FTO: linking m⁶A demethylation to adipogenesis

Polymorphism of the FTO gene encoding an N⁶-methyladenosine (m⁶A) RNA demethylase was robustly associated with human obesity; however, the mechanism by which FTO affects metabolism, considering its emerging role in RNA modification, is still poorly understood. A new study published in Cell Research reports novel functions implicating FTO in the regulation of mRNA alternative splicing in the control of adipogenesis.

Obesity is a major concern to western societies and to the growing number of overweight individuals which deal with serious health problems associated with overweight (e.g., diabetes, fatty liver, cardiovascular disease, stroke, and neurocognitive decline). Recently, a common variation of FTO (the fat-mass and obesity gene) was found to be robustly associated with increased body mass in humans¹. FTO encodes an RNA demethylase, whose major substrate is N⁶-methyladenosine (m⁶A)². m⁶A is the most prevalent mRNA internal modification with a highly conserved architecture across human and mouse³. This reversible epigenetic modification was shown to participate in regulating important cellular processes, including transport, degradation, and translation⁴. This diversity of functions is mediated by a variety of m⁶A reader proteins including the RNA-stabilizing protein, ELAVL1, and a number of YTH-family of proteins, YTHDF2 and YTHDF3, several of which have been identified to directly bind to m⁶A-containing transcripts³. A role for m⁶A in splicing regulation was demonstrated in several studies. Knockdown of components of the methylating complex (METTL3, METTL14 and WTAP) resulted in different isoforms of m⁶A-containing transcripts^3,4,5. In addition, m⁶A seems to alleviate the constraints imposed by larger exons on the exon-definition machinery and affect splicing efficiency, resulting in exon inclusion³. This is further supported by the fact that ALKBH5, another m⁶A demethylase, affects the rate of splicing⁴. In addition, ALKBH5, FTO, as well as the m⁶A methylation complex components, all display nuclear speckle localizations, further supporting a role for m⁶A in RNA processing^2,4,5. Collectively, these suggest that FTO may affect splicing of its target genes.

A causal link between FTO and obesity was recently established in animal studies. Overexpression of FTO led to increased food intake and obesity in mice⁶, while silencing of FTO protected against obesity⁷. A link between FTO-dependent m⁶A demethylation and gene expression was demonstrated in two independent studies. FTO-knockout mice exhibited changes in m⁶A levels in midbrain and striata, with higher m⁶A methylation level of key genes of the dopaminergic pathway and lower expression of these genes⁸. In addition, FTO risk allele was correlated with higher levels of the ghrelin intestinal hunger hormone, along with a reduced level of m⁶A in ghrelin mRNA transcripts after a test meal⁹. FTO is ubiquitously expressed with highest expression level during development as well as in adult brain, adipose and muscle tissues. Yet, the mechanisms by which FTO-dependent m⁶A demethylation affects key cellular activities operating in these tissues, and its roles in regulation of energy homeostasis as well as body weight remain to be elucidated.

Now, the pioneering study by Zhao et al.¹⁰ identified FTO as an important regulator of mRNA splicing and implicated its necessity in adipogenesis. The authors describe a role for FTO in differentiation of pre-adipocytes by demonstrating that its depletion interferes with pre-adipocyte differentiation, in contrast to an opposite effect observed in knockdown of METTL3. A decrease in FTO is observed along the course of adipocyte differentiation, and as expected the levels of m⁶A in mRNAs concomitantly rise through this process. The authors show that m⁶A RNA decoration during adipogenesis is inversely correlated with gene expression, in agreement with the recently demonstrated m⁶A-dependent mRNA decay mediated by YTHDF2⁴. Furthermore, they identified a subset of genes related to important cellular pathways including lipid synthesis and translation that respond in an opposite manner to FTO or METTL3 knockdown.

One of the most significant observations of the study by Zhao et al. is the identification of FTO as an important regulator of mRNA alternative splicing. As previously reported, significant differences in the proportion of single- and multi-isoform genes exists between m⁶A-methylated and non-methylated genes³, suggesting that m⁶A may pose an important mark for recruiting components of the splicing machinery. In accordance, the authors show that the splicing regulatory (SR) proteins Srsf1 and Srsf2 are recruited in an m⁶A-dependent manner. They show that m⁶A sites overlap with binding sites of Srsf1 and Srsf2, and that m⁶A is overrepresented in 5′ and 3′ exonic sequences flanking the splice sites. Consequentially, knocking down FTO enhances the binding affinity of Srsf2 to several of its targets, which coincides with high methylation levels of these targets, promoting exon inclusion.

During the course of differentiation of pre-adipocytes to adipocytes, the authors observed over 17 000 differentially expressed isoforms. Gene ontology (GO) analysis encompassing differentially expressed isoforms with opposed expression patterns in FTO vs METTL3 knockdown, revealed enrichment in pathways associated with sterol metabolism and organelle organization. Remarkably, the authors demonstrate that knockdown of FTO affects alternative splicing patterns of several genes, including two splice variants of the adipogenesis-related gene Runt-related transcription factor 1 (Runx1t1). Knocking down FTO dramatically leads to higher inclusion rate of exon 6 and generation of long Runx1t1 transcripts, with nearly complete elimination of the short splice variant (Figure 1). Notably, during the progression of adipocyte differentiation the expression level of the short splice isoform of Runx1t1 decreases in a manner similar to the one observed with FTO expression level. In addition, fat-deposits staining and triglyceride assays further support a role for the long Runx1t1 isoform in inhibiting differentiation, while the short isoform appears to promote it.

Figure 1

An illustration of alternative splicing of Runx1t1 mediated by FTO-dependent m⁶A demethylation and Srsf2. The typical removal of m⁶A by FTO (lower panel) minimizes the recruitment of Srsf2, which in its absence prompts the skipping of exon 6, resulting in a short product of Runx1t1. In turn, the short isoform of Runx1t1 induces pre-adipocyte differentiation. Conversely, m⁶A-containing transcripts (enhanced by FTO knockdown) are more likely to recruit Srsf2 and induce exon 6 inclusion, thereby inhibiting pre-adipocyte differentiation.

Collectively, the current study provides valuable evidence for the role of FTO, via m⁶A demethylation, in modulating pre-adipocyte differentiation. The present study makes an important advancement in our understanding of the link between FTO and metabolism, by unraveling a novel mechanism of splicing regulation of its methylated transcript targets. Interestingly, it identified the adipogenesis-related gene Runx1t1 as a prime example of an FTO-regulated alternatively spliced target. Further studies will be required in order to identify whether Srsf1 and Srsf2 are direct readers of m⁶A or are recruited through mediation of other m⁶A reader proteins. Similarly, in vivo examination of the effects of FTO on gene splicing and influence on adipose and other metabolic tissues in a whole animal model would prove valuable.