Evidence against a beneficial effect of irisin in humans

Abstract

Brown adipose tissue has gained interest as a potential target to treat obesity and metabolic diseases. Irisin is a newly identified hormone secreted from skeletal muscle enhancing browning of white fat cells, which improves systemic metabolism by increasing energy expenditure in mice. The discovery of irisin raised expectations of its therapeutic potential to treat metabolic diseases. However, the effect of irisin in humans is unclear. Analyses of genomic DNA, mRNA and expressed sequence tags revealed that FNDC5, the gene encoding the precursor of irisin, is present in rodents and most primates, but shows in humans a mutation in the conserved start codon ATG to ATA. HEK293 cells transfected with a human FNDC5 construct with ATA as start codon resulted in only 1% full-length protein compared to human FNDC5 with ATG. Additionally, in vitro contraction of primary human myotubes by electrical pulse stimulation induced a significant increase in PGC1α mRNA expression. However, FNDC5 mRNA level was not altered. FNDC5 mRNA expression in muscle biopsies from two different human exercise studies was not changed by endurance or strength training. Preadipocytes isolated from human subcutaneous adipose tissue exhibited differentiation to brite human adipocytes when incubated with bone morphogenetic protein (BMP) 7, but neither recombinant FNDC5 nor irisin were effective. In conclusion, our findings suggest that it is rather unlikely that the beneficial effect of irisin observed in mice can be translated to humans.

Figure 1. The human FNDC5 gene differs from other species by a mutation in the start codon.

(A) Multiple alignment of the exon 1 sequences: the conserved partial Kozak ATG start sequence of FNDC5 is bold and red. The mutated ATG to ATA in human is bold and blue. There is no other ATG present in exon 1. (B) Multiple sequence alignment of FNDC5 proteins of different species including two human versions. FNDC5_human_o: sequence published by Boström et al.; FNDC5_human_c: current version in Uniprot; red M = start methionines including the potential downstream human start site; light blue = irisin sequence; blue I = mutated start site claimed to be a non canonical start site; purple LRL = sequence shown in UniProt (Q8NAU1_old) as MRR. The underlined sequence indicates the transmembrane part of the protein. Green sequence = peptide used for the generation of the Abcam FNDC5 antibody.

Figure 2. Human FNDC5 with an ATA start codon is translated into full-length protein only at very low abundance.

(A) Schematic representation of the predicted FNDC5 protein structures. Using the first ATG/ATA as start codon of human FNDC5 tagged with GFP would result in full-length FNDC5 protein (a). The use of downstream ATG as start codon would result in truncated FNDC5 protein isoforms (b and c). Murine FNDC5 with ATG as start codon tagged with GFP (d) or without GFP (e). (B) Expression of FNDC5 in HEK293 cells. Cells transfected with constructs containing human FNDC5-GFP gene with ATA and ATG as start codon as well as mouse FNDC5-GFP gene were analyzed 24 h after transfection. Cell lysates were analyzed by immunodetection using antibodies against irisin/FNDC5 and GFP. Cell lysates were treated with PNGaseF to deglycosylate proteins. (C) Supernatants of primary human and C2C12 myotubes were collected for 24 h in serum-free medium and concentrated 60fold using centrifugal filter devices. Irisin protein levels in concentrated supernatants were measured using EIA kit. Medium alone showed no cross-reactivity with the kit; n = 5, *** p<0.001.

Figure 3. FNDC5 mRNA level is not contraction-regulated in skeletal muscle cells and is not increased by endurance or strength training in humans.

(A) and (B) Primary human skeletal muscle cells were differentiated in αMEM containing 2% (vol./vol.) horse serum, followed by overnight starvation, and subjected to EPS for 24 h in serum-free medium (1 Hz, 2 ms, 11.5 V). Relative gene expression of PGC1α, FNDC5 (A), MYH1, 2, and 7 (B) was measured by quantitative real-time PCR (qRT-PCR). All expression data were normalized to actin; n = 5 (A), n = 10 (B); **p<0.01. White bars, control (non-EPS); black bars, EPS. (C) qRT-PCR analysis of FNDC5 expression in m. vastus lateralis from young sedentary males before (Pre) and after 10 weeks (Post) of aerobic interval training (n = 6). (D) qRT-PCR analysis of FNDC5 expression in m. trapezius from sedentary males before (Pre) and after 11 weeks (Post) of strength training (n = 7). All expression data were normalized to RPLP0. Data are presented as mean values ± SEM.

Figure 4. BMP7 activates the brite fat gene program in human adipocytes, but not FNDC5 and irisin.

Isolated preadipocytes from human subcutaneous preadipocytes of different donors were differentiated in the presence of 50/ml BMP7, 200 ng/ml FNDC5 (Abnova), 200 ng/ml FNDC5 (Phoenix) and 60 ng/ml irisin (Phoenix). (A) Relative gene expression of PPARγ, UCP1, TCF21 and ZIC1 was measured by qRT-PCR after 12–14 days of differentiation. All expression data were normalized to actin; n≥4; ***p<0.001. (B) PGC1β and CYCS mRNA expression was assessed by using microfluidic card TaqMan gene expression assay, n ≥4, *p<0.05, **p<0.01. (C) Relative gene expression of CD137 was measured by qRT-PCR on day 0 of differentiation; n = 12; ***p<0.001. (D) The increase of UCP1 and PPARγ expression in six individual donors was compared after BMP7 and FNDC5 (Abnova) incubation, respectively. Preadipocytes with high CD137 expression showed a more robust activation of UCP1 compared to PPARγ after BMP7 incubation. (E) Cell lysates were analysed by immunodetection using an oxidative phosphorylation antibody cocktail. A representative blot is shown. (F) Signal intensities of all complexes of the oxidative phosphorylation were quantified, summed up and normalized to ß-actin, n = 3–5, *p<0.05. (A-F) White bars, control; black bars, BMP7; horizontally hatched bar, FNDC5 (Abnova), diagonally hatched bar, FNDC5 (Phoenix); crossed bar, irisin. Data are presented as mean values ± SEM.

Figure 5. Gene expression analysis of human adipocytes after incubation with BMP7, FNDC5 and irisin.

Isolated preadipocytes from human subcutaneous AT of different donors were differentiated in the presence of 50/ml BMP7, 200 ng/ml FNDC5 (Abnova), and 60 ng/ml irisin (Phoenix). Gene expression of 40 genes, related to adipocyte differentiation (A) and brite differentiation (B), was assessed by a microfluidic card TaqMan gene expression assay; n ≥4, *p<0.05, **p<0.01, ***p<0.001 vs control; n.s., not significant. White bars, control; black bars, BMP7; horizontally hatched bar, FNDC5; crossed bar, irisin. Data are presented as mean values ± SEM.