Loss of Mrap2 is associated with Sim1 deficiency and increased circulating cholesterol

Melanocortin receptor accessory protein 2 (MRAP2) is a transmembrane accessory protein predominantly expressed in the brain. Both global and brain-specific deletion of Mrap2 in mice results in severe obesity. Loss-of-function MRAP2 mutations have also been associated with obesity in humans. Although MRAP2 has been shown to interact with MC4R, a G protein-coupled receptor with an established role in energy homeostasis, appetite regulation and lipid metabolism, the mechanisms through which loss of MRAP2 causes obesity remains uncertain. In this study, we used two independently derived lines of Mrap2 deficient mice (Mrap2tm1a/tm1a) to further study the role of Mrap2 in the regulation of energy balance and peripheral lipid metabolism. Mrap2tm1a/tm1a mice have a significant increase in body weight, with increased fat and lean mass, but without detectable changes in food intake or energy expenditure. Transcriptomic analysis showed significantly decreased expression of Sim1, Trh, Oxt and Crh within the hypothalamic paraventricular nucleus of Mrap2tm1a/tm1a mice. Circulating levels of both high-density lipoprotein and low-density lipoprotein were significantly increased in Mrap2 deficient mice. Taken together, these data corroborate the role of MRAP2 in metabolic regulation and indicate that, at least in part, this may be due to defective central melanocortin signalling.


Introduction
Melanocortin receptor accessory protein (MRAP) and its paralogue MRAP2 are a recently identified class of small, single-pass transmembrane domain accessory proteins (Chan et al. 2009, Novoselova et al. 2013. Both MRAP and MRAP2 have been shown to interact with the melanocortin receptors (MCRs), a family of G protein-coupled receptors (GPCRs) with diverse physiological function stimulated by pro-opiomelanocortin (POMC) derived peptide agonists such as adrenocorticotropin hormone (ACTH) and α-MSH (Cone 2005, Chan et al. 2009). Of the five MCRs (MC1R-MC5R), only the function of the melanocortin 2 receptor (MC2R) is clearly recognized to be facilitated by MRAPs (Metherell et al. 2005, Chan et al. 2009), although in-vitro data suggests a broader role in conjunction with all the MCRs (Chan et al. 2009, Sebag & Hinkle 2009. MRAP is highly expressed in the adrenal gland and is essential for MC2R function. Mutations in MRAP are associated with familial glucocorticoid deficiency (OMIM#607398) (Metherell et al. 2005). MRAP2 is predominantly expressed in the central nervous system and hypothalamus, in particular within the paraventricular nucleus (PVN), a region known to have a role in energy homeostasis (Chan et al. 2009). Mice with global and brain-specific Mrap2 deletion developed marked obesity and rare loss-of-function or missense heterozygous variants in MRAP2 were also identified in humans with severe earlyonset obesity (Asai et al. 2013). This work demonstrated that MRAP2's role in the control of body composition and growth is via MC4R signalling (Asai et al. 2013). Further evidence for a link with MC4R signalling came from a study on the role of Mrap2 in zebrafish feeding and growth (Sebag et al. 2013).
Given these data, the phenotype observed in Mrap2deficient mice is likely, at least in part, to be driven by disruption of central melanocortin signalling. However, some areas of uncertainty remain. In particular, the paradoxical observation that the mutant mice become obese without detectable changes in food intake or energy

(ii) reduced hypothalamic
Mrap2 transcript in Mrap2 tm1a/+ and Mrap2 tm1a/tm1a compared with Mrap2 +/+ mice (n = 3 per genotype); mean plotted ± s.e.m.; **P < 0.005; ***P < 0.0005). (C) Expression of Mrap2 in the hypothalamus of the wild-type 129/Sv Mrap2 +/+ and Mrap2 tm1a/tm1a mice as shown by in situ hybridization using coronal brain sections (approx. bregma -0.6 mm). AS, antisense probe, S, sense probe as a negative control. Third ventricle indicated as 3 V, asterisk indicates position of the PVN, OC, optic chiasm; scale bars = 200 μm. (D) Weight curves of Mrap2 tm1a/tm1a in both genders and genetic backgrounds illustrated. 129/Sv Mrap2 tm1a/tm1a n > 8 per genotype and gender, C57BL/6N Mrap 2tm1a/tm1a n = 7 of each gender and genotype. (E) Total body weight gain of C57BL/6N Mrap2 tm1a/tm1a mice by the age of 16 weeks, n = 7 for each gender/genotype (left) and appearance of the mutant mice compared with the wild type (right). Data is presented as both box-and-whiskers plot (showing min-mean-max values, with the box representing the 25th and 75th percentiles), and as a scatter dot plot for individual values. P-values presented on graphs are either global P-values for genotype adjusted for multiple correction testing, or (in the cases of sexual dimorphism) the P-value is the impact of genotype for that sex.
balance (Asai et al. 2013) requires exploration, as does the potential role of MRAP2 in peripheral cholesterol and lipid metabolism, a function known to be regulated by melanocortins (Nogueiras et al. 2007, Perez-Tilve et al. 2010. In this study, we have used an independently derived line of Mrap2-deficient mice (Mrap2 tm1a/tm1a ) on two different genetic backgrounds to further study the role of MRAP2 in the regulation of energy homeostasis and the control of cholesterol and lipid metabolism.

Generation of Mrap2-deficient mouse
Mice carrying the knockout-first conditional-ready allele Mrap2 tm1a(EUCOMM)Wtsi (abbreviated to Mrap2 tm1a ) were generated on a C57BL/6N background as part of the Sanger Mouse Genetics Project (MGP) (Fig. 1A). Mice carrying the same Mrap2 tm1a allele were generated separately on a 129S5/ SvEvBrdWtsi;129P2/OlaHsdWtsi background (abbreviated to 129/Sv). Detailed description of the Sanger Mouse Genetics Project methodology has been reported (Skarnes et al. 2011). Briefly, a promoter-containing cassette (L1L2_ Bact_P) was introduced upstream of the critical Mrap2 exon 4 at position 87175333 of Chromosome 9, Build GRCm38 (Fig. 1A). The vectors containing Mrap2 tm1a were electroporated into C57BL/6N derived JM8F6 and 129P2/ OlaHsd derived E14Tg2a embryonic stem (ES) cells. Correct ES cell gene targeting was confirmed by long-range PCR and quantitative PCR. Targeted ES cells were microinjected into blastocysts and used to generate chimeras. Germ-line transmission was confirmed by genotyping PCR analyses (http://www.knockoutmouse.org/kb/25/). Mice obtained from heterozygous intercross were genotyped for the Mrap2 tm1a allele by PCR (Supplementary Table 1, see section on supplementary data given at the end of this article).

Animal husbandry
The care and use of all animals were carried out in accordance with the UK Home Office regulations, UK Animals (Scientific Procedures) Act 1986. Mice were kept under a standard 12 h light:12 h darkness cycle with food and water ad libitum unless otherwise stated. 129/Sv background mice were maintained in a facility at 22°C and fed a standard chow (SDS RM3, Essex, UK). Mice on a C57BL/6N background were maintained at 21°C ± 2°C, humidity 55% ± 10% and fed a standard rodent chow (LabDiets 5021-3, IPS, Richmond, VA, USA).

Metabolic phenotyping
Metabolic phenotyping was undertaken at two independent centres. In accordance to the 3R (replacement, reduction and refinement) principles of humane experimental technique and based on scientific objectives, not all procedures were performed on both lines. The genetic background of Mrap2 tm1a mice used in each experiment have been specified previously.
Phenotyping using C57BL/6N mice was performed at the Wellcome Trust Sanger Institute as part of the MGP (White et al. 2013), whilst studies using 129/Sv background mice were carried out at the University of Cambridge Metabolic Research Laboratories (MRL). For data arising from the MGP, a cumulative baseline was generated from controls of the same genetic background, age and sex. Seven male and seven female Mrap2 tm1a mice were processed in five batches for each sex (one to three mice per batch) and were phenotyped unblinded as part of a larger mixed genotype group that included weekly wild-type controls, with the individual mouse as the experimental unit. Animals for testing were randomly assigned to test sessions and operators. Mice were group housed to a minimum density of three per cage.
After overnight fasting (approximately 16 h), intraperitoneal glucose tolerance tests (IPGTT) were carried out on 13-week-old mice. After taking a baseline glucose measurement, mice were given a single glucose injection (2 g/kg) and blood glucose measured at 15, 30, 60 and 120 min (Accu-Chek Aviva, Roche).
Blood for plasma biochemistry was collected from 16-week-old C57BL/6N animals into lithium-heparin tubes. Animals were not fasted unless otherwise indicated. Clinical blood chemistry was carried out using an Olympus AU400 chemistry analyser (Olympus). Insulin levels were measured by Mesoscale Discovery array technology platform.
Additional data relating to the C57BL/6N Mrap2 tm1a/tm1a line can be found at http://www.mousepheno type.org/ data/genes/MGI:3609239. For studies at MRL, individual experiments were matched for age and sex of mice. The body weight and length were measured weekly since weaning. Food intake was carried out on 8-week-old single-housed acclimatized animals. Response to fasting was measured after mice were moved into clean cages, and food was removed at 07:00 for 24 h. Energy expenditure was determined at 8 weeks of age using indirect calorimetry. Animals were placed in a custom-built monitoring system based on their home cages (Ideas Studio, Cambridge, UK). Oxygen consumption and carbon dioxide production was measured, and samples were taken at 18 min intervals for 48 h. Energy expenditure was calculated using indirect calori metry with the Elia and Livesey constants for respiratory quotie nt (Elia & Livesey 1992). Activity was assessed by beam breaks (beams 1.25 cm apart) and measurements were taken as total, rather than consecutive beam breaks.

Behavioural tests
Open field assessment was used to quantify spontaneous locomotor behaviour in a novel environment. The open field, custom-designed-walled, infra-red backlit arena 75 cm 2 (Tracksys Ltd., Nottingham, UK) was subdivided into a centre zone (42 cm 2 ) with the remainder designated as border zone. Twenty-week-old 129/Sv background mice were recorded for a 20 min period using Noldus Ethovision-3 video tracking software. The position of the centre-point of the mouse within the open field was recorded. A mouse was considered to begin moving when its velocity surpassed 2 cm/s and stop moving when below 1.75 cm/s. Mrap2 tm1a/tm1a and their Mrap2 +/+ controls were littermates housed in single-sex groups of three to five.
To generate riboprobes for in-situ hybridization (ISH), RNA was extracted from hypothalamus and cDNA prepared. Full-length Mrap2 cDNA fragment (898 bp) was PCR amplified (Supplementary Table 1), ligated into pGEM-T easy vector (Promega), sequenced and then linearized with EcoRI or NotI (Promega). Digoxigenin (DIG)-labelled antisense and sense cRNA probes were synthesized by in-vitro transcription with T7 or SP6 RNA polymerases (Roche). Dissected brains were embedded into OCT and frozen in liquid nitrogen, 20 µm cryosections were cut onto slides and fixed with ice-cold 4% PFA for 20 min. Slides were then subjected to ISH as described previously (Gorrigan et al. 2011).

Laser-captured microdissection and RNA isolation
Mouse brains from 9-week-old, 129/Sv Mrap2 tm1a/tm1a and Mrap2 +/+ mice were dissected, immediately embedded into OCT and frozen in liquid nitrogen. Coronal sections (20 μm) covering the region from −0.58 mm to −1.22 mm caudal to bregma (Franklin & Paxinos 2012) were cut on a cryostat and mounted on Superfrost Plus slides (Thermo-Fisher). Frozen sections were fixed for 40 s in 95% ethanol and then rehydrated (75% and 50% ethanol, 30 s each). The slides were stained with 1% cresyl violet in 75% ethanol (w/v) for 45 s, dehydrated in a graded ethanol series (50, 75, 95, 100% for 30 s each), in 100% ethanol for 5 min and air-dried. Laser microdissection was performed using a P.A.L.M. MicroBeam (Zeiss). The PVN was collected into AdhesiveCap tubes (Zeiss). Total RNA was immediately isolated using the RNAqueous-Micro Kit (Ambion). Quality and quantity of the total RNA samples were determined by the Agilent BioAnalyzer using PicoChip. RNAse free technique and RNAase free reagents were used throughout.

RNA microarray hybridization and analysis
Fifteen nanograms of isolated RNA with the RNA Integrity Number of at least 6.5 (n = 4 for Mrap2 +/+ ; n = 3 for Mrap2 tm1a/tm1a ) was converted into cDNA using Ovation PICO SL System V2 (NuGEN), which was then fragmented and labelled using Encore BiotinIL Module (NuGEN). About 1500 ng of each labelled product was then hybridized with MouseRef-8v2.0 Expression BeadChip Kit according to the manual and scanned using iScan (Illumina, San Diego, CA, USA). Raw image data were converted to bsc format using Illumina GenomeStudio 2011.1 software. Bonferroni correction with Family-Wise Error Rate (FWER) of 0.05 was applied to identify statistical significance of gene expression changes. Pathway analysis was performed using DAVID6.7 (http://david.abcc.ncifcrf. gov/tools.jsp) and STRING 10 (http://string-db.org/).

Quantification of RNA by real-time quantitative PCR (Q-RT-PCR)
Dissected tissues were immediately frozen in liquid nitrogen, homogenized using Precellys24 (Precellys, Bertin Technologies, Paris, France) into RPL buffer (Qiagen) and the RNA was extracted with RNeasy Mini Kit (Qiagen). cDNA was produced with SuperScriptII (Life Technologies) and 50 ng of cDNA was used for RT-Q-PCR with TaqMan Universal MasterMix II and gene-specific TaqMan probes (Life technologies, Supplementary Table 2). The fold change related to Actin-b was calculated using 2 -ΔΔTh method (Livak & Schmittgen 2001).

Protein quantification
White and brown fat tissues were homogenized using Precellys24 in ice-cold RIPA buffer (Sigma) containing phosphotase (Roche) and protease inhibitors cocktail (Sigma). Lysates were centrifuged for 20 min at 4°C before separation of the lipid layer. The SDS-PAGE samples were prepared with 2× Sample buffer (Sigma), heated at 95°C for 5 min, centrifuged for 20 min at 4°C to separate samples from residual lipids and subjected to western blotting. The membrane was blocked with 5% bovine serum albumin in TBS (Life Technologies) for 1 h at 22°C followed by incubation at 4°C overnight with the primary antibody: anti-ACTB antibody 1/10,000 (Abcam), anti-UCP1 1/5000 (Abcam) and antibodies for Fatty Acid and Lipid Metabolism and Lipolysis Activation (8334, 8335 Cell Signaling Technology). After three washes, the membranes were probed with anti-mouse 680 and anti-rabbit 800 IRDye antibodies (LI-COR). The band intensities were quantified using Odyssey software.

Statistical analysis
All data were generated from the MGP utilized statistical analysis with RStudio running R version 3.1.2 and Phenstat package version 2.0.1. This uses a mixed-model framework (Karp et al. 2012) to assess the impact of genotype on phenotype. The analysis was performed by loading the model without body weight, therefore analysing the absolute differences between genotypes whilst accounting for sex, using the model: Y = Genotype + Sex + Genotype × Sex. Multiple correction testing was performed on the global P-values using the Hochberg correction. Data is presented as both box-and-whiskers plot (showing min-mean-max values, with the box representing the 25th and 75th percentiles), and as a scatter dot plot for individual values. P-values presented on graphs are either global P-values for genotype adjusted for multiple correction testing, or (in the cases of sexual dimorphism) the P-value is the impact of genotype for that sex.
For other data, males and females were assessed independently and the effect of genotype compared with wild-type controls was statistically tested using a twotailed Student's t-test. For calorimetry data, multiple linear regression analysis (ANCOVA) was used. Data is plotted as mean ± s.e.m. and analysed using Microsoft excel and GraphPad Prism.

Production of Mrap2-deficient mice
Mice carrying the mutant Mrap2 tm1a allele were viable with expected homozygous mutant offspring born from heterozygous matings (21% C57BL/6N Mrap2 tm1a/tm1a and 23% 129/Sv Mrap2 tm1a/tm1a ). Both female and male Mrap2 tm1a/tm1a mice were fertile and did not exhibit any changes in skin or hair colour/appearance. The introduction of the knockout-first Mrap2 tm1a allele resulted in targeted disruption of the critical exon 4 encoding the transmembrane domain of the protein. The predicted outcome would be a premature stop codon, thus producing a short 132 bp transcript that, if translated, would produce a 44 amino acid protein (predicted MW 5 kDa). Previous work demonstrated that such a protein was unlikely to be translated (Asai et al. 2013). However, generation of hypomorphic mice have been demonstrated previously using the 'knockout-first' strategy targeting other genes (McIntyre et al. 2012, Chen et al. 2013, White et al. 2013. We therefore determined the expression of Mrap2 by Q-RT-PCR analysis using a TaqMan probe spanning exons 4-5. cDNA generated from whole hypothalamus extracted from mice on an 129/Sv background revealed a low but detectable residual Mrap2 transcript of Mrap2 tm1a/tm1a within homozygous mice (13%, range 11-16%), whilst heterozygotes Mrap2 tm1a/+ mice had approximately half of the Mrap2 transcript expression compared with Mrap2 +/+ (Fig. 1B).

Mrap2 is predominantly expressed in the paraventricular nucleus of the hypothalamus
Mrap2 RNA expression was studied in wild-type mice tissues (Fig. 1B). The highest expression level was detected in the hypothalamus with substantial expression observed in the pituitary gland. Mrap2 expression was also detected in the cortex, cerebellum and adrenal gland. Kidney, testes, thymus and pancreas had very low expression levels whilst expression in white fat, liver, brown fat and skin was undetectable (Fig. 1B). ISH using a full-length Mrap2 probe showed visible Mrap2 RNA expression in the PVN of Mrap2 +/+ mice on a 129/Sv background, which was absent in Mrap2 tm1a/tm1a mice (Fig. 1C).

Mrap2 deficiency results in obesity in both C57BL/6N and 129/Sv background
Before weaning, there was no difference in body weight between wild-type and Mrap2 tm1a/tm1a mice (Fig. 1D).
However, in both genetic backgrounds and in both sexes, Mrap2 tm1a/tm1a mice had a significant increase in body weight from 6 weeks of age. By 16 weeks of age on a C57BL/6N background, mean body weight in male wild-type mice was 32.2 g, compared with 42.5 g in Mrap2 tm1a/tm1a mice; the corresponding weights in females were 25.4 g and 38.7 g, respectively (Fig. 1E). Similarly, on a 129/Sv background, mean body weight in male wild-type mice was 29.1 g, compared with 45.1 g in Mrap2 tm1a/tm1a mice; the corresponding weights in females were 24.4 g and 36.9 g, respectively.
In C57BL/6N, this increase in body weight was as a result of a significant increase in both fat and lean mass ( Fig. 2A and D). C57BL/6N Mrap2 tm1a/tm1a females had twice the fat/body weight ratio of Mrap2 +/+ controls and Mrap2 tm1a/tm1a males displayed a 1.5-fold increase ( Fig. 2A and B) with clear increase in adipocyte size macroscopically (Fig. 2C). No difference in bone mineral content or density was observed in either sex (Fig. 2E). Body length was significantly increased in female, but not male Mrap2 tm1a/tm1a for Mrap2 +/+ controls gender/genotype. P-values presented on graphs are either global P-values for genotype adjusted for multiple correction testing, or (in the cases of sexual dimorphism) the P-value is the impact of genotype for that sex. ***P = 0.0005; NS, not significant. mice (Fig. 2F). On the 129/Sv background, a significant increase in fat mass was recorded in both male and female Mrap2 tm1a/tm1a mice compared with wild-type controls with no change in lean mass observed (data not shown).

Mrap2-deficient mice display little difference in food intake and energy expenditure compared with wild-type control mice
Activation of the melanocortin system has a role in both feeding behaviour in both ad libitum conditions and in re-feeding after fasting. To determine if loss of Mrap2 affects feeding behaviour in either situation, the food intake and body weight of 8-week-old male and female 129/Sv mice were monitored over a period of 60 h. For the first 24 h, food was freely available, after which mice were fasted for 24 h followed by reintroduction of food (Fig. 3A, B, C and D).
Neither female nor male Mrap2 tm1a/tm1a mice exhibited changes in food intake when compared with wild-type controls over the entire period. Further, Mrap2 tm1a/tm1a mice did not show any difference in the rate of weight loss upon fasting compared with the wild-type mice. However, interestingly, during re-feeding after a fast, Mrap2 tm1a/tm1a male mice did not re-gain weight as fast as the wild-type males ( Fig. 3C and D).
Total energy expenditure measurements versus lean mass or total body weight did not show significant changes between the genotypes/sex (Supplementary Fig. 1). Analysis of respiratory quotient over a period of 48-h demonstrated that it did not differ between Mrap2 tm1a/tm1a and their Mrap2 +/+ littermates ( Supplementary Fig. 2). In keeping with a lack of change in energy expenditure, there was no difference between Mrap2 tm1a/tm1a and Mrap2 +/+ mice in the expression level of Ucp1 mRNA and UCP1 protein levels in brown adipose tissue of age-matched animals, despite differences in morphology ( Supplementary Fig. 3).
Locomotor activity measurements (average beam breaks in a 5 min time period) demonstrated that male 129/Sv Mrap2 tm1a/tm1a mice, compared with wild-type, moved significantly more during the daytime (Fig. 4A). No difference was observed in females.

Mrap2 tm1a/tm1a mice display behavioural changes when presented with a novel environment
To further examine the locomotor activity as well as novel environment exploration and anxiety-related behaviour, 20-week-old 129/Sv Mrap2 tm1a/tm1a mice were subjected to an open field exploration test during the light phase. This recapitulated the sex-specific difference in locomotor activity between female and male Mrap2 tm1a/tm1a mice ( Fig. 4B and C). Mrap2 tm1a/tm1a male mice spent more time moving and covered a greater distance compared with Mrap2 +/+ mice. Although Mrap2 tm1a/tm1a male mice appeared to spend more time traversing the centre of the open field Food intake and energy expenditure balance in Mrap2 +/+ , Mrap2 tm1a/+ and Mrap2 tm1a/tm1a mice. Body weight dynamics of female (A) Mrap2 +/+ n = 6; Mrap2 tm1a/+ n = 4; Mrap2 tm1a/tm1a n = 10 and male (B) Mrap2 +/+ n = 6; Mrap2 tm1a/+ n = 7; Mrap2 tm1a/tm1a n = 6 mutant mice and their food intake (C and D) in response to a 24 h fast (129/Sv background). P < 0.05. than controls, the difference was not significant, P = 0.075 ( Fig. 4C and D). There was no difference in thigmotactic behaviour in the females, and Mrap2 tm1a/tm1a mice of both sexes displayed no differences in gait, circling and rearing behaviour (data not shown).

Mrap2-deficient mice are Sim1 deficient
To further explore what might be the driving changes in body composition in Mrap2-deficient mice, we undertook transcriptomic analysis of laser microdissection PVN from 9-week-old 129/Sv Mrap2 tm1a/tm1a mice and wild-type littermates (Fig. 5A). Mice of a 129/Sv genetic background, less prone to developing obesity-related co-morbidity, were used to reveal the effect of Mrap2 deficiency without secondary changes caused by hyperinsulinaemia and/or elevated glucose. We confirmed the changes observed in laser-capture material by undertaking Q-RT-PCR on whole hypothalamus extracted from a separate, second population of 129/Sv Mrap2 tm1a/tm1a mice and wild-type littermates. Expression of genes that did not show any changes by microarray, such as Sf-1 and Pomc, were also confirmed by Q-RT-PCR as additional controls ( Supplementary Fig. 4).
We could not detect significant changes in Mc4r mRNA expression in the PVN of Mrap2 tm1a/tm1a mice due to the high variability between mice within each group (Fig. 5B). However, we found that Sim1 mRNA level in the PVN of Mrap2 tm1a/tm1a mice was less than 50% of that observed in wild-type littermates (Fig. 5C). Sim1 is responsible for the late stages of the differentiation of oxytocin (Oxt), arginine vasopressin (Avp), corticotrophin-releasing hormone (Crh), thyrotropin-releasing hormone (Trh) and somatostatin neurons (Sst) (Michaud et al. 1998). In keeping with this, in Mrap2 tm1a/tm1a mice, PVN expression levels of Oxt, Avp, Trh and Crh were significantly decreased compared with the wild-type. Sst expression in Mrap2 tm1a/tm1a was unchanged compared with Mrap2 +/+ mice, although results were variable within the cohort (n = 3 per group). Analysis of RNA from whole hypothalami (Fig. 5D) recapitulated these findings, except for Crh expression levels, which did not reach statistical significance. All changes were confirmed in both sexes (data not shown). It is known that SIM1 is implicated in the development of the PVN and Sim1 +/-mice exhibit a smaller PVN with reduced neuron number compared with their wild-type littermates (Michaud et al. 2001). We could not find morphological changes or a reduction in the number of neurons in the PVN of Mrap2 tm1a/tm1a mice compared with wild-type ( Fig. 5E and F), suggesting that unlike Sim1 +/-, a lack of Mrap2 does not cause underdevelopment of the PVN.

Mrap2 deficiency increases circulating HDL and LDL cholesterol
Macroscopically, the livers of Mrap2 tm1a/tm1a mice were visibly pale in both sexes in both 129/Sv and C57BL/6N backgrounds, and the histological analysis showed lipid accumulation ( Fig. 6A and B). There is recent evidence that the central melanocortin system directly controls peripheral lipid metabolism and circulating cholesterol (Nogueiras et al. 2007, Perez-Tilve et al. 2010. We therefore studied the cholesterol and lipid profile in C57BL/6N Mrap2 tm1a/tm1a mice. The blood triacylglyceride levels (TAG) were not significantly different in Mrap2 tm1a/tm1a mice compared with wild-type (Fig. 6C). However, total circulating cholesterol in Mrap2 tm1a/tm1a mice was significantly higher than in wild-type controls of both sexes (Fig. 6D). HDL was elevated in both sexes with a greater percentage increase in females (Fig. 6E). LDL was significantly increased in male and female Mrap2 tm1a/tm1a mice (Fig. 6F). NEFA-C levels were not significantly different between Mrap2 tm1a/tm1a and Mrap2 +/+ mice of either sex (Fig. 6G), whilst glycerol concentration was increased to a similar degree in mutant mice of both sexes (Fig. 6H). To investigate whether high cholesterol levels were due to a decrease in cholesterol re-uptake in the liver or an increase in cholesterol synthesis, we analysed the expression of the HDL scavenger receptor Scarb1, LDL receptor (Ldlr) and the key transcription factor of cholesterol biosynthesis Srebp2 (Shimomura et al. 1998) in the livers of 129/Sv Mrap2 tm1a/tm1a mice. Interestingly, Srebp2 mRNA levels were increased in the livers of female Mrap2 tm1a/tm1a mice, whereas Scarb1 and Ldlr levels were similar to the wild type (Fig. 6I). Expression analysis of livers from male Mrap2 tm1a/tm1a mice showed lower levels of Ldlr mRNA whereas Scarb1 and Srebp2 transcript were similar to the wild-type male littermates (Fig. 6I).
To study white fat function, we tested the protein levels and phosphorylarion state of enzymes involved in lipogenesis and fatty acid synthesis as well as phosphorylation of the rate-limiting enzyme for lipolysis of hormone sensitive lipase (HSL). Phosphorylation of ATP-citrate lyase (ACL), an enzyme responsible for the synthesis of cytosolic acetyl-CoA that serves lipogenesis and cholesterolgenesis pathways (reviewed in Chypre et al. 2012), was increased in white fat of female Mrap2 tm1a/tm1a mice but not in Mrap2 tm1a/tm1a male mice ( Fig. 7A and B, Supplementary Fig. 5A and B).
HSL phosphorylation is known to be important for the enzyme activation and therefore lipolysis (Egan et al. 1992). The proportion of HSL phosphorylated on S563, S565 and S660 was analysed and it was found that the proportion of S660 phosphorylated HSL was three times lower in white fat of female Mrap2 tm1a/tm1a mice compared with the wild type ( Fig. 7C and D). Phosphorylation on other residues was not changed and male Mrap2 tm1a/tm1a mice did not have changes in HSL phosphorylation on any residues tested ( Supplementary Fig. 5C, D, E and F).

Analysis of glucose homeostasis in Mrap2-deficient mice
Mrap2 tm1a/tm1a mice of both sexes on a C57BL/6N background fed on a chow diet from weaning display elevated fasting plasma insulin concentration with higher fasting blood glucose at 13 weeks of age compared with Mrap2 +/+ mice ( Fig. 7E and F). Glucose clearance in for female Mrap2 +/+ and 35 for Mrap2 +/+ controls gender/genotype. The P-values presented on graphs are either global P-values for genotype adjusted for multiple correction testing, or (in the cases of sexual dimorphism) the P-value is the impact of genotype for that sex. NS, not significant. *P < 0.05; **P < 0.005; ***P < 0.0005; NS, not significant. (I) Elevated expression levels of Srebp2 in Mrap2 tm1a/tm1a female mice, reduced Ldlr in male Mrap2 tm1a/tm1a mice and expression levels of Scarb1 in female mice and male mice (n = 4 for each genotype/gender, *P < 0.05, NS, not significant). response to an IP glucose bolus ( Fig. 7G and H) appeared delayed, although statistical analysis of the area under the curve was only significant in female mice.

Discussion
In this study, we report the generation and characterization of a murine model with a targeted Mrap2 allele (knockoutfirst Mrap2 tm1a(EUCOMM)Wtsi ). The construct used here is a Tm1a allele, which theoretically can still allow transcriptional read through (White et al. 2013). We detected low Mrap2 expression within Mrap2 tm1a/tm1a homozygous mice and in the absence of a suitable antibody we cannot rule out the possibility that the animals studied were indeed 'strong hypomorphs' rather than of complete knockouts.
Mrap2 tm1a/tm1a mutant mice on both C57/BL6N and 129/Sv background display severe early-onset obesity with a significantly increased fat mass, consistent with a recent report of Mrap2-knockout mice (Asai et al. 2013). Unlike Mrap2 -/mice on a 129/Sv genetic background (Asai et al. 2013), our C57BL/6N Mrap2 tm1a/tm1a display elevated fasting insulin and blood glucose concentrations. We believe that this is an evidence of an interesting interaction between the genetic backgrounds with Mrap2, which will form the basis of a future study.
Mrap2 -/mice have been reported previously to show no increased food intake or reduction in energy expenditure and thermogenesis to account for their increased body weight. In our assessment of 8-week-old Mrap2 tm1a/tm1a mice, we recapitulated and confirmed these findings. We calculate that female Mrap2 tm1a/tm1a mice gained more weight compared with the wild-type mice (females: 0.117 ± 0.041 g/day; males: 0.096 ± 0.023 g/day). To achieve this, a female mutant mouse would need to deposit 0.701 ± 0.246 kcal/day and a male mouse 0.578 ± 0.139 kcal/day (Flatt 1991). This would equate to an increase in food intake of 0.232 ± 0.081 g of standard chow (for females) and 0.191 ± 0.046 g (for males), which is within the measurable limits of food intake variation. Therefore, it is possible that the causative difference is below the threshold of the detection (Tschop et al. 2012, Speakman 2013. Indeed, older animals that are significantly more obese than their wildtype counterparts, and as a result would be expected to consume larger quantities, demonstrate a subtle increase in cumulative food intake when monitored over 50 days (Asai et al. 2013). Additionally, our behavioural tests on separate cohorts of animals, independently analysed on two separate platforms, demonstrated sex-specific increased daytime locomotor and exploratory activity in Mrap2 tm1a/tm1a male mice, which may indicate foodseeking behaviour. These lines of evidence would point to hyperphagia being the key driver in the development of obesity. However, importantly, obesity in Mrap2-deficient animals clearly precede any change in food intake and in paired feeding studies Mrap2-deficient animals continue to gain more weight than their wild-type counterparts (Asai et al. 2013). It is only when further food restriction was undertaken did weight gain in mutant mice become equivalent to that of wild-type mice (Asai et al. 2013). Intriguingly, this suggests a far more complex mechanism at play in Mrap2 null mice.
Our transcriptomic analysis of the PVN of the mutant mice also favoured increased energy intake as being a more likely promoter of increased body weight. It was found that Sim1 expression levels were low in the PVN of the Mrap2 tm1a/tm1a mice resembling Sim1 deficiency. The reduced expression of Sim1 and its associated neuropeptides cannot be secondary to obesity alone, as this was not observed in reported hypothalamic microarray data from obese mice fed in a high fat diet (Lee et al. 2010). SIM1 is a transcription factor that regulates development of the PVN, and Sim1 -/mice die due to the abnormal hypothalamic architecture (Michaud et al. 2001). Heterozygous Sim1 +/-mice exhibit a small PVN with reduced neuronal number and develop severe early-onset obesity due to hyperphagia and increased linear growth. They have an impaired response to MTII, an MC3R/MC4R agonist, indicative of a disrupted central melanocortin pathway (Holder et al. 2004, Kublaoui et al. 2006a,b, Tolson et al. 2010. Expression analysis of the PVN from Sim1 +/-mice has shown an 80% decrease in Oxt expression and 20-40% decrease in Trh, Crh, Avp and Sst expression (Kublaoui et al. 2008). Compared with Sim1 +/mice, we did not detect morphological changes within the PVN of Mrap2 tm1a/tm1a mice. However, the levels of Oxt, Avp and Trh in Mrap2 tm1a/tm1a mice PVN were reduced, consistent with low Sim1 expression levels. Interestingly, despite these changes and also the high expression of Mrap2 in the pituitary, we found no evidence of pituitary dysfunction in Mrap2 tm1a/tm1a mice. Progression through puberty and fertility appear unchanged in mutant mice, and thyroid hormone levels, T3 and T4, were normal (data not shown). Corticosterone concentrations were previously reported to be normal (Asai et al. 2013). This would suggest that the level of neuropeptide expression is sufficient for peptide production and physiological stimulation of downstream hormones, as exemplified by normal levels of T4 in the case of Trh.
In contrast to the downstream effects, the change in neuropeptide expression is likely to play a direct role in maintaining energy homeostasis as it is known that Oxt, Avp and Trh in the PVN have anorexigenic effects (reviewed in Valassi et al. 2008), and Oxt is thought to be key to the mechanism for the hyperphagia of Sim1 +/− mice (Kublaoui et al. 2008). Overall, the changes in Sim1 provide further evidence that a central melanocortin pathway deficiency exists in Mrap2 tm1a/tm1a mice as SIM1 has been considered to be downstream of MC4R signalling (Holder et al. 2004, Kublaoui et al. 2006a,b, Tolson et al. 2010. Modulation of MC4R has been shown to directly affect peripheral lipid metabolism. Mc4r -/mice had elevated plasma cholesterol and HDL levels (Nogueiras et al. 2007, Perez-Tilve et al. 2010. Both sexes of Mrap2 tm1a/tm1a mice displayed elevated circulating cholesterol although there is a suggestion that males and females partition cholesterol into HDL differently, consistent with reports of sex differences in the hepatic control of cholesterol metabolism (De Marinis et al. 2008). Mrap2 tm1a/tm1a female mice showed increased de novo hepatic lipogenesis; however, unlike female Mrap2 tm1a/tm1a mice, male Mrap2 tm1a/tm1a mice had low expression of liver LDL receptor, possibly reflecting elevated circulating LDL levels.

Figure 7
Phosphorylation of ACL and HSL are changed in female Mrap2 tm1a/tm1a mice and insulin insensitivity in both genders observed at 13 weeks of age. (A) Analysis of ACL phosphorylation in white fat of the female mutant mice compared with the wild type by immunoblotting. (B) A significant increase is demonstrated in mutant mice by using band densitometry analysis of the ratio of phosphorylated to non-phosphorylated ACL normalized to β-actin. (C) Phosphorylation of HSL on S660 in white fat of the female mutant mice is decreased as assessed by western blotting. (D) Densitometry analysis of phosphorylated to non-phosphorylated HSL normalized to β-actin was calculated (n = 3 per genotype; 129/Sv). (E) Fasting plasma glucose and (F) fed-state plasma insulin are higher in C57BL/6N Mrap2 tm1a/tm1a mutant mice of both genders associated with significantly elevated insulin levels (log transformed due to the range of values observed in Mrap2 tm1a/tm1a male mice. (G&H) IPGTT performed on 13-week-old mice of both genders, delayed glucose clearance demonstrated in Mrap2 tm1a/tm1a female mice but not statistically significant in male Mrap2 tm1a/tm1a mice. n = 7 for each Mrap2 tm1a/tm1a group, 39 for female Mrap2 +/+ and 35 for Mrap2 +/+ controls gender/genotype. P-values presented on graphs are either global P-values for genotype adjusted for multiple correction testing, or (in the cases of sexual dimorphism) the P-value is the impact of genotype for that sex. AUC, area under the curve; *P < 0.05; **P < 0.005; ***P < 0.0005; NS, not significant.