ORIGINAL RESEARCH published: 31 January 2019 doi: 10.3389/fphys.2019.00039 Neuregulin 4 Is a Novel Marker of Beige Adipocyte Precursor Cells in Human Adipose Tissue Ferran Comas 1,2† , Cristina Martínez 1† , Mònica Sabater 1,2 , Francisco Ortega 1,2 , Jessica Latorre 1,2 , Francisco Díaz-Sáez 3,4 , Julian Aragonés 5,6 , Marta Camps 3,4 , Anna Gumà 3,4 , Wifredo Ricart 1,2,7 , José Manuel Fernández-Real 1,2,7 and José María Moreno-Navarrete 1,2,7* 1 Department of Diabetes, Endocrinology and Nutrition, Institut d’Investigació Biomèdica de Girona, Girona, Spain, 2 CIBEROBN (CB06/03/010), Instituto de Salud Carlos III, Madrid, Spain, 3 Department of Biochemistry and Molecular Biomedicine, Faculty of Biology, Institute of Biomedicine of the University of Barcelona, Barcelona, Spain, 4 CIBER de Diabetes y Enfermedades Metabólicas Asociadas, Instituto de Salud Carlos III, Madrid, Spain, 5 Research Unit, Hospital of Santa Cristina, Research Institute Princesa, Autonomous University of Madrid, Madrid, Spain, 6 CIBER de Enfermedades Cardiovasculares, Carlos III Health Institute, Madrid, Spain, 7 Department of Medicine, University of Girona, Girona, Spain Edited by: Paula Oliver, Background: Nrg4 expression has been linked to brown adipose tissue activity and Universidad de les Illes Balears, Spain browning of white adipocytes in mice. Here, we aimed to investigate whether these Reviewed by: observations could be translated to humans by investigating NRG4 mRNA and markers Rushita Bagchi, of brown/beige adipocytes in human visceral (VAT) and subcutaneous adipose tissue University of Colorado Denver, United States (SAT). We also studied the possible association of NRG4 with insulin action. Yongguo Li, Technische Universität München, Methods: SAT and VAT NRG4 and markers of brown/beige (UCP1, UCP3, and Germany TMEM26)-related gene expression were analyzed in two independent cohorts (n = 331 *Correspondence: and n = 59). Insulin resistance/sensitivity was measured using HOMAIR and glucose José María Moreno-Navarrete infusion rate during euglycemic hyperinsulinemic clamp. jmoreno@idibgi.org † These authors have contributed Results: In both cohort 1 and cohort 2, NRG4 and thermogenic/beige-related gene equally to this work expression were significantly increased in VAT compared to SAT. Adipogenic-related Specialty section: genes followed an opposite pattern. In cohort 1, VAT NRG4 gene expression was This article was submitted to positively correlated with BMI and expression of UCP1, UCP3, TMEM26, and negatively Integrative Physiology, with adipogenic (FASN, PPARG, and SLC2A4)- and inflammatory (IL6 and IL8)-related a section of the journal Frontiers in Physiology genes. In SAT, NRG4 gene expression was negatively correlated with HOMAIR and Received: 17 August 2018 positively with UCP1 and TMEM26 gene expression. Multiple linear regression analysis Accepted: 14 January 2019 revealed that expression of TMEM26 gene was the best predictor of NRG4 gene Published: 31 January 2019 expression in both VAT and SAT. Specifically, NRG4 and TMEM26 gene expression was Citation: Comas F, Martínez C, Sabater M, significantly increased in VAT, but not in SAT stromal vascular fraction cells (p < 0.001). Ortega F, Latorre J, Díaz-Sáez F, In cohort 2, the significant association between NRG4 and TMEM26 gene expression in Aragonés J, Camps M, Gumà A, both VAT and SAT was confirmed, and SAT NRG4 gene expression also was positively Ricart W, Fernández-Real JM and Moreno-Navarrete JM (2019) correlated with insulin action and the expression of UCP1. Neuregulin 4 Is a Novel Marker of Beige Adipocyte Precursor Cells Conclusion: Current findings suggest NRG4 gene expression as a novel marker of in Human Adipose Tissue. beige adipocytes in human adipose tissue. Front. Physiol. 10:39. doi: 10.3389/fphys.2019.00039 Keywords: obesity, neuregulin 4, browning, adipose tissue, insulin sensitivity Frontiers in Physiology | www.frontiersin.org 1 January 2019 | Volume 10 | Article 39 Comas et al. Neuregulin 4 and Human Beige Adipocytes INTRODUCTION (Guo et al., 2017). These studies suggested that adipose tissue- derived Nrg4 could exert positive effects on obesity associated The modulation of brown adipose tissue activity and browning metabolic disturbances (Wang et al., 2014; Ma et al., 2016; of white adipose tissue has been proposed as a promising Chen et al., 2017; Guo et al., 2017), improving glucose tolerance therapeutic strategy in the treatment of obesity-associated and insulin sensitivity and attenuating adipose tissue and liver metabolic disturbances (Wu et al., 2012; Bartelt and Heeren, inflammation (Wang et al., 2014; Ma et al., 2016; Chen et al., 2017; 2014; Hepler et al., 2017; Rabhi et al., 2018; Zhang S. et al., 2018), Guo et al., 2017). with the intention of improving insulin sensitivity (Hepler et al., In humans, only one study investigates NRG4 mRNA levels 2017; Rabhi et al., 2018) and hepatic steatosis (Huang et al., 2017), in adipose tissue in association with body fat mass, liver lipid among others. content and glucose tolerance (Wang et al., 2014), but no Several studies pointed to neuregulins as an important previous studies investigated the relationship between NRG4 family of ligands that regulate diverse aspects of glucose and markers of adipose tissue browning in human adipose and lipid metabolism and energy balance. In skeletal muscle tissue. Since previous mice studies demonstrated that NRG4 cells, recombinant neuregulin administration stimulated glucose was a marker of BAT activity and browning of WAT (Rosell uptake in muscle cells (Suárez et al., 2001) in an alternative et al., 2014; Wang et al., 2014; Ma et al., 2016; Chen et al., insulin-independent mechanism, activating PI3K, PDK1, and 2017), in the present study we aimed to investigate the potential PKCzeta pathways (Cantó et al., 2004), and promoted glucose and relationship between human adipose tissue NRG4 and markers palmitate oxidation, enhancing mitochondrial oxidative capacity of brown/beige adipocytes. Furthermore, the impact of adipose (Cantó et al., 2007). In liver, neuregulin 1 (Nrg1) and neuregulin tissue NRG4 on human obesity and insulin sensitivity was also 4 (Nrg4) reduced gluconeogenesis and lipogenesis and increased evaluated. fatty acid oxidation, improving systemic insulin sensitivity and glucose tolerance (Wang et al., 2014; Ennequin et al., 2015; Ma et al., 2016; Chen et al., 2017; Zhang P. et al., 2018). In fact, the MATERIALS AND METHODS Nrg4/ErbB4 signaling pathway protects hepatocytes from stress- induced cell death, preventing the steatosis to steatohepatitis Human Adipose Tissue Samples progression (Guo et al., 2017). In human breast cancer cells, In cohort 1, a group of 331 [155 visceral (VAT) and NRG1 binding to ERBB4 activates SREBP-2 and led to increased 176 subcutaneous (SAT) adipose tissues] (Cohort 1) from expression of LDL uptake- and cholesterol biosynthesis-related participants with normal body weight and different degrees genes (Haskins et al., 2015). A recent study demonstrated of obesity, with body mass index (BMI) within 20 and that ErbB4 deletion accelerated the development of obesity, 68 kg/m2 , were analyzed. In a second cohort of morbidly dyslipidemia, hepatic steatosis, hyperglycemia, hyperinsulinemia obese (BMI > 35 kg/m2 ) subjects with different degrees of and insulin resistance after 24 week on a medium-fat diet (Zeng insulin action [measured using hyperinsulinemic-euglycemic et al., 2018). Nrg4, a specific ligand for ErbB4 involved in clamp (Moreno-Navarrete et al., 2013)], VAT (n = 34) and neurite growth, administration in 3T3-L1 adipocytes inhibited SAT (n = 25) samples (Cohort 2) were studied. Altogether lipogenesis and induced browning and glucose uptake, but these subjects were recruited at the Endocrinology Service of did not exert any effects on adipogenesis and lipolysis (Zeng the Hospital of Girona “Dr Josep Trueta.” All subjects were et al., 2018). In fact, Nrg4 has been proposed as a marker of Caucasian origin and reported that their body weight had of brown adipose tissue (BAT) activity in mice, being highly been stable for at least 3 months before the study. Subjects expressed in cold-induced BAT activity and white adipose tissue were studied in the post-absorptive state. BMI was calculated (WAT) browning (Rosell et al., 2014; Wang et al., 2014). NRG4 as weight (in kg) divided by height (in m) squared. They had was expressed in fully differentiated brown adipocytes, but no systemic disease other than obesity and all were free of any not in preadipocytes, and increased during brown adipocyte infections in the previous month before the study. Liver diseases differentiation (Rosell et al., 2014; Wang et al., 2014). In vitro (specifically tumoral disease and HCV infection) and thyroid experiments showed that brown adipocytes-derived NRG4 might dysfunction were specifically excluded by biochemical work-up. promote the growth of neurites in adipose tissue, increasing Samples and data from patients included in this study were sympathetic innervation, enhancing BAT activity and browning partially provided by the FATBANK platform promoted by the of WAT (Rosell et al., 2014). However, Wang et al. (2014) CIBEROBN and coordinated by the IDIBGI Biobank (Biobank reported that despite the abundant expression of Nrg4 in BAT, it IDIBGI, B.0000872), integrated in the Spanish National Biobanks seems dispensable for cold-induced hypothermia response, being Network and they were processed following standard operating Ucp1 and Dio2 induced to similar extent by cold exposure in WT procedures with the appropriate approval of the Ethics, External and NRG4KO mice. These findings indicated that Nrg4 did not Scientific and FATBANK Internal Scientific Committees. directly participate in BAT thermogenesis. Diet-induced obesity led to a significant decreased Nrg4 Ethics Statement gene expression in WAT but not BAT (Wang et al., 2014; Ma This study was carried out in accordance with the et al., 2016; Chen et al., 2017) in mice. A recent study also recommendations of the ethical committee of the Hospital showed that diet-induced non-alcoholic steatohepatitis (NASH) of Girona “Dr Josep Trueta.” The protocol was approved by the resulted in a significant reduced Nrg4 in both BAT and WAT ethical committee of the Hospital of Girona “Dr Josep Trueta.” Frontiers in Physiology | www.frontiersin.org 2 January 2019 | Volume 10 | Article 39 Comas et al. Neuregulin 4 and Human Beige Adipocytes All subjects gave written informed consent in accordance with peroxisome proliferator-activated receptor gamma (PPARG, the Declaration of Helsinki, after the purpose of the study was Hs00234592_m1), solute carrier family 2 (facilitated glucose explained to them. transporter), member 4 (SLC2A4 or GLUT4, Hs00168966_m1), AT samples were obtained from SAT and VAT depots perilipin 1 (PLIN1, Hs00160173_m1), PPARG coactivator 1 during elective surgical procedures (cholecystectomy, surgery of alpha (PPARGC1A, Hs00173304_m1), uncoupling protein abdominal hernia and gastric bypass surgery). Adipose tissue 1 (UCP1, Hs01084772_m1), uncoupling protein 3 (UCP3, samples were washed, fragmented and immediately flash-frozen Hs01106052_m1), transmembrane protein 26 (TMEM26, in liquid nitrogen before being stored at −80◦ C. Hs00415619_m1), interleukin 6 (IL6, Hs00174131_m1), C-X- The isolation of adipocyte and stromal vascular fraction cells C motif chemokine ligand 8 (CXCL8 or also named IL8, (SVF) was performed from 17 SAT and 20 VAT non-frozen Hs00174103_m1), and peptidylprolyl isomerase A (cyclophilin adipose tissue samples. These samples were washed three to A) (4333763, PPIA as endogenous control). four times with phosphate-buffered saline (PBS) and suspended in an equal volume of PBS supplemented with 1% penicillin- Statistical Analyses streptomycin and 0.1% collagenase type I prewarmed to 37◦ C. Statistical analyses were performed using the SPSS 12.0 software. The tissue was placed in a shaking water bath at 37◦ C with Unless otherwise stated, descriptive results of continuous continuous agitation for 60 min and centrifuged for 5 min at variables are expressed as mean and SD for Gaussian variables 400 g at room temperature. The supernatant, containing mature or median and interquartile range for non-Gaussian variables. adipocytes, was recollected. The pellet was identified as the SVF. Parameters that did not fulfill normal distribution criteria were Isolated mature adipocytes and SVF stored at −80◦ C for gene log transformed to improve symmetry for subsequent analyses. expression analysis. The relation between variables was analyzed by simple correlation (using Spearman’s and Pearson’s tests) and multiple linear Analytical Methods regression analyses. ANOVA and unpaired Student’s t-tests were Serum glucose concentrations were measured in duplicate used to compare clinical variables and gene expression relative to by the glucose oxidase method using a Beckman glucose obesity and type 2 diabetes (T2D). analyser II (Beckman Instruments, Brea, CA, United States). Intraassay and interassay coefficients of variation were less than 4% for all these tests. HDL cholesterol was quantified RESULTS by a homogeneous enzymatic colorimetric assay through the cholesterol esterase/cholesterol oxidase/peroxidase reaction Representative Ct values of analyzed genes were shown in (Cobas HDLC3). Total serum triglycerides were measured by an Table 1. enzymatic, colorimetric method with glycerol phosphate oxidase and peroxidase (Cobas TRIGL). We used a Roche Hitachi Cobas Cohort 1 c 711 instrument to perform the determinations. Anthropometric and clinical data from cohort 1 were detailed in Table 2. Similar to thermogenic/beige-related gene expression, RNA Expression NRG4 was significantly increased in VAT compared to SAT, RNA purification was performed using the RNeasy Lipid Tissue whereas adipogenesis-related genes followed an opposite gene Mini Kit (QIAGEN, Izasa SA, Barcelona, Spain) and the integrity expression pattern (Figure 1A). In cohort 1, VAT NRG4 gene was checked by the Agilent Bioanalyzer (Agilent Technologies, expression was significantly increased in participants with obesity Palo Alto, CA, United States). Gene expression was assessed by (Table 2), but no significant differences were found between real time PCR using a LightCycler 480 Real-Time PCR System R non-diabetic obese and obese participants with T2D (Table 2). (Roche Diagnostics SL, Barcelona, Spain), using TaqMan and R SYBR green technology suitable for relative genetic expression quantification. The RT-PCR reaction was performed in a final TABLE 1 | Representative Ct values of analyzed genes. volume of 12 µl. The cycle program consisted of an initial Mean ± SD denaturing of 10 min at 95◦ C then 40 cycles of 15 s denaturizing phase at 95◦ C and 1 min annealing and extension phase at 60◦ C. PPIA 23.22 ± 0.25 A threshold cycle (Ct value) was obtained for each amplification FASN 25.91 ± 1.83 curve and a 1Ct value was first calculated by subtracting the Ct PPARG 29.49 ± 0.85 value for human cyclophilin A (PPIA) RNA from the Ct value SLC2A4 27.71 ± 0.53 for each sample. Fold changes compared with the endogenous PLIN1 22.78 ± 0.49 control were then determined by calculating 2−1Ct , so that gene PPARGC1A 30.86 ± 0.71 expression results are expressed as expression ratio relative to UCP1 36.82 ± 0.91 PPIA gene expression according to the manufacturer’s guidelines. UCP3 34.11 ± 0.65 PPIA Ct values in both SAT and VAT were comparable TMEM26 34.16 ± 0.95 (23.48 ± 0.81 in SAT vs. 23.49 ± 1.28 in VAT, p = 0.9, IL6 31.81 ± 2.28 n = 152). Primer/probe sets used were: neuregulin 4 (NRG4, IL8 30.89 ± 1.92 Hs00163592_m1), fatty acid synthase (FASN, Hs00188012_m1), NRG4 35.57 ± 0.99 Frontiers in Physiology | www.frontiersin.org 3 January 2019 | Volume 10 | Article 39 Comas et al. Neuregulin 4 and Human Beige Adipocytes TABLE 2 | Anthropometric and clinical characteristics according to obesity and T2D in cohort 1. Non-obese Obese Obese + T2D p N 54 88 34 Age (years) 47.4 ± 10.1 45.6 ± 10.5 47.2 ± 9.5 0.5 BMI (kg/m2 ) 25.4 ± 3.8 43.9 ± 7.4∗ 44.7 ± 4.1∗ <0.0001 Fasting glucose (mg/dl)a 86 (80–94) 93 (84–100.5) 126 (93.5–169.5)∗# <0.0001 HOMAIR (n = 56)a 1.18 (0.79–1.76) 2.06 (1.44–3.39) 5.59 (3.93–7.05)∗# 0.001 Total-cholesterol (mg/dl)a 199 (174–219) 193 (167.5–218.7) 182 (166–214) 0.5 HDL-cholesterol (mg/dl)a 64.5 (50.7–77.5) 55 (45.5–62.6) 50.1 (42–62)∗ 0.04 LDL-cholesterol (mg/dl)a 114.5 (88.7–135.5) 116.8 (97.5–134.7) 101.5 (89.5–137.7) 0.5 Fasting triglycerides (mg/dl)a 79.5 (57.7–101.2) 98 (75–132) 136 (89.5–164.5)∗# <0.0001 VAT NRG4 (RU) ×10−3a 1.26 (0.217–4.11) 3.59 (2.24–5.56)∗ 4.58 (2.54−6.52)∗ 0.001 SAT NRG4 (RU) ×10−3a 0.141 (0.061–0.239) 0.168 (0.095–0.381) 0.083 (0.057–0.155) 0.2 VAT, visceral adipose tissue; SAT, subcutaneous adipose tissue; T2D, type 2 diabetes; HOMAIR , homeostasis model assessment – insulin resistance index; RU, relative gene expression units. a Median and interquartile range. ∗ p < 0.05 compared to non-obese participants after performing Bonferroni post hoc test. # p < 0.05 compared to obese participants after performing Bonferroni post hoc test. Bold values mean that p-value reached statistical significance. No significant differences were observed on SAT NRG4 gene Cohort 2 expression according to obesity or T2D. In VAT, NRG4 gene To examine the findings replication of cohort 1 excluding the expression was positively correlated with BMI, and negatively effects of obesity, an independent cohort (cohort 2) composed correlated with adipogenic-related genes (FASN, PPARG, and of morbidly subjects with different degrees of insulin action has SLC2A4) (Table 3). Interestingly, NRG4 gene expression was been analyzed. Anthropometric and clinical data from cohort significantly positively associated with expression of brown/beige 2 were detailed in Table 4. Similar to cohort 1, NRG4 and adipocyte activity-related (UCP1, UCP3, and TMEM26) and thermogenic/beige- related gene expression was increased in negatively with inflammatory-related (IL6 and IL8) genes VAT (Figure 1B). No significant differences on SAT or VAT (Table 3 and Figure 2A). In SAT, NRG4 gene expression was NRG4 gene expression according to glucose tolerance or T2D negatively correlated with HOMAIR and positively with UCP1 were found. VAT NRG4 gene expression were associated with and TMEM26 gene expression (Table 3 and Figure 2B). expression of TMEM26 gene (Table 5), and SAT NRG4 with In multiple linear regression analysis, TMEM26 (β = 0.58, insulin sensitivity (M) and expression of SLC2A4, UCP1 and p < 0.0001; model adjusted R2 = 0.37, p < 0.0001), UCP3 TMEM26 genes (Table 5). (β = 0.24, p = 0.03; model adjusted R2 = 0.13, p = 0.001), IL6 (β = −0.32, p = 0.01; model adjusted R2 = 0.16, p = 0.001), IL8 (β = −0.36, p = 0.008; model adjusted R2 = 0.17, p < 0.0001), DISCUSSION FASN (β = −0.42, p = 0.001, model adjusted R2 = 0.12, p = 0.001) and PPARG (β = −0.38, p = 0.005; model adjusted R2 = 0.11, To the best of our knowledge, this is the first study showing p = 0.008) significantly contributed to the variance of NRG4 gene a significant relationship between NRG4 and TMEM26 gene expression in VAT after controlling for BMI. In SAT, TMEM26 expression in human adipose tissue. Interestingly, this association (β = 0.77, p < 0.0001; model adjusted R2 = 0.61, p < 0.0001) was found in both VAT and SAT, and validated in a second significantly contributed to the variance of NRG4 gene expression independent cohort. TMEM26 has been described as a specific after controlling for BMI. Multiple linear regression analysis marker of brite/beige adipocytes (Wu et al., 2012; Torriani et al., revealed that expression of TMEM26 gene was the best predictor 2016; Finlin et al., 2017). We also found positive associations of NRG4 gene expression in both VAT and SAT. among NRG4 and markers of thermogenic activity (characteristic In addition, correlations between UCP1, a specific marker of both brown and beige adipocytes) such as expression of brown adipocytes, and clinical and metabolic parameters of UCP1 and UCP3 genes. In addition, VAT NRG4 gene were also explored. No significant correlation between SAT or expression was negatively correlated with expression of white VAT UCP1 gene expression and BMI, fasting glucose, HOMAIR , lipogenic/adipogenic (FASN and PPARG)- and inflammatory total-, LDL- and HDL-cholesterol, and fasting triglycerides were (IL6 and IL8)-related genes, even after controlling for BMI. observed. VAT, but not SAT, UCP1 was positively correlated with Since beige adipocytes have less lipogenic capacity compared to SLC2A4 (r = 0.43, p < 0.0001), PPARGC1A (r = 0.36, p = 0.001) white adipocytes (Aziz et al., 2017; Zuriaga et al., 2017), and and UCP3 (r = 0.31, p = 0.005) gene expression. browning/beiging of adipose tissue protected against visceral In adipose tissue fractions, NRG4 and TMEM26, but not adipose tissue inflammation (Wu et al., 2017; Gonzalez-Hurtado UCP1, gene expression was significantly increased in visceral et al., 2018), the negative association between NRG4 and white SVFs compared to visceral adipocytes, subcutaneous SVFs and adipogenic/inflammatory genes reinforced NRG4 as a marker of subcutaneous adipocytes (Figure 2C). beige adipocytes. In fact, these correlations were only observed Frontiers in Physiology | www.frontiersin.org 4 January 2019 | Volume 10 | Article 39 Comas et al. Neuregulin 4 and Human Beige Adipocytes FIGURE 1 | (A,B) Comparison of NRG4, UCP1, TMEM26, PPARGC1A, FASN, PPARG, and PLIN1 gene expression in paired VAT and SAT in cohort 1 (n = 132) (A) and cohort 2 (n = 20) (B). ∗ p < 0.05, ∗∗ p < 0.01, and ∗∗∗ p < 0.001 compared to gene expression in VAT. TABLE 3 | Correlation between NRG4 gene expression and anthropometric and clinical characteristics and selected gene expression in SAT (n = 176) and VAT (n = 155) from cohort 1. VAT SAT r P r p Age (years) −0.02 0.7 0.11 0.2 BMI (kg/m2 ) 0.28 <0.0001 0.06 0.5 Fasting glucose (mg/dl) 0.17 0.03 0.01 0.8 HOMAIR (n = 56) 0.17 0.3 −0.32 0.02 Total cholesterol (mg/dl) −0.10 0.2 0.11 0.2 HDL cholesterol (mg/dl) −0.01 0.9 0.15 0.1 LDL cholesterol (mg/dl) −0.09 0.3 0.08 0.4 Fasting triglycerides (mg/dl) 0.05 0.5 −0.09 0.3 FASN (RU) −0.37 <0.0001 −0.05 0.6 PPARG (RU) −0.39 <0.0001 0.05 0.6 SLC2A4 (RU) −0.33 0.002 0.12 0.2 PLIN1 (RU) −0.15 0.1 −0.10 0.4 PPARGC1A (RU) 0.06 0.5 0.18 0.05 UCP1 (RU) 0.30 0.005 0.30 0.001 UCP3 (RU) 0.29 0.005 0.13 0.1 TMEM26 (RU) 0.77 <0.0001 0.56 <0.0001 IL6 (RU) −0.45 <0.0001 0.06 0.5 IL8 (RU) −0.36 0.001 −0.05 0.6 Bold values mean that p-value reached statistical significance. Frontiers in Physiology | www.frontiersin.org 5 January 2019 | Volume 10 | Article 39 Comas et al. Neuregulin 4 and Human Beige Adipocytes FIGURE 2 | (A,B) Bivariate correlation between NRG4 and TMEM26 gene expression in VAT (A) and SAT (B). (C) NRG4, TMEM26, and UCP1 gene expression in adipose tissue fractions (SVF and adipocytes) in both SAT (n = 17) and VAT (n = 20). ∗∗ p < 0.01 and ∗∗∗ p < 0.001 compared to gene expression in VAT SVF cells. TABLE 4 | Anthropometric and clinical characteristics according to glucose tolerance in cohort 2. NGT IGT T2D p 11 10 13 Age (years) 41.6 ± 4.1 50 ± 8.5∗ 51.5 ± 7.5∗ 0.004 BMI (kg/m2 ) 46.3 ± 8.7 47.8 ± 3.2 44.7 ± 8.1 0.6 Fasting glucose (mg/dl)a 90 (83–98) 102.5 (96–107) 121 (100.5–132)∗ 0.003 M [mg/(kg.min)]a 4.35 (2.21–6.28) 3.41 (1.92–5.21) 2.79 (1.68–4.31) 0.6 Total-cholesterol (mg/dl)a 182 (163–221) 207 (184.5–252) 179 (162–209) 0.07 HDL-cholesterol (mg/dl)a 49 (40–58) 45 (34.5–55.5) 46 (40.5–49.5) 0.5 LDL-cholesterol (mg/dl)a 105 (95–129) 147 (110.7–176.2) 108 (84.5–133) 0.03 Fasting triglycerides (mg/dl)a 99 (68–134) 141.5 (66.7–190.7) 139 (87.5–183) 0.4 VAT NRG4 (RU) ×10−3a 4.04 (1.11–6.08) 2.97 (0.141–5.92) 3.37 (2.14–6.17) 0.4 SAT NRG4 (RU) ×10−3a 0.172 (0.072–0.287) 0.099 (0.081–0.129) 0.131 (0.039–0.182) 0.4 VAT, visceral adipose tissue; SAT, subcutaneous adipose tissue; NGT, normal glucose tolerance; IGT, impaired glucose tolerance; T2D, type 2 diabetes; M, insulin sensitivity obtained from hyperinsulinemic-euglycemic clamp; RU, relative gene expression units. a Median and interquartile range. ∗ p < 0.05 compared to NGT participants after performing Bonferroni post hoc test. Bold values mean that p-value reached statistical significance. in the samples with the highest correlation between NRG4 and and inflammatory-related gene expression were not replicated in TMEM26 (r = 0.77, p < 0.0001). However, the correlations morbidly obese participants (cohort 2). Further studies in human between VAT NRG4 and UCP1, UCP3, lipogenic/adipogenic - adipose tissue should be required to validate these correlations. Frontiers in Physiology | www.frontiersin.org 6 January 2019 | Volume 10 | Article 39 Comas et al. Neuregulin 4 and Human Beige Adipocytes TABLE 5 | Correlation between NRG4 gene expression and anthropometric and clinical characteristics and selected gene expression in SAT (n = 25) and VAT (n = 34) from cohort 2. VAT SAT r p r p Age (years) −0.07 0.7 −0.17 0.4 BMI (kg/m2 ) 0.06 0.7 −0.08 0.7 Fasting glucose (mg/dl) 0.13 0.4 −0.21 0.3 M [mg/(kg.min)] −0.03 0.9 0.43 0.04 Total cholesterol (mg/dl) −0.23 0.2 −0.22 0.3 HDL cholesterol (mg/dl) 0.07 0.7 0.11 0.5 LDL cholesterol (mg/dl) −0.07 0.7 −0.27 0.2 Fasting triglycerides (mg/dl) −0.14 0.4 −0.14 0.5 FASN (RU) 0.05 0.8 0.13 0.6 PPARG (RU) −0.09 0.6 0.17 0.4 SLC2A4 (RU) −0.09 0.6 0.49 0.01 PLIN1 (RU) 0.04 0.8 −0.15 0.5 PPARGC1A (RU) 0.37 0.1 0.04 0.8 UCP1 (RU) 0.29 0.1 0.43 0.04 UCP3 (RU) 0.13 0.6 0.03 0.8 TMEM26 (RU) 0.56 0.002 0.51 0.01 IL6 (RU) 0.03 0.9 −0.10 0.6 Bold values mean that p-value reached statistical significance. Furthermore, in both cohort 1 and 2, similar to beige contradictory data in relation to its thermogenic activity. Further adipocytes-related genes (TMEM26), NRG4 gene expression functional studies in human adipose tissue should be required to was significantly more expressed in VAT, whereas, as expected confirm the possible role of NRG4 in this process. adipogenic-related genes were more expressed in SAT (Sauma Another interesting finding of current study was the et al., 2007; Moreno-Navarrete et al., 2016; Zuriaga et al., 2017). positive association between SAT NRG4 gene expression and Contrary to mice, increased pattern of browning gene expression insulin sensitivity. In a previous study, SAT and VAT NRG4 in human VAT compared to SAT has been reported (Zuriaga was significantly decreased in patients with impaired glucose et al., 2017). Interestingly, NRG4 and TMEM26 gene expression tolerance (IGT) and T2D (Wang et al., 2014), but this study was enriched in SVFs from VAT compared to SVFs from SAT did not evaluate insulin sensitivity. Even tough, no significant or adipocytes from VAT or SAT. This finding points to a differences were found in relation to IGT or T2D, probably specific population of beige precursor cells in VAT, characterized due to the relatively low number of adipose tissue samples by increased NRG4 and TMEM26 gene expression, and could compared to the previous study (n = 642) (Wang et al., 2014). explain the increased expression of beige/browning-related genes The current study showed a positive association between SAT observed in this fat depot (current data and Zuriaga et al., NRG4 and insulin sensitivity in both cohort 1 and cohort 2017). Reinforcing this idea, previous studies demonstrated 2, evaluated by two different methods (HOMAIR in cohort that TMEM26 gene expression was also increased in SVF and 1 and hyperinsulinemic-euglycemic clamp in cohort 2). In decreased in the late stages of beige adipocyte differentiation, and agreement with these findings, mice studies demonstrated that indicated its abundance in the precursors of beige adipocytes (Lee liver and adipose tissue Nrg4 overexpression improved insulin et al., 2015; Garcia et al., 2016). sensitivity and glucose tolerance and prevented HFD-induced Altogether these findings indicated NRG4 as an additional hyperinsulinemia (Ma et al., 2016). In fact, two recent studies marker of beige adipocytes in human adipose tissue, and (López-Soldado et al., 2016; Zhang P. et al., 2018) demonstrated suggested a possible role of this factor in the development of that recombinant neuregulin administration improved glucose beige adipocytes in human fat depots. Supporting this hypothesis, tolerance in both control and diabetic rats by enhancing hepatic Rosell et al. (2014) suggested that NRG4 might promote the glucose utilization (López-Soldado et al., 2016) and insulin growth of neurites in adipose tissue, increasing sympathetic sensitivity in high fat-fed mice (Zhang P. et al., 2018). innervation and in consequence, enhancing browning of WAT. On the other hand, contrary to previous study that Regarding the possible role of NRG4 on thermogenic activity, demonstrated that SAT NRG4 was negatively correlated with Wang et al. (2014) demonstrated in Nrg4 deficient mice that body fat mass (Wang et al., 2014), in the current study no Nrg4 did not directly participate in BAT thermogenesis, but Ma significant association was found between SAT NRG4 and BMI. et al. (2016) showed that Nrg4 overexpression enhanced BAT In cohort 1, VAT NRG4 gene expression was increased in obese activity with an increase of ∼1◦ C body temperature, and BAT compared to non-obese participants, and positively correlated and iWAT thermogenic gene expression. These studies supported with BMI, but in cohort 2, VAT NRG4 gene expression was not a possible role of NRG4 in beiging of human adipose tissue, but correlated with BMI. Of note, similar NRG4 gene expression Frontiers in Physiology | www.frontiersin.org 7 January 2019 | Volume 10 | Article 39 Comas et al. Neuregulin 4 and Human Beige Adipocytes values were observed comparing obese participants from AUTHOR CONTRIBUTIONS cohort 1 vs. those from cohort 2. Strikingly, the positive effects of diet-induced weight loss reducing body fat mass were JF-R and JM-N participated in study design and analysis of data not associated with expression of brown/beige-related genes and wrote and edited the manuscript. FC, CM, MS, FO, JL, and (Barquissau et al., 2018). However, additional studies will be FD-S participated in acquisition of data. JA, MC, AG, and WR necessary to clarify the relationship between human adipose participated in interpretation of data. FC, CM, MS, FO, JL, FD- tissue NRG4 and obesity. S, JA, MC, AG, and WR revised the manuscript critically for A significant limitation of current study was the absence of important intellectual content. All authors participated in final VAT or SAT NRG4 protein analysis by scarce availability of approval of the version to be published. adipose tissue lysates for protein in the same tissue samples used for RNA analysis. Similar to this, NRG4 protein analysis was not evaluated in recent relevant studies that demonstrated the FUNDING importance of NRG4 in adipose tissue (Wang et al., 2014; Chen et al., 2017; Guo et al., 2017; Nugroho et al., 2018; Pellegrinelli This work was partially supported by research grants PI16/01173 et al., 2018). Thus, additional studies should be performed to from the Instituto de Salud Carlos III from Spain, FEDER investigate if NRG4 protein follows the same pattern of mRNA funds and by Fundació Marató de TV3 (201626-30). CIBEROBN expression in human adipose tissue. Interestingly and consistent Fisiopatología de la Obesidad y Nutrición is an initiative from the with current findings, increased NRG4 mRNA and protein Instituto de Salud Carlos III and Fondo Europeo de Desarrollo release in human beige adipogenesis of mural-like mesenchymal Regional (FEDER) from Spain. stem cell was more recently reported (Su et al., 2018), indicating that NRG4 gene expression were correlated with NRG4 protein levels and supporting NRG4 participation in beige adipocyte ACKNOWLEDGMENTS differentiation. However, it is important to note that expression of NRG4 and brown/beige adipose tissue markers (UCP1, UCP3, We acknowledge the technical assistance of Emili Loshuertos and TMEM26) were extremely low, suggesting that browning of (IdIBGi) and Oscar Rovira (IdIBGi). We also want to white adipose tissue in humans may have less relevance than in particularly acknowledge the patients, the FATBANK platform mice. promoted by the CIBEROBN and the IDIBGI Biobank In conclusion, all these observations suggest NRG4 gene (Biobanc IDIBGI, B.0000872), integrated in the Spanish expression as a novel marker of beige adipocytes in human National Biobanks Network, for their collaboration and adipose tissue. coordination. REFERENCES Garcia, R. A., Roemmich, J. N., and Claycombe, K. J. (2016). Evaluation of markers of beige adipocytes in white adipose tissue of the mouse. Nutr. Metab. 13:24. Aziz, S. A., Wakeling, L. A., Miwa, S., Alberdi, G., Hesketh, J. E., and doi: 10.1186/s12986-016-0081-2 Ford, D. (2017). Metabolic programming of a beige adipocyte phenotype Gonzalez-Hurtado, E., Lee, J., Choi, J., and Wolfgang, M. J. (2018). Fatty acid by genistein. Mol. Nutr. Food. Res. 61:1600574. doi: 10.1002/mnfr.2016 oxidation is required for active and quiescent brown adipose tissue maintenance 00574 and thermogenic programing. Mol. Metab. 7, 45–56. doi: 10.1016/j.molmet. Barquissau, V., Léger, B., Beuzelin, D., Martins, F., Amri, E. Z., Pisani, D. F., et al. 2017.11.004 (2018). Caloric restriction and diet-induced weight loss do not induce browning Guo, L., Zhang, P., Chen, Z., Xia, H., Li, S., Zhang, Y., et al. (2017). of human subcutaneous white adipose tissue in women and men with obesity. Hepatic neuregulin 4 signaling defines an endocrine checkpoint for steatosis- Cell. Rep. 22, 1079–1089. doi: 10.1016/j.celrep.2017.12.102 to-NASH progression. J. Clin. Invest. 127, 4449–4461. doi: 10.1172/JCI Bartelt, A., and Heeren, J. (2014). Adipose tissue browning and metabolic health. 96324 Nat. Rev. Endocrinol. 10, 24–36. doi: 10.1038/nrendo.2013.204 Haskins, J. W., Zhang, S., Means, R. E., Kelleher, J. K., Cline, G. W., Canfrán- Cantó, C., Pich, S., Paz, J. C., Sanches, R., Martínez, V., Orpinell, M., et al. (2007). Duque, A., et al. (2015). Neuregulin-activated ERBB4 induces the SREBP-2 Neuregulins increase mitochondrial oxidative capacity and insulin sensitivity cholesterol biosynthetic pathway and increases low-density lipoprotein uptake. in skeletal muscle cells. Diabetes Metab. Res. Rev. 56, 2185–2193. doi: 10.2337/ Sci. Signal. 8:ra111. doi: 10.1126/scisignal.aac5124 db06-1726 Hepler, C., Shao, M., Xia, J. Y., Ghaben, A. L., Pearson, M. J., Vishvanath, L., Cantó, C., Suárez, E., Lizcano, J. M., Griñó, E., Shepherd, P. R., Fryer, L. G., et al. et al. (2017). Directing visceral white adipocyte precursors to a thermogenic (2004). Neuregulin signaling on glucose transport in muscle cells. J. Biol. Chem. adipocyte fate improves insulin sensitivity in obese mice. eLife 6:e27669. doi: 279, 12260–12268. doi: 10.1074/jbc.M308554200 10.7554/eLife.27669 Chen, Z., Wang, G. X., Ma, S. L., Jung, D. Y., Ha, H., Altamimi, T., et al. (2017). Huang, L., Pan, D., Chen, Q., Zhu, L. J., Ou, J., Wabitsch, M., et al. (2017). Nrg4 promotes fuel oxidation and a healthy adipokine profile to ameliorate Transcription factor Hlx controls a systematic switch from white to brown diet-induced metabolic disorders. Mol. Metab. 6, 863–872. doi: 10.1016/j. fat through Prdm16-mediated co-activation. Nat. Commun. 8:68. doi: 10.1038/ molmet.2017.03.016 s41467-017-00098-2 Ennequin, G., Boisseau, N., Caillaud, K., Chavanelle, V., Etienne, M., Li, X., et al. Lee, M. W., Odegaard, J. I., Mukundan, L., Qiu, Y., Molofsky, A. B., Nussbaum, (2015). Neuregulin 1 improves glucose tolerance in db/db mice. PLoS One J. C., et al. (2015). Activated type 2 innate lymphoid cells regulate beige fat 10:e0130568. doi: 10.1371/journal.pone.0130568 biogenesis. Cell 160, 74–87. doi: 10.1016/j.cell.2014.12.011 Finlin, B. S., Zhu, B., Confides, A. L., Westgate, P. M., Harfmann, B. D., Dupont- López-Soldado, I., Niisuke, K., Veiga, C., Adrover, A., Manzano, A., Martínez- Versteegden, E. E., et al. (2017). Mast cells promote seasonal white adipose Redondo, V., et al. (2016). Neuregulin improves response to glucose tolerance beiging in humans. Diabetes Metab. Res. Rev. 66, 1237–1246. doi: 10.2337/db16- test in control and diabetic rats. Am. J. Physiol. Endocrinol. Metab. 310, E440– 1057 E451. doi: 10.1152/ajpendo.00226.2015 Frontiers in Physiology | www.frontiersin.org 8 January 2019 | Volume 10 | Article 39 Comas et al. Neuregulin 4 and Human Beige Adipocytes Ma, Y., Gao, M., and Liu, D. (2016). Preventing high fat diet-induced obesity tissue gene expression in HIV-infected patients. J. Clin. Endocrinol. Metab. 101, and improving insulin sensitivity through neuregulin 4 gene transfer. Sci. Rep. 1225–1234. doi: 10.1210/jc.2015-3993 6:26242. doi: 10.1038/srep26242 Wang, G. X., Zhao, X. Y., Meng, Z. X., Kern, M., Dietrich, A., Chen, Z., et al. (2014). Moreno-Navarrete, J. M., Escoté, X., Ortega, F., Serino, M., Campbell, M., The brown fat-enriched secreted factor Nrg4 preserves metabolic homeostasis Michalski, M. C., et al. (2013). A role for adipocyte-derived lipopolysaccharide- through attenuation of hepatic lipogenesis. Nat. Med. 20, 1436–1443. doi: 10. binding protein in inflammation- and obesity-associated adipose tissue 1038/nm.3713 dysfunction. Diabetologia 56, 2524–2537. doi: 10.1007/s00125-013-3015-9 Wu, J., Boström, P., Sparks, L. M., Ye, L., Choi, J. H., Giang, A. H., et al. (2012). Moreno-Navarrete, J. M., Jove, M., Ortega, F., Xifra, G., Ricart, W., Obis, È, et al. Beige adipocytes are a distinct type of thermogenic fat cell in mouse and human. (2016). Metabolomics uncovers the role of adipose tissue PDXK in adipogenesis Cell 150, 366–376. doi: 10.1016/j.cell.2012.05.016 and systemic insulin sensitivity. Diabetologia 59, 822–832. doi: 10.1007/s00125- Wu, W., Shi, F., Liu, D., Ceddia, R. P., Gaffin, R., Wei, W., et al. (2017). Enhancing 016-3863-1 natriuretic peptide signaling in adipose tissue, but not in muscle, protects Nugroho, D. B., Ikeda, K., Barinda, A. J., Wardhana, D. A., Yagi, K., Miyata, K., against diet-induced obesity and insulin resistance. Sci. Signal. 10:489. doi: et al. (2018). Neuregulin-4 is an angiogenic factor that is critically involved in 10.1126/scisignal.aam6870 the maintenance of adipose tissue vasculature. Biochem. Biophys. Res. Commun. Zeng, F., Wang, Y., Kloepfer, L. A., Wang, S., and Harris, R. C. (2018). ErbB4 503, 378–384. doi: 10.1016/j.bbrc.2018.06.043 deletion predisposes to development of metabolic syndrome in mice. Am. Pellegrinelli, V., Peirce, V. J., Howard, L., Virtue, S., Türei, D., Senzacqua, M., et al. J. Physiol. Endocrinol. Metab. 315, E583–E593. doi: 10.1152/ajpendo.00166. (2018). Adipocyte-secreted BMP8b mediates adrenergic-induced remodeling 2018 of the neuro-vascular network in adipose tissue. Nat. Commun. 9:4974. doi: Zhang, P., Kuang, H., He, Y., Idiga, S. O., Li, S., Chen, Z., et al. (2018). NRG1- 10.1038/s41467-018-07453-x Fc improves metabolic health via dual hepatic and central action. JCI Insight Rabhi, N., Hannou, S. A., Gromada, X., Salas, E., Yao, X., Oger, F., et al. (2018). 3:98522. doi: 10.1172/jci.insight.98522 Cdkn2a deficiency promotes adipose tissue browning. Mol. Metab. 8, 65–76. Zhang, S., Cao, H., Li, Y., Jing, Y., Liu, S., Ye, C., et al. (2018). Metabolic benefits doi: 10.1016/j.molmet.2017.11.012 of inhibition of p38α in white adipose tissue in obesity. PLoS Biol. 16:e2004225. Rosell, M., Kaforou, M., Frontini, A., Okolo, A., Chan, Y. W., Nikolopoulou, E., doi: 10.1371/journal.pbio.2004225 et al. (2014). Brown and white adipose tissues: intrinsic differences in gene Zuriaga, M. A., Fuster, J. J., Gokce, N., and Walsh, K. (2017). Humans and expression and response to cold exposure in mice. Am. J. Physiol. Endocrinol. mice display opposing patterns of “Browning” gene expression in visceral and Metab. 306, E945–E964. doi: 10.1152/ajpendo.00473.2013 subcutaneous white adipose tissue depots. Front. Cardiovasc. Med. 4:27. doi: Sauma, L., Franck, N., Paulsson, J. F., Westermark, G. T., Kjølhede, P., Strålfors, P., 10.3389/fcvm.2017.00027 et al. (2007). Peroxisome proliferator activated receptor gamma activity is low in mature primary human visceral adipocytes. Diabetologia 50, 195–201. doi: Conflict of Interest Statement: The authors declare that the research was 10.1007/s00125-006-0515-x conducted in the absence of any commercial or financial relationships that could Su, S., Guntur, A. R., Nguyen, D. C., Fakory, S. S., Doucette, C. C., Leech, C., be construed as a potential conflict of interest. et al. (2018). A renewable source of human beige adipocytes for development of therapies to treat metabolic syndrome. Cell Rep. 25, 3215.e9–3228.e9. doi: Copyright © 2019 Comas, Martínez, Sabater, Ortega, Latorre, Díaz-Sáez, Aragonés, 10.1016/j.celrep.2018.11.037 Camps, Gumà, Ricart, Fernández-Real and Moreno-Navarrete. This is an open- Suárez, E., Bach, D., Cadefau, J., Palacin, M., Zorzano, A., and Gumá, A. (2001). access article distributed under the terms of the Creative Commons Attribution A novel role of neuregulin in skeletal muscle. Neuregulin stimulates glucose License (CC BY). The use, distribution or reproduction in other forums is permitted, uptake, glucose transporter translocation, and transporter expression in muscle provided the original author(s) and the copyright owner(s) are credited and that the cells. J. Biol. Chem. 276, 18257–18264. doi: 10.1074/jbc.M008100200 original publication in this journal is cited, in accordance with accepted academic Torriani, M., Srinivasa, S., Fitch, K. V., Thomou, T., Wong, K., Petrow, E., et al. practice. No use, distribution or reproduction is permitted which does not comply (2016). Dysfunctional subcutaneous fat with reduced Dicer and brown adipose with these terms. Frontiers in Physiology | www.frontiersin.org 9 January 2019 | Volume 10 | Article 39