Blood serum metabolites reflect the physiological state of the body and allow measurement of changes occurring during disease or an infection. Metabolomics can be used to capture these changes and provide more accurate ways of diagnosing diseases, providing prognostic and allowing a better understanding of the pathogen’s mechanism of action. Viral infections are expected to change nucleotide, carbohydrate, lipid, and amino acid metabolisms. Although COVID-19 has been investigated employing metabolomics, our knowledge about the metabolic changes specifically occurring upon SARS-CoV-2 infection is still limited.
This work aimed to define serum metabolites that may differentiate patients that tested positive vs. negative for SARS-CoV-2 among two cohorts of patients – inpatients with severe symptoms under ventilation and ones with moderate symptoms. The first group of inpatients with severe respiratory symptoms, peripheral capillary oxygen saturation (SpO2) lower than 93% on room air or PaO2 lower than 83 (mmHg), were grouped into (A) the group with 21 individuals that tested positive for SARS-CoV-2 with golden standard RT-qPCR test, named positive patients (PP), and (B) the group with 20 individuals that tested negative for SARS-CoV-2, named PN, and those were compared with each other, and with the samples from healthy donors (46) collected before the pandemic, named healthy donors – H, matched by gender and age with the inpatient subgroups. The second group of patients (n = 235) showed moderate symptoms indicative for COVID-19, with 20% that tested negative to SARS-CoV-2 – named N (n = 47), while those that tested positive – named P (n = 188) – were treated at home and were compared with each other and to the samples from the healthy donors (n = 47, P vs N vs H). Informed consent was obtained from all participants, and the present study was approved by the Research Ethics Committees.
All serum samples (n = 369) were obtained from 5 mL of peripheral blood collected in a dry tube after peripheral venipuncture, realized for the routine hematological procedure. After sampling, blood was allowed to clot for 30 minutes, centrifuged at 4,500 ×g for 15 minutes at 4 °C, centrifugation supernatant was carefully separated and aliquoted (200 μL) into cryovials and stored at -80 °C. SARS-CoV-2 detection was done using the golden standard test by RT-qPCR from nasopharyngeal samples. Blood serum samples (200 μL) were thawed on ice, mixed with 250 μL of deuterium oxide (D2O 99.9%, Cambridge Isotope Laboratories, Inc., Andover, USA) at room temperature, centrifuged at 12,000 ×g for 2 minutes at 4 °C and transferred into 5 mm NMR tubes. High-resolution 1H-NMR NOESY1D 1D (noesy1dgppr1d), CPMG (cpmgpr1d), and edited by diffusion (stebpgp1s191d) spectra were acquired on the Bruker AVANCE III 600 MHz spectrometer using the inverse triple-core probe (TBI) at 25 °C. Non-polar and polar metabolites were isolated from serum samples using extraction. Lipids were dissolved in 450 μL of chloroform-d (CCl3D with 0.03% (v/v) of TMS), at room temperature, transferred into 5 mm NMR tubes, and high-resolution 1H-NMR (zg30) was acquired on the Bruker AVANCE III 600 MHz spectrometer using the inverse triple-core probe (TBI) at 25 °C. The serum 1H-NMR spectra had their baseline and phase-corrected manually, aligned, referenced (in the case of sera: 3H, δ 1.33, J = 7.0 Hz, doublet; and lipids – TMS signal: 12H, δ 0.00, s) in MestreNova Inc. software, and exported as data in .csv format for the chemometrics’ analyses. Groups of spectra were classified by type of donor, i.e., H – healthy, P – patients, PP – RT-qPCR positive severe inpatients, and PN – severe inpatients negative in RT-qPCR. The second group of the investigated individuals followed the grouping into positive to SARS-CoV-2 (P, n = 47) and negative (N, n = 47). Spectra were normalized by mean centering, HDO peak (4.50 – 5.20 pm) was excluded except for lipid extracts 1H-NMR and binned (0.005 ppm). Chemometrics were done using the MetaboAnalyst 3.0 software platform (https://www.metaboanalyst.ca/MetaboAnalyst/faces/home.xhtml). All 1H-NMR data sets showed the obvious class separation among the studied groups and the subgroups.
In severe COVID-19 inpatients at least fifty serum metabolites were found as strongly affected with the severity of the symptoms, and 60% of those were increased. For example, high concentrations of glucose, lipids, some amino acids, acids, cofactors were different in COVID-19 patients. Cholesterol and lysophosphocholines were decreased, but phosphorylcholines were increased. Serum proteins were decreased in COVID-19 patients and could be linked to liver, kidney, and heart altered functions, and/or malabsorption disorders. Besides, p-hydroxyphenyl acetic acid concentration showed to be positively correlated with severe COVID-19, and inversely correlated to lung function. Activation of the kynurenine pathway was positively associated with the disease severity and linked to significant down-regulation of T-cell lymphocyte function. Another important ratio between NAD(P)+ to glucose was decreased with the disease severity and lung dysfunction. All increased metabolites’ levels point to aggravation of symptoms in individuals with metabolic syndrome and shed a light on probable mitochondrial involvement in response to COVID-19.
On the other side, 1H-NMR differences in moderate COVID-19 patients were less evident, and visible in aliphatic spectral regions, principally in CPMG spectra. Serum lipids differed, although VLDL and LDL were somewhat less altered in moderate when compared to severe COVID-19 patients. Differences in serum proteins and glucose were less pronounced in moderate patients. So, there must be a connection between glucose excess, high VLDL, and low serum proteins with the disease severity. The blood serum metabolic changes point to probably two greatly affected cell organelles in COVID-19 patients: (1) endoplasmatic reticulum, which is responsible for VLDL assembly and production, as well anabolism of glycoconjugates (glycoproteins), and (2) mitochondria that suffer from low oxygen supply, adapt metabolism to new circumstances, but not rarely start the cytokine storm, because of the problems in membrane potential, transport, and change in catabolism. Finally, metabolomics by 1H-NMR could be successfully used to evaluate the disease severity, and CMPG and spectra edited by diffusion might be used to distinguish the severity of the symptoms, while discovered COVID-19 signatures might add-in greatly in clinical evaluation.