Effects of X-chromosome Tenomodulin Genetic Variants on Obesity in a Children’s Cohort and Implications of the Gene in Adipocyte Metabolism

1 Scientific RepoRts | (2019) 9:3979 | https://doi.org/10.1038/s41598-019-40482-0 www.nature.com/scientificreports Effects of X-chromosome Tenomodulin Genetic Variants on Obesity in a Children’s Cohort and Implications of the Gene in Adipocyte Metabolism Francisco Javier Ruiz-Ojeda1,2, Augusto Anguita-Ruiz 1,2,3, Azahara I. Rupérez 1,5, Carolina Gomez-Llorente 1,2,3, Josune Olza1,2,3, Rocío Vázquez-Cobela4, Mercedes Gil-Campos3,7, Gloria Bueno3,5,6, Rosaura Leis3,4, Ramón Cañete3,7, Luis A. Moreno3,5,6, Angel Gil 1,2,3 & Concepcion Maria Aguilera 1,2,3 Tenomodulin (TNMD) is a type II transmembrane glycoprotein that has been recently linked to obesity, and it is highly expressed in obese adipose tissue. Several sex-dependent associations have been observed between single-nucleotide polymorphisms (SNPs) of the TNMD gene, which is located in the X-chromosome, and obesity, type 2 diabetes mellitus (T2DM), and metabolic syndrome in adults. On the other hand, results are lacking for children. We aimed (i) to study the association between TNMD genetic variants and metabolic complications related to childhood obesity and (ii) to investigate the function of TNMD in human adipocytes. We conducted a case-control, multicenter study in 915 Spanish children and demonstrated significant positive associations between TNMD genetic variants and BMI z-score, waist circumference, fasting glucose, and insulin resistance in boys, highlighting the SNP rs4828038. Additionally, we showed a BMI-adjusted inverse association with waist circumference in girls. Second, in vitro experiments revealed that TNMD is involved in adipogenesis, along with glucose and lipid metabolism in differentiated adipocytes, and these effects may be mediated through AMPK activation. Hence, these results suggest that TNMD genetic variants could be potentially useful as early life risk indicators for obesity and T2DM. In addition, we support the fact that TNMD exhibits significant metabolic functions in adipocytes. Childhood obesity is a major health problem (GBD 2015 Obesity Collaborators) characterized by an expansion of the adipose tissue (AT)1. Many children who are overweight or suffer from obesity before puberty maintain obesity in early adulthood, which is associated with increased morbidity and mortality2. The expansion of AT implies metabolic alterations that are mainly related to glucose and lipid metabolism3. White adipose tissue (WAT) is 1Department of Biochemistry and Molecular Biology ii, institute of nutrition and food technology “José Mataix”, Center of Biomedical Research, University of Granada, Avda. del Conocimiento s/n., 18016, Armilla, Granada, Spain. 2instituto de investigación Biosanitaria iBS.GRAnADA, complejo Hospitalario Universitario de Granada, Granada, 18014, Spain. 3Spanish Biomedical Research centre in Physiopathology of Obesity and nutrition (ciBeRobn), Instituto de Salud Carlos III (ISCIII), Madrid, 28029, Spain. 4Unit of investigation in nutrition, Growth and Human Development of Galicia, Pediatric Department (USc), instituto de investigación Sanitaria de Santiago de compostela (iDiS), complexo Hospitalario Universitario de Santiago, Santiago de compostela, Spain. 5Growth, exercise, nUtrition and Development (GenUD) Research Group, Universidad de Zaragoza, Zaragoza, Spain. 6instituto Agroalimentario de Aragón (IA2), Instituto de Investigación Sanitaria de Aragón (IIS Aragón), Zaragoza, Spain. 7Department of Paediatrics, Reina Sofia University Hospital, Institute Maimónides of Biomedicine Investigation of Córdoba (IMIBIC), University of Córdoba, Avda Menéndez Pidal s/n, 14004, Córdoba, Spain. Francisco Javier Ruiz-Ojeda and Augusto Anguita-Ruiz contributed equally. Correspondence and requests for materials should be addressed to F.J.R.-O. (email: [email protected]) or c.M.A. (email: [email protected]) Received: 12 January 2018 Accepted: 18 October 2018 Published: xx xx xxxx OPEN 2 Scientific RepoRts | (2019) 9:3979 | https://doi.org/10.1038/s41598-019-40482-0 www.nature.com/scientificreports www.nature.com/scientificreports/ the main site for energy storage, but it is also an endocrine organ that secretes cytokines and adipokines4. White subcutaneous adipose tissue (SAT) and white visceral fat depots (VAT) represent 80% and 20% of total body fat storage, respectively. VAT size is strongly associated with insulin resistance, and it is well established that VAT and SAT are different with respect to adipocyte size and metabolic activity5. Tenomodulin (TNMD) was identified as a novel gene in 2001 by Brandau et al.6 and Shukunami et al.7, and it is located in the human Xq22 region, where it spans approximately 15 kb6,7. TNMD is a type II transmembrane protein; it is described as an angiogenesis inhibitor and is highly expressed in hypovascular connective tissues such as tendons and cartilage6,8. Indeed, TNMD contains a putative proteinase cleavage and two glycosylation sites where the C-terminus of the protein is cleaved in those tissues9–12. Furthermore, its expression in human AT has been recently observed to be higher in obesity and lower after diet-induced weight loss13. Analyses of AT TNMD expression in obese and lean subjects have also shown that TNMD mRNA is correlated with body mass index (BMI) in adults14–16. In line with these results, our research group previously found that TNMD was five-fold upregulated in the VAT of prepubertal children with obesity, compared with their normal-weight counterparts17. Furthermore, TNMD is known to promote human adipocyte differentiation and to act as a protective factor against insulin resistance in obese VAT18. Likewise, several studies have indicated that single-nucleotide polymorphisms (SNPs) in the TNMD gene are associated with BMI, serum low-density lipoprotein cholesterol (LDL-c) levels, and inflammatory factors in adults in a sex-specific manner19. Specifically, the SNPs rs2073162 and rs2073163 have been associated with type 2 diabetes mellitus (T2DM) in men, central obesity in women and inflammation in men and women19–23. On the other hand, results are lacking for children. At the GWAS level, none of the analyses that have been conducted on obesity traits have reported associations for TNMD SNPs. Since the X-chromosome has often been less scrutinized because of the unique statistical challenges it presents24,25, the X-chromosomal location of TNMD could be one of the reasons why its genetic variants have not been widely studied in the genetic context of obesity. Despite this, the X-chromosome has been proposed as a potential source of missing heritability and an important genomic region to be included into analyses26. Considering all this and the availability of new tools to overcome these complexities25,27–30, the present work was undertaken to study the effects of TNMD genetic variants in children with obesity and to evaluate the potential metabolic function of this gene in human adipocytes. First, we studied the association between TNMD genetic variants and metabolic complications related to childhood obesity. Second, through gene silencing, we aimed to demonstrate that TNMD is required for adipocyte metabolism in fully differentiated adipocytes. To the best of our knowledge, this is the first study to report an association between TNMD SNPs and childhood obesity while supporting the implication of TNMD in adipocyte metabolism. Results TNMD genetic variants are associated with BMI z-score in boys. The anthropometric, clinical, and metabolic characteristics of the children participating in the present study are shown in the supplementary material according to obesity status (Supplementary TableS1). Minor allele frequencies (MAFs) of all markers studied are listed in Table1. All SNPs showed MAFs above 5% regardless of the obesity class. Given the location of TNMD in a sex chromosome, all genetic analyses were conducted separately for boys and girls. The linkage disequilibrium (LD) pattern of the region of TNMD that was studied is presented in Fig.1; two previous literature-reported blocks were also identified in our population in a sex-stratified manner: haploblock-1 (rs11798018, rs5966709, and rs4828037) and haploblock-2 (rs2073162, rs2073163, rs4828038, and rs1155974)20. All SNPs within the haploblock-2 showed significant and positive association with the BMI z-score in boys but not in girls (Table1). Conversely, no association was identified between variants of the haploblock-1 and BMI z-score in any sex group. Among the associated SNPs within the haploblock-2, the rs2073162 and the rs4828038 exhibited the highest effect sizes and the most significant P values. All mentioned associations remained statistically significant after applying multiple-test correction by False Discovery Rate (FDR). Instead, only the rs2073162 association stood multiple-test correction by Bonferroni adjustment (Table1). No effects were reported for the haploblock-1 variants on any of the studied phenotypes, and rs4828038 was identified as a tag SNP within the haploblock-2 according to the Bakker’s method31; therefore, the paper will focus on all associations and findings that have been reported for this marker. A conditional joint multiple-SNP analysis for BMI z-score in boys further revealed there are no independent effects on the phenotype between all linked markers of the haploblock-2 (Supplementary Fig.S1). On this matter, our tag SNP rs4828038 might be a good representative marker for the region. Additionally, haplotype-based tests were performed to determine whether the reported associations remained statistically significant when each TNMD haploblock was analyzed as an allelic phase and not as independent single variants (Supplementary TableS2). As expected, the association between the haploblock-2 and the BMI z-score in boys remained statistically significant, even after applying a multiple-test correction. The tag SNP rs4828038 is associated with central adiposity and impaired glucose metabolism in boys, while it correlates with lower waist circumference in girls. To further explore the implication of TNMD in obesity and metabolic alterations, we studied the association between TNMD genetic variants and a range of additional anthropometric measurements and metabolic features including cardiovascular disease (CVD) and inflammation biomarkers (Table2 and Supplementary TableS3). Concerning anthropometric indicators of central obesity, a statistically significant and risky correlation was observed between the rs4828038-T-allele and waist circumference (WC) in boys, which disappeared after adjusting the model for BMI confounding. This finding did not remain statistically significant after applying FDR multiple-test correction neither. Conversely, we identified a protective association between the rs4828038-T-allele and WC in girls. The finding remained statistically significant after adjusting the model for BMI confounding 3 Scientific RepoRts | (2019) 9:3979 | https://doi.org/10.1038/s41598-019-40482-0 www.nature.com/scientificreports www.nature.com/scientificreports/ (Table2). Interestingly, this result also stood multiple-test correction and showed a 2.4% FDR value. Other central-adiposity indicators such as the waist-to-height ratio (WHR) reported equal findings in girls but did not reach multiple-test significance. Regarding glucose metabolism, the rs4828038-T-allele was associated with higher levels of fasting glucose and higher values of homeostatic model assessment for insulin resistance (HOMA-IR) in boys. In the same way, the rs4828038-T-allele was also associated with a lower quantitative insulin sensitivity check index (QUICKI) (Table2). All results were obtained under an additive model adjusted for age. Specifically, the correlation between our tag SNP and fasting glucose levels reached both nominal and multiple-test significance, exhibiting an FDR value of 4.9%. Significant results were also obtained after adjusting the model for BMI confounding but with only nominal statistical significance (P = 0.009). In relation to HOMA-IR, each rs4828038-T-allele copy increased the index value by 0.33 units in comparison to the rs4828038-C-allele. For the QUICKI index, we reported a risky and statistically significant correlation in boys that stood multiple-test FDR significance. The same results were obtained after adjusting the model for BMI confounding but with only nominal statistical significance (P = 0.039). No additional associations were reported regarding our tag SNP and any other glucose metabolism phenotype in neither boys nor girls. According to previous studies relating TNMD SNPs to inflammatory traits and diseases such as age-related macular degeneration (AMD) or T2DM21,32, we investigated the association between the rs4828038 SNP and inflammation, CVD risk markers and adipokines (Table2 and Supplementary TableS3). Interestingly, we reported a significant and positive correlation between the rs4828038-T-allele and interleukin (IL)-6 levels in girls, which remained significant after multiple-test correction (FDR = 4.7%). Concordant results were obtained after adjusting the model for BMI confounding but with only nominal significance (P = 0.021). Associations in line with this result have been previously identified in adult females in relation to TNMD haploblock-1 SNPs21, which suggests that TNMD could mediate its putative effects on obesity via low-grade inflammation. Nonsignificant results were observed for the rest of the analyzed traits of the block regardless of the sex. Other metabolic features such as blood pressure and lipid traits were also investigated. We detected a significant association between the rs4828038-T-allele and diastolic blood pressure (DBP) in girls when adjusting the model for BMI confounding (Table2). The result did not remain statistically significant after applying multiple-test correction, showing an FDR value of 10.2%. Remaining phenotypes of the block did not exhibit any significant correlation with analyzed markers (Table2 and Supplementary TableS3). TNMD is associated with adipogenesis and lipid metabolism in human adipocytes. Based on the data regarding the relationship between TNMD genetic variants and the BMI z-score and, especially, WC, SNP MAF P-value FDR adjusted P-value BONF adjusted P-value A1/ Normal-weight Overweight Obese β (95%CI)A2 rs11798018 A/C Females 0.272 0.269 0.263 0.11 (−0.13, 0.36) 0.369 0.4696 1 Males 0.283 0.242 0.280 0.14 (−0.34, 0.63) 0.561 0.6545 1 rs5966709 T/G Females 0.289 0.294 0.355 0.14 (−0.07, 0.35) 0.202 0.404 1 Males 0.325 0.286 0.316 0.07 (−0.39, 0.52) 0.762 0.8206 1 rs4828037 C/T Females 0.297 0.332 0.379 0.15 (−0.06, 0.36) 0.161 0.404 1 Males 0.350 0.281 0.332 0.01 (−0.43, 0.46) 0.952 0.952 1 rs2073162 A/G Females 0.447 0.468 0.463 0.11 (−0.08, 0.30) 0.267 0.4262 1 Males 0.395 0.328 0.469 0.65 (0.22, 1.07) 0.003 0.0233 0.042 rs2073163 C/T Females 0.445 0.479 0.476 0.11 (−0.09, 0.30) 0.274 0.4262 1 Males 0.402 0.345 0.471 0.59 (0.15, 1.03) 0.008 0.028 0.110 rs4828038 T/C Females 0.447 0.459 0.463 0.12 (−0.06, 0.31) 0.193 0.404 1 Males 0.395 0.311 0.465 0.64 (0.21, 1.06) 0.004 0.0233 0.056 rs1155974 T/C Females 0.447 0.454 0.443 0.09 (−0.10, 0.29) 0.339 0.4696 1 Males 0.400 0.311 0.466 0.62 (0.19, 1.04) 0.005 0.0233 0.070 Table 1. Association between TNMD SNPs and BMI z-score in children. BMI, body mass index; A1, minor allele; A2 major allele; MAF, minor allele frequency; β, Beta obtained under an additive model; CI, confidence interval. Linear regression analyses stratified by sex were performed under an additive model assuming TNMD locus escapes from the X-chromosome inactivation process. That is, while the female genotypes were coded 0, 1, or 2 according to 0, 1, or 2 TNMD SNP alleles, the genotypes for males were coded 0 or 1 according to 0 or 1 alleles. 4 Scientific RepoRts | (2019) 9:3979 | https://doi.org/10.1038/s41598-019-40482-0 www.nature.com/scientificreports www.nature.com/scientificreports/ together with our previously published results that described a highly significant upregulation of TNMD expression in the VAT of children with obesity17, we performed a functional in vitro study. The aim of this study was to elucidate the role of TNMD in human adipocytes, as the major constituent cells in AT. To assess TNMD gene and Figure 1. Location of selected markers in the TNMD gene and linkage disequilibrium (LD) analyses. (a) Light blue boxes represent exons, while the connecting blue lines are introns. Abbreviations: rs, reference SNP code; UTR, untranslated region. (b,c) show the LD pattern of the region in boys and girls, respectively. Left red triangles represent D′ values while right the black/gray triangles indicate R2 values. Triangle frames indicate observed haploblocks according to the solid spine of LD; the first consists of rs11798018, rs5966709, and rs4828037, and the second consists of rs2073162, rs2073163, rs4828038, and rs1155974. Between triangles, we listed each haplotype in a block along with its population frequency and connections from one block to the next. In the crossing areas, a value of multiallelic D′ is shown. This represents the level of recombination between the two haploblocks. 5 Scientific RepoRts | (2019) 9:3979 | https://doi.org/10.1038/s41598-019-40482-0 www.nature.com/scientificreports www.nature.com/scientificreports/ protein expression in human adipocytes from adipose-derived stem cells (ADSCs), we determined the mRNA and protein levels at various times during adipogenic differentiation. In agreement with previous reports13,18, we found that TNMD expression and protein levels were significantly upregulated in human differentiated adipocytes at day 14 compared with ADSCs at day 0. However, we did not find any significant differences between days 7, 10, and 14. Immunofluorescence images also showed the differences between TNMD expression at days 0 and 14 (Fig.2). In addition, it has been demonstrated that TNMD inhibition blocks adipogenesis in Simpson-Golabi-Behmel syndrome (SGBS) preadipocytes and benefits VAT expansion in mice. Since TNMD is required for adipocyte differentiation in SGBS adipocytes18, we first confirmed that TNMD was also required for human adipogenesis in ADSCs (Supplementary Fig.S2). Additionally, TNMD inhibition expression in fully differentiated adipocytes at day 14, downregulated the gene and protein expression of peroxisome proliferator-activated receptor gamma (PPARG), CCAAT/enhancer-binding protein alpha (CEBPA) and angiopoietin-like 4 (ANGPTL4) in TNMD-knocked-down adipocytes (Fig.3). Phenotype Genotypes Β (95%CI) PFDR adjusted P ΒBMI (95%CI) P BMI FDR adjusted P BMI T/T nT/T T/C nT/C C/C nC/C Waist circumference (cm) Females 75.68 (15.11) 124 77.13 (16.68) 168 77.15 (15.13) 168 −0.42 (−2.05, 1.04) 0.611a0.694 −1.22 (−2.06, −0.38) 0.005a0.024 Males 80.9 (17.4) 157 NA 077.04 (17.3) 220 3.76 (0.55, 6.98) 0.022a0.088 −0.1 (−1.54, 1.35) 0.897a0.962 Waist to Height Ratio Females 0.526 (0.09) 109 0.54 (0.10) 151 0.53 (0.09) 142 −0.002 (−0.01, 0.009) 0.701a0.876 −0.01 (−0.01, 0.001) 0.030a0.172 Males 0.54 (0.10) 130 NA 00.53 (0.10) 194 0.015 (−0.007, 0.04) 0.183a0.735 −0.002 (−0.01, 0.01) 0.678a0.753 Systolic BP (mmHg) Females 104.8 (14.43) 108 106.9 (12.73) 148 104.9 (13.81) 136 0.006 (−1.65, 1.66) 0.995b0.995 −0.46 (−1.95, 1.03) 0.542b0.995 Males 107.7 (15.06) 131 NA 0105.9 (16.31) 193 1.01 (−2.15, 4.17) 0.531b0.995 0.07 (−2.91, 3.05) 0.963b0.995 Diastolic BP (mm Hg) Females 62.82 (11.99) 108 65.17 (10.46) 148 65.46 (10.77) 136 −1.3 (−2.68. 0.09) 0.067b0.332 −1.60 (−2.91, −0.30) 0.016b0.102 Males 65.55 (11.31) 131 NA 063.82 (10.49) 193 1.53 (−0.82, 3.88) 0.204b0.446 0.96 (−1.32, 3.23) 0.410b0.708 Glucose (mg/dl) Females 84.7 (7.93) 132 83.64 (6.89) 170 84.37 (7.02) 172 0.16 (−0.66, 0.97) 0.707a0.980 0.18 (−0.62, 0.99) 0.658a0.906 Males 85.9 (8.45) 171 NA 083.71 (8.43) 235 2.22 (0.57, 3.87) 0.009a0.049 2.21 (0.55, 3.88) 0.009a0.054 Insulin (mU/dl) Females 12.85 (10.15) 131 13.5 (9.54) 164 11.63 (7.10) 167 0.81 (−0.14, 1.76) 0.094a0.196 0.72 (−0.15, 1.59) 0.104a0.443 Males 10.86 (7.67) 166 NA 09.50 (7.56) 224 1.39 (−0.02, 2.79) 0.053a0.196 0.74 (−0.55, 2.02) 0.261a0.443 HOMA-IR Females 2.76 (2.36) 131 2.82 (2.04) 163 2.44 (1.55) 166 0.2 (−0.01, 0.41) 0.064a0.186 0.18 (−0.02, 0.38) 0.075a0.386 Males 2.34 (1.79) 165 NA 02.02 (1.72) 224 0.33 (0.009, 0.66) 0.044a0.186 0.19 (−0.11, 0.50) 0.207a0.386 QUICKI Females 0.35 (0.05) 131 0.34 (0.04) 163 0.35 (0.04) 165 0.0001 (−0.004, 0.004) 0.953a0.952 0.0005 (−0.003, 0.004) 0.787a0.810 Males 0.35 (0.04) 165 NA 00.37 (0.05) 224 −0.01 (−0.02, −0.003) 0.006a0.038 −0.008 (−0.02, −0.0004) 0.039a0.232 TAG (mg/dl) Females 72.57 (36.18) 132 71.58 (32.59) 170 67.1 (27.12) 172 2.86 (−0.74, 6.46) 0.120a0.615 2.63 (−0.87, 6.13) 0.142a0.878 Males 65.53 (35.37) 172 NA 064.32 (35.51) 234 1.35 (−5.59, 8.29) 0.703a0.915 −1.00 (−7.74, 5.73) 0.771a0.905 HDL-c (mg/dl) Females 52.32 (14.03) 130 51.95 (13.77) 168 53.39 (13.69) 171 −0.6 (−2.17, 0.96) 0.451a0.683 −0.35 (−1.76, 1.06) 0.625a0.960 Males 55.25 (15.1) 169 NA 056.69 (16.53) 231 −1.53 (−4.67, 1.60) 0.338a0.683 0.20 (−2.64, 3.04) 0.892a0.960 IL-6 (ng/l) Females 3.86 (7.22) 106 2.78 (4.51) 152 2.21 (3.83) 151 0.80 (0.17, 1.44) 0.014a0.047 0.75 (0.12, 1.38) 0.021a0.073 Males 3.25 (6.29) 142 NA 03.53 (5.80) 205 −0.30 (−1.59, 0.98) 0.642a0.818 −0.38 (−1.68, 0.92) 0.568a0.727 Table 2. Association between rs4828038 TNMD and anthropometric, biochemical, and inflammation characteristics (mean (SD)) in children. BMI, body mass index; BP, blood pressure; HOMA-IR, homeostasis model assessment for insulin resistance; QUICKI, quantitative insulin sensitivity check index; TAG, triglycerides; HDL-C, high-density lipoprotein cholesterol; IL, interleukin; MAF, minor allele frequency; βBMI, Beta obtained under an additive model adjusted for BMI; P-valueBMI, P (P value) obtained under an additive model adjusted for BMI; CI, confidence interval; NA, not applicable. Linear regression analyses stratified by sex were performed under an additive model assuming TNMD locus escapes from the X-chromosome inactivation process. That is, while the female genotypes were coded 0, 1, or 2 according to 0, 1, or 2 TNMD SNP alleles, the genotypes for males were coded 0 or 1 according to 0 or 1 alleles. aAdjusted for age. bAdjusted for age and height. 6 Scientific RepoRts | (2019) 9:3979 | https://doi.org/10.1038/s41598-019-40482-0 www.nature.com/scientificreports www.nature.com/scientificreports/ Regarding lipid metabolism, knock-down of TNMD led to reduced lipolysis as observed by decreased extracellular glycerol levels in cell supernatants (Fig.3g). In addition, the expression levels of lipases such as hormone-sensitive lipase (HSL), adipose triglyceride lipase (ATGL), and perilipin (PLIN) were significantly downregulated upon shRNA-TNMD treatment (Fig.3d–f), as was ANGPTL4, which is a mediator of intracellular lipolysis in adipocytes (Fig.3c). Since TNMD is involved in the regulation of tenocyte proliferation, tendon development, and angiogenesis inhibition12, we investigated vascular endothelial growth factor A (VEGFA) gene and protein expression in shRNA-mediated knocked-down TNMD cells. However, we did not observe significant changes (Supplementary Fig.S3). In this sense, no differences in blood vessel morphology and density have been previously found in TNMD transgenic mice18. TNMD knock-down impairs glucose metabolism in human adipocytes. To test whether TNMD plays a role in glucose metabolism in adipocytes, TNMD was inhibited in fully differentiated human adipocytes at day 14. In this experiment, we observed that glucose transporter 4 (GLUT4) gene and protein expression were downregulated when TNMD expression was inhibited (Fig.4a,b). This finding was supported by immunofluorescence images that showed reduced GLUT4 protein levels in shRNA-TNMD-treated adipocytes (Fig.4g). However, TNMD inhibition did not affect basal glucose uptake in a significant manner; however, there was a tendency toward a reduction by approximately 1.5-fold (P-value: 0.08) (Fig.4c). Adiponectin mRNA and protein levels were also decreased in TNMD-knocked-down adipocytes (Fig.4d). Regarding the activation of kinases involved in glucose metabolism, we observed a lower activation of AMP-activated protein kinase-alpha (AMPKα) and no differences in AKT (phosphor-AKT, Ser473) in TNMD-inhibited adipocytes. Thus, the association between TNMD and glucose metabolism in human adipocytes could be mediated by AMPK. TNMD knock-down triggers inflammation in human adipocytes. Many studies have reported that inflammation occurs in adipocytes associated with obesity, which is further related to metabolic dysfunction and insulin resistance33,34. Thus, we next studied the inflammatory status of TNMD-inhibited human differentiated adipocytes. In our study, TNMD-knocked-down cells showed an upregulation of inflammatory markers such as IL1-β and tumor necrosis factor-α (TNF-α) mRNA. Furthermore, we observed a significant upregulation of the protein levels in the shRNA-TNMD-treated adipocytes compared with that in cells transfected with Figure 2. TNMD expression during adipogenic differentiation. (a) Gene expression of TNMD at various time points during adipogenic differentiation in human adipose-derived stem cells (ADSCs); mRNA levels were normalized to those of hypoxanthine-guanine phosphoribosyltransferase-1 (HPRT1) and presented as fold-change, as calculated using the Pfaffl method. (b) TNMD protein levels from cell lysates were analyzed by Western blotting using a specific antibody against TNMD (N-14), normalized to the internal control (α-tubulin), and expressed as fold-change; the lower section presents a representative crop blot. (c) Immunofluorescent staining of ADSCs (d0) and differentiated adipocytes at day 14 N-14 terminal domains of TNMD (green) and 4.6-diamidino-2-phenylindole (DAPI; blue; scale bar, 200 μm). All values are expressed as the means ± SEM of three independent experiments. Significant differences were identified using the nonparametric Mann-Whitney U test; *P-value < 0.05. 7 Scientific RepoRts | (2019) 9:3979 | https://doi.org/10.1038/s41598-019-40482-0 www.nature.com/scientificreports www.nature.com/scientificreports/ shRNA-control (P-value < 0.05) (Fig.5). However, we did not observe significant differences in the activation of the nuclear factor kappa-light-chain-enhancer of activated B cells (NF-κB) pathway through p65 subunit phosphorylation. Although TNMD could play a role against inflammation in adipocytes, more studies are needed to elucidate the underlying mechanism. Discussion In the present study, we show that X-chromosome TNMD genetic variants are associated with childhood obesity and metabolic alterations in a cohort of Spanish children. Particularly, we show that the tag SNP rs4828038 is associated with anthropometry, glucose metabolism alterations, and increased levels of pro-inflammatory biomarkers in a sex-specific manner. Furthermore, by in vitro gene silencing, we demonstrate that TNMD is required for an adequate glucose and lipid metabolism, and it plays a role in the control of inflammation in cultured human adipocytes. Genetic association studies are commonly focused on autosomal variants, and genetic polymorphisms in sex chromosomes are often neglected25. Notwithstanding, the study of sex chromosomes might help to clarify the role of several genes in the development of many diseases, especially complex human traits that exhibit gender disparity in risk or symptoms28,35,36. Studies that employ mouse models and allow the distinction of gonadal from chromosomal effects have revealed that X-chromosome dosage influences food intake, which in turn affects adiposity and the occurrence of adverse metabolic conditions such as hyperinsulinemia, hyperlipidemia, and fatty liver37. To date, the present study is the first to analyze and detect associations between TNMD X-chromosome genetic variants and obesity and its metabolic complications in a cohort of children. For that purpose, we quantified our power in 96.12% to detect small GWAS-size genetic effects (estimated in F2 = 0.02) at an alpha level of 0.05. Specifically, we found a risky correlation between the rs4828038, tag SNP of the TNMD haploblock-2, and the BMI z-score and WC in boys. Interestingly, findings related to BMI z-score remained statistically significant after applying multiple-test correction. Regarding these anthropometric measurements, others authors such as Tolppanen et al.20, have also reported TNMD associations20. Surprisingly, they have found a protective association between variants located in the haploblock-1 (rs11798018, rs5966709 and rs4828037) and BMI and weight in adult European men. Although these findings are reported for a different haploblock than our associated haploblock-2 SNP, such controversy merits special attention. A possible explanation might rely on the fact that both haploblock-1 and haploblock-2 could elicit contrasting effects on TNMD expression through, for example, the alteration of microRNA targets sites or the generation of different splicing patterns. Other sources of variability might also rely on the fact that we are studying a cohort of children, while Tolppanen et al. focus on adult population with an advanced status of impaired glucose tolerance (IGT) and T2DM. In this regard, further TNMD functional genetic studies are needed to clarify such an issue. Figure 3. TNMD promotes adipogenesis and impairs lipid metabolism in human adipocytes. Human adipocytes were transfected with an adenovirus-5 containing a shRNA-TNMD and shRNA-control (scrambled) at day 14 of adipogenesis induction. (a–c) Peroxisome proliferator-activated receptor gamma (PPARG), CCAAT/enhancer-binding protein alpha (CEBPA) and angiopoietin-like 4 (ANGPTL4) mRNA and protein levels were determined in the shRNA-TNMD and shRNA-control adipocytes. Protein levels in cell lysates were analyzed by Western blotting using specific antibodies against PPARG, CEBPA and ANGPTL4, normalized to the internal control (α-tubulin), and expressed as fold-change; the lower section shows a representative crop blot. (d–f) Hormone-sensitive lipase (HSL) gene expression, adipose triglyceride lipase (ATGL), and perilipin (PLIN) gene expression were determined in the shRNA-TNMD and shRNA-control adipocytes. (g) Glycerol levels (µM) in cell supernatants after treatment with shRNA-TNMD. All values are expressed as the means ± SEM of three independent experiments. Significant differences were identified using the nonparametric Mann-Whitney U test; *P-value < 0.05. 8 Scientific RepoRts | (2019) 9:3979 | https://doi.org/10.1038/s41598-019-40482-0 www.nature.com/scientificreports www.nature.com/scientificreports/ Regarding girls, we showed a protective correlation between our tag SNP rs4828038 and WC and WHR. In both associations, BMI was controlled for as a confounder. Specifically, the association between our tag SNP and WC further stood multiple-test correction (FDR = 2.4%). Interestingly, our findings are in accordance with prior works of Tolppanen et al.20, who detected an association between the rs2073162-A-allele (marker in complete LD with our tag SNP) and smaller horizontal diameters in adult females when adjusting the model by BMI20. Considering all this, we can see how the same region (haploblock-2) appears as a risk factor for obesity in boys, while at the same time, it acts as a protective element for central obesity parameters in girls. Such sex-specific behavior in our study reflects the typical sexual dimorphism of the X-chromosome very well, and it could arise from some X-chromosome particularities including differential gene dosage, the escape from the X-chromosome inactivation (XCI) and the existence of distinct genomic imprint mechanisms37–41. To account for all these X-chromosome particularities, several specific steps and procedures have been implemented following published recommendations25,27–30 (see the method section). Regarding glucose metabolism, several risky correlations were found for the rs4828038 SNP in boys. Particularly, we identified that the rs4828038-T-allele was associated with higher levels of fasting glucose and HOMA-IR, as well as lower values of QUICKI index. Interestingly, QUICKI and glucose associations remained statistically significant also after controlling for BMI confounding. In analyses without BMI-confounding adjustment, QUICKI and glucose insights further reached FDR multiple-test significance. Altogether, these associations are in concordance with previous findings that have been obtained by Tolppanen et al.20 during a 3-year follow-up study. For two TNMD SNPs (the rs2073163 and the rs1155974) in strong LD with our tag SNP, they showed that men carrying the C and T alleles (respectively) presented an altered oral glucose tolerance test in comparison to individuals with the T and C alleles. These markers, along with the rs2073162, were also associated with an increased risk to develop T2DM during a 5-year follow-up study conducted in men20. Considering all this, we could hypothesize that the small alterations detected in the glucose metabolism of our boys according to the TNMD genotypes may be a premature signal of future complications during adulthood such as IGT or even Figure 4. TNMD is involved in glucose metabolism in human differentiated adipocytes. Human adipocytes were transfected with an adenovirus-5 containing a shRNA-TNMD and shRNA-control (scrambled) at day 14 of adipogenesis induction. (a) Glucose transporter 4 (GLUT4) mRNA levels were normalized to those of hypoxanthine-guanine phosphoribosyltransferase-1 (HPRT1), and the data from three independent experiments are presented as the fold-change, which was calculated using the Pfaffl method. (b) GLUT4 protein levels from cell lysates were analyzed by Western blot using a specific antibody against GLUT4, normalized to the internal control (α-tubulin) and expressed as fold-change; the lower section shows a representative crop blot. (c) Glucose uptake levels in shRNA-TNMD-treated adipocytes compared with the shRNA- control or insulin (1 μM, 30 min) as a positive control. (d) Adiponectin (ADIPOQ) mRNA and protein levels expressed as fold-change. (e) Ratio phosphor-AMPKα/total-AMPKα. (f) Ratio phosphor-AKT/total-AKT. (g) Immunofluorescent staining of adipocytes at day 14 with GLUT4 (red) and 4.6-diamidino-2-phenylindole (DAPI; blue; scale bar, 200 μm) in the shRNA-control and shRNA-TNMD-treated adipocytes. All values are expressed as the means ± SEM of three independent experiments. Significant differences were identified using the nonparametric Mann-Whitney U test; *P-value < 0.05. 9 Scientific RepoRts | (2019) 9:3979 | https://doi.org/10.1038/s41598-019-40482-0 www.nature.com/scientificreports www.nature.com/scientificreports/ its progression to T2DM. On this matter, TNMD genetic variants could be potentially useful as early life risk indicators for T2DM in male subjects. For girls we did not observe significant results in any glucose metabolism outcome. Previous studies in adults have reported contradictory findings in this regard. In summary, although some of our genetic observations show a multiple-test level of statistical significance, it is important to stress, however, that they have not been corrected for between-trait multiple-test error. Thus, showed FDR corrected values are not study-wide robust and should be interpreted with caution. On the other hand, although we have taken some steps to account for main X-chromosome particularities, not all available suggestions were possible to incorporate in our study since this is a candidate-gene analysis instead of a GWAS approach. This, along with the fact that previous TNMD studies are statistically weak and barely accounted for X-chromosome specifications20–22, indicates that our study should be viewed as hypothesis-generating instead of a replication approach. On this matter, more detailed characterization in bigger and independent children samples as well as additional follow-up studies during adulthood are needed. According to these results in children, TNMD SNPs are associated with impaired glucose metabolism and we previously found TNMD overexpression in VAT from prepubertal obese children17. Other studies have also described that TNMD expression is highly upregulated in human AT, increased in obesity14–16 and downregulated after diet-induced weight loss13 and that TNMD expression is predominant in adipocytes compared with stromal vascular fraction (SVF) cells18. Furthermore, TNMD expression promotes preadipocyte proliferation and adipogenesis in SGBS adipocytes, and they improved insulin sensitivity in Tnmd transgenic mice, which suggesting the protective role of TNMD in VAT to alleviate insulin resistance in obesity18. Consistent with these results, TNMD knock-down led to lower gene expression and protein levels of important transcription factors that are involved in adipogenesis such as PPAR-γ and C/EBP-α. On the other hand, as TNMD is expressed in dense connective tissues as tendons and ligaments, and the C-terminal domain could be processed as a soluble factor, this fraction could reach the adipose tissue and promote the adipogenic differentiation in vivo. However, further studies are needed to clarify this possible effect. Therefore, our results confirm the fact that TNMD promotes adipogenic differentiation, and it could be implicated as a protective factor that contributes to AT expansion. The reduced lipolysis observed in TNMD-knocked-down adipocytes could be explained by the reduced gene expression and protein levels of PPAR-γ, as well as by the reduced AMPK activation because AMPK is the master regulator of metabolism. Indeed, it has been described that PPARG2-knocked-out adipocytes exhibit reduced lipolysis42; this could be explained by the lower expression of HSL43 and ATGL44, since they are both transcriptional targets of PPAR-γ. On the other hand, it has been reported that ANGPTL4 promotes the expression of genes involved in lipolysis in adipocytes45, and since ANGPTL4 gene and protein levels were significantly downregulated when TNMD was inhibited, the results indicate that TNMD could be directly associated with lipid metabolism through ANGPTL4. The association found between TNMD SNPs and fasting glucose levels in this study in children together with the higher 2-hour plasma glucose levels that was found in adults suggested a potential role for TNMD in adipocyte glucose metabolism. The confirmation of this hypothesis is another key finding in this study. We Figure 5. TNMD triggers inflammation in human differentiated adipocytes. (a) mRNA expression and protein levels of interleukin 1-β (IL-1B); (b) mRNA expression and protein levels of tumor necrosis factor-α (TNF-α), and mRNA levels were normalized to those of hypoxanthine-guanine phosphoribosyltransferase-1 (HPRT1); TNF-α and IL1B protein levels were analyzed by XMap technology (Luminex) as indicated in the methods section. (c) Phospho-NFκB p65 protein levels were analyzed by Western blot using a specific antibody against phospho-NFκB p65, normalized to the internal control (α-tubulin), and expressed as the fold-change. The lower section shows a representative crop blot. The data from three independent experiments are presented as the means ± SEM. Significant differences were identified using the Mann-Whitney U test; *P-value < 0.05. 16 Scientific RepoRts | (2019) 9:3979 | https://doi.org/10.1038/s41598-019-40482-0 www.nature.com/scientificreports www.nature.com/scientificreports/ 78. Ritchie, M. E., Liu, R. & Carvalho, B. S. Australia and New Zealand Multiple Sclerosis Genetics Consortium (ANZgene), T. A. and N. Z. M. S. G. C. & Irizarry, R. A. Comparing genotyping algorithms for Illumina’s Infinium whole-genome SNP Bead Chips. BMC Bioinformatics 12, 68 (2011). 79. Ling, H., Hetrick, K., Bailey-wilson, J. E. & Pugh, E. W. Application of sex-specific single-nucleotide polymorphism filters in genome-wide association data. BMC Proc. 5, 1–5 (2009). 80. Clayton, D. Testing for association on the X chromosome. Biostatistics 9, 593–600 (2008). 81. Clayton D. snpStats: SnpMatrix and XSnpMatrix classes and methods. R package version 1.32.0. (2018). 82. Kay, M. A., Glorioso, J. C. & Naldini, L. Viral Vectors for Gene Therapy. Nat Med 7, 33–40 (2001). 83. Pfaffl, M. W. A new mathematical model for relative quantification in real-time RT-PCR. Nucleic Acids Res. 29 (2001). 84. R Core Team. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. Available online at https://www.R-project.org/ (2018). Acknowledgements This work was supported by Plan Nacional de Investigación Científica, Desarrollo e Innovación Tecnológica (I + D + I), Instituto de Salud Carlos III-Fondo de Investigación Sanitaria (Projects numbers PI020826, PI051968, PI1102042, PI1600871), RETIC (Red SAMID RD12/0026/0015), Fondo Europeo De Desarrollo Regional (FEDER) and the Junta de Andalucía (project number CTS-6770); Secretaría General de Universidades, Investigación y Tecnología. Consejería de Economía, Innovación y Ciencia). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. This paper will be part of Augusto Anguita-Ruiz’s doctorate, which is being performed under the “Nutrition and Food Sciences Program” at the University of Granada. Author Contributions Responsible for child recruitment and anthropometric measures: R.V.-C., M.G.-C., G.B., R.L., R.C. and L.M. Conception and design of experiments: F.J.R.-O., C.G.-L., A.G. and C.M.A. Performed in vitro experiments: F.J.R.-O. Performed the SNP analysis results: A.A.-R. and A.I.R. Analyzed the results: F.J.R.-O., A.A.-R., A.I.R., J.O., C.G.-L., A.G. and C.M.A. Wrote the manuscript: F.J.R.-O., A.A.-R. and C.M.A. Additional Information Supplementary information accompanies this paper at https://doi.org/10.1038/s41598-019-40482-0. Competing Interests: The authors declare no competing interests. Publisher’s note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/. © The Author(s) 2019