Introduction

Urocortins are the endogenous ligands for the corticotropin-releasing factor receptor type 2 (CRFR2) [15]. Together with corticotropin-releasing factor, urocortins are important regulators of neurohormonal responses to stress [6], integrating adaptive changes in energy flux, and glucose and lipid metabolism [7]. However, chronic stress has been linked to an increased risk of the metabolic syndrome and its components (type 2 diabetes mellitus, dyslipidaemia, hypertension, obesity, atherosclerotic heart disease) [8].

Significant roles for CRFR2/urocortins in regulating metabolism have indeed been identified. Crfr2 (also known as Crhr2)-null mice resist high-fat diet (HFD)-induced obesity and insulin resistance, effects thought to be primarily centrally mediated, with increased sympathetic activity increasing thermogenesis in brown adipose tissue and hence the metabolic rate [9, 10]. Intriguingly, however, urocortins are also present at high levels in peripheral metabolic tissues, including skeletal muscle [11, 12], implying that these ‘stress regulators’ have local peripheral as well as central metabolic actions.

In vitro, skeletal muscle CRFR2 stimulates futile substrate cycling between de novo lipogenesis and lipid oxidation, with a requirement for glucose metabolism and AMP-activated protein kinase (AMPK) signalling [13]. This has spawned suggestions that CRFR2 agonists may protect muscle from excessive intracellular lipid storage and thus from lipotoxicity-induced insulin resistance. However, Ucn2-null mice are protected against diet-induced obesity, with increased insulin sensitivity in skeletal muscle, perhaps due to absence of paracrine antagonism of insulin by urocortin 2 [11]. Ucn3-null mice are also protected from glucose intolerance [14], yet pancreatic urocortin 3 leads to enhanced beta cell insulin production [12]. Thus, although urocortins are clearly potent regulators of glucose metabolism, underlying mechanisms and key tissue(s) of action are various, and the predominant effects unclear.

The necessary pharmacological studies of urocortin administration in vivo have to date only involved short-term treatments due to experimental constraints. Urocortin 3 is the most specific ligand known for CRFR2 [4, 5]. To model the effects of chronic pharmacological CRFR2 activation on metabolism in vivo, we exploited this specificity, and generated and phenotyped a transgenic mouse overexpressing Ucn3.

Methods

Production of transgenic mice

The Ucn3 transgene (Ucn3 +) was constructed by ligating the mouse Ucn3 open reading frame [4] into the EcoRI restriction site of the pBROAD3 vector (InvivoGen, San Diego, CA, USA) (Fig. 1a), placing it downstream of the mouse Rosa26 (also known as Gt(ROSA)26Sor) promoter and upstream of a human β-globin polyA signal. The expression cassette was injected into the pronucleus of fertilised (C57BL/6 × BALB/c) F1 mouse oocytes. Mice were genotyped by PCR of genomic DNA. Mouse Ucn3-specific sense (5′-ATGACAACCAAACGCACCTCCAGA-3′) and human ß-globin-specific antisense (5′-GGCCCTTCATAATATCCCCCAGTT-3′) primers resulted in a 450 bp transgene-specific amplification product. Five founder animals were identified. Transgene expression was quantified in F3 generations. The line with highest expression was used for subsequent analyses. Crfr2-null mice [15] were crossed with Ucn3 + mice for two generations to generate homozygous Crfr2-null mice carrying the Ucn3 + transgene for some experiments.

Fig. 1
figure 1

Ucn3 overexpression in tissues of Ucn3+ transgenic mice results in increased skeletal muscle mass with myocyte hypertrophy. a A construct for overexpression of Ucn3 was designed as shown, putting the transgene under the control of the mouse Rosa26 promoter. The position of primers used to genotype mice is shown. b Northern blots of RNA extracted from tissues of Ucn3+ mice and control littermates probed with Ucn3 and S16 cDNA. B, brain; H, heart; L, liver; K, kidney; M, muscle. c Growth curves of Ucn3+ mice (black squares) show they are heavier than controls (white squares) from 6 weeks of age (n = 17–19). Whole body composition analysis showed this to be due to (d) increased lean tissue mass with (e) no change in fat mass (n = 10). f Weight of individual skeletal muscles showing that Ucn3+ mice have increased skeletal muscle mass (n = 13–17). g Muscle histology (×200 magnification, scale bar 50 μm) and (h) myocyte cross-sectional area reveal increased muscle mass to be due to myocyte hypertrophy (n = 4). i Muscle fibre type distribution in the soleus muscle. Extensor digitorum longus and gastrocnemius muscles were >98% type 2 in control and Ucn3+ mice (not shown). *p < 0.05 vs control. di Black bars, Ucn3+; white bars, control

Animal husbandry

Mice were housed in temperature- and lighting-controlled rooms (lights on for 12 h) with free access to laboratory chow and water except where stated. Experiments were carried out on male mice between 16 and 26 weeks old. Ucn3 + mice were heterozygous for the transgene and controls were transgene-negative littermates. Principles of laboratory animal care (NIH No. 85-23, 1985) were followed. All procedures were approved by The Salk Institute Animal Use and Care Committee or carried out according to the UK Animals Scientific Procedures Act, 1986. Food intake was measured over 6 days in individually housed mice and the average intake per day was adjusted for body weight as per Kleiber’s allometric law.

In vivo studies

Procedures were carried out in the morning. ‘Fed’ is defined as food not withheld, ‘fasted’ is defined as 18 h food deprivation. Measurements were carried out fed except where stated, with the exception of glucose tolerance tests (GTTs) and insulin tolerance tests (ITTs), which were carried out following 6 or 3 h food deprivation, respectively. Glucose (2 g/kg) or insulin (0.75 IU/kg) was administered i.p. in saline and tail blood collected for immediate glucose measurement (Lifescan OneTouch Ultra glucometer; Johnson & Johnson, Milpitas, CA, USA). For insulin measurements, blood was collected by retro-orbital bleed. For other measurements, blood and tissues were collected immediately after animals had been killed. For glucagon response test and insulin signalling studies, see Electronic supplementary material (ESM), Methods.

High-fat feeding

Mice were fed an HFD (45% energy as fat) or control low-fat diet (LFD) (10% energy as fat) (products D12451 and D12450B, respectively; Research Diets, New Brunswick, NJ, USA) from weaning (3 weeks) until 16 weeks old and weighed weekly. Food intake was quantified at 15 weeks. GTT, ITT and time course measurement of blood glucose levels during fasting were performed, and pre- and post-fasting weight recorded. Mice were culled fasted or fed, and individual fat depots and hindlimb muscles were dissected out and weighed.

Metabolism cage studies

Rates of oxygen uptake (\( \dot{V}{{\text{O}}_{{2}}} \)), rates of carbon dioxide production (\( \dot{V}{\text{C}}{{\text{O}}_{{2}}} \)), food intake and locomotor activity (ESM Methods) were measured continuously and simultaneously over 3 days (LabMaster; TSE-Systems, Bad Homburg, Germany) after 48 h adaptation in singly housed mice.

Body composition analysis

Body composition of whole carcasses was determined using PIXImus densitometry (GE-Lunar, Madison, WI, USA) as previously validated [16].

Histology

Liver and quadriceps muscles were formalin-fixed, dehydrated in graded ethanol/xylene, paraffin-embedded, sectioned at 5 mm intervals and stained with haematoxylin and eosin. Liver was stained with periodic acid–Schiff's reagent (PAS) with and without diastase pre-treatment. For measurement of myocyte size, hindquarters were skinned and fixed, and decalcified in 10% (wt/vol.) formaldehyde, 8% (wt/vol.) formic acid and 1% (vol./vol.) methanol with the legs in mild flexion for 48 h. Transverse sections through the mid-femur were paraffin-embedded, sectioned at 4 mm intervals and stained with PAS. Areas of 20 myocytes/mouse were measured from ×300 digital images of rectus femoris muscle in cross-section using imaging software (QCapture Pro; QImaging, Surrey, BC, Canada).

Muscle fibre typing

Tissues were dissected and snap-frozen in liquid nitrogen-cooled melting isopentane. Transverse sections (8 μm) were stained with myosin ATPase to demonstrate fibre types classified as type 1 or type 2 as previously described [17]. Digital images (100×) covering the whole section and including, when possible, a minimum of 350 fibres, were analysed to quantify fibre type distribution using Image-Pro Plus software (Media Cybernetics, Bethesda, MD, USA).

Hormone measurements

Insulin and IGF-1 were measured using Sensitive Rat Insulin RIA (Linco Research, St Charles, MO, USA) and Mouse/Rat IGF-1 Quantikine ELISA (R&D Systems, Minneapolis, MN USA) kits, respectively. For corticosterone, glucagon and growth hormone measurements, see ESM Methods.

Combined GTT and [3H]glucose or 2-deoxy[3H]glucose tracer uptake into tissues

This was carried out based on methods previously described [18, 19]. Mice were fasted for 3 h and weighed. After tracer (3.7 MBq/ml d-[3-3H]glucose or 1.9 MBq/ml 2-deoxy-d-[2,6-3H]glucose solution [Amersham Pharmacia Biotech, Little Chalfont, UK]) with 2 g/kg glucose in 0.3 ml saline had been injected, blood glucose and plasma radioactivity were measured over 90 min and AUC calculated. At 90 min, mice were killed, and tissues dissected and snap-frozen. To estimate glucose uptake, tissues were homogenised in distilled H2O and centrifuged for 10 min at 600 g and 4°C. Total [3H]glucose and [3H]glucose 6-phosphate were counted in the supernatant fraction after separation using a 100 to 200 mesh anion exchange resin column (AG1-X8; Bio-Rad, Hercules, CA, USA). Blood radioactivity AUC and [3H]glucose uptake were used to calculate glucose clearance into each tissue. Total tissue glycogen was estimated as previously described [20]. For glucose incorporation into glycogen, radioactivity of digested samples was divided by the specific activity in serum over the GTT time course. For 2-deoxyglucose uptake, weighed tissues were digested at 50°C overnight in 1 ml Soluene-350 (Packard Bioscience, Groningen, the Netherlands). Uptake was calculated from the plasma profiles of 2-deoxy[3H]glucose and glucose, together with concentrations of radiolabelled tracer in tissue.

Hyperinsulinaemic clamps

Mice were implanted with an indwelling jugular catheter 3 days prior to experiments. Fasted mice received a 2 h basal saline infusion, then a primed continuous infusion of short-acting insulin consisting of a bolus of 100 mU/kg for 3 min, followed by a constant rate of 2.5 mU kg−1 min−1 (a standard protocol designed to achieve a high physiological insulin level of ~450 pmol/l [21]). Glucose (20% wt/vol. in saline) was infused at a variable rate (2.5–12 μl/min) to maintain euglycaemia (6–7 mmol/l) as far as possible. Tail blood samples were used to monitor glycaemia at baseline (0), and at 20, 40, 60, 70, 80, 90, 100, 110 and 120 min.

Phosphoenolpyruvate carboxykinase 1 activity

This was assayed in liver homogenates as previously described [22].

Northern blotting

Total RNA was extracted from tissues and northern blotting carried out as previously described [23] using deoxycytidine 5′-[32P]triphosphate-labelled mouse Ucn3 cDNA. Blots were rehybridised with S16 probes to control for loading.

Tissue lysates, SDS-PAGE and immunoblotting

Protein abundance and phosphorylation in tissue lysates was quantified by SDS-PAGE as previously described [19]. Primary antibodies were from Cell Signaling Technology (Beverley, MA, USA), except pY612-IRS-1 (BioSource International, Camarillo, CA, USA), total IRS-1 (Upstate Cell Signaling Solutions, Waltham, MA, USA), myostatin (Millipore, Watford, UK) and tubulin (Li-COR Biosciences, Lincoln, NE, USA). Primary antibodies against phosphorylated proteins were used at 1:500 dilution, those against tubulin at 1:10,000 and all other primary antibodies at 1:1,000. For AMPK, see ESM methods.

Quantitative real-time PCR

Total RNA was extracted from tissues and 500 ng reverse-transcribed to cDNA using a kit (QuantiTect DNase; Qiagen, Crawley, UK). PCR cycling was carried out using a LightCycler480 (Roche Diagnostics, Mannheim, Germany). Standard curves were constructed for each assay using serial dilution of cDNA pooled from all samples. Reactions were performed in triplicate. Results were corrected for 18S, cyclophilin A and Hprt1 (also known as Hprt) mRNA expression levels, which were not different between groups. For cyclophilin A, Ppargc1a, Ucn2 and Ucn3, assays Mn02342429-g1, Mm00440939-m1, Mm01227928-s1 and Mm00453206-s1 (Applied Biosystems, Carlsbad, CA, USA) were employed. Other gene-specific assays and probes were designed using the Universal ProbeLibrary Assay Design Center (Roche). Polymerase, buffer and dNTPs were provided from the LightCycler480 Probes Master kit (Roche). Details of primer sequences and probes are available upon request.

Statistical analysis

Statistical analyses employed the Student’s two-tailed t test, or one- or two-way ANOVA with post hoc analysis using Bonferroni’s or Dunnett’s test as appropriate. Data are presented as mean ± SEM. Differences were considered statistically significant at p < 0.05. For further details, see ESM, Methods.

Results

Generation of transgenic mice overexpressing Ucn3

Ucn3+ mice were generated by placing the coding region of mouse Ucn3 under the control of the ubiquitously expressed Rosa26 promoter (Fig. 1a). Progeny from founders were screened and a line selected that expressed Ucn3 mRNA at high levels in a wide range of organs, as reflected in elevated plasma and tissue peptide levels (Fig. 1b and Table 1). Mice carrying the transgene were born at the expected Mendelian ratio and showed no obvious signs of ill health until at least 18 months old or any significant pathology on routine post-mortem examination (not shown).

Table 1 Urocortin 3-like immunoreactivity in tissues of control and Ucn3 + mice

Ucn3+ mice have increased skeletal muscle mass

Ucn3+ mice were heavier than wild-type littermates by 6 weeks old. By 12 weeks this difference stabilised with Ucn3+ mice weighing ~12% more than controls (Fig. 1c). Carcass analysis revealed that this was entirely due to increased lean tissue mass (Fig. 1d, e). Weights of parenchymal organs, femoral length and nose to tail length were not significantly different between genotypes (not shown). Individual skeletal muscle weights (whether predominantly type 1, type 2 or mixed fibre type) were greater in Ucn3+ mice (Fig. 1f), whereas fat pad weights were unaltered (ESM Fig. 1). Histology revealed muscle hypertrophy with increased myofibre diameter (Fig. 1g, h), and a shift towards type 2 glycolytic muscle fibre type (Fig. 1i).

Ucn3+ mice have an increased energy requirement

Food intake was significantly greater in Ucn3 + mice, even when adjusted for body weight, signifying an increased energy requirement (Fig. 2a). \( \dot{V}{{\text{O}}_{{2}}} \) and \( \dot{V}{\text{C}}{{\text{O}}_{{2}}} \) were ~7% higher in Ucn3 + mice (Fig. 2b, c), but individual variation did not allow conclusions to be drawn on the significance of this. Increased respiratory exchange ratio (RER) indicated higher proportional use of carbohydrate as a metabolic fuel source (Fig. 2d). While voluntary running wheel activity was not different, Ucn3 + mice showed increased spontaneous activity in metabolism cages (ESM Fig. 1). This may in part account for their increased energy requirements, but cannot explain other aspects of their phenotype including glucose metabolism as described below.

Fig. 2
figure 2

Ucn3+ mice have an increased energy requirement and uptake of glucose to tissues. a Food intake is higher in Ucn3+ mice (black bars) than controls (white bars) (n = 10–11). b\( \dot{V}{\text{C}}{{\text{O}}_{{2}}} \), (c) \( \dot{V}{{\text{O}}_{\text{2}}} \) and (d) RER are elevated in Ucn3+ mice (n = 5–7). e Blood glucose levels during fasting (n = 10–12) and (f) during GTT (n = 10) showed lower glucose in Ucn3+ mice (black squares) than in controls (white squares). g Body weight, (h) fasting blood glucose and (i) GTT in Ucn3+ mice on a Crfr2-null background (n = 5–6) show that these aspects of phenotype are mediated via CRFR2. Light grey bars and circles, Crfr2-null; dark grey bars and circles, Crfr2-null/Ucn3+. j Glucose uptake to peripheral tissues, and (k) total glycogen and (l) acute incorporation of glucose into glycogen in skeletal muscle were increased in Ucn3+ mice in a GTT (n = 8). m Phosphorylation of glycogen synthase at Ser641 indicated increased enzyme activity (n = 7). *p < 0.05 vs control and p < 0.05 for Crfr2-knockout vs control; p = 0.07 for Ucn3+ vs control

Ucn3+ mice show more rapid falls in blood glucose during fasting and a GTT

Ucn3+ mice had lower fed blood glucose levels, a more rapid drop in glucose at the onset of fasting and significantly lower glucose levels during fasting (Fig. 2e). Ucn3+ mice showed improved glucose tolerance across a GTT (Fig. 2f). To determine whether the effects of Ucn3+ were mediated, as anticipated, by CRFR2, Ucn3+ mice were crossed with Crfr2-null mice. Importantly, the body mass increase, and fasting and GTT glucose-lowering effects of the Ucn3+ transgene were negated by the Crfr2-null mutation, confirming that muscle hypertrophy and glucose phenotype were dependent on CRFR2 (Fig. 2g–i). Ucn3+ mice had increased uptake of 2-deoxy[3H]glucose to tissues per unit mass, particularly skeletal muscle, during a GTT (Fig. 2j), demonstrating that their accelerated disposal of a glucose load was not due to increased muscle mass alone. Ucn3+ muscle showed increased glycogen storage and incorporation of [3H]glucose into glycogen during the GTT (Fig. 2k, l). Decreased phosphorylation of glycogen synthase at residue S641 indicated increased glycogen synthase activity (Fig. 2m).

Ucn3+ mice exhibit adequate counter-regulatory responses upon fasting

Since lower blood glucose levels were consistently found in fed and fasted Ucn3 + mice (Fig. 2e), we examined their counter-regulatory responses. Following fasting, circulating corticosterone and glucagon levels in Ucn3 + mice, as well as glucose production in response to glucagon administration to fed mice, were the same as in controls (ESM Fig. 2). Moreover, in Ucn3 + mice, fasting insulin was lower than in controls, showing appropriate downregulation in response to hypoglycaemia (Fig. 3a). Ucn3 + mice lost a greater proportion of liver weight (a measure of glycogen mobilisation and depletion) and showed less hepatic glycogen staining after fasting (Fig. 3b, c). Pepck (also known as Pck1) mRNA expression was upregulated appropriately in Ucn3 + mice after fasting (Fig. 3d) and phosphoenolpyruvate carboxykinase 1 (PEPCK) enzyme activity levels were increased when fed and fasted, consistent with gluconeogenesis being active at all times (Fig. 3e).

Fig. 3
figure 3

Ucn3+ mice have an adequate counter-regulatory response to fasting. a Fasting insulin downregulation (n = 9–12) in Ucn3+ mice is appropriate for blood glucose levels. b Decrease in liver weight (indicative of glycogen mobilisation) following fasting (n = 6–12) is increased in Ucn3+ mice. c PAS staining showed relative depletion of glycogen in the periportal area of Ucn3+ livers compared with controls (×4 magnification scale bar 500 μm; ×10 magnification scale bar 200 μm). dPepck mRNA and (e) PEPCK activity in fed and fasted Ucn3+ mice (n = 6–8). a, b, d, e Black bars, Ucn3+; white bars, control. *p < 0.05 for Ucn3+ vs control

Ucn3+ mice have normal insulin sensitivity

Lower fasting insulin/glucose ratio indicated the possibility of increased insulin sensitivity in Ucn3 + mice (Fig. 4a). However, during an ITT there was no difference in the magnitude of response to insulin, implying no difference in insulin sensitivity (Fig. 4b). Moreover, basal (fed) activating phosphorylation of the key signalling intermediates, Akt and IRS-1, in skeletal muscle and liver was decreased in Ucn3 + mice, presumably due to their lower insulin levels (Fig. 4c). While exogenous insulin caused the expected increase in IRS-1 and Akt phosphorylation, there were no differences between genotypes (ESM Fig. 2), again indicating no increase in insulin sensitivity. Total levels of IRS-1 and Akt did not differ with genotype (not shown). p70 S6 kinase (p70S6) phosphorylation as a measure of Akt–mechanistic target of rapamycin (mTOR)–p70S6 signalling pathway activation was not different from controls under basal conditions (ESM Fig. 2). Finally, euglycaemic–hyperinsulinaemic clamps demonstrated increased glucose uptake in Ucn3 + mice at a fixed insulin dose (Fig. 4e). However, euglycaemia could not be achieved even at maximal glucose infusion rates in Ucn3 + mice at this standard high physiological insulin dose (~450 pmol/l [21]), whereas euglycaemia was achieved in controls (Fig. 4d). This technique relies on the assumption that non-insulin-dependent glucose uptake is the same between experimental groups, but our in vivo data suggest that this was not the case in Ucn3 + mice. Hence further comparisons using this standard method for measuring insulin sensitivity were not attempted.

Fig. 4
figure 4

Ucn3+ mice do not show increased insulin sensitivity and are protected against adiposity and hyperglycaemia when on an HFD. a Insulin:glucose ratios are decreased in fasted Ucn3+ mice (black bars and squares) compared with controls (white bars and squares) (n = 10–12). b Change in blood glucose levels during an ITT in fed mice (n = 13–14). c Activating phosphorylation of insulin signalling intermediates in tissues from fed mice (n = 7). d Mean glucose level and (e) total glucose infused from 70–120 min in a glucose clamp (n = 5). f Body weight and (g) energy intake in kJ per 100 g body weight (BW) at 16 weeks of age. h Weight loss and (i) blood glucose levels following fasting, with (j) AUC glucose and (k) blood glucose levels during a GTT in mice fed LFD or HFD (n = 13–17). *p < 0.05 vs control and p < 0.05 vs LFD. f–k White, control LFD; black, Ucn3+ LFD; light grey, control HFD; dark grey, Ucn3+ HFD

Ucn3+ mice resist adiposity and hyperglycaemia on HFD

On control LFD from weaning to 16 weeks, Ucn3 + mice were heavier than controls, as seen in animals on normal chow. However, while controls gained weight despite reducing their energy intake on HFD, Ucn3 + mice resisted weight gain while maintaining their energy intake (Fig. 4f, g). Moreover, LFD-fed Ucn3 + mice lost a greater proportion of body weight following fasting, implying increased mobilisation of energy stores, a process that was not attenuated by HFD as in controls (Fig. 4h). Ucn3 + mice resisted the increase in fat mass on HFD (Table 2). Ucn3 + mice had lower fasting blood glucose than controls and resisted HFD-induced hyperglycaemia, showing lower glucose on both diets (Fig. 4i–k). In an ITT, there was no significant difference in the initial decrement between genotypes and therefore again no evidence of increased insulin sensitivity in Ucn3 + mice (ESM Fig. 2).

Table 2 Adiposity of control and Ucn3 + mice fed LFD or HFD

Mechanism for metabolic phenotype and muscular hypertrophy

Ucn2 (the endogenous urocortin in muscle) and Crfr2 mRNA expression were unchanged in Ucn3+ skeletal muscle (ESM Fig. 3). To further characterise the metabolic pathways involved in the phenotype of Ucn3+ mice, candidate gene expression was examined in skeletal muscle and liver. AMPK may improve glucose homeostasis [24], but its activity and mRNA expression were not altered in Ucn3+ muscle or liver (Table 3, ESM Fig. 3). Glut1 (also known as Slc2a1) and Glut4 (also known as Slc2a4) expression was also unchanged in Ucn3+ mice (Table 3, ESM Fig. 3). However, Ucp2 and Ucp3 mRNA was upregulated in muscle (Fig. 5a), suggesting the possibility of increased metabolic uncoupling or fatty acid metabolism. Cpt1a mRNA was greatly upregulated in liver (Fig. 5b), consistent with increased hepatic fat oxidation. mRNA levels of other metabolic genes were examined under conditions of HFD and LFD, and when fed or fasted (Table 3). While these experimental conditions regulated gene expression as expected, no effect of genotype was observed.

Table 3 Genes not altered in Ucn3 + mice compared with controls
Fig. 5
figure 5

aUcp2 and Ucp3 mRNA in muscle (n = 11–15). bCpt1a mRNA in liver (n = 6–8). c Quantification of myostatin protein levels in (d) western blot (C, control; U3, Ucn3+) (n = 11). eIgf1 mRNA (n = 11–15), and (f) IGF-1 protein in muscle and (g) IGF-1 plasma levels (n = 11). mRNA levels (a, b, e) are expressed as percentage of the average value of all samples. *p < 0.05 vs control and p < 0.05 vs fed. White, control fed; black, Ucn3+ fed; light grey, control fasted; dark grey, Ucn3+ fasted

Myostatin mRNA levels were unchanged in Ucn3 + muscle. However, myostatin propeptide levels were decreased (Fig. 5c, d). Intriguingly, Igf1 expression was upregulated ~80% in Ucn3 + skeletal muscle, but unchanged in liver (Fig. 5e, f). In contrast, serum levels of IGF-1 were only marginally increased (16%) (Fig. 5g). Circulating growth hormone levels were unchanged in Ucn3 + mice (ESM Fig. 3), thus indicating that CRFR2 may stimulate IGF-1 in muscle via a paracrine action that causes muscle hypertrophy and also stimulates glucose uptake independently of the growth hormone–IGF-1 axis.

Discussion

The model of chronic urocortin 3 overabundance presented here shows a novel metabolic phenotype, with lean body composition, increased skeletal muscle mass, increased disposal and use of glucose in muscle, and increased energy intake with reduced adiposity on an HFD. Strikingly, Ucn3 + mice have lower blood glucose levels and increased energy intake even under basal conditions, which Crfr2-, Ucn2- and Ucn3-knockout models do not exhibit until challenged with HFD or aged. Interestingly, Ucn3 + mice do not show increased insulin sensitivity, suggesting that augmented glucose uptake occurs largely independently of insulin action. This again contrasts with the knockout mice, all of which exhibit increased insulin sensitivity, albeit via different mechanisms [911, 14]. Accordingly, the metabolic alterations in Ucn3 + mice appear to be primarily in carbohydrate metabolism, with accelerated glucose uptake and metabolism by muscle, and increased RER reflecting greater proportional carbohydrate use [25].

An AMPK-dependent role for CRFR2 in stimulating thermogenesis in skeletal muscle has been described [13], suggesting a possible mechanism by which glucose uptake to muscle may be enhanced in Ucn3 + mice. However, we did not find evidence for increased AMPK activity, but did observe upregulation of uncoupling proteins (UCP) 2 and 3. These mitochondrial proteins are normally produced in skeletal muscle, where their physiological functions are highly controversial [26, 27]. UCP3 has been proposed to facilitate fatty acid translocation and oxidation, and to mitigate reactive oxygen species production. Its primary role is not generally considered to be thermogenesis, but it does have the capacity to stimulate heat production and glucose oxidation in vivo, all roles compatible with the phenotype of Ucn3 + mice. UCP2 is even less well understood, but is generally proposed to fulfil similar functions. Increased UCP activity, as observed in transgenic mice, will result in increased mitochondrial oxidative phosphorylation and glucose metabolism and thus uptake, as well as increased fatty acid metabolism [28, 29]. Conversely, enhanced glucose transport achieved by overabundance of GLUT4 results in increased UCP production and activity in skeletal muscle, which in turn raises glucose use and lowers blood glucose [30]. This is plausibly the effective mechanism in Ucn3 + mice, with the increase in glucose metabolism responsible for protection against the diet-induced obesity seen in control mice.

Transgenic mouse models of increased UCP gene expression or increased metabolic rate normally exhibit decreased body weight due to lean body composition, whereas a notable feature of Ucn3 + mice is their increased body weight, entirely due to increased muscle mass. This muscular hypertrophy helps explain their metabolic phenotype. Increased IGF-1 abundance in Ucn3 + skeletal muscle provides a plausible link between the muscular and metabolic phenotypes of the mice. Decreased myostatin expression may also contribute to muscle hypertrophy; indeed, myostatin-null mice show similar fasting glucose and body composition responses to HFD as do Ucn3 + mice [31]. However, myostatin is quantitatively much less affected than IGF-1 in Ucn3 + mice.

In addition to its anabolic effects, IGF-1 has long been known to exert metabolic effects similar to insulin, with increased glucose uptake and glycogen synthesis, and hypoglycaemia as seen in Ucn3 + mice [32, 33]. These effects are mediated via the IGF-1 receptor in skeletal muscle [34]. Overabundance of IGF-1 in muscle can cause a shift towards glycolytic metabolism [35], which may improve body composition and insulin sensitivity [36]. Thus IGF-1 may be responsible for the observed shift towards type 2 muscle fibres in Ucn3 + mice. Transgenic or pharmacological models of increased IGF-1 levels show evidence of increased flux through various signalling pathways including the phosphatidylinositol 3-kinase pathway in tissues [37, 38], whereas decreased IRS-1 and Akt phosphorylation was seen in tissues from Ucn3 + mice following fasting. This may be because the observed increase in IGF-1 production, and hence presumably action, is of a lesser magnitude to that seen in these other models, and the IGF-1 axis in Ucn3 + mice will still be subject to regulation by endogenous factors including feeding status. Likewise, no difference in Akt–mTOR–p70S6 protein synthesis pathway signalling was observed in Ucn3 + mice under basal conditions when adaptation of signalling activity might have been expected. Detailed dynamic signalling pathway studies under conditions of CRFR2 stimulation should further clarify the role of these signalling pathways.

While exogenous urocortin 1 administration can stimulate pituitary growth hormone release [39], the mechanism by which IGF-1 production is stimulated in Ucn3 + mice is more likely to be paracrine. Neither serum growth hormone nor hepatic IGF-1 levels were increased in Ucn3 + mice. It is possible that subtle differences in growth hormone may have been missed on the single time point sample due to the pulsatile nature of growth hormone secretion. However, while growth hormone or Igf1 transgenic mice show organomegaly and other aspects of an acromegalic phenotype [40, 41], transgenic mice overexpressing Igf1 specifically within muscle show myofibre hypertrophy without systemic features of acromegaly [42, 43]. CRFR2 agonists ameliorate muscle atrophy in a variety of pathological conditions, including denervation atrophy and cancer cachexia, and will also increase mass of normal muscles [4446]. While the mechanisms underlying this effect have not been described, it has been shown to be a paracrine effect, strongly suggesting that urocortin 3 acting locally at CRFR2 is responsible for the myocyte hypertrophy observed in Ucn3 + mice. IGF-1 produced within skeletal muscle appears to be released to the circulation, but this source is a minor contributor in comparison to liver [47]. The source of the small increase in serum IGF-1 in Ucn3 + mice is thus likely to be their increased muscle content.

Beneficial effects of IGF-1 in type 2 diabetes and other insulin-resistant states have been observed [37, 48]. Decreased hepatic gluconeogenesis contributes to this; however, liver contains no IGF-1 receptors and the main effect of IGF-1 is mediated in skeletal muscle, with effects on liver metabolism being secondary [49]. Likewise, evidence for altered hepatic fuel metabolism is seen in Ucn3 + mice when CRFR2 has not been demonstrated in hepatocytes [50]. The Ucn3 + liver phenotype is apparently one of increased fatty acid oxidation and gluconeogenesis, which would seem appropriate compensatory adaptations for the increased fuel catabolism in muscle tissue. The Crfr2-knockout phenotype has been ascribed to increased sympathetic drive from the central nervous system upregulating brown adipose tissue UCP1 [10]. However, this does not seem to be occurring in Ucn3 + mice. Nevertheless, altered sympathetic drive to splanchnic organs including liver cannot be entirely ruled out as contributing in part to their phenotype.

Overall, Ucn3 + mice demonstrate that CRFR2 can drive skeletal muscle hypertrophy and glucose uptake even in the face of dietary challenge. CRFR2 agonists may be useful therapies to maintain muscle mass and bypass the muscle insulin resistance that so commonly occurs, most notably with ageing.