TGT GC, Gdf R: AAAACCATGAAAGGAGTGGG. Successful POMC-specific deletion of MeCP2 has been validated in previously published reports [10] but using immunohistochemistry we show decreased expression of MeCP2 in the ARC of POMC-Cre MeCP2 KO mice (Figure 1A). Previously published data indicate decreased ARC Pomc mRNA expression in POMC-Cre MeCP2 KO mice, with increased DNA methylation of the Pomc promoter [10]. This may account for the decreased POMC expression in the ARC of our POMC-Cre MeCP2 KO mice in Figure 1A.

Figure 1.

Representative images of MeCP2 KO in POMC neurons and experimental timeline. (A) Representative images of arcuate nucleus. Top two rows are taken at 10x magnification (scale bar 200 μm); bottom two rows are taken at 20x magnification (scale bar 100 μm). (B) Mice were food restricted until 90-95% of free feeding weight after which time mice underwent operant conditioning and conditioned place preference (CPP). Operant conditioning was done until mice completed progressive ratio (~30 d). After operant conditioning, mice did 13 days of CPP. Mice were then allowed ad libitum access to food after which mice then completed elevated plus maze (EPM), forced swim test (FST), and finally novel object recognition (NOR) on separate days. Mice were exposed to high fat diet (HFD) after completion of NOR for ~9 weeks. After HFD exposure, mice were food deprived overnight and sacrificed to assay for plasma leptin, corticosterone, and aldosterone. Brains were extracted for analysis of hippocampal brain derived neurotrophic factor (BDNF). KO: knockout

Two cohorts of 8-15 week old, single-housed WT and POMC-Cre MeCP2 KO male mice were used for experiments (1^st cohort: n=6/group; 2^nd cohort: n=8/group; N=28 male mice) and were housed in ventilated cages with ad libitum access to food and tap water unless otherwise noted. Male mice were maintained on a 12 h light/dark cycle at ambient temperature of 23°C ± 3°C) and humidity (50% ± 20%). Mice were sacrificed at the conclusion of behavioral experiments at ~15-22 weeks of age. All animal procedures were in accordance with the Guide for the Care and Use of Laboratory Animals and were approved by Texas Woman’s University, Institutional Animal Care and Use Committee.

The first cohort was used for operant conditioning and conditioned place preference experiments. Body weight and food intakes were collected from the first cohort of mice. The second cohort was used for the following experiments (in this order): operant conditioning, conditioned place preference, elevated plus maze, forced swim test, and novel object recognition test (Refer to Figure 1B for timeline of experiments). After the conclusion of the novel object recognition test, the second cohort was given high fat diet for 9 weeks before being sacrificed to assay plasma levels of hormones and BDNF.

2.2. Immunohistochemistry

Mice were transcardially perfused with 10% sucrose and then 10% buffered formalin (Fisherbrand, 245-684). Dissected brains were post-fixed in 10% formalin overnight at 4°C and then transferred to a 30% sucrose solution. Coronal sections (30 μm) were cut on a freezing microtome and stored in glycerol solution (25% glycerol, 25% ethylene glycol, 50% PBS) at −20°C until further processing. Hypothalamic sections were placed in blocking buffer (BB: 2% normal goat serum, PBS and 0.2% Triton X) and then probed for POMC (Phoenix Pharmaceuticals, H-029-30; 1:250 in BB) overnight at room temperature (RT) before rinsing (3 washes at 5 min in PBS and 0.2% Triton) and then placed in secondary goat-anti-rabbit antibody (Invitrogen Alexa Fluor 568, A11036; 1:500 in BB). Sections were then blocked in 1% BSA and PBS/0.1% Triton X for 1 hr at RT. Sections were rinsed and incubated with Fab antibody (Invitrogen, 31239; 5 μg/ml in BSA BB). Sections were rinsed and then probed for MeCP2 (Invitrogen, PA1-887; 1:1000 in BB) overnight at RT, rinsed, and then placed in goat-anti-rabbit secondary (Invitrogen Alexa Fluor 488, A11034; 1:500 in BB). Sections were rinsed before mounting on gelatin-coated microscope slides. Slides were coverslipped with ProLong Gold (Invitrogen, P36941) and then imaged on a Leica fluorescent microscope (DM2000; Figure 1A).

2.3. Food intakes and body weights

Prior to high fat diet exposure, mice were food restricted to 95 to 90% of their ad libitum body weight and then underwent operant conditioning and conditioned place preference testing. After behavioral testing was completed, mice were placed on high fat diet (HFD; TestDiet, 58Y1, 18.1% Protein, 61.6% Fat, 20.3% Carbohydrates) for 9 weeks, ad libitum, at which time body weights and food intakes were weighed in grams and recorded in a subset of mice (n=6/group). Body weights were taken once a week while food intakes were taken 3 days a week on Mondays, Wednesdays and Fridays and then averaged for weekly food intakes. A timeline of experiments is provided in Figure 1B.

2.4. Operant conditioning

Past research has demonstrated that mice with Mecp2 mutations have increased food intake [8; 10; 11], however it is not clear if this hyperphagia is the result of a change in the motivational properties of food reward. Therefore, operant conditioning was used to determine if POMC-Cre MeCP2 KO (n=5) mice were more motivated to acquire food reward relative to WT littermate controls (n=5). Prior to conditioning, mice were food restricted until they reached 95-90% of their baseline body weights in order to increase motivation for food reward. Approximately 5 g of chocolate food pellets (BioServ, F05301) were placed in home cages 2-3 days before conditioning to reduce neophobic responses to food pellets. Fixed ratio schedules were used to train mice in which mice were required to press a lever a set number of times to obtain food reward in the form of chocolate pellets (Harvard Apparatus, LE1002CP). During fixed ratio (FR) 1, mice received food reward after 1 lever press, FR3 after 3 lever presses, FR5 after 5 lever presses. Mice were trained on the following schedule: FR1, FR3, and FR5 for 1 day, 2 consecutive days, and 3 consecutive days, respectively. Each mouse had to achieve at least 25 lever presses for FR1, 75 lever presses for FR3, and 150 lever presses for FR5 to successfully complete a training session. After the third consecutive FR5 training day, mice moved to progressive ratio testing in which mice were required to incrementally increase the number of lever presses to obtain food reward over a 2.5 h period such that every 1, 2, 4, 6, 9, etc lever press(es) was reinforced with food reward (following the equation: [5e(R*0.2)]-5). PR schedule performance was considered stable when rewards earned in 1 hr deviate less than 10% for at least 3 consecutive days [31]. The dependent variable of interest was “breakpoint” or the maximum number of lever presses before the mouse stopped pressing the lever for access to food reward [31].

2.5. Conditioned Place Preference (CPP)

Previously, it has been shown that female heterozygous null Mecp2 mice have enhanced CPP responses [11]. However, changes in CPP responses have not been assessed using specific knockdown of Mecp2 in POMC neurons. CPP was, thus used to determine if male mice (n=8) had an increased preference for HFD relative to WT controls (n=8) [32]. Prior to CPP, mice were food restricted until they reached approximately 95-90% of their free feeding body weight. On the first day, mice were given access to all 3 chambers of a CPP box (20”x8.5”x8.25”; middle compartment length: 4”; 2 side compartments length: 8”, in which 2 compartments were made distinct by different flooring and wall patterns and a middle chamber which was considered neutral. On even numbered days (Day 2, 4 etc), a small amount of high fat diet was placed in a mouse’s least preferred side as determined by day 1 testing and mice were given 30 min sessions in the HFD-paired chamber. On odd numbered days (Day 3, 5, etc), normal chow was placed on the other side for 30 min. Conditioning took place over 12 days after which mice were given access to all three chambers and time spent in each chamber was assessed by EthoVision XT. Side preference was determined by the amount of time spent in the HFD-paired side divided by the total amount of time spent in both chambers, multiplied by 100.

2.6. Elevated plus maze (EPM)

It has been well established that forebrain deletions of Mecp2 produce anxiety-like behavior in mice [4; 22; 33] which recapitulates the anxiety phenotype common in RTT patients [34; 35; 36]. It is not yet known if neuron-specific knockdown of MeCP2 produces a similar phenotype in mice. The purpose of this experiment was to assess anxiety-like behavior in mice with POMC-specific knockdown of MeCP2. WT and POMC-Cre MeCP2 KO mice (n=8/group) were placed in the center of an elevated plus maze (Length of arm: 30”; Width of arm: 2”; Height of arm wall: 6”; Height from floor: 20”; Noldus) for 5 min before being placed back into its home cage. Time spent in open and closed arms was assessed by EthoVision XT Videotracking Software (Noldus). Less time spent in open arms is associated with increased anxiety-like behavior [37].

2.7. Forced swim test (FST)

Given the high co-morbidity of anxiety and depression in clinical human populations [38], we decided to also test for behavioral despair or depressive-like behavior in POMC-Cre MeCP2 KO and WT mice (n=8/group). Mice were placed in a 4 L beaker full of ~3 L of room temperature water (23±2°C) for a total of 6 min and time spent immobile was quantified by EthoVision XT. Increased time spent immobile is indicative of increased behavioral despair or depressive-like behavior [39].

2.8. Novel object recognition (NOR)

Because some of our mice were excluded from analysis in operant conditioning, we wanted to determine if there were other issues with learning and memory processes using a recognition memory test. Thus, NOR was used to determine if there were deficits in hippocampal-based episodic memory [40] as a function of POMC-specific MeCP2 KO. Mice were familiarized to a large rectangular box (16.5”x16.5”x10”) for 10 min. Four hours later, mice were familiarized to two of the same object (A) for 10 min (familiarization phase). The following day mice were introduced to one novel object and one familiar object for 10 min and the time and frequency spent with each object was quantified. A recognition index was calculated for all groups [41] by dividing the time spent with novel object with the total time spent on novel and familiar objects and then multiplying that by 100. Time spent with objects was assessed by EthoVision XT. Mice that did not interact with familiar or novel objects during training/testing phases were excluded from analysis. Decreased time spent with the novel object is indicative of deficits in NOR.

2.9. Enzyme-Linked Immunosorbent Assay (ELISA)

Given that previous studies have shown elevated plasma leptin levels in POMC-Cre MeCP2 KO mice after ad libitum access to normal chow, we wanted to test if these differences were maintained when KO mice were challenged with HFD. It is also not known if long-term HFD access could produce changes in stress hormone levels, therefore we also assayed levels of adrenal stress hormones, CORT and aldosterone (ALDO) to determine if there were differences between WT and POMC-Cre MeCP2 KO mice. Given that prolonged, elevated levels of CORT can produce changes in hippocampal Brain-Derived Neurotrophic Factor (BDNF) levels, we assayed hippocampal BDNF levels in our POMC-Cre MeCP2 KO and WT mice [42; 43]. Plasma levels of leptin (Cat# 90030, Crystal Chem), aldosterone (Cat# KGE016, R&D Systems), corticosterone (Cat# KGE009, R&D Systems), and total hippocampal BDNF protein levels (DBNT00, R&D Systems) were quantified according to manufacturer’s instructions. All mice were food deprived 12-16 hours before sacrifice to ensure that baseline levels of leptin were accurately measured. Food intake, in general, can affect leptin levels differentially [44] therefore it is important to control for this variability. Plasma levels of these proteins were quantified at terminal end points.

2.10. Statistics

Independent t-tests were used to determine group differences for the elevated plus maze, forced swim test, plasma leptin, CORT, ALDO, hippocampal BDNF levels, novel object recognition, breakpoint for operant conditioning and conditioned place preference. Repeated measures ANOVA was used to assess differences in body weight and food intake. A Bonferroni correction was used to identify significant post-hoc differences in time points between groups if repeated measures ANOVA yielded significant F values. Statistical significance was set at p≤0.05. Shapiro-Wilk tests of normality were conducted for all data sets and were found to be p>0.05. Levene’s test of equality of error variance was also conducted on all data sets and all values were p>0.01 indicating homoscedasticity of our data. Statistical analyses were performed by IBM SPSS Version 28.

3. Results

3.1. Body weight and food intake

POMC-MeCP2 KO mice weighed significantly more than WT littermate controls all throughout HFD exposure (~9 weeks; Figure 2A). A repeated measures ANOVA did not reveal significant interaction effects but did reveal a significant main effect of time (F(9,90)=130.03, p<0.001). Using a Bonferroni correction, we show significant differences at baseline and during weeks 1-7 (p<0.025) between WT (n=6) and POMC-Cre MeCP2 KO (n=6) mice.

Figure 2. Body weight and food intake after prolonged HFD exposure between WT controls and POMC-Cre MeCP2 KO mice.

(A) Weekly body weight data at baseline and after exposure to high fat diet (HFD) in adult male mice. Green arrow indicates start of HFD. Body weight data indicate that POMC-Cre MeCP2 KO (n=6) mice have significantly higher body weights relative to WT (n=6) littermate controls after ~9 weeks of high fat diet exposure from weeks 1-9. (B) Average weekly food intake in kcal for the duration of the experiment. Data are presented as mean ± SEM. *p<0.025 using a Bonferroni correction; **p<0.007 using a Bonferroni correction; KO: POMC-Cre MeCP2 knockout; WT: wildtype

A repeated measures ANOVA on high fat diet intake (as measured by kcal consumed weekly) demonstrates a significant group by time interaction effect (F(7,70)=3.497, p=0.003) and significant main effect of time (F(7,70)=15.986, p<0.001). Post-hoc analyses, however, did not show significant differences at any time point between WT and POMC-Cre MeCP2 KO mice (Figure 2B). Using a Bonferroni’s correction, post-hoc within subjects analyses demonstrate significant differences between week 1 versus weeks 2, 3, 4, 5, and 7 for POMC-Cre MeCP2 KO mice. We also show significant differences between week 1 versus week 2 food intake in WT control mice. It is not clear why both WT and POMC-Cre MeCP2 KO mice decrease food intakes after their first exposure to HFD. It is possible that these mice may have been “stressed” by behavioral testing and may have responded with elevated food intakes. Finger et al [45] show elevated HFD intake that then attenuated over time after chronic psychosocial stress, similar to the food intake pattern in our mice.

3.2. Operant conditioning

A total of 28 mice were used for operant conditioning of which there were 14 WT and 14 POMC-Cre MeCP2 KO mice. One KO mouse was excluded based on a Grubbs outlier analysis. Five WT and 5 POMC-Cre MeCP2 KO mice successfully completed fixed ratio training and moved on to progressive ratio testing (See Figure 3A for protocol). The other 17 did not differ significantly in motivation to obtain food reward (Figure 3B–3D). More specifically, over 5 days of progressive ratio testing, we did not find significant differences between WT and POMC-Cre MeCP2 KO mice, indicating that POMC-Cre MeCP2 KO mice were not more motivated to obtain food reward relative to WT controls.

Figure 3. POMC-Cre MeCP2 KO mice do not have an increased motivation for food reward.

(A) Operant training protocol. Food restricted mice were trained on a fixed ratio (FR) 1 schedule in which one lever press resulted in one food pellet being delivered. After successful completion of FR1 mice then progressed to FR3, (3 lever presses 1 food pellet), and then to FR5 (5 lever presses 1 food pellet). After successful completion of FR5, mice were then given a progressive ratio (PR) test in which mice had to incrementally increase the number of lever presses in order to receive a food pellet. (B) Average number of lever presses over 5 days of PR testing (p=0.10; WT vs KO). (C) Breakpoint in PR was determined by the maximum number of lever presses prior to the mouse stopping lever pressing. POMC-Cre MeCP2 KO mice (n=5) pressed the lever less than WT littermate controls (n=5) but this was not statistically significant (p=0.11, independent t-test). (C) The average number of reinforcements (food reward) obtained over 5 PR testing sessions. Data are presented as mean ± SEM. *p<0.05; KO: POMC-Cre MeCP2 knockout; WT: wildtype

3.3. CPP

Twelve mice (n=7 WT; n=5 KO) were excluded from this experiment as a result of not acquiring a CPP. Perello et al [32] show that WT mice on a C57BL/6J background develop a CPP to HFD-paired side and given that the WT mice of the first cohort did not show a CPP for the HFD-paired side, we replaced the plastic flooring in our CPP chambers with metal floors. In the original iteration of the experiment, WT mice displayed a ~45% preference for the HFD-paired side, indicating a potential conditioned place aversion. After modifying our CPP chambers (See Figure 4A for protocol), we ran an additional 16 mice (n=8/group) and demonstrated that the WT and POMC-Cre MeCP2 KO mice acquired a CPP for the HFD-paired side. The WT mice showed a 57.8% preference for HFD-paired side while the POMC-Cre MeCP2 KO mice preferred the HFD-paired side 66.5% of the time spent in the CPP chamber, suggesting that the KO mice developed a stronger place preference than WT littermate controls (Figure 4B; independent t-test, p=0.035).

Figure 4. POMC-Cre MeCP2 KO mice display an increased preference for HFD.

(A) Conditioned place preference (CPP) protocol. Food restricted mice were habituated to 3 compartments (1 neutral and 2 distinct) on the first day of training after which time high fat diet (HFD) and normal chow (NC) were paired with their least preferred and most preferred sides, respectively. HFD was paired with one side on even numbered days while NC was paired with the other side on odd numbered days. On testing day, mice were allowed to freely explore all chambers and time spent in each chamber was assessed. Side preference was determined by the amount of time spent in the HFD-paired side divided by the total amount of time spent in both chambers. (B) CPP. POMC-Cre MeCP2 KO mice (n=8) significantly preferred the HFD-paired side compared to WT controls (n=8) as determined by percent preference. Data are presented as mean ± SEM. *p<0.05; KO: POMC-Cre MeCP2 knockout; WT: wildtype

3.4. EPM

There were no significant differences in anxiety-like behavior as POMC-Cre MeCP2 KO mice spent similar amounts of time in the open arms of the EPM as WT littermate control mice (n=8/group; Figure 5A and 5B).

Figure 5. POMC-Cre MeCP2 KO mice do not show anxiety-like or depressive-like phenotypes.

(A) Elevated plus maze (EPM). Adult male POMC-Cre MeCP2 KO mice do not show increased anxiety relative to WT controls as determined by frequency and time spent in the open arms of an elevated plus maze (n=8/group). (B) % time spent in open arms of EPM was not different between WT and KO mice. (C) Forced swim test (FST). POMC-Cre MeCP2 KO mice show a slight increase in immobility in the FST compared to WT mice (n=8/group), but these data are not statistically significant (p=0.13, independent t-test). Data are presented as mean ± SEM. *p<0.05; KO: POMC-Cre MeCP2 knockout; WT: wildtype

3.5. FST

POMC-Cre MeCP2 KO mice showed a slight increase in immobility relative to WT controls (p=0.13, independent t-test; Figure 5C; n=8/group) however these data are not statistically significant.

3.6. NOR

Our current data show that POMC-Cre MeCP2 KO (n=6) mice do not have impairments in recognition memory (See Figure 6A for protocol) as POMC-Cre MeCP2 KO mice were able to recognize novel objects in the NOR test as well as WT controls (n=8) (Figure 6B, 6C).

Figure 6. POMC-Cre MeCP2 KO mice do not have deficits in hippocampal-based novel object recognition (NOR) compared to WT.

(A) Protocol for NOR. Mice were habituated to an open field for 10 min. During the familiarization phase four hours after habituation, mice were allowed to explore 2 of the same object. The following day (testing phase), mice were presented with 1 familiar object and 1 novel object and time spent with each object was quantified. (B) Time (sec) spent with each object during the testing phase. POMC-Cre MeCP2 KO (n=6) mice did not spend more time with the novel object compared to WT controls (n=8). (C) Frequency with which mice explored novel vs familiar objects. POMC-Cre MeCP2 KO mice interacted with the novel object the same number of times as WT controls. (D) Recognition index in NOR. POMC-Cre MeCP2 mice did not show impairments in recognizing the novel object as assessed by the recognition index (time spent with novel object/total time spent with both objects) relative to WT controls. Data are presented as mean ± SEM. KO: POMC-Cre MeCP2 knockout; WT: wildtype

The recognition index was determined by taking the time spent with the novel object and dividing it by the total amount of time spent with both objects. Based on these data, POMC-Cre MeCP2 KO mice did not display impairments in recognizing new objects (Figure 6D).

3.7. ELISA

Plasma leptin was significantly increased in POMC-Cre MeCP2 KO (n=6) mice with plasma leptin levels more than 2 fold higher than WT (n=5) littermate control mice (Figure 7A, independent t-test, p=0.025). One blood sample from the WT group was excluded from analysis due to lysing of red blood cells which precludes accurate plasma leptin measurements.

Figure 7. Plasma leptin and corticosterone are significantly elevated in POMC-Cre MeCP2 KO mice compared to WT controls.

(A) Plasma leptin levels at the conclusion of the experiment. Plasma leptin was significantly higher in POMC-Cre MeCP2 KO (n=5) mice versus WT (n=6) controls (p=0.025). (B) Plasma corticosterone at the conclusion of experiments. Adult male POMC-Cre MeCP2 KO mice had significantly more plasma corticosterone relative to WT littermate controls (n=8/group). (C) Plasma aldosterone. POMC-Cre MeCP2 KO mice (n=6) had similar levels of plasma aldosterone compared to WT littermate controls (n=7). (D) Hippocampal Brain Derived Neurotrophic Factor (BDNF). POMC-Cre MeCP2 KO mice (n=10) did not have altered BDNF levels in the hippocampus compared to WT controls (n=11). Data are presented as mean ± SEM. *p<0.05; KO: POMC-Cre MeCP2 knockout; WT: wildtype

Based on an independent t-test, we saw significant differences between WT and POMC-Cre MeCP2 KO mice (n=8/group) with KO mice having higher plasma corticosterone relative to WT littermate controls (p=0.04; Figure 7B). We assayed levels of aldosterone and hippocampal BDNF levels to see if elevated corticosterone may have had downstream effects on these proteins. We did not find any statistically significant differences using independent t-tests in plasma aldosterone or hippocampal BDNF levels between WT (n=7 aldosterone; n=11 BDNF) and POMC-Cre MeCP2 KO (n=6 aldosterone; n=10 BDNF) mice (Figure 7C and 7D, respectively). Intra-assay variability (or %CV) was 19% for leptin, 29% for CORT, 26% for ALDO, and 5% for BDNF ELISA assays.

4. Discussion

Here we demonstrate that POMC-specific MeCP2 knockdown leads to substantial weight gain (Figure 2A), a commensurate increase in plasma leptin (Figure 7A), as well as an enhanced conditioned place preference for HFD relative to WT littermate controls (Figure 4B). Our data also show that knockdown of MeCP2 in POMC neurons results in elevated plasma corticosterone levels (Figure 7B) and a slight increased immobility in the forced swim test, although these data are not statistically significant (Figure 5C). These results suggest that Mecp2 mutations in POMC neurons alter preferences for food reward, the effects of which may be dissociable from the elevated corticosterone we observed in POMC-Cre MeCP2 KO mice.

Previous work has shown that knockout of MeCP2 either globally [9] or in specific populations of hypothalamic Sim1 neurons [8] results in an overweight phenotype. Using a different transgenic line in which Mecp2 (B6.129P2-Mecp2^tm1Bird) was knocked out in POMC neurons, Wang et al [10] demonstrated that KO mice also have an overweight phenotype but these mice were maintained on normal chow, not high fat diet. Our current data show that this overweight phenotype persists using HFD and a different strain of floxed MeCP2 mice [9; 30]. We show increased plasma leptin levels (Figure 7A), replicating Wang et al’s findings but contrary to Wang et al’s study we did not observe an increase in food intake (Figure 2B). It is worth noting that POMC expression is not restricted to ARC but is also expressed in the pituitary gland [46] and the nucleus of the solitary tract (NTS) [47]. Thus, knockdown of MeCP2 may be occurring in the pituitary gland and NTS as well. Nonetheless, our data indicate a consistent overweight phenotype when MeCP2 is specifically knocked down in POMC neurons (Figure 2A) and that this is not necessarily the result of an increase in high fat food intake (Figure 2B).

Our conditioned place preference data (Figure 4B) demonstrate that POMC-Cre MeCP2 KO mice show an increased preference for HFD but interestingly were not necessarily more motivated to obtain food reward, as evidenced by decreased breakpoint in the progressive ratio task (Figure 3C). Although not statistically significant, the data showed a decreased breakpoint relative to WT controls. This decreased breakpoint was not the result of learning impairments as these mice performed equally well as WTs in the novel object recognition test (Figure 6B-6D), a test of hippocampal episodic memory [40]. Our CPP data replicate earlier findings using a different animal model in which MeCP2 was knocked out centrally in female heterozygous Mecp2 mice [11]. Overall, our findings suggest that MeCP2 knockdown does not alter the motivational properties of food reward (Figure 3B–3D) but could heighten preferences to HFD (Figure 4B). Our results demonstrate dissociable effects of POMC-specific knockdown of MeCP2 on preferences to food reward (i.e., CPP) as well as its motivational properties (i.e., operant conditioning).

Given that our POMC-Cre MeCP2 KO mice exhibit more “liking” (as assessed by CPP) than “wanting” (as determined by operant conditioning) of food reward, this indicates that the “hedonic impact” of food reward may be of more significance than the “craving” aspect of food reward [48]. Previously published work has shown that changes in Mecp2 expression, specifically in reward-related areas of the brain (i.e., nucleus accumbens and ventral tegmental area) lead to alterations in the rewarding properties of cocaine and amphetamine [18; 49]. Collectively, these results suggest that knockdown of Mecp2 in POMC neurons may have downstream effects on reward-processing areas of the brain although MeCP2’s role in the context of natural reward (i.e., food) still has yet to be determined, particularly as it relates to POMC function. Based on our CPP data, it is plausible that MeCP2 contributes to the processing of natural reward (i.e., high fat diet).

It is not clear why we see elevated plasma CORT levels in our POMC-Cre MeCP2 KO mice relative to WT mice (Figure 7B). POMC deficiency in general results in decreased ACTH, corticosterone, and aldosterone levels [50; 51; 52] which is consistent with the premise that ACTH stimulates release of these adrenal hormones. Previous work has shown that POMC-specific knockout of MeCP2 decreases ARC Pomc mRNA expression which would presumably result in a decrease in ACTH, a peptide product of POMC [50; 51; 52]. If this is the case, then we would expect to see a commensurate decrease in CORT as ACTH stimulates the release of corticosterone/cortisol from the adrenal cortex [53]. The increased CORT in our POMC-Cre MeCP2 KO mice (Figure 7B) contradicts this idea thus elevations in CORT must be occurring through a different mechanism in our transgenic mice. Past data have shown that Mecp2 loss-of-function mutations increase and Mecp2 gain-of-function mutations decrease Crh mRNA in the PVH [21; 22]. Based on these results, it is possible that increased CRH, as a function of MeCP2 knockdown, may be causing this increase in CORT levels in our POMC-Cre MeCP2 KO mice as CRH independently stimulates CORT release [54; 55]. These results suggest that MeCP2 expression has bidirectional effects on CORT release whereby MeCP2 overexpression results in decreased CORT and Mecp2 loss-of-function mutations lead to increased plasma CORT [22] presumably through its effects on CRH. Our data are the first to demonstrate that MeCP2’s impact on CORT release can be localized to its effects on POMC neurons. While we show a significant increase in plasma CORT in our POMC-Cre MeCP2 KO mice relative to WT mice, our %CV (coefficient of variability) is high for CORT and for ALDO (29% and 26%, respectively) indicating variability between replicates in our ELISA assay. The high %CV is a potential limitation to our data and may be due to edge effects of ELISA plates, temperature of reagents used in the ELISA assays, pipetting error, etc [56].

Chronic exposure to glucocorticoids such as CORT, either peripherally or centrally, leads to depression and a number of studies have established that exogenous administration of CORT or exposure to chronic stress produces depression-like behaviors in rodent models [24; 25; 26; 27; 28]. Using a mouse model of RTT bearing a truncated Mecp2 allele, mutant mice expressed higher levels of plasma CORT as well as an increase in anxiety-like behavior [22]. We did not observe anxiety-like or depressive-like phenotypes as assessed by the elevated plus maze (Figure 5A and 5B) or the forced swim test (Figure 5C), respectively. However, this does not preclude the possibility that our MeCP2 KO mice do not have these behavioral phenotypes as other tests of anxiety-like and/or depression-like behaviors may yield different results. Future experiments could use other behavioral measures of anxiety and/or depression using this transgenic mouse model to determine conclusively if an anxiety/depression phenotype exists.

Circulating levels of corticosterone/cortisol will bind to GC receptors or MC receptors in the hippocampus thereby leading to decreased expression of BDNF, a neurotrophic factor implicated in cell survival, neuronal health, neurogenesis and other processes associated with neuron function [42; 43]. Increased CORT also compromises hippocampal function/plasticity by decreasing BDNF levels [57; 58; 59]. We did not observe changes in BDNF expression (Figure 7D) in our POMC-Cre MeCP2 KO mice. Given that our KO mice had elevated CORT levels, it is unclear at what point our KO mice had increased CORT. Future work could pinpoint the time at which POMC-Cre MeCP2 KO mice have elevated CORT and if this then leads to changes in hippocampal BDNF levels as previous studies have demonstrated that prolonged exposure to increased glucocorticoid levels is associated with alterations in BDNF expression [60; 61].

Our results are the first to demonstrate that POMC-specific MeCP2 knockdown produces elevated plasma corticosterone and leptin levels, enhanced conditioned place preference as well as increased body weight after long-term exposure to high fat diet. To our knowledge, our data are also the first to show that MeCP2 knockdown in POMC neurons alters the rewarding properties of food as demonstrated by CPP while not necessarily affecting its motivational qualities. Taken together, these results indicate the dynamic effects of POMC-specific MeCP2 dysfunction on diverse biological/psychological processes which could contribute to a number of disease states such as obesity and its associated sequelae.

Highlights.

• Knockdown of MeCP2 in arcuate POMC neurons leads to increased body weight as well as increased plasma levels of the anorectic hormone leptin after prolonged high fat diet exposure

• High fat diet preference is enhanced as assessed by conditioned place preference in mice with arcuate POMC-specific knockdown of MeCP2 relative to wildtype littermate controls

• POMC-specific knockdown of MeCP2 in mice leads to elevated levels of the glucocorticoid, corticosterone after high fat diet exposure

• Our data suggest that aberrant expression of MeCP2 in arcuate POMC neurons may contribute to obesity pathophysiology and reveal other pathologies associated with POMC-specific knockdown of MeCP2