Studies on the fat mass and obesity-associated (FTO) gene and its impact on obesity-associated diseases

Abstract

Obesity has become a major health crisis in the past ∼50 years. The fat mass and obesity-associated (FTO) gene, identified by genome-wide association studies (GWAS), was first reported to be positively associated with obesity in humans. Mice with more copies of the FTO gene were observed to be obese, while loss of the gene in mice was found to protect from obesity. Later, FTO was found to encode an m⁶A RNA demethylase and has a profound effect on many biological and metabolic processes. In this review, we first summarize recent studies that demonstrate the critical roles and regulatory mechanisms of FTO in obesity and metabolic disease. Second, we discuss the ongoing debates concerning the association between FTO polymorphisms and obesity. Third, since several small molecule drugs and micronutrients have been found to regulate metabolic homeostasis through controlling the expression or activity of FTO, we highlight the broad potential of targeting FTO for obesity treatment. Improving our understanding of FTO and the underlying mechanisms may provide new approaches for treating obesity and metabolic diseases.

Introduction

Obesity, a condition of an abnormally high ratio of body fat, is a risk factor for type 2 diabetes, hypertension, cardiovascular diseases, and other metabolic illnesses.¹ Obesity and its associated sicknesses are expected to influence more than 1 billion people by the year 2030.² Researchers considered that the changes in dietary patterns and increasingly obesogenic environment might account for the obesity pandemic.³ Moreover, the present costs of obesity are estimated at nearly $2 trillion annually from lost economic productivity and direct health care costs. These costs are equivalent to 2.8% of the world's gross domestic product (GDP).⁴ Hence, to mitigate this epidemic and develop safer and more effective treatments, intensive research has been sparked to clarify the underlying mechanisms of obesity.

Over the past decades, however, strategies for obesity treatment and prevention, both at the population and individual level, have not hit the mark due to the complex pathogenesis of obesity.¹ To date, the fundamental drivers of obesity, which is caused by an intricate interplay between behavioral, genetic, environmental, epigenetic, physiological, socio-cultural, and economic factors, are widely considered the consequence of long-term energy disequilibrium between too much food intake and few physical activities (Fig. 1).^1,5, 6, 7, 8 In 2007, four investigations for genetic risk factors in obesity have successfully associated single nucleotide polymorphism (SNPs) in the first intron of fat mass and obesity-associated (FTO) gene with obesity-related features concerning anthropometric traits, type 2 diabetes, early-onset and severe obesity.9, 10, 11, 12 Subsequently, the FTO protein was functionally confirmed as the first RNA demethylate targeting the substantial N⁶-methyladenosine (m⁶A) residues in RNA in vitro.¹³ FTO-dependent m⁶A demethylation was proved to be essential for adipogenesis.¹⁴ Notably, the story of FTO research is interesting, intricate, and instructive: A large body of literatures on FTO has achieved controversial consequences.^15,16 For instance, although most studies confirmed a significant relationship between FTO and obesity, one study found that FTO deficient mice have comparable fat mass as the wild-type controls.^15,17, 18, 19 Besides, both FTO-overexpressing mice and FTO-knockout mice exhibited hyperphagia.^17,19

Figure 1.

Key factors involved in the regulation of energy balance. Body weight will gain when energy intake exceeds energy expenditure.

In this review, we aim to summarize studies on the roles of FTO in adipogenesis and obesity-associated metabolic diseases. Special focus will be given to discuss the ongoing debates including the association between FTO polymorphisms and obesity. Considering the disproportional high expression of FTO and abundant epigenetic modification in the central nervous system,20, 21, 22 we highlight the role of FTO in the control of energy balance. Moreover, the role of nutritional genomics containing epigenomic signatures is a young field with accumulating interest to uncover the causes of obesity and for precision nutrition to fight or prevent obesity and obesity-related diseases.^5,23, 24, 25 Thus, we also analyzed the potential of targeting FTO therapeutically by dietary compounds, especially polyphenols, for the management of obesity.

The history of FTO

In 1994, to explore the process of programmed cell death during limb development, the Fused toes (Ft) mouse was created by transgenic insertional mutagenesis involved in a 1.6 Mb deletion on mouse chromosome 8.^26,27 Heterozygous carriers for the Ft mutation displayed a fusion of forelimb.²⁷ Due to the deletion, the entire IrxB gene cluster (Irx3, Irx5, and Irx6 from the Iroquois gene family) combined with three other genes (FT1, Fasto, and Fantom) were affected by the Ft mutation.²⁶ Among the three genes, which were all largely unknown at that time, FT1 (now called Fts) has been detected in the first place.²⁸ Secondly, Peters et al reported the identification of Fasto (Fto) in the study of the Ft mouse as a result of exon trap analysis of a genomic clone encoding wild-type sequences.²⁹ The third gene, whose transcription initiation site was found very close to the initiation site of Fto, was labeled as Fantom (Ftm, also called KIAA1005 and RPGR interacting protein 1 [RPGRIP1L]) because its sequence was elusive to characterize technically.^11,26,30 When first cloned in 1999, Fto was found to express throughout embryonic development and speculated as a candidate gene related to the establishment of left-right asymmetry, craniofacial development, and programmed cell death.²⁹ In 2007, three independent genome-wide association studies (GWAS) showed a clear link between human obesity and variants within introns of FTO.9, 10, 11 From then on, the mechanisms underlying the association between FTO and obesity have been extensively studied. Notably, great challenges were present at the very start. FTO was a gene of unidentified function with its expression throughout the whole body of both fetal and adult.^11,12,20 Besides, there was no clear genetic mechanism to elucidate how the genetic variation affected the expression or function of FTO. The possibility that the predisposing variants were involved in the development of obesity by affecting the expression or function of genes adjacent to FTO or more distant genes has existed from the very beginning.¹¹ It is also noteworthy, however, that FTO protein was subsequently confirmed to be a member of AlkB family of 2-oxyglutarate and Fe(II) dependent oxidative RNA/DNA demethylases.^31,32 Gerken et al reported that FTO catalyzed demethylation of 3-methylthymine in single-stranded DNA and was highly expressed in the brain, particularly in hypothalamic nuclei regulating energy balance.²⁰ Three following studies in the next three years further identified the association between FTO and obesity.^19,33,34 Afterward, He et al first identified FTO as an RNA demethylase,¹³ thus initiating a wave of research on epigenetic modifications of RNA (Fig. 2). Since then, the diverse and complicated functions of FTO protein in the progress of metabolic diseases, especially obesity, have been gradually elucidated.

Figure 2.

A brief timeline of discovery and exploration of FTO.

FTO expression in various species and tissues

FTO is a large gene containing more than 4,000 kb and 9 exons and is located on human chromosome 16q12.2.³⁵ The ancient evolutionary origin of this gene can be traced back to 450 million years ago.³⁶ Sequence homology comparisons have revealed that the FTO gene was found only in marine algae and vertebrates including fish, chicken, rabbits.36, 37, 38 The pig FTO gene is located on chromosome 6 where numerous quantitative trait loci for the control of fat have been mapped.³⁹ The cDNA sequence of FTO in the pig shared nearly 80% homology with the cDNA sequences of both humans and mice.^40,41 The FTO gene of the human encodes a 505 amino-acid protein with its sequence highly conserved in organisms ranging from green algae to mammals.^20,42 While the mouse FTO protein contains 502 amino acids, with a molecular weight of 58,007 Da (Q8BGW1-1) compared with 58,282 Da in humans (Q9C0B1-1). The FTO protein consists of two well-defined domains: a C-terminal domain (residues 327–498, exons 6–9) (called CTD) and an N-terminal domain (residues 32–326, exons 1–5) (called NTD).^38,43 The CTD plays a critical role in stabilizing the conformation of the NTD.⁴³ NTD is composed of nucleotide recognition lids (NRL1 and NRL2), a highly conserved double-stranded β-helix (DBSH) domain and an extra loop.^43,44 Among the nine human AlkB homologs, FTO is the only member that contains this type of loop (Fig. 3).⁴⁵ The unique loop selectively blocks the ability of dsRNA/DNA to act as a physiological substrate for FTO.⁴⁶ The evolutionary conserved regions and structure of FTO protein have indicated that high similarity of whole-genome alignments exists between humans and mice.⁴⁷ It was reported that although FTO expressed widely in both mice and human tissues,^11,20,36 its mRNA was most abundant in the brain, particularly in hypothalamic nuclei regulating energy balance.²⁰ In human embryos, a nearly ubiquitous expression of FTO was observed, with higher expression in the liver and the central nervous system.⁴⁸ A strong expression of FTO in the mitral and semilunar valves, ventricular myocardium, pituitary, frontonasal and mandibular mesenchyme was also detected. This wide spatiotemporal expression pattern may related to the broad spectrum of clinical manifestations of many diseases like hypertonicity, ventricular septal defect and neuro sensory deafness.⁴⁸ FTO was first shown to be a nuclear protein,²⁰ but afterward, it was reported to exist in both the cytoplasm and cell nucleus in various mammalian cell lines.49, 50, 51 Recent evidence from the pig model suggested that FTO protein also existed in the cytoplasm of specific tissues and cells, e.g., in the pancreatic β-cells in a pig model.⁵² Abundant FTO protein expression was also found in the cerebellum, kidney, and salivary gland of adult pigs. However, FTO protein was not detected in bile, saliva, and blood.⁵² Furthermore, FTO expression in tissues relies on energy intake, metabolic status, and age, although the mechanisms governing this dependency are unknown yet.⁵²

Figure 3.

Functional domains of the human FTO protein. FTO contains 505 amino acids and consists of an N-terminal domain (called NTD) (residues 32–326) and a C-terminal domain (called CTD) (residues 327–498). The catalytic core of the NTD is mainly composed of a highly conserved double-stranded β-helix (DSBH) fold which is homologous to those of Fe(II)/2-oxoglutarate-dependent oxygenase (for a review of structural information of these enzymes, see ref. 41). NTD also possesses two nucleotide recognition lids (NRL1 and NRL2). Residues of NRL1 (residues 77–102), NRL2 (residues 103–116), DSBH, CTD (residues 327–498) and the unique loop of FTO (residues 210–223) are labeled with bule, green, brown, red, and orange, respectively. The highly conserved residues His 231, Asp 233 and His 307 are coordinated to Fe²⁺. Residues (Asn205, Asp 233, Val 244, Tyr 295, His 307, Val 309, Arg 316, Ser 318, Thr 320, Arg 322) involved in interactions with α-KG analog N-oxalylglycine (NOG) are also described.

RNA demethylase of FTO in adipogenesis

Although Gao et al found that FTO deficient mice had comparable fat mass as the wild type controls,¹⁸ numerous studies have demonstrated that FTO plays a vital role in adipogenesis. In vitro, overexpression of FTO promotes adipogenesis in 3T3-L1 preadipocytes, porcine intramuscular preadipocytes and mouse embryonic fibroblasts (MEFs).53, 54, 55, 56, 57, 58, 59 Overexpression of FTO led to an obese phenotype in mice,¹⁷ while loss of FTO in mice generated by replacing exons 2 and 3 of the FTO gene caused a significant decrease in adipose tissue.^19,60,61 A recent study indicated that the protein expression level of FTO in the white adipose tissue of fat pigs is significantly higher than that in lean pigs.⁶² However, its molecular mechanisms via which FTO affects adipogenesis remain largely unknown.

Without a change in the DNA sequence, epigenetics mainly affects the functions and characteristics of genes through controlling the translation or transcription processes.^63,64 The study of epigenetics and its involvement in adipogenesis and metabolic diseases is a young research field.^24,25,65 In the last few decades, more attention has been paid to RNA modification with the help of modern high throughput sequencing techniques. N⁶-methyladenosine (m⁶A), accounting for more than 80% of all RNA base methylations, is the most prevalent modification in eukaryotic mRNAs with a highly conserved architecture across mice and humans.^66,67 Since the FTO protein was confirmed as an RNA demethylase,¹³ whether FTO influences adipogenesis by controlling m⁶A modification has been hypothesized by our group. To verify this, we have conducted a series of studies and established the central role of the RNA demethylase FTO in adipogenesis. For example, we found that deficiency of m⁶A demethylase FTO attenuated adipogenesis in porcine and mouse preadipocytes through Janus kinase 2 (JAK2)/signal transducer and activator of transcription 3 (STAT3)/transcription co-activators CCAT/enhancer-binding protein β (C/EBPβ) signaling.⁵⁶ Mechanistically, knockdown of FTO inhibited JAK2 expression and STAT3 phosphorylation, causing debilitated transcription of C/EBPβ which is necessary for adipocyte differentiation in the early stage. Furthermore, it was forced expression of wild-type (FTO-WT) plasmid, not catalytic inactive mutant FTO (FTO-MUT, R96Q) plasmid that can dramatically enhance JAK2 protein expression and stimulate STAT3-C/EBPβ signal pathway in porcine preadipocytes at the early stage of differentiation.⁵⁶ Similarly, it has been shown that FTO could influence adipogenesis during the process of mitotic clonal expansion (MCE), a prerequisite for adipocyte differentiation, which also appears at an early stage after adipogenic stimulation.⁵⁴ The effect of FTO overexpression on MCE appears to be mediated via increased expression of the pro-adipogenic S inform of runt related transcription factor 1 (RUNX1T1).⁵⁴ While our group discovered that the negative effect of FTO knockdown on MCE may partly attribute to impaired cell cycle progression at the early stage of adipogenesis.⁵⁷ The expression of cyclin A2 (CCNA2) and cyclin-dependent kinase 2 (CDK2), two pivot cell cycle regulators promoting cells from S phase to G2 phase,^68,69 was drastically decreased by FTO knockdown in preadipocytes, leading to a delayed entry of MDI-induced cells into S phase.⁵⁷ Unregulated m⁶A levels of CCNA2 and CDK2 mRNA and decreased triglyceride (TG) deposition were observed following FTO knockdown.⁵⁷ These results suggest that high m⁶A levels lead to less lipid accumulation (Fig. 4). Indeed, our group recently has identified m⁶A is negatively related to fat deposition in muscle tissue by using pigs as a model.⁷⁰ To explore the effects of m⁶A on the adipogenesis of intramuscular preadipocytes and associated mechanisms, we compared m⁶A methylome of the longissimus dorsi muscles (LDMs) between Jinhua pigs (obese-type breed) and Landrace pigs (lean-type breed).⁷⁰ The LDM of Jinhua (J-LDM) pigs had a higher intramuscular fat content than that of Landrace (L-LDM) pigs, whereas the m⁶A/A ratio was higher for L-LDM than for J-LDM. Consistently, the expression level of FTO was higher in J-LDM than in L-LDM.⁷⁰

Figure 4.

The regulation of gene expression by m⁶A modification in adipogenesis. The diagram describes how FTO (blue) and YTHDF2 (green) are engaged in the regulation of various pathways and genes in adipogenesis (modified from ref. 26).

Just like many other chemical modifications on biological macromolecules, m⁶A can also be recognized by specific readers.^71,72 For instance, our group found that deficiency of m⁶A modification on family with sequence similarity 134 member B (FAM134B) enhanced its protein abundance through m⁶A reader YTHDF2-dependent manner and stimulated differentiation of porcine preadipocytes (Fig. 4).⁷³ A recent study by our group showed that FTO could cooperate with YTHDF2, the first identified m⁶A binding reader,⁷⁴ to regulate the progress of adipogenesis.⁷⁵ FTO played an important role in regulating macroautophagy and adipogenesis through targeting autophagy related 5 (ATG5) and ATG7. Knockdown of FTO caused an increase in the m⁶A levels of ATG5 and ATG7 with their transcripts captured by m⁶A-binding protein YTHDF2, which resulted in mRNA degradation and reduction of protein expression. As a result, autophagy and adipogenesis were alleviated (Fig. 4).⁷⁵ In brief, our group uncovered molecular mechanisms responsible for the regulatory role of FTO in adipogenesis which may be potentially significant in the development of new prevention or treatment modalities for obesity.

FTO and energy homeostasis

FTO was highly enriched in neurons and widely expressed in the brain, paraventricular, hypothalamic arched, ventromedial nucleus, dorsomedial, which were engaged in the regulation of energy homeostasis and food intake.^20,76 Global loss of FTO as well as neural-specific FTO knockout mice both had postnatal growth retardation with altered energy intake and expenditure. FTO-deficient mice had apparent hyperphagia.¹⁹ Evidence up-to-date showed that variants in FTO influence appetite and increase food intake through impaired central nervous system satiety processing.77, 78, 79 In addition, within the arcuate nucleus, the expression of FTO was decreased nearly 60% following a fasting period of 48 h and increased after 10 weeks of consuming a high-fat diet.^20,80 Moreover, although the role of FTO in neurogenesis and neurodevelopment is still largely unknown, FTO deficiency in mice led to reduced proliferation and neuronal differentiation of adult neural stem cells and exhibited impaired dopaminergic function which may potentially affect body-weight regulation and energy homeostasis.^81,82 Taken together, these investigations support the apparent primacy of the function of FTO, particularly in modulating energy homeostasis, is partly mediated by the brain. Furthermore, a recent study by our group provided new insight into understanding the relationship between FTO and energy balance.⁸³ Adipose-specific deletion of FTO predisposed mice to prevent high-fat diet (HFD)-induced obesity by enhancing energy expenditure. Additionally, loss of FTO in vitro promoted thermogenesis and white to beige adipocytes transition. Mechanistically, FTO deficiency increased the m⁶A levels of hypoxia inducible factor 1 A (HIF1A) mRNA, which is recognized and bound by m⁶A-binding protein YTHDC2, thus facilitating mRNA translation and protein abundance. HIF1A activated transcription of thermogenic genes including PR domain containing 16 (PRDM16), and peroxisome proliferator activated receptor gamma (PPARG), thereby promoting UCP1 expression and browning process.⁸³ Collectively, these results unveiled an epigenetic mechanism by which m⁶A-mediated HIF1A expression controls browning of white adipocytes and thermogenesis, providing an attractive target to counteract obesity and metabolic diseases.

FTO SNPs and physical activity

The first two GWAS for obesity-related genes that confirmed the FTO gene as a promising candidate identified different SNPs in the gene's first intron significantly associated with BMI.^9,11 Subsequently, the association between multiple FTO SNPs and obesity was further identified by a myriad of population-based European studies.84, 85, 86, 87, 88, 89 Studies have also identified the association between FTO SNPs and BMI in non-European populations, specifically, Asian populations,90, 91, 92, 93 Africans,⁹⁴ and African-Americans.⁹⁵ After the wave of association studies were followed by reports considering how do these FTO SNPs exert their effect on BMI and whether FTO SNPs are associated with physical activity and food intake, the two main mediators of energy balance. Evidence supporting a link between physical activity and FTO on BMI is increasing and has been reviewed.96, 97, 98, 99, 100 Data from observational studies in human beings reported that the FTO SNPs effect on adiposity could be modified by physical activity (PA) level.^100,101 A meta-analysis of 218,166 adults and 19,268 children concluded that the relevance of the FTO rs9939609 variant with the odds of obesity was decreased in physically active adults.¹⁰¹ In another study, European adolescents who meet the daily PA recommendations (exercising for 60 min per day or longer) could abolish the deleterious effect of the FTO rs9939609 on body fat estimates.¹⁰² In Korean, active exercise in humans carrying rs9939609 (AT + AA) genotypes also reduced the risk of obesity by nearly 2-fold compared with those carrying the wild type (TT) homozygote.¹⁰³ Recently, a meta-analysis of 200,452 adults discovered robust evidence of interaction with PA for the strongest known obesity-risk locus in the FTO gene region, of which the body mass-increasing effect is decreased by nearly 30% in physically active individuals compared to inactive individuals.¹⁰⁴ Notably, the adverse impacts of the FTO rs1421085 on BMI may also be overcome by PA.¹⁰⁵ The association with PA and rs1421085 has been observed in a multi-ethnic prospective cohort that included 17,423 people from 17 countries who were followed for 3.3 years.¹⁰⁵ Moreover, low PA has been reported to accentuate the effect of the FTO polymorphism on body fat accumulation.^106,107 In sum, these data provide evidence for genetic variability in response to vigorous PA on weight loss, although clinical applications of these studies require further explorations.

FTO SNPs and food intake

Since time immemorial, the regulation of body weight and food intake has been simply considered as an issue of willpower instinct and self-control.^108,109 After all, Gluttony is one of the Seven Deadly Sins which not only involved a habitual excess of eating and drinking but also referred to being fussy about food and drink. So as obesity is becoming increasingly prevalent,¹ public in turn blame these patients with obesity for a lack of moral principle.^108,109 “Rectify your eating habits and do more exercise” so the advice for patients with obesity goes. However, this sage piece of advice, as far as goes, is distinct not working.^108,110 For the public, two complex and meaningful problems to be settled urgently are: why some people eat more than others and why do we prefer some food over others. Nowadays, an increasing number of genetics-related researches provided new insights into understanding these issues and suggested that the FTO SNPs are related to increased energy intake,^77,111,112 food preference,^33,113 overeating,¹¹⁴ increased intake of dietary fat or protein.^115,116 For example, a GWAS conducted by Tanaka et al for macronutrient intake in more than 70,000 participants confirmed variants in FTO to be significantly relevant to increased protein intake.¹¹⁶ Chuang et al firstly reported the pleiotropic longitudinal effects of the FTO gene and its influence on personality, brain function, and diet in an older population.¹¹⁷ FTO rs1421085 was associated with decreased brain function in the medial prefrontal cortex which mediated a greater preference for dietary fat over time during aging.¹¹⁷ In addition, a study carried on 196 overweight adults of Iran reported that carriers of the rs9939609 AA genotype had significantly higher carbohydrate, calorie, and fat intake than the carriers of the TT genotype.¹¹⁵ While the FTO rs9939609 AA subjects in Emirati were reported to have higher carbohydrate but lower fat intake compared to those with other genotypes.¹¹⁸ Mechanically, decreased levels of ghrelin and IL6 postprandially in individuals with rs9939609 AA genotype may be responsible for their increased energy intake.¹¹⁹ Moreover, a recent study using cross-sectional data from 985 older people found that serum levels of leptin were inversely related to the number of FTO C risk alleles in participants with distinct rs17817449 genotypes.¹²⁰ Taken together, the present findings suggest that FTO SNPs may affect weight gain by shifting the endocrine balance and are highly associated with food intake.

FTO SNPs and genes adjacent to FTO

The obesity-associated FTO region is located within a large topologically associated domain of approximately 2 Mb containing FTO, the IRXB cluster, and RPGRIP1L which encodes a protein localized at the transition zone of the primary cilium.121, 122, 123, 124 As a result of this arrangement, the obesity-associated SNPs may affect the regulation of all or any of these genes. The possibility that the mechanistic basis of interactions between FTO SNPs and obesity involved neighboring genes was raised right from the start.¹¹ Recently, there is some evidence to support this possibility. For example, FTO and RPGR1P1L were both regulated by isoforms P110 and P200 of the transcription factor cut-like homeobox 1 (CUX1) whose binding site overlapped SNP rs8050136 in intron 1 of FTO,⁹ which in turn affected binding affinities of P200 and P110.¹²⁵ In mice, homozygosity for a null allele of RPGR1P1L was embryonically lethal.¹²⁶ While mice heterozygous for a null allele of RPGR1P1L would be fatter than wild-type animals.¹²⁷ Moreover, Rpgrip1l^+/− mice displayed diminished suppression of food intake in response to leptin administration.¹²⁷ Mechanistically, the number of type III adenylyl cyclase (ACIII)-positive cilia in the axonemal part of the cilium was diminished in the hypothalamus of Rpgrip1l^+/− mice, accompanied by diminished pStat3 in response to leptin and impaired convening of the leptin receptor to the vicinity of the cilium.¹²⁷ Smemo et al also supported this possibility by demonstrating that the promoters of the homeobox gene IRX3, as well as FTO, directly interacted with the obesity-associated FTO region in the human, mouse, and zebrafish genomes.¹²⁸ Furthermore, obesity-associated SNPs were related to the expression of IRX3, but not FTO, in human brains.¹²⁸ The expression of IRX3 was positively linked to regulation of body mass and composition demonstrated by a bodyweight reduction of 25% or more in Irx3-deficient mice, primarily through the increase in basal metabolic rate with browning of white adipose tissue and loss of fat mass.¹²⁸ Most recently, Laber et al engineered a deletion of the rs1421085 conserved cis-regulatory module (CRM) in mice and identified that the CRM affects mitochondrial function and Irx3 and Irx5 gene expression in an adipose depot–dependent manner.¹²⁹ The CRM also modulated physical phenotypes that were associated with obesity, including decreased whole-body fat mass and high-fat diet-induced weight gain.¹²⁹ Interestingly, data from an unpublished study indicated that the rs1421085 variant within FTO may promotes thermogenic capacity and was potentially associated with human migration.¹³⁰ Notably, all these results concerning FTO rs1421085 were consistent with previous data from human subcutaneous adipocyte model systems which indicated that adipocyte thermogenesis regulation involving AT-rich interaction domain 5B (ARID5B), rs1421085, IRX3, and IRX5.¹³¹ However, these data did not offer an explanation immediately for the well-known association of variants within FTO with food preference and eating behavior.^33,78,131 To address this question and ascertain the pattern of interactions between the obesity-associated region and genes in the FTO-IRXB locus, a recent study performed by Sobreira et al found that obesity-resistant Irx3^−/− mice exhibited a decreased preference for sucrose, but not protein or lipid, compared with WT animals.¹³² It has also established a central nervous system role of IRX3 in the regulation of metabolism and eating behavior analogous to phenotypes associated with allelic variants of obesity-associated SNPs within FTO in humans.¹³² Nonetheless, there are two important limitations of the study. Firstly, the selection of immortalized cell lines, but not primary cells, for the reporter assays may mask the allelic effects of SNPs seen in primary cells.¹³² Secondly, the manipulation of candidate genes in mice may cause organismal phenotypes that are quantitatively and qualitatively different from the small-effect phenotypes elicited by allelic variants of SNPs associated with the human trait.¹³² Moreover, in which cell groups of the brain, the expression of IRX5 and IRX3 was determined by allelic variants and enhancers in the obesity-associated region are unknown yet. Further studies are needed to address this outstanding and critical question.¹³²

FTO and Type 2 diabetes

Type 2 diabetes (T2D), accounting for approximately 90% of diabetes mellitus, is a complicated metabolic disease characterized by dyslipidemia and hyperglycemia.¹³³ At the first time, FTO was reported to be strongly associated with T2D in an adiposity-dependent manner in the UK and Finnish populations.^11,134,135 Later, one study found that FTO gene expression was not increased in subcutaneous adipose tissues (SAT) and visceral adipose tissues (VAT) from individuals with T2D. Furthermore, VAT FTO mRNA expression was related only to fasting glucose, but not to insulin action.¹³⁶ While some other studies found evidence that levels of FTO were increased in skeletal muscle and adipose tissue from patients with T2D and observed no correlation of both omental adipose tissues (OAT) and SAT FTO expression with either fasting glucose or insulin levels.^137,138 The discrepancy of these studies is difficult to interpret, but it is worth noting that small sample sizes and ethnic differences may account for part of these discrepancies observed.

Yang et al showed that the FTO mRNA expression in white blood cells was positively correlated with fasting glucose and negatively connected with the m⁶A content in all 102 participants with T2D.¹³⁹ FTO expression in patients with T2D with hyperglycemic emergency was significantly higher than those with hypoglycemic emergency. A previous study has shown that m⁶A strongly stimulates glucose oxidation in adipocytes of rats.¹⁴⁰ Recent research showed that the m⁶A content in RNA in patients with T2D and diabetic rats was strongly decreased.¹⁴¹ These data suggest that the dynamic m⁶A level regulated by FTO is related to blood glucose. However, the effect of high-glucose stimulation on FTO expression is still controversial. Hepatic FTO mRNA levels were increased in fasting mice which are influenced by reduced blood glucose levels and body weight.¹⁴² Correspondingly, level of FTO mRNA in liver was significantly lesser in hyperglycemic mice compared to normoglycemic mice.¹⁴³ In humans, levels of FTO mRNA and protein were significantly higher in livers of patients with nonalcoholic fatty liver disease (NAFLD) who are also hyperglycemic compared to healthy patient controls.¹⁴⁴ Specie differences may account for these inconsistent results. Additional investigations should be done to thoroughly explore the effect of long and short-term hyperglycemia on FTO expression in liver tissue.

Forkhead box protein O1 (FOXO1), FASN, glucose-6-phosphatase-alpha (G6PC), and diacylglycerol O-acyltransferase 2 (DGAT2) are involved in glucose and lipid metabolism.145, 146, 147, 148 The transcription factor FOXO1 plays a critical role in the insulin and insulin-like growth factor 1 (IGF-1) signaling pathway.¹⁴⁹ FASN is known to have a key role in T2D.¹⁴⁶ G6PC catalyzes the hydrolysis of the intracellular glucose-6-phosphate to glucose in the final step of gluconeogenesis.¹⁵⁰ DGAT catalyzes the terminal step in the major pathway of triacylglycerol biosynthesis.¹⁴⁷ The expression of FTO was positively relevant to FOXO1, FASN, G6PC, and DGAT2 in HepG2 cells.¹³⁹ Meanwhile, previous data from m⁶A-Seq in HepG2 cells have indicated that FASN and FOXO1 have 2 and 5 m⁶A loci, respectively.⁶⁷ To sum up, these data imply that FTO may participate in the process of T2D by controlling the expression of these genes.

The biological regulation of islet cells, especially pancreatic β-cells, is critical for glucose homeostasis.¹⁵¹ mRNA m⁶A methylation has been demonstrated to be vital for β-cell biology and neonatal β cell mass establishment and contribute significantly to the pathogenesis of T2D.^152,153 It was reported that m⁶A modification can also regulate mature β-cells insulin secretion and survival.^154,155 Depletion of m⁶A levels in EndoC-βH1 cells caused cell cycle arrest and impaired insulin secretion by decreasing pancreatic duodenal homeobox 1 (PDX1) protein levels and AKT serine/threonine kinase (AKT) phosphorylation.¹⁵² As the only hormone in the human body that can lower blood glucose, insufficient secretion of insulin in β-cells is the culprit of T2D. Several cell lines have been used as a useful model to exploit the biological function of FTO in the pancreatic β cells and the interrelated molecular mechanism. The FTO protein was proved to have a quick turnover in the clonal β-cell line INS-1 and its overexpression enhanced insulin secretion.¹⁵⁶ By contrast, the insulin secretion of MIN6 cells was significantly inhibited by FTO overexpression which promoted reactive oxygen species (ROS) production and nuclear factor-kappaB (NF-κB) activation, but FTO silence did not affect insulin secretion. Inhibition of intracellular ROS production could alleviate NF-κB activation and normalize insulin secretion.¹⁵⁷ Recent studies have illustrated that FTO expression was reduced in islets from donors with T2D and that knockdown of FTO expression in GRINCH cells attenuated insulin secretion stimulated by glucose.¹⁵⁸ These contradictory results could be explained by the reasons that cell lines are isolated from different tissues and cells, and as a result, often differ genetically and phenotypically from each other. Thus, data only generated from cell lines should be interpreted cautiously since it could not fully represent the results obtained from in vivo studies. Collectively, these studies implied that the levels of FTO in the pancreas may directly influence insulin release from beta cells. Additional investigations should pinpoint the exact role of the FTO gene as a regulator of pancreatic β-cell function.

FTO and nonalcoholic fatty liver disease

NAFLD is a common co-morbidity related to obesity.¹⁵⁹ Excess fat accumulation is the contributor and common feature of this disease.¹⁵⁹ The expression of FTO was considerably increased in the liver of both animal models and patients with NAFLD.^144,160,161 Increased hepatic FTO mRNA and protein levels in patients with NAFLD were reported to be involved in lipid deposition and oxidative stress, which characterize NAFLD.¹⁶⁰ Overexpression of FTO led to increased triglyceride accumulation in both HepG2 and L02 cells and caused increased malondialdehyde (MDA) levels and decreased superoxide dismutase (SOD) activity which were both studied as markers of tissue injury and oxidative stress in L02 cells.^160,162,163 However, the precise regulatory mechanisms via which FTO controls oxidative stress remain unknown. In conclusion, up-regulation of FTO may contribute to the pathogenesis and progression as well as increased liver damage of NAFLD. Indeed, FTO could enhance lipid accumulation by FTO/Sterol regulatory element binding protein-1c (SREBP1c)/cell death-inducing DNA fragmentation factor-α-like effector c (CIDEC) signaling pathway in hepatocytes in a demethylation-dependent manner. FTO overexpression significantly upregulated the transcriptional levels of SREBP cleavage-activating protein (SCAP) and site 2 proteases (S2P) which are involved in the processing and nuclear translocation of SREBP1c.¹⁶³ Levels of key regulatory factors in lipogenesis such as SREBP1c, CIDEC, fatty acid synthase (FASN), and stearoyl-CoA desaturase (SCD1) were improved followed by transfection of FTO, but not catalytically inactive FTO mutant (R316A).¹⁶³ Consistently, a recent study revealed that FTO reduced mitochondria content, decreased m⁶A levels, and promoted lipid deposition, but the FTO (R316A) mutant did not have this effect.¹⁶² Mechanistically, overexpression of FTO in HepG2 cells enhanced expression levels of mitofusin 1/2 (MFN1/2) and OPA1 mitochondrial dynamin like GTPase (OPA1) (related to mitochondrial fusion), but inhibited expression levels of mitochondrial fission 1 (FIS1), dynamin related protein 1 (DRP1), and mitochondrial fission process protein 1 (MTP18) (regulation of mitochondrial fission), and decreased expression of PPARγ coactivator 1 alpha (PGC-1α) and mitochondrial transcription factor A (TFAM) (involved in mitochondrial biogenesis) (Fig. 5). Taken together, FTO may modulate mitochondrial content by inhibiting mitochondria fission and promoting mitochondrial fusion. Long-term exposure to excessive glucocorticoids (GC) might have a pathogenic role in the progression of NAFLD.¹⁶⁴ Recently study discovered that FTO also participates in the GC-induced NAFLD in a chicken model.¹⁶⁵ Glucocorticoid receptor (GR)-mediated m⁶A modification on mRNA of lipogenic genes and transactivation of FTO contributed to the accumulation of lipids in primary chicken hepatocytes induced by oleic acid/dexamethasone and lipogenic gene activation in corticosterone-induced chicken fatty liver.¹⁶⁵ These studies suggest the possibility that inhibition of FTO expression and/or activity may serve as a new therapeutic method to treat NAFLD.

Figure 5.

Role of FTO in the development of nonalcoholic fatty liver disease (NAFLD). FTO-mediated m⁶A demethylation controls lipogenesis and mitochondrial content. One study indicated that FTO-mediated m⁶A demethylation enhanced the expression of SREBP1c and CIDEC and promoted lipogenesis, resulting in increased lipid accumulation. Another study showed that FTO-mediated m⁶A demethylation promoted expression levels of MFN1/2 and OPA1, but inhibited expression levels of FIS1, DRP1, and MTP18, and decreased expression of PGC-1α, TFAM, thereby reducing mitochondrial content. For oxidative stress, MDA and SOD, the biomarkers of oxidative stress, are changed in the development of NAFLD accompanied by increase expression of FTO. Red arrow: Stimulation. Green arrow: Inhibition.

FTO and cardiovascular diseases

Cardiovascular diseases (CVD) are the leading cause of global mortality and a principal contributor to disability.¹⁶⁶ Although m⁶A has long been considered as the most abundant chemical mRNA posttranscriptional modification in mammalian cells and has an impact on various fundamental bioprocesses, the effect of this process in the heart remains largely unknown. Very recently, Doen et al (2019) identified that m⁶A methylation was essential for normal hypertrophic response in cardiomyocytes indicating that m⁶A also takes part in the cardiac homeostasis.¹⁶⁷ Notably, increased levels of m⁶A were associated with significantly reduced expression of FTO in the ischemic heart.¹⁶⁸ It is also interesting to note that FTO has been associated with many cardiac defects including heart hypertrophy,¹⁶⁹ atrioventricular defects,⁴⁸ arrhythmias,¹⁷⁰ and coronary heart disease.¹⁷¹ Moreover, cardiac ventricular levels of FTO in human embryos were higher in addition to brain and liver tissues even though FTO is expressed ubiquitously.⁴⁸ Therefore, the cardiac FTO mRNA expression level may relate to the physiological state of the heart. Recently study showed that FTO expression was decreased in failing mammalian hearts and hypoxic cardiomyocytes. Forced expression of FTO could block ischemia-induced m⁶A increase and rescued myocardial contractile function by selectively demethylating transcriptions associated with cardiac contraction, controlling Ca²⁺, and stabilizing mRNA to affect myocardial dynamics.¹⁷² Importantly, FTO overexpression in mouse models of myocardial infarction resulted in decreased fibrosis and enhanced angiogenesis.¹⁷² Transgenic overexpression of FTO caused an upregulation of myosin heavy chain associated RNA transcript (Mhrt) and reduced m⁶A modification of Mhrt in the hypoxia/reoxygenation (H/R)-treated myocardial cells.¹⁷³ FTO upregulation repressed apoptosis of H/R-treated myocardial cells, implying that FTO may be a potential target gene for heart failure treatment.¹⁷³ Correspondingly, FTO deficiency leads to diminished heart function.¹⁶⁹ Global knockout of the mouse FTO gene will lead to an imbalance of the autonomic neural modulation of cardiac function and a higher risk for the proarrhythmic remodeling of electrical and structural properties of the mouse heart.¹⁷⁰ Collectively, the regulation of m⁶A modification on the transcription process may be a target for the treatment of cardiovascular diseases.

However, there is an ongoing conflict for the role of FTO in body weight and obesity with some investigations revealing either negative or positive association for FTO with body mass.^128,131 In the context of the negative effects of obesity on almost all the major CVD risk factors,¹⁷⁴ although serval studies showed the possible cardioprotective function of FTO,^169,172 a deeper understanding of the impact of FTO in body mass is needed to make the best of its clinical potential for the treatment of human heart diseases. Furthermore, the learning of m⁶A methylation in myocardial fibers is unknown, and it is a field to be investigated urgently.¹⁷⁵

RNA m⁶A demethylase FTO: a new bridge between nutrition and health

Because of the obtainable and edible- and medicinal-safety properties, extracts from plants, such as vegetables, tea, fruits, have been attracted attention for their potential in the prevention and treatment of metabolic diseases for decades.⁸ Recently, several studies showed that plant-derived polyphenols may regulate adipogenesis and attenuate obesity in an m⁶A-dependent manner. Due to the dynamic and reversible nature of m⁶A modification, understanding more about the mechanisms responsible for their anti-obesity effects will help to develop new approaches or medicines for treating obesity-related disorders.

• 1)Since 2006, using resveratrol as a small molecule was proven to improve health and survival of middle-aged mice on a high-calorie diet and shift their physiology,¹⁷⁶ the naturally occurring polyphenol has been continuously studied. Dietary resveratrol supplementation affected liver function and changed the hepatic m⁶A abundance in piglets.¹⁷⁷ Currently, the beneficial outcome of resveratrol on high-fat diet-induced lipid metabolism disorder was proved that may be due to a decrease of m⁶A RNA methylation and a significant increase of peroxisome proliferator-activated receptor α (PPARα) mRNA.¹⁷⁸ Resveratrol supplementation of high-fat diet-fed mice decreased the level of YTH domain family 3 (YTHDF3) and m⁶A abundance, whereas it upregulated the transcript levels of methyltransferase like 3 (METTL3), alkB homolog 5 (ALKBH5), FTO, and YTH domain family 2 (YTHDF2) in the liver.¹⁷⁸
• 2)The anti-obesity properties of curcumin, a yellow-colored polyphenol from the curcuminoids, are similar to resveratrol, through inhibiting adipocyte differentiation, lipogenesis in adipose tissue.¹⁷⁹ Na Lu and colleagues observed that dietary curcumin affected the expression of METTL14, METTL3, FTO, ALKBH5, and YTHDF2 mRNA, and improved the abundance of m⁶A in the liver of piglets.¹⁸⁰ Further investigations are required to explore the precise mechanisms of the effect of curcumin on the profile of gene-specific m⁶A RNA methylation. Most recently, our group reported that curcumin prevents HFD-induced obesity by decreasing the expression of m⁶A demethylase, ALKHB5, which leads to higher m⁶A-modified TNF receptor-associated factor 4 (TRAF4) mRNA. TRAF4 with higher m⁶A modification was bound and recognized by YTHDF1 resulting in enhanced translation of TRAF4 which inhibits adipogenesis owing to the improved degradation of adipocyte differentiation regulator PPARγ by a ubiquitin-proteasome pathway.¹⁸¹ Our group has also reported that Epigallocatechin gallate (EGCG), the most biologically active and abundant polyphenols in green tea,¹⁸² inhibited adipogenesis by blocking the MCE at the initial stage of adipocyte differentiation.¹⁸³ EGCG decreased the expression of FTO and increased the m⁶A levels of CCNA2 and CDK2 mRNA in 3T3-L1 cells. Interestingly, EGCG increased the expression of YTHDF2, which recognized and decayed the methylated CCNA2 and CDK2 mRNA, resulting in decreased protein levels of CCNA2 and CDK2.¹⁸³

Conclusion

Genetic variation in the first intron of FTO is largely related to adiposity.^11,12 Despite many exciting and profound scientific breakthroughs have been made (Fig. 2), there are still some outstanding questions: what are the mechanisms linking FTO SNPs to obesity and the relevant cell types and target genes? Do sequences of the intronic regions of the FTO gene regulate the expression of genes associated with whole body energy balance? If or how does FTO-determined changes in IRX3 expression influence whole body energy metabolism? Is there a possibility for targeting IRX3 and/or other FTO neighboring genes as a therapeutic strategy against obesity? Do the FTO risk alleles influence the FTO protein or not? Further studies should provide advances in the detailed characterization of these problems. Nevertheless, FTO undoubtedly plays an important part in the development of obesity and contributes to many metabolic diseases. N⁶-Methyladenosine (m⁶A), found in the mid-1970s,^184,185 is the most abundant internal modification in eukaryotic mRNA.¹⁸⁶ m⁶A has been reported to impact various biological and pathological processes.186, 187, 188, 189, 190 The discovery of FTO as an m⁶A demethylase has revealed the significance of FTO in epigenetic regulation. Developing effective inhibitors to target FTO's m⁶A demethylase activity or searching for plant extracts to inhibit the expression of FTO may serve as a feasible way to remedy obesity and associated diseases. Intriguingly, several small-molecule FTO inhibitors have been studied and they exhibit powerful anti-tumor effects in multiple types of cancers.191, 192, 193, 194 In addition, entacapone, a Food and Drug Administration-approved drug, was confirmed as a potent FTO chemical inhibitor by using biochemical and structural experiments. Entacapone administration lowered fasting blood glucose concentrations and decreased body weight in diet-induced obese mice.¹⁹⁵ These investigations provide evidence in targeting FTO for metabolic disorders therapy. However, considering that the ubiquitous expression of FTO and the negative phenotype in humans and mice with FTO deficiency,^48,82,196 further explorations are needed to dig into the underlying mechanisms between FTO and metabolic diseases in-depth, which could help to develop more FTO inhibitors or/and other efficacious drugs to propel the field of epitranscriptomics forward.

Conflict of interests

There is no conflict of interests to declare.

Funding

The work was supported by the National Key R&D Program (No. 2018YFD0500405).

Author contributions

Writing and original draft preparation: Chaoqun Huang; Review & editing: Wei Chen; Revision and Supervision: Xinxia Wang. All authors have read and agreed to the published version of the manuscript.