Production of r-irisin.

Production of r-irisin in yeast was performed as described previously (46, 49).

Harvest of scWAT and perirenal BAT.

Human scWAT from breast fat of 28 female donors ages 17–73 yr and perirenal BAT from four donors (3 males and 1 female) were obtained from surgical specimens through the University of Florida, Department of Pathology Grossing Room, according to a protocol approved by the Institutional Review Board. All scWAT donors were females undergoing breast reduction for cosmetic reasons, and all perirenal fat donors were undergoing nephrectomy due to malignancy (renal cell carcinoma). Fat tissues were processed within 6 h after surgery. Adipose tissue (2–5 g/donor) was dissected from surrounding fibrous-vascular tissue and cut into 2- to 3-mm pieces. Adipose tissue fragments were used to obtain preadipocytes from mature adipocytes by ceiling culture or placed directly in flasks for tissue culture. Since a variable amount of adipose tissue was obtained from donors, we triaged our experiments based on sample availability, aiming at five to six donors per group.

Isolation and expansion of human primary preadipocytes.

Isolation and expansion of preadipocytes from human scWAT were performed as described with minor modifications (Fig. 1) (19). Briefly, fresh fat tissue fragments were cultured in T25 flasks filled with basic medium (BM; DMEM-F-12 medium + 10% fetal bovine serum, 100 U/ml penicillin-streptomycin). Fat tissue floated to the top surface of the flask (“ceiling” cultures), and mature adipocytes were attached to the top of the flasks. Preadipocytes dedifferentiated from mature adipocytes were split at ∼70% confluence at a 1:4 ratio for three to four passages to generate sufficient cells for our experiments, which usually took 2–4 mo.

Differentiation of human preadipocytes into mature adipocytes and osteoblasts.

Preadipocytes derived from mature adipocytes are known to possess mesenchymal stem cell (MSC) properties and exhibit multilineage differentiation capacity (20). Human preadipocytes were cultured in six-well plates until confluent and then induced toward adipogenic differentiation in differentiation medium (DM; BM + 0.5 mmol/l isobutylmethylxanthine, 0.25 μmol/l dexamethasone, and 10 μmol/l insulin) for 14 days, with fresh medium added every 3rd day. After 14 days, cells were treated with irisin or vehicle for 4 more days to examine whether irisin can induce browning in human mature adipocytes. For osteogenic differentiation, the DM was composed of BM + 0.1 μmol/l dexamethasone, 0.2 mmol/l ascorbic acid, and 10 mmol/l β-glycerophosphate. Irisin was included in the DM for the entire differentiation period to demonstrate the effect of irisin in osteogenesis. Adipogenic and osteogenic differentiation was confirmed by Oil Red O and alizarin red staining (34).

Mitochondrial bioenergetics analysis.

Preadipocytes were differentiated into mature adipocytes in adipogenic DM for 14 days and then seeded into XF96 Microplates (15,000/well) for 3–4 days in the absence/presence of various irisin concentrations, rosiglitazone (Sigma-Aldrich), or CL316243 (Sigma-Aldrich). Metabolic analyses were performed using a Seahorse Bioscience XF96 Analyzer, which enables real-time simultaneous measurement of oxygen consumption rate (OCR) and extracellular acidification rates (ECAR). Following basal respiration, the mitochondrial effectors (oligomycin, FCCP, and rotenone) were injected sequentially according to our previously established protocols (46).

Western blotting.

Anti-phosphroylated (p)-ERK1/2 (no. 9101), anti-p-p38 MAPK (no. 9211), and anti-p-STAT3 (no. 9145) antibodies (Cell Signaling Technology), anti-β-actin (A5316; Sigma-Aldrich), and rabbit anti-UCP1 (AB155117; Abcam) were used for Western blotting. Immunoreactive bands were quantified by densitometry (6, 13).

RNA isolation and quantitative real-time PCR.

Quantitative RT-PCR was performed with β-actin RNA as an internal control by the 2^−ΔΔC_T method (46, 49). Primer sequences are available upon request.

Immunocytochemistry and immunofluorescence.

Adipocytes were fixed with 4% parformaldehyde for 10 min and blocked in the presence of hydrogen peroxide. Cells were incubated with anti-UCP1 antibody (1:500) overnight at 4°C. After washing, the cells were incubated with horseradish peroxidase-conjugated [for immunocytochemistry (ICC)] or Alexa fluor 555-conjugated [for immunofluorescence (IF)] secondary antibody (1-h, 22°C). Nuclei were stained by hematoxylin (ICC) or 4,6-diamidino-2-phenylindole (DAPI; IF).

Adipose tissue culture and browning factors treatment.

Fresh human scWAT fragments were cultured overnight in 12-well plates in BM to restore quiescence. For short-term signaling, scWAT fragments were bathed in BM containing either irisin (50 nmol/l), vehicle, or rosiglitazone (1 μmol/l) or CL316243 (1 μmol/l) for various times with agitation. For browning, scWAT fragments were bathed in BM containing irisin (0.5, 5, or 50 nmol/l) for various times (2, 3, or 4 days) with agitation. Total proteins and RNAs were collected for Western blotting of UCP1 and qRT-PCR gene expression studies.

Statistical analysis.

Results are presented as means ± SE of at least three independent experiments. Each experiment was conducted in triplicate. Statistical significance among multiple groups was analyzed using Prism 5 software by one-way ANOVA, followed by post hoc analyses, and between two groups by Student's t-test. P < 0.05 was considered significant.

Irisin has browning effects on human mature adipocytes derived from scWAT.

Human white adipocytes obtained by “ceiling” culture after 3-day culture exhibited mature adipocyte features, including cytoplasmic lipid droplets and dedifferentiated into spindle-shaped MSC-like preadipocytes. The preadipocytes were expanded and redifferentiated into mature adipocytes following 14-day culture in adipogenic DM. Most cells differentiated into mature adipocytes, as evidenced by cytoplasmic lipid droplets (phase) and Oil Red O-stained lipid droplets (Fig. 1, A–C).

We next examined irisin's browning effect on mature primary adipocytes differentiated from preadipocytes of six donors' scWAT (donors 1–6; Fig. 1D). After adipogenic differentiation, adipocytes treated with irisin (50 nmol/l) for 4 days showed upregulation of browning-related genes (UCP1, PGC1A, and PRDM16) but exhibited individual differences. In contrast, the expression of general adipose genes (PPARG and ADIPOQ) showed no difference between the control and irisin groups (not shown). Next, we analyzed UCP1 protein expression after irisin treatment by ICC (Fig. 1E) and IF (Fig. 1F). UCP1 expression was enhanced and the percentage of UCP1-positive cells increased (57 vs. 12% in control group), strongly suggesting that irisin induced browning of mature adipocytes, consistent with the observations of Lee et al. (26).

Irisin increased thermogenesis of human mature adipocytes.

To examine effects of irisin on thermogenesis in mature adipocytes isolated from scWAT, we measured mitochondrial respiration. Figure 2 shows that irisin increased OCR (Fig. 2, A and B) and ECAR (Fig. 2, C and D) at effective concentrations as low as 5 nmol/l. Mitochondrial respiration (OCR) of irisin-treated adipocytes was different from that of untreated controls or low dose of irisin-treated cells. In contrast to 0.5 nmol/l irisin, adipocytes treated with 50 nmol/l irisin showed increased basal, uncoupled (oligomycin), and maximal (FCCP) mitochondrial respiration, whereas 5 nmol/l irisin increased basal and uncoupled cellular respiration but not maximal mitochondrial respiration. Adipocytes treated with irisin (5 or 50 nmol/l) also increased glycolysis (ECAR). These data demonstrated that irisin increased cellular thermogenesis by upregulating UCP1, uncoupling oxidative phosphorylation from ATP synthesis, resulting in energy dissipation in the form of heat (15).

Irisin upregulated UCP1 expression in scWAT by activating ERK/p38 MAPK signaling.

The potentially beneficial effects of irisin-mediated browning on human adipocytes led us to probe its potential therapeutic significance. Although assessing irisin's effects directly in human subjects was impractical, fresh human scWAT fragments are suitable for modeling irisin's functions. Our previous murine studies show that irisin activates p38 MAPK and ERK signaling (49). As shown in Fig. 3, A and B, phosphorylation of p38 MAPK (p-p38) and ERK (p-ERK) was enhanced by irisin treatment for 60 min. In five donors (donors 7–11), irisin increased p-ERK 2.37 ± 0.41-fold at 60 min and 2.74 ± 0.25-fold at 90 min. p-p38 Levels increased similarly, whereas p-STAT3 was unchanged. To verify the involvement of p38/ERK signaling in UCP1 expression, scWAT fragments were treated with p-p38 (SB-203580) or p-ERK (U-0126) inhibitor for 30 min, followed by irisin. Both abolished irisin-induced upregulation of UCP1 protein (Fig. 3C), suggesting that activation of p38/ERK MAPK signaling is required for UCP1 expression.

In human adults, adipocytes derived from perirenal adipose tissue were smaller and expressed higher UCP1 compared with scWAT, suggesting that perirenal fat in human adults acts as brown adipose tissue (27, 29, 41). We also examined whether irisin regulates human perirenal fat, which is classified as BAT. Figure 3D shows that the ERK and p38 were already phosphorylated in untreated perirenal fat and that irisin did not enhance further phosphorylation. Further irisin treatment did not affect UCP1 translation (n = 4; Fig. 3, E and F) or UCP1 and PRDM16 transcription (n = 3; Fig. 3G). Thus, our study suggests that human BAT from perirenal fat has high levels of UCP1 protein, and its regulation appears to be irisin independent.

Next, we explored irisin's browning effect in scWAT fragments. Treatment of scWAT with irisin for 4 days increased UCP1 protein (Fig. 4A) and brown-associated transcripts (UCP1 and PRDM16; Fig. 4B) in >75% of scWAT samples. Interestingly, two irisin nonresponders (donors 13 and 17) had relative higher levels of basal UCP1 protein than the irisin responders, suggesting that these fat tissues might be blended with BAT. Interestingly, irisin treatment of scWAT increased FNDC5 transcript in all donors except for donor 13, suggesting positive autoregulation of irisin, in agreement with previous studies (11, 37).

To determine the lowest effective dose and shortest duration of irisin treatment, we examined the dose response and time course. As shown in Fig. 4C, irisin increased UCP1 protein (1- and 3-fold at 5 and 50 nmol/l, respectively, P < 0.01). Surprisingly, as early as day 2 of irisin treatment, UCP1 was increased (P < 0.01; Fig. 4D), suggesting that irisin plays an early role in UCP1 expression.

Correlation of basal levels of beige gene expression with irisin-induced browning.

Responses to irisin can be fat depot specific (5, 26, 45). Given these results and our finding that UCP1 protein levels varied among donors after irisin treatment (Fig. 4A), we hypothesized that the genetic background may affect irisin responsiveness. We measured transcription levels of UCP1 and PRDM16 in eight donors' scWAT after irisin treatment (Fig. 4B). Responsiveness varied considerably, from no response (donors 13 and 17) to intermediate (donors 12 and 15 for UCP1, 2- to 4-fold increase) to high responsiveness (donors 10, 11, 14, and 16, >10-fold increase). Next, we determined UCP1, TMEM26, PRDM16, CD137, and FNDC5 expression in scWAT from seven donors prior to irisin treatment (scWAT from donor 10 was insufficient). CD137 and TMEM26 are characteristically expressed by beige cells (45). scWAT showed highly variable basal levels of all five genes (Fig. 5A), suggesting that the abundance of beige adipocytes in scWAT differs among individuals and that the relative expression changes of UCP1 and PRDM16 in irisin-treated scWAT were positively correlated with basal levels of UCP1, TMEM26, PRDM16, CD137, and FNDC5 (Fig. 5B). In contrast, the basal levels of CEBPB and ADIPOQ were similar (not shown). There was no correlation between irisin responsiveness in scWAT and donors' BMI in our limited samples.

Irisin suppressed adipogenic differentiation and promoted osteogenic differentiation.

To further explore irisin's contradictory browning effects in human cell models (23, 26, 35), we examined the longitudinal effects irisin treatment during adipogenic or osteogenic differentiation in six donors (donors 10 and 18–22; Fig. 6A). After 18 days, mature adipocytes (by morphology and Oil Red O staining) in the irisin-treated group were reduced by 20–60% compared with controls (Fig. 6B). Furthermore, irisin decreased CEBPB and ADIPOQ expression (Fig. 6C). CEBPB and ADIPOQ regulate adipocyte differentiation (25, 44, 48), and suppression of their expression suggests that irisin exerts an inhibitory effect on adipogenesis. This conclusion was further supported by the fact that irisin decreased expression of browning-related UCP1 and PRDM16 by 10–65% (Fig. 6C). Consistent with the transcription results, UCP1 protein expression was reduced when irisin was included throughout adipogenic differentiation (Fig. 6D, bottom). In contrast, its level increased dramatically when irisin was included only in the last 4 days (Fig. 6D, middle). Thus, irisin not only induces browning of mature human adipocytes but also inhibits adipogenic differentiation of preadipocytes. Our data suggest that irisin's browning effect is observed only after formation of mature adipocytes, which may explain, at least partly, the conflicting reports of irisin's effects on human adipocytes.

In addition, we examined irisin's effect on osteogenesis in humans since exercise is known to prevent bone loss and reduce fracture risks. Exposure to irisin (50 nmol/l) during differentiation pushed preadipocytes toward osteoblastic differentiation, as evidenced by mineral deposition highlighted by Alizarin Red staining (Fig. 6E), and enhanced the expression of osteogenic genes RUNX2, OSTERIX, and OSTEOPONTIN in five of six donors (Fig. 6F). Thus, irisin promotes human osteoblastic differentiation, consistent with recent published murine data (10, 11).

Comparison of “browning” effect among browning factors.

PPARG ligand agonists such as rosiglitazone and β₃-adrenergic receptor agonists such as CL316243 can induce a beige fat gene program in scWAT (4). To compare the browning capacity of irisin with traditional browning factors, we tested rosiglitazone and CL-316243. We treated human scWAT fragments from three different donors with 50 nmol/l irisin, 1 μmol/l rosiglitazone, or 1 μmol/l CL316243 for 3 days. As shown in Fig. 7, A and B, top, both UCP1 protein and gene expression were increased significantly after all three treatments, and irisin seemed more potent on a molar basis than rosiglitazone and CL-316243. In addition, the cotranscriptional regulator PGC1A induces mitochondrial biogenesis by activating nuclear respiratory factors (NRF1 and -2) (32). We found that not only rosiglitazone and CL-316243 but also irisin stimulated PGC1A and NRF1 expression (Fig. 7B, middle and bottom), suggesting that irisin also has a positive effect on mitochondrial biogenesis and cellular metabolism.

Next, we compared the thermogenic effect of rosiglitazone, CL-316243, and irisin on human mature adipocytes isolated from scWAT. Figure 7, C–E, shows that both OCR and ECAR at basal and peak levels increased significantly following treatment with irisin, rosiglitazone, or CL-316243. The browning effect of irisin is comparable with rosiglitazone and CL-316243.