Pathway-based metabolomics study of sarcopenia-related traits in two US cohorts

Introduction

Sarcopenia, characterized by the age-related loss of muscle mass and strength, is a major contributor to the risk of physical frailty, loss of independence, and hospitalization with poor health outcomes in aging older adults [1]. It has become a major public health challenge with the rapid expansion of the world’s older population and is responsible for substantial healthcare expenditure [2]. Sarcopenia was recognized as an independent condition with an assigned ICD-10 code in 2016, which is a milestone for the increased awareness of its importance in human health [3]. Age-related sarcopenia is a multifactorial disease, and some possible causes have been suggested, such as decreased nerve input, protein intake, and physical activity [4–7]. However, the biological mechanisms underlying the development of sarcopenia are still largely unknown, and no specific drugs have been approved for the treatment of sarcopenia [1].

Metabolomics is an emerging technology to comprehensively profile small molecules in biofluids, cells, and tissues and has contributed to our understanding of a number of disorders [8–11], such as diabetes, cardiovascular disease, and cancer. Metabolomics methodologies fall into two distinct groups, untargeted and targeted metabolomics, each with its own inherent advantages and disadvantages. Untargeted metabolomics is the comprehensive analysis of all the measurable analytes in a sample. It is especially suitable for diseases/conditions with very limited information about its metabolic mechanisms. By contrast, targeted metabolomics is the measurement of defined groups of chemically characterized and biochemically annotated metabolites [12]. Although most studies have used either the untargeted or targeted approach, it is reasonable, or even ideal, to use the untargeted metabolomics approach to prioritize metabolic pathways (hypothesis generation) for further comprehensive targeted metabolomics analysis in additional samples which can validate findings of the untargeted analysis and discover more disease-related metabolic changes.

To investigate the mechanisms of sarcopenia, we have conducted an untargeted metabolomics study of sarcopenia-related traits, muscle mass and strength, in a cohort of young white women. In the pathway analysis, two amino acids, aspartic acid and glutamic acid, were both mapped to five metabolic pathways [13]. To further validate the associations of these two metabolites with muscle mass (measured by body mass index-adjusted appendicular lean mass, ALM/BMI) and muscle strength (measured by hand grip strength, HGS) and identify additional trait-associated metabolites, we designed a targeted metabolomics assay to assess the key metabolites from these pathways in the original discovery cohort of white young women and a validation cohort including older black and white as well as men and women.

Results

The clinical characteristics of the study participants in the discovery and validation cohorts are shown in Table 1. Across all the sex and race subgroups, the participants of the validation cohort were older, had a higher rate of current smoking, and had less alcohol drinking compared to those of the discovery cohort. The validation cohort had similar levels of ALM/BMI but smaller hand grip strength, especially among white and black women, than the discovery cohort.

Table 2 shows the metabolites selected and measured from the five metabolic pathways involving aspartic acid and glutamic acid. Twelve out of the 15 metabolites on the targeted liquid chromatography-mass spectrometry (LC-MS) assay were successfully detected in the study samples. Figure 1 shows the correlation coefficients between the 12 detected metabolites in the discovery and validation cohorts, respectively. In general, the correlations among the metabolites were very different between the subjects in the discovery cohort and the validation cohort. Additionally, the correlations of metabolites varied substantially among different race- and sex-subgroups as well (Figure 2). These different correlations of metabolites within each metabolic pathway among different age, race, and sex groups may suggest the functions of the pathways probably vary across age, sex, and race as well.

The multivariate analysis showed that aspartic acid, identified in our previous untargeted metabolomics study, was still significantly associated with sarcopenia traits in the discovery cohort, but not in the validation cohort (Table 3). Higher aspartic acid levels were associated with lower ALM/BMI and HGS (Table 4). However, glutamic acid, another metabolite identified in our previous untargeted metabolomics study, was significantly associated with sarcopenia traits in both the discovery and validation cohorts even after adjusting for multiple testing (FDR q-value < 0.05) (Table 3). Specifically, glutamic acid was associated with ALM/BMI (P = 0.011) in the discovery cohort but associated with both ALM/BMI (P = 0.024) and HGS (P = 0.001) in the validation cohort. Glutamic acid was negatively associated with ALM/BMI in both the discovery and validation cohorts and negatively associated with HGS in the validation cohort (Table 4). We did not observe any race- or sex-specific effects of glutamic acid on the sarcopenia traits in the validation cohort (P values for the interaction terms of the metabolite with race and sex > 0.05) (Table 3).

N-Acetyl-L-aspartic acid (NAA) and carnosine were two potential cohort-specific metabolites associated with HGS. NAA was significantly associated with HGS in the discovery cohort (P for joint analysis = 0.006; effect = 2.2071 and P = 0.002 for HGS), and carnosine was suggestively associated with HGS in the validation cohort (P for joint analysis = 0.043; effect = -1.0917 and P = 0.038 for HGS) (Tables 3, 4). There were also suggestive (not significant after adjusting for the multiple testing) sex-specific effects of both NAA (P for interaction with sex = 0.046) and carnosine (P for interaction with sex = 0.022) on the sarcopenia traits in the validation cohort (Table 3). Especially, the effects on HGS were opposite in women and men for both N-acetyl-L-aspartic acid and carnosine (Table 4).

Discussion

This targeted metabolomics study replicated the study findings of a previous untargeted metabolomics study, the association of glutamic acid with sarcopenia traits, not only in the original discovery cohort (for the previous untargeted metabolomics study) but also in another independent validation cohort. Although we did not replicate the association of aspartic acid, which was another metabolite identified in our previous untargeted metabolomics study, with sarcopenia traits in the validation cohort, we did replicate it in the discovery cohort using the targeted metabolomics method. Additionally, we identified two novel metabolites, NAA and carnosine, which were potentially associated with the sarcopenia trait, muscle strength, by measuring more metabolites from the metabolic pathways involving glutamic acid and aspartic acid.

Although untargeted metabolomics analysis has its advantages in discovering novel biomarkers without the need for a priori metabolic hypothesis over targeted metabolomics analysis, it still has many challenges or pitfalls, especially compound identification [14, 15]. In addition to a large proportion of unknown metabolites, the structures of metabolites identified through existing MS databases still need further verification through comparing with authentic standards tested using the same instrument as study samples, a “gold standard” for determining metabolite identities [16]. Also, the structure verification of disease-related metabolites is necessary for subsequent functional studies for the ultimate translational goal of metabolomics study findings. Therefore, targeted metabolomics analysis designed using chemical standards of metabolites identified in untargeted metabolomics studies will help to further validate the metabolite identities and their associations with traits/diseases. In this study, we also included additional metabolites from the disease-related metabolic pathways found in the previous untargeted metabolomics study in the targeted metabolomics assay to further identify novel biomarkers for sarcopenia traits.

In this study, we replicated the associations of aspartic acid and glutamic acid with sarcopenia traits in the discovery cohort of young white women using the new targeted LC-MS assay. Glutamic acid was also associated with sarcopenia traits in the new validation cohort of older subjects, including both black and white and men and women. These findings might suggest glutamic acid influences these sarcopenia traits across age, sex, and race. However, aspartic acid might be a sarcopenia trait-related metabolite that is specific to younger individuals. Aspartic acid and glutamic acid are not essential amino acids but among the 20 proteinogenic amino acids. Aspartic acid can be made from glutamic acid by enzymes using vitamin B6. Both aspartate and glutamate (the anions of the amino acids) are major excitatory neurotransmitters [17], and glutamate also plays roles in the skeletal neuromuscular junction. Participating in modulating cholinergic transmission and plastic changes [18]. Glutamate in skeletal muscle also participates in various metabolic pathways, such as glutathione synthesis, insulin production, tricarboxylic acid cycle, and purine nucleotide cycle [19]. Additionally, glutamic acid can be converted to γ-aminobutyric acid (GABA) by the enzyme glutamic acid decarboxylase. GABA is the most abundant inhibitory transmitter in the brain. Oral supplementation of GABA has been reported to increase growth hormone and muscle protein synthesis, potentially contributing to dynamic protein turnover [20]. The association of glutamic acid with muscle mass we identified was in line with the finding in a sample of UK women [21]. However, we added further evidence for its association with muscle strength in humans as well. In addition, animal studies have shown that aspartate inhibits inflammation-induced muscle loss through regulating phosphorylation of Akt, AMPKα, and FOXO1 [22, 23]. However, we identified aspartate were negatively associated with muscle mass and strength in humans. These inconsistent findings warrant more studies to clarify its role in muscle regulation.

NAA is a derivative of aspartic acid and the second most concentrated molecule in the brain after the amino acid glutamic acid. NAA is essential for normal brain operation and declines in several neurodegenerative and neuropsychiatric diseases [24]. Also, NAA is reduced in the aging spinal cord that contributes to loss of innervation and downstream degenerative processes that occur in skeletal muscle [25]. NAA is synthesized in neuronal mitochondria but can efflux from the brain to the circulation. It has been observed that serum levels of NAA decreased with aging [26], suggesting its effects on health conditions/diseases might vary with age. This might partially explain why it was associated with muscle strength only in the discovery cohort of young subjects, but not in the validation cohort of older subjects in this study.

Carnosine is a dipeptide of the amino acids beta-alanine and histidine with a high concentration in mammalian skeletal muscle. In skeletal muscle cells, carnosine can be synthesized by carnosine synthase from beta-alanine and histidine. Interestingly, muscle carnosine loading leads to improved performance in high-intensity exercise in both untrained and trained individuals [27]. Also, carnosine has the potential to suppress many of the biochemical changes that accompany aging, such as protein oxidation, glycation, and cross-linking, and associated pathologies [28]. In this study, we observed a suggestive sex difference in the effects of plasma carnosine levels on muscle strength among older subjects. Carnosine was positively associated with muscle strength in women, but negatively in men. Blood carnosine is partially influenced by dietary factors, such as intakes of carnosine and beta-alanine [29]. Therefore, reverse causation might be an explanation for the negative association between carnosine and muscle strength in men. For example, carnosine supplementation was taken among the elderly with reduced muscle function.

Our study has several strengths. First, we implemented a novel strategy of metabolomics study, which was to conduct targeted metabolomics analysis following up the study findings of an untargeted metabolomics study. Second, a pathway-based approach was used in the targeted metabolomics analysis to further identify sarcopenia trait-related metabolites. Third, two diverse study cohorts were included to identify common and specific outcome-related metabolites for different age, race, and sex subgroups. Finally, we investigated two major sarcopenia traits, muscle mass and strength, at the same time instead of muscle mass only which most of the previous studies focused on. In fact, muscle strength is better than muscle mass in predicting adverse outcomes [30]. Nevertheless, our study also has some limitations. The sample size for each race-sex subgroup in the validation cohort was relatively small, especially for white men and black women, which might have limited the study power for examining the potential race and sex differences in metabolites associated with sarcopenia traits. Although we have used comprehensive pathway analysis to select metabolites from the metabolic pathways of interest for the targeted metabolomics analysis, it is still possible that we have missed some important metabolites that were not tested in the targeted LC-MS assay.

In conclusion, this targeted metabolomics study replicated some findings from the previous untargeted metabolomics study of sarcopenia traits and also identified novel metabolites associated with sarcopenia traits. Our findings suggest that glutamic acid might be a risk factor for sarcopenia traits across all age, sex, and race groups. However, the effects of aspartic acid, NAA, and carnosine might vary with age. The biological mechanisms underlying the relationship between these metabolites and sarcopenia traits still need to be clarified in future studies.

Materials and Methods

Supplementary Materials

Supplementary Table 1