Materials.

NKA activator SSA412 is a polyclonal antibody generated against the extracellular ⁸⁹⁷DVEDSYGQQWTYEQR⁹¹¹ region of rat α-subunit of NKA and is further purified through an affinity column directed against the same synthetic peptide (42). SSA412 increases the ATP-hydrolyzing activity of NKA and consequently modulates intracellular Ca²⁺ transients (42). Western blots show that SSA412 recognizes the α-subunit of three isoforms of NKA. Ouabain was from Wako Pure Chemical Industries (Osaka, Japan). Ghrelin was from Peptide Institute (Osaka, Japan). All other reagents were purchased from Sigma-Aldrich (St. Louis, MO).

Animals.

Adult male Sprague-Dawley (SD) rats were purchased from Japan SLC (Hamamatsu, Japan) and were maintained on a 12:12-h light-dark cycle and given conventional food (CE-2; CLEA Japan, Tokyo, Japan) and water ad libitum. The animal protocols for this study were approved by the Animal Care and Use Committee of Jichi Medical University.

Measurements of NKA's ATP-hydrolyzing activity and ATP content in the hypothalamus and in vitro slices.

For in vivo assays, bilateral portions of ARC, LHA, and VMH were dissected from rats aged 8 wk under ad libitum-fed or 24-h-fasted conditions. For in vitro assays, coronal brain slices containing ARC were prepared from rats aged 5 wk by using a microslicer. Bilateral ARC portions were dissected from slices, incubated in HEPES-buffered Krebs-Ringer bicarbonate buffer (HKRB) solution containing 0.5, 2.8, or 8.3 mM glucose for 10 min at 37°C. The ATPase assay was performed by the method of Kyte (20), with modifications as previously reported (42). The ATP-hydrolyzing activity of NKA was defined as the ouabain-sensitive hydrolysis of Mg²⁺ATP (3 mM) in the presence of both Na⁺ (100 mM) and K⁺ (20 mM) with or without ouabain (2 mM). Briefly, the incubation mixture containing samples in a final volume of 0.2 ml was maintained at 37°C, and reaction was initiated by adding Mg²⁺ATP and was stopped after 30 min by adding 0.75 ml of quench solution and 0.02 ml of developer. The color was allowed to develop for 30 min at room temperature. The free phosphate was then determined by reaction with molybdate, giving a product measured by spectrophotometer at 700 nm. For ATP content assay, samples were homogenized in 10 mM HEPES buffer (pH 7.5) and incubated with 100 μl of lysis buffer (Toyo B-net, Tokyo, Japan). The luminescence of each sample was measured with Luminoimage Analyzer LAS-1000 (Fuji Film, Tokyo, Japan) and normalized to the DNA content extracted from each homogenized solution.

Western blotting.

Bilateral portions of ARC were dissected from ad libitum-fed or 24-h-fasted rats and lysed. The 5 μg of proteins were subjected to 10% SDS-PAGE and transferred to nitrocellulose filters. NKA-α1, -α2, and -α3 isoforms were detected with specific antibodies (Table 1). Immunoreactive proteins were detected with HRP-conjugated secondary antibody and the ECL system (Amersham, Arlington Heights, IL). The luminescence of each sample was calculated by Luminoimage Analyzer. Protein levels were normalized to those of β-actin from the same samples. β-Actin was detected with goat anti-β-actin polyclonal antibody (C-11) (sc-1615; Santa Cruz Biotechnology, Santa Cruz, CA).

Measurements of plasma ghrelin and blood glucose concentrations.

Blood was sampled from the tails of 10-wk-old rats under ad libitum-fed and 24-h-fasted conditions in the same animals. Blood glucose and plasma ghrelin (acylated ghrelin) concentrations were measured using a GlucoCard DIA meter (Arkray, Kyoto, Japan) and enzyme-linked immunosorbent assay (ELISA) kits (Mitsubishi Kagaku Iatron, Tokyo, Japan), respectively, as described previously (14).

Intra-ARC and third ventricle injections and measurements of food intake.

For intra-ARC injections, a guide cannula was inserted with the tip located at 2.8 mm caudal, 0.3 mm lateral to the bregma and 8.8 mm below the skull (38) in 9- to 10-wk-old rats. Ouabain (250 pmol/0.5 μl) was injected during the light phase (10:00) in ad libitum-fed animals, immediately followed by feeding experiments, in which food intake for 1 and 2 h was measured. After the feeding experiment, bilateral portions of ARC were immediately collected from hypothalamic slices for mRNA measurement. SSA412 (1.1 μg/0.5 μl) was injected after 24 h of fasting and 30 min prior to the start of feeding experiments, in which food intake for 1 and 2 h was also measured. For third ventricular injections, a guide cannula was inserted with the tip located at 2.5 mm caudal to the bregma in the midline and 8 mm below the surface of the skull, and ouabain (0.5 nmol/5 μl) and NPY Y1 receptor antagonist BIBP-3226 (30 nmol/5 μl) were injected for feeding experiments.

Measurements of c-fos expression.

Measurements of c-Fos expression were performed as previously reported (25). Briefly, at 2 h after ouabain injection, rats were transcardially perfused and coronal sections processed. Anti-c-Fos antiserum Ab-5 (dilution 1:40,000; Calbiochem, San Diego, CA), anti-AgRP antiserum (1:100; Neuromics Antibodies, Northfield, MN), and anti-proopiomelanocortin (POMC) antiserum (1:5,000, provided by Dr. Tanaka) were used for immunohistochemistry. This anti-POMC antiserum was produced by Dr. Shigeyasu Tanaka (40) and shown to be highly specific to POMC (25). The number of c-Fos-positive cells per section in ARC was counted.

Measurements of [Ca²⁺]_i and immunocytochemistry in single neurons of ARC.

Measurements of [Ca²⁺]_i in single neurons and subsequent immunocytochemical staining followed previous reports (18, 25, 28). In brief, single neurons were isolated from ARC of rats aged 5–7 wk by treatment with papain. The cells were loaded with 2 μM fura 2-acetoxymethyl ester (AM) (Dojin Chemical, Kumamoto, Japan). They were excited at 340 and 380 nm, respective emission signals at 510 nm (F340 and F380) were detected, and ratio (F340/F380) images were produced by Aquacosmos (Hamamatsu Photonics, Hamamatsu, Japan). After [Ca²⁺]_i measurements, neurons were immunocytochemically stained with rabbit antisera against AgRP (1:200; Phoenix Pharmaceuticals, Burlingame, CA), NPY (1:5,000; ImmunoStar, Hudson, WI), and POMC (1:1,000) (40).

Measurements of NAD(P)H level in single neurons of ARC.

Autofluorescence of NAD(P)H in single neurons isolated from ARC was measured with excitation at 360 nm and emission at 470 nm, as previously reported (43). Thereafter, the cells were loaded with fura 2-AM, and [Ca²⁺]_i was measured in the same neurons to examine the responses to LG.

Measurements of [Na⁺]_i in single neurons of ARC.

[Na⁺]_i was measured as previously reported (16), with slight modifications. Briefly, the cells were incubated with Na⁺ indicator SBFI-AM (2 μM; Molecular Probes, Carlsbad, CA) and Pluronic F-127 (0.0125%; Molecular Probes) for 2 h at 34°C. The procedures for ratiometric SBFI microfluorometry followed those for fura 2. After measurement of [Na⁺]_i, [Ca²⁺]_i was measured using fluo-4 with distinct excitation (490 nm) and emission wavelengths (520 nm) from SBFI. The cells were incubated with 2 μM fluo-4-AM (Molecular Probes) for 1 h at room temperature.

Immunohistochemistry of NKA α-isoforms and AgRP.

Coronal sections of hypothalamus containing ARC were prepared from SD rats aged 6 wk and were double-imuunostained using antibodies against NKA-α1, -α2, or -α3 and against AgRP (Table 1) according to previously described methods (25). Fluorescence images were acquired with an Olympus FV1000 confocal microscope (Tokyo, Japan).

Knockdown.

Target sequences for shRNAs of rat NKA-α1, -α2, and -α3 isoforms were chosen from accession nos. NM012504, NM012505 and NM012506: 5′-CCATCGTAAATACGGAACA-3′, 5′-GCCATCCAATGACAATTTAT-3′, and 5′-GCGATACCTGTTAGTGATGA-3′. Briefly, double-stranded DNA oligonucleotides consisted of the shRNA sense sequence, loop sequence TAGTGCTCCTGGTTG, shRNA antisense sequence, and string of six thymidine residues. They were cloned downstream of the mouse U6 promoter in a pBAsi-mU6 vector (Takara Bio, Tokyo, Japan). Cloned plasmid DNA was transfected using in vivo jetPEI (Polyplus-transfection SA, Illkirch, France). The mixture of plasmid and jetPEI was injected stereotaxically into bilateral ARC, located at 2.8 mm caudal, 0.3 mm lateral to the bregma, and 8.8 mm below the skull. Four days after operation, food intake following 16-h fasting was measured. Five days after operation, bilateral portions of ARC were immediately collected from hypothalamic slices for mRNA measurement. To examine cell viability, ARC slices and single ARC neurons were collected 5 days after shRNAs injection. Immunohistochemistry and [Ca²⁺]_i measurements were performed as described above.

Measurements of mRNA.

mRNA was measured by real-time PCR as previously reported (13). Primers were as follows (forward and reverse): NPY, 5′-TGGCCAGATACTACTCCGCTC-3′ and 5′-ATGGAAGGGTCTTCAAGCCT-3′; AgRP, 5′-GGTGCTAGATCCACAGAACCG-3′ and 5′-CCAAGCAGGACTCGTGCAG-3′; POMC, 5′-CCTCCATAGACGTGTGGAGCTG-3′ and 5′-AAGGGCTGTTCATCTCCGTTG-3′; GAPDH, 5′-GGCACAGTCAAGGCTGAGAATG-3′ and 5′-ATGGTGGTGAAGACGCCAGTA-3′; NKA-α1, 5′-TGCATTGACTTGGGCACTGAC-3′ and 5′-TTCAGCCTGTTCATAGGCCAGAG-3′; NKA-α2, 5′-CACCTGCTGAACCCACGTTATG-3′ and 5′-ATGTTCCCGAAAGTCAGGGATG-3′; NKA-α3, 5′-ACCGCACTGTCAACGACCTG-3′ and 5′-AAGGCTGTGTGGAACGTGAACTC-3′.

Statistics.

Factorial ANOVA followed by Bonferroni post hoc test was employed in four-group experiments, and paired or unpaired Student's t-tests were employed in two-group experiments. Data are shown as means ± SE. N signifies number of rats examined, and n signifies number of cells, slices, and sections examined. P < 0.05 was considered statistically significant.

Fasting suppresses NKA's ATP-hydrolyzing activity and substrate ATP level in ARC.

We examined whether the NKA enzyme activity in ARC is physiologically controlled under fed vs. fasted conditions. We measured the NKA's ATP-hydrolyzing activity and substrate ATP content, the two factors that determine the final NKA enzyme activity in living cells. In 24-h-fasted rats compared with ad libitum-fed rats, the ouabain-sensitive ATP-hydrolyzing activity of NKA decreased in ARC [100.0 ± 9.7% for fed (N = 11) vs. 54.0 ± 3.0% for fasted (N = 12), P < 0.05], LHA [100.0 ± 6.9% for fed (N = 11) vs. 57.8 ± 18.4% for fasted (N = 12), P < 0.05], and VMH [100.0 ± 6.0% for fed (N = 11) vs. 74.6 ± 2.8% for fasted (N = 12), P < 0.05], and the magnitude of reduction was greater in ARC than in LHA and VMH (Fig. 1A). However, the expressions of the catalytic NKA-α1, -α2, and -α3 proteins in ARC were unchanged (Fig. 1B). These results indicate that the ATP-hydrolyzing enzyme activity but not the expression of NKA molecule was suppressed. In rats fasted 24 h, the ATP content markedly decreased in ARC [100.0 ± 26.7% for fed (N = 5) vs. 32.6 ± 6.0% for fasted (N = 5), P < 0.05], whereas it only tended to decrease in LHA [100.0 ± 25.5% for fed (N = 5) vs. 62.2 ± 23.2% for fasted (N = 5)] and VMH [100.0 ± 19.4% for fed (N = 5) vs. 56.7 ± 18.6 for fasted (N = 5); Fig. 1C]. Furthermore, the ATP content in ARC of individual fed or fasted rats correlated significantly with the blood glucose concentration of the corresponding animal, indicating that ARC ATP content reflects the systemic energy state (Fig. 1D). These results demonstrate that the systemic energy state influences the final NKA enzyme/pump activity in ARC by regulating both the ATP-hydrolyzing activity and substrate ATP level.

Fasting-induced plasma ghrelin rise and glucose fall reduce NKA's ATP-hydrolyzing activity and substrate ATP level, respectively, in ARC.

We next investigated the messengers that link systemic energy states to NKA enzyme activity in ARC in rats. Under 24-h-fasted conditions compared with fed conditions, plasma level of ghrelin, the candidate hunger hormone considered to be a physiological meal initiator (2, 12, 33), was increased [36.2 ± 8.7 fmol/ml for fed (N = 6) vs. 68.7 ± 9.4 fmol/ml for fasted (N = 6), P < 0.05], and blood glucose level was decreased [95.0 ± 5.0 mg/dl for fed (N = 6) vs. 60.2 ± 2.8 mg/dl for fasted (N = 6), P < 0.01] (Fig. 1, E and F). In isolated ARC slices, administration of ghrelin significantly decreased the ATP-hydrolyzing activity of NKA [100.0 ± 18.7% for vehicle (n = 6) vs. 47.0 ± 13.1% for ghrelin (n = 6), P < 0.05; Fig. 1G], whereas ATP content was unaltered (Fig. 1H). Conversely, ATP content measured in ARC slices was significantly decreased by incubation in 2.8 mM glucose (2.8G) compared with 8.3G [100.0 ± 14.0% for 8.3G (n = 8) vs. 56.1 ± 9.7% for 2.8G (n = 8), P < 0.05] and in 0.5G compared with 2.8G [100.9 ± 10.4% for 2.8G (n = 10) vs. 74.5 ± 3.8% for 0.5G (n = 10), P < 0.05] (Fig. 1J), whereas ATP-hydrolyzing activity of NKA was not different between 8.3G and 2.8G and between 2.8G and 0.5G (Fig. 1I). These results suggest that fasting downregulates NKA enzyme/pump activity in ARC via two modes (see Fig. 8): increased plasma ghrelin reduces ATP-hydrolyzing activity of NKA, and lowered blood glucose reduces its substrate ATP level.

Lowering glucose reduces intracellular NAD(P)H level and increases Na⁺ concentration in ARC GI neurons.

To assess whether the fasting- and low blood glucose-induced reduction of ATP level or energy state in ARC slices also takes place at the neuronal level, we measured the autofluorescence of NAD(P)H, a molecule that couples glucose metabolism to intracellular energy production in islet β-cells and hypothalamic neurons (1, 15). We found that LG from 8.3G to 2.8G reduced NAD(P)H level in a single ARC neuron, wereas a subsequent shift from 2.8G to 8.3G increased it (Fig. 2A, left). This neuron, after being loaded with fura 2-AM, subsequently responded to the same LG protocol with increases in [Ca²⁺]_i (Fig. 2A, right; n = 5), showing that this neuron was a GI neuron. Average NAD(P)H autofluorescence at 2.8G was significantly smaller than that at 8.3G in GI neurons (n = 5, Fig. 2B). These results indicate that LG decreases the intracellular energy state and increases neuronal activity in GI neurons of ARC. Since more than 90% of the ARC GI neurons are IR to NPY (28), the observed responses to LG may take place, at least partly, in NPY/AgRP neurons.

NKA activity was assessed by measuring [Na⁺]_i with SBFI. LG from 8.3G to 2.8G increased [Na⁺]_i in a single ARC neuron (Fig. 2C, left). This neuron, after being loaded with Fluo-4, subsequently responded to the same LG protocol with [Ca²⁺]_i increases (Fig. 2C, right), showing that this neuron was a GI neuron. By contrast, LG from 8.3G to 2.8G little altered [Na⁺]_i (Fig. 2D, left) in the ARC non-GI neurons that did not respond to LG with [Ca²⁺]_i increases (Fig. 2D, right). Average amplitude of [Na⁺]_i increases was significantly larger in GI than in non-GI neurons (Fig. 2E). According to a calibration curve, LG from 8.3G to 2.8G increased [Na⁺]_i by 7.5 ± 1.1 mM in GI neurons, consistent with a previous report (35). These results showed that LG induces [Na⁺]_i increases and neuronal activation in ARC GI neurons.

Intra-ARC injection of NKA inhibitor promotes feeding and NPY mRNA expression in ARC.

The results that fasting suppressed NKA enzyme activity, most markedly in ARC, prompted us to examine whether the energy state-sensing ARC NKA is related to feeding behavior. For this, we manipulated NKA in ARC by injecting its specific inhibitor and activator. Focal injection of the NKA inhibitor ouabain (250 pmol/0.5 μl) into ARC in ad libitum-fed rats during the light phase markedly increased food intake for 1 and 2 h (Fig. 3A). This condition also increased NPY, tended to increase AgRP, and tended to decrease POMC mRNA expressions in ARC (Fig. 3B). Thus, ouabain increased mRNA expression for orexigenic peptides and tended to decrease that for anorexigenic peptide. These changes could account for the feeding stimulation.

Third ventricular injection of NKA inhibitor induces feeding in an NPY-dependent manner and c-fos expression in AgRP neurons in ARC.

Since fasting suppressed ATP-hydrolyzing activity of NKA not only in ARC but in LHA and VMH, and the ARC NPY/AgRP neurons project to various brain regions, the effect of intracerebroventricular injection of ouabain and NPY antagonist was examined. The third ventricular injection of the NKA inhibitor ouabain, compared with vehicle, increased food intake, and the ouabain-induced food intake was blocked by simultaneous third ventricular injection of NPY Y1 receptor antagonist BIBP-3226 (Fig. 3C). The third ventricular injection of ouabain also induced c-Fos expression in AgRP neurons in ARC (Fig. 3, D and E), and the c-Fos induction was much greater in AgRP (Fig. 3F) than in POMC neurons (Fig. 3G).

NKA inhibitor ouabain activates NPY/AgRP neurons more potently than POMC neurons.

Administration of ouabain (100 μM) increased [Ca²⁺]_i in single ARC neurons that were subsequently shown to be immunoreactive (IR) to NPY (Fig. 4A), AgRP (Fig. 4B), and POMC (Fig. 4C). Ouabain evoked [Ca²⁺]_i responses in 15 of 22 (59.1%) NPY-IR neurons, 17 of 32 (53.1%) AgRP-IR neurons, and 13 of 36 (36.1%) POMC-IR neurons. These results indicate that ouabain activates NPY/AgRP neurons in ARC. The rest of [Ca²⁺]_i and immunohistochemical experiments were performed mainly for AgRP rather than NPY (neurons), since this peptide is expressed exclusively in ARC and is colocalized with NPY (10). Ouabain at a lower concentration of 1 μM induced solid increases in [Ca²⁺]_i in AgRP-IR neurons (Fig. 4D) but much smaller increases in [Ca²⁺]_i in POMC-IR neurons (Fig. 4E). The average amplitude of the [Ca²⁺]_i responses to ouabain at 1 and 100 μM were both much greater in AgRP-IR neurons than in POMC-IR neurons (Fig. 4F). These results taken together suggest that NKA suppression is linked to cellular activation in NPY/AgRP neurons more efficiently than in POMC neurons in ARC.

NKA activator SSA412 suppresses fasting-induced feeding behavior and LG-induced neuronal activation.

We found that fasting decreases NKA activity in ARC. If this change is related to the fasting-evoked appetite, the food intake following fasting would be supressed by activation of NKA. When the specific NKA activator SSA412 (1.1 μg/0.5 μl) was focally injected into ARC, food intake at 1 and 2 h of refeeding following 24-h fasting was significantly attenuated (Fig. 5A). Furthermore, whether the decreased NKA activity mediates the LG-induced activation of GI neurons was investigated. In control experiments, LG from 8.3G to 2.8G induced [Ca²⁺]_i increases twice, first immediately after the treatment with control nonimmune IgG (Fig. 5B, left) and subsequently following 70 min of incubation in HKRB (Fig. 5B, right). The majority of these GI neurons were IR to NPY (Fig. 5C). In test experiments, soon after the treatment with SSA412, LG from 8.3G to 2.8G failed to significantly increase [Ca²⁺]_i in an ARC neuron (Fig. 5D, left). This neuron, after incubation in HKRB to washout the influence of the antibody, responded to the same LG protocol with [Ca²⁺]_i increases (Fig. 5D, right), showing the recovery of the response to LG. Thus, [Ca²⁺]_i responses to LG in GI neurons were significantly inhibited by SSA412 compared with that with control IgG and that after washout incubation (Fig. 5E).

NKA-α isoforms are present in ARC and localized abundantly in NPY/AgRP neurons.

The finding that ouabain induced much greater increases in [Ca²⁺]_i in NPY/AgRP neurons than in POMC neurons prompted us to study the molecular identity of NKA in ARC NPY/AgRP neurons. We examined the colocalization of the catalytic NKA-α isoforms α1, α2, and α3 with AgRP in ARC. As shown in Fig. 6, in the first, third, and fifth lanes, double immunohistochemistry revealed that the neuron-preferential isoforms α1 and α3 proteins (8) were abundantly located in ARC, confirming a previous report (3). The glial cell-preferential isoform α2 (8) was also observed. In an expanded scale, the three NKA-α isoforms were located in AgRP-IR neurons in ARC (Fig. 6, 2nd, 4th, and 6th lanes). The immunoreactivities for NKA-α3 and -α1 were frequently and intensely colocalized with the AgRP immunoreactivity (Fig. 6, 6th lane), indicative of high-level expression of α3 and α1 in ARC AgRP neurons.

ARC-selective knockdown of NKA-α3 impairs feeding behavior without altering the number and function of AgRP neurons.

To further assess the role of these α-subunit isoforms, shRNAs for NKA-α1, -α2, and -α3 were focally injected into ARC, which resulted in reduction of the mRNA expressions of NKA-α1, -α2, and -α3 to the levels ∼ 30% (Fig. 7A). In rats injected with shRNA for α3, food intake during the 2-h refeeding period following 16 h of fasting was significantly reduced, whereas that during the 1-h refeeding period tended to decrease and 24-h food intake was not changed (Fig. 7B). Treatment with shRNAs for NKA-α1 and -α2 tended to decrease food intake during 1- and 2-h refeeding periods but not statistically significantly. The treatment with NKA-α3 shRNA also significantly decreased NPY mRNA expression in ARC at 2 h of refeeding following 16 h of fasting (Fig. 7C). In the rats treated with NKA-α3 shRNA, the number of AgRP-IR neurons in ARC were not altered (Fig. 7, D and E), and the [Ca²⁺]_i responses to glutamate in ARC neurons were not altered (Fig. 7, F and G) compared with control rats treated with scramble shRNA. These results indicate that the treatment with NKA-α3 shRNA little affected the survival and functionality of ARC AgRP neurons. These results taken together suggest that NKA-α3 is implicated in the fasting-induced feeding and NPY expression in ARC.