Deletion of the Feeding-Induced Hepatokine TSK Ameliorates the Melanocortin Obesity Syndrome

Endocrine signaling via secreted hormones and metabolites is a fundamental feature of nutrient and energy homeostasis in mammals. Dedicated endocrine cells, such as pancreatic β-cells, sense the nutritional status of the body and release hormones that act on peripheral tissues and the central nervous system to regulate metabolic, behavioral, autonomic, and neuroendocrine systems to maintain homeostasis. Beyond these professional endocrine cells, peripheral metabolic tissues also provide an important source of endocrine factors. Adipose tissue hormones, such as leptin and adiponectin (1–3), hepatokines (4,5), nonendocrine gut-derived factors (6,7), and myokines (8), participate in nutrient sensing and coordinate key aspects of nutrient and energy metabolism. A common feature of these metabolic hormones is that their release and action are exquisitely responsive to nutritional cues. For example, leptin relays the status of energy storage in adipose tissue and acts on food intake and energy expenditure in part through the hypothalamic melanocortin-4 receptor (MC4R) pathway (9,10). The hepatokine fibroblast growth factor 21 (FGF21) is inducible in response to starvation and certain nutrients such as fructose (11,12). FGF21 elicits pleiotropic effects on tissue metabolism and nutrient preference by acting on peripheral tissues as well as the central nervous system (13,14). As such, a distributed endocrine network is emerging as the foundation for hormonal signaling that governs nearly all aspects of metabolic physiology.

We recently identified Tsukushi (TSK) as an inducible hepatokine that is responsive to stimulation of energy expenditure (15). TSK is a leucine-rich repeat-containing protein that has been implicated in neural development (16–18). Hepatic TSK gene expression and its plasma levels are upregulated by thyroid hormone, cold exposure, and adrenergic stimulation. Mice lacking TSK exhibit a hypermetabolic phenotype and are remarkably resistant to diet-induced obesity and its associated metabolic disorders. TSK-null mice display features of augmented thermogenesis in brown fat as a result of increased sympathetic innervation and adrenergic stimulation. Accordingly, expression of genes involved in thermogenesis, including uncoupling protein 1 (UCP1), was elevated in brown fat from TSK-deficient mice (15). TSK-null mice were also protected from diet-induced nonalcoholic steatohepatitis (NASH) (19). These observations raise the possibility that TSK may elicit its metabolic effects by acting on multiple target tissues, perhaps including central nervous system sites involved in the central control of energy expenditure. The profound effects of TSK inactivation on whole-body energy balance prompted us to explore the physiological mechanisms of TSK action, particularly its cross talk with central melanocortin signaling. In this study, we examined the role of TSK in a mouse model of the melanocortin obesity syndrome and elucidated its effects on feeding behavior and energy expenditure phenotypes previously characterized in both the mouse and human models of the syndrome.

Mice were maintained on Teklad 5001 laboratory rodent chow under 12/12 light/dark cycles. To induce NASH, mice of 2–3 months of age were fed the Amylin (Amylin Liver NASH [AMLN]) diet containing 40 kcal% fat (mostly palm oil), 20% kcal% fructose, and 2% cholesterol (D091 00310, Research Diets) for 5–6 months, as previously described (20,21). The TSK- and MC4R-null mutant mouse strains were generated and described in previous studies (17,22). Double heterozygous TSK- and MC4R-knockout (KO) mice were bred to generate the following four experimental groups: wild-type (WT), TSK-KO, MC4R-KO, and TSK- and MC4R-double KO (DKO) mice.

PCR amplification for TSK and MC4R genotyping was performed on templates of tail lysate with the following set of primers (F, forward; R, reverse): 5′-cccagcagtagcaacaacaa-3′ (TSK-F), 5′-gagcttgtaagtcccttgga-3′ (TSK-R), 5′-gatccccatcaagattatcg-3′ (LacZ-R), 5′-cactcggagcttccctgacccag-3′ (MC4R-F), 5′-gaccatggtttccgacccatt-3′ (MC4R-R), and 5′-ttccaagcctctgagcccaga-3′ (Neo-R). TSK-F, TSK-R, and LacZ-R amplified 267 base pair (bp) and 428 bp fragments identifying the WT and null allele, respectively, whereas MC4R-F, MC4R-R, and Neo-R amplified ∼400 and 540 bp fragments corresponding to the WT and null allele, respectively.

Mice were individually housed for food intake studies. The number of mice used in each experiment is described in figure legends. Unless otherwise stated, the study used male C57BL/6 mice aged 3–4 months. All mouse studies were performed according to procedures approved by the University of Michigan Institutional Animal Care and Use Committee.

Mice were put under isoflurane anesthesia and stereotaxically implanted with a stainless steel cannula (Plastics One, Roanoke, VA) into their right lateral ventricle (coordinates from bregma: lateral, 1.00 mm; anteroposterior, −0.460 mm; ventral, −2.20 mm). Following recovery from surgery, each mouse was administered 20 ng of angiotensin II (Sigma-Aldrich, St Louis, MO), and positive cannulation was confirmed by angiotensin-induced water intake. HS014 (Tocris Bioscience, Minneapolis, MN) was dissolved in sterile PBS and infused into the lateral ventricle at 0.5 µg per mouse in 500 nL volume around during the light cycle. Vehicle groups received the same volume of PBS. Blood was collected 6 h postinfusion in EDTA-containing tubes. Plasma was separated following centrifugation and stored at −80°C until further processing.

Setmelanotide was dissolved in PBS at a concentration of 0.5 mg/mL. Mice were injected by intraperitoneal (i.p.) route at 1 mg/kg setmelanotide (or PBS as control) prior to the dark cycle. Plasma was collected 5–6 h after injections.

Adipose tissue and hepatic gene expression analysis was performed using quantitative PCR (qPCR) analysis, as previously described (23,24). Briefly, total RNA from white adipose tissue was isolated using the PureLink RNA kit (Thermo Fisher Scientific). Total RNA from other tissues was isolated using TRIzol method. Then, 2 μg of total RNA was reverse-transcribed using Moloney murine leukemia virus-RT, followed by real-time qPCR using SYBR Green (Thermo Fisher Scientific).

Total lysates were prepared in a lysis buffer containing 50 mmol/L Tris (pH 7.5), 150 mmol/L NaCl, 5 mmol/L NaF, 25 mmol/L β-glycerol phosphate, 1 mmol/L dithiothreitol (DTT), and freshly added protease inhibitors. Protein concentrations were quantified by the bicinchoninic acid method using Protein Assay Dye Reagent Concentrate (Bio-Rad). Rabbit polyclonal TSK antibodies were generated using mouse TSK peptides and affinity purified before use. Additional antibodies used in this study were UCP1 (UCP11-A, Alpha Diagnostic), tyrosine hydroxylase (ab112, Abcam), and Hsp90 (sc-7947, Santa Cruz Biotechnology).

Liver and brown and white adipose tissues were fixed with 4% formalin, embedded in paraffin, and sectioned for hematoxylin-eosin staining. Images were taken using a light microscope. For brain immunostaining, transcardiac perfusion was performed with 4% paraformaldehyde before harvesting brain tissue. The brain tissues were fixed in 4% paraformaldehyde for 4 h at 4°C, followed by incubation with 30% sucrose overnight at 4°C before embedding in optimal cutting temperature compound. The brain and liver tissues were cut into 30-μm and 12-μm sections. The frozen sections were permeabilized with 0.1% Triton-X100 in PBS for 15 min at room temperature, rinsed with PBS, and treated with blocking buffer (10% normal goat serum in PBS with 0.01% Triton-X100) for 1 h before incubation with the first antibody c-fos (1:300; #2250, Cell Signaling Technology) for brain sections and decorin (1:200; AF-1060, R&D Systems) for liver sections at 4°C overnight. The slides were washed with PBS and subsequently incubated with Alexa Fluor conjugated with secondary antibody (1:300; Invitrogen) in blocking buffer for 1 h. The slides were washed and incubated with DAPI for 1 min and were mounted with mounting medium. Fluorescent images were obtained using a confocal microscope (Nikon A1). The periventricular nucleus (PVN) was identified using the 3rd edition of The Mouse Brain in Stereotaxic Coordinates by Paxinos and Franklin. The selected slices for each mouse corresponded to bregma −0.56 mm to −1.22 mm. Neurons immunopositive for Fos were counted manually after the PVN boundaries were outlined. Neurons positive for c-Fos were determined only when the c-Fos signal was colocalized with DAPI staining.

Liver triglyceride was extracted and measured using a commercial assay kit (Sigma-Aldrich). The hydroxyproline level in the livers was measured using the Hydroxyproline Colorimetric Assay Kit (BioVision), as previously described (25). Triglyceride and collagen content was normalized to liver tissue weight.

Statistical analysis was performed using GraphPad Prism 8 software. The two-tailed Student t test was used to analyze the differences between two groups. For the glucose tolerance test, two-way ANOVA with multiple comparisons was used for statistical analysis. P values of <0.05 were considered statistically significant.

Resources generated during this study are available from the corresponding author upon reasonable request. No data sets were generated during the current study.

We previously demonstrated that mice lacking TSK are remarkably resistant to diet-induced obesity and its associated metabolic disorders and exhibit augmented adipose tissue sympathetic innervation and thermogenesis (15). Despite these intriguing findings, whether TSK exerts its effects on energy balance through cross talk with the central regulatory circuitry remains unknown. To address this, we first explored nutritional regulation of hepatic TSK gene expression and its plasma levels under different feeding conditions. We subjected WT C57BL/6J mice to ad lib feeding, overnight starvation, or overnight starvation, followed by 6 h of refeeding. Analysis by qPCR of hepatic gene expression indicates that, as expected, mRNA expression of Srebp1c, a key regulator of de novo lipogenesis, was suppressed by fasting and increased in response to refeeding, while Fgf21 mRNA expression exhibited the opposite pattern of regulation (Fig. 1A). Tsk mRNA level was reduced by ∼50% after overnight starvation and rapidly recovered to ad lib levels after refeeding. Consistently, immunoblotting analysis revealed that starvation decreased TSK levels in circulation, which subsequently recovered to levels comparable to the ad lib group following refeeding (Fig. 1B). Induction of Tsk by feeding was also observed during diurnal feeding cycles in mice. Tsk mRNA expression in the liver exhibited a robust diurnal rhythm that peaked at the transition from the light to dark cycle, coinciding with the bulk of feeding activities in mice (Fig. 1C). Plasma TSK reached peak levels during the early dark phase and gradually returned to baseline levels observed in the light phase (Fig. 1D). This diurnal pattern of Tsk expression was nearly abolished in mice lacking Bmal1 in the liver (26); the latter serves as a core component of the molecular clock (Fig. 1E). As such, TSK is a downstream target of circadian timing cues and potentially serves as an effector endocrine factor of the body clock.

Feeding is known to trigger hypothalamic leptin-melanocortin signaling to suppress food intake, stimulate thermogenesis, and restore energy homeostasis (9,10). To determine whether hepatic TSK secretion is modulated by the central regulatory circuitry, we examined plasma TSK levels in mice lacking leptin (ob/ob), leptin receptor (db/db), or MC4R. Compared with respective littermate control, plasma TSK levels were elevated in all three models of defective leptin-melanocortin signaling (Fig. 2A).

We next examined whether pharmacological manipulations of MC4R regulate plasma TSK levels. As expected, i.p. injection of setmelanotide, an agonist for MC4R (27,28), acutely suppressed food intake in mice (Fig. 2B), whereas intracerebroventricular injection of Hs014, an MC4R antagonist (29), elicited an opposite effect on feeding (Fig. 2C). Immunoblotting analysis indicated that setmelanotide treatment decreased plasma TSK levels in mice fed ad lib (Fig. 2D). In accordance, pharmacological inhibition of MC4R by Hs014 resulted in elevated plasma TSK levels (Fig. 2E). The effects of MC4R on plasma TSK levels were not observed under fasting conditions, suggesting that melanocortin signaling exerts its effects on TSK via a mechanism dependent on food intake. These findings provide strong support for the regulation of circulating TSK levels by feeding activities and central melanocortin signaling.

We previously demonstrated that TSK-KO mice were hypermetabolic and resistant to diet-induced obesity (15,19). While activation of brown fat thermogenesis contributes to the lean phenotype, TSK-KO mice gained significantly less body weight than WT littermates when housed at thermoneutral conditions. These results suggest that TSK deficiency may regulate whole-body energy balance beyond adipose thermogenesis. We postulated that TSK may serve as a hepatic endocrine signal that acts on the central nervous system to elicit multifaceted effects on energy balance. To test this, we generated TSK- and MC4R-DKO mice to dissect potential cross talk between TSK and the melanocortin signaling pathway. Immunoblotting analysis indicated that the TSK protein was absent in plasma from TSK-KO and DKO mice (Supplementary Fig. 1). As expected, MC4R-KO mice exhibited hyperphagia and accelerated weight gain and developed severe obesity when fed standard rodent chow (Fig. 3A and Supplementary Fig. 1). TSK inactivation nearly completely reversed these canonical manifestations of MC4R deficiency in TSK- and MC4R-DKO mice. In fact, body weight gain and cumulative food intake in DKO mice were comparable with WT control (Fig. 3B and C). As such, TSK deficiency corrects hyperphagia and severe obesity in MC4R-null mice, suggesting that TSK may normally exert an inhibitory effect on the physiological actions of MC4R signaling.

MC4R neurons located in the PVN of the hypothalamus play an important role in mediating melanocortin signaling and its effects on energy homeostasis. Activation of the PVN MC4R neurons by feeding and dietary fat is associated with increased expression of c-Fos, an indirect marker of neuronal activity (30,31). To address whether TSK modulates neuronal activity in the PVN in response to feeding, we subjected a cohort of WT and TSK-KO mice to 2 h of refeeding following overnight starvation and examined c-Fos expression by immunofluorescence staining. As shown in Fig. 3D, we observed a markedly increased number of PVN neurons positive for c-Fos expression in TSK-KO mice compared with WT littermate control mice. Baseline c-Fos expression in fasted mice was comparable between the two groups. This increase in neuronal activity is linked to reduced food intake by TSK-KO mice during the refeeding period compared with WT control mice (Fig. 3E). To further evaluate the relationship between TSK and MC4R, we next examined c-Fos expression in WT, TSK-KO, MC4R-KO, and DKO mouse brains under the refed condition. Interestingly, the stimulatory effect of TSK deficiency on PVN neuronal activity was diminished in mice lacking MC4R (Fig. 3F), suggesting that TSK may mediate its effects in part via an MC4R-dependent mechanism. These results demonstrate that TSK deficiency alters neuronal response to feeding and support a potential role of TSK in modulating central melanocortin signaling.

We next performed metabolic analyses on TSK- and MC4R-DKO mice to assess the extent to which TSK inactivation restores metabolic health in MC4R-deficient mice. Compared with WT control, MC4R-deficient mice exhibited elevated blood glucose and plasma total cholesterol levels (Fig. 4A). While TSK-KO mice had comparable blood glucose and lipid levels as control mice, TSK ablation essentially normalized these parameters in DKO mice. Glucose tolerance test results indicated that TSK inactivation restored impaired glucose tolerance observed in MC4R-KO mice (Fig. 4B). Measurements of tissue weight indicated that expansion of liver and adipose tissue mass in MC4R mice was significantly reduced in DKO mice (Fig. 4C), reflecting reversal of obesity as a result of TSK inactivation. Similar to previous reports (32), chow-fed MC4R-KO mice developed hepatic steatosis, as evidenced by the presence of abundant lipid droplets in hepatocytes and increased liver triglyceride content (Fig. 5A and B). In contrast, liver histology and fat content in DKO mice were comparable to WT control mice. Hepatic steatosis in MC4R-KO livers was associated with increased mRNA expression for genes involved in lipid metabolism (Srebp1c, Cd36, and Cidec), inflammatory response (Ccl2, Ccl5, and Trem2), and liver fibrosis (Tgfb, Col1a1, Col1a2, Col3a1, and Mmp12). Remarkably, elevated expression of these genes in MC4R-null mouse livers was significantly diminished in the livers from DKO mice (Fig. 5C). These results illustrate that TSK inactivation is sufficient to restore key aspects of metabolic defects caused by MC4R deficiency.

Mice lacking MC4R exhibit reduced peripheral sympathetic tone that is associated with impaired adipose tissue thermogenesis (33,34). As such, brown adipocytes from MC4R-KO mice contained larger lipid droplets compared with WT control mice (Fig. 6A). In contrast, brown adipocytes from TSK-KO mice exhibited reduced lipid content that is linked to enhanced sympathetic innervation and thermogenic activation. We observed smaller lipid droplets in DKO brown fat than those observed in MC4R-KO brown adipocytes. mRNA expression of several genes involved in fuel oxidation and brown fat thermogenesis was partially restored, including Dio2, Ppargc1a, Cox7a1, Cox8b, and Cidea (Fig. 6B). Notably, mRNA and protein expression of UCP1 in DKO brown fat was elevated to levels observed in TSK-KO mice (Fig. 6B and C). The protein level of tyrosine hydroxylase, a marker for sympathetic nerve fibers, was also more abundant in brown fat from DKO mice than that of MC4R-deficient mice. In epididymal white adipose tissue, mRNA expression of Pparγ, Ppargc1a, and Glut4 was markedly decreased in MC4R-null mice; these impairments of gene expression were largely rescued by TSK inactivation. In parallel, mRNA levels for genes involved in adipose tissue inflammation (Ccl2, Cd68, and Adgre1) were drastically increased in obese MC4R-deficient mice, consistent with the presence of abundant crown-like structures (Figs. 5A and 6D). The induction of these inflammation-related genes was greatly attenuated in epididymal white adipose tissue from DKO mice. Together, these results illustrate functional interaction between TSK and MC4R signaling in the regulation of systemic energy balance and metabolic health.

A major manifestation of MC4R deficiency is the development of progressive hepatic steatosis, liver injury, and inflammation characteristic of NASH (32). To determine whether TSK may exert effects on this aspect of MC4R signaling, we fed WT, TSK-KO, MC4R-KO, and DKO mice the Amylin diet, a high-fat high-fructose diet that has been previously demonstrated to induce prominent NASH pathologies in mice (20,21,25). Compared with WT control, MC4R-KO mice developed more severe obesity, while TSK-KO mice exhibited reduced diet-induced weight gain (Fig. 7A). Body weight in mice lacking both MC4R and TSK reached levels similar to the WT group, indicating that TSK inactivation attenuated the obesogenic effect of MC4R deficiency in mice fed the Amylin diet. Similarly, liver weight and the concentrations of plasma triglycerides and total cholesterol were significantly lower in DKO mice than in the MC4R-KO group (Fig. 7B and D). A core feature of NASH pathogenesis is the presence of liver injury. Measurements of plasma levels of ALT and AST, two surrogate markers of liver injury, revealed that as expected, MC4R-KO mice developed more severe liver injury than WT control mice (Fig. 7E). This exacerbation of liver injury in MC4R-KO mice was significantly improved in mice lacking both MC4R and TSK.

Examination of liver histology indicated that while hepatic steatosis remained similar among the four genotypes, prominent liver fibrosis was observed in MC4R-KO mice (Fig. 8A). Remarkably, TSK inactivation suppressed diet-induced liver fibrosis in both the WT and MC4R-KO background. In support of this, Sirius red staining, quantitation of hydroxyproline content, and immunofluorescent staining for decorin, an extracellular matrix protein associated with liver fibrosis, revealed marked improvements of liver fibrosis in DKO mice compared with the MC4R-KO group (Fig. 8B and C). Accordingly, TSK inactivation also partially normalized the expression of genes involved in inflammatory response (Ccl2 and Tnfa) and fibrosis (Tgfb, Acta2, Timp1, Col1a1, Col3a1, Col4a1, and Col6a1) in DKO mouse liver (Fig. 8D). These results collectively demonstrate that TSK inactivation restores the impairments of whole-body energy balance and ameliorates the severity of metabolic disorders caused by MC4R deficiency.

TSK is an emerging endocrine factor secreted by the liver. Similar to other metabolic hormones, plasma concentrations of TSK are highly responsive to nutritional and metabolic status in the body. In this study, we demonstrate that hepatic TSK expression and its plasma levels were rapidly induced by feeding following overnight starvation in mice. These observations prompted us to generate TSK- and MC4R-DKO mice and explore potential cross talk between TSK and the melanocortin-MC4R signaling pathway. Remarkably, TSK inactivation restores energy balance, ameliorates hyperphagia, and improves peripheral metabolic parameters in MC4R-deficient mice. These findings strongly suggest that TSK may exert an inhibitory effect on central melanocortin signaling and its downstream physiological responses, including food intake and energy expenditure (Fig. 7E).

While the molecular identity of TSK receptor and its relationship with MC4R remains unknown, this study illustrates robust functional cross talk between these two pathways in the regulation of whole-body energy balance and metabolic physiology. Previous studies have established that mice lacking MC4R develop severe obesity as a consequence of hyperphagia and reduced energy expenditure (9,10). Further, MC4R-deficient mice develop pronounced insulin resistance, hepatic steatosis, and inflammation in adipose tissue and the liver. Genetic ablation of TSK rescued essentially all of these behavioral, autonomic, and metabolic defects caused by MC4R deficiency, suggesting that TSK may act to exert an inhibitory effect on the physiological action of melanocortin-MC4R signaling. In support of this, we observed increased c-Fos expression in PVN neurons following refeeding. Previous studies have demonstrated that MC4R neurons located in the PVN are highly responsive to feeding by upregulating c-Fos expression (30,31). Whether these c-Fos–positive neurons in TSK-KO mice represent bona fide MC4R neurons remains to be confirmed in future studies using MC4R reporter mouse strains. The central energy circuit receives diverse hormonal cues that collectively tune food intake and energy expenditure to maintain homeostasis. In this regard, TSK likely serves as a liver-derived metabolic signal that counterbalances the activation of the hypothalamic melanocortin circuit in response to feeding.

TSK is emerging as an inducible hepatokine that elicits potent effects on systemic energy balance. TSK-null mice are highly resistant to diet-induced obesity when placed on a high-fat diet (15). This striking resistance to diet-induced weight gain was also observed in TSK-KO mice fed a Western diet (unpublished data) and the Amylin NASH diet (19). Given this robust energy balance phenotype in both diet-induced and genetic obesity models, it is surprising that these effects were not observed in an independently generated TSK-null mouse strain (35). In contrast, both KO strains exhibited improved liver pathologies following diet-induced NASH (19). It is likely that differences in targeting strategies, genetic background, and housing conditions may contribute to the discrepancy of diet-induced obesity. Previous studies have demonstrated that TSK physically interacts with extracellular signaling molecules such as TGFβ and FGF (36). As such, it is possible that TSK may act as an intrahepatic paracrine signaling factor that modulates liver injury response and local tissue remodeling during NASH pathogenesis. Compared with MC4R-KO mice, which developed several core features of NASH pathologies including liver injury, inflammation, and fibrosis, TSK MC4R-DKO mice displayed a greatly improved NASH phenotype. Whether this amelioration of NASH severity is the consequence of central or local TSK action remains to be established. MC4R mutations have been associated with obesity and other metabolic disorders in humans. As such, it is tempting to speculate that blockade of TSK signaling may elicit beneficial metabolic effects in this population.

Acknowledgments. The authors thank Dr. Liangyou Rui (University of Michigan) for providing technical and scientific advice.

Funding. This work was supported by National Institutes of Health, National Institute of Diabetes and Digestive and Kidney Diseases (DK070332 to R.D.C., and DK118731 and DK102456 to J.D.L.). Core services used in this study were supported by the Michigan Nutrition and Obesity Research Center and the Michigan Diabetes Research Center National Institutes of Health National Institute of Diabetes and Digestive and Kidney Diseases grants P30-DK020572 and P30-DK089503.

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. Q.W., P.Z., I.C., and L.M. performed the studies. Q.W., R.D.C., and J.D.L. conceived the project and designed research. Q.W. and J.D.L. analyzed the data and wrote the manuscript. J.D.L. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of data analysis.