AbstractObjective

Four-and-a-Half-LIM-domain-protein 2 (FHL2) modulates multiple signal transduction pathways but has not been implicated in obesity or energy metabolism. In humans, methylation and expression of the FHL2 gene increases with age, and high FHL2 expression is associated with increased body weight in humans and mice. This led us to hypothesize that FHL2 is a determinant of diet-induced obesity.

Methods

FHL2-deficient (FHL2−/−) and wild type male mice were fed a high-fat diet. Metabolic phenotyping of these mice, as well as transcriptional analysis of key metabolic tissues was performed. Correlation of the expression of FHL2 and relevant genes was assessed in datasets from white adipose tissue of individuals with and without obesity.

Results

FHL2 Deficiency protects mice from high-fat diet-induced weight gain, whereas glucose handling is normal. We observed enhanced energy expenditure, which may be explained by a combination of changes in multiple tissues; mild activation of brown adipose tissue with increased fatty acid uptake, increased cardiac glucose uptake and browning of white adipose tissue. Corroborating our findings in mice, expression of FHL2 in human white adipose tissue positively correlates with obesity and negatively with expression of browning-associated genes.

Conclusion

Our results position FHL2 as a novel regulator of obesity and energy expenditure in mice and human. Given that FHL2 expression increases during aging, we now show that low FHL2 expression associates with a healthy metabolic state.

1. IntroductionThe worldwide prevalence of obesity is reaching epidemic levels and represents a major health issue. The principal drivers in the rise of obesity are a sedentary lifestyle and an increased high-caloric food consumption, which are subject to lifestyle interventions. Obesity increases the risk for developing, among others, Type 2 Diabetes (T2D), cardiovascular diseases and some types of cancer. However, these pathologies do not always develop in subjects with obesity, because other genetic and environmental factors contribute substantially to their etiology. Many genes have been correlated to body-mass index, a proxy for obesity, but only a small proportion of these genes has been directly positioned within specific biological pathways [[1]Locke A.E. Kahali B. Berndt S.I. Justice A.E. Pers T.H. Day F.R. et al.Genetic studies of body mass index yield new insights for obesity biology.].In addition to genetic factors, current evidence supports the notion that epigenetics, and specifically DNA methylation, play a role in obesity and associated comorbidities [[2]De Toro-Martín J. Guénard F. Tchernof A. Hould F.S. Lebel S. Julien F. et al.Body mass index is associated with epigenetic age acceleration in the visceral adipose tissue of subjects with severe obesity.]. Hypermethylation of certain DNA regions is characteristic of the aging process, and it is well-recognized that obesity is highly prevalent among the elderly. Moreover, it has been hypothesized that obesity can promote epigenetic aging in metabolic tissues, something that so far has been proven in the liver [[3]Obesity and related consequences to ageing.,[4]Horvath S. Erhart W. Brosch M. Ammerpohl O. Von Schönfels W. Ahrens M. et al.Obesity accelerates epigenetic aging of human liver.]. In humans, consistent age-related DNA hypermethylation of the LIM-only-protein-Four-and-a-Half LIM domain protein 2 (FHL2) gene is observed in blood cells, white adipose tissue (WAT), liver and pancreatic islets [5Rönn T. Volkov P. Gillberg L. Kokosar M. Perfilyev A. Jacobsen A.L. et al.Impact of age, BMI and HbA1c levels on the genome-wide DNA methylation and mRNA expression patterns in human adipose tissue and identification of epigenetic biomarkers in blood., 6Bacos K. Gillberg L. Volkov P. Olsson A.H. Hansen T. Pedersen O. et al.Blood-based biomarkers of age-associated epigenetic changes in human islets associate with insulin secretion and diabetes., 7Bysani M. Perfilyev A. De Mello V.D. Rönn T. Nilsson E. Pihlajamäki J. et al.Epigenetic alterations in blood mirror age-associated DNA methylation and gene expression changes in human liver.]. Notably, increased methylation in the FHL2 locus itself results in increased FHL2 expression [[6]Bacos K. Gillberg L. Volkov P. Olsson A.H. Hansen T. Pedersen O. et al.Blood-based biomarkers of age-associated epigenetic changes in human islets associate with insulin secretion and diabetes.].FHL2 is a protein that integrates signaling pathways to regulate transcriptional responses through its association with an array of target proteins in a signal- and cell-dependent manner [[8]Tran M.K. Kurakula K. Koenis D.S. de Vries C.J.M. Protein-protein interactions of the LIM-only protein FHL2 and functional implication of the interactions relevant in cardiovascular disease.]. FHL2 is expressed in several organs and cell types throughout the body. Its expression is highest in the heart, which prompted investigation of its role in cardiac function [[9]Kong Y. Shelton J.M. Rothermel B. Li X. Richardson J.A. Bassel-Duby R. et al.Cardiac-specific LIM protein FHL2 modifies the hypertrophic response to β-adrenergic stimulation.]. Whole-body FHL2-deficient mice (FHL2−/−) are viable and have normal cardiac development and function. However, these mice show an exaggerated response upon induction of cardiac hypertrophy as compared to wild type mice. Whether FHL2 plays a role in energy metabolism is largely unknown. However, anecdotal reports implicate a role for FHL2 in fatty acid and cholesterol metabolism [[10]Ramayo-Caldas Y. Ballester M. Fortes M.R.S. Esteve-Codina A. Castelló A. Noguera J.L. et al.From SNP co-association to RNA co-expression: novel insights into gene networks for intramuscular fatty acid composition in porcine.,[11]Kurakula K. Sommer D. Sokolovic M. Moerland P.D. Scheij S. van Loenen P.B. et al.LIM-only protein FHL2 is a positive regulator of liver X receptors in smooth muscle cells involved in lipid homeostasis.]. This has prompted us to address the role of FHL2 in whole body energy metabolism and obesity.

We report here that FHL2-deficient mice are protected from high-fat diet (HFD)-induced obesity and posit that this can be attributed to increased lipid uptake by brown adipose tissue (BAT), increased uptake of glucose by the heart and a marked browning of WAT. Importantly, to substantiate the results in mice, our analysis of human cohorts revealed that FHL2 expression in WAT correlates positively with obesity, and negatively with expression of browning genes. Together, our data position the signal transduction intermediary protein FHL2 as a novel modifier of obesity.

2. Materials and methods2.1 Human data analysisThe publicly accessible human datasets used in this study were: a) GSE59034, from subcutaneous white adipose tissue from female individuals before going through bariatric surgery (n = 16) and lean controls that never had obesity (n = 16) [[12]Petrus P. Mejhert N. Corrales P. Lecoutre S. Li Q. Maldonado E. et al.Transforming growth factor-β3 regulates adipocyte number in subcutaneous white adipose tissue.], and b) GSE70353, from subcutaneous white adipose tissue of 770 male individuals (age: 45–73 years old) with varying BMI who were part of the METSIM study [[13]Stǎácaková A. Javorský M. Kuulasmaa T. Haffner S.M. Kuusisto J. Laakso M. Changes in insulin sensitivity and insulin release in relation to glycemia and glucose tolerance in 6,414 finnish men.,[14]Civelek M. Wu Y. Pan C. Raulerson C.K. Ko A. He A. et al.Genetic regulation of adipose gene expression and cardio-metabolic traits.]. These datasets were uploaded to R2: Genomics Analysis and Visualization Platform (http://r2.amc.nl) for analysis. Analysis for genes that significantly associate with FHL2 expression was performed using this software.2.2 F2 intercross mice study (BHF2) dataset analysisDataset from F2 intercross between C57BL/J6 and C3H/HeJ mice on ApoE−/− background (BHF2 population) generated at The University of California Los Angeles (UCLA) was retrieved from the publicly available GeneNetwork database (http://www.genenetwork.org). Pearson correlations were obtained from male mice adipose tissue mRNA and described phenotypes.2.3 Mouse strainsAll animal experiments were approved by an ethic committee of the Amsterdam University Medical Center, The Netherlands (permit number DBC287) and were performed in accordance with European directive 2010/63/EU guidelines. Working protocols for every experiment reported here were approved by the ethic committee of the Amsterdam University Medical Center. FHL2-deficient mice were generated by R. Bassel-Duby (University of Texas Southwestern Medical Center, Dallas, TX) and were bred onto a C57BL/6 background for more than 11 generations. In all experiments, male littermates of 8- to 20-week-old were used and n refers to the number of single animals. We did not include female mice in our study because of the knowledge that especially female mice have a bone defect upon FHL2 deficiency, which may affect body weight [[15]Govoni K.E. Baylink D.J. Chen J. Mohan S. Disruption of four-and-a-half LIM 2 decreases bone mineral content and bone mineral density in femur and tibia bones of female mice.]. Sample size for each experiment was determined by power calculation using the nQuery software (https://www.statsols.com/). Littermates from wild-type and FHL2-deficient genotype were randomly separated in ventilated cages with free access to water and food (60% fat diet Open Source diets #D12492). High-fat diet feeding was started at the age of 8–10 weeks. Mice were terminated by intraperitoneal injection of a lethal dose of ketamine (166 mg/kg) plus xylazine (24 mg/kg). Both number and suffering of animals was minimized as much as possible, such as group housing except for the metabolic cage experiments and cage enrichment. Health was monitored weekly by researchers and animal caretakers and humane endpoints were taken into account under standard procedure from the animal research facility. After termination, mouse tissues were rinsed with ice-cold PBS through trans-cardiac perfusion, harvested and stored at −80 °C for further analysis.2.4 Glucose and insulin tolerance

For the intraperitoneal glucose tolerance test (ipGTT), mice were fasted 4 h prior receiving an injection of glucose (2 g/kg body weight). For insulin tolerance tests (ITT), mice were fasted for 4 h prior injection of an intraperitoneal insulin dose (1 IU/kg body weight, Sigma Aldrich). In all experiments blood was collected from the tail vein at baseline and every 15 min for a period of 120 min. Blood glucose was measured using an automatic Stat Strip glucometer (Nova Biomedical). At indicated times, blood samples were collected in EDTA-coated capillary tubes, centrifuged and plasma samples were stored at −80 °C for further measurements.

2.5 Biochemical analyses

Mouse blood was collected from the tail vein using EDTA-coated capillary tubes after 4 h of fasting. Blood was centrifuged at room temperature and plasma was isolated and stored at −80 °C until analysis. Plasma triglyceride (TG), total cholesterol (TC), adiponectin and leptin levels were measured using: Triglyceride GPO Method Assay Kit (Biolabo), Cholesterol CHOD-PAP Method Assay kit (Biolabo), mouse adiponectin ELISA kit (Crystal Chem) and mouse leptin ELISA kit (Crystal Chem), respectively, following the manufacturer's protocols.

2.6 Absorption of dietary fat

Dietary fat absorption was measured in 4 h fasted chow- or HFD-fed mice at noon. Plasma triglycerides were measured before and after an oral bolus of olive oil (400 μl) administered by intragastric gavage; blood samples were taken at indicated time points.

2.7 Fecal energy and lipid extraction

To measure fecal energy and lipid content, 24 h feces was collected from individually housed mice to quantitatively determine food intake and feces production. Feces were weighed, freeze-dried, and grinded prior to analysis. First, in order to measure the energy excreted, feces from 3 mice was pooled in a total of 3 measurable samples. A bomb calorimeter (IKA C1) was used for combustion and a food sample was taken along as reference (Lovelady and Stork, 1970).

Fecal lipids were extracted from freeze-dried feces in a mixture of methanol and 10 M NaOH and incubated at 90 °C. Afterwards, a solution of 6 M HCl and hexane was added for lipid solubilization. After centrifugation, a second step of hexane was needed before the samples were dried under a nitrogen stream at 40 °C and dissolved in 2% Triton X-100. Triglycerides and total cholesterol were measured as described in the previous section. Non-esterified fatty acids (NEFA) were measured using HR Series NEFA-HR(2) (Wako Diagnostics).

2.8 Indirect calorimetryWild type and FHL2−/− mice were randomly and individually housed in PhenoMaster Indirect Calorimetry System (TSE Systems, Bad Homburg, Germany) for 5 days where they had access to food and water ad libitum. Mice adapted to the PhenoMaster system for 48 h before the start of data analysis. The measurements in the metabolic cages are performed automatically, so that blinding is not necessary. Allocation and measurements were performed at the same time for wild type and FHL2−/− mice. Here, we analyzed bodyweight, locomotor activity, respiratory exchange ratio (RER), O2 consumption and CO2 production, energy expenditure (EE). Rates of oxygen consumption (VO2, ml/h) and carbon dioxide production (VCO2, ml/h) were calculated by TSE software and used to calculate the RER (RER = VCO2/VO2) and EE (kcal/h). Locomotor activity was measured by infrared beams at the long side (X-frame) and at the short side (Y-frame) of the cage (expressed as total beam breaks (both X and Y) per hour). Energy Expenditure (EE) ANCOVA analysis was performed using the Energy Expenditure Analysis page (http://www.mmpc.org/shared/regression.aspx) of the NIDDK Mouse Metabolic Phenotyping Centers (MMPC).2.9 Histology

Organs were fixed in 4% paraformaldehyde (Roth), embedded in paraffin, sectioned and mounted onto StarFrost glass slides (Thermo Scientific). Sections (5 μm thickness) were deparaffinized and rehydrated. Tissue morphology was assessed by hematoxylin and eosin staining (H&E) (Sigma). Sections were visualized with a Leica DM6 microscope and quantified using Leica LAS-X Software. Sectioning and staining was performed by a researcher and lipid content quantification of histological sections was performed blindly by another researcher.

2.10 In vivo triglyceride and glucose clearance

TG-rich lipoprotein (TRL)-like particles (80 nm), radiolabeled with glycerol tri[3H]oleate (3.7 MBq) were prepared as described before (Li et al., 2020), and stored at 4 °C under argon until use at the second day after preparation. TRL-like particles were mixed 2-[1-14C]deoxy-d-glucose ([14C]DG) in a 4:1 ratio (3H:14C). Mice at 12 weeks old (4 weeks of HFD) were injected via the tail vein with the combination of TRL-like particles (1 mg TG) and deoxyglucose (200 μL/mouse). After 15 min, mice were killed by CO2 inhalation, transcardially perfused with ice-cold PBS, tissue was collected and a piece was dissolved in 500 μl of Solvable (Perkin Elmer) overnight at 56 °C. The tissue uptake of 3H and 14C was determined using scintillation counting (Ultima Gold XR, Perkin Elmer).

2.11 RNA extraction and reverse transcription (RT)-qPCR

Total RNA was isolated using Trizol reagent (Invitrogen) according to the manufacturer's protocol. cDNA was synthesized using iScript cDNA synthesis kit (BioRad). Quantitative PCR was carried out using SensiFAST SYBR No-ROX Kit (Bioline) on a LightCycler 480 II PCR platform (Roche). Cycle quantification and primer set amplification efficiency were calculated using the LinRegPCR software package (Ruijter et al., 2009). Target gene expression was normalized by dividing the geometric mean of the gene expression of Rplp0 and β-actin. Primer sequences are listed in Table S1.

2.12 Western blotting

Frozen tissue samples were powdered in liquid nitrogen. Proteins were isolated using RIPA buffer (150 mM NaCl, 50 mM Tris pH 7.4, 1% Nonidet-P40, 0.5% sodiumdeoxycholate, 0.1% SDS and Roche cOmplete™ protease inhibitor cocktail) and quantified using the DC protein assay (BioRad). Equal amounts of protein lysate were loaded on to 12% SDS-PAGE gels along with the protein ladder standard (Precision Plus Protein™ All Blue Prestained Protein Standards from Bio Rad) and transferred to nitrocellulose membranes. Membranes were blocked with 5–10% non-fat milk for 1 h and subsequently incubated with primary antibody (Rabbit anti-UCP1; Abcam #ab10983 and Mouse anti-alpha-tubulin; Cedarlane #CLT9002 as loading control) at 4 °C overnight. Membranes were washed and incubated with appropriate HRP-conjugated secondary antibodies for 1 h at room temperature and protein bands were visualized using Supersignal West Pico PLUS Chemiluminescent Substrate (Thermo Scientific) and ImageQuant LAS 4000 imager (GE Healthcare). Quantification of band intensity was performed with reference to loading control (alpha-tubulin) using ImageJ Gel Analysis program.

2.13 Differentiation of pre-adipocytes from the stromal vascular fraction of white adipose tissue

Pre-adipocytes from the stromal vascular fraction (SVF) of white adipose tissue (WAT) were isolated from WT and FHL2−/− mice white. Fat pads were minced in 1.5 mg/ml collagenase solution (Sigma-Aldrich; #C6885) and homogenates were digested for 60 min at 37 °C on a shaking platform. After digestion, homogenates were filtered through a 100 μm strainer and centrifuged at 1600 rpm for 10 min. Cell pellets were resuspended in red blood cell lysis buffer (Roche; # 11814389001) and neutralized with DMEM supplemented with 10% FBS and 1% penicillin-streptomycin. Cells were again centrifuged at 1200 rpm for 5 min and seeded in cell culture plates for differentiation. When cells reached full confluency they were cultured for two more days before the addition of culture medium supplemented with 3-isobutyl-1-methylxanthine (500 μM), dexamethasone (1 μM), insulin (170 nM) and rosiglitazone (1 μM) to induce differentiation. After this, every two days medium supplemented only with insulin was added and left until day 8 of differentiation where RNA was isolated and Oil Red O staining was performed.

2.14 RNA sequencing (RNA-seq)Heart and gonadal (g)WAT of 4 HFD wild-type and 4 HFD FHL2−/− mice (n = 3 in case of gWAT) were used for mRNA isolation using Trizol reagent (Invitrogen) according to the manufacturer's protocol. RNA concentration was measured using the Nanodrop ND-1000 spectrophotometer (Nanodrop Technologies) and RNA quality was determined using the Eukaryote Total RNA Nano assay from Agilent 2100 Bioanalyzer (Agilent Technologies). Strand-specific libraries were generated using the Kapa mRNA Hyperprep kit (Roche). Libraries were sequenced on an Illumina HiSeq4000 in single-end 50 bp reads. Reads were subjected to quality control (FastQC, dupRadar, Picard Tools), trimmed using Trimmomatic v0.32 and aligned to the genomes using HISAT2 (v2.1.0). Counts were obtained using HTSeq (v0.11.0) using the corresponding GTFs [20Chu P.H. Yeh H.I. Wu H.H. Hong R.C. Shiu T.F. Yang C.M. Deletion of the FHL2 gene attenuates the formation of atherosclerotic lesions after a cholesterol-enriched diet., 21Hojayev B. Rothermel B.A. Gillette T.G. Hill J.A. FHL2 binds calcineurin and represses pathological cardiac growth., 22Kummitha C.M. Kalhan S.C. Saidel G.M. Lai N. Relating tissue/organ energy expenditure to metabolic fluxes in mouse and human: experimental data integrated with mathematical modeling.]. Statistical analyses were performed using the edgeR and limma/voom R packages [[23]Schulze P.C. Drosatos K. Goldberg I.J. Lipid use and misuse by the heart.,[24]Neves F.A. Cortez E. Bernardo A.F. ABM Mattos Vieira A.K. Malafaia T de O. et al.Heart energy metabolism impairment in Western-diet induced obese mice.]. Count data were transformed to log2-counts per million (logCPM), normalized by applying the trimmed mean of M-values method and precision weighted using voom. Differential expression was assessed using an empirical Bayes moderated t-test within limma's linear model framework including the precision weights estimated by voom. Resulting P values were corrected for multiple testing using the Benjamini-Hochberg false discovery rate. Genes were re-annotated using biomaRt using the Ensembl genome databases. Geneset enrichment is performed with MSigDB genesets using CAMERA approach as implemented in limma. The resulting DEGs, expression plots, geneset enrichment results are shown in an in-house made Shiny-app. Ingenuity Pathway Analysis (IPA) was used to identify the significant canonical pathways arising from the gene expression changes observed. Only genes with a differential expression of adjusted p-value GSE156027.2.15 Statistical analysis

Statistical analyses were performed using GraphPad Prism software. Data are reported as mean ± SEM. Two-tailed unpaired Student's t-test was used when comparing two groups and Two-way analysis of variance (ANOVA) was used to analyze time course experiments, with a p-value <0.05 being considered significant and levels of significance being indicated as follows: *p < 0.05; **p < 0.01; ***p < 0.001.

4. DiscussionIn this study, we demonstrate to our knowledge for the first time that FHL2 expression is higher in WAT of individuals with obesity compared to lean humans. In addition, we prove that FHL2-deficient mice have an improved metabolic phenotype in response to HFD feeding compared to WT mice. In response to HFD, FHL2−/− mice are protected against weight gain and hepatic steatosis, confirming our hypothesis that FHL2 has a substantial role in energy metabolism. We applied the whole-body FHL2-deficient mice, which may be considered a major limitation of current study. However, the overall beneficial phenotype we revealed is most likely determined by changes in multiple tissues, rather than by only one specific tissue or organ. This fits our present knowledge that FHL2 is expressed in various tissues and is a specific and often subtle modulator of several cellular signal transduction pathways. After only a few weeks of HFD the difference in weight gain between groups becomes already apparent with the body weight of FHL2−/− mice remaining lower, whereas food intake and fecal output are similar to wild-type mice. Together with the reduction in weight, hepatic steatosis is significantly reduced in the absence of FHL2. In line with the observation that silencing of FHL2 in an insulinoma cell line reduces insulin secretion [[6]Bacos K. Gillberg L. Volkov P. Olsson A.H. Hansen T. Pedersen O. et al.Blood-based biomarkers of age-associated epigenetic changes in human islets associate with insulin secretion and diabetes.], we observed that after HFD the fasting plasma insulin level is reduced in FHL2-deficient mice (Fig. 2C). However, there is no difference in glucose or insulin tolerance between WT and FHL2−/− mice and also total insulin secretion during the ipGTT is similar. Therefore, we conclude that glucose handling does not explain the difference in metabolic phenotype in response to HFD between WT and FHL2−/− mice. Indirect calorimetry showed that heat production/energy expenditure is significantly higher in case of FHL2-deficiency under HFD conditions.Given that FHL2 expression is highest in the heart [[45]Chu P.H. Ruiz-Lozano P. Zhou Q. Cai C. Chen J. Expression patterns of FHL/SLIM family members suggest important functional roles in skeletal muscle and cardiovascular system.], we decided to analyze this tissue in the setting of diet-induced obesity. The discovery of increased glucose uptake by the FHL2−/− heart although not expected in this context, is in line with our knowledge that FHL2 deficiency promotes an exaggerated response in the heart when it is challenged. This finding suggests that the diet and consequent weight gain can be considered a sufficient cardiac stressor to change the preferential substrate from fatty acid to glucose in the absence of FHL2, similar to a pathologic state. This was confirmed by transcriptomic analysis where several pathways involved in cardiac hypertrophy were listed. Analysis of possible upstream regulators based on differential gene expression from transcriptomic data, showed several activated regulators that have been described to regulate cell stress or have a role in heart failure. Among them we found MAPK1, which is an activated regulator found in FHL2−/− heart, also a known as an interactor of FHL2. It was previously shown that FHL2 represses MAPK1 activity in cardiomyocytes and this was related to the cardiac hypertrophy response, but MAPK1 is also involved in glucose diffusion and development of insulin resistance in obesity [[46]Purcell N.H. Darwis D. Bueno O.F. Müller J.M. Schüle R. Molkentin J.D. Extracellular signal-regulated kinase 2 interacts with and is negatively regulated by the LIM-only protein FHL2 in cardiomyocytes.,[47]Bengal E. Aviram S. Hayek T. P38 mapk in glucose metabolism of skeletal muscle: beneficial or harmful?.].Energy expenditure can be increased by the process of adaptive thermogenesis of BAT, but we found no structural differences, with a trend towards higher expression of the key thermogenic protein UCP1 in FHL2−/− mice and a significant preference for the uptake from the circulation of fatty acids over glucose. It is noteworthy to mention that apart from the widely studied UCP1-dependent thermogenesis, there are other factors that contribute to the thermogenic machinery, such as calcium cycling, creatine signaling or UCP1-independent proton leak that could play a role in the study we present here [[48]UCP1-independent thermogenesis.].When body composition is compared between the two mouse groups, the biggest difference relies in the amount of WAT, being consistently lower in various depots along the FHL2−/− mouse body. Analyzing genome-wide expression of genes from gWAT the most remarkable difference is a strong upregulation of PGC1α in the absence of FHL2. PGC1α is a master regulator of mitochondrial biogenesis and energy expenditure, working as a coregulator of transcription in a wide variety of high energy-demanding metabolic tissues [[49]Fernandez-Marcos P.J. Auwerx J. Regulation of PGC-1α, a nodal regulator of mitochondrial biogenesis.]. The role of PGC1α is vastly described in BAT where its expression is induced upon cold stimuli. In the case of WAT, ectopic PGC1α expression has been shown to induce browning of adipocytes providing them the ability of producing heat [[50]Yuan D. Xiao D. Gao Q. Zeng L. PGC-1α activation: a therapeutic target for type 2 diabetes?.]. We pursued the white adipocyte browning hypothesis in FHL2−/− gWAT resorting to RNA-seq and making use of BATLAS software, a useful tool to estimate the brown adipocyte content in a heterogeneous sample [[32]Perdikari A. Leparc G.G. Balaz M. Pires N.D. Lidell M.E. Sun W. et al.BATLAS: deconvoluting brown adipose tissue.]. The results confirmed that the proportion of brown adipocyte signature gene expression was significantly higher in FHL2−/− gWAT than in WT tissue, which is likely an important part of the energy expenditure increase observed. Importantly, in the absence of FHL2, expression of AdrB3 and Ebf2 genes in gWAT was remarkably higher than in WT. This may contribute to the browning phenotype of this tissue in mice since activation of both genes is known to drive brown fat induction in WAT [[33]Jiang Y. Berry D.C. Graff J.M. Distinct cellular and molecular mechanisms for β3 adrenergic receptor-induced beige adipocyte formation.,[34]Wang W. Kissig M. Rajakumari S. Huang L. Lim H.W. Won K.J. et al.Ebf2 is a selective marker of brown and beige adipogenic precursor cells.].We wish to highlight the fact that in FHL2−/− mice increased expression of PGC1α was found both in WAT and heart (the latter not shown) after HFD. Similar as for FHL2, the role of PGC1α differs in a tissue-dependent context. Its function in BAT was already mentioned, but it is also known that fasting induces PGC1α gene expression in the liver to increase the process of gluconeogenesis, or that in the skeletal muscle and cardiomyocytes it regulates metabolic homeostasis [[52]Cheng C.F. Ku H.C. Lin H. Pgc-1α as a pivotal factor in lipid and metabolic regulation.]. Therefore, the finding of an upregulation of PGC1α gene across different tissues in our mouse model proposes the involvement of FHL2 in the co-regulation of PGC1α gene transcription. Although regulation of PGC1α gene expression varies depending on the tissue and energy requirements, it has been investigated that the PGC1α promoter has specific binding sites for FoxO1, MEF2, CREB and ATF2, all of which enhance its transcription [[49]Fernandez-Marcos P.J. Auwerx J. Regulation of PGC-1α, a nodal regulator of mitochondrial biogenesis.]. From this list of transcription regulators, FHL2 is known to interact with CREB and FoxO1 [[8]Tran M.K. Kurakula K. Koenis D.S. de Vries C.J.M. Protein-protein interactions of the LIM-only protein FHL2 and functional implication of the interactions relevant in cardiovascular disease.]. The interaction between FHL2 and CREB stimulates transcriptional activity of CREB target genes when co-expressed in mammalian cells [[53]Fimia G.M. De Cesare D. Sassone-Corsi P. A family of LIM-only transcriptional coactivators: tissue-specific expression and selective activation of CREB and CREM.], so in the hypothetical case that FHL2 was involved in PGC1α transcription regulation it would not be through its interaction with CREB. On the other hand, FHL2 interacts with FoxO1 through deacetylase Sirt1 thereby inhibiting FoxO1 transcriptional activity [[54]Yang Y. Hou H. Haller E.M. Nicosia S.V. Bai W. Suppression of FOXO1 activity by FHL2 through SIRT1-mediated deacetylation.], making it a plausible hypothesis for PGC1α increased expression in the absence of FHL2. Another possibility inferred from our data could be that the increase of AdrB3 expression in FHL2−/− adipocytes causes a higher oxidative capacity upregulating genes involved in mitochondrial activity such as PGC1α and inducing browning of WAT [[55]Granneman J.G. Li P. Zhu Z. Lu Y. Metabolic and cellular plasticity in white adipose tissue I: effects of β3-adrenergic receptor activation.].

In the present study, we answered three main questions concerning FHL2. Firstly, does FHL2 have a role in energy metabolism and obesity development, secondly, what function could FHL2 be exerting in metabolically relevant tissues, and lastly are these observations relevant for human. The first question was successfully answered by confirming the advantageous phenotype of FHL2−/− mice in comparison to wild type mice after HFD and next, we uncovered new roles for FHL2 in energy metabolism of heart and WAT. Finally, low FHL2 expression correlates with a lean phenotype in humans, adding FHL2 to the list of epigenetic factors involved in the development of this complex disease, in line with its role in obesity-related comorbidities.