Discussion

Considering severe early onset obesity might enrich more low-frequency deleterious variants, we performed whole-exome sequencing and association study on obesity in Chinese children. We found two missense mutations (rs1059491 and rs189326455) are significantly associated with common obesity. The rs1059491 is not only associated with BMI, but also in association with FMI, FMP, FFMI, and LDL. It might result in obesity by affecting the expression level of target genes other than alternating protein structure because it is predictively unharmful to protein structure but acts as an eQTL that can dramatically change the expression of several obesity-related genes evidenced by transcriptome sequencing.
Previous GWAS studies have identified multiple SNVs associated with increased BMI(Cotsapas et al., 2009; Locke et al., 2015; Speliotes et al., 2010; Thorleifsson et al., 2009). We noted that our most significant SNV, rs1059491, was located in a previously reported obesity locus with lead SNV of rs7359397 which was identified in European(Speliotes et al., 2010). However, a fine mapping study on the locus failed to find that rs1059491 was in association with obesity in European children and adolescents(Volckmar et al., 2015). Explanations for such contradiction are as follows three reasons. Firstly, although rs1059491 is highly linked (r2 > 0.75) to rs7359397 in European, no linkage disequilibrium (r2 = 0.00029) was observed between them in Chinese. Secondly, the allele frequency of rs1059491 varies drastically among different populations, which has been exemplified in 1,000 Genomes Project (1KG) (Supplementary Figure 2 ). Minor allele frequency was 0.3434 in European while was around 0.1 in Asian populations. In our findings, minor allele frequency was 0.0421 in the control group, which was similar to that of 0.0571 in southern Han Chinese in 1KG. Thirdly, our conditional analysis found that rs1059491 (conditional P=7.13E-21) was independently associated with obesity in Chinese children when controlling rs7359397. Taken together, rs1059491 was a population specific obesity-associated SNV and significantly associated with children obesity in Chinese.
The rs1059491 was a conserved missense mutation and was predicted as deleterious (PolyPhen score = 1), while our findings show that it might result in obesity through altering gene expression other than affecting protein structure. Chip-seq have revealed that rs1059491 resided in transcription factors binding sites of PPARγ2 and RXRA in ENCODE datasets(Kheradpour & Kellis, 2014). Irregular behavior of PPARγ2 and RXRA can result in abnormal lipid metabolism which lead to obesity(Akyurek et al., 2013; Lima et al., 2013; Stossi et al., 2019; Takahashi, Morita, Yokoyama, Suzuki, & Yamamoto, 2012). Therefore, rs1059491 might affect the binding affinity to obesity-related transcription factors and subsequently alter obesity-related gene’s expression. Our transcriptome data and GETx data have shown that rs1059491 is an eQTL affecting the expression of many genes, such asSULT1A2 , ATXN2L , and TUFM . It’s worthy to note thatSULT1A2 and ATXN2L were strongly implicated in obesity(Voisin et al., 2015; Volckmar et al., 2015).SULT1A2is up-regulated in adipose tissues of individuals carrying G allele of rs1059491, and high expression level of SULT1A2 is a characteristic not only in lipid uptake tissues as duodenum and intestine, but also in fat tissues (GTEx database). On the other hand, rs1059491 regulates ATXN2L on translational level, and subsequently affects its interaction partner ATXN2 , the down-regulation of which would likely lead to increased body size and fat levels in Caenohabditis elegans (Bar et al., 2016; Kaehler et al., 2012), and ATXN2 knock-out mice show adult-onset obesity(Kiehl et al., 2006; Lastres-Becker et al., 2008; Meierhofer, Halbach, Sen, Gispert, & Auburger, 2016). In brief, the most likely mechanism of rs1059491 association with obesity is expression regulation of obesity-related genes on the transcription level and post-transcription level (Figure 4 ).
We also found that rs189326455 was another variant significantly associated with obesity. It located in the SH3 domain of MAP3K21, which was responsible for controlling or regulating protein-protein interactions in the signal transduction pathways, such as cytoplasmic signaling(Koch, Anderson, Moran, Ellis, & Pawson, 1991; Schlessinger, 1994). MAP3K21 is a member of the mixed lineage kinases family that acts as a specific modulator to inhibit lipopolysaccharide (LPS)-induced activation of c-Jun N-terminal kinase (JNK), or extracellular signal-regulated kinase (ERK)(Craige, Reif, & Kant, 2016), or as a negative regulator of Toll-like receptor 4 (TLR4) signaling(Seit-Nebi, Cheng, Xu, & Han, 2012). It’s well known that JNK, ERK, and TLR4 are all implicated with obesity through elevating gene expression level(Ahmad et al., 2012; Hirosumi et al., 2002), promoting insulin resistance(Hirosumi et al., 2002; Ozaki et al., 2016; Pal et al., 2012; Solinas & Becattini, 2017), or stimulating adipogenic differentiation(Gu et al., 2015). It’s worthy to note that rs189326455 is also suggestively (P = 5.63E-07) associated with blood triacylglycerol levels which is highly associated with insulin resistance(Ormazabal et al., 2018). Thus, we speculate that rs189326455 might alter the affinity of SH3 domain and subsequently affects interaction between MAP3K21 and mis-activating down-stream signaling pathway proteins, such as JNK, ERK, and TLR4 (Figure 4 ), and results in insulin resistance and obesity(Seit-Nebi et al., 2012).
In this study, we investigated causal variants in severe early onset obesity by taking advantage of that child obesity is less affected by cumulative environmental factors and the heritability is higher compared with adulthood obesity. We identified two potential variants related to obesity with whole exome sequencing. We found that rs1059491 was an eQTL which might result in obesity through altering the expression of obesity-related genes, and rs189326455 might affect the regulation of MAP3k21 to subsequent obesity-related genes. Although the speculation should be proved by more functional experiments, our study is among the first to use exome sequencing technology to identify functional variants of both common obesity and abdominal obesity in Chinese children. Chinese people are more likely to develop abdominal obesity, which has been associated with a higher risk of developing type 2 diabetes and cardiovascular disease(Thomas et al., 2004). Therefore, identification of the SNVs associated with abdominal obesity is of great impact on prevention and control of adiposity-based chronic diseases.
Acknowledge. This work was supported by grants from the Beijing Natural Science Foundation Program and Scientific Research Key Program of Beijing Municipal Commission of Education(KZ202010025039) and National Natural Science Foundation of China (81973110, 81502872, and 31671312)
Author Contributions. K.M., M.Z., and J.C. wrote the manuscript, researched, analysed and interpreted the data, contributed to the discussion, and reviewed/edited the manuscript. J.M., and Y.Z. designed the study, and made critical revisions of the manuscript. C.M. revised the figures. M.Z., X.Z., L.G., and L. F. were involved in the sample collection, selection and preparation. H.C. collected the phenotype data. W.L., and Z.Z. designed the bioinformatics and experimental sections, coordinated the data collection, maintained project procedure and carried out bioinformatics analysis. M.Z. and L.G. performed the genotyping analysis. X.X and X.S. made critical revisions of the manuscript. All the authors contributed to the final paper.
Duality of Interest. The authors have nothing to declare.
Data Availability Statement. The data that support the findings of this study are available from the corresponding author upon reasonable request.