Urine Metabolomics Exposes Anomalous Recovery After Maximal Exertion in Female ME/CFS Patients Part 2

Why we will be tired? How can we solve the fatigue problems?

【Contact】Email: george.deng@wecistanche.com / WhatsApp:008613632399501/Wechat:13632399501

2.4. A Pathway Topology Analysis Highlights Altered Carbohydrates and Amino Acid Metabolism as a Result of Exercise 

Differences between ME/CFS and control subjects at the metabolic pathway level were assessed using the pathway analysis module within the Metaboanalyst 5.0 webtool (www. metaboanalyst.ca, accessed on 14 December 2022), which combines quantitative enrichment analysis and pathway topology analysis. In total, 453 out of 734 known compounds of the 1154 compounds analyzed were included in this analysis, due to limitations in Human Metabolome Database (HMDB) ID matching. This analysis was carried out twice for each comparison, with the Kyoto Encyclopedia of Genes and Genomes (KEGG) and then the small molecule pathway database (SMPDB) human reference metabolomes to define pathways. The quantitative enrichment analysis uses the global test to compare the two groups, which employs a logistic regression method to test whether the metabolites in the pathway help to improve the classification of the samples as ME/CFS or control, with the null hypothesis that no metabolite in the pathway has a different concentration in either group [31,32]. The p-values from this test are adjusted for multiple comparisons using the BH procedure,  with q < 0.2 as the significance threshold. This threshold was chosen to allow a review of potentially interesting findings considering the small sample size while ensuring that all significant pathways still have p < 0.05. This analysis is different from the ChemRICH enrichment analysis above in that it allows overlap, so several pathways may have the same key compounds. When evaluating the key compounds, we also report their significance as individual compounds in the LMM, where we employed a more stringent cutoff (q < 0.1). The outcome of the topology analysis is an impact score ranging from 0 to 1. This score is derived from metabolite node importance values calculated using the relative betweenness centrality measure, which is then normalized so that the maximum importance of each pathway is 1. The impact score is the sum of the importance measures for each matching metabolite node in a pathway.

Cistanche can act as an anti-fatigue and stamina enhancer, and experimental studies have shown that the decoction of Cistanche tubulosa could effectively protect the liver hepatocytes and endothelial cells damaged in weight-bearing swimming mice, upregulate the expression of NOS3, and promote hepatic glycogen synthesis, thus exerting anti-fatigue efficacy. Phenylethanoid glycoside-rich Cistanche tubulosa extract could significantly reduce the serum creatine kinase, lactate dehydrogenase, and lactate levels, and increase the hemoglobin (HB) and glucose levels in ICR mice, and this could play an anti-fatigue role by decreasing the muscle damage and delaying the lactic acid enrichment for energy storage in mice. Compound Cistanche Tubulosa Tablets significantly prolonged the weight-bearing swimming time, increased the hepatic glycogen reserve, and decreased the serum urea level after exercise in mice, showing its anti-fatigue effect. The decoction of Cistanchis can improve endurance and accelerate the elimination of fatigue in exercising mice, and can also reduce the elevation of serum creatine kinase after load exercise and keep the ultrastructure of skeletal muscle of mice normal after exercise, which indicates that it has the effects of enhancing physical strength and anti-fatigue. Cistanchis also significantly prolonged the survival time of nitrite-poisoned mice and enhanced the tolerance against hypoxia and fatigue.

Click on chronic fatigue syndrome

Again, there were no significant differences between the urine metabolites at baseline in the ME/CFS and control groups. However, comparing ME/CFS patients and controls at the postexercise time point and the change over time using the within-subject post-exercise/baseline fold change, both had several significantly altered metabolic pathways (Figure 4).

At 24 h post-exercise, seven pathways in the SMPDB and one pathway in KEGG were significantly altered in the ME/CFS patients vs. controls (Figure 4A, q < 0.2). Five of the eight significantly altered pathways are involved in sugar metabolism, which is related to energy production. Fructose is the key compound in all of the sugar metabolism-related pathways, with p < 0.01 in the LMM when comparing the ME/CFS patients and controls post-exercise; however, fructose is not significant in the univariate analysis after multiple comparisons correction. The pathways with the highest impact scores are catecholamine biosynthesis followed by tyrosine metabolism. Both pathways have dehydroascorbate (p < 0.01) as a key compound, but again, this compound is not significant in the LMM after FDR correction. For tyrosine metabolism, ascorbate (vitamin C) (p < 0.01) is an additional key compound but is also not significant after FDR correction.

Ten pathways, all in the KEGG database, had significantly different post-exercise/baseline ratios in the ME/CFS patients vs. controls, indicating they are changing differently over time (Figure 4B, q < 0.2). The four pathways with the highest impact are all involved in amino acid metabolism: arginine and proline metabolism, cysteine and methionine metabolism, lysine degradation, and aminoacyl-tRNA biosynthesis. All of the proteinogenic amino acids are involved in the aminoacyl-tRNA biosynthesis pathway, so it makes sense that the pathway is significantly affected since several of these amino acids are significantly different in this comparison (see Supplementary Figure S2). For the KEGG arginine and proline metabolism pathway, proline is the only compound that is also significant in the LMM (q < 0.1) and also has one of the highest importance scores.

Cysteine changes differently after exercise in the ME/CFS patients vs. controls in the LMM model (q < 0.1) and is the key compound in the KEGG pathways cysteine and methionine metabolism and pantothenate and Coenzyme A (CoA) biosynthesis. CoA is a ubiquitous cofactor that is required for fatty acid metabolism and the tricarboxylic acid (TCA) cycle. Its biosynthesis requires cysteine, which is a unique amino acid as it is the only one that contains a thiol group. Valine is also involved in the significantly altered pantothenate and Coenzyme A (CoA) biosynthesis pathway and has q < 0.1 in the LMM.

The most highly significant pathway for this comparison was ether lipid metabolism, but it had a low impact score. Although only two metabolites that we detected were matched to this pathway, they are both significant in the LMM with q < 0.1: glycerophosphoethanolamine and glycerophosphorylcholine (GPC) (both categorized as phospholipid metabolism by Metabolon®, see Supplementary Figure S2). Purine metabolism was also highly significant, with adenosine 30,50 -cyclic monophosphate (cAMP) as the key compound (q < 0.1 in LMM).

2.5. Acyl Glycines Have Lower Concentrations in the Urine of ME/CFS Patients Compared to Controls 24 Hours Post-Exercise 

The four metabolites found at significantly lower concentrations in ME/CFS vs. control subjects at the post-exercise time point in the univariate LMM analyses (q < 0.1) were 3- hydroxyoctanoylglycine, hexanoylglycine (C6), 2-octenoylglycine, and unknown X–24334. The three known compounds are all in the Metabolon® subpathway of acyl glycine fatty acid metabolism. The heatmap in Figure 5A shows the osmolality-normalized concentration post-exercise for these metabolites for every subject. Using agglomerative hierarchical clustering, the subjects clustered into three groups: (1) six control subjects, (2) two control subjects and one ME/CFS subject, and (3) the remaining nine ME/CFS subjects. The cluster of control subjects predominantly shows higher concentrations for all four metabolites while the cluster of ME/CFS subjects shows lower concentrations for all four metabolites. The small cluster with subjects from both groups shows intermediate values. The boxplots in Figure 5B demonstrate the minimal amount of overlap between the ME/CFS and control groups for these metabolites. Out of those four metabolites, only X-24334 is also changing significantly differently over time between controls and ME/CFS patients in the LMM and is increased after exercise in the control group.

2.6. Metabolites That Are Changing Differently during Exercise Recovery in ME/CFS Patients vs. Controls Are Predominantly Amino Acids and Lipids 

The super pathways with the most altered compounds are amino acid and lipid when considering the 110 compounds that have a significant interaction (q < 0.1) between disease status (ME/CFS vs. control) and time (baseline vs. post-exercise) (Supplementary Figure S2). Figure 6 shows the data for every subject in both groups at both baseline and post-exercise for the significantly altered compounds in several amino acid subpathways (Figure 6A–D). Figure 7 shows the data for all subjects at both baseline and post-exercise for the significantly altered compounds in the selected lipid subpathways (Figure 7A, B). For this figure, we combined the three subpathways involved in acylcarnitine fatty acid metabolism (dicarboxylate, hydroxy, and medium chain) and the four steroid subpathways (androgenic, cortico-, pregnenolone, and progestin). Every compound shown in Figures 6 and 7 is also significantly increasing in urine post-exercise vs. baseline in the sedentary controls.

Four compounds in the urea cycle; arginine and proline metabolism subpathway change differently after exercise in the ME/CFS patients and controls: carboxy-methylarginine, proline, symmetric dimethylarginine (SDMA), and dimethylarginine (ADMA) (Figure 6A). Proline is a building block of collagen and is therefore a key component of connective tissues. SDMA and ADMA are both regulators and competitive inhibitors of nitric oxide (NO) production. NO aids in vascular maintenance in healthy individuals [33],  and decreased NO production is associated with endothelial dysfunction and cardiovascular disease [34]. ADMA can be removed through urinary excretion or it can be degraded in the liver [35]. The increased excretion of SDMA and ADMA in controls but not in patients after exercise implies that controls may be removing excess NO synthase inhibitors to maintain vascular homeostasis and that this beneficial adaptation to exertion may not be occurring in patients. The relationship between NO and ME/CFS is unclear; plasma from ME/CFS  subjects at baseline was found to induce less NO production by endothelial cells in vitro [36], but it is unknown whether or not that was due to higher levels of ADMA or SDMA in ME/CFS plasma, as they were not measured in that study and NO regulation is complex.

Three compounds in the methionine, cysteine, S-adenosylmethionine (SAM), and taurine subpathway are significantly altered: methionine sulfone, cysteine, and s-methylcysteine sulfoxide (Figure 6B). Cysteine is a unique amino acid in that it contains a thiol group and can participate in redox reactions [37]. Cysteine can also be converted into pyruvate, the starting point for the TCA cycle.

The polyamine metabolism subpathway has five significantly different metabolites:  asicoga, (N(1)+N(8))-acetylspermidine, diacetylspermidine, N1, N12-diacetylspermine, and N-acetyl-isoputreanine (Figure 6C). Polyamines have a wide variety of biological functions and are involved in cellular proliferation, differentiation, and apoptosis [38]. An increase in polyamines is part of the normal response to stressors, including exercise [39].

The leucine, isoleucine, and valine metabolism subpathway also has five significantly altered metabolites: 3-methylglutarylcarnitine, tiglyl carnitine, valine, beta-hydroxyisovalerate, and beta-hydroxyisovaleroylcarnitine (Figure 6D). Leucine, isoleucine, and valine are the  branch chain amino acids (BCAAs). These essential amino acids promote protein anabolism in human muscle which helps build muscle following exercise [40]. The catabolism of the three BCAAs leads to energy metabolism pathways and valine is glucogenic, meaning it is converted into glucose precursors which can enter the TCA cycle. While most amino acids are catabolized in the liver, BCAAs are mostly catabolized in other tissues including skeletal muscle, brain,  kidney, and adipose tissue [41]. Isoleucine is both glucogenic and ketogenic, and leucine is ketogenic. These significantly different metabolites are produced during the catabolism of all three BCAAs. 3-methylglutarylcarnitine, as well as 3-hydroxyhexanoylcarnitine (which is categorized as an acylcarnitine by Metabolon®, see Figure 7A), are produced as leucine is converted to acetyl-CoA and acetoacetate. During isoleucine degradation into acetyl-CoA or propionyl-CoA, tiglylcarnitine is produced in two different steps. Beta-hydroxy isovalerate is produced at three different steps of the valine degradation pathway.

Five of the significantly altered compounds between ME/CFS patients and controls are involved in acyl carnitine fatty acid metabolism: pimeloylcarnitine/3-methyladipoylcarnitine, 3- hydroxyhexanoylcarnitine, hexanoylcarnitine, suberoylcarnitine, and 3-hydroxyoctanoylcarnitine (Figure 7A). Acyl carnitines play a key role in long-chain fatty acid β-oxidation, which is the primary mode of energy metabolism during aerobic exercise [42].

Five of the significantly altered compounds are classified as steroids (Figure 7B). 11- ketoetiocholanolone glucuronide is an androgenic steroid. 3alpha,21-dihydroxy-5betapregnane-11,20-dione 21-glucuronide and cortolone glucuronide are corticosteroids. 17alphahydroxypregnanolone glucuronide is a pregnenolone steroid. pregnanediol-3-glucuronide is a progestin steroid. Glucuronides are produced in the liver to aid in the excretion of substances by making them more water-soluble. Corticosteroids function as signaling molecules in a variety of processes, including promoting protein catabolism during exercise or other stressors [43], mediating responses to inflammation [43], and maintaining healthy salt and fluid levels [44]. Altered corticosteroid metabolism could be contributing to orthostatic intolerance, another ME/CFS symptom. Progestin steroids, androgenic steroids, and corticosteroids were also found at lower concentrations in female ME/CFS patient plasma compared to controls in another study, although that study investigated baseline levels only [11].

2.7. The Same Metabolites in Urine and Plasma Are Highly Correlated 

Our group previously published plasma metabolomics data from these same subjects [25]. These subjects underwent the complete two-day CPET protocol and along with urine collection, blood was drawn from each subject at four-time points: baseline (P1), 15–30 min after the CPET (P2), 24 h after the CPET (P3), and 15–30 min after the second CPET (P4) which was performed 24 h after the first CPET (Figure 1A in [25]). Out of the 1403 urine metabolites and 1157 plasma metabolites detected by Metabolon®’s platforms, 727 compounds were measured in both urine and plasma. The relationship between the urine and plasma metabolomes as well as the influence of exercise on this relationship was evaluated by calculating the Pearson correlation coefficients®) between the urine and plasma datasets for each metabolite at all possible time point and time point ratio combinations (Supplementary File S3). For the time points, we chose to focus on the combinations of urine and plasma from the same day (baseline urine (U1) with P1 and P2, and post-exercise urine (U3) with P3 and P4). We also examined correlations between the post-exercise/baseline ratio in urine (U3/U1) and three different post-exercise ratios in plasma (P4/P1, P3/P2, P3/P1) to explore how the metabolite levels changed during the 24 h recovery period in urine vs. plasma. The p-values were calculated for each correlation using a t-test with the null hypothesis of R = 0 (BH FDR correction, q < 0.15). The number of strong correlations, which we defined as R > 0.7 or R < −0.7, amounts to approximately 40% of the 727 metabolites when looking at time point correlations (Figure 8). Notably, for all time point pairs, there are very few strong negative correlations between urine and plasma. For the ratio correlations, the number of strong negative correlations is increased in the healthy controls. For all correlations, the number of strong positive correlations is higher in the ME/CFS patients than in the controls.

2.8. Probing Compounds with Correlations between Urine and Plasma That Are Different in ME/CFS Patients and Controls 

We proceeded to screen for compounds with the most significant differences between controls and patients, using the following stringent criteria: (1) |R| > 0.7, p < 0.05, and q < 0.15  in either ME/CFS patients or controls; (2) R < 0.3 with the same sign or an R-value with an opposite sign (i.e., negative if the significant correlation was positive) in the other cohort (controls or patients); (3) compounds that had extreme outliers usually affecting the linear relationship were removed (modified z-score method of outlier detection, with a threshold of z > 6). When the outlier was only found in the time point data, that compound was removed for all but only the time point comparisons. When the outlier was found in the ratio data,  that compound was removed for all but only the ratio comparisons. The summary of the compounds that met the above criteria is displayed as a heatmap of R values in Figure 9.

The heatmap contains metabolites spanning 35 subpathways with an overrepresentation of subpathways in the amino acid superpathway, 11 out of 15. Within the amino acid  subpathways, tryptophan metabolism as well as leucine, isoleucine, and valine metabolism,  had the most affected compounds, with 9/21 (43%) and 8/27 (30%), respectively

Kynurenate is part of the tryptophan pathway and is one of the metabolites with the most drastic difference in correlation coefficients between the ME/CFS and the control cohorts (Figure 9, P4/P1 with U3/U1). The kynurenate correlation graphs for all comparisons from the heatmap of Figure 9 are shown in Figure 10A. We can see the inverted correlations in the “P4/P1 with U3/U1” graph where R = 0.74 (ME/CFS) and R = −0.77 (controls). The strong positive correlation in the ME/CFS patients is consistent throughout the ratio correlations, whereas the negative correlation in the controls is not as consistent. The time point correlations show a similar pattern, with a strong positive correlation for the ME/CFS group. While kynurenate is the only tryptophan compound with correlations in both the time point and the ratio sides of the heatmap, the differences seen in plasma and urine correlations in the other eight compounds, which appear at various locations in the tryptophan pathway, attest to a profound dysregulation of this pathway in the ME/CFS patients compared to the controls. Additionally, another compound on the heatmap, quinolinate, is the metabolite that links tryptophan metabolism to nicotinate and nicotinamide metabolism, which is a crucial pathway for the formation of NAD+ and NADP+. A dysregulation in the kynurenate pathway has been hypothesized to be the underlying cause of ME/CFS pathophysiology due to its central role in cellular energy production and involvement in mediating the immune response as reviewed by Kavyani et al. [45].

Eight compounds in the leucine, isoleucine, and valine subpathway had differences in the correlations between ME/CFS patients and controls for the selected time point and ratio comparisons (Figure 9). For five of these compounds, the differences were in the postexercise ratios, with four out of the five compounds having differences when correlating the U3/U1 (24 h post-exercise/baseline urine) with P3/P2 (the 24 h post-exercise/15 min postexercise plasma). Beta-hydroxy isovalerate is one such compound, with a strong and significant positive correlation in the ME/CFS patients between U3/U1 and P3/P2 and a weak, non-significant negative correlation in the controls (Figure 10B). Beta-hydroxy isovalerate also changes significantly over time in the urine of the ME/CFS patients compared to the controls in the LMM (Figure 6D). These eight compounds span all three branches of the BCAA catabolism pathway. Isovalerylglycine and isovalerylcarntine are produced during leucine catabolism (ketogenic). 2-methylbutrylcarnitine and 3-methyl-2-isovalerate are produced during isoleucine catabolism (ketogenic and glucogenic). Betahydroxyisovalerate, as mentioned above, is downstream of valine catabolism (glucogenic). This is further evidence that there is dysfunctional metabolic recovery from exercise in the ME/CFS patients related to BCAA catabolism, which is affecting all three BCAAs. The three BCAAs have a common enzyme involved in the first step of the pathway, BCAA  aminotransferase. However, considering that there is dysregulation in so many amino acid subpathways, it is likely that this is evidence of a more global metabolic problem.

Within the lipid superpathway, four subpathways about steroids also caught our attention. Indeed, eight steroid compounds from four subpathways had different correlations between plasma and urine in the ME/CFS patients and controls, including androgenic steroids, corticosteroids, pregnenolone steroids, and progestin steroids (Figure 9). Pregnanediol-3-glucuronide, which is a progestin steroid and a product of progesterone catabolism, has a strong and significant positive correlation (R = 0.78) between U3/U1 and P3/P1 in the ME/CFS patients and a strong and significant negative correlation in the healthy controls (R = −0.8) (Figure 10C). In the ME/CFS patients, when pregnanediol- 3-glucuronide increases in plasma 24 h post-exercise, it also increases in the urine and vice versa. Whereas in the healthy controls, the subjects with the largest increases in urine concentration of pregnanediol-3-glucuronide 24 h post-exercise have a decrease in plasma levels. This same trend is seen in the urine and plasma correlation for the other ratio comparisons. This compound is also changing significantly differently in the LMM between ME/CFS patients and controls, where the controls have a consistent post-exercise increase in urine concentration that is not seen in the ME/CFS patients (Figure 7B). At all four-time point comparisons, the plasma and urine levels of pregnanediol-3-glucuronide are highly correlated in both groups of subjects, which has been shown before [46]. It is only when examining the change over time after exercise that the differences between the ME/CFS patients and healthy controls emerge. Although pregnanediol-3-glucuronide levels are not reported in acute exercise studies, it has been measured throughout menstrual cycles in exercising vs. sedentary females, and the exercising females typically have lower urinary levels overall compared to sedentary females [46,47]. Given our results, it is possible that acute exercise initially leads to an increase in urinary pregnanediol-3-glucuronide levels in healthy sedentary females as they are excreting it and not replacing it. This healthy response to exercise is not occurring in the ME/CFS patients, which is yet further evidence for their overall altered metabolic response to exercise.

We generated another heatmap that contains unknowns, partially characterized molecules, and food components meeting the same criteria used to generate Figure 9 (Supplementary Figure S5). This is provided as additional information to illustrate the potential of some yet-to-be-identified metabolites. As an example, X–25524 consistently shows a strong positive correlation for the ME/CFS group but no correlation for the control group regardless of exercise. Identifying such a compound could potentially help develop a diagnostic marker for ME/CFS by measuring blood and urine concentrations.

【Contact】Email: george.deng@wecistanche.com / WhatsApp:008613632399501/Wechat:13632399501