Leptin, a polypeptide hormone produced mainly by adipocytes, has diverse effects in both the brain and peripheral organs, including suppression of feeding. Other than mediating leptin transport across the blood-brain barrier, the role of the endothelial leptin receptor remains unclear. We recently generated a mutant mouse strain lacking endothelial leptin receptor signaling, and showed that there is an increased uptake of leptin by brain parenchyma after its delivery by in situ brain perfusion. Here, we tested the hypothesis that endothelial leptin receptor mutation confers partial resistance to diet-induced obesity. These ELKO mice had similar body weight and percent fat as their wild-type littermates when fed with rodent chow, but blood concentrations of leptin were significantly elevated. In response to a high-fat diet, wild-type mice had a greater gain of body weight and fat than ELKO mice. As shown by metabolic chamber measurement, the ELKO mice had higher oxygen consumption, carbon dioxide production, and heat dissipation, although food intake was similar to that of the wild-type mice and locomotor activity was even reduced. This indicates that the partial resistance to diet-induced obesity was mediated by higher metabolic activity in the ELKO mice. Since neuronal leptin receptor knockout mice show obesity and diabetes, the results suggest that endothelial leptin signaling shows opposite effects from that of neuronal leptin signaling, with a facilitatory role in diet-induced obesity.
- blood-brain barrier
- metabolic phenotype
leptin is produced mainly by adipose tissue and has pleiotropic functions in peripheral organs and central nervous system (CNS), the brain being a major target for feeding suppression. The importance of CNS leptin signaling is shown by diabesity in neuronal leptin receptor knockout mice. These mice are generated by breeding of CamKIIα-Cre recombinase heterozygote mice with the leptin receptor LepR-floxed mice, resulting in deletion of exon 17 of LepR that encodes a Janus kinase 2 (JAK2) docking site for activation of signal transducer and activator for transcription (STAT)-5B. There is also a frameshift mutation disrupting the terminal exon 18b, which contains a binding site of the long isoform LepRb (ObRb) for STAT3 and STAT1. Because all membrane-bound isoforms of LepR contain exon 17, its deletion results in the loss of signal transduction capabilities of all membrane-bound LeptR isoforms. These neuronal-specific LepR mutant mice are obese and develop diabetes early (19).
Endothelial leptin receptors are also important, particularly in receptor-mediated transport of leptin across the blood-brain barrier (BBB) (2, 18, 23, 30). During the course of transport, leptin also activates endothelial signaling through its receptors ObRa (LEPRa) or ObRb (LEPRb), an event potentiated by certain G protein-coupled receptors, including corticotropin releasing hormone receptors and melanocortin receptors (26, 32). ObRa has a short cytoplasmic tail and is the most abundant receptor isoform in cerebral microvessels (4). However, the long cytoplasmic tail isoform ObRb is also present and shows developmental changes at the BBB as well as hypothalamus (23).
It is not known, however, whether endothelial leptin signaling affects the metabolic phenotype or CNS function. Endothelial specific leptin receptor mutant (ELKO) mice are generated by crossing Tie2-cre heterozygous mice with LepR-floxed mice and characterized in our previous study (15). When fed with regular rodent chow, ELKO mice show normal body weight and apparent influx of leptin across the BBB, although brain parenchymal uptake of leptin is increased in studies by in situ brain perfusion (15). Here we test the overall hypothesis that endothelial leptin signaling modulates the susceptibility of a mouse to diet-induced obesity.
MATERIALS AND METHODS
High-fat diet challenge and metabolic phenotyping.
The generation and characterization of the ELKO mice with endothelial specific mutant ObR expression are described in detail elsewhere (15). Following a protocol approved by the Institutional Animal Care and Use Committee, male ELKO and littermate controls were group-housed and fed with either 45% high-fat diet (HFD) or control diet (CD; Research Diets, New Brunswick, NJ) after measurement of baseline body weight (BW) and fat composition at age 6 wk. The HFD (D12451) contains 20 kcal% protein, 35 kcal% carbohydrate, and 45 kcal% fat. The control diet (D12450B) has the same total energy content of 20 kcal% protein, 70 kcal% carbohydrate, and 10 kcal% fat, with a higher percent corn starch and sucrose and a lower percent maltodextrin 10 and lard, whereas all other components are the same as in the HFD. BW was monitored weekly and body composition (fat and muscle mass expressed as %BW) was determined monthly until 7 mo of age by nuclear magnetic resonance (NMR, Bruker Optics, Billerica MA) as described previously (25).
At 7 mo old, the mice were placed in Oxymax metabolic chambers (Columbus Instrument, Columbus, OH) that allow real-time monitoring of food intake, oxygen consumption (V̇o2), carbon dioxide production (V̇co2), respiratory quotient (RER, equivalent to V̇co2/V̇o2), heat dissipation, and activity. The mice were adapted to single-caged metabolic chambers for 3 days before initiating data collection for 3 consecutive days. To determine food intake, pellets were processed in a grinder and the crushed food was placed in round trays inside the metabolic chambers.
Circadian rhythms of each variable were analyzed by the cosinor method, as described previously (3, 12, 27). Midline Estimating Statistic of Rhythm (MESOR), amplitude, and acrophase were determined. Least-squares spectra were computed for blocks of 3-day spans, in the frequency range of 1 cycle/3 days to 10 cycles/day. Phase-unweighted average spectra were obtained by averaging the amplitudes at each frequency across all mice in a given group [wild type (WT) or ELKO; CD or HFD]. Parameter tests were also performed, taking the phase information into account. Student t-tests were used to compare amplitudes at each frequency between ELKO and WT animals on the CD or HFD.
Serum glucose measurement and ELISA.
Arterial blood was collected to prechilled tubes by dissecting a common carotid artery after mice are anesthetized, immediately before decapitation. Serum was obtained by centrifugation of whole blood after overnight storage at 4°C. To measure leptin-like immunoreactivity (LI), 50 μl of diluted serum from ELKO (n = 3; 2 male and 1 female) or WT control mice (n = 4; 2 male and 2 female) 2–2.5 mo old or 50 μl of standard was added to a Quantikine Mouse leptin ELISA kit (R&D Systems, Minneapolis, MN; cat. no. MOB00). The range of the standards was 62.5 pg/ml–4 ng/ml, and the control contained assay diluents only. Triplicates were analyzed following the assay protocol to obtain an average for each sample, and optical density was obtained at 450 nm on a microplate reader.
Insulin-LI was measured by a mouse Ultrasensitive ELISA kit from ALPCO Diagnostics (Salem, NH; cat. no. 80-INSMSU-E01). The range of the standards is 0.188–6.9 ng/ml. Glucose measurement was achieved by use of a ReliOn Confirm Blood glucose monitoring system. The reading from the glucometer was extrapolated from standards.
Microvessel isolation, leptin treatment, and immunocytochemistry (ICC) for ObRb and pSTAT3.
Cerebral microvessels were harvested following an established protocol (11) with modifications. Adult WT or ELKO mice were anesthetized and decapitated. Brain was dissected to obtain cerebral cortex devoid of circumventricular organs. The cortices were maintained at 4°C in oxygenated PBS buffer (2.7 mM KCl, 1.46 mM KH2PO4, 136.9 mM NaCl, and 8.1 mM Na2HPO4, supplemented with 5 mM d-glucose and 1 mM sodium pyruvate, pH 7.4), in a concentration of one mouse cortex/1 ml PBS buffer. The cortices were homogenized with a Dounce homogenizer, and mixed with 3 ml of 20% Ficoll (Sigma, St. Louis, MO) for a final concentration of 15%. After centrifugation at 5,800 g for 20 min at 4°C, pellets was resuspended in PBS containing 1% bovine serum albumin (BSA), and passed through 100 μm cell restrainer. The flow-through then passed a 40-μm cell restrainer. The unfiltered microvessels were collected by reversing the restrainer and washes with PBS/BSA to a BSA precoated petri dish. The purified microvessels were then collected to a 15 ml tube and centrifuged at 300 g for 10 min at 4°C. The pellet was further suspended in 1 ml PBS buffer without BSA but containing 5 mM d-glucose and 1 mM sodium pyruvate.
For ICC, the microvessels were treated at 37°C for 15 min with either leptin (100 ng/ml, R&D Systems) or PBS only (vehicle control). The microvessels were then fixed with 4% paraformaldehyde at room temperature for 15 min, and allowed to air dry on a slide for 1 h. After three PBS washes, the microvessels were permeabilized with 0.1% Triton X-100 at room temperature for 10 min. After six PBS washes, 10% normal donkey serum was applied to block nonspecific binding. Primary antibody incubation at 4°C overnight used either chicken anti-ObRb (1:100, Neuromics), or rabbit anti-pSTAT3 (Tyr705, 1:100, Cell Signaling). After six PBS washes, the microvessels were incubated with the respective secondary antibody at room temperature for 1.5 h, including FITC-conjugated donkey anti-chicken (1:200, Jackson Laboratory) or Alexa594-conjugated donkey anti-rabbit IgG (1:400, Invitrogen). The slides were then thoroughly washed with PBS eight times and coverslipped with Prolong Gold mounting medium. Images with captured by use of an Olympus confocal microscope with a 60× oil lens with additional zoom-in and z stacking.
Differences in body weight and fat composition among the four groups were determined by repeated-measures ANOVA, followed by Tukey's post hoc test. For accumulated food intake, or accumulative activities across the same circadian cycle, two-way ANOVA was performed to determine the effects of strain (ELKO mutation vs. WT), diet (HFD vs. CD), and their interactions. For blood glucose, leptin, and insulin concentration measurements, independent t-tests were performed to compare the difference between naive ELKO and WT mice.
Adiposity and food intake in the ELKO mice on CD and HFD.
Four groups of mice were studied simultaneously across 21 wk for body weight and percent fat (per g of body weight): WT on CD, ELKO on CD, WT on HFD, and ELKO on HFD (n = 6 /group). The body weights of the ELKO and WT mice were similar at the initial measurement starting when the mice were 6 wk old. When fed with CD, both groups showed steady weight gain. ELKO mice showed a slightly higher body weight than the WT littermate controls while on CD. HFD increased body weight in both ELKO and WT mice; after 5 wk of feeding, the increase in body weight was consistently greater in the WT mice than the ELKO mice. This was most pronounced after age 20 wk (14 wk into HFD; Fig. 1A). The amount of fat was the same among groups at baseline, but at ages of 15 and 28 wk, the percent fat was lower in the ELKO group than the WT group on HFD (Fig. 1B).
Toward the end of the weekly fat and weight monitoring, mice were habituated in metabolic chambers for measurement of food intake and other metabolic activities. The ELKO mice had reduced food intake during the light span; the difference in the 12-h food intake, determined by t-test, was significant in the light span (P = 0.05 for day 1 and P < 0.05 for day 2). However, the circadian rhythm and accumulative 24-h food intake were not significantly different between the ELKO and WT groups. HFD reduced food intake, in both amount and calories, compared with the CD for the respective strain (P < 0.05). There was no difference between the ELKO and the WT mice (Fig. 2, A and B). Two-way ANOVA showed that the change of food intake was caused by diet (P < 0.0005 for each day), but not by strain, although there was a minor (P < 0.05 for day 2 only) interaction between diet and strain.
Circadian rhythm of V̇o2, and the effects of ELKO mutation and HFD.
Circadian rhythm of V̇o2 was measured in the metabolic chamber, to determine the effects of strain and diet and their interactions (n = 5–6/group). In all groups, V̇o2 showed a 24-h rhythm with a peak in the dark span and a nadir in the light span. ELKO alone did not affect the circadian characteristics of V̇o2. When both strains of mice were placed on HFD, the difference of MESOR of V̇o2 in the ELKO and WT mice was even more pronounced (P < 0.05). In the WT mice, HFD induced a decrease of circadian amplitude (P = 0.01). However, the ELKO mice did not show a significant reduction of the circadian amplitude of V̇o2 while fed with HFD. There was a larger circadian amplitude when the ELKO HFD group was compared with the WT HFD group (P < 0.05) (Fig. 3A).
In all four groups of mice, a prominent 24-h rhythm of V̇o2 was shown by a significant major peak in the least-squares spectra. ELKO mice had higher V̇o2 than WT mice, both on CD (P < 0.05) and HFD (P < 0.05). Besides the 24-h rhythm, ELKO mice also had larger amplitudes of the 12-h (P < 0.05) and 6-h (P < 0.05) components of the circadian rhythm of V̇o2 while on CD (Fig. 3B).
In summary, ELKO on CD did not differ significantly from the WT mice on CD on 24-h oxygen consumption, but they did have greater amplitudes of the 12-h and 6-h ultradian rhythms. HFD reduced V̇o2 circadian amplitude significantly in the WT mice, but to a much lesser extent in the ELKO mice. Thus ELKO mutation appeared to help the mice preserve circadian features of oxygen consumption, both under CD and HFD.
ELKO and HFD modulate the circadian rhythm of V̇co2.
ELKO mice on CD did not differ from WT mice on CD in regard to their MESOR and circadian amplitude of V̇co2 (Fig. 4A). However, the ELKO mice showed larger amplitudes of the 12-h (P < 0.05) and 6-h (P < 0.05) components (Fig. 4B). Compared with the control diet, HFD was associated with a lower V̇co2 in both WT (P < 0.01) and ELKO (P < 0.01) mice. The HFD was also associated with a smaller circadian amplitude of V̇co2 in both WT (P < 0.05) and ELKO mice (P = 0.052) (Fig. 4B). The amplitude of other spectral components was also decreased on the HFD, notably that of the 12-h component of ELKO mice (P < 0.05). The alteration of the spectral structure on HFD is particularly evident for WT on HFD, with diminished amplitudes of the 8-h (P < 0.05), 4-h (P = 0.01), and 3.4-h (P < 0.05) components.
In summary, the V̇co2 was similar in the ELKO and WT mice on control diet. A HFD reduced the V̇co2 in both strains, but the decrease was more pronounced in the WT mice. Thus, although the obese mice showed a reduction in metabolic activity reflected by decreased V̇co2, ELKO mutation helped to preserve the circadian rhythm of V̇co2.
Circadian rhythm of respiratory quotient and effects of ELKO and HFD.
There was no difference in the circadian parameters of RER between WT and ELKO mice (Fig. 5A). Compared with the CD, the HFD was associated with a lower RER in both WT (P < 0.001) and ELKO mice (P < 0.05) (Fig. 5B). The amplitude of the 12-h amplitude in the ELKO mice was decreased on the HFD (P < 0.05).
Altogether, the ELKO mutation did not cause a significant increase of fat oxidation at baseline, as the reduction of RER did not reach statistical significance. The HFD decreased RER in both strains as a result of higher fat utilization.
Effects of ELKO and HFD on circadian rhythm of heat production and activity levels.
Accumulative heat dissipation was higher in the dark span than the light spans for either ELKO or WT mice while on CD. There was more heat dissipation in the ELKO mice than in the WT mice (P < 0.05). HFD blunted the circadian rhythm of heat dissipation, particularly in the ELKO group, which had higher heat dissipation than the WT group (Fig. 6A). The infracadian rhythm persisted in all groups, and ELKO mice on HFD showed higher amplitudes, but the difference was not statistically significant, probably resulting from wide individual variation (Fig. 6B). Nonetheless, cosinor analysis indicated that the 12-h amplitude was larger in ELKO mice on the control diet (P < 0.05, Fig. 6C). Overall, the results show that ELKO mice had more heat dissipation while on CD and HFD conditions.
Activity level (x-axis) did not show significant changes among groups (Fig. 7A). However, accumulative activity level was lowest in ELKO mice on the HFD, indicating that the lower fat percent and body weight in this group of mice was not caused by increased activity (Fig. 7, A and B).
Blood leptin concentration was increased in ELKO mice.
Serum leptin concentration was estimated by ELISA in ELKO and WT littermate controls 2–2.5 mo old. There was a significant increase of leptin-like immunoreactivity (LI) in the ELKO mice, both male and females. The pooled male and female results are shown in Fig. 8A. The 15-fold increase of leptin concentration contrasts with a lack of significant change of serum insulin and blood glucose in the same mice (data not shown).
To confirm that cerebral microvessels from the ELKO mice do not have leptin-induced signaling, microvessels from the ELKO mice and WT mice were subjected to leptin or vehicle treatment for 15 min, before ICC for ObRb and pSTAT3. ObRb immunoreactivity was mainly present in the basolateral side of microvessels from the WT mice but the fluorescent intensity was diminished in the ELKO mice. Leptin treatment for 15 min reduced ObRb fluorescence in the WT microvessels (Fig. 8B). Leptin treatment also induced an increase of pSTAT3 immunoreactivity in WT microvessels, with pSTAT3 translocated to nucleus costained with DAPI. By contrast, ELKO microvessels lacked pSTAT3 immunoreactivity, and leptin treatment did not induce nuclear pSTAT3 activation (Fig. 8C). The results indicate that leptin-induced pSTAT3 signaling in microvessels was indeed abolished in ELKO mice.
The endothelial leptin receptors play important biological functions, at least in the cerebral microvessels composing the BBB, where they mediate leptin transport (2, 5, 13, 31). There are developmental changes in the BBB with significantly higher levels of expression of ObRa as well as ObRb in neonatal mice (23), and upregulation of ObRa in rats fed with a HFD (5). We have already shown that all membrane-bound ObR isoforms can effectively endocytose leptin given sufficient levels of expression (30). This contrasts with an antagonistic role of the soluble receptor ObRe in inhibiting leptin transport (29). We have also shown in vitro that even in the absence of a signaling cytoplasmic domain, truncated LEPR can still mediate leptin endocytosis and intracellular trafficking (28). Thus our endothelial specific ObR mutant mice provide an ideal model to test the consequence of lack of endothelial leptin signaling on metabolic phenotypes, as leptin can still reach its CNS targets in these mice (15).
While on a control diet, the ELKO mice showed normal body weight and percent fat, and the body weight gain was only slightly higher than the littermate controls. While on a HFD, ELKO mice showed a consistently lower body weight than the WT mice, suggesting partial resistance to diet-induced obesity. This was confirmed by measurement of the percent body fat that showed a significantly smaller increase at the end of the HFD. To determine the mechanisms underlying the body phenotype changes, we tested all arms of metabolic activity: food intake, substrate utilization, heat dissipation, and activity levels. Metabolic chamber analyses were performed in ELKO and WT mice in the basal state and after metabolic challenge by HFD feeding.
Circadian rhythm analysis of metabolic variables by the cosinor method provides a more accurate estimate of metabolic activity across 24-h cycles than conventional methods. In combination with calculation of 12-h accumulative changes, cosinor analysis showed that on a CD the ELKO mutation is associated with a reduction of food intake in the light span, no change in oxygen consumption or carbon dioxide production, and a mild increase of heat dissipation without a change in activity level. By contrast, the HFD resulted in decreased food intake in both light and dark spans, reduction of VO2 present only in WT mice (not in ELKO mice), reduction of VCO2 in both strains but more so in WT than ELKO mice, blunted circadian rhythm of heat dissipation more apparent in ELKO than WT mice, and a mild increase of activity in WT but decrease in the ELKO mice. Altogether, the metabolic chamber analyses showed that ELKO mice had a mild increase of heat dissipation while on the CD, but a prominent protective role against HFD-induced obesity, an effect mainly explained by an increased energy expenditure not resulting from reduction of food intake or higher locomotor activity.
The ELKO mice also showed different infracadian parameters than the WT mice. In particular, the 12-h and 6-h components of V̇o2 and V̇co2 both had higher amplitudes in the ELKO mice on control diet. Heat dissipation also showed a larger 12-h circadian amplitude in the ELKO mice. HFD differentially regulated the infracadian components, and overall the ELKO mice showed better preserved infracadian as well as circadian profiles.
This partial resistance to obesity in the ELKO mice contrasts with the extreme obese phenotype and diabetes in neuronal specific LEPR knockout mice (19). Thus, while the neuron-specific leptin receptor transgene completely rescues obesity, diabetes, and infertility in db/db mice lacking LEPRb expression (8), indicating a protective role of neuronal LEPR against obesity and metabolic/endocrine disturbance, endothelial leptin signaling appears to play an opposite role.
Different, and even opposite, effects of cell-type specific leptin signaling is not unusual. In the brain, we have recently shown that astrocytic activity appears to attenuate neuronal leptin signaling in agouti viable yellow (Avy) mice with adult-onset obesity, in that inhibition of astrocytic activity increases neuronal leptin uptake (21, 24). ObR is present in astrocytes (16) that are an integral part of the BBB. We have also shown that both Avy mice (22) and diet-induced obese mice (14) have upregulation of their astrocytic ObR. Thus leptin signaling is cell-specific and each cell type makes unique contributions to the overall CNS metabolic and neurobehavioral phenotype. Endothelial leptin signaling may be beneficial during famine and cachexia but appears to be deleterious to brain function during energy surplus.
We further hypothesize that increased blood leptin concentration is a mechanism by which ELKO show partial resistance to diet-induced obesity. Indeed, a significantly higher leptin concentration, or at least leptin-like immunoreactivity shown by ELISA assay, was seen in the naive ELKO mice that have similar body weight and fat percent as the WT littermates. The higher blood concentration of leptin in the ELKO mice is consistent with end-organ resistance in the absence of cellular signaling (17). It is possible that ELKO also have reduced leptin excretion from the kidney, which is responsible for 80% of clearance, although megalin as well as leptin receptors may be involved in renal degradation and clearance of leptin (7, 9, 10, 20). Leptin is known to increase lipid oxidation (1, 6). The higher metabolic activity of the ELKO mice mainly resides in great energy expenditure without a reduction of food intake or increase of activity. Thus increased amount or activity of brown adipose tissue with higher uncoupling protein expression might be potential targets of the increased leptin.
In summary, the ELKO mice showed an absence of pSTAT3 signaling in response to leptin in endothelial cells, reduced food intake in the light span, and a better metabolic profile in response to a high-fat diet as a result of a greater increase of energy expenditure. The partial resistance of the ELKO mice to diet-induced obesity shows a novel role of endothelial leptin signaling in CNS activity, with endothelial ObR functioning as a negative regulator of diet-induced obesity.
This research was supported by National Institutes of Health Grants DK-54880 and DK-92245 to A. J. Kastin and NS-62291 to W. Pan.
No conflicts of interest, financial or otherwise, are declared by the author(s).
Author contributions: W.P. and A.J.K. conception and design of research; W.P., H.H., B.J., Y.W., and X.W. performed experiments; W.P., H.H., G.G.C.-G., B.J., F.H., and A.J.K. analyzed data; W.P., H.H., G.G.C.-G., F.H., and S.C.C. interpreted results of experiments; W.P., H.H., and G.G.C.-G. prepared figures; W.P. and A.J.K. drafted manuscript; W.P., H.H., G.G.C.-G., and A.J.K. edited and revised manuscript; W.P., G.G.C.-G., B.J., Y.W., F.H., X.W., S.C.C., and A.J.K. approved final version of manuscript.
- Copyright © 2012 the American Physiological Society