Adropin Is a Brain Membrane-bound Protein Regulating Physical Activity via the NB-3/Notch Signaling Pathway in Mice*

Background: The mechanism of action of adropin on metabolism remains elusive. Results: Adropin is a plasma membrane protein that can bind to the brain-specific, non-canonical Notch1 ligand NB-3. Conclusion: Adropin regulates physical activity, motor coordination, and cerebellum development in mice via the NB-3/Notch1 signaling pathway. Significance: Adropin is a highly conserved polypeptide that might also be important for cerebellum development in mammals. Adropin is a highly conserved polypeptide that has been suggested to act as an endocrine factor that plays important roles in metabolic regulation, insulin sensitivity, and endothelial functions. However, in this study, we provide evidence demonstrating that adropin is a plasma membrane protein expressed abundantly in the brain. Using a yeast two-hybrid screening approach, we identified NB-3/Contactin 6, a brain-specific, non-canonical, membrane-tethered Notch1 ligand, as an interaction partner of adropin. Furthermore, this interaction promotes NB3-induced activation of Notch signaling and the expression of Notch target genes. We also generated and characterized adropin knockout mice to explore the role of adropin in vivo. Adropin knockout mice exhibited decreased locomotor activity and impaired motor coordination coupled with defective synapse formation, a phenotype similar to NB-3 knockout mice. Taken together, our data suggest that adropin is a membrane-bound protein that interacts with the brain-specific Notch1 ligand NB3. It regulates physical activity and motor coordination via the NB-3/Notch signaling pathway and plays an important role in cerebellum development in mice.

Adropin is a highly conserved polypeptide that has been suggested to act as an endocrine factor that plays important roles in metabolic regulation, insulin sensitivity, and endothelial functions. However, in this study, we provide evidence demonstrating that adropin is a plasma membrane protein expressed abundantly in the brain. Using a yeast two-hybrid screening approach, we identified NB-3/Contactin 6, a brain-specific, non-canonical, membrane-tethered Notch1 ligand, as an interaction partner of adropin. Furthermore, this interaction promotes NB3-induced activation of Notch signaling and the expression of Notch target genes. We also generated and characterized adropin knockout mice to explore the role of adropin in vivo. Adropin knockout mice exhibited decreased locomotor activity and impaired motor coordination coupled with defective synapse formation, a phenotype similar to NB-3 knockout mice. Taken together, our data suggest that adropin is a membrane-bound protein that interacts with the brain-specific Notch1 ligand NB3. It regulates physical activity and motor coordination via the NB-3/Notch signaling pathway and plays an important role in cerebellum development in mice.
Adropin, composed of 76 amino acids, is encoded by the energy homeostasis-associated (Enho) gene that is remarkably down-regulated in the livers of mice with high fat diet-induced obesity or genetic obesity because of melanocortin receptor deficiency (Mc3r Ϫ/Ϫ ) or leptin deficiency. It was initially identified on microarray screening for genes that were up-regulated or down-regulated in Mc3r Ϫ/Ϫ mice (1). The expression of hepatic Enho mRNA is tightly regulated by energy intake in mice (1). Transgenic overexpression or synthetic adropin treatment in diet-induced obesity mice attenuated hepatosteatosis and insulin resistance (1), whereas global adropin knockout (adrKO) 4 mice, generated by the same group of researchers, were associated with increased hepatosteatosis, adiposity, and insulin resistance (2). In addition, adropin has been found to have a protective effect on endothelial function in the mouse (3). However, because the beneficial effects of adropin against hepatosteatosis and hyperinsulinemia associated with obesity and endothelial dysfunction in mice were determined on the basis of a single publication, most of the findings remain to be validated.
Although previous studies suggest a role of adropin as a possible metabolic regulator, the underlying molecular mechanism for the actions of adropin remains elusive at this stage. Published works have mainly focused on studying the action of adropin on the liver, white adipose tissue, and endothelium (1)(2)(3). However, on the basis of the mRNA expression profiles of adropin, it is expressed most abundantly in the brain in both mice and humans (15), whereas the central action of adropin has not been investigated. In this study, we characterized the expression and biochemical properties of adropin and investigated the central action of adropin in the mouse.

Cells and Plasmid Constructs
HEK293 and HeLa cells were obtained from the ATCC. Cells were maintained in DMEM supplemented with 10% fetal bovine serum (Invitrogen) and 1% penicillin/streptomycin in a humidified atmosphere of 5% CO 2 at 37°C. The ORF for adropin used in this study was obtained from the C57BL/6 wild-type mouse liver cDNA. To isolate the cDNA fragment that contains the full-length adropin protein coding sequence, PCRs using primers (forward, 5Јgaa ttc gcc acc atg ggg gca gcc atc tcc 3Ј; reverse, 5Јgga tcc tta ctt atc gtc gtc atc ctt gta atc ggg ctg cag cag gta gct gcc 3Ј) were performed. After PCR, the DNA fragment was digested with restriction enzymes (EcoRI and BamHI) and inserted into the pcDNA3.1 vector (Invitrogen) for transient expression in mammalian cells. To clone the DNA fragment encoding adropin protein (C-terminal, amino acids 30 -76) for yeast two-hybrid screening, oligonucleotide primers (forward, 5Јgaa ttc tgc cat tct cga tct gct gac gtc 3Ј; reverse, 5Јgga tcc tta ggg ctg cag cag gta gct gcc 3Ј) were used to perform PCR. After PCR, the DNA fragment was digested with restriction enzymes (EcoRI and BamHI) and inserted into the pGBKT7 vector (Clontech) for transient expression in yeast. All expression plasmids were verified by sequencing. The jetPEI in vitro transfection reagent (Polyplus) was used to transfect DNA into HEK293 and HeLa cells following the instructions of the manufacturer.

Generation of Anti-adropin Antibody
A His-tagged cDNA fragment encoding mouse adropin (amino acid residues 30 -76) was subcloned into the pROEX-HTb vector, which was then used to transform into host Escherichia coli BL21 cells. Oligonucleotide primers (forward, 5Јgga tcc atg cat cat cac cat cac cac tgc cat tct cga tct gct gac gtc3Ј; reverse, 5Јgaa ttc tta ggg ctg cag cag gta gct gcc 3Ј) were used to perform PCR. The expression was induced by addition of 1 mM isopropyl 1-thio-␤-D-galactopyranoside. His-tagged adropin fragment was purified from the bacterial lysates using a nickelnitrilotriacetic acid-agarose column as described previously (16). The purity of the protein was confirmed by SDS-PAGE and HPLC. The polyclonal antibody against the recombinant mouse adropin was raised in female New Zealand White rabbits as described previously (16). The specificity of the antibodies was verified by Western blotting.

Generation of Mice with Targeted Disruption of Adropin
The adrKO targeting vector was designed to disrupt the expression of adropin by inserting a neomycin resistance gene into the 5Ј region of the adropin open reading frame in the adropin gene. The targeting vector was linearized and electroporated into 129Sv/J-derived CJ7 ES cells. The neomycin-resistant clones were identified by PCR, and correct ES cell clones were microinjected into the blastocyst stage of C57BL/6 embryos to generate chimeric mice that were subsequently backcrossed with C57BL/6 mice to generate F1 animals heterozygous for the mutated adropin allele. The offspring was routinely genotyped by PCR using 35 cycles of 95°C for 30 s, 55°C for 30 s, and 72°C for 45 s with the following primers: 5Ј ATG GTT GGC CAC CCC AGA 3Ј (forward) and 5Ј ACT AGT GAG ACG TGC TAC TTC 3Ј (reverse). AdrKO mice were generated by the Shanghai Research Center for Model Organisms.

Animal Maintenance
All animal experimental procedures were approved by the Committee on the Use of Live Animals for Teaching and Research of the University of Hong Kong and were carried out in accordance with the Guide for the Care and Use of Laboratory Animals. AdrKO mice were backcrossed onto a C57BL/6 genetic background for at least six generations before investigation. AdrKO mice and WT littermates were housed in a 12-h light/dark cycle (07:00 -19:00) room under controlled temperatures (23 Ϯ 1°C) with free access to water and standard chow (20% kcal protein, 10% kcal fat, and 70% kcal carbohydrates) or high-fat diet (HFD; Research Diet; 20% kcal protein, 45% kcal fat, and 35% kcal carbohydrates), respectively. Mice were sacrificed by cardiac puncture at the indicated time points. Serum and tissues were collected for further analysis.

Indirect Calorimetry and Body Composition
Energy expenditure was measured using the comprehensive laboratory animal monitoring system (CLAMS, Columbus Instruments, Columbus, OH) as described previously (17,18). Briefly, mice were housed singly in CLAMS cages and acclimated for 48 h, and then data on oxygen consumption (VO 2 ), carbon dioxide consumption (VCO 2 ), respiratory exchange ratio (VCO 2 /VO 2 ), food intake, and locomotor activity were recorded simultaneously for a further 72 h. Body composition (fat mass and lean mass) was measured using 1 H magnetic resonance spectroscopy (Bruker BioSpin, Billerica, MA) as described previously (19).

Glucose Homeostasis
Glucose homeostasis was assessed in standard chow-or HFD-fed AdrKO mice and their WT littermates. Intraperitoneal glucose tolerance tests were performed at 10:00 a.m. in mice fasted for 16 h using a blood glucose meter (Accu-CHEK Advantage, Roche) to detect glucose levels in the blood sampled from the tail vein after intraperitoneal injection of D-glucose (Sigma-Aldrich, St. Louis, MO) (1 g/kg of body weight). 20 l of blood samples were collected at 0, 30, 60, and 90 min for insulin ELISA (Antibody and Immunoassay Services, The University of Hong Kong). For insulin tolerance tests, a similar procedure was followed, except that mice were fasted for 6 h, and insulin (Eli Lilly and Co., Indianapolis, LN) (1 unit/kg of body weight) was injected intraperitoneally.

Behavior Analysis
Rotarod Test-Mice were placed on a horizontally oriented, rotating rod (20 rpm/min). The length of time (latency) that a given mouse stayed on the rotating rod was recorded, and the maximum time was set to 180 s.
Wheel Running Test-Mice were housed singly in standard cages equipped with running wheels (Lafayette Instruments) with food and water ad libitum and with a 12-h light/12-h dark cycle for 10 days. Wheel revolution counts were recorded automatically by computer.
Hanging Wire Test-Four paws were placed onto a metal wire by handling the tail. When the mouse had grasped the wire, it was gently turned upside-down to face upward, and its tail was released. The time until the mouse fell to the floor was recorded, and the maximum time was set to 120 s.

RNA Extraction and Real-time PCR
Total RNA was extracted from various mouse tissues with TRIzol reagent (Invitrogen), and treated with RNase-free DNase (Promega, Madison, WI) at 37°C for 30 min to remove genomic DNA. For reverse transcription, 1 g of the total RNA was converted to first-strand complementary DNA in 20 l reactions using a cDNA synthesis kit (Promega). Quantitative real-time PCR was performed in duplicate in a total reaction volume of 20 l with SYBR Green PCR Master Mix (Qiagen, Hilden, Germany) in an ABI Prism 7000 instrument (Applied Biosystems) with the following parameters: 95°C for 5 min, followed by 40 two-step cycles at 95°C for 10 s and 60°C for 30 s. The forward and reverse primers used for PCR amplification of mouse HES1, HES5, HEY1, Cdkn1a, ␤-actin, and GAPDH were as follows: HES1, 5Ј-CAG CCA GTG TCA ACA CGA CAC-3Ј (forward) and 5Ј-TCG TTC ATG CAC TCG CTG AG-3Ј (reverse); HES5, 5Ј-CGC ATC AAC AGC AGC ATA GAG-3Ј (forward) and 5Ј-TGG AAG TGG TAA AGC AGC TTC-3Ј (reverse); HEY1, 5Ј-CAC GCC ACT ATG CTC AAT GT-3Ј (forward) and 5Ј-TCT CCC TTC ACC TCA CTG CT-3Ј (reverse); Cdkn1a, 5Ј-CTT GCA CTC TGG TGT CTG AG-3Ј (forward) and 5Ј-GCA CTT CAG GGT TTT CTC TTG-3Ј (reverse); GAPDH, 5Ј-CTT TGT CAA GCT CAT TTC CTG G-3Ј (forward) and 5Ј-TCT TGC TCA GTG TCC TTG C-3Ј (reverse); and ␤-actin 5Ј-CTA AGG CCA ACC GTG AAA G-3Ј (forward) and 5Ј-ACC AGA GGC ATA CAG GGA CA-3Ј (reverse). Analysis was performed with ABI Prism 7000 SDS software, and the mRNA expression of each gene was calculated after being normalized with ␤-actin expression using the comparative Ct method.

Yeast Two-hybrid Assay
The C-terminus of mouse adropin (amino acids 30 -76) was used for yeast two-hybrid screening. The screening procedures were conducted following the instructions of the manufacturer. The cDNA library used for this screening was a universal mouse (normalized) library (catalog no. 630482, Clontech). About 1 million cotransformants were screened for the four reporter genes HIS, lacZ, ADE, and AUR1-C on plates containing the respective selective medium.

Western Blotting and Cellular Fractionation
Cells were lysed in radioimmune precipitation assay buffer (50 mM Tris-Cl, 150 mM NaCl, 1% Nonidet P-40, 0.5% sodium deoxycholate, and 0.1% SDS), and total protein concentration was quantified by Bradford assay. Equal amounts of cellular protein were separated by SDS-PAGE, and immunoblotting was carried out as described previously (17,19). For subcellular fractionation, the Qproteome plasma membrane kit (Qiagen) was used according to the protocol of the manufacturer.

Immunofluorescence
HeLa cells grown on coverslips were transfected with pcDNA3.1-adropin. After 72 h, cells were fixed with 4% paraformaldehyde in PBS and permeabilized with 0.2% Triton X-100. The coverslips were blocked with 5% BSA in PBS for 1 h at 37°C and incubated with anti-adropin in a 1:50 dilution. Mouse monoclonal anti-cadherin antibody (Affinity Bioreagents) and rabbit anti-FLAG (Sigma-Aldrich) were used for immunostaining. TRITC (red)-conjugated goat anti-mouse and FITC (green)-conjugated goat anti-rabbit antibodies were used for visualization with a confocal microscope.

Luciferase Reporter Assay
Transfections were done in 6-cm plates (5 ϫ 10 5 cells/plate), and the cells were harvested after 48 h and lysed in a cell lysis buffer (Promega). Luciferase activity was measured using a luciferase reporter assay system (Promega), and readings were taken on a luminometer.

Statistical Analysis
All experiments were performed using male mice in cohorts consisting of 6 -10 mice with data presented as mean Ϯ S.E. Statistical significance was determined by one-or two-way analysis of variance or Student's t test, where appropriate, using the Statistical Package for Social Sciences version 16.0 (SPSS, Chicago, IL). In all statistical comparisons, p Ͻ 0.05 was used to indicate a significant difference.

RESULTS
Adropin Is a Membrane-bound Protein Highly Expressed in the Brain-Quantitative PCR (qPCR) analysis showed that the expression of adropin mRNA is highest in the brain, followed by the liver (Fig. 1A; 6-fold higher in the brain than in the liver), which is in agreement with a previous study (1) and the gene expression database Expression Atlas (15). To further explore the tissue distribution of the adropin protein, we generated a specific antibody against adropin for Western blot analysis when the same amount of protein was loaded (Fig. 1B). Because adropin had been proposed as a hepatocyte-secreted protein (1), it was expected that adropin could be found in the liver and blood samples. The controversial findings regarding circulating adropin (6,8,9,13) led us to reinvestigate the biochemical properties of adropin.
First, prediction of transmembrane topology and the signal peptide of adropin were performed by bioinformatics analysis, including Phobius and Signal IP 4.1 (20,21). From amino acids 9 -30, the score for the transmembrane was highest (Ͼ0.9). Among all four possibilities, Phobius predicted this region to be a transmembrane domain (Fig. 1C). This may be due to the fact that this region is enriched in hydrophobic residues (including seven leucine, three isoleucine, and three valine residues of 21 amino acids) with a very poorly defined protease cleavage site (20 -23).
To verify the bioinformatics-based prediction, in vitro experiments were performed to examine the presence and location of adropin in HEK293 cells harboring the plasmid overexpressing adropin. To our surprise, no adropin was detected in the conditioned medium (Fig. 1D, lane 2), even in the sample concentrated 4-fold by TCA precipitation (Fig. 1D, lane 3). The adropin protein was only detected in total cell lysates (Fig. 1D, lane 1). As a comparison, adiponectin, a well known antidiabetic hormone secreted from fat cells (24), was detected in both the cell lysates and the conditioned culture medium (Fig. 1D, lanes 4 and 5) and further enriched by TCA precipitation (Fig. 1D, lane 6).
Because adropin was detected in cell lysates but not in the conditioned medium, we proceeded to investigate the subcellular localization of adropin. Subcellular protein fractionation analysis with transfected HEK293 cells overexpressing adropin was performed, and the selected fractions were examined by Western blotting (Fig. 1E, left panel). In addition, to prevent the expression level from affecting its subcellular localization, mouse brain lysate for endogenous adropin was used (Fig. 1E, right panel). Both overexpressed and endogenous adropin were found to cofractionate predominately with the plasma membrane-bound protein pan-cadherin (Fig. 1E,  lanes 2 and 5).
In addition, the membrane marker pan-cadherin was used to examine whether adropin localizes to the plasma membrane by confocal microscopy analysis (Fig. 1, F-I). Colocalization of pan-cadherin (stained red in Fig. 1F) and adropin (stained green in Fig. 1G) was detected as yellow fluorescence in Fig. 1I. Our data showed that adropin was mainly localized to the plasma membrane (Fig. 1, F versus G). The evidence from subcellular protein fractionation and confocal microscopy analysis clearly indicates that adropin is indeed a plasma membrane-bound protein, as predicted by bioinformatics. In B, the arrangement of the samples is the same as in A. sWAT, subcutaneous white adipose tissue; eWAT, epididymal white adipose tissue; BAT, brown adipose tissue. C, the subcellular localization of adropin was predicted by the Phobius program. On the basis of the analysis, for the N-terminal from amino acids 1-9, because the probability of cytoplasmic is higher than the non-cytoplasmic, this region should be localized in the cytoplasm. From amino acids 9 -30, the score for the transmembrane is the highest (Ͼ0.9) among all four possibilities, so this region was suggested to be the transmembrane domain. For the C-terminal from amino acids 30 -76, because the probability of "non-cytoplasm" is highest, it should be localized outside of the surface of the plasma membrane. The score for the probability of the N-terminal amino acids from 1-33 as a signal peptide is lower than 0.1. D, Western blotting of adropin and adiponectin (a secretory protein used as a control) in HEK293 cells. C, cell lysate; M, culture medium; T, concentrated protein by TCA precipitation (4ϫ) of the culture medium. E, subcellular fractionation of overexpressed adropin in HEK293 cells and endogenous adropin in mouse brain tissue. C, cytosol; Me, membrane; CM, culture medium. F-I, adropin is localized mainly on the plasma membrane. HeLa cells were grown on cover slips and transfected with a plasmid expressing adropin. After 24-h transfection, cells were fixed with 4% paraformaldehyde and stained with anti-pan-cadherin and anti-adropin antibodies. The immunofluorescence analysis was performed by confocal microscopy. Endogenous pan-cadherin (membrane marker) was visualized in red (F), adropin in green (G), and DNA in blue (H). I, colocalization of adropin and pan-cadherin was detected as yellow fluorescence.

Adropin Binds to NB-3 and Activates Notch Signaling in
Vitro-Because information about the molecular functions of adropin was scarce, we performed yeast two-hybrid screening to explore the function of adropin via its binding proteins. Because the N-terminal transmembrane domain (amino acids 1-29) of adropin is too hydrophobic, the C terminus of adropin (amino acids 30 -76) was used as bait for yeast two-hybrid screening. In line with our finding of adropin as a membranebound protein, most of the adropin binding partners identified by this approach were also membrane proteins. Among them, we focused on the prey, neural recognition molecule 3 (NB-3, also known as Cntn6) in this study because NB-3 is a brainspecific glycosylphosphatidylinositol anchor protein (25,26). To verify the interaction, the plasmid GAL4 DNA-binding domain (GAL4-BD)-fused adropin and/or GAL4 activation domain (GAL4-AD)-fused NB-3 were transformed into the yeast strain Y2HGold (auxotrophic for histidine because the endogenous HIS3 gene was deleted). Only the yeast cells transformed with both the GAL4-BD-adropin and GAL4-AD-NB-3 plasmids could grow under selective conditions lacking histidine ( Fig. 2A, right panel, lane 3 versus lanes 1 and 2) because the interaction of GAL4-BD-adropin and GAL4-AD-NB-3 drove the transcription of the exogenously introduced Gal4responsive HIS3 gene.
The physical interaction of NB-3 and adropin was also validated by coimmunoprecipitation analysis in a mammalian cell culture system. The cell lysates of HEK293 cells overexpressing FLAG-tagged adropin and/or HA-tagged NB-3 were used (Fig.  2B). Without FLAG-tagged adropin, the HA-tagged NB-3 could not be pulled down by anti-FLAG beads (Fig. 2B, lane 8), and all unbounded NB-3 remained in the supernatant (Fig. 2B, lane 9). However, with FLAG-tagged adropin, NB-3 was found mainly in the pellet fraction (Fig. 2B, lane 5 versus lane 8) instead of in the supernatant (Fig. 2B, lane 6 versus lane 9). Our results clearly demonstrate that adropin physically binds to NB-3.
In addition, the colocalization of NB-3 and adropin was confirmed further by confocal microscopy. As shown in Fig. 2C, both adropin and NB-3 are colocalized to the plasma membrane. In addition, we also performed coimmunoprecipitation to confirm that endogenous adropin binds to NB-3 in mouse brain samples. As shown in Fig. 2D, in brain tissue lysate from WT mice, both the adropin and NB-3 proteins were enriched in the pulldown samples (Fig. 2D, lane 3). However, in the lysate from adrKO mice, because there is no adropin protein in the lysate (Fig. 2D, lane 2), no NB-3 was pulled down (Fig. 2D, lane 4). These data suggest that endogenous NB-3 and adropin interact with each other.
NB-3 functions as a membrane-tethered Notch1 ligand and belongs to the contactin family that mediates cell surface interaction during nervous system development (27). NB-3 promotes Notch1 activation and, subsequently, drives the expression of Notch1 target genes such as HES1 and HEY1 (28,29). To decipher the effects of adropin on NB-3-mediated Notch1 signaling, adropin and/or NB-3 were overexpressed in HEK293 cells harboring both a firefly luciferase (F-Luc) reporter driven by the HES1 promoter and a Renilla luciferase (R-Luc) reporter with the CMV promoter as an internal control. As shown in Fig.  2E, overexpressing the Notch1 ligand NB-3 could induce the expression of HES1-F-Luc (28). Furthermore, NB-3 and adro-pin apparently synergized with each other to further increase the activity of the HES1 promoter.
The induction of expression of the endogenous HES1 and HEY1 mRNA levels was also examined by qPCR. Overexpressing both NB-3 and adropin in 3T3L1 cells could induce the expression of HES1 and HEY1 mRNA (Fig. 2F). To further verify that the activation of the transcription of HES1 and HEY1 is through the Notch signaling pathway, cells transfected with adropin-and NB-3-overexpressing plasmids were treated with the ␥-secretase inhibitor N-[N- (3,5-difluorophenacetyl)-L-alanyl]-S-phenylglycine t-butyl ester (DAPT), where ␥-secretase is an enzyme required for Notch activation. As shown in Fig. 2F, the induction of both HES1 and HEY1 mRNA expression by adropin and NB-3 could be repressed by treatment with the DAPT. Interestingly, it is not necessary to express both adropin and NB-3 in the same cell to activate the Notch signaling pathway. Cocultured cells expressing either adropin or NB-3 could also induce the expression of HES1 and HEY1, and their expression levels were similar to cells transfected with both adropinand NB-3-overexpressing plasmids. This may be due to the fact that NB-3 is a lipid-anchored protein covalently attached to a fatty acid that serves to anchor NB-3 to the cell membrane face, and it has been reported that the glycophosphatidylinositol anchor NB-3 protein could be transferred intercellularly from one cell to another by a process termed "cell painting" (30).
Generation of Adropin Knockout Mice-To further explore the role of adropin in Notch/NB-3 signaling in vivo, we generated adrKO mice to determine its physiological role. As mentioned in a previous study (2), the adropin gene comprises two exons, and its ORF is located in exon 2, which overlaps with exon 20 of the adjacent gene dynein, axonemal, intermediate chain 1 (Dnaic1) (Fig. 3A). To minimize the adverse effect on the expression of Dnaic1 by disrupting transcription termination sites and/or 3ЈUTR of Dnaic1 located on the exon 2 of adropin gene, DNA sequence of Neomycin resistance gene (NEO) and 126 bp (including the loxP recombination site) were inserted into the 5Ј and the 3Ј region of the adropin ORF, respectively (Fig. 3A, lower  panel), to disrupt the translation of adropin protein. No DNA sequence in the exon 2 of adropin gene was deleted.
The relative position of the primers for the genotyping is shown in Fig. 3A, and a typical genotyping result is shown in Fig.  3B. RT-PCR was performed to ensure that there were no changes in the expression of Dnaic1 and adropin mRNAs in both WT and adrKO mice (Fig. 3C). Because of the insertion of the NEO gene into the 5Ј region of the adropin ORF, no adropin protein was translated from this bicistronic mRNA. As expected, in our Western blot analysis (Fig. 3D), no adropin protein was detected in the brain tissues of adrKO mice.
AdrKO Mice Have Defects in Physical Activity-After successfully obtaining the adrKO mouse line, we further examined its phenotype. The body weight and adiposity of the adrKO mice were similar to WT littermates under both standard chow conditions and after chronic HFD treatment (Fig. 3, E and F). No significant changes in glucose tolerance and insulin sensitivity were observed by glucose tolerance test and insulin tolerance test, respectively (Fig. 3, G-I). Because adropin is highly expressed in the brain (Fig. 1) and binds to NB-3 (Fig. 2), which is important for brain development and motor coordination (25), we next investigated whether adropin regulates physical activity via the central nervous system. CLAMS was used to monitor the motor performance of adrKO mice. A significant reduction in locomotor activity (Fig. 4A) and a slight decrease in food intake (Fig. 4B) were observed in adrKO mice compared with their WT littermates, especially during the dark period. The phenotype of reduced physical activity of our adrKO mice was consistent with the findings of a previous study (2).
To further confirm the physically inactive phenotype of adrKO mice, voluntary wheel running experiments were performed to monitor the movement of the mice for 7 days. The results showed that the running distance of the adrKO mice was significantly less than that of their WT littermates from days 4 -7 (Fig. 4C). In addition, the locomotor activity of adrKO mice was significantly less than that of their WT littermates in the dark phase (Fig. 4D). This may be due to the fact that mice are usually more active in feeding FIGURE 2. Adropin can bind to NB-3 and modulate the Notch1 signaling pathway. A, validation of the interaction between NB-3 and adropin by yeast two-hybrid assay. Yeast (Y2HGold, Clontech) cells expressing both GAL-BD-adropin and GAL-AD-NB-3 could grow on the selective medium without leucine, tryptophan, and histidine (LWH) but not cells expressing either GAL-BD-adropin or GAL-AD-NB-3. SC, synthetic complete. B, validation of the interaction of NB-3 and adropin by coimmunoprecipitation. FLAG-tagged adropin and/or HA-tagged NB-3 were expressed in HEK293 cells. Immunoprecipitation was performed with anti-FLAG beads, and precipitates were immunoblotted with anti-NB-3 or anti-adropin antibody. I, input; P, pellet; S, supernatant. C, confirmation of plasma membrane colocalization of both NB-3 and adropin. HeLa cells were grown on coverslips and transfected with adropin-and NB-3-overexpressing plasmids. After 24-h transfection, cells were fixed with 4% paraformaldehyde and stained with anti-NB-3 and anti-adropin antibodies. Immunofluorescence analysis was performed by confocal microscopy. Adropin is shown in green (first panel), and NB-3was visualized in red (second panel). Colocalization of adropin and NB-3 was detected as yellow fluorescence (third panel) with bright field images to show where the cells are localized. D, validation of the interaction of endogenous adropin and NB-3 by mouse brain tissue samples. Brains from 8-week-old WT and adrKO mice were used. Coimmunoprecipitation was performed with anti-adropin antibody, and precipitates were immunoblotted with either anti-NB-3 or anti-adropin antibodies. E and F, adropin can induce the expression of Notch-regulating genes. E, HEK239 cells were transfected with the HES1 F-Luc reporter plasmid with NB-3 and/or adropin. The readings were normalized with CMV-R-Luc. The empty vector pcDNA3.1 was used, and corrected luciferase values were arbitrarily set to a value of 1 and served as a reference for comparison of fold differences. Error bars indicate mean Ϯ S.D. of triplicate experiments. *, p Ͻ 0.05. F, qPCR was performed to check the mRNA expression levels of HES1 and HEY1 mRNA. 3T3L1 cells were transfected with NB-3-and/or adropin-overexpressing plasmid. The empty vector pcDNA3.1 was used as control and served as a reference (set to a value of 1) for comparison of fold differences. We also checked whether cocultured 3T3L1 cells transfected with NB-3-or adropin-overexpressing plasmid could induce the expression of HES1 and HEY mRNA. The ␥-secretase inhibitor DAPT was used to inhibit the Notch signaling pathway. Cells were treated with 1 M DAPT for 24 h before being harvested. Error bars indicate mean Ϯ S.D. of triplicate experiments. *, p Ͻ 0.05 compared with control. and movement during the dark phase, so that the changes became more obvious. In summary, adrKO mice were less active than their WT littermates in the dark phase.
To explore whether the reduction of physical activity in adrKO mice is due to a decrease in muscle strength and/or endurance capacity, forced treadmill running tests were performed (31). There was a significant increase in time (Fig. 4E) and number (Fig. 4F) of shocks for adrKO mice compared with their WT littermates. Furthermore, the total run distance in adrKO mice was reduced significantly compared with wild-type littermates (Fig. 4G). To further examine the possible defects in muscle strength, hanging wire tests were performed. No significant change in the latency of fall between WT and adrKO mice was observed (Fig. 4H).
Deletion of Adropin Impairs Motor Coordination and Synapse Formation in the Cerebellum-Because a previous study demonstrated that NB-3 knockout (nb-3KO) mice exhibited impaired motor coordination (32), rotarod tests were performed to evaluate the motor coordination of our adrKO mice. WT mice had a significant improvement in the latency to fall after 13 trials (Fig. 5A, black columns), but adrKO mice were inferior to WT mice in their ability to walk on the rotating rod (Fig. 5A, white versus black columns), indicating that adrKO mice were impaired in balance control. The poor motor coordination phenotype of adrKO mice matched the respective phenotype in nb-3KO mice reported previously (25). The coordination defect of adrKO mice could also explain their poor performance in the forced treadmill running tests (Fig. 4, E and  F).
Next, we explored whether the deletion of adropin would affect cerebellum development, as observed in nb3-KO mice (27,33). No gross abnormality in the brain architecture of adrKO mice was observed (data not shown). Then we examined the expression level of NB-3 in the cerebellum of adrKO mice. There was a modest but significant decrease (ϳ15%) in the expression level of NB-3 in the cerebellum of adrKO mice compared with the WT controls (Fig. 5B, left panel, top, lanes 3 and  4 versus lanes 1 and 2, p Ͻ 0.05). Previous studies suggested that NB-3 plays a role in synaptic formation in the developing cerebellum (33), possibly through the Notch1 signaling pathway (28,29,34,35). Therefore, we also measured the expression level of Notch1 in the cerebellum of adrKO mice by Western blot analysis. A modest increase (ϳ20%) in the level of Notch1 in the cerebellum of adrKO mice was observed compared with WT littermates (Fig. 5B, left panel). In line with the findings, the basal expression levels of the Notch1-regulated genes HES1, HES5, and Cdkn1a were down-regulated in the brain tissue samples of adrKO mice as determined by qPCR (Fig. 5C, p Ͻ 0.05).
A previous study suggested that the impairment of motor coordination in nb-3KO mouse was due to a reduction in synaptic density between Purkinje cells and granule cells located along parallel fibers in the cerebellum of nb-3KO mice (27) because Purkinje cell-granule cell parallel fiber synaptic transmission mediated by the neurotransmitter glutamate is critical for normal cerebellar function, including motor control (36). Therefore, we also measured the synaptic density via the vesicular glutamate transporter (VGlut1)-positive area in the cerebellum of adrKO and WT mice by immunohistochemistry (Fig.  5D). Calbindin was used as a marker of Purkinje cells. The den-sity and number of Purkinje cells in adrKO mice were similar to their WT littermates (Fig. 5D, left panel, cells stained green). However, the density of VGlut1-positive puncta (Fig. 5D, left panel, cells stained red) was decreased slightly but significantly in the cerebellum of adrKO mice (p Ͻ 0.05) compared with the respective regions of their WT littermates. The observation of the defect in the Purkinje cell synapse is similar to the reported phenotype of nb-3KO mice (27) and could at least partially explain the defects of locomotor activity and coordination in adrKO mice.

DISCUSSION
In this study, we report that adropin is a membrane-anchored protein (Fig. 1). Although several papers published recently have reported that adropin from various mammals could be detected in serum and that its circulating level was altered under different pathophysiological conditions (1)(2)(3)(4)(5)(6)(7)(8)(9)(10)(11)(12)(13)(14), we cannot exclude the possibility that a tiny amount of adropin presents in the circulation under certain pathological conditions. Nevertheless, our findings strongly suggest that the majority of adropin protein localizes in the brains of mice and that adropin is in a membrane-bound state under physiological conditions.
To explore the function of adropin in vivo, we generated adrKO mice. In agreement with a previous report (2), adrKO mice display reduced physical activity. Our adrKO mice displayed defects in both locomotor activity and coordination. The movement disorders may be due to the decrease in the reduction of NB-3 in the cerebellum and, hence, cerebellar synapse formation in adrKO mice. We hypothesize that adropin binds to a brain-specific membrane-bound protein, NB-3 (Fig.  2, A-D), and activates the Notch1 signaling pathway in vitro (Fig. 2, E and F). Because NB-3 is required for normal cerebellum development in mice, we predict that the phenotype of impaired motor activity (Fig. 4, A, C, and G) and coordination (Fig. 5A) in our adrKO mice might be due to defects in the NB-3-mediated Notch1 signaling pathway (Fig. 5C) and, hence, granule cell-Purkinje cell synapse formation during cerebellar development (Fig. 5D). A previous study has demonstrated that adropin is detected at granular and Purkinje cells in the cerebellum (37). Unfortunately, we could not find a suitable antiadropin antibody to perform a similar immunohistochemical FIGURE 5. Impaired motor coordination and cerebellar synapse formation in adrKO mice. A, the rotarod test was used to assess motor coordination and balance. The latency time indicates how long mice were able to walk on an accelerating rotarod at speeds from 20 rpm. The adrKO mice had a motor coordination defect. The difference was more significant after the mice were given training with more than 13 trials. Male mice at 8 weeks of age were used (n ϭ 14 per group). B, the levels of the endogenous NB-3 and Notch1 in the cerebellum of WT and adrKO mice were examined by Western blotting (left panel). Cerebellum were isolated from 8-week-old male WT or adrKO mice, followed by Western blot analysis to detect the expression levels of NB-3 and Notch1. Right panel, densitometric analysis for the relative amount of NB-3 (n ϭ 6). *, p Ͻ 0.05. C, qPCR to determine the expression levels of HES1, HES5, Cdkn1a, and GAPDH mRNA in cerebellum of 8-week-old male WT and adrKO mice. The expression level of ␤-actin mRNA was used for normalization of the target gene expression levels. The corrected values of WT mice were arbitrarily set to a value of 1 (n ϭ 5) *, p Ͻ 0.05. D, cerebellar synaptic defects in adrKO mice. Calbindin was used as a marker of Purkinje cells. VGlut1-positive puncta around Purkinje cells in adrKO and WT mice are shown. The density of the VGlut1 signal was quantified using the JACoP plugin of ImageJ (38). The corrected values of WT mice were arbitrarily set to a value of 1. There was a significant change in the density of VGlut1 puncta in adrKO mice (n ϭ 6). *, p Ͻ 0.05. FIGURE 6. Working model. A, in wild-type mice, adropin cooperates with NB-3 to regulate the Notch1 signaling pathway for normal cerebellar development. Adropin may be important for NB-3 recruitment, enrichment, and/or binding affinity to Notch1 receptor and then affects brain development, locomotor activity, and coordination. B, in adropin knockout mice, less NB-3 was recruited, and the Notch1 signaling pathway was repressed. This may affect brain development and, hence, locomotor activity and coordination. study to identify the cell types and precise region(s) in the brain where the adropin protein is expressed.
In conclusion, our data suggest that adropin cooperates with NB-3 to regulate Notch1 signaling in brain development and, hence, the locomotor activity and coordination of mice (Fig. 6). It would be of great interest to further experimentally delineate the exact mechanism underlying how adropin modulates the Notch1 signaling pathway and, hence, brain development via NB-3 in vivo. Because the amino acid sequences of human and mouse adropin are identical, the findings of the mouse adropin research will also apply to humans.