FGF21 Improves the Adipocyte Dysfunction Related to Seipin Deficiency

Bscl2^−/− mice were generated as previously described (14). LY2405319 (kindly provided by Lilly) was delivered by osmotic pumps (model 1004; Alzet Inc.). Mice (n = 8–11 per group) were killed after a 6-h fasting period. Insulin tolerance tests were performed as previously described (14). Adiponectin plasma levels were measured using ELISA (Crystal Chem, Downers Grove, IL).

Seipin knockdown was achieved by infecting 3T3-L1 cells with SMARTvector inducible lentiviral short hairpin RNA (Dharmacon, St. Leon-Rot, Germany). A level of 2 μg/mL puromycin selection was maintained throughout the experiments. Adipocyte differentiation protocol was followed (12), but on day 4, doxycycline (5 μg/mL) was added in the indicated wells. Treatment was performed using LY2405319 (500 nmol/L) added at day 6.

For Western blot analyses and immunolabeling, we used phosphorylated (p) P38, p-38 MAPK antibodies, and cleaved caspase-3 (Cell Signaling). Caspase-3 activity was evaluated using the N-Acetyl-Asp-Glu-Val-Asp-7-amido-4-methylcoumarin (Ac-DEVD-AMC) caspase-3 substrate that fluoresces following cleavage. RNA expression was analyzed as previously described (14).

All data are reported as mean ± SEM. Data sets were analyzed for statistical significance using the nonparametric Mann-Whitney U test, Kruskal-Wallis, or two-way ANOVA analysis.

Bscl2^−/− mice were previously shown to display severe lipodystrophy at 3 months of age. Studying these mice at a younger age, we found that WAT loss was progressive. Plasma adiponectin levels decreased by 10-fold between 4 and 14 weeks of age in Bscl2^−/− mice, concomitantly with the worsening of random-fed hyperglycemia (Fig. 1A and B). A 50% decrease in inguinal fat pad weight and a complete disappearance of the gonadal fat pad were observed in Bscl2^−/− mice between 4 and 12 weeks of age (Fig. 1C). Gene expression of several mature adipocyte markers (PPARγ, Adipoq, aP2, Hsl, and Plin1) were significantly decreased in inguinal WAT from 12-week-old compared with 4-week-old Bscl2^−/− mice (Fig. 1D). Transmission electron microscopy imaging of WAT from 4-week-old Bscl2^−/− mice revealed the presence of autophagic vacuoles and extended ER tubular network (Fig. 1E). Finally, ex vivo caspase-3 activity (Fig. 1F) and immunohistochemistry with cleaved caspase-3 staining (Fig. 1G) indicated a high level of apoptosis in adipocytes. These observations suggest that seipin-deficient WAT is undergoing cellular stress, compromising its functionality.

Fgf21 mRNA levels were strongly increased in 4-week-old Bscl2^−/− mice compared with wild-type mice but were decreased over time in 12-week-old Bscl2^−/− mice (Fig. 1H). Pioglitazone treatment, which improves both WAT gene expression pattern and metabolic profile in Bscl2^−/− mice (14), increased Fgf21 mRNA levels in Bscl2^−/− WAT (Fig. 1I). On the basis of these results, we hypothesized that FGF21 might counteract the adipocyte dysfunction related to seipin deficiency.

We treated 6-week-old Bscl2^−/− and Bscl2^+/+ mice with the FGF21 analog LY2405319 for a period of 4 weeks. LY2405319 reduced the random-fed glucose levels in Bscl2^+/+ mice and corrected the hyperglycemia in Bscl2^−/− mice (Fig. 2A). LY2405319 improved insulin sensitivity (Fig. 2B) and induced a 2.5-fold increase in plasma adiponectin levels (Fig. 2C) in Bscl2^−/− mice (1,562 ± 164 mg/mL vs. 636 ± 48 mg/mL). LY2405319 increased the mRNA expression of mature adipocyte markers in the inguinal WAT from Bscl2^−/− mice, including aP2, Pparγ, Hsl, Atgl, Lep, and Adipoq (Fig. 2F), without a significant change in adipose tissue mass (Fig. 2D and E). In addition, LY2405319 increased Ucp1 and Pgc1a mRNA levels (two- and threefold, respectively) in Bscl2^−/− WAT but not in BAT (Supplementary Fig. 1A). LY2405319 did not alter hepatic triglyceride content or the catabolic gene expression pattern and had only a mild lowering effect on the mRNA levels of the lipogenic genes Scd1 and Chrebp in the liver (Supplementary Fig. 1C–E).

We next treated older Bscl2^−/− mice that already displayed metabolic abnormalities. FGF21-treated mice maintained their initial adiponectin levels, whereas nontreated animals displayed a 2.5-fold decrease of plasma adiponectin levels (Fig. 2G). In older Bscl2^−/− mice, the effect of FGF21 on random-fed glycemia was milder and the improvement of insulin sensitivity failed to achieve statistical significance (Supplementary Fig. 2A and B).

Altogether these results demonstrate that FGF21 treatment is able to prevent the WAT dysfunction associated with seipin deficiency, i.e., to increase adiponectin secretion and to improve the metabolic profile of young Bscl2^−/− mice.

To assess whether FGF21 might directly correct the adipocyte dysfunction due to seipin deficiency, we generated a cell line derived from 3T3-L1 preadipocytes with conditional Bscl2 knockdown (SKD). Because seipin is required for full adipocyte differentiation (11–14), this cell model allowed us to study the effect of seipin deficiency in mature adipocytes. Using Oil Red O staining, we observed that at day 8 the differentiation levels (∼90%) were identical in all conditions (Supplementary Fig. 3A and B). Adipoq and aP2 expression levels were similar in all conditions, whereas doxycycline-treated SKD displayed an 80% reduction in Bscl2 mRNA levels, validating our mature adipocyte seipin-deficient model (Fig. 3A). At day 21, SKD doxycycline-treated cells displayed a strong reduction in Adipoq and aP2 mRNA levels (Fig. 3B) and a decrease in Oil Red O–positive cells (Fig. 3C), suggesting a progressive loss of mature adipocytes. We then investigated some other stress-activated signaling pathways that may potentially be involved in this apoptotic response. We noticed that seipin deficiency did not alter the phosphorylation levels of JNK and EIF2A (ER stress key player) (data not shown). However, doxycycline-treated SKD cells displayed a threefold increase in p38 MAPK phosphorylation as well as a 2.5 induction in the mRNA levels of one of its target genes, the death receptor FASR (16) (Fig. 3D–F). Seipin-deficient mature adipocytes displayed a sevenfold increase in caspase-3 activity (Fig. 3G) and stained positively for both cleaved caspase-3 and bodipy (lipid droplets) (Supplementary Fig. 3C). It is noteworthy that treatment with the p38 MAPK inhibitor SB203580 decreased by 35% the induction of Fasr mRNA expression and decreased by 20% the caspase-3 activity in seipin-deficient cells (Fig. 3H and I).

We tested whether FGF21 could directly prevent the loss of mature seipin-deficient adipocytes. Oil Red O staining at day 21 of differentiation revealed that LY2405319 treatment (from day 6) partially prevented adipocyte loss (Fig. 4A and B). LY2405319 maintained Adipoq and aP2 mRNA expression levels (Fig. 4C and D, respectively). In addition, LY2405319 was able to reduce p38 activation, caspase-3 activity, and Fasr mRNA levels in doxycycline-treated SKD adipocytes (Fig. 4E–G). Finally, we demonstrated that LY2405319 was able to reduce Fasr mRNA levels in vivo in WAT from Bscl2^−/− mice (Fig. 4H).

We assessed whether treatment with FGF21 analog was able to correct the metabolic abnormalities associated with seipin deficiency (i.e., BSCL2), the most severe form of lipodystrophy. We showed that FGF21 prevents the progressive dysfunction of WAT in Bscl2^−/− mice, potentiating an improvement in insulin sensitivity. Those effects appear to be mainly mediated by the WAT. By using a unique cellular model of seipin knockdown in differentiated adipocytes, we demonstrated that seipin is critical for the maintenance of mature adipocytes. Indeed, seipin deficiency is associated with cellular stress activation and apoptosis that can be partly prevented by FGF21 treatment.

Whether lipodystrophy is due exclusively to an alteration in the development of adipose tissue or to its rapid disappearance remains unknown. MRI exploration in BSCL patients sometimes shows residual WAT (17). To date, seipin deficiency has been mainly shown to impair adipogenesis and to increase unstimulated basal lipolysis (13,14,18,19). In this report, we show for the first time that there is a loss of adipose tissue in Bscl2^−/− mice between 4 and 12 weeks of age, meaning that the adipogenesis impairment cannot alone explain the lipodystrophic phenotype previously described in adult mice. In vitro, inducible seipin knockdown in fully mature adipocytes increases p38 MAPK phosphorylation and caspase-3 activity, both indicative of a substantial loss of mature adipocytes. The double staining of cleaved caspase-3 and bodipy demonstrates that apoptosis is taking place in the mature adipocyte. Chronic treatment with a low dose of the p38 inhibitor SB203580 significantly reduces the Fasr mRNA induction and the caspase-3 activity, supporting the notion that p38 MAPK activation contributes at least partly to the induction of apoptosis observed in doxycycline-treated SKD cells. However, the effects of SB203580 failed to fully restore the phenotype of SKD cells. This could be explained by the lower dose used here to avoid cellular toxicity upon a chronic treatment (i.e., 2.5 μmol/L compared with 20 µmol/L in acute studies [20]). Alternatively, this could indicate that some p38 MAPK–independent pathways contribute to the apoptosis in these cells. Our results are consistent with two recent studies reporting that in vivo seipin deficiency in mature adipocytes leads to progressive lipodystrophy (21,22). Although further studies are needed to highlight the causes of the cellular stress, it is tempting to speculate that apoptosis induction takes part in the adipocyte loss in vivo and contributes to the lipodystrophic phenotype.

The observation that FGF21 expression is transiently induced in WAT of 4-week-old Bscl2^−/− mice led us to hypothesize that FGF21 might be locally overexpressed as a compensatory response to tissue stress. Importantly, pharmacological treatment with the FGF21 analog LY2405319 is able to prevent the progressive decrease in adiponectin secretion and the deterioration of the WAT expression pattern. This might explain the unexpected increase in Fgf21 expression levels in Bscl2^−/− mice only. Indeed, in Bscl2^−/− mice from 4 to 12 weeks of age, there was a threefold decrease in Fgf21 mRNA levels along with the degradation of the WAT. LY2405319 prevented this degradation and might therefore maintain the Fgf21 mRNA levels at their steady state, resulting in a 2.5-fold increase in 12-week-old Bscl2^−/− mice treated with LY2405319 compared with untreated ones. When we treated 6-week-old Bscl2^−/− mice that display only a threefold decrease in adiponectin levels with LY2405319, we were able to improve insulin sensitivity and hyperglycemia. Previous reports have shown that the metabolic effects of FGF21 are mediated by an increase in adiponectin production, as the metabolic efficacy of FGF21 is abolished in adiponectin-deficient mice (6,7). Notably, when we treated older mice (i.e., 8 weeks old) that display a 10-fold decrease in adiponectin circulating levels, FGF21 treatment still prevented a further decrease in plasma adiponectin concentrations but failed to improve insulin sensitivity. Previous results showed that FGF21 treatment does not improve glucose tolerance in lipodystrophic aP2–nuclear SREBP1c mice. In the same study, WAT transplantation restored the beneficial effects of FGF21 (23). In our study, WAT appears to be the main target of FGF21 as there is no major effect on the BAT or the liver. Regarding the lack of effect on BAT, this is consistent with two recent reports showing that FGF21 similarly improves glucose homeostasis in wild-type and UCP1 KO mice (24,25). Altogether, these data support the hypothesis that WAT is the key mediator of the beneficial effect of FGF21 and that a minimum amount of functional WAT is required to enable its effect.

Finally, we demonstrate that in vitro FGF21 treatment is able to maintain adiponectin mRNA levels, to reduce p38 MAPK chronic activation, and to limit the mature adipocyte loss. As the strongest effect of FGF21 treatment is on the induction of Adipoq mRNA, we may assume that adiponectin exerts an anticellular stress autocrine effect, as previously suggested (26,27). Further works are required to demonstrate the direct effect of adiponectin on p38 activation and apoptosis induction. Nevertheless, these in vitro results are consistent with those in vivo that place adiponectin at the center of the beneficial effect of FGF21 in our lipodystrophic model. As LY2405319 treatment lowers FAS-R expression in Bscl2^−/− WAT, it is tempting to speculate that FGF21 limits the apoptosis induction in vivo.

To our knowledge, this is the first report suggesting that a FGF21 analog might exert its protective effect in WAT through an anticellular stress effect on mature adipocytes. Our study is important not only for the metabolic complications associated with BSCL2 but also for the adipocyte dysfunction in the context of obesity and metabolic syndrome. Notably, from a clinical perspective regarding lipodystrophic patients, we might speculate that the FGF21 analog represents a novel therapeutic approach if used at an early stage of the disease or in patients with partial lipodystrophy.

Acknowledgments. The authors thank M. Takahashi, J.-F. Deleuze, and M. Lathrop (Centre d'Energie Atomique, Centre de Génotypage, Evry, France) for donating the mouse model and C. Dumet, S. Suzanne, and S. Lemarchand-Minde (Animal Facility, l'Institut du Thorax, Nantes, France) for animal care. The authors thank Lisa Oliver for her precious advice for the caspase-3 assay. They also thank the MicroPICell core facility (Structure Fédérative de Recherche François Bonamy, Nantes, France) for microscopy and Guillaume Mabilleau for helpful technical assistance for transmission electron microscopy performed at SCIAM (Service Commun Imagerie et d’Analyse Microscopique), Université d’Angers.

Funding. This work was supported by grants from the INSERM and the French associations Association pour la Recherche sur le Diabète, Société Francophone du Diabète, and Fondation de France. L.D. is supported by a PhD fellowship from Ministère de la Recherche et de la Technologie. C.L. is funded by a PhD fellowship from the region Pays de la Loire and INSERM Grand Ouest. X.P. was awarded the European Foundation for the Study of Diabetes/Lilly Research Fellowship Programme 2015. LY2405319 was provided by Lilly.

Duality of Interest. T.C., A.C.A., and R.E.G. are Lilly employees involved in LY2405319 development. B.C. has received research funding from Sanofi; has received honoraria from Amgen, AstraZeneca, Pierre Fabre, Janssen, Eli Lilly, Merck Sharp & Dohme, Novo Nordisk, Sanofi, and Takeda; and has acted as a consultant/advisory panel member for Amgen, Eli Lilly, Merck Sharp & Dohme, Novo Nordisk, Sanofi, and Regeneron. No other potential conflicts of interest relevant to this article were reported.

Author Contributions. L.D. researched data and wrote the manuscript. C.L., S.L.L., A.A., and Q.V. researched data. T.C., A.C.A., and R.E.G. contributed to the scientific discussion and edited the manuscript. C.L.M. contributed to scientific discussion. J.M. wrote the manuscript. B.C. contributed to scientific design and wrote the manuscript. X.P. researched data, designed the experiments, and wrote the manuscript. X.P. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.