Data handling for interactive metabolomics: tools for studying the dynamics of metabolome-macromolecule interactions

All published metabolomics studies investigate changes in either absolute or relative quantities of metabolites. However, blood plasma, one of the most commonly studied biofluids for metabolomics applications, is a complex, heterogeneous mixture of lipoproteins, proteins, small organic molecules and ions which together undergo a variety of possible molecular interactions including metal complexation, chemical exchange processes, micellular compartmentation of metabolites, enzyme-mediated biotransformations and small-molecule-macromolecule binding. In particular, many low molecular weight (MW) compounds (including drugs) can exist both ‘free’ in solution and bound to proteins or within organised aggregates of macromolecules. To study the effects of e.g. disease on these interactions we suggest that new approaches are needed. We have developed a technique termed ‘interactive metabolomics’ or i-metabolomics. i-metabolomics can be defined as: “The study of interactions between low MW biochemicals and macromolecules in heterogeneous biosamples such as blood plasma, without pre-selection of the components of interest”. Standard 1D NMR experiments commonly used in metabolomics allow metabolite concentration differences between samples to be investigated because the intensity of each peak depends on the concentration of the compound in question. On the other hand, the instrument can be set-up to measure molecular interactions by monitoring the diffusion coefficients of molecules. According to the Stokes–Einstein equation, the diffusion coefficient of a molecule is inversely proportional to its effective size, as represented by the hydrodynamic radius. Therefore, when low MW compounds are non-covalently bound to proteins, the observed diffusion coefficient for the compound will be intermediate between those of its free and bound forms. By measuring diffusion by NMR, the degree of protein binding can be estimated for either low MW endogenous biochemicals or xenobiotics. This type of experiment is referred to as either Diffusion-Ordered Spectroscopy (DOSY) or Diffusion-Edited Spectroscopy, depending on the type of post-acquisition data processing applied to the spectra. Results presented in this paper demonstrate approaches for the non-selective modelling of metabolite-macromolecule interactions (i-metabolomics), whilst additionally highlighting some of the all too frequently ignored issues associated with interpretation of data derived from profiling of blood plasma.     

Solute Size Effects on the Diffusion in Biofilms of Streptococcus mutans

The diffusion of poly(ethylene glycol) (PEG) (MW varying between 200 and 10,000), and of three different types of micelles was examined in Streptococcus mutans biofilms using infrared spectroscopy. PEGs were used because they show limited interactions with biological materials and their weight can be selected in order to cover a wide range of size. The study showed that a considerable fraction at the base of the biofilm was not accessible to the diffusing solute molecules and this inaccessible fraction was very dependent on the size of the diffusing molecules. In parallel, it was found that the diffusion coefficients of these solutes in the biofilms were less than those in water and this reduction was less pronounced for large macromolecules, an effect proposed to be related to their limited penetration. Triton X-100, a neutral detergent, forms micelles that behave like PEG, suggesting that the behaviour observed for neutral macromolecules can be extrapolated to neutral macroassemblies. However, the diffusion, as well as the penetration of sodium dodecylsulphate micelles (a negatively charged surfactant) and cetylpyridinium chloride micelles (positively charged), in the biofilms appeared to be significantly influenced by electrostatic interactions with biofilm components. The present findings provide useful insights associated with the molecular parameters required to efficiently penetrate bacterial biofilms. The study suggests a rationale for the limited bactericidal power of some antibiotics (the large ones). The restricted accessibility of macromolecules and macroassemblies to biofilms must be examined carefully in order to offer guidelines in the development of novel antibacterial treatments.     

PhoE protein pore of the outer membrane of Escherichia coli K12 is a particularly efficient channel for organic and inorganic phosphate

This study was undertaken to investigate the proposed in vivo pore function of PhoE protein, an Escherichia coli K12 outer membrane protein induced by growth under phosphate limitation, and to compare it with those of the constitutive pore proteins OmpF and OmpC. Appropriate mutant strains were constructed containing only one of the proteins PhoE, OmpF or OmpC, or none of these proteins at all. By measuring rates of nutrient uptake at low solute concentrations, the proposed pore function of PhoE protein was confirmed as the presence of the protein facilitates the diffusion of P_i through the outer membrane, such that a pore protein deficient strain behaves as a K_m mutant. Comparison of the rates of permeation of P_i, glycerol 3-phosphate and glucose 6-phosphate through pores formed by PhoE, OmpF and OmpC proteins shows that PhoE protein is the most effective pore in facilitating the diffusion of P_i and phosphorus-containing compounds. The three types of pores were about equally effective in facilitating the permeation of glucose and arsenate. Possible reasons for the preference for P_i and P_i-containing solutes are discussed.     

Dynamic light-scattering studies of DNA. I. The coupling of internal modes with anisotropic translational diffusion in congested solutions

A model for the coupling between internal modes, or molecular rotation, and anisotropic translational diffusion in congested solutions is proposed to account for the anomalously slow component that has appeared ubiquitously in reported autocorrelation functions of Rayleigh scattered light from solutions of DNA's with molecular weights greater than about 10⁷. The predicted existence of an anomalously slow mode in addition to a faster “normal” mode, as well as the predicted relative amplitudes of both fast and slow components, are qualitatively in agreement with the observations. For sufficiently long-wavelength fluctuations all of the amplitude appears in the slower mode, which then exhibits an appropriately averaged translational diffusion coefficient. In support of the model it is shown in the Appendix that nonideal central interactions between macromolecules are by themselves insufficient to generate isolated internal mode relaxation terms in the autocorrelation function, unless translational ordering of the macromolecules extends over the illuminated observation region.     

Bound-State Diffusion due to Binding to Flexible Polymers in a Selective Biofilter

Selective biofilters are used by cells to control the transport of proteins, nucleic acids, and other macromolecules. Biological filters demonstrate both high specificity and rapid motion or high flux of proteins. In contrast, high flux comes at the expense of selectivity in many synthetic filters. Binding can lead to selective transport in systems in which the bound particle can diffuse, but the mechanisms that lead to bound diffusion remain unclear. Previous theory has proposed a molecular mechanism of bound-state mobility based only on transient binding to flexible polymers. However, this mechanism has not been directly tested in experiments. We demonstrate that bound mobility via tethered diffusion can be engineered into a synthetic gel using protein fragments derived from the nuclear pore complex. The resulting bound-state diffusion is quantitatively consistent with theory. Our results suggest that synthetic biological filters can be designed to take advantage of tethered diffusion to give rapid, selective transport.     

Multilevel structural characteristics for the natural substrate proteins of bacterial small heat shock proteins

Small heat shock proteins (sHSPs) are ubiquitous molecular chaperones that prevent the aggregation of various non‐native proteins and play crucial roles for protein quality control in cells. It is poorly understood what natural substrate proteins, with respect to structural characteristics, are preferentially bound by sHSPs in cells. Here we compared the structural characteristics for the natural substrate proteins of Escherichia coli IbpB and Deinococcus radiodurans Hsp20.2 with the respective bacterial proteome at multiple levels, mainly by using bioinformatics analysis. Data indicate that both IbpB and Hsp20.2 preferentially bind to substrates of high molecular weight or moderate acidity. Surprisingly, their substrates contain abundant charged residues but not abundant hydrophobic residues, thus strongly indicating that ionic interactions other than hydrophobic interactions also play crucial roles for the substrate recognition and binding of sHSPs. Further, secondary structure prediction analysis indicates that the substrates of low percentage of β‐sheets or coils but high percentage of α‐helices are un‐favored by both IbpB and Hsp20.2. In addition, IbpB preferentially interacts with multi‐domain proteins but unfavorably with α + β proteins as revealed by SCOP analysis. Together, our data suggest that bacterial sHSPs, though having broad substrate spectrums, selectively bind to substrates of certain structural features. These structural characteristic elements may substantially participate in the sHSP–substrate interaction and/or increase the aggregation tendency of the substrates, thus making the substrates more preferentially bound by sHSPs.     

On the low viscosity blood of two cold water,marine sculpins

Summary The viscosities of blood from shorthorn sculpin (Myoxocephalus scorpius), longhorn sculpin (Myoxocephalus octodecemspinosus) and winter flounder (Pseudopleuronectes americanus) were compared using a cone-plate viscometer. Both species of sculpin were almost identical with respect to blood and plasma viscosity at the temperatures (0 and 15°C) and shear rates (2.3–90/s) examined. In contrast, the viscosities of winter flounder blood and plasma were considerably greater than those observed in the sculpins. This difference in blood viscosity between the shorthorn sculpin and the winter flounder persisted over the hematocrit range of 0 to 40% red blood cells. The viscosity of the plasma and the interactions between plasma proteins and red blood cells appeared to be the major reasons for the relatively high viscosity of the flounder blood. Although a proportion of the flounder blood viscosity was attributable to fibrinogen, other plasma proteins also appeared to play a significant role. The relatively low blood viscosity of the sculpin species may confer a circulatory advantage during periods of low water temperatures.     

Penetration of macromolecules into contracted blood clot.

The effectiveness of thrombolytic therapy is determined by accessibility of thrombus compartments to plasminogen activators and, therefore, depends on permeability of thrombus to blood born macromolecules. Accumulation of 125I labeled proteins with molecular massess ranging from 150 to 450 kD into partly contracted blood clot or plasma clot was consistent with diffusion coefficients 3.2 x 10(-11) and 2.7 x 10(-11) m2 s-1, respectively. So far as the model conditions imitated those for venous thrombi, these data indicate that such thrombi are porous enough for immunoconjugates of relatively big size.     

Band centrifugation of macromolecules in self-generating density gradients. III. Conditions for convection-free band sedimentation

The conditions for convection-free hand sedimentation are analyzed in terms of the negative density gradients associated with the leading edge of a band and the positive density gradients generated during the experiment. The amount of material necessary to perform a band-centrifugation experiment depends on the diffusion coefficient of the macromolecules, which determines the rate at which the concentration at band maximum decreases, and on the extinction coefficient at the wavelength of observation. The maximum negative gradients in bands of macromolecules with diffusion and extinction coefficients typical of proteins, nucleic acids, and viruses are calculated. Positive density gradients generated by diffusion of small molecules between the thin lamella and the bulk solution are calculated for bulk solutions such us 1M NaCl or 95% D₂O. These “diffusion gradients” are generally adequate to stabilize bands of the above macromolecules. Positive density gradients generated by sedimentation of salts within the bulk solution may be significant in providing stability near the bottom of the cell. The effects of inadequate stabilizing gradients are discussed, and are found to cause forward spreading of the band.     

Chymotrypsin inhibitor from potatoes: interaction with target enzymes

The interactions of chymotrypsin, subtilisin and trypsin with a low MW proteinase inhibitor from potatoes were investigated. The K_i value calculated for the binding of inhibitor to chymotrypsin was 1.6 ± 0.9 × 10^?10M, while the second-order rate constant for association was 6 × 10⁵ M^?1/sec. Although binding was not observed to chymotrypsin which had been treated with diisopropyl fluorophosphate or with l-tosylamide-2-phenylethyl chloromethyl ketone, the 3-methylhistidine-57 derivative bound inhibitor with a K_i value of 9.6 × 10^?9 M. The inhibitor also exhibited a tight association with subtilisin (K_i < 4 × 10^?9 M). In contrast, little inhibition of trypsin was observed, and this was believed to be due to low levels of a contaminant in our preparations. No evidence for reactive site cleavage was observed after incubation of the inhibitor with catalytic amounts of chymotrypsin or subtilisin at acid pH.     

功能代谢组学技术在微生物研究中的应用

功能代谢组学是以代谢组学技术发现关键代谢物为基础,结合体内体外实验和分子生物学等技术手段,研究差异代谢物及相关蛋白、酶和基因的功能,从而揭示生物体内在的分子调控机制。功能代谢组学技术具有精准识别关键调控代谢物及其相关基因或酶的特性,近年来在微生物相关疾病的防控和工业化生产等方面受到了广泛的关注。本文介绍了功能代谢组学技术的分析流程、相关研究方法与平台及其在微生物研究方面的应用,其中重点阐述了真核、原核以及病毒微生物的代谢特性、调控靶点及相关防控策略等。最后,提出功能代谢组学研究在未来面临的问题与挑战,为后续功能代谢组学的研究与发展提供新的思路。     

Metabolome responses of Enterococcus faecium to acid shock and nitrite stress

Enterococcus faecium is gaining increasing interest due to its virulence and tolerance to a range of stresses (e.g., acid shock and nitrite stress in human stomach). The chemical taxonomy and basic structural features of cellular metabolite can provide us a deeper understanding of bacterial tolerance at molecular level. Here, we used hierarchical classification and molecular composition analysis to investigate the metabolome responses of E. faecium to acid shock and nitrite stress. Our results showed that considerable high biodegradable compounds (e.g., dipeptides) were produced by E. faecium under acid shock, while nitrite stress induced the accumulations of some low biodegradable compounds (e.g., organoheterocyclic compounds and benzenoids). Complete genome analysis and metabolic pathway profiling suggested that E. faecium produced high biodegradable metabolites responsible for the proton-translocation and biofilm formation, which increase its tolerance to acid shock. Yet, the presence of low biodegradable metabolites due to the nitrite exposure could disturb the bacterial productions of surface proteins, and thus inhibiting biofilm formation. Our approach uncovered the hidden interactions between intracellular metabolites and exogenous stress, and will improve the understanding of host-microbe interactions.     

Application of Two‐Dimensional NMR Spectroscopy to Metabotyping Laboratory Escherichia coli Strains

NMR Spectroscopy has been established as a major tool for identification and quantification of metabolites in a living system. Since the metabolomics era began, one‐dimensional NMR spectroscopy has been intensively employed due to its simplicity and quickness. However, it has suffered from an inevitable overlap of signals, thus leading to inaccuracy in identification and quantification of metabolites. Two‐dimensional (2D) NMR has emerged as a viable alternative because it can offer higher accuracy in a reasonable amount of time. We employed ¹H,¹³C‐HSQC to profile metabolites of six different laboratory E. coli strains. We identified 18 metabolites and observed clustering of six strains according to their metabolites. We compared the metabolites among the strains, and found that a) the strains specialized for protein production were segregated; b) XL1‐Blue separated itself from others by accumulating amino acids such as alanine, aspartate, glutamate, methionine, proline, and lysine; c) the strains specialized for cloning purpose were spread out from one another; and d) the strains originating from B strain were characterized by succinate accumulation. This work shows that 2D‐NMR can be applied to identify a strain from metabolite analysis, offering a possible alternative to genetic analysis to identify E. coli strains.     

Estimation of binding parameters for the protein–protein interaction using a site-directed spin labeling and EPR spectroscopy

Sensitivity of the electron paramagnetic resonance (CW EPR) to molecular tumbling provides potential means for studying processes of molecular association. It uses spin-labeled macromolecules, whose CW EPR spectra may change upon binding to other macromolecules. When a spin-labeled molecule is mixed with its liganding partner, the EPR spectrum constitutes a linear combination of spectra of the bound and unbound ligand (as seen in our example of spin-labeled cytochrome c ₂ interacting with cytochrome bc ₁ complex). In principle, the fraction of each state can be extracted by the numerical decomposition of the spectrum; however, the accuracy of such decomposition may often be compromised by the lack of the spectrum of the fully bound ligand, imposed by the equilibrium nature of molecular association. To understand how this may affect the final estimation of the binding parameters, such as stoichiometry and affinity of the binding, a series of virtual titration experiments was conducted. Our non-linear regression analysis considered a case in which only a single class of binding sites exists, and a case in which classes of both specific and non-specific binding sites co-exist. The results indicate that in both models, the error due to the unknown admixture of the unbound ligand component in the EPR spectrum causes an overestimation of the bound fraction leading to the bias in the dissociation constant. At the same time, the stoichiometry of the binding remains relatively unaffected, which overall makes the decomposition of the EPR spectrum an attractive method for studying protein–protein interactions in equilibrium. Our theoretical treatment appears to be valid for any spectroscopic techniques dealing with overlapping spectra of free and bound component.     

The role of cholesterol and sphingolipids in the dopamine D1 receptor and G protein distribution in the plasma membrane

G proteins are peripheral membrane proteins which interact with the inner side of the plasma membrane and form part of the signalling cascade activated by G protein-coupled receptors (GPCRs). Since many signalling proteins do not appear to be homogeneously distributed on the cell surface, they associate in particular membrane regions containing specific lipids. Therefore, protein–lipid interactions play a pivotal role in cell signalling. Our previous results showed that although Gαs and Gαi₃ prefer different types of membrane domains they are both co-localized with the D₁ receptor. In the present report we characterize the role of cholesterol and sphingolipids in the membrane localization of Gαs, Gαi₃ and their heterotrimers, as well as the D₁ receptor. We measured the lateral diffusion and membrane localization of investigated proteins using fluorescence recovery after photobleaching (FRAP) microscopy and fluorescence resonance energy transfer (FRET) detected by lifetime imaging microscopy (FLIM). The treatment with either methyl-β-cyclodextrin or Fumonisin B₁ led to the disruption of cholesterol–sphingolipids containing domains and changed the diffusion of Gαi₃ and the D₁ receptor but not of Gαs. Our results imply a sequestration of Gαs into cholesterol-independent solid-like membrane domains. Gαi₃ prefers cholesterol-dependent lipid rafts so it does not bind to those domains and its diffusion is reduced. In turn, the D₁ receptor exists in several different membrane localizations, depending on the receptor's conformation. We conclude that the inactive G protein heterotrimers are localized in the low-density membrane phase, from where they displace upon dissociation into the membrane-anchor- and subclass-specific lipid domain.     

A theoretical model of the pinocytotic vesicular transport process in endothelial cells

A new theoretical model for vesicular transport in single endothelial cells is described using a kinetic molecular approach in which the vesicle diffusion process is coupled with the vesicle attachment/detachment process occurring at the cell plasmalemmal boundaries. Rate constants k^d_i, k_i characterizing a two stage reaction sequence in the attachment/detachment region and the vesicle diffusion coefficient D are obtained by comparison of the theory with the results of tracer studies. For the condition of rapid vesicle loading/discharge of macromolecules it is found that the permeability of endothelial cells to macromolecules tends to be controlled by the vesicular attachment/detachment process rather than the vesicle diffusion process. The rate limiting step in the vesicle attachment/detachment process tends to be the reaction process involving the rate at which a vesicle and the plasmalemmal membrane are brought into/separated from intimate contact rather than that involving the rate of formation/dissolution of the membrane diaphragm of an attached vesicle. Estimated relaxation times for processes occurring in the attachment/detachment region and in the diffusion region, the vesicle transit time in the diffusion region, and the viscosity of the cytoplasm in the diffusion region are deduced. Fair agreement is obtained between the predicted and the observed temperature dependence of the permeability.     

Changes in plasma membrane glycoproteins during differentiation of an established myoblast cell line and a non-fusing α-amanitin-resistant mutant

The glycoproteins of purified plasma membranes from mononucleated myoblasts and from myotubes of the L₆ line were characterized according to their apparent molecular weight (MW) and to their ability to bind concanavalin A (conA). We identified 25 proteins in membranes from mononucleated myoblasts and fused myotubes which specifically bound the lectin. Comparison with the pattern of membrane glycoproteins of a non-fusing mutant allowed us to identify developmentally regulated changes in the accumulation of 8 proteins with an apparent MW of 160, 80, 60, 51.5, 43, 40, 38, and 27 Kilodalton (kD), and changes in the glycosylation of six others which migrate at 215, 150, 135, 90, 85, and 32 kD. Two of them (160 and 38 kD) appeared at fusion, whereas the 51.5 kD band could not be identified in plasma membrane from myotubes. As conA inhibits fusion of myoblasts, it is suggested that at least some of these proteins might be involved in this process.     

Identification of homologous binding proteins in porcine and bovine gametes

The purpose of this study was to determine if sperm and oocyte proteins that mediate plasma membrane interaction during mammalian fertilization are conserved among porcine and bovine gametes. We examined homologous and heterologous sperm and zona-free oocyte interactions to determine the extent of cross-reactivity between the gametes of these two ungulate species. First, the numbers of ejaculated porcine and bovine sperm bound to the oocyte plasma membrane of intact porcine and bovine oocytes were determined in vitro. There was no significant difference between the number of porcine or bovine sperm that bound to porcine or bovine oocytes (P > 0.25). Second, individual porcine and bovine sperm plasma membrane proteins were identified by binding of homologous or heterologous oocyte plasma membrane to whole sperm plasma membrane on Western ligand blots. The relative amount of labeled oocyte plasma membrane bound to individual sperm plasma membrane proteins was analyzed by laser densitometry. Eight porcine sperm plasma membrane proteins and seven bovine sperm plasma membrane proteins were bound by both porcine and bovine oocyte plasma membrane. A significantly greater relative amount of porcine oocyte plasma membrane than bovine oocyte plasma membrane was bound to the 14- and 10-kD porcine sperm plasma membrane proteins (P < 0.001 and P < 0.01, respectively). A 27-kD bovine sperm plasma membrane protein bound proportionally more bovine oocyte plasma membrane probe than porcine oocyte plasma membrane probe (P < 0.04). These results are consistent with conservation of similar receptor ligand interactions at the gamete plasma membrane among porcine and bovine gametes.     

Phosphoproteins and the phosphoenolpyruvate: sugar phosphotransferase system in Salmonella typhimurium and Escherichia coli: evidence for IIImannose, IIIfructose, IIIglucitol, and the phosphorylation of enzyme IImannitol and enzyme IIN-acetylglucosamine

Phosphoproteins produced by the incubation of crude extracts of Salmonella typhimurium and Escherichia coli with either [32P]phosphoenolpyruvate or [gamma 32P]ATP have been resolved and detected using sodium dodecyl sulphate polyacrylamide gel electrophoresis and autoradiography. Simple techniques were found such that distinctions could be made between phosphoproteins containing acid-labile or stable phosphoamino acids and between N1-P-histidine and N3-P-histidine. Phosphoproteins were found to be primarily formed from phosphoenolpyruvate, but because of an efficient phosphoexchange, ATP also led to the formation of the major phosphoenolpyruvate-dependent phosphoproteins. These proteins had the following apparent subunit molecular weights: 65,000, 65,000, 62,000, 48,000, 40,000, 33,000, 25,000, 20,000, 14,000, 13,000, 9,000, 8,000. Major ATP-dependent phosphoproteins were detected with apparent subunit molecular weights of 75,000, 46,000, 30,000, and 15,000. Other minor phosphoproteins were detected. The phosphorylation of the 48,000- and 25,000-MW proteins by phosphoenolpyruvate was independent of the phosphoenolpyruvate:sugar phosphotransferase system (PTS). The PTS phosphoproteins were identified as enzyme I (soluble; MW = 65,000); enzyme IIN-acetylglucosamine (membrane bound; MW = 65,000); enzyme IImannitol (membrane bound; MW = 62,000); IIIfructose (soluble; MW = 40,000); IIImannose (partially membrane associated; MW = 33,000); IIIglucose (soluble; MW = 20,000); IIIglucitol (soluble; MW = 13-14,000); HPr (soluble; MW = 9,000); FPr (fructose induced HPr-like protein (soluble; MW = 8,000). HPr and FPr are phosphorylated on the N-1 position of a histidyl residue while all the others appear to be phosphorylated on an N-3 position of a histidyl residue. These studies identify some previously unknown proteins of the PTS and show the phosphorylation of others, which although previously known, had not been shown to be phosphoproteins.     

Serum Metabolomic Profiling in Acute Alcoholic Hepatitis Identifies Multiple Dysregulated Pathways

Background and Objectives

While animal studies have implicated derangements of global energy homeostasis in the pathogenesis of acute alcoholic hepatitis (AAH), the relevance of these findings to the development of human AAH remains unclear. Using global, unbiased serum metabolomics analysis, we sought to characterize alterations in metabolic pathways associated with severe AAH and identify potential biomarkers for disease prognosis.

Methods

This prospective, case-control study design included 25 patients with severe AAH and 25 ambulatory patients with alcoholic cirrhosis. Serum samples were collected within 24 hours of the index clinical encounter. Global, unbiased metabolomics profiling was performed. Patients were followed for 180 days after enrollment to determine survival.

Results

Levels of 234 biochemicals were altered in subjects with severe AAH. Random-forest analysis, principal component analysis, and integrated hierarchical clustering methods demonstrated that metabolomics profiles separated the two cohorts with 100% accuracy. Severe AAH was associated with enhanced triglyceride lipolysis, impaired mitochondrial fatty acid beta oxidation, and upregulated omega oxidation. Low levels of multiple lysolipids and related metabolites suggested decreased plasma membrane remodeling in severe AAH. While most measured bile acids were increased in severe AAH, low deoxycholate and glycodeoxycholate levels indicated intestinal dysbiosis. Several changes in substrate utilization for energy homeostasis were identified in severe AAH, including increased glucose consumption by the pentose phosphate pathway, altered tricarboxylic acid (TCA) cycle activity, and enhanced peptide catabolism. Finally, altered levels of small molecules related to glutathione metabolism and antioxidant vitamin depletion were observed in patients with severe AAH. Univariable logistic regression revealed 15 metabolites associated with 180-day survival in severe AAH.

Conclusion

Severe AAH is characterized by a distinct metabolic phenotype spanning multiple pathways. Metabolomics profiling revealed a panel of biomarkers for disease prognosis, and future studies are planned to validate these findings in larger cohorts of patients with severe AAH.