Gender-dependent metabolomics in children and adults
A total of 291 identified metabolites were subjected to further analysis (Table S1). Gender-differences were assessed using PCA and OPLS-DA models. Scatter plots showed that gender differences in the urine metabolites of adults are more apparent than in children (OPLS-DA: children: R2Y = 0.599, Q2 = 0.379 ; adults: R2Y = 0.928, Q2 = 0.768. Fig. S1b and S1c). Metabolites with VIP values above the threshold value 1 and p values below the significance threshold 0.05 were considered gender dependent. Overall 42 metabolites were found to be gender-dependent in children, and 98 metabolites in adults (Fig. 2a and 2b, Tables S2 and S3). A total of 17.9% (21) of these metabolites showed gender differences in both children and adult populations with the same change trend in male/boy and female/girl (Fig. 2c). Guanidoacetic acid, 5-hydroxyindoleacetic acid, dopamine, 5'-methylthioadenosine and indoleacrylic acid showed higher levels in females. The metabolites deoxyinosine, cotinine glucuronide, dopamine glucuronide and L-formylkynurenine showed higher levels in males. Gender differences of these common differential metabolites in the adult population were larger than those in the children population (Fig. 2c). Specifically, the metabolites cortisol, uric acid, 18-hydroxycortisol, deoxycholic acid and glycine conjugate were found to be gender-dependent only in the children population. Metabolites of deoxyuridine, pantothenic acid, riboflavin, and 3-hydroxytetradecanedioic acid were gender-dependent only in the adult population. These metabolites suggested differential metabolic status between children and adults.
Pathway enrichment analysis would provide an overview of gender-dependent metabolism status in children and adults (Fig. 2d and 2e). Consistent with previous studies (Chiu et al., 2016; Fan et al., 2018; Liu et al., 2018), amino acid metabolism, including tryptophan metabolism and arginine and proline metabolism, showed gender differences in both children and adults. In females/girls, tryptophan metabolites, indoleacrylic acid, indolylacryloylglycine and 5-hydroxyindoleacetic acid showed higher levels compared with males/boys. 5-Hydroxyindoleacetic acid is a breakdown product of serotonin that is excreted in the urine and is mainly involved in serotonin degradation and serotonin receptor signaling regulation. The higher level in females could be partly explained by greater precursor, such as tryptophan, availability (Akhmadeev and Kalimullina, 2013). Studies on experimental animals have revealed the effects of gonadal hormones on the indoleamine system, and in the rat brain, 5-hydroxyindoleacetic acid is higher in females than in males (Giulian et al., 1973). These results were consistent with our present urine results. Additionally, children and adults showed different gender-dependent metabolic statuses. In the child population, purine metabolites, steroid metabolites and amino acid metabolites showed specific gender dependence. For example, uric acid, an oxidation product of purine, showed a significantly higher level in boys compared with girls but showed no gender variations in adults. Renal excretion of uric acid in children differs from that in adults. It was reported that the younger the child, the greater the excretion of uric acid (Baldree and Stapleton, 1990). Altered urine uric acid level is an indispensable marker in detecting rare inborn errors of metabolism (Jasinge et al., 2017). The level of uric acid in urine was higher in boys than girls, probably contributing to the higher prevalence of hyperuricemia in boys than in girls (Niegawa et al., 2017). These results indicated the commonness and differences of gender-dependent metabolism characteristics in children and adults, probably resulting from different physiological characteristics, dietary habits or other environmental factors 8. The detailed gender-associated pathways in each age stage are listed in Table S4.
Age-dependent metabolomics in children and adults
Age is another important factor in metabolism status. PCA was firstly performed, showing apparent separation of different groups (Fig. S2a). To determine metabolites related to age, PLS-DA modeling was performed on urine metabolite profiles of populations with different age stages. Since the metabolism status of the population was different between sexes, the analyses for age in children and adults were performed separately for males and females.
In the children population, the PLS-DA three-component score plot showed clear associations of metabolite profiles with age in boys (R2Y = 0.719, Q2 = 0.528, Fig. 3a, Fig. S2b), indicating significant metabolism differences among different age stages. The same trend was observed in girls (R2Y = 0.78, Q2 = 0.564, Fig. 3b, Fig. S2b). In both boys and girls, two clusters of metabolites were found, one cluster with metabolites decreasing with age (Component 1); the second cluster with metabolites showing the highest level in the primary school aging 7 to 12 (Component 2) (Fig. 3c). The first cluster contributes most to age variation in both girls and boys. In the adult population, clear separations were also observed for the three age groups in males and females (PLS-DA-male: R2Y = 0.652, Q2 = 0.17; PLS-DA-female: R2Y = 0.583, Q2 = 0.362, Fig. S2b, Fig. S2c and Fig. S2d). Similar to children, metabolites decreasing with age contributed most to age differences (Fig S2e). According to the significance threshold, a total of 98 and 76 metabolites were selected as age-dependent in boys and girls, respectively. For the adult population, 55 and 80 metabolites were age-dependent in males and females, respectively (Table S5, S6, S7 and S8). These metabolites were further submitted for further pathway analysis.
For children and adults, we found many common age-dependent pathways between boys (males) and girls (females), indicating common metabolism status variations with age in humans. For instance, pantothenate and CoA biosynthesis, fatty acid biosynthesis and tryptophan metabolism were found to be changed with age in both children and adults. These pathways correspond to energy demand changes with aging (Chiu et al., 2016). Interestingly, tryptophan metabolism was only found to be age dependent in boys and males, while no significant changes were found in girls and females. Metabolites of tryptophanol and 5-methoxytryptophan were reported to be associated with increased cellular anti-inflammatory and blood circulation properties (Cheng et al., 2012), probably reflecting age-associated metabolism activity differences between genders. In addition, we found that the pathway of steroid hormone biosynthesis was changed with age in adults and children. As a whole, steroid hormone biosynthesis showed high activity during the adolescent and youth stages, corresponding to sexual development (Steensma et al., 2013).
Additionally, children and adults showed different age-dependent metabolic statuses. Particularly in the children population, the fatty acid biosynthesis pathway was found to be age dependent. The results showed a positive correlation with increasing age. Fatty acids constitute a large energy source for the body. Increased fatty acid metabolism indicated high ATP generation with age in a children population (Wakil and Abu-Elheiga, 2009). In adults, pyrimidine metabolism and caffeine metabolism were found to be age dependent. Pyrimidine metabolism was found to be positively correlated with aging, showing the highest level in the old adults. Deoxyuridine, a naturally occurring nucleoside, is considered to be an antimetabolite that is converted to deoxyuridine triphosphate during DNA synthesis. Disturbance of DNA synthesis may modulate the aging process and contribute to the high incidence of cancer with aging (Kirsh et al., 1986). The detailed age-dependent metabolism pathways for males (boys) and females (girls) are listed in Table S4.
Metabolic characteristics for gender and age
According to the above results, the urine metabolites of children and adults showed both gender and age dependent effects. Furthermore, we detailed the metabolic differences affected by both gender and age. The p values of OPLS-DA for gender separation during each stage were used to evaluate the significance of gender difference. The results showed a parabolic trend of gender differences during life span, with less significance in the pre- and primary school stages, high significance during the secondary school, youth and middle stages, and less significance during the old stage (Fig 4).
Second, we examined the main metabolic characteristics marked with metabolites with highest intensity for each age group in males (boys) and females (girls) (Fig 4 and Table 2).
Metabolomic characteristics during the pre- and primary school stages During early life in the preschool and primary school stages, gender differences were relatively smaller than in other age stages, as shown in Fig 4. During the preschool stage, pathways of pantothenate and CoA biosynthesis, pyrimidine metabolism, vitamin B6 and alanine metabolism showed high activity in girls and boys. These active pathways were associated with energy and nutrient supply. These metabolic characteristics correspond to the physiological characteristics during this life stage, high metabolism activity for rapid growth and development demands (Chiu et al., 2016). Research on children aged from 2-7 years old suggested that pantothenate and CoA biosynthesis, pyrimidine metabolism and vitamin B6 metabolism were significantly related to autism spectrum disorder (ASD) (Gevi et al., 2016). These results highlight the importance for these pathways to maintain homeostasis during preschool.
During primary school stage, the most active metabolism pathway is tryptophan metabolism, lipid metabolism, nicotinate and nicotinamide metabolism and histidine metabolism. These metabolic features were corresponding to the main physiological characteristic-visual development, blood circulation increasing and rapid metabolism activity during this stage. Although gender differences were small during the primary school stage, metabolites with the highest level in boys and girls showed specific features. In boys, tryptophan metabolites, such as 5-methoxytryptophan, were found to show the highest level in boys. 5-Methoxytryptophan is an endogenous tryptophan metabolite with anti-inflammatory properties (Cheng et al., 2012). In addition, 5-methoxytryptophan was reported to be involved in the cyclic metabolism of the retina (Leino and Airaksinen, 1985), ventricular remodeling and maintaining liver function (Rossignol and Frye, 2011; Lin et al., 2016). In girls, urocanic acid, a breakdown (deamination) product of histidine, showed the highest level. Urocanic acid is one of the essential components of human skin (Wezynfeld et al., 2014). It could accumulate in the epidermis and may be both a UV protectant and an immunoregulator. A higher level of these metabolites in girls may contribute to skin development during this age stage. It was reported that skin disorders are more common among girls than boys aged 6 to 17 years (Sula et al., 2014), which could be affected by the immunomodulatory effects of urocanic acid (Finlay-Jones and Hart, 1997).
Metabolomic characteristics during adolescence and youth During adolescence and the youth stage, gender differences become more significant, partly due to changes in hormone and endocrine levels. The main metabolic feature during these stages was fatty acid oxidation and biosynthesis, androgen and estrogen metabolism and steroidogenesis. These metabolic characteristics correspond to pubertal development, neurodevelopmental changes and heightened stress sensitivity during adolescence and youth stages. Cortisol, androstenol, testosterone, and their glucuronide metabolites showed higher levels during this period. Cortisol is the main glucocorticoid secreted by the adrenal cortex and is involved in the stress response. Synergies between cortisol reactivity and testosterone were reported to influence antisocial behavior in young adolescence. The youth with high diurnal testosterone, combined with high or moderate cortisol reactivity, were significantly higher on antisocial behavior and attention behavior problems (Susman et al., 2017).
In addition to the common metabolic features during adolescence and the youth stage, boys and girls also showed specific metabolic characteristics. During adolescence, spermidine biosynthesis was higher in boys. One of the involved metabolites was 5-methylthioadenosine, a byproduct of polyamine synthesis in DNA turnover cycles that increases with inflammation to modulate cellular stress. It has been shown to influence the regulation of gene expression, proliferation, differentiation, and apoptosis (Avila et al., 2004). Higher serum levels of 5-methylthioadenosine have been reported in youth boys when compared to girls (Perng et al., 2017). A direct association between 5-methylthioadenosine and high metabolic risk was found in boys, possibly driven by proinflammatory pathways (Guasch-Ferre et al., 2016). While in adolescent girls, fatty acid oxidation and biosynthesis showed high activity. Acylcarnitines showed the highest level, indicating high activity of carnitine acetyltransferase in mitochondria (Zammit et al., 2009). These results indicated the preferred metabolic fuel-from fatty acid oxidation in adolescent girls (Zammit et al., 2009).
In the youth males aged 20 to 30 years, linoleic acid metabolites showed higher levels in addition to steroid metabolism. The involved metabolite eicosapentaenoic acid serves as the precursor for prostaglandin-3. It could enhance the production of the cytoprotective prostanoid 15d-PGJ2 (Davidson et al., 2013), which corresponds to the elevated prostaglandin production in youth males (Pace et al., 2017). In contrast to youth males, arachidonic acid metabolites showed higher levels in youth females. It was reported that in females, arachidonic acid metabolism could rescue anti-inflammatory and butyrate-producing microbiota, then upregulate GPR41 and GPR109A and control hypothalamic inflammation (Zhuang et al., 2017).
Metabolomic characteristics during the middle and old stages During the middle age stage, males and females showed the most significant gender differences. In males, the main metabolic features were vitamin B6 and purine metabolism. 8-Hydroxy-7-methylguanine, a methylated purine base, showed higher levels in males. High methylated purine bases were found in tumor-bearing patients compared to healthy controls (Morris et al., 1986). Urine alkylated purines were partly derived from covalently bound adducts in DNA formed from exposure to carcinogenic alkylating agents (Choi and Guengerich, 2006). Purine disorders may be associated with serious, sometimes life-threatening consequences. In females, the metabolism pathway of steroidogenesis and caffeine metabolism showed high activity. Menopausal symptom are an unavoidable problem in females during this period, which could contribute to some metabolic disorders. Caffeine metabolism was reported to be associated with menopausal symptoms, particularly vasomotor symptoms (Faubion et al., 2015).
For the stage aged above 50 years, the gender difference decreased. During this period, organ metabolism activity gradually slows down. Energy-supply metabolism pathways, such as fatty acids and amino acids, showed low levels. In contrast, steroidogenesis, caffeine and pyrimidine metabolism showed high levels. Higher levels of pyrimidine metabolites in the old stage have adverse effects on health (Choi and Guengerich, 2006). Additionally, several cognitive impairment-related metabolites, including acetylhistidine and steroid hormones, were found to be higher in the old population, partly contributing to the high incidence of cognitive impairment diseases at older ages (Bressler et al., 2017).
In conclusion, we provide a comprehensive view of human metabolism status across the life span. To our knowledge, this is the first systematic study to analyze metabolism characterization based on population across a considerable age range. This study showed that gender differences existed from early life stages, and these differences were much smaller than those in adults. Age is another recognized confounder. Metabolism characteristics for each age group could reflect the metabolism status during different life stages and possibly contribute to some age-dependent disease incidences. Our present study would be helpful to understand the age- and gender-dependent metabolism differences, which will be a critical component for the development of metabolomics-based systems biology as a population screening and precision medicine research.
Additionally, several limitations still exist and need to be further validated in the future. First, urine samples of children and adults were collected from two different hospitals. Although sampling operation is strictly controlled, batch effects resulting from center differences could not be completely eliminated. Considering batch effects, data from children and adults were first analyzed separately and then a cross comparison was performed, resulting in an age gap between second school and the youth stages. Second, in the children population, each stage could show specific features for their rapid development. However, due to sample size limitations, we provide an overview of the average metabolism status over a period of 6 years. Third, the influences of diets and circadian rhythm on urine metabolomics could not be completely eliminated, though all subjects were from the same region. For future validation analysis, these influences should be systematically evaluated and analyzed.