https://orcid.org/0000-0003-1048-6932
Figures
Abstract
Purpose
To analyze the causal relationship between 486 human serum metabolites and the active tuberculosis (ATB) in European population.
Methods
In this study, the causal relationship between human serum metabolites and the ATB was analyzed by integrating the genome-wide association study (GWAS). The 486 human serum metabolites were used as the exposure variable, three different ATB GWAS databases in the European population were set as outcome variables, and single nucleotide polymorphisms were used as instrumental variables for Mendelian Randomization. The inverse variance weighting was estimated causality, the MR-Egger intercept to estimate horizontal pleiotropy, and the combined effects of metabolites were also considered in the meta-analysis. Furthermore, the web-based MetaboAnalyst 6.0 was engaged for enrichment pathway analysis, while R (version 4.3.2) software and Review Manager 5.3 were employed for statistical analysis.
Results
A total of 21, 17, and 19 metabolites strongly associated with ATB were found in the three databases after preliminary screening (P < 0.05). The intersecting metabolites across these databases included tryptophan, betaine, 1-linoleoylglycerol (1-monolinolein) (1-LG), 1-eicosatrienoylglycerophosphocholine, and oleoylcarnitine. Among them, betaine (I2 = 24%, P = 0.27) and 1-LG (I2 = 0%, P = 0.62) showed the lowest heterogeneity among the different ATB databases. In addition, the metabolic pathways of phosphatidylethanolamine biosynthesis (P = 0.0068), methionine metabolism (P = 0.0089), betaine metabolism (P = 0.0205) and oxidation of branched-chain fatty acids (P = 0.0309) were also associated with ATB.
Conclusion
Betaine and 1-LG may be biomarkers or auxiliary diagnostic tools for ATB. They may provide new guidance for medical practice in the early diagnosis and surveillance of ATB. In addition, by interfering with phosphatidylethanolamine biosynthesis, methionine metabolism, betaine metabolism, oxidation of branched-chain fatty acids, and other pathways, it is helpful to develop new anti-tuberculosis drugs and explore the virulence or pathogenesis of ATB at a deeper level, providing an effective reference for future studies.
Citation: Xian X, Li L, Ye J, Mo W, Liang D, Huang M, et al. (2024) Betaine and I-LG may have a predictive value for ATB: A causal study in a large European population. PLoS ONE 19(7): e0306752. https://doi.org/10.1371/journal.pone.0306752
Editor: Naseem Ahamad, UT Health San Antonio: The University of Texas Health Science Center at San Antonio, UNITED STATES
Received: March 19, 2024; Accepted: June 22, 2024; Published: July 5, 2024
Copyright: © 2024 Xian et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: The exposure datasets generated during the current study are available in IEU [http://ftp.ebi.ac.uk/pub/databases/gwas/summary_statistics/GCST90199001-GCST90200000]. The outcome database can be searched and downloaded from the IEU website based on tuberculosis [https://gwas.mrcieu.ac.uk/]. The original contributions presented in this study are included in the article/supplementary material, which can be queried via the corresponding link. Corresponding authors can be contacted for further inquiries.
Funding: this study received financial support from the National Natural Science Foundation of China (No. 82260656, 82360660) for the decision to publish data. The funders had no role in study design, data collection and analysis, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
Abbreviation: AIDS, Acquired immune deficiency syndrome; ATB, Active tuberculosis; ATP, Adenosine Triphosphate; BCAAs, branched-chain amino acids; KEGG, Kyoto Encyclopedia of Genes and Genomes; CI, confidence interval; Cys, Cysteine; GWAS, Genome Wide Association Study; HIV, Human Immunodeficiency Virus; IEU, Integrative Epidemiology Unit; IVs, Instumental variables; IVW, Inverse variance weighting; Kb, Kilobase; KEGG, Kyoto Encyclopedia of Genes and Genomes; LD, Linkage disequilibrium; LG, 1-linoleoylglycerol (1-monolinolein); M.tb, Mycobacterium tuberculosis; mmBCFAs, monomethyl-branched-chain fatty acids; MR, Mendelian randomization; OR, Odds ratio; PE, phosphatidylethanolamine; SAM, S-adenosylmethionine; SMPDB, Small Molecule Pathway Database; SNPs, Single nucleotide polymorphisms; TB, Tuberculosis; TBSA, Tuberculous stearic acid; TNF-α, Tumor Necrosis Factor α; WM, Weighted median
Introduction
It was estimated that a quarter of the global population is in a state of latent infection with Mycobacterium tuberculosis (M. tb), which is at risk of developing active tuberculosis (ATB) disease when host immune regulation is out of balance [1, 2]. While about 90–95% of people infected with M. tb do not develop ATB and remain asymptomatic [3]. Once developed into ATB, it most often involves the lungs, but also easily causes lymphadenopathy, skin diseases, neurological deficits, or other disseminated diseases, which not only causes serious damage to human health but also increases the economic burden and psychological burden of patients. According to the Global TB Report 2023, about 1.3 million people will die from ATB worldwide in 2022 [4]. Although great progress has been made in the prevention and control of ATB, it remains a public health problem worldwide.
In addition to environmental, pathogen, and socioeconomic factors [5, 6], host genetic diversity plays a significant role in the development of ATB [7]. At the same time, emerging metabolomics has been widely used to identify potential diagnostic biomarkers for various diseases, including ATB [8, 9]. Excepting diagnosis, these metabolites can also be used as biomarkers to elucidate disease mechanisms, as they can truly reflect the adaptation of bacteria to the metabolome due to growth in vivo and host response to infection and disease [10]. Studies have shown that inosine can be used as a marker of latent infection of tuberculosis, and metabolites such as betaine, branch-chain amino acids, and serum alanine have important effects on the pathogenesis of TB [11]. Similarly, tuberculous stearic acid (TBSA), branched-chain fatty acids, and vitamin D are also frequently studied as metabolites of ATB related [7, 12]. These metabolites may participate in the metabolic pathways of amino acids and fatty acids, and the abnormal metabolism of these metabolites may lead to the disorder of pro-inflammatory cytokine response and the change of anti-inflammatory response, leading to the decline of immunity, and then promote the accelerated proliferation of bacteria and the spread of infectious lesions.
However, the majority of these reports were derived from cross-sectional observational studies, which limits the strength of causal inference. Therefore, the present study collected serum metabolomic data from a comprehensive European population and utilized metabolism-associated single nucleotide polymorphisms (SNPs) as instrumental variables (IVs) in mendelian randomization (MR) validation in genomics to assess the causal influence of genetic replacement elements in metabolic pathways and the development of ATB, thereby providing new insights into the underlying mechanisms of ATB. Therefore, this study included three different European ATB GWAS datasets to comprehensively analyze the influence of serum metabolites on ATB pathogenesis. This study is of great significance for the monitoring and diagnosis of ATB in the future.
Methods
Study design
In this study, we comprehensively evaluated the relationship between 486 human serum metabolites and ATB based on a rigorous MR design. At its core, MR used genetic data as IVs to explore causal associations between a given exposure and outcome. A scientific MR study must include the testing of three hypotheses [13]: 1) the association hypothesis: genetic IVs are strongly associated with the study’s exposure; 2) the independence hypothesis: genetic IVs are not associated with outcome and are independent of any confounding factors, known or unknown; 3) the exclusive hypothesis: IVs can only affect the outcome through exposure factors. Therefore, the association obtained by MR is not affected by causal inversion and is not affected by confounding factors. In addition, the advantages of MR include the small measurement error of genetic variation and its effects, and the availability of publicly available GWAS data. However, MR can also have disadvantages, on the one hand, MR studies may be error-prone if the genetic variation is pleiotropy, and on the other hand, there may be linkage disequilibrium reactions (LD), since not all genes determining the trait are independently and randomly isolated. So, this study will set some constraints to reduce the consequences of LD. A schematic of this two-sample MR study is shown in Fig 1. At the same time, we used the STROBE-MR checklist to ensure the reliability and scientific of the study [14] (S1 Table).
(We selected significant IVs for 486 metabolites and ATB and explored causality. Acyclic graphs are used to illustrate the three basic assumptions of the MR Analysis).
Data sources
GWAS data of human serum metabolites (exposure).
Shin et al. [15] obtained genome-wide association aggregate datasets for 486 human serum metabolites in this study. The service platform is a collection of relatively complete European human serum metabolomics data and can be from the GWAS server publicly available (https://gwas.mrcieu.ac.uk/). Developed by the Integrated Epidemiology Unit (IEU) at the University of Bristol, this resource is a manually curated collection of the complete GWAS summary datasets and is available for download as an open-source file or by querying the complete database online. A total of 7,824 participants for the GWAS of human serum metabolites were recruited from two European population cohorts, including 1,768 participants from the KORA and 6,056 from the TwinUK, and approximately 2.1 million SNPs were included in the study. Out of the 486 metabolites, 309 have been named, according to the Kyoto Encyclopedia of Genes and Genomes (KEGG) database definition of genomes [16]. They can be divided into 8 Super-pathways (Amino acids, Carbohydrates, Cofactors and Vitamins, Energy, Lipids, Nucleotides, Peptides, and Xenobiotics) for integration and interpretation of large molecular datasets. The chemical properties of an additional 177 unknown metabolites have not been fully determined.
GWAS data for ATB (outcome).
In this study, we will use three different ATB databases to match 486 human serum metabolites and screen out the SNPs with genetic significance. To avoid racial differences, metabolites, and ATB genetic information were selected from GWAS datasets of the European population. The different ATB GWAS pooled data from the ieu open GWAS project were finn-b-AB1_TUBERCULOSIS database, finn-b-TBC_RESP database, and ebi-a-GCST90018892 database. These inclusion criteria for the cohort were based on the diagnostic protocol of ATB [17]. The GWAS pooled data included a total of 2,937 cases and 911,722 controls, with 56.95 million SNPs included in the analysis. We used SNPs associations of 486 human serum metabolites with ATB and their significant variables to explore the causal relationship.
Selecting of IVs
The selection of IVs in this MR Analysis were based on three basic premises. First, for each metabolite, we set P < 1×10−5 as a genome-wide significance threshold to select strongly correlated SNPs [18]. Second, a clustering program was implemented using R software to identify independent variants. Within a range of 10,000 kilobases (kb), R2 ≤ 0.01 were used to indicate LD, and identify independent SNPs [19]. Finally, to quantitatively verify whether the selected SNPs were powerful tools, the F statistic for each metabolite was calculated in this study. In general, an F ≥ 10 threshold was adopted for the following analyses. Moreover, considering the influence of alleles, we performed direct deletion of SNP palindromes [20]. MR of polygenic variants were performed using the pooled data.
MR analysis
The standard inverse variance weighted (IVW) method was employed to assess the causal association between metabolites and ATB. When all three main assumptions of IVs are met, the IVW method can provide a more precise estimate of the causal effect of exposure and is considered as the approach for robust causal detection [15]. If the null hypothesis were rejected, indicating that one or more variants may be multivariable, a random effects IVW was performed instead of a fixed effects IVW [21]. Conversely, if some IVs do not conform to the hypothesis, the analysis may gave inaccurate results. Therefore, we performed the following sensitivity analysis: 1) Q tests were performed on both IVW and MR-Egger methods to detect potential violation of heterogeneity assumption among individual IVs. Cochran’s Q test, commonly used in studies assessing diagnostic accuracy heterogeneity, was utilized [22]; 2) MR-Egger intercept was estimated to evaluate horizontal pleiotropy and ensure that genetic variants were independently associated with both metabolites and ATB outcomes [23]. The MR-Egger method was employed for interpreting the results obtained from MR analysis; 3) Additional analyses such as weighted median (WM) and weighted mode were implemented to enhance reliability and stability in hypothesis testing; 4) Single SNP analysis along with leave-one-out test were conducted to assess the plausibility of observed associations attributed to individual SNP.
Meta-analysis
If the results obtained from different databases were controversial, to ensure more convincing results, we conducted heterogeneity tests on these metabolites. Review Management 5.3 software was engaged to conduct a meta-analysis of the individual human serum metabolites that were identified as having a causal relationship with ATB and examined the overall estimated effect size. In this study, we used the odds ratio (OR) as the scale indicator to estimate the unknown data in the exposure group, and finally presented the results through forest plots. When I2 ≥ 50% or P < 0.05, high heterogeneity was indicated, and a random effects model should be used; when I2 < 50% and P ≥ 0.05, a fixed effects model should be used.
Enrichment analysis
MetaboAnalyst 6.0, a web-based tool for analyzing metabolomics data, was utilized to perform metabolic pathway analysis, interpretation, and integration with other omics data (https://www.metaboanalyst.ca/) [24]. The pathway and enrichment analysis module were applied to screen out metabolite clusters or super pathways that may be associated with metabolic processes and analyze their potential associations with ATB. Refer to the Small Molecule Pathway Database (SMPDB) and the Kyoto Encyclopedia of Genes and Genomes (KEGG) database. The significance level of pathways was less than 0.05.
Statistical analysis
All MR Analyses were performed using the "TwoSampleMR" package in R (version 4.3.2). Both the volcano and veen plots were drawn using Bioladder (https://www.bioladder.cn/), and the heat plots were drawn using Chiplot (https://www.chiplot.online/). The P < 0.05 of the IVW, since the threshold of the P value was artificially specified, even the smallest P value will give false positive results. Therefore, to reduce the chance of making Class I errors, we needed to perform multiple tests to eliminate false positives by correcting the threshold of P-values by Bonferroni, ensuring that the metabolites selected were diagnostically significant. The correction formula is P* (1/N), where "P" is the original threshold and "N" is the total number of tests [15]. Similarly, Bonferroni-P < 0.05 was considered statistically significant. The OR is used to estimate the size and direction of the metabolite effect and its corresponding 95% confidence interval (CI).
Results
Selection of genetic IVs for the three GWAS
In this study, a total of 486 serum metabolites data were matched to three the ATBGWAS online databases, and the selection of strongly associated SNPs was based on a genome-wide significance threshold (P < 1×10−5). Unfortunately, we were unable to identify relevant human serum metabolites in our reverse causal assessment. Consequently, different databases yielded varying numbers of associated metabolites based on the forward MR Study. Consequently, different databases yielded varying numbers of related metabolites. Among them, 485 metabolites were selected for both finn-b-AB1_TUBERCULOSIS and finn-b-TBC_RESP, and 486 metabolites were selected for ebi-a-GCST90018892. However, metabolites that passed the IVW test and Bonferroni P < 0.05 were found in 21 in finn-b-AB1_TUBERCULOSIS, 17 in finn-b-TBC_RESP, and 19 in ebi-a-GCST90018892. These metabolites were involved in seven super-pathways. In comparison, Lipid was the highest constituent, followed by Amino acids and Xenobiotics, and Carbohydrates and Nucleotides were the lowest (S2 Table). Finally, the IVs of finn-b-AB1_TUBERCULOSIS database contained a total of 2346 SNPs, with a median of 20 SNPs. The IVs of finn-b-TBC_RESP database contain 400 SNPs, and the median SNPs was 15. Meanwhile, the IVs of ebi-a-GCST90018892 database contain only 18,364 SNPs, and the median was 18. All three databases had F > 10, indicating that weak tool bias was unlikely to be significant (Table 1).
Impact of metabolites on ATB.
Metabolites were screened in three databases, and 57 human serum metabolites were found to be significantly associated with ATB (S3 Table). All metabolic analyses used random-effects IVW as the primary analysis method, with no evidence of heterogeneity and no weak tools. The preliminary analysis identified 21 metabolites significantly associated with ATB in finn-b-AB1_TUBERCULOSIS database, with 12 showing a positive association and 9 showing a negative association. In finn-b-TBC_RESP database, 17 metabolites were found to be statistically significant, 10 were positively associated with ATB, and 7 were negatively associated with ATB. There were 19 significant metabolites in ebi-a-GCST90018892 database, 10 were positively associated with ATB and 9 were negatively associated. A volcanic plot of human serum metabolites associated with ATB in the three databases is shown in Fig 2.
(The plot includes the OR on a log2 scale and the negative log10 scale of the P-values, which are estimated using the IVW method. A, finn-b-AB1_TUBERCULOSIS; B, finn-b-TBC_RESP; C, ebi-a-GCST90018892).
Relationship between three intersection metabolites and ATB
The heat plot shows the metabolites with statistical significance in any database and the corresponding OR values in different databases (Fig 3). The known metabolites that were positively correlated with ATB regardless of statistical significance of P values were tryptophan, tyrosine, pantothenate, urea, betaine, choline, 1-linoleoylglycerol (1-monolinolein) (1-LG), 2-hydroxybutyrate (AHB), serine, alanine, propionylcarnitine, and 2-linoleoylgly cerophosphocholine*. The negative correlations were included inosine, oleate (18:1n9), taurocholate, 1,5-anhydroglucitol (1,5-AG), X-12990—docosapentaenoic acid (n6-DPA), and oleoylcarnitine.
In addition to the 22 unknown metabolites, the other 35 known metabolites have different intersections between different databases (Fig 4). The specific metabolite names and related OR (95%CI) are shown in Tables 2 and 3. The significance of two metabolites, tryptophan and betaine, was observed across all three databases, indicating their super-pathway as amino acids. Both of these metabolites were positively correlated with ATB and were risk factors for ATB. The metabolites that were significant in finn-b-AB1_TUBERCULOSIS and finn-b-TBC_RESP databases included 1-LG, 1-eicosatrienoylgly cerophosphocholine, and oleoylcarnitine, in which 1-LG is positively correlated with ATB and has a promoting effect, while 1-eicosatrienoylcholine and oleoylcarnitine are negatively correlated with ATB and are protective factors.
Results of meta-analysis
As can be seen from Table 3, the OR values of metabolites in different databases were different. To explore the combined effect size of each metabolite, a meta-analysis was conducted. The results suggested low heterogeneity of betaine (I2 = 24%, P = 0.27), and even complete homogeneity of I-LG (I2 = 0%, P = 0.62). However, the final combined effect of the five metabolites was statistically significant (P < 0.05). Among them, tryptophan, betaine, and I-LG were risk factors, and 1-eicosatrienoylgly cerophosphocholine and oleoacylcarnitine were protective factors for the development of ATB (Fig 5).
Enrichment analysis results of known metabolites
35 known metabolites significantly associated with ATB were entered into the Metabolic Analyzer 6.0 platform to identify various potential metabolic pathways involved in the pathogenesis of ATB. Table 4 and Fig 6 demonstrate the interaction network among the metabolic pathways involved in this study. The metabolic pathways with high enrichment degree and proportion were the phosphatidylethanolamine (PE) biosynthesis (P = 0.0068), methionine metabolism (P = 0.0089), betaine metabolism (P = 0.0205) and oxidation of branched chain fatty acids (P = 0.0309).
(Colors range from light yellow to dark red, indicating the degree of enrichment significance, and the size of the circles reflects the degree of enrichment proportion).
Discussion
This study employed a two-sample MR model, utilizing genetic variants as IVs, to elucidate the causal relationship between human serum metabolites and ATB. Firstly, 15 known metabolites (tryptophan, tyrosine, pantothenate, pyroglutamylglycine, serine, alanine, propionylcarnitine, betaine, 2-hydroxybutyrate (AHB), 1-LG, X-14189—leucylalanine, urea, choline, X-11593—O-methylascorbate, and 2-linoleoylgly cerophosphocholine) were found to be risk factors for the development of ATB, while 13 known metabolites (inosine, oleate (18:1n9), 1,5-anhydroglucitol (1,5-AG), homocitrulline, glycerol 2-phosphate, 1-eicosatrienoylglycerophosphocholine, oleoylcarnitine, carnitine, taurocholate, butyrylcarnitine, salicyluric glucuronide, X-12990—docosapentaenoic acid (n6-DPA), and 4-vinylphenol sulfate) were found to be protective factors for ATB. Secondly, the meta-analysis showed that betaine and 1-LG showed little heterogeneity in different databases, suggesting that ATB was more related to these two metabolites. Finally, this study found that the abnormal pe biosynthesis, metabolism of methionine and betaine, and oxidation pathway of branch chain fatty acids were associated with the pathogenesis of ATB.
In recent years, significant advancements have been made in the field of metabolomics, particularly in the areas of AIDS (Acquired immune deficiency syndrome) [25], exercise [26] and health [27], and has also been applied to the diagnosis of diseases [28]. In addition to diagnosis, these metabolites can also be used as biomarkers to elucidate disease mechanisms, as they can truly reflect the adaptation of bacteria to the metabolome due to growth in vivo and host response to infection and disease [9]. Betaine (or trimethylglycine) is similar to choline (trimethylaminoethanol) but differs in that the carboxylate group at the end of choline is reduced to a hydroxyl group. Betaine can be obtained from dietary sources such as whole grains, beets, and spinach, and can also be synthesized from choline in the liver and kidneys [29, 30]. Firstly, choline is oxidized to betaine aldehyde by mitochondrial choline oxidase (choline dehydrogenase). Then, betaine aldehyde dehydrogenase oxidizes betaine aldehyde to betaine in the mitochondria or cytoplasm. In the liver, betaine acts similarly to the methyl donors of choline, folate, S-adenosylmethionine (SAM), and vitamin B12. Methyl donors are important for liver function, cell replication, and detoxification response. Betaine is also involved in the production of carnitine to protect the kidneys from damage and acts as an osmoprotector in the endomedulla. Betaine is a derivative of glycine that is produced endogenously through choline metabolism, or obtained exogenously through dietary intake. One of its roles is to help regulate homocysteine levels in the blood, which is also an amino acid that, when present at high levels, has been linked to an increased risk of neurovascular diseases, dementia, migraines, developmental disorders, or epilepsy [31–33]. The second is to convert homocysteine to methionine and detoxify it in the human liver and kidney, acting as a methyl donor and having osmoprotective properties [34]. Studies have found that choline, carnitine, and betaine are associated with inflammation and cardiometabolic biomarkers [35]. But they play a good role in chronic diseases such as obesity, diabetes, cancer, and Alzheimer’s disease [36–41]. Somashekar [42] has identified betaine as a potential biomarker for ATB detection, which aligns with the findings of this study. Shin. [43] conducted an in vitro trial using culturedguinea pig lung tissue and observed a significant decrease in betaine metabolism, suggesting the involvement of this metabolite in ATB infection. The fact that betaine was shown to be a risk factor for ATB in this study may be due to its abnormal metabolism leading to a decrease in the production of pro-inflammatory cytokines, which blocks the release of anti-inflammatory cytokines and leads to a decline in human immunity. This makes it difficult for the body to combat the accelerated proliferation of M.tb and the spread of infectious foci. However, the exact mechanism needs to be further explored with more studies in the future.
Jeffery et al. [44] demonstrated for the first time an association between specific lipid biosynthesis pathways and mycobacterium ATB colony morphology and virulence. Then, Garton et al. [45] found that liposome-positive (fatty triacylglycerol) acidobacter is a biomarker for non-replicating (inert) M.tb cells in sputum, so identifying this persistent group of M.tb bacteria in sputum provides an exciting and tractable new opportunity to study chemotherapy response and the spread of ATB. At the same time, it has been found that by increasing levels of monoacylglycerol (1-LG) and/or n-acylethanolamine, inhibitors of endocannabinoid hydrolysis may increase levels of their metabolites. Cannabinoids inhibit the release of TNF-α (Tumor Necrosis Factor), which is a type of tumor necrosis factor that causes the killing and necrosis of tumor cells [46]. The risk of 1-LG is acknowledged, but the role of 1-LG in ATB has not been reported in more literature.
PE is the second most abundant phospholipid in eukaryotic cell membranes and is concentrated in the inner lobules of cell membranes together with phosphatidylserine. In Homo sapiens, PE exists in two biosynthetic pathways [47, 48]. The PE biosynthesis from ethanolamine via the Kennedy pathway for the first. In the beginning, choline is catalyzed to choline phosphate by the cytosol-locating enzyme choline or ethanolamine kinase. Then, the conversion to CDP-choline is catalyzed by choline-phosphate cytotransferase, which is localized to the endoplasmic reticulum. Finally, PE is synthesized by choline/ethanolamine phosphotransferase and cofactors such as magnesium ion or manganese ion. The second pathway is the methylation of phosphatidylserine to synthesize PE. Choline, folate, and methionine metabolism are related because they all affect the production of SAM, which is a common donor of methyl groups in biological reactions [49]. Choline primarily contributes to the formation of cell membranes and neurotransmitters [30].
Cysteine (Cys) and methionine are sulfur-containing amino acids. Cys is synthesized from serine, and in bacteria and plants. Cys is converted from serine (via acetyl serine) by the transfer of hydrogen sulfide. In animals, methionine-derived homocysteine is used as a sulfur source, and its condensation products with serine (cysteine thionine) are converted to cysteine. Inhibition of Cys synthesis leads to reduced fitness and pathogenesis in bacteria [50–53]. Cys acts as a central regulator of REDOX homeostasis through the synthesis of Mycobacterium mercaptan (MSH), the main antioxidant buffer of mycobacterium [54]. In addition, Cys-derived Fe-S clusters [55, 56], sulfatolides (SL-1) [57], and hydrogen sulfide (H2S) [58], are essential for the respiration and persistence of M.tb and for resistance to antibiotics [59, 60]. Higher circulating homocysteine (Hcys) levels are consistently associated with HIV-infected patients [61]. Bandyopadhyay [62] theorizes that cysteine beta synthase may contribute to the survival of M. tb during HIV/TB co-infection. Methionine is an essential amino acid that cannot be synthesized by animals. In bacteria and plants, methionine is synthesized from aspartate. SAM is a methyl donor synthesized from methionine and Adenosine Triphosphate (ATP) and acts as a methyl donor in many important transfer reactions, including DNA methylation. SAM may also be used to regenerate methionine in the methionine recycling pathway. Methionine is involved in countless cellular functions, and methionine is essential for immunity against tumors. Methionine metabolism is involved in many cellular functions, including methylation reactions, redox maintenance, and folate metabolism. Acetylation and methylation are important for controlling antimicrobial mechanisms within macrophages (MΦ) during intracellular infections such as ATB [63, 64]. The inactivation of cysteine and methionine biosynthesis in M.tb significantly reduces the persistence and virulence of the chronic infection phase in mouse models [65].
In addition, patients with obesity and insulin resistance are often accompanied by abnormal increases in branched-chain amino acids (BCAAs) and their metabolites. Metabolic disorders of BCAAs are strongly associated with the development of obesity and insulin resistance. In addition, levels of monomethyl-branched-chain fatty acids (mmBCFAs) have also been linked to the development of metabolic syndrome and cancer. However, there are few studies on mmBCFAs at present, and the specific mechanism of mmBCFAs’ action in disease is not fully understood, further studies are needed to reveal its role in disease development. Only Kniazeva [66, 67] studies have mentioned that lipids containing mmBCfAs play a key role in the postembryonic growth, neuronal development, and foraging behavior of Caenorhabditis elegans.
It is worth mentioning that the ATB bacterium has a special cell wall structure containing branched-chain fatty acids (mycolic acid), which play a key role in the formation and stability of the cell wall. The ATB bacterium maintains the integrity of its cell wall by synthesizing branched fatty acids, thereby protecting itself from attack by the host immune system. The metabolism of branched-chain fatty acids is closely related to the synthesis and assembly of cell wall polysaccharides, and together these molecules make up an important part of the cell wall of the ATB bacterium. Therefore, the metabolism of branch-chain fatty acids interacts with the synthesis of cell wall polysaccharides to affect the growth, reproduction, and pathogenicity of ATB bacteria.
In conclusion, in the future, when developing and diagnosing ATB drugs, it may be considered to start from the above metabolites and monitor them through multiple stages using in vitro and in vivo models to evaluate their role in the progression of ATB. At the same time, although this study only focused on the significance of PE biosynthesis, methionine metabolism, betaine metabolism, and branched-chain fatty acid metabolism pathways, the role of these pathways in the pathogenesis of ATB should be further explored in the process of drug development.
However, there may be several limitations to this study. Firstly, in the current environment, it was inevitable to encounter the limitation of unavailability of individual behavioral and biological characteristics due to the utilization of a pooled dataset. Secondly, there is a possibility that certain metabolites associated with ATB might have been excluded due to the pleiotropic level limitation of instrumental variables or limitations in the sample size. Lastly, some significant metabolites and pathways were omitted as they lacked proper naming or annotation in pathway databases. Therefore, further comprehensive exploration will be necessary in future studies.
Conclusion
Through a causal analysis based on a large European sample, we found that betaine and 1-LG may be biomarkers or auxiliary diagnostic tools for ATB. The findings suggest that these two human serum metabolites may lead to new guidance for medical practice in the early diagnosis and surveillance of ATB. In addition, by interfering with PE synthesis, methionine metabolism, betaine metabolism branch chain fatty acid metabolism, and other pathways, it is helpful to develop new anti-tuberculosis drugs and explore the virulence or pathogenesis of ATB at a deeper level, providing an effective reference for future studies.
Supporting information
S1 Table. STROBE-MR checklist of recommended items to address in reports of mendelian randomization studies.
https://doi.org/10.1371/journal.pone.0306752.s001
(CSV)
S2 Table. The super-pathway of metabolites selected by IVs.
https://doi.org/10.1371/journal.pone.0306752.s002
(CSV)
S3 Table. The causal relationships between 57 metabolites and the risk of ATB are estimated by three MR models, which also conduct tests for heterogeneity and horizontal pleiotropy.
https://doi.org/10.1371/journal.pone.0306752.s003
(CSV)
References
- 1. Houben RMGJ Dodd PJ. The Global Burden of Latent Tuberculosis Infection: A Re-estimation Using Mathematical Modelling. PLoS Med. 2016;13: e1002152. pmid:27780211
- 2. Jin BC, Moon HJ, Kim SW. Latent and Active Tuberculosis Infection in Patients with Inflammatory Bowel Disease. Korean J Gastroenterol. 2022 Aug 25;80(2):72–76. Korean. pmid:36004634
- 3.
Jilani TN, Avula A, Zafar Gondal A, Siddiqui AH. Active Tuberculosis. 2023 Jan 26. In: StatPearls [Internet]. Treasure Island (FL): StatPearls Publishing; 2024.
- 4. Bagcchi S. WHO’s Global Tuberculosis Report 2023. Lancet Microbe. 2023 Jan;4(1):e20.
- 5. Kharwadkar S, Attanayake V, Duncan J, et al. The impact of climate change on the risk factors for tuberculosis: A systematic review. Environ Res. 2022 Sep;212(Pt C):113436. pmid:35550808
- 6. Tian Y, Fu MJ, Bei CL. Review on risk factors associated with tuberculosis death. Zhonghua Jie He He Hu Xi Za Zhi. 2022 Sep 12;45(9):941–946. Chinese. pmid:36097933
- 7. Cai L, Li Z, Guan X, Cai K, Wang L, Liu J, et al. The Research Progress of Host Genes and Tuberculosis Susceptibility. Oxid Med Cell Longev. 2019 Aug 14;2019:9273056. pmid:31485302
- 8. Dunn WB, Bailey NJ, Johnson HE. Measuring the metabolome: current analytical technologies. Analyst. 2005 May; 130(5): 606–25. pmid:15852128
- 9. Preez ID, Luies L, Loots DT. Metabolomics biomarkers for tuberculosis diagnostics: current status and future objectives. Biomark Med. 2017 Feb;11(2):179–194. pmid:28097879
- 10. Diboun I, Cyprian FS, Anwardeen NR, et al. Identification of Prognostic Metabolomic Biomarkers at the Interface of Mortality and Morbidity in Pre-Existing TB Cases Infected With SARS-CoV-2. Front Cell Infect Microbiol. 2022 Jul 22;12:929689. pmid:35937683
- 11. Li YX, Zheng KD, Duan Y, et al. Mass spectrometry-based identification of new serum biomarkers in patients with latent infection pulmonary tuberculosis. Medicine (Baltimore). 2022 Dec 2;101(48):e32153. pmid:36482536
- 12. Selvaraj P, Harishankar M, Afsal K. Vitamin D: Immuno-modulation and tuberculosis treatment. Can J Physiol Pharmacol. 2015 May;93(5):377–84. pmid:25744368
- 13. Burgess S, Butterworth A, Thompson SG. Mendelian randomization analysis with multiple genetic variants using summarized data. Genet Epidemiol. 2013;37(7):658–665. pmid:24114802
- 14. Skrivankova VW, Richmond RC, Woolf BAR, et al. Strengthening the Reporting of Observational Studies in Epidemiology using Mendelian Randomisation (STROBE-MR): Explanation and Elaboration. BMJ. 2021;375:n2233. pmid:34702754
- 15. Shin SY, Fauman EB, Petersen AK, et al. An atlas of genetic influences on human blood metabolites. Nat Genet. 2014 Jun;46(6):543–550. Epub 2014 May 11. pmid:24816252; PMCID: PMC4064254.
- 16. Kanehisa M, Goto S, Sato Y, Furumichi M, Tanabe M. KEGG for integration and interpretation of large-scale molecular data sets. Nucleic Acids Res. 2012;40:D109–114. pmid:22080510
- 17. Sakaue S, Kanai M, Tanigawa Y, et al. A cross-population atlas of genetic associations for 220 human phenotypes. Nat Genet. 2021 Oct;53(10):1415–1424. Epub 2021 Sep 30. pmid:34594039
- 18. Guo Z, Zhang T, Yun Z, et al, Assessing the causal relationships between human blood metabolites and the risk of NAFLD: A comprehensive mendelian randomization study. Front. Genet. 2023, 14:1108086. pmid:37056283
- 19. Zhao L, Wu R, Wu Z, et al. Genetically predicted 486 blood metabolites concerning risk of systemic lupus erythematosus: a Mendelian randomization study. Sci Rep. 2023 Dec 18;13(1):22543. pmid:38110541
- 20. Gill D., Brewer C. F., Monori G., et al. (2019). Effects of genetically determined iron status on risk of venous thromboembolism and carotid atherosclerotic disease: A mendelian randomization study. J. Am. Heart Assoc. 8, e012994. pmid:31310728
- 21. Burgess S., Butterworth A., Thompson S. G. (2013). Mendelian randomization analysis with multiple genetic variants using summarized data. Genet. Epidemiol. 37, 658–665. pmid:24114802
- 22. Cohen JF, Chalumeau M, Cohen R, Korevaar DA, Khoshnood B, Bossuyt PM. Cochran’s Q test was useful to assess heterogeneity in likelihood ratios in studies of diagnostic accuracy. J Clin Epidemiol. 2015;68(3):299–306. pmid:25441698
- 23. Yelamanchi SD, Surolia A. Targeting amino acid metabolism of Mycobacterium tuberculosis for developing inhibitors to curtail its survival. IUBMB Life. 2021 Apr; 73(4): 643–658. pmid:33624925
- 24. Chong J, Xia J. Using MetaboAnalyst 4.0 for Metabolomics Data Analysis, Interpretation and Integration with Other Omics Data. Methods Mol Biol. 2104; 11 337–360.
- 25. Lao D, Liu R, Liang J. Study on plasma metabolomics for HIV/AIDS patients treated by HAART based on LC/MS-MS. Front Pharmacol. 2022 Aug 29; 13: 885386. pmid:36105186
- 26. Khoramipour K, Sandbakk Ø, Keshteli AH, et al. Metabolomics in Exercise and Sports: A Systematic Review. Sports Med. 2022 Mar; 52(3): 547–583. pmid:34716906
- 27. Abdelhafez OH, Othman EM, Fahim JR, et al. Metabolomics analysis and biological investigation of three Malvaceae plants. Phytochem Anal. 2020 Mar; 31(2): 204–214. pmid:31390115
- 28. Yelamanchi SD, Surolia A. Targeting amino acid metabolism of Mycobacterium tuberculosis for developing inhibitors to curtail its survival. IUBMB Life. 2021 Apr;73(4):643–658. pmid:33624925
- 29. Day CR, Kempson SA: Betaine chemistry, roles, and potential use in liver disease. Biochim Biophys Acta. 2016 Jun;1860(6):1098–106. pmid:26850693
- 30. Zhao G, He F, Wu C, et al. Betaine in Inflammation: Mechanistic Aspects and Applications. Front Immunol. 2018 May 24;9:1070. pmid:29881379
- 31. Arumugam M.K., Paal M.C., Donohue T.M., et al. Beneficial effects of betaine: A comprehensive review. Biology. 2021;10:456. pmid:34067313
- 32. Truitt C., Hoff W.D., Deole R. Health functionalities of betaine in patients with homocystinuria. Front. Nutr. 2021;8:627. pmid:34568401
- 33. Hermann A, Sitdikova G. Homocysteine: Biochemistry, Molecular Biology and Role in Disease. Biomolecules. 2021 May 15;11(5):737. pmid:34063494
- 34. Williams KT, Schalinske KL. New insights into the regulation of methyl group and homocysteine metabolism. J Nutr. 2007 Feb;137(2):311–4. pmid:17237303
- 35. Fu BC, Hullar MAJ, Randolph TW, et al. Associations of plasma trimethylamine N-oxide, choline, carnitine, and betaine with inflammatory and cardiometabolic risk biomarkers and the fecal microbiome in the Multiethnic Cohort Adiposity Phenotype Study. Am J Clin Nutr. 2020 Jun 1;111(6):1226–1234. pmid:32055828
- 36. Konstantinova SV, Tell GS, Vollset SE, Nygård O, Bleie Ø, Ueland PM. Divergent associations of plasma choline and betaine with components of metabolic syndrome in middle age and elderly men and women. J Nutr. 2008 May;138(5):914–20. pmid:18424601
- 37. Schartum-Hansen H, Ueland PM, Pedersen ER, Meyer K, Ebbing M, Bleie Ø, et al. Assessment of urinary betaine as a marker of diabetes mellitus in cardiovascular patients. PLoS One. 2013 Aug 6;8(8):e69454. pmid:23936331
- 38. Lever M, Slow S, McGregor DO, Dellow WJ, George PM, Chambers ST. Variability of plasma and urine betaine in diabetes mellitus and its relationship to methionine load test responses: an observational study. Cardiovasc Diabetol. 2012 Jul 11;11:34. pmid:22510294
- 39. Ying J, Rahbar MH, Hallman DM, Hernandez LM, Spitz MR, Forman MR, et al. Associations between dietary intake of choline and betaine and lung cancer risk. PLoS One. 2013;8(2):e54561. pmid:23383301
- 40. Zeng FF, Xu CH, Liu YT, et al. Choline and betaine intakes are associated with reduced risk of nasopharyngeal carcinoma in adults: a case-control study. Br J Cancer. 2014 Feb 4;110(3):808–16. pmid:24169354
- 41. Madsen SK, Rajagopalan P, Joshi SH, et al. Alzheimer’s Disease Neuroimaging Initiative (ADNI). Higher homocysteine associated with thinner cortical gray matter in 803 participants from the Alzheimer’s Disease Neuroimaging Initiative. Neurobiol Aging. 2015 Jan;36 Suppl 1:S203–10.
- 42. Somashekar BS, Amin AG, Tripathi P, et al. Metabolomic signatures in guinea pigs infected with epidemic-associated W-Beijing strains of Mycobacterium tuberculosis. J Proteome Res. 2012 Oct 5;11(10):4873–84. pmid:22873951
- 43. Shin JH, Yang JY, Jeon BY, et al. (1)H NMR-based metabolomic profiling in mice infected with Mycobacterium tuberculosis. J Proteome Res. 2011 May 6; 10(5): 2238–47. pmid:21452902
- 44. Cox Jeffery S., Chen Bing, et al. "Complex Lipid Determines Tissue-specific Replication of Mycobacterium Tuberculosis in Mice", NATURE, 1999. (IF: 7) pmid:10573420
- 45. Garton NJ, Waddell SJ, Sherratt AL, et al. Cytological and transcript analyses reveal fat and lazy persister-like bacilli in tuberculous sputum. PLoS Med. 2008 Apr 1;5(4):e75. pmid:18384229
- 46. Mitoma H, Horiuchi T, Tsukamoto H, Ueda N. Molecular mechanisms of action of anti-TNF-α agents—Comparison among therapeutic TNF-α antagonists. Cytokine. 2018 Jan;101:56–63.
- 47. Alatorre-Cobos F, Cruz-Ramírez A, Hayden CA, et al. Translational regulation of Arabidopsis XIPOTL1 is modulated by phosphocholine levels via the phylogenetically conserved upstream open reading frame 30. J Exp Bot. 2012 Sep;63(14):5203–21. pmid:22791820
- 48. Henneberry AL, Wistow G, McMaster CR. Cloning, genomic organization, and characterization of a human cholinephosphotransferase. J Biol Chem. 2000 Sep 22;275(38):29808–15. pmid:10893425
- 49. Zeisel SH. Choline: critical role during fetal development and dietary requirements in adults. Annu Rev Nutr. 2006;26:229–250. pmid:16848706
- 50. Poyraz O, Jeankumar VU, Saxena S, et al. Structure-guided design of novel thiazolidine inhibitors of O-acetyl serine sulfhydrylase from Mycobacterium tuberculosis. J Med Chem 2013; 56:6457–6466. pmid:23879381
- 51. Jean Kumar VU, Poyraz Ö, Saxena S, et al. Discovery of novel inhibitors targeting the Mycobacterium tuberculosis O-acetylserine sulfhydrylase (CysK1) using virtual high-throughput screening. Bioorg Med Chem Lett 2013; 23:1182–1186. pmid:23391589
- 52. Spyrakis F, Singh R, Cozzini P, et al. Isozyme-specific ligands for O-acetylserine sulfhydrylase, a novel antibiotic target. PLoS One 2013; 8:e77558. pmid:24167577
- 53. Nagpal I, Raj I, Subbarao N, Gourinath S. Virtual screening, identification and in vitro testing of novel inhibitors of O-acetyl-L-serine sulfhydrylase of Entamoeba histolytica. PLoS One 2012; 7:e30305. pmid:22355310
- 54. Buchmeier N. A., Newton G. L., Koledin T., et al. Association of mycothiol with protection of Mycobacterium tuberculosis from toxic oxidants and antibiotics. Mol. Microbiol. 47, 1723–1732 (2003). pmid:12622824
- 55. Huet G., Daffé M., Saves I., Identification of the Mycobacterium tuberculosis SUF machinery as the exclusive mycobacterial system of [Fe-S] cluster assembly: Evidence for its implication in the pathogen’s survival. J. Bacteriol. 187, 6137–6146 (2005). pmid:16109955
- 56. Rybniker J., Pojer F., Marienhagen J., Kolly G. S., Chen J. M., van Gumpel E., et al. The cysteine desulfurase IscS of Mycobacterium tuberculosis is involved in iron-sulfur cluster biogenesis and oxidative stress defence. Biochem. J. 459, 467–478 (2014). pmid:24548275
- 57. Seeliger J. C., Holsclaw C. M., Schelle M. W., Botyanszki Z., Gilmore S. A., Tully S. E., et al. Elucidation and chemical modulation of sulfolipid-1 biosynthesis in Mycobacterium tuberculosis. J. Biol. Chem. 287, 7990–8000 (2012). pmid:22194604
- 58. Pal V. K., Bandyopadhyay P., Singh A., Hydrogen sulfide in physiology and pathogenesis of bacteria and viruses. IUBMB Life 70, 393–410 (2018). pmid:29601123
- 59. Voskuil M. I., Visconti K. C., Schoolnik G. K., Mycobacterium tuberculosis gene expression during adaptation to stationary phase and low-oxygen dormancy. Tuberculosis (Edinb.) 84, 218–227 (2004). pmid:15207491
- 60. Schnappinger D., Ehrt S., Voskuil M. et al. Transcriptional adaptation of Mycobacterium tuberculosis within macrophages: Insights into the phagosomal environment. J. Exp. Med. 198, 693–704 (2003). pmid:12953091
- 61. Deminice R., Silva T. C. V., de Oliveira V. H. F., Elevated homocysteine levels in human immunodeficiency virus-infected patients under antiretroviral therapy: A meta-analysis. World J. Virol. 4, 147–155 (2015). pmid:25964880
- 62. Bandyopadhyay P, Pramanick I, Biswas R, et al. S-Adenosylmethionine-responsive cystathionine β-synthase modulates sulfur metabolism and redox balance in Mycobacteriumtuberculosis. Sci Adv. 2022 Jun 24;8(25):eabo0097. Epub 2022 Jun 24. pmid:35749503
- 63. Darrah PA, Zeppa JJ, Maiello P, et al. Prevention of tuberculosis in macaques after intravenous BCG immunization. Nature. 2020;577:95–102. pmid:31894150
- 64. Irvine EB, O’Neil A, Darrah PA, et al. Robust IgM responses following intravenous vaccination with Bacille Calmette-Guerin associate with prevention of Mycobacterium tuberculosis infection in macaques. Nat Immunol. 2021;22:1515–1523.
- 65. Fontán P, Aris V, Ghanny S, et al. Global transcriptional profile of Mycobacterium tuberculosis during THP-1 human macrophage infection. Infect Immun. 2008 Feb;76(2):717–25. pmid:18070897
- 66. Kniazeva M, Euler T, Han M. A branched-chain fatty acid is involved in post-embryonic growth control in parallel to the insulin receptor pathway and its biosynthesis is feedback-regulated in C. elegans. Genes Dev. 2008 Aug 1;22(15):2102–10. pmid:18676815
- 67. Kniazeva M, Zhu H, Sewell AK, et al. A Lipid-TORC1 Pathway Promotes Neuronal Development and Foraging Behavior under Both Fed and Fasted Conditions in C. elegans. Dev Cell. 2015 May 4;33(3):260–71.