Adipocytes control food intake and weight regain via Vacuolar-type H+ ATPase

Abstract

Energy metabolism becomes dysregulated in individuals with obesity and many of these changes persist after weight loss and likely play a role in weight regain. In these studies, we use a mouse model of diet-induced obesity and weight loss to study the transcriptional memory of obesity. We found that the 'metabolic memory' of obesity is predominantly localized in adipocytes. Utilizing a C. elegans-based food intake assay, we identify 'metabolic memory' genes that play a role in food intake regulation. We show that expression of ATP6v0a1, a subunit of V-ATPase, is significantly induced in both obese mouse and human adipocytes that persists after weight loss. C. elegans mutants deficient in Atp6v0A1/unc32 eat less than WT controls. Adipocyte-specific Atp6v0a1 knockout mice have reduced food intake and gain less weight in response to HFD. Pharmacological disruption of V-ATPase assembly leads to decreased food intake and less weight re-gain. In summary, using a series of genetic tools from invertebrates to vertebrates, we identify ATP6v0a1 as a regulator of peripheral metabolic memory, providing a potential target for regulation of food intake, weight loss maintenance and the treatment of obesity.

Fig. 1. Identification of metabolic memory genes in adipose tissue in formerly obese mice.

a Schematic diagram of diet switch mouse study. b RNA seq of adipocytes, c stromal vascular cells (SVCs), from lean(LF, black circles), obese (HF, red squares) and formerly obese (SW, blue triangles) mice. Gene expression was analyzed using the statistical algorithm limma and differential expression between LF and HF fed mice determined as lfdr < 0.05, n = 3 per group. d C. elegans based food intake assay used to determine if genetic strains lacking orthologous mouse metabolic memory genes play a role in food intake. Strains with significantly different food intake from wild type “N2” strain was determined by one-way ANOVA with Dunnetts multiple comparison test. Values are expressed as box and whisker plots, with boxes indicating the Q1 and Q3 ranges, center line representing the mean, and the whiskers as minimum and maximum values, *p < 0.05, n = 17–23 wells per strain. e unc-32/ATP6v0a1 strain responds to hyperphagic stimulus serotonin as determined by unpaired student t-test (2 tailed), (average n = ∼250 worms per strain, p < 0.0001). Values are expressed box and whisker plots, with boxes indicating the Q1 and Q3 ranges, center line representing the mean, and the whiskers as minimum and maximum values, *p < 0.05, n = 20–21 wells per strain. Statistical parameters can be found in the Supplementary Data 1–5.

Fig. 2. V-ATPase expression in obesity.

a Expression of Atp6v0a1 in lean (LF, black bars), obese (HF, red bars) and formerly obese (SW, blue bars) mice determined by RNA seq, n = 3/group, Adipocyte p = 0.005. b Expression of V-ATPase subunits in lean, obese and formerly obese mice determined by RNA seq, *denotes differential expression compared to LF fed mice determined as lfdr < 0.05, n = 3/group, (p < 0.004 for all LF vs HF comparisons). c Expression of Atp6v0a1 in WT (open bars) and ob/ob mice (hatched bars) fed with normal chow (black/white) and high-fat diet (HFD, red) determined by RNA seq (GEO GSE167264,), n = 3/group, p = 0.0002. d Adipocyte gene expression of V-ATPase subunits in lean (black bars) and humans with obesity (red bars), (n = 9–10), from Gene Expression Omnibus (GEO: GSE2508) and differential expression between groups determined using GEO2R (default parameters including Benjamini and Hochberg calculation of the false discovery rate, fdr)* = fdr < 0.4 (V0A1 fdr = 0.36, V1B2 = 0.18, V0D1 = 0.35, v0E1 = 0.16, V1F = 0.19). e AT6V0a1 expression in subcutaneous adipose tissue from lean participants (n = 4) and individuals with obesity (n = 5) (BMI: t = 6.915, df = 7, CI = 6.942–14.16, p = 0.0002; Atp6V0a1 protein expression: t = 2.074, df = 7, CI = −0.009–0.15, p = 0.07) was analyzed using 2-tailed student’s t-test. f Adipose gene expression of ATP6v0a1 in participants with obesity before (red) and after 5%, 10%, and 15% weight loss (shades of blue). Microarray expression data was downloaded from GEO GSE70529, (n = 9/group). g, h Western blot and densinometric quantification of membrane-bound and cytoplasmic Atp6V1a (V1a) and Atp6V0a1 (V0a1) subunits in gonadal WAT from lean (LF), obese (HF), and formerly obese (SW) mice (n = 3/group; MemV1: F = 2.319, df = 8, p = 0.09; MemV0: F = 4.786, df = 8, p = 0.03; CytoV1: F = 4.860, df = 8, p = 0.03; CytoV0: F = 6.324, df = 8, p = 0.02). Densinometric quantification is expressed as mean ± SEM and was analyzed by one-way ANOVA followed by Two-stage linear step-up procedure of Benjamini, Krieger and Yekutieli. * denotes statistical significance at p < 0.05.

Fig. 3. Adipocyte-specific Atp6v0a1 depletion in vitro and in vivo.

Atp6v0a1 mRNA quantification in WAT and BAT (f = 7.221, df = 3, p = 0.0013) adipose tissue in KO (white bars) and WT controls (red bars), n = 7 per group. b Western blots and densinometric quantification of Atp6V0a1 in adipocytes (t = 16.86, df = 4, CI = −0.244 to −0.175, p < 0.001) and SVCs (t = 0.41, df = 6, CI = −0.028 to 0.02, p = 0.34) from gonadal WAT from WT (adiponectin Cre) and KO mice (n = 3–4/group). c Food intake (F = 3.03, df = 28, p = 0.03), cre, n = 7, FF n = 12 KO, n = 10. d Body weight (F = 9.86, df = 24, p < 0.001), of adiponectin Cre (Cre, n = 7) (red bars), Atp6v0a1^fl/fl (F/F, n = 9) (gray bars) and KO (n = 9, white bars) mice, e weight of WAT (t = 2.416, df = 30, p = 0.02) and liver (t = 0.88 df = 30, p = 0.37) as % of total body weight of WT and KO mice (n = 7–12/group). All mice were fed normal chow. a, c–e were analyzed by one-way ANOVA followed by two-stage linear step-up procedure of Benjamini, Krieger and Yekutieli with false discovery rate of 0.10. b was analyzed by students t-test (2 tailed). Data is expressed as mean ± SEM, *denotes statistical significance at p < 0.05.

Fig. 4. Adipocyte-specific Atp6v0a1 KO mice fed HFD gain less weight due to reduced food intake.

a Average daily food intake (WT n = 20, KO n = 24, FF n = 12, F = 7.03, df = 55, p = 0.002), b weight gain (WT, n = 13, KO n = 16, FF n = 12, F = 5.06, df = 266, p = 0.01) in WT (red), KO (white) and FF (gray) mice. c Body composition (n = 5/group, lean mass: t = 0.11, df = 7, CI = −1.70–1.536, p = 0.91; fat mass: t = 2.07, df = 7, CI = −6.06 to 0.409, p = 0.04), d adipocyte size (t = 0.21, df = 3665, CI = −1055 to −8791, p < 0.0001), n = 5 per group, e energy expenditure (genotype: Wald Chi-square = 298, df = 1, p < 0.001; BW: Wald Chi-square = 7.119, df = 1, p = 0.08; time: Wald Chi-square = 1943, df = 372, p < 0.001), (n = 5/group). f Respiratory quotient (t = 0.345, df = 8, CI = −0.02–0.015, p = 0.37), (n = 5/group). g Spontaneous activity as determined by beam breaks averaged over 3 days (t = 0.08, df = 8, CI = −22429–20849, p = 0.47), in WT and KO fed 60% fat diet for 8 weeks. Quantitative PCR measurement of gene expression within h hypothalamus (n = 7/group; Agrp: t = 0.44, df = 12, p = 0.33; Npy: t = 2.28, df = 12, p = 0.02; Pomc: t = 7.29, df = 12, p < 0.001; Cart: t = 1.30, df = 12, p = 0.11) and i gonadal WAT (n = 5/group; Fasn: t = 1.81, df = 8, p = 0.05; Cpt1a: t = 0.31, df = 8, p = 0.38; Adipoq: t = 3.55, df = 8, p = 0.004; Lep: t = 2.34, df = 8, p = 0.03; Atf4: t = 2.99, df = 6, p = 0.01; Ddit3: t = 2.01, df = 6, p = 0.05). j Gdf15 transcript levels in WAT and plasma protein levels (WT n = 5, KO n = 6 group; qPCR: t = 2.51, df = 9, p = 0.02; plasma: t = 2.27, df = 9, p = 0.03). k Western blot and densinometric quantification of membrane-bound and cytoplasmic Atp6v1a (V1a) and Atp6v0a1 (V0a1) subunits in gonadal WAT from WT and KO mice fed with 60% fat diet for 8 weeks (n = 4; MemV1: t = 1.35, df = 6, p = 0.23; MemV0: t = 3.21, df = 6, p = 0.02; CytoV1: t = 1.03, df = 6, p = 0.34; CytoV0: t = 1.80, df = 8, p = 0.12). a, b were analyzed by one-way ANOVA followed by two-stage linear step-up procedure of Benjamini, Krieger and Yekutieli with false discovery rate of 0.10. c, d, f–k were analyzed by students t-test (2 tailed). e was analyzed using ANCOVA with body weight as covariate using SPSS v22. Data is expressed as mean ± SEM, * denotes statistical significance at p < 0.05.

Fig. 5. ATP6v0a1 depletion in adipocytes results in improved glucose homeostasis.

a Glucose tolerance test (GTT; genotype: F = 6.15, df = 1, p = 0.02; time: F = 63.69, df = 4, p < 0.001; genotype X time: F = 1.40, df = 4, p = 0.24). b Body weight at time of GTT (t = 2.52, df = 19, p = 0.01). c Glucose-stimulated plasma insulin (genotype: F = 1.14, df = 1, p = 0.30; time: F = 11.37, df = 1, p = 0.003; genotype X time: F = 0.06, df = 1, p = 0.80). d Insulin tolerance test (ITT), presented as change from baseline (genotype: F = 6.76, df = 1, p = 0.01; time: F = 59.95, df = 5, p < 0.001; genotype X time: F = 2.18, df = 5, p = 0.06). e Body weight at time of ITT, (n = 9-12/group; t = 3.78, df = 17, p < 0.001). Immunoblots of total AKT and phosphorylated AKT at Ser473 in WAT, liver and skeletal muscle in f the basal state g insulin stimulated state, of WT and KO mice fed with 60% fat diet for 8 weeks (n = 4/group). h Densinometric quantification of basal blots (WAT: t = 1.35, df = 8, p = 0.11; Liver: t = 0.53, df = 8, p = 0.30; Skeletal muscle: t = 0.81, df = 8, p = 0.22). i Densinometric quantification of insulin stimulated blots (WAT: t = 2.37, df = 8, p = 0.04; Liver: t = 1.55, df = 8, p = 0.16; Skeletal muscle: t = 2.15, df = 8, p = 0.06). WT (red circles, red bars), KO (white squares, open bars). Data is expressed as mean ± SEM. a, c, d were analyzed by repeated measures two-way ANOVA followed by two-stage linear step-up procedure of Benjamini, Krieger and Yekutieli with false discovery rate of 0.10. b, e, h, i were analyzed by student t-test (2 tailed). * denotes statistical significance at p < 0.05.

Fig. 6. Inhibition of V-ATPase abrogates HFD-induced weight gain which is dependent on the adipocyte expression of Atp6v0a1.

a Average daily food intake (F = 10.21, df = 3, p < 0.001), (WT-VEH n = 7, WT-BFM n = 7, KO- VEH n = 6, KO-BFM n = 6). b weight gain in WT (solid bars) and KO (open bars) mice after 7 days treated with BFM (black) or VEH (red) and fed HFD (F = 3.27, df = 3, p = 0.03), (WT-VEH n = 7, WT-BFM n = 7, KO-VEH n = 6, KO-BFM n = 6). c, d GTT after 3 days of treatment (group: F = 1.19, df = 3, p = 0.33; time: F = 206, df = 4, p < 0.001; genotype X time: F = 4.48, df = 12, p < 0.001), (WT-VEH n = 8, WT-BFM n = 8, KO-VEH n = 7, KO-BFM n = 8), e, f Glucose stimulated insulin secretion, (group: F = 0.35, df = 3, p = 0.78; time: F = 38.62, df = 1, p < 0.001; genotype X time: F = 1.46, df = 3, p = 0.25), (WT-VEH n = 8, WT-BFM n = 8, KO-VEH n = 7, KO-BFM n = 8). g, h ITT after 7 days treatment with BFM (WT-VEH n = 8, WT-BFM n = 7, KO-VEH n = 8, KO-BFM n = 7: F = 8.02, df = 3, p = 0.006; time: F = 74.23, df = 5, p < 0.001; genotype X time: F = 4.25, df = 15, p < 0.001). Relative gene expression in i. hypothalamus (n = 6/group; Agrp: t = 2.506, df = 10, p = 0.03; Npy: t = 0.17, df = 10, p = 0.86; Pomc: t = 4.055, df = 10, p = 0.002; Cart: t = 0.45, df = 10, p = 0.66) and j WAT (n = 7/group; Adipoq: t = 1.71, df = 12, p = 0.11; Lep: t = 1.39, df = 12, p = 0.19; Acaca: t = 0.41, df = 12, p = 0.69; Gdf15: t = 4.33, df = 12, p = 0.001) in vehicle or BFM-treated WT mice after 2 weeks of treatment. Data is expressed as mean ± SEM. a, b were analyzed by one-way ANOVA followed by two-stage linear step-up procedure of Benjamini, Krieger and Yekutieli with false discovery rate of 0.10. c–h were analyzed by repeated-measures two-way ANOVA followed by two-stage linear step-up procedure of Benjamini, Krieger and Yekutieli with false discovery rate of 0.10. i, j were analyzed by student t-test (2 tailed), * denotes statistical significance at p < 0.05.

Fig. 7. Inhibition of V-ATPase abrogates HFD-induced weight re-gain in formerly obese mice.

a Schematic diagram of weight rebound mouse experiment. b Average daily food intake (t = 4.31, df = 26, p < 0.001) and c HFD-induced weight re-gain (n = 14/group; group: F = 18.29, df = 1, p = 0.002; time: F = 119, df = 10, p < 0.001; genotype X time: F = 13.78, df = 10, p < 0.001). d WAT as percentage of body weight (n = 8; t = 2.74, df = 14, p = 0.01). e Relative gene expression in hypothalamus as determined by qPCR (Agrp: t = 0.18, df = 12, p = 0.43; Npy: t = 1.53, df = 12, p = 0.15; Pomc: t = 2.27, df = 12, p = 0.04; Cart: t = 0.04, df = 12, p = 0.96). f Gdf15 gene levels in WAT (n = 7/group; t = 4.46, df = 12, p < 0.001) and plasma concentration (n = 6/group; t = 2.32, df = 10, p = 0.04). g Relative gene expression in WAT (Fasn: t = 3.93, df = 13, p = 0.001; Acaca: t = 2.58, df = 12, p = 0.02; Cpt1a: t = 2.73, df = 10, p = 0.02; Lep: t = 6.74, df = 13, p < 0.001; Adipoq: t = 0.21, df = 12, p = 0.84; Atf4: t = 2.86, df = 12, p = 0.02; Ddit3: t = 2.11, df = 13, p = 0.05). h Western blot and densinometric quantification of membrane-bound and cytoplasmic Atp6V1a and Atp6V0a1 in gonadal WAT (n = 3/group; MemV1: t = 19.18, df = 4, p < 0.001; MemV0: t = 2.22, df = 4, p = 0.04; CytoV1: t = 0.76, df = 4, p = 0.25; CytoV0: t = 0.43, df = 4, p = 0.34) of WT mice treated with either Veh (Blue) or BFM (white) during weight rebound (Rb). Data is expressed as mean ± SEM and * denotes statistical significance at p < 0.05. b, d–h were analyzed by student t-test (2 tailed), c was analyzed by repeated-measures two-way ANOVA followed by two-stage linear step-up procedure of Benjamini, Krieger and Yekutieli with false discovery rate of 0.10.

Fig. 8. Adipocyte-specific knockdown of Atp6v0a1 increases Gdf-15 expression in 3T3L1 adipocytes.

a Relative Atp6v0a1 gene expression in 3T3-derived adipocytes treated with SiRNA targeting Atp6v0a1 (blue) or control siRNA (red) (control n = 9, SiRNA n = 8/group; t = 5.10, df = 15, p < 0.001). b Western blots and densinometric quantification of ATP6v0a1 (control, n = 8, SiRNA n = 6/group; t = 2.78, df = 12, p = 0.008). c Relative gene expression of leptin, adiponectin and Gdf-15 (Lep: t = 0.67, df = 12, p = 0.51; Adipoq: t = 0.76, df = 12, p = 0.46; Gdf15: t = 2.86, df = 12, p = 0.01), control, n = 6, SiRNA n = 9/group. d Western blots and densinometric quantification of Gdf15 (n = 6/group; t = 2.15, df = 10, p = 0.03). e Relative gene expression of Atf4 and Ddit3 (n = 6/group; Atf4: t = 2.15, df = 19, p = 0.02; Ddit3: t = 1.94, df = 19, p = 0.04). f Schematic representation on how the inhibition of adipocyte V-ATPase activity by bafilomycin or genetic deletion of Atp6v0a1 leads to increased circulating Gdf15 levels that is associated with increased Pomc expression, reduced food intake and blunted weight gain/re-gain. Data is expressed as mean ± SEM and was analyzed using students t-tests (2 tailed), * denotes statistical significance at p < 0.05.