Diabetes Volume 63, November 2014 3955

Christian Benedict,1 Tomas Axelsson,2 Stefan Söderberg,3 Anders Larsson,4 Erik Ingelsson,5,6 Lars Lind,7 and Helgi B. Schiöth1

Fat Mass and - Associated (FTO)Is Linked to Higher Plasma Levels of the Hunger Hormone Ghrelin and Lower Serum Levels of the

Satiety Hormone in Older GENETICS/GENOMES/PROTEOMICS/METABOLOMICS Adults

Diabetes 2014;63:3955–3959 | DOI: 10.2337/db14-0470

The mechanisms through which common polymorphisms gain in by shifting the endocrine balance from in the fat mass and obesity-associated gene (FTO) drive the satiety hormone leptin toward the hunger-promoting the development of obesity in humans are poorly under- hormone ghrelin. stood. Using cross-sectional data from 985 older people (50% females) who participated at age 70 years in the Prospective Investigation of the Vasculature in Uppsala Population-based studies have repeatedly shown that Seniors (PIVUS), circulating levels of ghrelin and leptin subjects carrying specific single nucleotide polymorphisms were measured after an overnight fast. In addition, sub- (SNPs) in the fat mass and obesity-associated gene (FTO) jects were genotyped for FTO rs17817449 (AA, n = 345 are more prone to gain weight and to develop obesity and [35%]; AC/CA, n = 481 [48.8%]; CC, n = 159 [16.1%]). associated comorbidities (1–4). Using functional MRI, a re- Linear regression analyses controlling for sex, self- cent small clinical trial involving normal-weight young reported physical activity level, fasting plasma glucose, men revealed that the FTO obesity-risk rs9939609 A and BMI were used. A positive relationship between the —located in the first of this gene—is associated number of FTO C risk and plasma ghrelin levels with an enhanced brain response to hedonic food stimuli was found (P = 0.005; relative plasma ghrelin difference (5). Furthermore, a divergent neural responsiveness to between CC and AA carriers = ∼9%). In contrast, serum circulating acyl ghrelin within brain regions that regulate levels of the satiety-enhancing hormone leptin were in- appetite, reward processing, and incentive motivation was versely linked to the number of FTO C risk alleles (P = observed between AA and TT subjects (5). Finally, by us- 0.001; relative serum leptin difference between CC and ing cell models, the authors demonstrated that FTO over- AA carriers = ∼11%). These associations were also expression reduced ghrelin mRNA N6-methyladenosine found when controlling for waist circumference. The methylation, concomitantly increasing ghrelin mRNA present findings suggest that FTO may facilitate weight and peptide levels (5). Ghrelin, which is mainly produced

1Department of Neuroscience, Uppsala University, Uppsala, Sweden 6Wellcome Trust Centre for Genetics, University of Oxford, Oxford, U.K. 2Department of Medical Sciences, Molecular Medicine and Science for Life 7Department of Medical Sciences, Cardiovascular Epidemiology, Uppsala Univer- Laboratory, Uppsala University, Uppsala, Sweden sity, Uppsala, Sweden 3 Department of Public Health and Clinical Medicine and Heart Center, Umeå Corresponding author: Christian Benedict, [email protected]. University, Umeå, Sweden Received 21 March 2014 and accepted 27 May 2014. 4Department of Medical Sciences, Clinical Chemistry, Uppsala University, Uppsala, Sweden © 2014 by the American Diabetes Association. Readers may use this article as 5Department of Medical Epidemiology and Science for Life Laboratory, Uppsala long as the work is properly cited, the use is educational and not for profit, and University, Uppsala, Sweden the work is not altered. 3956 FTO and Ghrelin and Leptin Diabetes Volume 63, November 2014 by the stomach, causes hyperphagia while decreasing the Clinical and Biochemical Investigation energy expenditure (6). Therefore, a ghrelin-driven shift At the age of 70 years, blood samples were collected in the in the energy balance to positive values could be hypoth- morning after an overnight fast. No medication or smoking esized to promote weight gain in those who carry the was allowed after midnight. Plasma ghrelin and serum FTO obesity-risk rs9939609 A allele. However, this pre- leptin were analyzed with commercially available assays vious observation that FTO is linked to ghrelin metabo- (Linco Research, St. Charles, MO). Fasting plasma glucose lism is based on a small sample of young men. Further, levels were measured by standard laboratory techniques. only carriers homozygous for either rs9939609 A allele Height and weight was used to calculate BMI (kg/m2). or rs9939609 T allele were included in the study (5). As previously reported (10), the number of self- However, approximately 50% of the general population reported light (non2sweat-inducing) and hard (sweat- carries only one copy of the FTO obesity-risk rs9939609 A inducing) exercise activities with a duration of at least allele (5). Thus, the finding that FTO is linked to ghrelin 30 min per week was used to assign each participant’s metabolism requires further validation in studies involv- physical activity level to one of the four physical activity ing larger samples, as well as using all three available categories: very low, low, medium, and high. genotypes (i.e., AA, AT/TA, and TT). Contrary to ghrelin, Statistical Analysis leptin that mainly stems from subcutaneous adipose tis- Data were analyzed using linear regression models. Ghrelin sue leads to lower food intake (7). In addition, it reduces and leptin variables were naturally log-transformed and brain activation in regions linked to hunger (e.g., insula) subsequently standardized to approach normal distribution while enhancing activation in regions linked to inhibition (Fig. 1B). In the case of leptin, serum levels were standard- and satiety (e.g., prefrontal cortex) (8). ized by sex due to the large sex differences in circulating Against this background, in the present population- levels (11). If not otherwise specified, linear regression based study involving 985 men and women at age 70, we analyses were adjusted for sex (on models on ghrelin examined the link between the FTO obesity-risk only), fasting plasma glucose, BMI, and self-reported rs17817449 C allele and circulating levels of ghrelin and physical activity level. Reported regression coefficients leptin measured in the morning after an overnight fast. are unstandardized coefficients 6 SE. Overall, two-tailed FTO rs17817449 was chosen instead of FTO rs9939609 P values , 0.05 were regarded as significant. as it reached a higher genotyping success rate in the cur- rent study (n = 985 vs. n = 800). Importantly, correla- RESULTS tional analysis in the subsample of 800 participants Descriptive Characteristics revealed singularity (i.e., perfect linkage disequilibrium) Descriptive characteristics for the cohort are shown in between the FTO rs9939609 SNP and FTO rs17817449 Table 1. No differences in sex ratio and self-reported SNP (r2 = 1). All analyses were controlled for BMI, self- physical activity level were found between the three reported physical activity, sex, and fasting plasma glucose FTO rs17817449 groups (P . 0.05 for all x2 compari- levels. sons). For 51 individuals, no physical activity data were RESEARCH DESIGN AND METHODS available (5%), i.e., their data were not entered into the main regression analysis. No association was found be- Design Overview, Participants, and Genotyping tween BMI and the FTO rs17817449 SNP (P . 0.05, Details of the Prospective Investigation of the Vascula- adjusted for sex). In contrast, fasting plasma glucose con- ture in Uppsala Seniors (PIVUS) have been reported centrations significantly differed between the three FTO previously (9). Briefly, all subjects aged 70 years and rs17817449 groups (b [SE] = 20.167 [0.071], P = 0.018, living in the community of Uppsala, Sweden, were eligi- adjusted for sex, BMI, and self-reported physical activity ble. The subjects were chosen from the Total Population level). Register and were invited in a randomized order from the start of the study in April 2001 to the last included Plasma Ghrelin and Serum Leptin Concentrations subject in June 2004. Of the 2,025 subjects invited, Linear regression analysis demonstrated a positive asso- 1,016 subjects (507 men and 509 women) were investi- ciation between the number of FTO rs17817449 risk gated (50%). Of these individuals, 985 (97%) were suc- alleles (i.e., allele C) and plasma ghrelin concentrations cessfully genotyped for the FTO rs17817449 SNP measured after an overnight fast (P = 0.005; Fig. 1C, ( 16) as a part of a custom Illumina iSelect left panel). In contrast, serum leptin concentrations genotyping array (average call rate of 99.9%). Testing for were lower the more FTO rs17817449 risk alleles an in- Hardy-Weinberg equilibrium (using a x2 test, 1 df) dividual had (P = 0.001; Fig. 1C, right panel). Importantly, revealed that the SNP did not deviate from expected the use of waist circumference instead of BMI as a mea- genotype proportion (Fig. 1A). sure of body adiposity did not change the direction and The study was approved by the Uppsala University significance of the observed associations (leptin: b [SE] = ethics committee. All participants gave their written 20.83 [0.035], P = 0.017; ghrelin: b [SE] = 0.124 [0.049], informed consent. The study was conducted according P = 0.011; adjusted for sex [when appropriate], waist to Declaration of Helsinki principles. circumference, self-reported physical activity level, and diabetes.diabetesjournals.org Benedict and Associates 3957

Figure 1—Link between the obesity-risk FTO rs17817449 C allele and circulating levels of ghrelin and leptin in a Swedish population of men and women at age 70 years. A: Genotype frequency of FTO rs17817449 in participants at age 70 years from PIVUS. B: Distribution of standardized circulating levels of ghrelin and leptin (standardized by sex due to the large sex differences in circulating levels [17]) following natural log-transformation. C: Significant association between the number of obesity-risk FTO rs17817449 C alleles and circulating levels of ghrelin and leptin (raw values are shown). Linear regression analyses were adjusted for sex (on models on ghrelin only), fasting plasma glucose, BMI, and self-reported physical activity level. Reported regression coefficients (i.e., b) are unstandardized coefficients 6 SE. Overall, two-tailed P values < 0.05 were regarded as significant.

fasting plasma glucose levels). Moreover, imputation of the stomach and is integrally involved in the hedonic missing values for self-reported physical activity did not value of food (12). In contrast, leptin is an adiposity signal change the results (data not shown). Finally, a Pearson that circulates in proportion with fat mass (13). Leptin correlational analysis revealed that serum leptin (r2 = administration can decrease food intake, increase energy 0.719, P , 0.001) but not plasma ghrelin (r2 = 20.07, expenditure, and cause weight loss, whereas deficiencies P = 0.830) were significantly correlated with BMI. in leptin are associated with obesity (14). Against this background, the results of the current study indicate DISCUSSION that a shift from satiety-enhancing toward appetite- We demonstrate in elderly subjects that a common stimulating hormones in the circulation may increase the obesity-susceptibility variant (rs17817449) located in an susceptibility for energy surplus in subjects carrying the FTO intron region of the FTO gene is linked to higher plasma common obesity-risk rs17817449 C allele. levels of the hunger-promoting hormone ghrelin after an Comparison With the Literature overnight fast. Concomitantly, the FTO risk allele was as- Our findings of increased plasma ghrelin concentrations sociated with lower serum levels of the satiety-enhancing after an overnight fast complement results of a previous adipokine leptin. Importantly, the observed associations experiment (5) involving 10 homozygous carriers of the were independent of body adiposity, as indicated by the FTO obesity-risk rs9939609 A allele and 10 homozygous robustness of our results when adjusting for waist circum- carriers of the FTO rs9939609 T allele (this SNP occurred ference instead of BMI. Ghrelin is primarily released by in perfect linkage in a subsample of our cohort with 3958 FTO and Ghrelin and Leptin Diabetes Volume 63, November 2014

Table 1—Raw data of study participants at age 70 years, stratified by the number of obesity-risk FTO rs17817449 C alleles FTO rs17817449 genotype AA AC/CA CC n 345 481 159 Women, n (% genotype group) 164 (48) 240 (50) 88 (55) Self-reported physical activity level, n (% genotype group) Very low 27 (8) 66 (14) 17 (11) Low 213 (62) 267 (56) 100 (63) Medium 71 (20) 97 (20) 28 (17) High 16 (5) 24 (5) 6 (4) Missing values 18 (5) 27 (5) 8 (5) BMI (kg/m2) 27.3 6 4.4 27.0 6 4.5 26.8 6 3.9 Blood glucose levels (mmol/L) 5.6 6 2.1 5.2 6 1.2 5.2 6 1.6 Data are mean 6 SD, unless indicated otherwise. rs17817449). While fasting ghrelin levels did not signifi- with an increased of FTO in brain struc- cantly differ between FTO genotypes in this small clinical tures involved in appetite control. However, evidence for trial, homozygous carriers of the FTO obesity-risk a regulatory effect of FTO on the expression, secretion, or rs9939609 A allele exhibited attenuated postprandial sup- degradation of leptin remains to be uncovered. Against pression of both hunger and circulating ghrelin levels this background, using animal models will help to deci- compared with TT carriers (2). In this study, no genotypic pher the molecular mechanisms through which an FTO differences were observed for either fasting or postprandial overexpression or functional long-range targets of obesity- serum leptin levels (5). The discrepancies in results between associated variants within FTO (e.g., IRX3 [18]) may pre- this study and ours might be explained by differences in dispose humans to exhibit higher plasma levels of ghrelin subject characteristics, e.g., our sample consisted of elderly and lower serum levels of leptin. men and women at age 70 years, whereas the other study The finding that serum levels of leptin were linked to involved young adult men (i.e., age ,30 years) (5). both BMI and FTO rs17817449 and no link between FTO In another study from the Quebec City metropolitan rs17817449 and BMI was found might indicate that this area, it has been demonstrated in 359 men and women SNP is more closely related to leptin than BMI. One ex- that the risk allele of FTO was associated with higher planation could be that FTO or its functional long-range plasma leptin, but this was abolished after adjusting for targets (18) exerts a more direct effect on fat mass (which BMI (15). The divergence between our results and those can be better reflected by leptin) rather than overall obesity. of the Quebec Family Study (15) may be explained by dif- Limitations ferences in participant age (70 years vs. 41 years), inclusion In contrast to large population-based genome-wide asso- of confounders in the analysis, and sample size that was ciation studies (1), no link was observed between FTO and entered into the final analysis (ours was ;2.7-fold greater). BMI, indicating that the lacking association between FTO and BMI is just a consequence of our relatively small sam- Potential Mechanisms for the Observed Associations FTO Although we cannot establish causality in our observa- ple size. While was linked to higher plasma ghrelin and lower serum leptin concentrations after an overnight tional study, there are findings from previous studies that fast—a condition that is conducive for increased hunger may explain by which mechanisms FTO enhances plasma (19)—we cannot draw firm conclusions by which extent ghrelin in humans. For instance, in cell models, FTO over- such endocrine alterations would lead to energy surplus, expression reduced ghrelin mRNA N6-methyladenosine as we neither measured acute food intake or hunger at the methylation, concomitantly increasing ghrelin mRNA time when blood was sampled. Another limitation is that and peptide levels (2). fl FTO we measured total, but not active, ghrelin in blood. Finally, Using double-immuno uorescence staining, also fi has been colocalized with the leptin receptor long isoform generalization of our ndings to other age-groups or eth- nic groups may not be appropriate. in arcuate nucleus of and the nucleus of the solitary tract (16), two brain regions that are very impor- Conclusions tant for the central nervous system control of food intake The present cross-sectional analysis provides a strong and energy expenditure (17). Interestingly in this previ- rationale for hypothesizing that FTO may facilitate weight ous study, leptin administration also downregulated FTO gain by tipping the scale of circulating signals involved in in vitro arcuate nucleus of hypothalamus cultures and central nervous system food intake control toward appetite- in vivo wild-type mice (16). This suggests that lower cir- stimulating factors. However, unless independent cohorts culating leptin levels—as observed in subjects carrying the can replicate that the FTO risk allele is linked to higher obesity-risk FTO allele in our population—may concur plasma ghrelin and lower serum leptin levels, caution is diabetes.diabetesjournals.org Benedict and Associates 3959 needed before generalizing our results to other age-groups 7. Friedman JM, Halaas JL. Leptin and the regulation of body weight in or ethnicities. For instance, in a study involving young men mammals. Nature 1998;395:763–770 (5), FTO was linked to an altered postprandial plasma 8. Baicy K, London ED, Monterosso J, et al. Leptin replacement alters brain fi ghrelin response, whereas fasting levels of this hormone— response to food cues in genetically leptin-de cient adults. Proc Natl Acad Sci USA 2007;104:18276–18279 contrary to our findings—were not linked to this gene. 9. Lind L, Fors N, Hall J, Marttala K, Stenborg A. A comparison of three dif- ferent methods to evaluate endothelium-dependent vasodilation in the elderly: the Prospective Investigation of the Vasculature in Uppsala Seniors (PIVUS) study. Funding. This work is funded by the Swedish Research Council (E.I., L.L., Arterioscler Thromb Vasc Biol 2005;25:2368–2375 H.B.S.), Swedish Brain Research Foundation (C.B., H.B.S.), Novo Nordisk Foun- 10. Benedict C, Brooks SJ, Kullberg J, et al. Association between physical dation (C.B.), and Åke Wibergs Foundation (C.B.). The funding sources had no activity and brain health in older adults. Neurobiol Aging 2013;34:83–90 role in the design and conduct of the study; collection, management, analysis, 11. Gonzalez M, Lind L, Söderberg S. Leptin and endothelial function in the and interpretation of the data; and preparation, review, or approval of the elderly: the Prospective Investigation of the Vasculature in Uppsala Seniors manuscript. (PIVUS) study. Atherosclerosis 2013;228:485–490 Duality of Interest. No potential conflicts of interest relevant to this article 12. Dickson SL, Egecioglu E, Landgren S, Skibicka KP, Engel JA, Jerlhag E. The were reported. role of the central ghrelin system in reward from food and chemical drugs. Mol Author Contributions. T.A., S.S., A.L., E.I., and L.L. researched data. All Cell Endocrinol 2011;340:80–87 authors wrote the manuscript. C.B. is the guarantor of this work and, as such, 13. Maffei M, Halaas J, Ravussin E, et al. Leptin levels in human and rodent: had full access to all the data in the study and takes responsibility for the integrity measurement of plasma leptin and ob RNA in obese and weight-reduced sub- of the data and the accuracy of the data analysis. jects. Nat Med 1995;1:1155–1161 14. Zhang Y, Proenca R, Maffei M, Barone M, Leopold L, Friedman JM. Posi- References tional cloning of the mouse obese gene and its human homologue. Nature 1994; 1. Fall T, Ingelsson E. Genome-wide association studies of obesity and met- 372:425–432 abolic syndrome. Mol Cell Endocrinol 2014;382:740–757 15. Do R, Bailey SD, Desbiens K, et al. Genetic variants of FTO influence adi- 2. Ho AJ, Stein JL, Hua X, et al.; Alzheimer’s Disease Neuroimaging Initiative. A posity, insulin sensitivity, leptin levels, and resting metabolic rate in the Quebec commonly carried allele of the obesity-related FTO gene is associated with reduced Family Study. Diabetes 2008;57:1147–1150 brain volume in the healthy elderly. Proc Natl Acad Sci USA 2010;107:8404–8409 16. Wang P, Yang FJ, Du H, et al. Involvement of leptin receptor long isoform 3. Melka MG, Gillis J, Bernard M, et al. FTO, obesity and the adolescent brain. (LepRb)-STAT3 signaling pathway in brain fat mass- and obesity-associated Hum Mol Genet 2013;22:105021058 (FTO) downregulation during energy restriction. Mol Med 2011;17:523–532 4. Hess ME, Hess S, Meyer KD, et al. The fat mass and obesity associated 17. Schwartz MW, Woods SC, Porte D Jr, Seeley RJ, Baskin DG. Central nervous gene (Fto) regulates activity of the dopaminergic midbrain circuitry. Nat Neurosci system control of food intake. Nature 2000;404:661–671 2013;16:1042–1048 18. Smemo S, Tena JJ, Kim KH, et al. Obesity-associated variants within FTO 5. Karra E, O’Daly OG, Choudhury AI, et al. A link between FTO, ghrelin, and form long-range functional connections with IRX3. Nature 2014;507:371–375 impaired brain food-cue responsivity. J Clin Invest 2013;123:3539–3551 19. Nogueiras R, Tschöp MH, Zigman JM. Central nervous system regulation 6. Nakazato M, Murakami N, Date Y, et al. A role for ghrelin in the central of energy metabolism: ghrelin versus leptin. Ann N Y Acad Sci 2008;1126: regulation of feeding. Nature 2001;409:194–198 14–19