Obesogenic and Diabetogenic Effects of High-Calorie Nutrition Require Adipocyte BK Channels

Obesity is caused by high calorie intake, typically calories derived from dietary fats or sugars. Over time, an imbalance between consuming nutrients and burning calories leads to a massive increase in fat mass, which is—among other factors—a major cause of insulin resistance and diabetes (1,2). In addition, genetic mutations may cause excessive lipid accumulation and thereby a morbid increase in body weight (BW), emphasizing the multifactorial etiology of chronic excessive weight gain (3,4). It has become increasingly clear that a variety of K⁺ ion channels—that is, K⁺ channels expressed within the brain and in the periphery, possibly by complex effects on appetite and satiety, energy expenditure, glucose balance, and/or fat cell function—are involved in the pathophysiology of obesity and related disorders (5). Accordingly, a genome-wide association study recently identified the human KCNMA1 gene, encoding for the pore-forming subunit of the large-conductance Ca²⁺-activated K⁺ channel (BK), as a novel susceptibility locus for obesity (6). BK channel mRNA levels in variant carriers were significantly increased in white adipose tissue (WAT) and adipose tissue–derived cells, suggesting a pathogenic role for fat cell BK in metabolism. Others reported electrophysiological evidence for BK channels in human preadipocytes, and BK knockdown, or its pharmacological inhibition, further revealed a possible link between channel activity and the proliferative capacity of these cells (7), whereas Ca²⁺-activated K⁺ channels of the IK type seem to be dominant in the widely studied murine preadipocyte model 3T3-L1 (8). However, because both peripheral and central organs involved in the control of metabolism express BK, it is difficult to estimate the impact of preadipocyte BKs on fat cell formation and adiposity in vivo. For example, a direct modulation of BK by fatty acids and their metabolites seems to provide a possible link between lipid-mediated effects on the channel and altered vascular functions in hypertriglyceridemia secondary to obesity (9–13). Previous analyses of global BK knockout mice imply that both glucose-induced insulin secretion and endocrine output from the hypothalamic-pituitary-adrenal (HPA) axis require endogenous BK channels (14,15). However, it has not been determined whether these dysregulations in β-cell or HPA function caused metabolic abnormalities in vivo (16). Moreover, several studies found that BK promoted the inhibitory effects of leptin signaling, via phosphoinositide 3-kinase in hippocampal neurons (17) and in mouse chromaffin cells (18), suggesting that neuronal circuits controlling appetite and energy expenditure may also depend on functional BK. A malfunction of neuronal circuits has been widely appreciated as causing excessive fat storage (19), and indeed a number of studies found evidence for BK in the brainstem, different hypothalamic nuclei, parts of the cortex and limbic system—hence in brain regions that are implicated in the central control of food intake, appetite, and energy expenditure (20,21).

The recent discovery of functional channels with properties similar to BK in fat cell progenitors (7), together with the link between high-fat-cell BK mRNA expression levels and morbid adiposity (6), suggest that BK activity in the adipose tissue plays physiologically and pathophysiologically important roles for weight control, although these specific roles remain to be discovered. To establish endogenous BK channels as potential modulators of fat deposition and excessive weight gain, herein we assess the susceptibility of global and adipocyte-specific BK mouse mutants to genetic and dietary causes of adiposity. Our approach to validate BK as a novel player in the response to obesogenic factors in vivo should also direct future studies on the pharmacotherapy of adiposity and related disorders.

All animal experiments were performed with the permission of the local authorities and conducted in accordance with German and/or accordingly U.K. legislation on the protection of animals. Animals were housed in cages under controlled environmental conditions, with temperatures maintained between 21°C and 24°C, humidity at 45–55%, and 12-h light/12-h dark cycle.

Global BK channel–deficient mice (genotype BK^L1/L1) and their heterozygous (genotype BK^L1/+) and wild-type littermates (genotype BK^+/+) on a C57BL/6N background were generated and maintained at the Institute of Pharmacy, Department of Pharmacology, Toxicology and Clinical Pharmacy, University of Tübingen, Tübingen, Germany, as described previously (22). To produce BK^L1/L1 mice that are also unable to produce functional leptin (genotype BK^L1/L1;B6.V-Lep^ob/ob [BK^L1/L1;ob/ob]), BK^L1/+ animals were crossed with mice carrying a heterozygous mutation of the leptin gene B6.V-Lep^+/ob (ob/+) obtained from Charles River Laboratories (Sulzfeld, Germany). Intercrossing of double-heterozygous animals (genotype BK^L1/+;B6.V-Lep^ob/+ [BK^L1/+;ob/+]) produced homozygous obese BK^L1/L1;ob/ob, obese BK^+/+ (genotype BK^+/+;B6.V-Lep^ob/ob [BK^+/+;ob/ob]), lean BK^L1/L1 (BK^L1/L1;B6.V-Lep^+/+ [BK^{L1/ L1}; ^+/+]), and lean BK^+/+ (BK^+/+;B6.V-Lep^+/+ [BK^+/+;^+/+]) offspring from the same litters. To obtain tissue-specific mutants lacking BK in various adipocyte populations, heterozygous BK mice (BK^L1/+) on a C57BL6/N background, carrying a tamoxifen-inducible adipose tissue–specific Cre recombinase (genotype adiponectin-CreERT2^tg/+) under control of the adiponectin promoter (23), were first mated to homozygous floxed BK animals (genotype BK^L2/L2). Adipocyte-specific controls (genotype adiponectin-CreERT2^tg/+;BK^+/L2 [adipoqBK^+/L2]) and premutant BK animals (genotype adiponectin-CreERT2^tg/+;BK^L1/L2 [adipoqBK^L1/L2]) derived from this breeding were injected with tamoxifen (1 mg/day) for 5 consecutive days to induce Cre-mediated recombination at an age of 8 weeks (Figs. 3 and 4C–H) or 19 weeks (Fig. 4I).

The specificity and efficiency of the adiponectin-CreERT2-derived recombination was supported by studying double-fluorescent ROSA26-Tomato reporter mice (genotype ROSA26^{mTomato,mEGFP/+} [tom/+]) obtained from Charles River (24); these mice expressed the adipoqCreERT2 transgene (genotype adiponectin-CreERT2^tg/+; ROSA26^{mTomato,mEGFP/+}; [adipoq^tg/+;tom/+]). Double-transgenic animals were either injected with tamoxifen for 5 days (as described) to induce the recombination of the reporter allele—that is, a switch from cell membrane-localized red fluorescence to green fluorescence proteins—or received solvent. Genotyping was performed using specific primers to identify the Cre transgene (23), the single nucleotide mutation in the leptin gene (according to a protocol for stock no. 000632 from The Jackson Laboratory), and the L1, L2, and wild-type alleles of the BK on DNA samples obtained either from different fat depots, control tissue, or tail-tip biopsies from mice that were left untreated or received tamoxifen, as described previously (22).

Before administering the different research diets, all experimental mice received ad libitum tap water and a commercial chow obtained from Altromin (Lage, Germany). Dietary feeding trials were performed in male mice that were allowed to adapt to a defined control diet (CD) formulation containing 10% of calories from fat for 2 weeks before the long-term feeding experiments. Ten-week-old experimental mice received either a high-fat diet (HFD) containing 45% or 60% of calories from fat or continued feeding on the CD for another 18 weeks. Progressive BW gain in the monogenetic model of obesity was monitored between 4 and 24 weeks of age while eating normal chow diet.

The core body temperature and the locomotor activity were measured using a telemetry setup (ETA-F10 with two-lead electrocardiogram transmitters of the implant removed because they were not required to monitor the activity and temperature; Data Sciences, Inc.). The surgical procedure was described previously (25). In brief, anesthesia was induced by inhalation of 4% isoflurane in oxygen and was then maintained with 1–2% isoflurane. The implants were placed in the peritoneal cavity using an aseptic technique and the skin was closed with 5-0 surgical silk. During recovery, mice were placed on a warming pad before they were returned to their home cages for at least 3 days before recordings began. Activity and core body temperature were assessed continuously for 96 h in experimental mice that had received the HFD for 18 weeks.

Food intake was measured using a nonautomated and modified metabolic cage setup. To reduce environmental stress in the feeding chamber of the metabolic cage, mice housed individually were placed on paper bedding that was renewed daily. After 2 days of acclimatization, food consumption was assessed for 4 days using an analytic balance. Food pellets were replaced every 24 h. The amount of food consumed by each individual mouse was determined by calculating the weight differences between the initial weight of the pellets and the final weight 24 h later. Food pieces dropped by the animals on the paper bedding were thoroughly collected and included in the final weight determination. Food intake was assessed in a subset of 10-week-old experimental mice before the dietary feeding or after the CD or HFD feeding protocols.

Preadipocytes derived from inguinal (iWAT) or epididymal WAT (eWAT) of 10-week-old BK^+/+ and BK^L1/L1 mice were isolated using an established collagenase liberation protocol (26). Upon 50 min of digestion in collagenase type I/HEPES buffer at 37°C, 2 × 10⁴ or 5 × 10⁴ cells/cm² from iWAT or eWAT, respectively, were plated in DMEM/Ham F12 with 5% FCS. At 90–100% cell confluence, maturation was induced by adding induction medium containing insulin (170 nmol/L), dexamethasone (1 μmol/L), isobutylmethylxanthin (500 μmol/L), and indomethacin (30 μmol/L) to eWAT cultures. Differentiation of iWAT preadipocytes was stimulated by adding rosiglitazone (2.5 μmol/L) and triiodothyronine (1 nmol/L) to the eWAT induction medium. After 48 h of induction, the medium was exchanged with insulin (170 nmol/L)–containing medium in DMEM/Ham F12 with 10% FCS for another 48h, followed by maintenance medium (DMEM/Ham F12 with 10% FCS). Maintenance medium was changed every second day until cells were mature, usually 14 to 16 days after plating. Triglyceride incorporation was detected by Oil Red O (ORO) staining. ORO content was quantified by photometry using the Implen system at 518 nm. Digital images of the maturation process were acquired using AxioVision software version 4.8 (Carl Zeiss, Jena, Germany). Cell growth and proliferation assays were analyzed in real time using the xCELLigence impedance setup (Omni Life Science, Bremen, Germany) during the first 60 h upon plating in the medium (DMEM/Ham F12 with 5% FCS). Data were acquired and analyzed using RTCA software version 1.2.1 (ACEA Biosciences Inc.).

The conventional whole-cell mode of patch clamp electrophysiology was used to analyze BK currents in undifferentiated and differentiated 3T3-L1 cells, as well as from preadipocytes and induced matured adipocytes prepared from primary cultures of mouse eWAT. The pipette solution contained 140 mmol/L KCl, 10 mmol/L HEPES, 30 mmol/L glucose, 1 mmol/L 1,2-Bis(2-Aminophenoxy)ethane-N,N,N′,N′-tetra acetic acid (BAPTA), 2 mmol/L MgCl, and free calcium buffered to 0.5 μmol/L at pH 7.2. The standard bath (extracellular) solution contained: 140 mmol/L NaCl, 5 mmol/L KCl, 10 mmol/L HEPES, 20 mmol/L glucose, 2 mmol/L MgCl, and 1 mmol/L CaCl; pH was adjusted to 7.4. Electrophysiological recordings were performed at room temperature (18–22°C) using pCLAMP 9 (Molecular Devices), with a sampling rate of 10 kHz and filtering at 2 kHz. Patch pipettes were fabricated from borosilicate glass (Garner) with resistances between 2 and 3 mol/LΩ. Drugs were applied to cells using a gravity perfusion system with a flow rate of 1–2 mL/min to minimize flow-induced artifacts. For analysis of BK currents, the BK channel–specific blocker paxilline (1 μmol/L) was applied and the paxilline-sensitive outward current was analyzed at the membrane voltages indicated in the respective figure legends.

For histological examination, tissues were fixed in 4% paraformaldehyde for 1.5 h directly after harvesting and were then embedded in NEG 50 (Richard-Allan Scientific, Thermo Scientific) for sectioning following sucrose-gradient cryoprotection. Serial 14-µm cryosections of different WAT depots were prepared using an HM 560 cryostat (Thermo Scientific, Waltham, MA) and stored at −80°C until further processing. To determine the size of the fat cells before and after the administration of different diets, their perimeter was determined from digital images of the iWAT depot using the AxioVision software package version 4.8 (Carl Zeiss, Oberkochen, Germany). The uncoupling protein 1 (UCP1) expression pattern was compared between iWAT cryosections obtained from HFD-fed adipoqBK^L1/L2 and adipoqBK^+/L2 mice using a commercial antibody (dilution 1:400; Santa Cruz Biotechnology, Dallas, TX). BK immunodetection was performed in iWAT and eWAT sections derived from 10-week-old global and adipocyte-specific BK mutants and control litters using a primary antibody specific for the BK channel α-subunit (dilution 1:1,000) and hormone-sensitive lipase (dilution 1:400), according to a previously published protocol (27). In brief, antigen-antibody complexes were detected by a fluorescence-labeled secondary antibody (antirabbit AlexaFluor568; dilution 1:500) using the AxioCam MR system (Carl Zeiss) or by appropriate secondary antibodies conjugated to horseradish peroxidase (dilution 1:500).

To determine total fat mass and mass of different individual depots, fat pads were dissected unfixed upon dietary feeding then weighed, immediately immersed in liquid nitrogen, and finally stored at −80°C for further analysis. For analysis of total body composition, total BW was determined before and after the corpse was dried for 24 h at 85°C. Water content was calculated as the difference between dried corpse weight and total weight of the corpse at death. Fat content was determined in dried corpses using the Soxhlet procedure according to a previously published protocol (28).

Blood was collected from mice before or after dietary feeding. Leptin, interleukin (IL)-6, and adiponectin concentrations were measured in serum samples using mouse immunoassay kits (Merck Millipore, Darmstadt, Germany), according to manufacturer’s instructions. Blood glucose was determined using the GlucoCheck ADVANCE system (TREND Pharma GmbH, Saalfeld/Saale, Germany) immediately before an intraperitoneal glucose challenge (2 g/kg of body-weight) and 15, 30, 60, and 120 min after the injection in mice fasted overnight. At each time point, additional blood (25 µL) was collected via the tail vein for subsequent insulin determination. Plasma insulin was measured using the Ultrasensitive Mouse Insulin ELISA (Mercodia, Uppsala, Sweden), according to the manufacturer’s instructions.

Total mRNA was extracted from WAT samples by a guanidineisothiocyanat-phenol-chloroform extraction protocol using peqGold RNAPure (Peqlab Biotechnologie, Erlangen, Germany), according to the manufacturer’s instructions. Following DNase treatment for 30 min to remove traces of genomic DNA, RNA samples were quantified; 0.5 µg RNA was used to generate cDNA using an iScript cDNA synthesis kit (Bio-Rad, Hercules, CA). Quantitative real-time PCR was performed in triplicate using the comparative 2^−ΔΔC(t) method, with C(t) indicating the cycle number at which the signal of the PCR product crosses the threshold, set within the exponential phase of the detected fluorescence signal. BK and IL-6 levels were normalized to β-actin to determine their relative quantities. Respective primer sequences were designated: BK forward: 5′-ACGCCTCTTCATGGTCTTC-3′, BK reverse: 5′-TAGGAGCCCCCGTATTTCTT-3′; IL-6 forward: 5′-CTTCAACCAAGAGGTAAAAG-3′, IL-6 reverse: 5′-CCAGCTTATCTGTTAGGAGAG-3′. To extract total mRNA from 3T3-L1 cells or preadipocytes and differentiated adipocytes in culture, the Roche High Pure mRNA Isolation Kit was used according to the manufacturer’s instructions following shearing of cells in lysis buffer using a 25-gauge needle. Following DNase treatment as described above, 0.25 µg RNA was used to generate cDNA using the Roche Transcriptor High Fidelity cDNA Synthesis Kit with random hexamers and oligo-dT used at a 2:1 ratio. Quantitative RT-PCR was performed using the comparative 2^−ΔΔC(t) method, as above, with mRNA levels normalized to importin-8 (Ipo8). Respective primers were used: BK forward: 5′-GTCTCCAATGAAATGTACACAGAATATC-3′; BK reverse: 5′-CTATCATCAGGAGCTTAAGCTTCACA-3′; GLUT4 using the Quantitect Assay (Mm_Slc2a4_2_SG); and Ipo8 using Quantitect assay (Mm_Ipo8_2_SG) from Qiagen.

Undifferentiated (UD) 3T3-L1 fibroblasts were maintained in DMEM containing 10% FBS and 1% penicillin/streptomycin at 37°C at 10% CO₂; the fibroblasts were passaged every 4 days after reaching 60–80% confluence. Differentiation was initiated in confluent monolayers in DMEM/FBS containing 1 μg/mL insulin, 0.25 μmol/L dexamethasone, 1μmol/L troglitazone, and 500 μmol/L isobutylmethylxanthine for 3 days, followed by DMEM/FBS containing 1 μg/mL insulin and 1 μmol/L troglitazone. After 3 days the cell medium was replaced with DMEM/FBS, and the medium was refreshed every 3 days until differentiated (DF) cells were analyzed (typically 10–14 days after differentiation was initiated).

iWAT was dissected and processed as described above to generate total protein lysates for subsequent immunoblot analyses (27). Total protein was obtained from UD and DF 3T3-L1 fibroblasts. The proteins were separated by molecular weight via gel electrophoresis using 12% SDS gels. For immunodetection, various primary antibodies were used: UCP1 (dilution 1:1,000; Santa Cruz Biotechnology), GAPDH (dilution 1:1,000; Cell Signaling Technologies), GLUT4 (1:1,000; a gift from Gwyn Gould, Glasgow, U.K.), BK (1:1,000; NeuroMab, University of California, Davis, Davis, CA), and β-actin (1:500; Sigma-Aldrich, St. Louis, MO).

Statistical analysis was performed on experimental data using a two-tailed Student t test for paired or unpaired comparison or ANOVA with the post hoc Dunnett test for multiple comparisons, where appropriate. All data are presented as mean ± SEM. For all tests, P values less than 0.05 were considered as significant.

BW gain was investigated in global BK-deficient (BK^L1/L1) and age- and litter-matched male wild-type mice (BK^+/+) (Fig. 1A) receiving either a purified, low-fat CD or an HFD. Before the different diets were administered, we observed a small but highly significant (P < 0.001) difference in the BWs between the two genotypes, as previously reported (29,30). Starting with 5 weeks of HFD feeding, however, a substantial weight gain in male BK^+/+ mice became apparent, resulting in a total BW gain of 16.44 ± 0.83 g for BK^+/+ mice at the end of an 18-week HFD feeding protocol (Supplementary Fig. 1A), whereas BK^L1/L1 mutants gained 8.37 ± 0.45 g during the same observation period. Importantly, the differences in BW gain among BK^L1/L1 mutants were about ∼10% for the groups fed the CD (33.34 ± 2.38%) and the HFD (44.26 ± 2.68%), whereas the respective difference in the diet-induced BW gain for BK^+/+ mice was 27.76% (Supplementary Fig. 1B). Neither daily food intake of BK^+/+ and BK^L1/L1 mice under CD or HFD conditions (Supplementary Fig. 1C) nor the activity or body temperature of the animals under HFD conditions differed significantly (Supplementary Fig. 1D and E), suggesting that the lean phenotype of the BK^L1/L1 mice was, although not primarily, the result of abnormal hyperactivity or dysfunction in the central control of peripheral circuits regulating food intake and reward behaviors. Body composition revealed that the HFD did not affect wet body mass and carcass dry mass, but rather stimulated a substantial increase in the total body fat mass in BK^+/+ mice, whereas the respective body composition of CD- and HFD-fed BK^L1/L1 mice did not differ (Fig. 1B), supporting the notion that the genotype-specific differences in BW upon CD and HFD feeding (Fig. 1A) are largely the result of reduced accumulation of fat in the absence of BK. Less total body fat mass of CD- and HFD-fed BK^L1/L1 mice was accompanied by a small but significant shift toward lower dry mass values and reduced tibial length (TL) (Supplementary Fig. 1F), both of which confirm the lower overall body size in the global absence of BK channels (Fig. 1C). Analyses of fat mass normalized to TL revealed significant weight differences for various WAT depots as well as interscapular brown adipose tissue (BAT), indicating that the lean phenotype of the BK^L1/L1 mice was related to a decrease in multiple cell populations in adipocyte tissue (Fig. 1D). So far, the data suggest that high-caloric nutrition-induced BW gain requires functional BK channels. In addition, the reduced TL in the global BK knockouts is in accordance with previous reports, which collectively imply a mild growth defect that may contribute, at least in part, to the differences observed between BK^+/+ and BK^L1/L1 BW during dietary feeding (30). We next tested whether BK plays a role in the excessive fat accumulation in the monogenic ob/ob model of morbid obesity. BW gain (Supplementary Fig. 2A), body composition (Supplementary Fig. 2B and C), and individual weights of different fat depots (Supplementary Fig. 2D) were studied in BK-negative ob/ob double-mutants and age- and litter-matched BK^+/+ ob/ob controls. In line with the HFD-fed BK^+/+ and BK^L1/L1 mice, we found that key parameters of the progressively developing superobese phenotype, including BW, body composition, and fat depot masses, are attenuated in the absence of BK (Supplementary Fig. 2A–D). By contrast, differences in BW gain between lean control mice in the absence (Δ16.83 ± 0.55 g) and presence (Δ15.73 ± 0.55 g) of BK were not significant (P = 0.17). Protection against overwhelming BW gain in the BK-deficient ob/ob model was related to a consistent effect on total fat masses and nonfat components of the body (Supplementary Fig. 2B). Indeed, lower initial and final BWs of BK-negative lean and ob/ob double-mutant mice (Supplementary Fig. 2A), as well as differences in the mean TL (Supplementary Fig. 2E), suggest that BK plays a role in normal growth (Supplementary Fig. 2) and morbid obesity resulting from leptin deficiency.

Given the previous reports of the role of K⁺ channels in adipocytes (6,7) and our consistent observation of lower fat masses of multiple fat depots (Fig. 1D and Supplementary Fig. 2D), we considered that protection against overwhelming weight gain stemmed from the adipocyte itself. To address this possibility, we first assessed the BK mRNA and protein expression in iWAT and eWAT. Fat pads were studied using primer pairs, allowing for specific detection and quantification of BK mRNA (Fig. 1E and F). Compared with the internal reference, BK levels in iWAT and eWAT depots were low in BK^+/+ mice, but reliable amplification was accomplished in all samples tested. Importantly, signal traces detected in the respective fat depots derived from BK^L1/L1 mice were well within the range of nonspecific PCR product formation in the absence of reverse transcriptase (Supplementary Fig. 3A and B). Because adipose tissue is heterogeneous, comprising fat and nonfat cells, we next examined the cellular distribution of BK channel protein in frozen adipose tissue sections obtained from 10-week-old BK^+/+ and BK^L1/L1 mice. These analyses revealed evidence for BK in unilocular BK^+/+ cells, characteristic of white adipocytes, whereas BK^L1/L1 iWAT and eWAT remained BK-negative (Fig. 1G and H). Importantly, our specific BK antibodies and an antibody for the white fat cell marker hormone-sensitive lipase marked the same cells in BK^+/+ eWAT (Supplementary Fig. 3C), supporting the notion that mature adipocytes express BK channels.

To explore the functional attributes of endogenous BK channels in adipocytes, we adopted a previously established protocol of primary culture and differentiation of murine adipocyte precursor cells (26). Adipocyte differentiation was induced at 90–100% confluence of BK^+/+ and BK^L1/L1 precursor cells derived from either eWAT or iWAT depots. ORO incorporated into lipid droplets revealed strong accumulation of lipid-filled adipocyte-like cells under adipogenic maintenance conditions in cell cultures derived from both BK-negative and BK^+/+ eWAT (Fig. 2A–C) and iWAT (Fig. 2D–F). In iWAT-derived cultures we did not observe differences in adipogenic differentiation among genotypes (Fig. 2D–F), whereas significantly more ORO was incorporated in BK^L1/L1 than BK^+/+ cultures established from eWAT (Fig. 2A–C), suggesting a depot-specific role for BK in in vitro lipid storage. Moreover, genetic and pharmacological blockades of BK channels enhanced the growth of eWAT-derived preadipocytes (Fig. 2G and H).

To further validate differentiated adipocytes from eWAT express functional BK channels, we assayed BK mRNA, protein, and ionic currents in the differentiated cultures. BK channel, GLUT4 transporter, mRNA, and protein were strongly upregulated in differentiated cells compared with precursor cells (Fig. 2I and J). Importantly, outward potassium currents that are sensitive to the specific BK channel blocker paxilline (1 μmol/L) were significantly upregulated in differentiated adipocytes (Fig. 2K). Increases in BK channel mRNA, protein, and currents were also observed in the established 3T3-L1 preadipocyte cell model following the induction of differentiation (Supplementary Fig. 4A–C). In addition, the inhibition of BK channels with paxilline increased the growth of 3T3-L1 preadipocytes and the amount of ORO incorporated into lipid droplets (Supplementary Fig. 4D and E). Taken together, these data suggest that fat cell BK channels control preadipocyte expansion and adipogenic conversion of preadipocytes in an adipocyte depot–specific manner.

Because homeostasis of mature adipocytes and adipocyte differentiation in vivo are distinct from the above-studied in vitro models, we next studied the lean phenotype by generating an adipocyte-specific BK knockout mouse model (adipoqBK^L1/L2). In an initial series of experiments we confirmed the tissue-specific recombination efficacy of the recently established adiponectin promotor–driven, tamoxifen-inducible CreERT2 mouse model (23) using a two-color fluorescent Cre reporter system (24). Before Cre-mediated excision, we observed ubiquitous expression of the cell membrane–targeted red fluorescent protein Tomato (mT) in eWAT, iWAT, and all other tissues studied (Fig. 3A and B and data not shown). Tamoxifen-induced Cre-mediated excision of the floxed mT led to an almost complete switch to green fluorescent protein (mG) expression in eWAT and iWAT fat cells, whereas changes in mT labeling were not observed in various other organs such as the brain, skeletal muscle, and liver (data not shown). High efficiency and specificity of the adiponectin-CreERT2 approach was further confirmed by a BK-specific primer set designed to identify the three different BK alleles—wild-type (+), floxed (L2), and knockout (L1)—within one sample. In line with the Cre reporter assay (Fig. 3A and B), PCR products indicative of a recombination event were only observed in WATs and BATs of adiponectin-CreERT2 transgenic BK^+/L2 mice (adipoqBK^+/L2), whereas analysis of multiple other cell types and organ systems did not reveal significant Cre activity, regardless of whether tamoxifen was applied (Fig. 3C). Compared with adipoqBK^+/L2 controls, BK mRNA expression levels in eWAT and iWAT were reduced in tissue-specific adipoqBK^L1/L2 mutants upon tamoxifen application (Fig. 3D). Accordingly, adiponectin-CreERT2 activation by tamoxifen resulted in a nearly complete ablation of the BK protein in the different WAT tissues (Fig. 3E and F). Hence the adipocyte-specific knockout model of BK should allow us to test the role of fat cell BK channels for BW gain under CD and HFD feeding conditions in the absence of changes in satiety or adiposity signals arising from endocrine or neuronal circuits potentially involving hypothalamic BK channels, among others. To induce site-specific recombination of the L2 BK gene locus, adipoqBK^L1/L2 mice and adipoqBK^+/L2 control age-matched mice and littermates were subjected to the tamoxifen injection 2 weeks before the experimental diets were administered, that is, a CD or an HFD with 60% of its calories derived from fat (Fig. 4A and B). Consistent with our previous findings in the global BK^L1/L1 model (Supplementary Fig. 1C), the tissue-specific ablation of BK did not affect the amount of food (CD or HFD) consumed by the animals at the start or end of the feeding experiment (Fig. 4C and D). Starting BWs and BW gain in the CD-fed group were not different between adipoqBK^+/L2 and adipoqBK^L1/L2 mice (Fig. 4E and G), whereas the lack of adipocyte BK channels afforded partial protection against HFD-induced BW gain (Fig. 4F and H), an effect that reached a significant level after 4 weeks of HFD feeding and was maintained until the mice reached their final BWs (Fig. 4F), indicative of a lower susceptibility of adipoqBK^L1/L2 mutant mice to high-calorie challenges.

We next tested whether excess BW gain is also affected by fat cell BK in mice that have already gained significant weight. To test this, tamoxifen treatment was commenced at an age of 19 weeks, when the adipoqBK^+/L2 control and adipoqBK^L1/L2 pre mutants had reached a BW of 32.35 ± 1.05 and 31.91 ± 0.89 g, respectively. Total BW gain was not different between age- and litter-matched adipoqBK^+/L2 and adipoqBK^L1/L2 mice before the tamoxifen treatment (weeks_10–19 8.18 ± 1.00 g for adipoqBK^+/L2; weeks_10–19 7.71 ± 0.76 g for adipoqBK^L1/L2 [P = 0.7]), whereas upon five repetitive tamoxifen injections, the extent of the weight gain showed a clear tendency toward lower values in adipoqBK^L1/L2 mice (Fig. 4I) (weeks_20–30 5.33 ± 0.56 g for adipoqBK^L1/L2; weeks_20–30 7.40 ± 1.03 g for adipoqBK^+/L2 [P < 0.06]).

In-depth analyses of the individual fat depots (Fig. 5A and G) and the total fat mass (Fig. 5B) did not reveal statistical weight differences between adipoqBK^+/L2 and adipoqBK^L1/L2 mice fed a CD diet. HFD feeding, however, confirmed protection against excessive fat storage in adipoqBK^L1/L2 mice. Interestingly, the lack of adipocyte BK was associated with smaller total fat masses and smaller masses of various fat depots (Fig. 5C–G).

Because fat mass is determined by both adipocyte number and size, we next studied fat storage at the cellular level. iWAT cell size was calculated by assessing the cell perimeters in different cryosectional areas obtained from adipoqBK^+/L2 and adipoqBK^L1/L2 mice. Before the different diets were administered (i.e., at an age of 10 weeks) genotype-specific differences in fat cell size were not detectable in different areas of the iWAT sections (Fig. 6A and C). Compared with adipoqBK^+/L2 fat cells, BK-deficient adipocytes were smaller upon CD and HFD feeding (Fig. 6A and C).

Because enlarged, hypertrophic fat cells in the adipose depot relate to obesity, inflammation, and insulin resistance, we next tested whether the observed changes in adipocyte morphology (Fig. 6A–C) were associated with a more proinflammatory state. Indeed, we found reduced IL-6 mRNA expression in iWAT of HFD-fed adipoqBK^L1/L2 mice and tendentially lower serum IL-6 concentrations (P = 0.11) (Fig. 6E). Moreover, HFD-fed adipoqBK^L1/L2 mice exhibited lower serum leptin concentrations as a marker of body mass (Fig. 6F), whereas adiponectin concentrations were not different between diets or genotypes (Fig. 6G). In accordance with the markers of inflammation, small adipose tissue mass, and low fat cell hypertrophy, adipoqBK^L1/L2 mice showed improved glucose clearance (Fig. 6H) with lower insulin concentrations (Fig. 6J) upon intraperitoneal glucose challenge after HFD feeding, whereas a lack of adipocyte BK did not affect glucose handling before the feeding protocol (Fig. 6I). Change in the cellularity of the iWAT may also be considered an indicator for accelerated browning to promote energy expenditure, which counters obesity and its metabolic consequences. Lack of fat cell BK resulted in strong induction of UCP1 (Fig. 7A and C) as a hallmark of uncoupled respiration and heat production by brown or brown adipocyte–like fat cells. Higher levels of UCP1 protein under HFD conditions were also reflected by a significant increase in core body temperature during the night (Fig. 7D), whereas during the day, while the mice were sleeping, body temperatures of adipoqBK^L1/L2 and adipoqBK^+/L2 mice were lower, suggesting a reduced burning of stored fats, with no apparent differences between the two genotypes (Fig. 7D).

In summary, adipocyte BKs promote fat cell size and fat pad mass in vivo. Adipocyte tissue growth in the presence of endogenous BK channels was related to a noninfectious activation of adipose tissue inflammation and metabolically unfavorable effects on fat cell functions, which together may result in insulin resistance and amplified HFD-induced adiposity.

Using combined analysis of different BK channel mutant mouse lines we uncovered a novel function for adipocyte BKs in fat cell biology and metabolism under different nutritional conditions. We found resistance to HFD-induced BW gain, a smaller total fat mass, and thereby improved glucose handling upon ablation of endogenous BK channels in various adipose depots in different parts of the body. These data imply that the development of obesity caused by nutrient excess is promoted by the BK channel (Fig. 4). Accordingly, a previously identified BK gene variant was associated with elevated levels of fat cell BK mRNA in morbidly obese human subjects, suggesting a causal relationship between the amount of adipocyte BKs and weight gain (6). To the best of our knowledge, BK levels in other organ systems of the affected patients have not been assessed; therefore, it remains unclear whether the obesogenic effects attributed to amplified BK expression resulted exclusively from an effect on adipocyte cell function or whether BK channels present in nonadipocyte cells contributed to morbid weight gain. Interestingly, our HFD-fed BK^L1/L1 mutants did not exhibit any changes in dietary food consumption or body temperature (Supplementary Fig. 1C and D), whereas the respective fat cell–specific mutants exhibited a lean phenotype that was related to an increase in UCP1 level and energy expenditure (Fig. 7). It is therefore tempting to speculate that BK channels in the brain or other nonfat cells do not play a role in the chemical and neural signals regulating calorie intake and/or energy expenditure and thereby body composition. For several reasons, however, we find it is too early to draw such conclusions.

1. Our previous analyses of the global BK^L1/L1 mouse model revealed multiple defects in various systems of these animals, including in β-cells (15), the HPA axis (14), and cerebellar neurons (29), which separately or together could affect energy balance and BW development. For example, a lack of cerebellar BK channels in Purkinje cells, among other deficits, was shown to cause motor learning impairment and signs of ataxia, with the latter being related to muscle shivering and trembling (29), a common cause of increased energy expenditure. Irrespective of the dietary fat content, we did not observe an increase in the core body temperatures of BK^L1/L1 mice (Supplementary Fig. 1D), suggesting that their very low body fat (Fig. 1) allows them to maintain their body temperature at a physiological level. Potential changes in body temperature regulation (i.e., adjustments due to changes in [non]shivering thermogenesis) may be superimposed by the loss of heat in BK^L1/L1 mutants. Yet BK^L1/L1 mice present with a lean phenotype with CD and HFD feeding regimes, and we observed a lower propensity of the BK/ob double-mutants to accumulate excessive BW and fat mass. This suggests that the global lack of BK channels for the entire life span prevents excessive BW gain, an effect that was more obvious in the presence of genetic or nutritional risk factors for developing obesity. In this complex mouse model, the dysregulation of multiple pathways (e.g., in fat cells, hypothalamus, liver, sympathetic neurons, β-cells) involving BK may be related to this phenotype.

2. The protection in terms of BW gain and fat accumulation was less pronounced in adipocyte-specific BK knockout groups compared with the BK^L1/L1 or BK/ob double-mutant models (compare Fig. 1A and Supplementary Fig. 2A and Fig. 4F), implying that nonfat cell BKs are also important for weight control in vivo. Indeed, dietary-related changes in energy expenditure involving, for example, hypothalamic control mechanisms may be positively or negatively regulated by BK channels of neuronal nuclei that respond to satiety or hunger signals. Lack of GIRK4, a G protein–gated, inwardly rectifying K⁺-channel, in mice, for instance, resulted in late-onset obesity through hypothalamic mechanisms (31), whereas ATP-sensitive K⁺ channel (K_ATP) knockout mice showed hyperphagia but were resistant to HFD-induced BW and visceral fat mass gains (32). Along these lines, K_ATP conductance in diet-induced obesity has been reported in pro-opiomelanocortin-positive nuclei of the hypothalamus, and neuronal excitability and thereby the release of peptides that control food intake and BW were sensitive to central K_ATP inhibition (33). In addition to K_ATP, BK channels have been shown to modulate the excitability of hypothalamic neurons in response to insulin and leptin under physiological conditions (34), suggesting they may be involved in central circuits regulating BW via energy intake and expenditure. Given the complexity of the different metabolic pathways, future studies are needed to clarify the functional roles of hypothalamic and other nonfat cell BK channels in adiposity and thereby the mechanism(s) underlying the lean phenotype of BK^L1/L1 mice. Resistance to genetic and diet-induced models of obesity (i.e., hypothalamic-driven obesity) has also been reported in mice with a Shaker family voltage-dependent K⁺ channel gene disruption in Kv1.3 (35,36), further highlighting the importance of K⁺ channels for the hypothalamic-regulation of BW (37). More recently, however, the therapeutic benefits on obesity and insulin resistance upon Kv1.3 inhibition using a blocker that was unable to cross the blood-brain barrier have been attributed to peripheral mechanisms stemming from changes in liver and WAT metabolism and BAT (38). Obviously, peripheral and central K⁺ channels affect the development of adiposity.

Studying adipocyte-specific BK knockout mice, we did not find any evidence for tamoxifen-induced recombination in different brain regions, including the hypothalamus and other tissues involved in total body metabolism, such as muscle and liver (Fig. 3), suggesting that partial protection against diet-induced obesity (Fig. 4F) results from the lack of BK channels in adipocytes (Fig. 3). Further, we observed amplified BK expression and currents with adipogenic differentiation of mouse adipocytes from both eWAT fat cell precursors and the preadipocyte 3T3-L1 cell line (Fig. 2K and Supplementary Fig. 4C). Pharmacological blockade of BK channels stimulated both growth and lipid incorporation, supporting a role for BK channels as regulators of cell cycle progression in human preadipocytes (7). By assessing BK-deficient eWAT- and iWAT-derived preadipocytes in vitro, however, we recognized fat depot–specific functions for endogenous BK channels potentially involved in the control of cell growth and lipid accumulation (Fig. 2). Because adiposity is usually induced through hypertrophic expansion of existing adipocytes, leading to dysfunctional cells, an increase in the number of fat cells (i.e., hyperplasia) is proposed as a mechanism that preserves metabolic fitness (39). Accordingly, lower levels of the proinflammatory cytokine IL-6 (Fig. 6D and E), significantly lower BW (Fig. 4F) and smaller iWAT cell size together support the notion that a lack of BK leads to the production of more “healthy” adipocytes that may afford protection against excessive BW gain, thereby maintaining a regular response to the action of insulin. Adipocytes reportedly produce and release IL-6 (40). In obesity the fraction of adipose tissue–derived IL-6 may promote a chronic state of low-grade inflammation, affecting metabolism and the development of insulin resistance (41). In comparison with adipoqBK^L1/L2 mutant mice, we found higher IL-6 adipose tissue mRNA (P < 0.05) and nonsignificantly elevated serum concentrations (P = 0.11) in HFD-fed adipoqBK^+/L2 control mice; hence, it seems that fat cell BK channel activity is positively related to the adipose tissue IL-6 pathway, providing a link between BK and obesity-related comorbidities such insulin resistance (Fig. 6D, E, and H). Interestingly, high plasma and adipose tissue concentrations of IL-6 were previously reported for the ob/ob mouse model (42), suggesting that BK channels may interfere with both the genetic and dietary causes of adiposity through an IL-6-dependent mode of action. Along the same lines, HFD-induced overnutrition among adipoqBK^L1/L2 mice revealed smaller total fat mass and smaller fat cells in iWAT depots. Of note, iWAT cells and depots differed in size and mass only in vivo (Fig. 6), whereas in (pre)adipocyte cultures originating from iWAT pads, we did not detect differences in terms of various adipogenic parameters (Fig. 2). The reason for this apparent discrepancy is unclear, although several mechanisms could be involved. First, the adipocyte-specific lack of BK channels in vivo results in a decrease in the iWAT depot, but not the eWAT depot, fat mass. A “normal” mass of the adipoqBK^L1/L2 eWAT depot in vivo (Fig. 5A and C) may result from direct effects of BK on the lipid storage and cell growth capacities; both parameters were augmented in the respective BK^L1/L1-derived fat cells in vitro (Fig. 2A–C, G, and H). By contrast, a smaller eWAT depot mass in the global BK knockouts (Fig. 1D) may result from dysregulation of BK-dependent mechanisms in nonadipocytes. Interestingly, our observations in HFD-fed adipoqBK^L1/L2 and adipoqBK^+/L2 mice imply a small but significant increase (approximately 0.3°C) in the body’s core temperature in the absence of fat cell BKs (Fig. 7D), which indicates that these channels are involved in the thermogenic program of adipose cells. The presence of UCP1 in iWAT is characteristic of a process known as “browning,” whereby this depot acquires brown adipocyte–like catabolic functions (43); we found the adipocyte-specific ablation of BK channels to result in elevated mRNA (data not shown) and protein levels of iWAT UCP1 (Fig. 7A–C). The burning of stored fats by expansion or activation of BAT or by browning of WAT may result in a metabolically favorable phenotype (44). Accordingly, we observed a large fraction of smaller fat cells in iWAT, smaller iWAT and total fat depot weights, as well as an improved glucose homeostasis and insulin sensitivity in HFD-fed adipoqBK^L1/L2 mice (Fig. 7A–C and H–J).

Clearly, the mechanistic details underlying the lean phenotype of HFD-fed adipoqBK^L1/L2 mutants require future investigation. In general, it is widely accepted that BK channels are activated either by an increase in intracellular Ca²⁺ ([Ca²⁺]_i) or by membrane depolarization. [Ca²⁺]_i dynamics have a multifaceted role in brown adipocyte–like functions and in the formation of mature fat-laden adipocytes (45–52); however, relatively little is known about Ca²⁺ signaling in white and brown adipocytes despite its suggested importance (53). Yet the efflux of K⁺ via BK in nonexcitable cells such as adipocytes may serve as a positive regulator for Ca²⁺ entry as well as a number of Ca²⁺-dependent processes, whereas in excitable cells BK is usually part of a negative feedback loop limiting the influx of Ca²⁺ via voltage-dependent Ca²⁺ channels (54). Transient receptor potential vanilloid 4 (TRPV4) channels have recently been identified as a major feature of the influx of extracellular Ca²⁺ into fat cells (55). TRPV4 seems to negatively regulate the thermogenic and proinflammatory programs in adipocytes, and mice with a TRPV4 ablation were protected from diet-induced obesity; hence TRPV4 deficiency results in a phenotype very similar to that caused by adipocyte-specific BK ablation. A functional coupling of BK and TRPV4 was reported in different systems (56,57). Although this study did not examine this link, we hypothesize that BK may suppress the acquisition of brown adipocyte–like features in white fat cells by an effect on Ca²⁺ entry and subsequent signaling pathways involving TRPV4 or other Ca²⁺-permeable members of the large TRP channel protein family (50,51,58).

Finally, BK channels are common targets of various intracellular signaling molecules including cGMP and cAMP (59,60). It is well established that both cAMP and cGMP pathways play differential roles in lipolysis, browning of WAT, and brown fat cell thermogenesis (44,61), among other processes, which should allow for complex adjustments of BK function in the different fat depots in vivo under (patho)physiological conditions.

Together, the data presented here show that loss of fat cell BK channels supports the browning of fat cells to enhance thermogenesis in response to nutritional excess, thereby limiting excessive weight gain. Targeting fat cells with BK channel inhibitors may thus have the potential to reduce the pathological features of excessive weight gain and related disorders.

Acknowledgments. The authors thank Clement Kabagema-Bilan, Isolde Breuning, Michael Glaser, and Katrin Junger (all from the Department of Pharmacology, Toxicology and Clinical Pharmacy, Institute of Pharmacy, University of Tübingen, Tübingen, Germany) for excellent technical help.

Funding. This work was funded in part by the Deutsche Forschungsgemeinschaft (to P.R. and R.L.), the Wellcome Trust (to P.R. and M.J.S.), and Diabetes UK (to L.H.C. and M.J.S.).

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

Author Contributions. J.I., L.T., H.M., M.W., and M.J.S. conducted experiments. J.I., M.J.S., and R.L. analyzed data and wrote the manuscript. L.H.C., V.L., S.O., and P.R. contributed to discussions and edited the manuscript. V.L., A.S., and S.O. contributed new tools. P.R., M.J.S., and R.L. designed the research. All authors approved the final manuscript. M.J.S. and R.L. are the guarantors of this work and, as such, had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.