An Essential Role for the K+-dependent Na+/Ca2+-exchanger, NCKX4, in Melanocortin-4-receptor-dependent Satiety*

Background: The melanocortin-4 receptor, MC4R, regulates satiety, but the signaling mechanism is unknown. Results: K+-dependent Na+/Ca2+-exchanger, NCKX4, knock-out mice are lean and hypophagic and display MC4R-dependent, Ca2+-sensitive, constitutive activation of paraventricular nucleus neurons. Conclusion: Ca2+ signaling is required for MC4R action, and its dysregulation in Nckx4−/− mice leads to anorexia. Significance: MC4R-mediated satiety requires NCKX4-modulated Ca2+ signaling. K+-dependent Na+/Ca2+-exchangers are broadly expressed in various tissues, and particularly enriched in neurons of the brain. The distinct physiological roles for the different members of this Ca2+ transporter family are, however, not well described. Here we show that gene-targeted mice lacking the K+-dependent Na+/Ca2+-exchanger, NCKX4 (gene slc24a4 or Nckx4), display a remarkable anorexia with severe hypophagia and weight loss. Feeding and satiety are coordinated centrally by melanocortin-4 receptors (MC4R) in neurons of the hypothalamic paraventricular nucleus (PVN). The hypophagic response of Nckx4 knock-out mice is accompanied by hyperactivation of neurons in the PVN, evidenced by high levels of c-Fos expression. The activation of PVN neurons in both fasted Nckx4 knock-out and glucose-injected wild-type animals is blocked by Ca2+ removal and MC4R antagonists. In cultured hypothalamic neurons, melanocyte stimulating hormone induces an MC4R-dependent and sustained Ca2+ signal, which requires phospholipase C activity and plasma membrane Ca2+ entry. The Ca2+ signal is enhanced in hypothalamic neurons from Nckx4 knock-out animals, and is depressed in cells in which NCKX4 is overexpressed. Finally, MC4R-dependent oxytocin expression in the PVN, a key essential step in satiety, is prevented by blocking phospholipase C activation or Ca2+ entry. These findings highlight an essential, and to our knowledge previously unknown, role for Ca2+ signaling in the MC4R pathway that leads to satiety, and a novel non-redundant role for NCKX4-mediated Ca2+ extrusion in controlling MC4R signaling and feeding behavior. Together, these findings highlight a novel pathway that potentially could be exploited to develop much needed new therapeutics to tackle eating disorders and obesity.

K ؉ -dependent Na ؉ /Ca 2؉ -exchangers are broadly expressed in various tissues, and particularly enriched in neurons of the brain. The distinct physiological roles for the different members of this Ca 2؉ transporter family are, however, not well described. Here we show that gene-targeted mice lacking the K ؉ -dependent Na ؉ /Ca 2؉ -exchanger, NCKX4 (gene slc24a4 or Nckx4), display a remarkable anorexia with severe hypophagia and weight loss. Feeding and satiety are coordinated centrally by melanocortin-4 receptors (MC4R) in neurons of the hypothalamic paraventricular nucleus (PVN). The hypophagic response of Nckx4 knock-out mice is accompanied by hyperactivation of neurons in the PVN, evidenced by high levels of c-Fos expression. The activation of PVN neurons in both fasted Nckx4 knock-out and glucose-injected wild-type animals is blocked by Ca 2؉ removal and MC4R antagonists. In cultured hypothalamic neurons, melanocyte stimulating hormone induces an MC4R-dependent and sustained Ca 2؉ signal, which requires phospholipase C activity and plasma membrane Ca 2؉ entry. The Ca 2؉ signal is enhanced in hypothalamic neurons from Nckx4 knock-out animals, and is depressed in cells in which NCKX4 is overexpressed. Finally, MC4R-dependent oxytocin expression in the PVN, a key essential step in satiety, is prevented by blocking phospholipase C activation or Ca 2؉ entry. These findings highlight an essential, and to our knowledge previously unknown, role for Ca 2؉ signaling in the MC4R pathway that leads to satiety, and a novel non-redundant role for NCKX4mediated Ca 2؉ extrusion in controlling MC4R signaling and feeding behavior. Together, these findings highlight a novel pathway that potentially could be exploited to develop much needed new therapeutics to tackle eating disorders and obesity.
Ca 2ϩ homeostasis is a critical determinant of neuronal excitability via an influence on resting membrane potential, neurotransmitter release, and synaptic efficiency (1). Although ion channel activation is responsible for a rapid increase in Ca 2ϩ levels, longer term regulation of Ca 2ϩ depends mostly on extrusion pathways (2). Principal among these, particularly in neurons (3,4), is the family of K ϩ -dependent Na ϩ /Ca 2ϩ -exchangers (NCKX1 to -5), 2 encoded by the slc24 genes (5,6). NCKX family members are broadly expressed in various tissues, and particularly enriched in neurons of the brain. With the exception of NCKX1 (rod photoreceptors (7)) and NCKX5 (skin pigmentation (8)), the distinct physiological roles for the different family members are not well described. Recently published studies have shown an important role for NCKX4 (the product of the slc4a4 or Nckx4 gene) in olfactory function (9) and in tooth enamel formation (10). Our laboratory has independently generated Nckx4 Ϫ/Ϫ mice, and we demonstrate here that adult animals display a remarkable hypothalamic defect that results in a dramatic anorexic and hypophagic phenotype.
The hormonally regulated and neuronally encoded network that controls food intake, satiety, and energy homeostasis is the major determinant of body weight in both animal models and humans (11)(12)(13). Peripherally produced circulating factors, such as insulin and leptin, influence energy balance through their action on neuronal centers within the hypothalamus, primarily in the arcuate nucleus (ARC) (12). Here, two classes of neurons have antagonistic effects on feeding behavior. One class express neuropeptide Y and agouti-related protein (AgRP) and induce feeding, whereas the other class express pro-opiomelanocortin (POMC; the precursor to melanocortins, such as ␣-melanocyte stimulating hormone (MSH), adrenocorticotropic hormone and the endorphins) and suppress feeding (14). Leptin acts to inhibit AgRP neurons whereas stimulating POMC neurons. The targets of POMC and AgRP neurons are principally melanocortin-4 receptor (MC4R) expressing neurons found in the paraventricular nucleus (PVN) as well as other hypothalamic areas (11)(12)(13). MSH, released from POMC neurons, is an agonist for MC4R and acts to stimulate the firing rates of downstream neurons (15). AgRP, on the other hand, is an antagonist of MC4R, thereby inhibiting these neurons (11)(12)(13)15), although AgRP neurons have also been shown to induce feeding directly, without melanocortin involvement (14,16).
Loss of function of the MC4R in mouse models causes profound obesity, due to both hyperphagia and reduced energy expenditure (17). Targeted knock-out and re-expression of MC4R in the PVN has established this as the key hypothalamic site for the regulation of feeding behavior (18,19). Importantly, mutations in MC4R are the most abundant monogenetic cause of obesity in humans, and a relatively common cause of severe obesity in childhood (20). Moreover, there is emerging evidence that MC4R variants contribute to obesity more generally (21). These observations have focused attention on the central melanocortin system as the essential control network for the regulation of energy homeostasis (11), with MC4R-expressing neurons as the core element (22). The MC4R is a G-protein coupled receptor that can stimulate G␣ s and consequently activate adenylyl cyclase to increase cAMP levels (22). However, a causative role for cAMP in MC4R signaling that leads to satiety has not to our knowledge been demonstrated. Indeed, loss of G␣ s function in PVN does not influence feeding behavior (23,24). Thus, the mechanism that links MC4R signaling to changes in neuronal firing, and particularly to the control of feeding behavior, remains unknown.
We demonstrate here that genetically targeted Nckx4 Ϫ/Ϫ mice lacking the NCXK4 protein display a striking anorexic phenotype as a consequence of Ca 2ϩ -dependent heightened sensitivity of MC4R in the PVN. These data also demonstrate an essential role for Ca 2ϩ signaling downstream from MC4R in mediating feeding behavior and satiety.

EXPERIMENTAL PROCEDURES
Animals-Mice were maintained and treated in accordance with the guidelines of the Canadian Council on Animal Care as determined by the University of Calgary Animal Care Committee. Mc4r knock-out mice (loxTB Mc4r mice (19)) were purchased as heterozygous animals from Jackson Laboratory (stock number 006414). C57Bl/6 mice used for breeding with chimeric animals were obtained from Charles River. All mice were kept in a pathogen-free environment with a 12-h light/ dark schedule, and were maintained ad libitum on a commercial diet (20% protein, 9% fat, and 4% fiber; PMI Nutrition International LLC, catalog number 5062), with free access to water.
Generation of Nckx4 Ϫ/Ϫ Knock-out Mice-A conventional gene targeting approach was taken to produce a constitutive knock-out of the mouse slc24a4 (Nckx4) gene by deletion of exons 6 and 7, as illustrated in Fig. 1. Briefly, Nckx4 fragments were amplified from mouse 129/SvJ genomic DNA by PCR using a high-fidelity polymerase, and combined in pBluescript II SK(Ϫ) (Agilent Technologies) with a polymerase II promoter-driven neomycin resistance cassette (25), and the herpes simplex virus promoter-driven thymidine kinase gene derived from the pflox vector (26) (see Fig. 1). A linear NotI fragment of 11.7 kb was isolated, electroporated into E14 embryonic stem cells derived from the 129P2/Ola mouse line, and recombinants were screened using positive/negative selection as previously described (25). Targeted cells were then screened by Southern blot as described below to confirm homologous recombination. Blastocyst injections led to two male chimeras that were then bred with C57Bl/6 females, and agouti animals were screened by Southern blot to identify recombinant Nckx4 heterozygous animals. Subsequent animal genotyping was conducted using a mixed PCR primer protocol (see below). Both chimeric mice resulted in targeted Nckx4 heterozygotes, creating two independent lines. Most studies were performed on F2/F3 animals from one line, using littermates of different genotypes for comparison. The anorexia and hypophagia reported here was confirmed in animals from the second line (data not shown). Embryonic stem cell targeting and chimeric mouse production were conducted at the Clara Christie Center for Mouse Genomics, Faculty of Medicine, University of Calgary.
Mc4r ϩ/Ϫ heterozygous male mice were bred with Nckx4 ϩ/Ϫ heterozygous females to produce animals that were doubly heterozygous for both genes. Then the double heterozygotes were bred together sequentially to produce the double knockout. Mice at 3-4 months of age were used for food intake experiments and c-Fos staining, body weight was measured in 5-6month-old animals.
Chemicals-All chemicals used were analytical or molecular biology grade or better, and generally obtained from Sigma or VWR International, unless indicated otherwise.
Construction of Lentivirus Containing Mouse Nckx4 -The Nckx4 cDNA in pcDNA3.1ϩ (Invitrogen) has been previously described (27). A fragment from the pWPI vector (a kind gift of Dr. Didier Trono obtained as Addgene plasmid 12254) encompassing the central polypurine tract (cPPT), internal ribosome entry site (IRES), and enhanced green fluorescent protein (EGFP) loci was inserted at the 3Ј end of the Nckx4 cDNA. A fragment comprising the CaMKIIa promoter from the pLenti-CamKII-ChETA-EYFP (a kind gift of Dr. Karl Deisseroth obtained as Addgene plasmid 26967) (28) was inserted at the 5Ј end of the Nckx4 cDNA. The entire CaMKIIa-promoter-Nckx4-IRES-EGFP construct was then excised as an XbaI-EcoRI fragment and used to replace the corresponding CaMKIIa-promoter-ChETA-EYFP fragment from the pLenti vector. This construct was propagated in Stbl3 Escherichia coli cells (Invitrogen). A control construct was also generated in which most of the NCKX4 coding region was removed by BamHI digestion and ligation. These constructs were then packaged and amplified by transfection into HEK293T cells using the helper plasmids pMD2.G and psPAX2 (both kind gifts of Dr. Didier Trono obtained as Addgene plasmids 12259 and 12260) as described on the Trono laboratory website. Final viral particles were concentrated by centrifugation and collected on a 20% sucrose cushion in PBS.
Molecular Analyses-All molecular biology procedures were conducted essentially according to standard protocols (29). Genomic DNA was isolated by phenol/chloroform extraction from adult mouse liver, targeted ES cells, or mouse tail snips for genotyping. For Southern blot analysis, ϳ10 g of genomic DNA was digested overnight with NsiI, separated on a 1% agarose gel, transferred to nylon, and probed with a digoxigeninlabeled genomic fragment located upstream of the targeting Nckx4 sequence. This probe recognizes a 9.9-kb band in genomic DNA from wild-type animals, and a 4.9-kb band in the targeted allele (see Fig. 1). For mixed PCR genotyping, we employed a forward primer (P1: CCATACTAACGATGAT-GATGGTG) immediately upstream of exon 6, a reverse primer (P2: GAACACAGCAGAGCCCACAAT) located within exon 6, and a reverse primer (P3: CGGGCCTCGTTCATGAAT-ATTC) from the murine polymerase II promoter at the 5Ј-end of the neomycin resistance cassette. Amplification from wildtype genomic DNA resulted in a band of 175 bp between primers P1 and P2, whereas amplification from the targeted locus resulted in a band of 350 bp between primers P1 and P3. The PCR genotyping analysis of the loxTB Mc4r mice was performed in a similar manner, essentially according to the published report (19).
Total RNA was isolated by guanidinium isothiocyanate extraction and CsCl centrifugation. For Northern blot analysis, 10 g of total RNA was separated by formaldehyde-agarose gel electrophoresis, transferred to nylon, and analyzed at high stringency essentially as described previously (30) using a digoxigenin-labeled probe corresponding to exon 6 and exon 7 of mouse Nckx4. The probes and conditions for detection of Ncx1, Ncx2, Ncx3, Nckx2, and Nckx3 transcripts were as described previously (30).
For immunoblot analysis, a membrane fraction, enriched in synaptic markers, was isolated from homogenized whole brains of mice of each genotype according to published procedures (30). Aliquots of 25 g of this membrane fraction were separated by SDS-PAGE, transferred to nitrocellulose membranes, and probed with an antibody against NCKX4. The NCKX4 antibody was prepared by immunizing rabbits with a keyhole limpet hemocyanin-conjugated peptide (CVKEKPPYGKTPV-VMV) from the cytoplasmic loop of the mouse NCKX4 protein (amino acids 294 -308; N-terminal Cys residue added for conjugation) in the University of Calgary Faculty of Medicine Antibody Services Facility. Anti-NCKX4 antibody was affinity purified from serum with immobilized peptide using the SulfoLink kit (Pierce-Thermo Scientific).
In situ hybridization was performed with digoxigenin-labeled antisense/sense riboprobes corresponding to nucleotides 680 -1398 of mouse Nckx4 essentially as described previously (30) on 7-m coronal paraffin sections cut sequentially across the entire hypothalamic region of mouse brain.
Behavioral Screen-Ten age-matched mice of each genotype and sex at 4 -5 months of age were characterized using established methods (31). Measures of general health and behavior were determined: general appearance, body weight, whisker length and appearance, and fur appearance. Nest building, social huddling, and general behaviors including social interactions and fighting were monitored in the home cage for any aberrant activity. Evidence of tremors, twitches, or seizures was recorded. The following neurological and sensory reflexes were observed for individual animals: righting reflex, postural reflex, eye blink reflex, ear twitch reflex, whisker orienting reflex, forelimb placing reflex, hind limb placing reflex, acoustic startle reflex, tail flick response, and visual placing test. The animals were also subject to an open field test and measures of distance traveled, time spent resting, and time spent in ambulatory activity recorded (Auto-Track system; Columbus Instruments). No abnormalities were observed in any of the general health and behavioral parameters, nor were there statistically significant differences in the open field measures. A wire-hang test was also performed, but the difference in mass between genotypes confounded the analysis of these data, which are not reported.
Weight Gain Curves-A subset of mice were weighed immediately after birth (P0) and a tail snip taken for genotyping. All other mice were routinely weighed upon weaning when they were separated into cages by sex, identified by ear punch and genotyped. Subsequently, these mice were weighed individually once a week in the morning of the same day, from weaning up to 32 weeks of age. Separate sets of animals were allowed free access to either the normal "enriched" lab chow containing 9% fat by weight (PMI Nutrition International LLC, catalogue 5062), "low fat" regular lab chow (4.5% fat by weight; Harlan Teklad, catalogue 5002), or a high fat diet (22.8% fat by weight; Harlan Teklad, catalogue TD.06415). Diets were introduced at the time of mating of the heterozygous parents of the animals used in the studies. The data were classified according to gender and genotype to construct weight gain curves. Not all animals of all genotypes on all diets were measured at all time periods. Data are thus only reported up to 24 weeks of age. The high fat diet tended to be rejected such that animals fed this diet did not weigh more than those on the enriched 9% fat lab chow. These data are therefore not reported.
Animal Morphology-A select set of adult mice (5-6 months old) were sacrificed by injection with a lethal dose of sodium pentobarbital. The animals were then weighed, and their length determined from nose to anus by gently straightening the spine. Heart, kidneys, liver, and total abdominal cavity fat mass (bilateral epididymal, inguinal, and retroperitonal fat pads) were dissected from these animals, blotted dry, and weighed. For tissue histology, mice were perfused transcardially with heparin (0.8 USP/ml) in 0.1 M phosphate buffer, pH 7.5, for 30 min, followed by 4% paraformaldehyde in 0.1 M phosphate buffer. Tissue was removed, fixed overnight in the same solution, paraffin embedded, and 7-m sections were stained with hematoxylin and eosin. Tissue histology was performed in the Department of Pathology and Laboratory Medicine core laboratory at the University of Calgary.
Food and Water Intake-Food intake was measured on 5-6month-old mice, each housed two or three to a cage, and allowed to habituate for at least a week before any measurements were taken. Over a 2-week period, a sufficient amount of food for the period was weighed and provided to the mice. Each weekday morning, the remaining food was measured, for a total of 10 measurements. Cages were carefully monitored for spillage, which was negligible. For water consumption, a measuring tube was constructed by inserting a normal mouse drinking tube with a rubber stopper into a 50-ml Falcon tube bearing volume marks. Daily water intake was measured for mice housed two to a cage each weekday morning over 2 weeks, for a total of 10 measurements. Special attention was given to limit cage movement, to prevent water draining and spilling. The 10 measurements of food or water intake were averaged and normalized per mouse to generate an individual datum point per animal.
Hidden Item Olfaction Tests, Urine-scented Litter-This test was performed on 5-6-month-old mice, using a protocol adapted from published methods (32). A small amount of urine-scented litter from an animal of the opposite sex was hidden near one end of a clean cage (43 ϫ 25 ϫ 20 cm) filled with fresh litter to a depth of about 3 cm. An individual free-fed mouse was released at the other end of the cage, and the time taken before the mouse located and dug up the urine-scented litter was noted.
Buried Food Pellet-This test was performed on 7-9-monthold mice according to a previously published protocol (32). For three consecutive days before the experiment, a piece of Oreo cookie (Nabisco) was put into each cage, and checked next morning for consumption. Each tested mouse was then fasted for 24 h, and transferred into a clean cage containing fresh bedding for 5 min to allow the subject to acclimate to the cage. One piece of cookie (80 -90 mg) was buried ϳ1 cm beneath the surface, in a corner of a clean testing cage containing fresh bedding as above. Then an individual mouse was transferred to the far end of the test cage, and the time taken to find and start to eat the cookie piece was recorded.
Blood Hormone and Chemistry Measurements-Blood samples were collected from 4 -5-month-old overnight-fasted mice via cardiac puncture, clotted at room temperature for 30 min, and then spun down at 3000 ϫ g for 15 min. The serum samples were supplemented with 1/10 volume of proteinase inhibitor mixture plus final concentrations of 200 M PMSF and 0.3 M aprotinin, divided into aliquots, and stored at Ϫ80°C. Serum levels for glucose, triglyceride, cortisol, MSH, insulin, leptin, and total thyroxine were measured according to the manufacturer's instructions using kits from Sigma for glucose and triglyceride, ALPCO Diagnostics for cortisol and MSH, Linco Research for insulin and leptin, and MP Biomedicals for total thyroxine.
Glucose Tolerance Test-5-6-Month-old mice were fasted overnight and then injected with glucose solution (2 mg glucose/g body weight) into the peritoneal cavity. Glucose levels were measured in blood collected from the tail immediately before and 15, 30, 60, and 120 min after the injection using OneTouch blood glucose strips (OneTouch FastTake, Life Scan).
Body Temperature Measurements-Body temperature in 5-6-month-old mice was measured at ϳ9 -10 a.m. each morning rectally using a BAT-12 microprobe thermometer from Physitemp Instruments.
Immunofluorescent Staining-Mice were anesthetized with pentobarbital and perfused transcardially with heparin (0.8 USP/ml) in 0.1 M sodium phosphate, pH 7.5, for 30 min, followed with 4% paraformaldehyde in 0.1 M sodium phosphate buffer, and the brains were removed and fixed overnight in the same solution. For antibody staining following in vitro perfusion conditions (treatments with insulin, MSH (Calbiochem, catalog number 05-23-075), SHU9119 (Bachem, catalog number H-3952), SKF96365 (Sigma), U73122 (EMD-Millipore), and Ca 2ϩ -free Krebs-Henseleit plus 0.5 mM EGTA), mice were perfused transcardially for 30 min with heparin (0.8 USP/ml) in Krebs-Henseleit solution (124 mM NaCl, 3 mM KCl, 26 mM NaHCO 3 , 2 mM CaCl 2 , 2 mM MgCl 2 , 1.25 mM NaH 2 PO 4 , 10 mM glucose, gassed with 95% O 2 , 5% CO 2 ). Following treatment, animals were fixed by perfusion with 4% paraformaldehyde in 0.1 M sodium phosphate buffer, and the brains were removed and fixed overnight in the same solution. Immunofluorescent staining was performed on 50-m free-floating brain coronal sections encompassing the entire hypothalamus area cut by vibratome. The sections were blocked for 2 h in working solution (3% donkey serum, 2% dimethyl sulfoxide, and 0.2% Tween 20 in 0.1 M sodium phosphate), incubated overnight at 4°C with primary antibodies (against MSH (Bachem, catalog number T-4434) used at a dilution of 1:4000; AgRP (Phoenix Pharmaceuticals, catalog number H-003-57) used at a dilution of 1:500; c-Fos (Calbiochem, catalog number PC38 (Ab-5)) used at a dilution of 1:5000; oxytocin (EMD-Millipore, catalog number MAB5296) used at a dilution of 1:600) in working solution. After washing, the sections were incubated overnight at 4°C with 1:500 dilutions in working solution of the appropriate Cy3 (Jackson ImmunoResearch Laboratories) species-related secondary antibody. After final washing, the sections were mounted in Vectashield anti-fade medium (Vector Labs) and viewed on a Zeiss Axioskop 2 microscope and fluorescent images were captured with a Spot RT3 camera system (Diagnostic Instruments).
Cell Culture, Ca 2ϩ Imaging, and cAMP Determination-All cell culture solutions and reagents were obtained from Invitrogen. Immortalized murine hypothalamic GT1-7 cells (33) (a kind gift from Dr. Richard Weiner, University of California at San Francisco) were cultured in DMEM supplemented with 10% fetal bovine serum. Cells in either multiwall dishes or on poly-D-lysine-coated coverslips were grown to confluence and then used for all assays.
For lentiviral infection, GT1-7 cells at ϳ70 -80% confluence were washed with Opti-MEM, then treated with lentivirus (multiplicity of infection of about 1) and Polybrene (8 g/ml) in Opti-MEM for 6 h. Three volumes of regular culture medium was added to the virus-treated wells overnight, and this mixture was replaced with fresh culture medium the next morning. The cells were incubated a further 2 days before being used for all assays.
Primary hypothalamic neuronal cell cultures were obtained by dissecting the hypothalamus from E18 -19 fetal or P0 neonatal mouse brain in ice-cold Hank's balanced salt solution (using the anterior margin of the optic chiasm, the posterior margin of the mammillary bodies, and the lateral hypothalamic fissures as guides), then digested with papain (Worthington Biochemicals) at 61.3 units/ml in basal medium Eagle (BME) containing 10% fetal bovine serum, and triturated with a fine glass pipette. The dissociated cells were pooled from the brains of 6 pups from each pregnant female, evenly plated on poly-Dlysine-and laminin-coated coverslips in 24-well plates and cultured in Neurobasal medium supplemented with 2% B27. The cultured neurons reached full confluence at 12-15 days, and were used for Ca 2ϩ imaging and cAMP assays at this point.
For Ca 2ϩ imaging, the cells were loaded with 5 M fura-2 AM (Molecular Probes) for 45 min at room temperature, mounted in a Warner Instruments chamber on a Zeiss Axioskop FS2 microscope, and perfused with physiological solution containing 135 mM NaCl, 5 mM KCl, 2 mM MgCl 2 , 2 mM CaCl 2 , 10 mM Hepes, pH 7.5, and 30 mM glucose. Fura-2 ratio images were acquired at 10-s intervals at room temperature using the EasyRatioPro system from Photon Technology International. Data were analyzed by placing regions of interest around responding cells.
For the cAMP assay, cultured cells were washed and incubated with Opti-MEM containing 0.5 mM isobutylmethylxanthine for 10 min, then stimulated with various concentrations of MSH for 15 min at 37°C. The medium was aspirated, and the cells were lysed in 0.1 M HCl. cAMP levels were measured following the manufacturer's instructions using ELISA kits from Biovision Research Products.
Statistics-Data were analyzed by either t test (in the case of only two test groups) or analysis of variance followed with the Bonferroni post-test to determine statistical significance, calculated using either GraphPad Prism or Microsoft Excel.

RESULTS
Nckx4 Ϫ/Ϫ Mice Are Anorexic and Hypophagic-The expression of NCKX4 was disrupted constitutively using a conventional gene targeting approach directed to the slc24a4 (Nckx4) gene (see Fig. 1). This was confirmed at the transcript level in brain, eye, and lung (tissues that normally express higher levels of Nckx4 (34)) using a probe from the deleted exons, and at the protein level in brain using an anti-peptide antibody. These data demonstrate complete absence of the NCKX4 protein in targeted Nckx4 Ϫ/Ϫ animals (Fig. 1C). The Nckx4 gene is broadly expressed in brain (see Fig. 2A), but there was no obvious difference in brain morphology in the knock-out animals (see Fig. 2B), nor any compensatory change in expression of paralogous Na ϩ /Ca 2ϩ -exchanger genes in Nckx4 Ϫ/Ϫ mice (Fig. 2C). The knock-out mice were viable, overtly healthy and fertile, and heterozygous breeding resulted in the expected Mendelian ratio of Nckx4 inheritance. A basic standard behavioral screen (31) displayed no abnormalities in the knock-out animals.
In contrast to the otherwise normal phenotype of the mice, an obvious reduction in body mass of the Nckx4 Ϫ/Ϫ animals was immediately evident (Fig. 3). Although not present at birth (Fig. 3B), the difference was already significant at weaning (4 weeks of age; Fig. 3B), and reached a sustained reduction of about 25-30% at 4 months of age (Fig. 3C). A similar fractional reduction in body mass was observed when knock-out animals were fed either standard lab chow (4.5% fat) or an enriched diet (9% fat) (Fig. 3A). Heterozygous Nckx4 animals were not different in weight from wild-type. In contrast to the reduction in the weight of knock-out animals, there was no change in their body length (Fig. 3C). The reduction in body weight was accompanied by a proportional reduction in food consumption, and a modest but statistically significant reduction in water intake (Fig. 3D). The knock-out mice also displayed reduced abdominal fat tissue with smaller adipocytes, and smaller, less fatty, livers (Fig. 4). The Nckx4 Ϫ/Ϫ animals did not display elevated body temperature (Fig. 4A) nor higher than normal T4 levels (Fig. 6A), and thus the reduced body mass appears to be almost entirely a result of the hypophagic phenotype.
The Hypophagia of Nckx4 Ϫ/Ϫ Mice Is Not Secondary to Other Health Issues-The knock-out mice do not display any pervasive development defect, as indicated by their normal birth weight, normal length, and normal brain morphology (see Figs.  2 and 3). In addition, the knock-out animals display improved glucose utilization, and a modestly extended lifespan (Fig. 5, A  and B), consistent with previous studies on animals fed calorierestricted diets (35). Moreover, motivational problems do not seem to underlie the hypophagia. First, Nckx4 knock-out mice demonstrated a paradoxical small increase in food consumed immediately following a 24-h fast (Fig. 5C). Similar behavior has been observed in other genetically lean mice (36), and has been attributed to depleted energy reserves in fasted animals. Second, the knock-out mice performed slightly better than controls in the task of finding hidden urine-scented litter (Fig.  5D). On the other hand, the Nckx4 Ϫ/Ϫ mice took longer to find a buried food pellet (Fig. 5D).

MC4R-dependent and Ca 2ϩ -sensitive Hyperactivation of PVN Neurons Underlies the Nckx4
Ϫ/Ϫ Phenotype-Chronic elevation of peripheral adiposity factors would lead to hypophagia (11)(12)(13). However, insulin, leptin, glucose, and MSH were all significantly lower in the Nckx4 Ϫ/Ϫ mice than in wild-type controls (Fig. 6A). The direction of these changes is indicative of a fasted, or semi-starved, state that would normally increase feeding, rather than reduce it.
In the hypothalamic ARC, satiety and feeding cessation are signaled from POMC neurons by an increase in the synthesis and release of anorexigenic MSH, and by inhibition of the synthesis and release of orexigenic AgRP and neuropeptide Y from their cognate neurons (11)(12)(13). Immunofluorescent analysis of brain sections illustrates the typical punctate pattern of packaged peptide hormone expression (Fig. 6B). However, the knock-out animals displayed paradoxically decreased MSH and increased AgRP levels in the ARC compared with wild-type, which therefore rules out these mediators as the cause for hypophagia.
The PVN is one of the major sites of neuronal activation upon feeding (37), and a selective role in food intake has been demonstrated here for the MC4R (11,18,19). Therefore, we examined neuronal activation in PVN under different physiological states in both wild-type and Nckx4 mutant mice, using c-Fos expression as a marker. As seen in Fig. 7, glucose infusion into fasted wild-type mice induced an obvious and highly significant 4-fold increase in the number of c-Fos positive cells, which was blocked by treatment with the MC4R antagonist, SHU9119. Infusion of insulin to previously fasted mice also induced a similar increase in c-Fos expression in the PVN, suggesting the effect of glucose may be in part secondary to insulin release. In contrast, fasted Nckx4 knock-out mice displayed a density of c-Fos staining in the PVN that was equivalent to the numbers observed in the glucose-or insulin-treated wild-type controls (Fig. 7). Moreover, the constitutive activation of PVN neurons in the knock-out was blocked by SHU9119 treatment, even though MSH release is expected to be suppressed under these conditions. This observation suggests that the MC4R in PVN is constitutively activated, or hyper-sensitized, in Nckx4 knock-out mice. Ca 2ϩ chelation with EGTA also reduced the level of c-Fos expression observed in the knock-out, indicating a role for altered Ca 2ϩ homeostasis due to loss of the NCKX4 protein. These data indicate that constitutive hyperactivation of PVN neurons, which is both MC4R-dependent and Ca 2ϩsensitive, plays a key role in the hypophagic phenotype of Nckx4 Ϫ/Ϫ mice.
To test further the role for MC4R in the Nckx4 knock-out phenotype, Nckx4 Ϫ/Ϫ and Mc4r Ϫ/Ϫ (19) mice were bred together to create a double knock-out. While Nckx4 Ϫ/Ϫ mice displayed decreased food consumption and body weight, and FIGURE 2. Nckx4 knock-out does not change brain morphology or the expression of other Na ؉ /Ca 2؉ -exchangers. A, serial coronal sections through the hypothalamus of wild-type adult mice were labeled for Nckx4 transcripts with an antisense or a sense control probe, as indicated. B, hematoxylin and eosin-stained coronal paraffin sections through the PVN (wild-type, ϩ/ϩ; Nckx4 knock-out, Ϫ/Ϫ). No differences were observed between genotypes in this or in other brain regions examined in either coronal or parasagittal stained sections (data not shown). In both panels A and B, the right-hand images are magnified views of the area in the box in the corresponding left-hand images. Scale bars: left-hand images, 1 mm; right-hand images, 200 m. C, Northern blots of 10 g of total RNA samples isolated from adult brain of wild-type (ϩ/ϩ), heterozygous (ϩ/Ϫ), or Nckx4 knock-out (Ϫ/Ϫ) mice were analyzed with probes for the indicated Na ϩ /Ca 2ϩ -exchangers genes. The blots were then stripped and reprobed for Gapdh as a loading control. All images in this figure are representative of independent experiments from 3 different animals of each genotype.

NCKX4 Function Is Essential for Normal Feeding Behavior
increased c-Fos expression in the PVN, Mc4r Ϫ/Ϫ mice displayed the opposite responses: increased food consumption and body weight, and decreased c-Fos expression in PVN (Fig.  8). The double knock-outs were indistinguishable in response to the Mc4r Ϫ/Ϫ mice and significantly different from either Nckx4 Ϫ/Ϫ mice or wild-type controls (Fig. 8). These data support an essential role for MC4R signaling in the etiology of the Nckx4 knock-out phenotype.

MC4R-induced Ca 2ϩ Signals in Hypothalamic Neurons-
The loss of exchanger-driven Ca 2ϩ extrusion would be anticipated to influence Ca 2ϩ homeostasis and signaling. Therefore, Ca 2ϩ signaling was examined in cultured hypothalamic neurons from wild-type and Nckx4 knock-out mice. Treatment with MSH induced a Ca 2ϩ response in a minority of hypothalamic neurons, consistent with the fraction of cells expected to express MC4R (19). The character of the response was highly variable from cell to cell within one microscopic field, with individual cells displaying either a single transient spike; repetitive, prolonged, oscillatory behavior; or delayed but sustained Ca 2ϩ elevation (Fig. 9A). Extensive time control experiments confirmed that these Ca 2ϩ signals were dependent upon treatment with MSH. The Nckx4 Ϫ/Ϫ -derived cultures consistently displayed a larger number of cells responding to MSH stimulation per field (Fig. 9B). Otherwise, there were no obvious alterations in the magnitude or time course of Ca 2ϩ signals in the knockout cells, including their response to KCl treatment. Collectively, these data indicate that loss of NCKX4 leads to a highly selective MSH hyper-responsive state that involves altered Ca 2ϩ signaling, and which can account for the constitutive activation of PVN neurons and downstream satiety pathways in the Nckx4 Ϫ/Ϫ animals.
MC4R-positive neurons represent only a minor component of all hypothalamic neurons, and so to study further the Ca 2ϩ response to receptor stimulation, we utilized an immortalized murine hypothalamic cell line, GT1-7 (33), known to express MC4R (38), but which has lost NCKX4 expression. GT1-7 cells were infected with lentivirus expressing mouse NCKX4 or a control construct, and Ca 2ϩ signals were recorded in response to MSH stimulation. As seen in Fig. 9C, these signals were very similar, in their variable and oscillatory temporal behavior, to those observed in the cultured hypothalamic neurons. Expression of NCKX4 in GT1-7 cells caused a significant reduction in the number of responding cells (Fig. 9D), analogous to the observations in hypothalamic neurons from wild-type versus knock-out mice.
The Ca 2ϩ signals in response to MSH treatment were examined in more detail, as shown in Fig. 10. The response typically continued long after wash-out of MSH (Fig. 10, A, D, G, and H), and was completely blocked by the MC4R antagonist, SHU9119 (Fig. 10B). The induced Ca 2ϩ signals required the presence of extracellular Ca 2ϩ , such that its removal both abolished previously activated signals and prevented the development of signals when removed prior to the MSH stimulus (Fig. 10, C and  D). Importantly, there was no evidence for transient Ca 2ϩ release from ER stores in response to MSH in the absence of extracellular Ca 2ϩ (Fig. 10D). Subsequent re-addition of Ca 2ϩ after MSH had been washed out led to resumption of the typical oscillatory Ca 2ϩ response (Fig. 10D). The MSH-induced Ca 2ϩ response was also prevented by prior treatment with the phospholipase C (PLC) inhibitor, U73122, or with the TRPC-selective Ca 2ϩ -entry inhibitor, SKF96365, but not with the voltagedependent Ca 2ϩ channel inhibitor, nifedipine, or the cAMP FIGURE 3. Nckx4 ؊/؊ mice are anorexic and hypophagic. Knock-out (Ϫ/Ϫ) and wild-type (ϩ/ϩ) and heterozygous (ϩ/Ϫ) littermates, as indicated, were fed ad libitum either standard or enriched laboratory chow (4.5% or 9% fat). A, animals were weighed weekly from 4 weeks of age onwards (n ϭ 12-15 for each group; however, measurements were not made for all animals for all time points, all conditions or both sexes; therefore, some data points correspond to a smaller n value). B, animals were weighed at birth (P0; n ϭ 10) and then at weaning (n ϭ 63, 107, and 56 for wild-type, heterozygous, and knockout, respectively). C, weight (n ϭ 8) and length (nose to anus; n ϭ 10) were measured on adult animals. D, food (n ϭ 8 for females and n ϭ 10 for males) and water consumption (n ϭ 6) were measured in adult animals. In all cases the mean Ϯ S.E. is plotted. Statistical significance of differences between knock-out and wild-type is indicated where present: *, p Ͻ 0.05; **, p Ͻ 0.01; ***, p Ͻ 0.001. antagonist, R p -cAMP (Fig. 10, E-H, respectively). Furthermore, MSH induced a dose-dependent increase in cAMP levels in GT1-7 cells, which was not influenced by prior treatment with either U73122 or SKF96365 (Fig. 9E). These data support the generation of dual and mutually independent MC4R-mediated cAMP and Ca 2ϩ signals, similar to dual signaling reported for the metabotropic glutamate receptor (39). The MC4R Ca 2ϩ signal does not arise from intracellular release, but depends upon surface entry activated by a long-lived PLC-generated mediator, possibly diacylglycerol. Such a pathway for activation of Ca 2ϩ entry has been shown for the human transient potential-like channels, TRPC3 and TRPC6 (40).
Ca 2ϩ -and NCKX4-dependent Regulation of Oxytocin Expression in the PVN-To determine whether altered MC4R-induced Ca 2ϩ signaling is responsible for the anorexic phenotype of the Nckx4 Ϫ/Ϫ mice, we examined oxytocin (OXT) expression in PVN neurons, where it appears to play an essential MC4R-dependent role in feeding behavior, downstream from both ARGP-and POMC-neurons of the ARC (16,41). Consistent with these data, we found that treatment of fasted wild-type mice with MSH induced a more than 2-fold increase in the number of PVN neurons expressing high levels of OXT (Fig.  11). Fasted Nckx4 Ϫ/Ϫ mice contained a similar number of neu-rons expressing high levels of OXT as the MSH-treated wildtype, consistent with a constitutively induced level of expression. In both cases, inhibition of PLC with U73122 or surface Ca 2ϩ entry channels with SKF96365, treatments that prevent MSH-induced Ca 2ϩ signals but not cAMP signals, reduced the number of neurons expressing high levels of OXT to control values (Fig. 11). Thus, MSH-induced OXT expression in hypothalamic PVN neurons and the consequent suppression of food intake depend upon M4CR-generated Ca 2ϩ signals. These signals are present constitutively in the Nckx4 knockout mice, resulting in the observed hypophagic phenotype.

DISCUSSION
In this article we have demonstrated that loss of the K ϩ -dependent Na ϩ /Ca 2ϩ -exchanger protein, NCKX4, results in a specific and dramatic hypophagic response, leading to reduced fat content and lower weight in Nckx4 Ϫ/Ϫ mice. We observed similar reductions in feeding and weight in two independently derived Nckx4 knock-out lines and similar data on body weight were published recently by another group using a different gene targeting strategy (9). Our data rule out changes in circulating adiposity molecules and ARC mediators as a cause for the hypophagic behavior (see Fig. 6). The animals do not demonstrate a pervasive developmental defect, any serious health issues, nor any obvious change in morphology that might lead to the observed reduction in food intake (Figs. 2, 4, and 5). Neither do motivational factors, which can influence food intake, appear to influence the behavior of the Nckx4 knock-out mice (Fig. 5). Thus we conclude that the loss of NCKX4 protein expression leads directly to highly specific neuronal changes underlying the reduction in food intake.
A recently published study (9) demonstrated a similar reduction of body weight in global Nckx4 knock-out mice, an altered olfactory response, and a partial deficit in olfactory function that led to an ensuing nursing defect. Olfactory-specific Nckx4 knock-out animals had lower body weight at weaning, although the authors did not report body weight in adult olfactory-specific knock-out animals. Our experiments on Nckx4 knock-out animals revealed a similar deficit in olfactory performance when presented with hidden food following a fast, but not with hidden urine-scented litter (Fig. 5D). Although our data are consistent with a partial loss of olfactory function in Nckx4 Ϫ/Ϫ mice that contributes to hypophagia, such a deficit cannot underlie the main change in feeding behavior and body weight in adult animals for the following reasons. Unlike Stephan et al. (9), we note a progressively increasing fractional difference in body weight from weaning age to adulthood (Fig. 3A). We have also directly measured a reduction in food consumption by adult Nckx4 Ϫ/Ϫ mice (Fig. 3D). Because these animals have no difficulty finding visible food, as demonstrated in a re-feeding following the fasting experiment (Fig. 5C), we conclude that reduced olfactory function does not lead to reduced food consumption in an ad libitum feeding scenario for adult animals. Finally and most importantly, we have demonstrated an increase in both c-Fos and OXT expression in the hypothalamic PVN neurons of fasted Nckx4 Ϫ/Ϫ mice that could not arise as a consequence of an olfactory deficit leading to reduced food intake (a scenario that would be expected to reduce c-Fos and OXT expression, as observed in fasted wild-type mice, see Figs. 7 and 11).
Instead, our data demonstrate that the main change in feeding behavior in the Nckx4 Ϫ/Ϫ mice is caused by a constitutive activation of PVN neurons, in an MC4R-dependent and Ca 2ϩsensitive manner, which induces an increase in OXT expression and a consequent anorexigenic response. That loss of NCKX4-driven Ca 2ϩ efflux could have such a selective behavioral response is surprising. We had previously demonstrated that Nckx4 expression is quite widely distributed among brain regions (34). This is also shown in Fig. 2A, where Nckx4 tran- FIGURE 5. Health parameters, motivation and olfaction in Nckx4 ؊/؊ knock-out and wild-type (؉/؉) mice. A, a glucose tolerance test was performed on 5-6-month-old animals following overnight fasting. Glucose (2 mg/g body weight) was injected intraperitoneally, and blood glucose was measured just prior and at intervals thereafter. Average values Ϯ S.E. are plotted (n ϭ 6). B, a Kaplan Meier survival curve is plotted for 18 wild-type (ϩ/ϩ) and 13 Nckx4 knock-out (Ϫ/Ϫ) animals. A log-rank (Mantel-Cox) test indicates a significant difference between genotypes, p Ͻ 0.05. C, cumulative food consumption was measured in adult mice immediately following a 24-h fasting period (n ϭ 11). D, time taken to find buried urine-scented litter (left) or buried food pellet (right) was measured for adult mice (n ϭ 11-13). Statistical significance of the differences between genotypes determined by analysis of variance using the Bonferroni post-test is indicated where present: *, p Ͻ 0.05; ***, p Ͻ 0.001. scripts can be readily observed in the PVN. Various other Na ϩ / Ca 2ϩ -exchangers are also broadly expressed in brain neurons, although their expression is not influenced by the loss of the NCKX4 protein (Fig. 2C). Moreover, no anorexic phenotype comparable with that of the Nckx4 Ϫ/Ϫ mice has been observed in other Na ϩ /Ca 2ϩ -exchanger knock-out models (30,(42)(43)(44)(45). Furthermore, we do not observe a general defect in Ca 2ϩ extrusion from neurons isolated from Nckx4 Ϫ/Ϫ animals (data not shown). Thus, our results suggest that within PVN neurons NCKX4 plays a special and non-compensated role in regulating Ca 2ϩ flux and signaling associated tightly with MC4R signaling linked with feeding behavior and satiety. PVN neurons have previously been shown to play a critical role in the MSH-induced anorexic response (18,19). This nucleus comprises a heterogenic population of neurons extending processes to other hypothalamic areas, to the pituitary, as well as to nuclei in the brain stem and spinal column (46).  Although a fully integrated feeding response clearly requires multiple pathways and neuronal connections, recent studies have demonstrated an essential role for PVN OXT neurons in normal satiety-induced suppression of feeding (16,41). Because intracerebroventricular delivery of OXT is effective in suppressing feeding, it is likely that OXT actions are mediated by pathways that connect the PVN to the brainstem and/or spinal cord, rather than those that project to the posterior pituitary (16,41). Our data demonstrating an MSH-dependent increase in PVN OXT expression (Fig. 11) are consistent with those findings. In humans, the obesity of Prader-Willi syndrome is thought to be due to a deficiency in OXT neurons (47), and OXT administration may also be effective in the treatment of obesity (48).
The MC4R is a G-protein coupled receptor capable of signaling through G␣ s to cAMP and protein kinase A (22). However, loss of G␣ s expression in PVN neurons (23,24), or selective deletion of the cAMP-response-element binding protein in PVN neurons (49), does not change feeding behavior indicating that G␣ s signaling downstream of MC4R is not required for the satiety response. Furthermore, mutations in MC4R that are associated with changes in adipocity do not always result in consistent changes to G␣ s signaling, suggesting that other pathways may also contribute to MC4R action (41,50). Dual coupling has been demonstrated previously for other G-protein coupled receptors, such as the metabotropic glutamate receptor and the thyroid stimulating hormone receptor (39, 51).  . MSH-induced Ca 2؉ signals. A, the fura-2 340/380 fluorescent excitation ratios are shown for individual neurons, each plotted with a uniquely colored trace, from primary hypothalamic cultures from either wild-type (WT) or Nckx4 knock-out (KO) mice. Cells were continually perfused with Hepesbuffered saline, containing either 1 M MSH or 75 mM KCl where indicated. B, the number of cells in each microscopic field that responded with a change in fura-2 ratio, as shown in panel A, following MSH treatment is plotted. Averages Ϯ S.E. for 18 -20 microscopic fields from six independent cultures are shown. ***, p Ͻ 0.001. C, GT1-7 cells were infected with lentivirus expressing mouse NCKX4, or a deleted control construct, and subsequently assayed for Ca 2ϩ response 2 days later. The fura-2 340/380 excitation ratios for individual responding cells are shown, each plotted with a uniquely colored trace. D, the average number of responding cells per field is summarized. n ϭ 9; **, p Ͻ 0.01. E, GT1-7 cells in Opti-MEM were treated with vehicle only (control) or with 10 M SKF96365 or 10 M U73122 at 37°C for 15 min with the indicated concentration of MSH, followed by extraction and determination of cAMP levels (n ϭ 5-8).
Although there have been previous reports showing Ca 2ϩ signals in response to MSH treatment of cultured cells (52,53), to our knowledge no previous report has demonstrated an essential role for MC4R-mediated Ca 2ϩ signals in suppression of feeding. Our hypothesis for the mechanism linking MC4R FIGURE 10. MSH treatment induces Ca 2؉ signals in hypothalamic GT1-7 cells. Confluent GT1-7 cells were analyzed individually for Ca 2ϩ responses as described in the legend to Fig. 9. The average fura-2 ratio Ϯ S.E. is plotted for the responding cells. A, the normal response of the cells to 300 s perfusion with 1 M MSH in Hepes-buffered saline perfusate containing 2 mM CaCl 2 (n ϭ 17). B, cells were pretreated with 1 M SHU9119 for 15 min prior to application of MSH (n ϭ 14). C, following activation by MSH treatment in normal perfusate (ϩCa), 200 s later the perfusate was switched one without CaCl 2 containing 0.1 mM EGTA (Ca-free) (n ϭ 15). D, cells were switched from normal perfusate (ϩCa) to one without CaCl 2 containing 0.1 mM EGTA for 500 s (Ca-free) during which MSH was applied for 300 s (n ϭ 9). E, cells were pretreated with 10 M SKF96365 for 15 min prior to application of MSH (n ϭ 11). F, cells were pretreated with 10 M U73122 for 15 min prior to application of MSH (n ϭ 11). G, cells were pretreated with 50 M nifedipine for 15 min prior to application of MSH (n ϭ 9). H, cells were pretreated with 100 M R p -cAMP for 15 min prior to application of MSH (n ϭ 17). These data are representative of at least six different coverslips from three different experiments for each condition. Anywhere from 7 to 25 cells per microscopic field, with an average of 15 cells, responded to MSH stimulation in each experiment. The number of responding cells was not different for cells pretreated with nifedipine or R p -cAMP. The oxytocin-positive cells were scored as being of high or low staining intensity, and the numbers pooled across all sections from each of five animals per treatment group. The total number of cells scored varied from 522 to 1288 per animal, but the averaged values were not different between groups (800 cells). The averaged ratios of high/low stained cells Ϯ S.E. are shown in the panel at lower right. Statistical significance is indicated where present: compared with wild-type untreated: **, p Ͻ 0.01; ***, p Ͻ 0.001. Scale bar, 100 m. stimulation with neuronal activation that leads to satiety signaling, and the role for NCKX4 activity in this process, is illustrated in the schematic of Fig. 12. The central findings from this study that support our model are as follows. First, we demonstrate that MSH treatment induces prominent oscillatory and long-lived Ca 2ϩ signals that are sensitive to the MC4R antagonist, SHU9119 (see Fig. 10). These Ca 2ϩ signals are blocked by the PLC inhibitor, U73122, or by the TRPC-selective Ca 2ϩ entry inhibitor, SKF96365, but not the voltage-dependent Ca 2ϩ channel inhibitor, nifedipine, suggesting that MC4R can couple through G␣ q to the activation of PLC, the production of diacylglycerol, and the activation of surface Ca 2ϩ entry channels, such as TRPC3 or TRPC6 (40). Second, MSH also induces an increase in cAMP, a response that is not influenced by treatment with either U73122 or SKF96365 (Fig. 9E), suggesting that MC4R can signal independently via both G␣ s and G␣ q , as also demonstrated for other G-protein coupled receptors (39,51). Third, the MSH induced production of OXT in PVN can be prevented by U73122 or SKF96365 (Fig. 11), thus implicating Ca 2ϩ , but not cAMP, signals in the normal suppression of feeding during satiety. In support of this essential role for Ca 2ϩ signaling, the Nckx4 knock-out mice have a constitutive activation of OXT expression, which can also be blocked by U73122 or SKF96365 (Fig. 11).
Our data clearly demonstrate that loss of the Ca 2ϩ transporter, NCKX4, in the Nckx4 Ϫ/Ϫ mice leads to a specific change in the melanocortin signaling axis that results in the constitutive activation of PVN neurons and the induction of anorexia. In our studies this is evidenced by an elevation of c-Fos expression and OXT expression in PVN neurons, and by a change in the number of neurons that respond with Ca 2ϩ signals to MSH stimulation. Because K ϩ -dependent Na ϩ /Ca 2ϩ -exchangers like NCKX4 are thermodynamically poised to catalyze Ca 2ϩ extrusion, loss of this pathway in Nckx4 Ϫ/Ϫ animals may destabilize normal Ca 2ϩ homeostasis, so that normally subthreshold signals through the MC4R trigger a sustained response. This may be particularly important for MC4R function, because the receptor is known to possess constitutive activity even in the absence of ligand (54), and because once a Ca 2ϩ response has been initiated, it remains long lived (see Figs. 9 and 10). Previous studies have demonstrated that NCKX exchangers typically play a significant role in Ca 2ϩ homeostasis in environments where Ca 2ϩ flux is high and intracellular Ca 2ϩ levels rise substantially, in a temporally and/or spatially restricted manner (3,55). This suggests that NCKX4 may be localized in close proximity to the entry pathways that control Ca 2ϩ signaling downstream from the MC4R (see model in Fig. 12). Such a specialized location may also explain why NCKX4 has a non-compensated FIGURE 12. A model for MC4R-dependent Ca 2؉ signaling and the role of NCKX4. MSH binding to the MC4R is shown to active G␣ q , via the exchange of GTP for GDP (upper row). It is important to note that MC4R may constitutively activate G␣ q even in the absence of ligand (54). So activated, G␣ q goes on to bind and stimulate PLC-␤, which subsequently cleaves phosphatidylinositol 1,4-bisphosphate (PIP 2 ) to diacylglycerol (DAG) and inositol 1,4,5-trisphosphate (IP 3 ). The membrane-bound diacylglycerol then activates the Ca 2ϩ -selective entry channel, TRPC6, which results in an elevation of cytosolic [Ca 2ϩ ] and consequent neuronal activation, which may include the initiation of action potentials (15) and the release of oxytocin (16, 41) (lower row). Normally, NCKX4 acts locally to extrude the Ca 2ϩ that enters via TRPC6, thus moderating the signaling pathway so only strong signals drive neuronal activation. In the absence of this moderating influence, even weak, constitutive, signaling via MC4R can result in sufficient Ca 2ϩ entry to initiate neuronal activation. and selective role in Ca 2ϩ homeostasis associated with the hypothalamic melanocortin axis. Selective roles for NCKX4 in other cells and tissues may also depend upon the precise subcellular distribution of the NCKX4 protein in relationship to components of other signaling pathways. These concepts await further investigation.