Cholinergic and lipid mediators crosstalk in Covid-19 and the impact of glucocorticoid therapy

64 Cytokine storms and hyperinflammation, potentially controlled by glucocorticoids, occur 65 in COVID -19; the roles of lipid mediators and acetylcholine (ACh) and how 66 glucocorticoid therapy affects their release in Covid-19 remain unclear . Blood and 67 bronchoalveolar lavage (BAL) samples from SARS-CoV-2- and non -SARS-CoV-2-68 infected subjects were collected for metabolomic/lipidomic, cytokine s, soluble CD14 69 (sCD14), and ACh, and CD14 and CD36 -expressing monocyte/macrophage 70 subpopulation analyses. Transcriptome reanalysis of pulmonary biopsies was performed 71 by assessing coexpression, differential expression, and biological networks. Correlations 72 of lipid mediators, sCD14, and ACh with glucocorticoid treatment were evaluated. This 73 study enrolled 190 participants with Covid-19 at different disease stages, 13 hospitalized 74 non-Covid-19 patients, and 39 healthy -participants. SARS -CoV-2 infection increased 75 blood levels of arachidonic acid (AA), 5 -HETE, 11 -HETE, sCD14, and ACh but 76 decreased monocyte CD14 and CD36 expression. 5 -HETE, 11-HETE, cytokines, ACh, 77 and neutrophils were higher in BAL than in circulation (fold-change for 5-HETE 389.0; 78 11-HETE 13.6; ACh 18.7, neutrophil 177.5 , respectively ). Only AA was higher in 79 circulation than in BAL samples (fold-change 7.7). Results were considered significant 80 at P<0.05, 95%CI. Transcriptome data revealed a unique gene expression profile 81 associated with AA, 5 -HETE, 11 -HETE, ACh, and their receptors in C ovid-19. 82 Glucocorticoid treatment in severe/critical cases lowered ACh without impacting disease 83 outcome. We first report that pulmonary inflammation and the worst outcomes in Covid-84 19 are associated with high levels of ACh and lipid mediators. Glucocorticoid therapy 85 only reduced ACh , and we suggest that treatment may be started early, in combination 86 with AA metabolism inhibitors, to better benefit severe/critical patients. 87 All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for thisthis version posted January 15, 2021. ; https://doi.org/10.1101/2021.01.07.20248970doi: medRxiv preprint Page 5 of 84

88 Individuals C ovid-19 may present asymptomatically or with manifestations 89 ranging from acute respiratory distress syndrome to systemic h yperinflammation and 90 organ failure, events attributed to cytokine storms1. Free polyunsaturated fatty acids, such 91 as AA and derivative eicosanoids, regulate inflammation2,3, yet their role in Covid-19 has 92 not been well investigated. 93 ACh, which is released by nerves 4, leukocytes 5, and airway epithe lial cells 6, 94 regulates metabolism 7, cardiac function 8, airway inflammation 9, and cytokine 95 production10, all of which occur in Covid-19. It is known that eicosanoids stimulate ACh 96 release3, but crosstalk between cholinergic and lipid mediator pathways in Covid-19 still 97 need to be clarified. 98 In this study, levels of lipid mediators, ACh, and other inflammatory markers in 99 blood and BAL from patients with Covid-19 who were treated or not with glucocorticoids 100 were compared to those of non-Covid-19 and healthy-participants. Moreover, lung biopsy 101 transcriptome reanalysis data from Covid-19 and non-Covid-19 patients corroborated our 102 findings. 103

104 Study design and blood collection 105 This observational, analytic, and transversal study was conducted from June to 106 November 2020. All participants were over 16 years old and chosen according to the 107 inclusion and exclusion criteria described in Table S1 and in the protocol, after providing 108 signed consent. Blood samples collected from patients posi tive for Covid -19 (n=190) 109 were analyzed by RT -qPCR (Biomol OneStep/Covid -19 kit; Institute of Molecular 110 Biology of Paraná - IBMP Curitiba/PR, Brazil) using nasopharyngeal swabs and/or 111 All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for thisthis version posted January 15, 2021. ; https://doi.org/10.1101/2021.01.07.20248970doi: medRxiv preprint Page 6 of 84 serological assays to detect IgM/IgG/IgA (SARS -CoV-2® antibody test; Gua ngzhou 112 Wondfo Biotech, China). Samples obtained from a cohort of SARS -CoV-2-negative 113 healthy participants were used as controls (n=39). Participants positive for Covid-19 were 114 categorised as asymptomatic -mild (n=43), moderate (n=44), severe (n=54), or crit ical 115 (n=49). The criteria for the clinical classification of patients were defined at the time of 116 sample collection, as shown in Table S1. Peripheral blood samples were obtained by 117 venous puncture from patients upon their first admission and/or during the period of 118 hospitalisation at two medical centres, Santa Casa de Misericordia de Ribeirão Preto and 119 Hospital Sao Paulo at Ribeirão Preto, São Paulo State, Brazil. Blood samples from 120 healthy controls and asymptomatic -mild non -hospitalized participants were c ollected 121 either at the Centre of Scientific and Technological Development “Supera Park” 122 (Ribeirão Preto, São Paulo State, Brazil) or in the home of patients receiving at -home 123 care. The plasma was separated from whole blood samples and stored at −80°C. For 124 lipidomic and metabolomic analyses, 250 µL of plasma was stored immediately in 125 methanol (1:1 v/v). Lipidomic and metabolomic analyses were performed by mass 126 spectrometry (LC-MS/MS), while a cytokines, sCD14, and ACh were quantified using 127 CBA flex Kit (flow cytometer assay) or commercial ELISA. The expression levels of 128 CD14, CD36, CD16, and HLA-DR in cells were evaluated by flow cytometry following 129 the gate strategy (Figure S2). 130 Ethical considerations 131 All participants provided written consent in accordance with the regulations of the 132 Conselho Nacional de Pesquisa em Humanos (CONEP) and the Human Ethical 133 Committee from Faculdade de Ciências Farmacêuticas de Ribeirão Preto (CEP-FCFRP-134 USP). The research protocol was approved and received the certificate of Pre sentation 135 and Ethical Appreciation (CAAE: 30525920.7.0000.5403). The sample size was 136 All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for thisthis version posted January 15, 2021. ; https://doi.org/10.1101/2021.01.07.20248970doi: medRxiv preprint Page 7 of 84 determined by the convenience of sampling, availability at partner hospitals, agreement 137 to participate, and the pandemic conditions within the local community (more information 138 in the Protocol). 139 Bronchoalveolar lavage fluid (BAL) collection and processing 140 BAL fluids were collected from hospitalised Covid-19 patients at the severe or critical 141 stages of disease (n=32) to assess their lung immune responses. Control samples we re 142 obtained from hospitalised intubated donors negative for SARS -CoV-2 (n=13) (as 143 certified by SARS-CoV-2-negative PCR), referred to as non-Covid-19 patients, who were 144 intubated because of the following primary conditions: bacterial pneumonia, abdominal 145 septic shock associated with respiratory distress syndrome, pulmonary atelectasis due to 146 phrenic nerve damage, or pulmonary tuberculosis. BAL fluid was collected as previously 147 described11, using a siliconized polyvinylchloride catheter (Mark Med, Porto Alegre, 148 Brazil) with a closed Trach Care endotracheal suction system (Bioteque Corporation, 149 Chirurgic Fernandes Ltd., Santana Parnaíba, Brazil) and sterile 120 mL polypropylene 150 flask (Biomeg-Biotec Hospital Products Ltd., Mairiporã, Brazil) under aseptic conditions. 151 Approximately 5–10 mL of bronchoalveolar fluid was obtained and placed on ice for 152 processing within 4 h. The BAL fluids were placed into 15-mL polypropylene collection 153 tubes and received half volume of their volume of phosphate buffered saline (PBS) 0.1 154 M (2:1 v/v) in relation to the total volume of each sample. After centrifugation (700 × g, 155 10 min), the supernatants of the BAL fluid were recovered and stored at –80°C. For 156 lipidomic and metabolomic analyses, 250 µL of these supernatants were st ored 157 immediately in methanol (1:1 v/v). Subsequently, the remaining BAL fluid was diluted in 158 10 mL of PBS and gently filtered through a 100 -µm cell strainer (Costar, Corning, NY, 159 USA) using a syringe plunger. The resulting material was used for cytokine an d 160 acetylcholine (ACh) quantification. The BAL fluids were centrifuged (700 × g, 10 min) 161 All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for thisthis version posted January 15, 2021. ; https://doi.org/10.1101/2021.01.07.20248970doi: medRxiv preprint Page 8 of 84 and the red blood cells were lysed using 1 mL of ammonium chloride (NH 4Cl) buffer 162 0.16 M for 5 min. The remaining airway cells were washed with 10 mL of PBS, 163 resuspended in PBS–2% heat-inactivated foetal calf serum, and counted with Trypan blue 164 using an automated cell counter (Countess, Thermo Fisher Scientific, Waltham, MA, 165 USA). The leukocyte numbers were adjusted to 1 × 109 cells/L for differential counts and 166 1 × 106 cells/mL for flow cytometry analysis. All procedures were performed in a Level 167 3 Biosafety Facility (Departamento de Bioquímica e Imunologia, Faculdade de Medicina 168 de Ribeirão Preto, Universidade de São Paulo). 169 Data collection 170 The electronic medical recor ds of each patient were carefully reviewed. Data 171 included sociodemographic information, comorbidities, medical history, clinical 172 symptoms, routine laboratory tests, immunological tests, chest computed tomography 173 (CT) scans, clinical interventions, and outcomes (more information in the Protocol). The 174 information was documented on a standardised record form, as indicated in Tables S1, 175 S2, and S3. Data collection of laboratory results included first -time examinations within 176 24 h of admission, defined as the primary endpoint. The secondary endpoint was clinical 177 outcome (death or recovery). 178 Clinical laboratory collection 179 For hospitalised patients, blood examinations were performed by clinical analysis 180 laboratories at their respective hospitals. Blood examinations of healthy participants and 181 non-hospitalized patients were performed at Serviço de Análises Clínicas (SAC), 182 Departamento de Análises Clínicas, Toxicológicas e Bromatológicas of the Faculdade 183 de Ciências Farmacêuticas de Ribeirão Preto, Univers idade de São Paulo, Ribeirão 184 All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for thisthis version posted January 15, 2021. ; https://doi.org/10.1101/2021.01.07.20248970doi: medRxiv preprint Page 9 of 84 Preto, São Paulo, Brazil. The blood samples were used to measure for liver and kidney 185 function, myocardial enzyme spectrum, coagulation factors, red blood cells, 186 haemoglobin, platelets, and total and differential leukocytes using automated equipment. 187 Similarly, the absolute numbers of leukocytes in the BAL fluid were determined in a 188 Neubauer Chamber with Turkey solution. For the counts of differential leukocytes in the 189 BAL, 100 µL of the fluid was added to cytospin immediately after collection to avoid any 190 interference on cell morphology. Differential leukocyte counts were conducted using an 191 average of 200 cells after staining with Fast Panoptic (LABORCLIN; Laboratory 192 Products Ltd, Pinhais, Brazil) and examined under an optical microscope (Zeiss EM109; 193 Carl Zeiss AG, Oberkochen, Germany) with a 100× objective (immersion oil) equipped 194 with a Veleta CCD digital camera (Olympus Soft Imaging Solutions Gmbh, Germany) 195 and ImageJ (1.45s) (National Institutes of Health, Rockville, MD, USA)12. Lymphocytes, 196 neutrophils, eosinophils, and monocytes/macrophages were identified an d 197 morphologically characterised, and their lengths and widths were measured (100×). 198 High-performance liquid chromatography coupled with tandem Mass Spectrometry 199 (LC-MS/MS) assay 200 Reagents 201 Eicosanoids, free fatty acids (AA, EPA, and DHA), and metabolites as molecular 202 weight standards (MWS) and deuterated internal standards were purchased from Cayman 203 Chemical Co. (Ann Arbor, MI, USA). HPLC -grade acetonitrile (ACN), methanol 204 (MeOH), and isopropanol were purchased from Merck (Kenilworth, NJ, USA). Ultrapure 205 deionised water (H 2O) was obtained using a Milli -Q water purification system (Merck -206 Millipore, Kenilworth, NJ, USA). Acetic acid (CH 3COOH) and ammonium hydroxide 207 (NH4OH) were obtained from Sigma Aldrich (St. Louis, MO, USA). 208 All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for thisthis version posted January 15, 2021. ; https://doi.org/10.1101/2021.01.07.20248970doi: medRxiv preprint Page 10 of 84 Sample preparation and extraction 209 The plasma (250 μL) in EDTA -containing tubes (Vacutainer ® EDTA K2; BD 210 Diagnostics, Franklin Lakes, NJ, USA) and BAL (250 μL) samples were stored in MeOH 211 (1:1, v/v) at ‒80°C. Three additional volumes of ice-cold absolute MeOH were added to 212 each sample overnight at ‒20°C for protein denaturation and after lipid solid -phase 213 extraction (SPE). To each sample, 10 μL of internal standard (IS) solution was added, 214 centrifuged at 800 × g for 10 min at 4°C. The resulting supernatants were collected and 215 diluted with deionised water (ultrapure water; Merck-Millipore, Kenilworth, NJ, USA) to 216 obtain a MeOH concentration of 10% (v/v). In the SPE extractions, a Hypersep C18-500 217 mg column (3 mL) (Thermo Scientific-Bellefonte, PA, USA) equipped with an extraction 218 manifold collector (Waters -Milford, MA, USA) was used. The diluted samples were 219 loaded into the pre -equilibrated column and washing using 2 mL of MeOH and H 2O 220 containing 0.1% acetic acid, respectively. Then, the cartridges were flushed with 4 mL of 221 H2O containing 0.1% acetic acid to remove hydrophilic impurities. The lipids that had 222 been adsorbed on the SPE sorbent were eluted with 1 mL of MeOH containing 0.1% 223 acetic acid. The eluates solvent was removed in vacuum (Concentrator Plus, Eppendorf, 224 Germany) at room temperature and reconstituted in 50 μL of MeOH/H2O (7:3, v/v) for 225 LC-MS/MS analysis. 226 LC-MS/MS analysis and lipids data processing 227 Liquid chromatography was performed using an Asce ntis Express C18 column 228 (Supelco, St. Louis, MO, USA) with 100 × 4.6 mm and a particle size of 2.7 μm in a high-229 performance liquid chromatography (HPLC) system (Nexera X2; Shimadzu, Kyoto, 230 Japan). Then, 20 μL of extracted sample was injected into the HPLC column. Elution was 231 carried out under a binary gradient system consisting of Phase A, comprised of H 2O, 232 All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for thisthis version posted January 15, 2021. ; https://doi.org/10.1101/2021.01.07.20248970doi: medRxiv preprint Page 11 of 84 ACN, and acetic acid (69.98:30:0.02, v/v/v) at pH 5.8 (adjusted with NH4OH), and Phase 233 B, comprised of ACN and isopropanol (70:30, v/v). Gradient elution was performed for 234 25 min at a flow rate of 0.5 mL/min. The gradient conditions were as follows: 0 to 2 min, 235 0% B; 2 to 5 min, 15% B; 5 to 8 min, 20% B; 8 to 11 min, 35% B; 11 to 15 min, 70% B; 236 and 15 to 19 min, 100% B. At 19 min, the gradient was returne d to the initial condition 237 of 0% B, and the column was re -equilibrated until 25 min. During analysis, the column 238 samples were maintained at 25°C and 4°C in the auto -sampler. The HPLC system was 239 directly connected to a TripleTOF 5600+ mass spectrometer (SCIEX-Foster, CA, USA). 240 An electrospray ionisation source (ESI) in negative ion mode was used for high-resolution 241 multiple-reaction monitoring (MRM HR) scanning. An atmospheric -pressure chemical 242 ionisation probe (APCI) was used for external calibrations of the calibrated delivery 243 system (CDS). Automatic mass calibration (<2 ppm) was performed periodically after 244 each of the five sample injections using APCI Negative Calibration Solution (Sciex -245 Foster, CA, USA) injected via direct infusion at a flow rate of 300 μL/min. Additional 246 instrumental parameters were as follows: nebuliser gas (GS1), 50 psi; turbo gas (GS2), 247 50 psi; curtain gas (CUR), 25 psi; electrospray voltage (ISVF), ‒4.0 kV; temperature of 248 the turbo ion spray source, 550°C. The dwell time was 10 ms, an d a mass resolution of 249 35,000 was achieved at m/z 400. Data acquisition was performed using Analyst TM 250 software (SCIEX- Foster, CA, USA). Qualitative identification of the lipid species was 251 performed using PeakViewTM (SCIEX-Foster, CA, USA). MultiQuantTM (SCIEX-Foster, 252 CA, USA) was used for the quantitative analysis, which allows the normalisation of the 253 peak intensities of individual molecular ions using an internal standard for each class of 254 lipid. The quantification of each compound was performed using in ternal standards and 255 calibration curves, and the specific mass transitions of each lipid were determined 256 All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for thisthis version posted January 15, 2021. ; https://doi.org/10.1101/2021.01.07.20248970doi: medRxiv preprint Page 12 of 84 according to our previously published method 13. The fin al concentration of lipids was 257 normalised by the initial volume of plasma or BAL fluid (ng/mL). 258 Metabolomics analysis 259 Metabolite was extracted and samples were transferred to autosampler vials for 260 LC–MS analysis using TripleTOF5600+ Mass Spectrometer (Sciex-Foster, CA, USA) 261 coupled to an ultra -high-performance liquid chromatography (UHPLC) system (Nexera 262 X2; Shimadzu, Kyoto, Japan). Reverse -phase chromatography was performed similarly 263 to lipids analyses above. Mass spectral data were acqui red with negative electrospray 264 ionisation, and the full scan of mass -to-charge ratio ( m/z) ranged from 100 to 1500. 265 Proteowizard software 14 was used to convert the wiff files into mz XML files. Peak 266 peaking, noise filtering, retention time, m/z alignment, and feature quantification were 267 performed using apLCMS15. Three parameters were used to define a metabolite feature: 268 mass-to-charge ratio ( m/z), retention time (min), and intensity values. Data were log 2 269 transformed and only features detected in at least 50% of samples from one g roup were 270 used in further analyses. Missing values were imputed using half the mean of the feature 271 across all samples. Mummichog (version 2) was used for metabolic pathway enrichment 272 analysis (mass accuracy under 10 ppm)16. 273 Acetylcholine measurement 274 ACh was measured in heparinized plasma (SST ® Gel Advance ®; BD Diagnostics, 275 Franklin Lakes, NJ, USA) and in BAL using a commercially available 276 immunofluorescence kit (ab65345; Abcam, Cambridge, UK) according to the 277 manufacturer’s instructions. Briefly, ACh was converted to choline by adding the enzyme 278 acetylcholinesterase to the reaction, which allows for total and free-choline measurement. 279 The amount of ACh present in the samples was calculated by subtracting the free choline 280 All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for thisthis version posted January 15, 2021. ; https://doi.org/10.1101/2021.01.07.20248970doi: medRxiv preprint Page 13 of 84 from the total choline. The products formed in the assay react with the choline probe and 281 can be measured by fluorescence with excitation and emission wavelengths of 535 and 282 587 nm, respectively (Paradigm Plate Reader; SpectraMax, San Diego, CA, USA). The 283 concentration of ACh was analysed using SoftMax ® software (SpectraMax, Molecular 284 Devices, Sunnyvale, CA, USA), expressed as pmol.mL-1. 285 Soluble CD14 (sCD14) measurement 286 Samples from heparinized plasma (SST ® Gel Advance®; BD Biosciences, Franklin 287 Lakes, NJ, USA) were placed in 96 -well plates. The concentration of sCD14 was 288 determined using an ELISA kit (DY383; R&D Systems, Minneapolis, MN, USA), 289 following the manufacturer’s instructions, expressed as pg.mL-1. 290 Flow Cytometry 291 Uncoagulated blood samples in EDTA-containing tubes (Vacutainer® EDTA K2; BD 292 Biosciences) were processed for flow cytometry analysis of circulating leukocytes. 293 Whole blood (1 mL) was separated and red blood cells were lysed using RBC lysis buffer 294 (Roche Diagnostics GmbH, Mannheim, GR). Leukocytes were washed in PBS containing 295 5% foetal bovine serum (FBS) (Gibco™, USA), centrifuged, and resuspended in Hank’s 296 balanced salt solution (Sigma-Aldrich, Merck, Darmstadt, Germany) containing 5% FBS, 297 followed by surface antigen staining. Similarly, cells obtained from BAL fluid were 298 processed for flow cytometry assays. Briefly, cells were stained with Fixable Viability 299 Stain 620 (1:1000) (BD Biosciences) and incubated with monoclonal antibodies specific 300 for CD14 (1:100) (M5E2; Biolegend), HLA-DR (1:100) (G46-6; BD Biosciences), CD16 301 (1:100) (3G8; Biolegend), and CD36 (1:100) (CB38, BD Biosciences) for 30 min at 4°C. 302 Stained cells were washed and fixed with BD Cytofix™ Fixation B uffer (554655; BD 303 Biosciences, San Diego, CA, USA). Data acquisition was performed using a LSR -304 All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for thisthis version posted January 15, 2021. ; https://doi.org/10.1101/2021.01.07.20248970doi: medRxiv preprint Page 14 of 84 Fortessa™ flow cytometer (BD Biosciences, San Jose, CA, USA) and FACS -Diva 305 software (version 8.0.1) (BD Biosciences, Franklin Lakes, NJ, USA). For the analysis, 306 300,000 events were acquired for each sample. Data were evaluated using FlowJo ® 307 software (version 10.7.0) (Tree Star, Ashland, OR, USA) to calculate the cell frequency, 308 dimensionality reduction, and visualisation using t -distributed stochastic neighbour 309 embedding. Gate strategy performed as described before 17, as shown in Figure S2. 310 Cytokine Measurements 311 The cytokines interleukin (IL) -6, IL -8, IL -1ß, IL -10, and tumour necrosis factor 312 (TNF) were quantified in heparinized plasma and BAL fluid samples using a BD 313 Cytometric Bead Array (CBA) Human Inflammatory Kit (BD Biosciences, San Jose, CA, 314 USA), according to manufacturer’s instructions. Briefly, after sample processing , the 315 cytokine beads were counted using a flow cytometer (FACS Canto TM II; BD 316 Biosciences, San Diego, CA, USA), and analyses were performed using FCAP Array 317 (3.0) software (BD Biosciences, San Jose, CA, USA). The concentrations of cytokines 318 were expressed as pg.mL-1. 319 Re-analysis of transcriptome data from lung biopsies of patients with Covid-19 320 To gain a better understanding of the correlation between the altered concentrations 321 of ACh, AA, and AA-metabolites detected in the plasma and BAL fluid of severe/critical 322 Covid-19 patients, we performed a new analysis by re -using a previously published 323 transcriptome open dataset 18, deposited in the Gene Expression Omnibus repository 324 under accession no. GSE150316 19. We used transcriptome data from lung samples 325 (n=46) from patients with Covid -19 (n=15), seven of which displayed a low viral load 326 and eight a high viral load, and non -Covid-19 patients (n=5) with other pulmonary 327 illnesses (negative control). Patients with a high viral load had meantime periods of 328 All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for thisthis version posted January 15, 2021. ; https://doi.org/10.1101/2021.01.07.20248970doi: medRxiv preprint Page 15 of 84 hospital stays (3.6±2.2 days) and duration of illness (7.2±3.02 days) shorter than the 329 patients with low viral load (14±7.9 and 19±4.9 days, respectively), as described by the 330 authors of the public data source 18. Hence, for analysis purposes, all patient samples were 331 grouped into four classifications: Covid-19 (CV), Covid-19 low viral load (CVL), Covid-332 19 high viral load (CVH), and non -Covid-19 (NCV). The strategy for reanalysing the 333 transcriptome was implemented according to three consecutive steps: (i) co -expression 334 analysis, (ii) differential expression analysis, and (iii) biological network construction. 335 Initially, for the co-expression study, normalised transcriptome data in log 2 of reads per 336 million (RPM) were filtered by excluding non-zero counts in at least 20% of the samples. 337 Next, the selected genes were explored in the R package Co -Expression Modules 338 identification Tool (CEMITool) 20, using a p-value of 0.05 as the threshol d for filtering. 339 Then, the co-expression modules were analysed for the occurrence of ACh and AA genes 340 list obtained from the Reactome pathways 21, as w ell as the Covid -19-related genes 341 obtained from the literature (Supplementary Appendix I). Next, differential gene 342 expression between samples from the lung biopsy transcriptome (CV, NCV, CVL, and 343 CVH) was measured using the DESeq2 package 22, with p-values adj usted using the 344 Benjamini and Hochberg method 23. The list of differentially express genes (DEGs) 345 generated for all comparisons was filtered from the genes listed in Supplementary 346 Appendix I, considering the values of log 2 of fold -change (FC) greater than 1 347 (|log2(FC)|>1) and adjusted p<0.05. Finally, a first -order biological network was 348 constructed using co-expression module(s) containing genes associated with the ACh and 349 AA pathways to characterise the interplay between these mediators in combination with 350 Covid-19 severity markers, as well as to identify relevant DEGs and hub genes in this 351 network, using the BioGRID repository 24. The networks were constructed, analysed, and 352 graphically represented using the R packages igraph 25, Intergraph 26, and ggnetwork 27. 353 All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for thisthis version posted January 15, 2021. ; https://doi.org/10.1101/2021.01.07.20248970doi: medRxiv preprint Page 16 of 84 Due to the substantial inf luence of glucocorticoid treatment on the levels of some 354 mediators, we measured the sensitivity of genes from the differential expression analysis 355 between CV samples from patients who underwent treatment (CTC, three patients and 356 ten samples) and patients w ho were not treated (NCTC, 12 patients and 36 samples), as 357 previously described 18. 358 Statistical Analysis 359 Two-tailed tests were used for the statistical analysis, with a significance value of 360 p <0.05 and a confidence interval of 95%. The data were evaluated for a normal 361 distribution using the Kolmogorov –Smirnov test. The parametric data were analysed 362 using unpaired t-tests (for two groups) or one-way ANOVA followed by Tukey’s multiple 363 comparison tests for three or more groups simultaneously. For data that did not display a 364 Gaussian distribution, Mann -Whitney (for two -group comparisons) or Kruskal -Wallis 365 testes were used, followed by Dunn’s post-tests for analysis among three or more groups. 366 The cytokine network data in patients with Covid -19 were analysed using significant 367 Spearman’s correlations at p<0.05. Data were represented by connecting edges to 368 highlight positive stro ng (r ≥ 0.68; thick continuous line), moderate (0.36 ≥ r r r ≤ ‒0.36; thinner dashed line), or weak 371 (-0.36 0; thin dashed line), as proposed previously 28,29. The absence of a line 372 indicates the non-existence of the relationship. The Venn diagrams were elaborated using 373 the online tool Draw Venn Diagram (http://bioinformatics.psb.ugent.be/webtools/Venn/). 374 The results were tabulated using GraphPad Prism software (version 8.0) and the 375 differences were considered statistically significant at p<0.05. See the Additional 376 Statistical Report section for mor e information. Some of the confounding variables 377 associated with Covid -19 (age, sex, obesity, hypertension, and diabetes mellitus) were 378 All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for thisthis version posted January 15, 2021. ; https://doi.org/10.1101/2021.01.07.20248970doi: medRxiv preprint Page 17 of 84 analysed for their potential impacts on the main analytical procedures of this study, such 379 as ACh, AA, 5 -HETE, and 11 - HETE measurements of the plasma (healthy, 380 asymptomatic-to-mild, moderate, severe, and critical patients) and BAL (severe and 381 critical patients) samples. This analysis was performed using the Kruskal-Wallis, Mann-382 Whitney, Spearman’s correlation, or Chi-square (χ2) tests (Table S7-S9). 383

384 Study Population 385 This study enrolled 39 healthy -participants, 13 hospitalized non -Covid-19, and 386 190 Covid-19 patients aged 16-96 years from April to November 2020. The 190 C ovid-387 19 patients were categorized as having asymptomatic-to-mild (n=43), moderate (n=44), 388 severe (n=54), or critical (n=49) disease (Table S2). 389 Covid-19 Modifies Circulating Soluble Mediators and Cell Populations 390 To determine whether SARS -CoV-2 infection alter s the metabolism of lipid 391 mediators, we used high -resolution sensitive mass spectrometry to perform targeted 392 eicosanoid analysis and nontargeted metabolomics using plasma from healthy -393 participants and Covid-19 patients. In total, 8,791 metabolite features were present in at 394 least 50% of a ll samples, and t he relative abundance of 595 metabolite features (FDR 395 adjusted P<0.05) was altered in the groups studied (Figure 1A). Two -way hierarchical 396 clustering based on these significant metabolite features resulted in three clear clusters : 397 one for severe/critical C ovid-19 patients, one for healthy -participants and one for 398 asymptomatic-to-mild and moderate Covid-19 (Figure S1A). Pathway analysis revealed 399 the top significant metabolic pathways to be enriched in features involved in fatty acid 400 biosynthesis, metabolism, activation , and oxidation (Figure 1B). Compared to healthy -401 All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for thisthis version posted January 15, 2021. ; https://doi.org/10.1101/2021.01.07.20248970doi: medRxiv preprint Page 18 of 84 participants, tentative metabolite annotations suggested an increased abundance of fatty 402 acids (FFAs), such as linoleic acid, tetradecanoate, dodecanoate and AA, in COVID -19 403 (Figure 4C-F). Among the identified lipids, AA was the most abundant, and its levels 404 correlated with the severity of Covid-19 (Figure 1F). Linoleic acid can be metabolized to 405 AA, which in turn is a substrate for eicosanoids, such as 5 -hydroxyeicosatetraenoic acid 406 (5-HETE) and 11 -hydroxyeicosatetraenoic acid (11 -HETE) (Figure 1G , 1 H); both 407 molecules with function in the immune response. Overall, the increased plasma levels of 408 AA in Covid-19 indicate that it predicts disease severity. 409 As eicosanoids induce cell recruitment and regulate immune responses, we next 410 determined the profile of immune cells and soluble mediators in whole blood and plasma 411 of patients with C ovid-19 and healthy-participants. According to whole-blood analysis, 412 absolute leukocyte and ne utrophil counts (Figure 2A , 2B) were significantly higher but 413 lymphocyte counts (Figure 2C) significantly lower in patients with severe/critical disease 414 than in those with moderate disease. Eosinophil (Figure 2D) and basophil counts (Figure 415 2E) were reduced severe disease compared to asymptomatic and moderate disease. 416 Although no differences in total monocyte counts among the groups (Figure 2F) 417 were observed based on CD16, CD14 and HLA-DR, expression of membrane CD14 was 418 reduced in all SARS-CoV-2-infected patients compared to healthy -participants (Figure 419 2G; Figure S2 , S3). In parallel, CD36 expression was decreased in monocytes from 420 patients with severe/critical disease (Figure 2H), but sCD14 was increased only in plasma 421 from critical patients (Figure 2I). We did not detect differences in classic or non -classic 422 monocytes, but the percentage of intermediate monocytes was decreased in all SARS -423 CoV-2-infected participants compared to healthy-participants (Figure 2J, 2K, 2L). Based 424 on plasma cytokine level analysis, patients with moderate, severe, and critical disease 425 All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for thisthis version posted January 15, 2021. ; https://doi.org/10.1101/2021.01.07.20248970doi: medRxiv preprint Page 19 of 84 share a Covid-19 cytokine profile defined by increased IL-8, IL-6, and IL-10 (Figure 2M, 426 2N, 2Q) levels, with unaltered IL-1β and TNF levels (Figure 2O, 2P). 427 Covid-19 Induces Strong Lung Responses 428 We performed measurements of BAL from hospitalized Covid-19 and non - 429 Covid-19 patients. Changes in l ung metabolomics induced by SARS -CoV-2 infection 430 (Figure 3A ; Figure S1B ) included alterations in sphingolipids, beta oxidation of 431 trihydroxyprostanoil-CoA, biosynthesis and metabolism of steroidal hormones, vitamin 432 D3, and glycerophospholipids (Figure 3B). We also evaluated AA and its metabolites . 433 Despite no differences in AA, levels of 5-HETE and 11-HETE in BAL were significantly 434 higher in Covid-19 than in non-Covid-19 patients, though other metabolites did not differ 435 between these groups (Figure 3C). When assessing leukocytes in BAL between the 436 patient groups, we found no differences in total or differential counts, with the exception 437 of lymphocyte numbers (Figure 3D, 3E). In contrast to eicosanoids, cytokine profiles in 438 Covid-19 and non -Covid-19 patients were similar (Figure 3F), suggesting that lipid 439 mediators contribute to the pathophysiological processes induced by SARS -CoV-2. 440 Interestingly, we observed a significant reduction in classical (Figure 3G) and 441 intermediate (Figure 3H) monocytes in BAL from Covid-19 patients. In parallel, CD14 442 and CD36 expression in monocyte was lower in BAL of Covid-19 than in that of non -443 Covid-19 patients (Figure 3I, 3J). 444 Acetylcholine, Cytokines and Eicosanoids are Higher in the Lung Micro -445 Environment 446 We compared systemic and lung responses only in patients with severe/critical 447 Covid-19 and found significantly higher levels of cytokines in BAL than in blood (Figure 448 4A). When comparing lipid mediator levels in either compartment from the same Covid-449 All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for thisthis version posted January 15, 2021. ; https://doi.org/10.1101/2021.01.07.20248970doi: medRxiv preprint Page 20 of 84 19 patients, we detected lower AA but higher levels of 5 -HETE and 11-HETE in BAL 450 than in blood (Figure 4B , 4C, 4 D). Previous results from our group 3 have shown that 451 eicosanoids contribute to ACh release. Therefore, we next measured ACh in patients with 452 Covid-19 and observed higher plasma levels of ACh in patients with Covid-19; in 453 addition, ACh was elevated in patients with severe/critical disease compared with those 454 with asymptomatic/moderate disease or healthy-participants (Figure 4E). Unexpectedly, 455 in patients with severe/critical Covid-19, ACh levels in BAL were 10 -fold higher than 456 those in serum, and patients treated with glucocorticoids showed decreases in ACh in 457 both compartments (Figure 4F, 4G). Interesting, neutrophil counts were higher in BAL , 458 as these cells produce high levels of 5-HETE and IL-1β, both of which are mediators of 459 ACh release 30,31 (Figure 4H). Correlation analysis was then performed to evaluate the 460 relationship between eicosanoids, cytokines, sCD14 and ACh in patients with Covid-19. 461 When comparing blood samples from all patients, we detected strong correlations 462 between ACh versus IL-1β and moderate correlations between ACh versus AA. A 463 substantial number of interactions between AA and its metabolites and between cytokines 464 and eicosanoids were observed (Figure 4I ; Figure S4 A). Correlations among all 465 parameters were also observed in BAL (Figure 4J; Figure S4B). To evaluate the benefit 466 of glucocorticoid treatments and their relationship with eicosanoids, CD14 and ACh, we 467 analysed the intersections in a Venn diagram of blood and BAL from severe/critical 468 Covid-19. The results showed that treatment of Covid-19 with glucocorticoids did not 469 have a significant influence on eicosanoid or sCD14 release in patients with more severe 470 stages of disease; in critical patients, however, reductions of 44% and 65% in ACh levels 471 in blood and BAL, respectively , were observed (Figure 4K -4Q; Figure S5). These 472 findings suggest that the use of glucocorticoids ha s a positive effect on resolution of the 473 All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for thisthis version posted January 15, 2021. ; https://doi.org/10.1101/2021.01.07.20248970doi: medRxiv preprint Page 21 of 84 inflammatory process because they reduce ACh release, which directly or indirectly 474 stimulates cell recruitment and proinflammatory mediator release. 475 Altered Expression of Acetylcholine and Arachidonic Acid Pathway Genes in Lung 476 Biopsies From Some Covid-19 Patients 477 In this study, we reanalysed the lung biopsy transcriptome from Covid-19 patients 478 to evaluate expression of ACh and AA pathway genes (Supplementary Appendix I) . 479 These genes were coexpressed only in a single module (M1) that included genes related 480 to the ACh release cycle, AA metabolism, cholinergic and eicosanoid receptors, and 481 biomarkers of C ovid-19 severity (Figure 5A ; Table S4 ; Supplementary Appendix II ). 482 Furthermore, expression of nearly all genes was upregulated in some deceased Covid-19 483 patients with long hospital stays and low viral loads (Figure 5B ; Figure S6A; 484 Supplementary Appendix I II). These differentially expressed genes populated the 485 biological network and are likely under the action of some hubs (Figure 5C ; 486 Supplementary Appendix IV), such as oestrogen receptor II (ESR2) and albumin (ALB). 487 We identified a unique proinflammatory gene expression profile in lung biopsies re lated 488 to cholinergic and eicosanoid receptors (Figure 5H, 5I; Figure S6B, S6C). In combination 489 with the altered levels of ACh, AA, and AA metabolites found in patient from our cohort, 490 the transcriptome data reported here in strengthen the likelihood that these mediators 491 contribute to Covid-19 severity (Figure 5D-5G). Interestingly, lung samples from Covid-492 19 patients with short hospital stays and high viral loads did not present this unique 493 expression profile of ACh or AA pathway genes (Figure 5H, 5I; Figure S6B, S6C). Some 494 Covid-19 patients treated versus not with glucocorticoid showed transcript expression of 495 monoglyceride lipase (MGLL) and N-acylethanolamine acid amidase (NAAA) and up-496 regulation of fatty acid amide hydrolase (FAAH), which are involved in the production 497 of AA from endocannabinoid (cases 3, 9, and 11; Figure 5 B; Supplementary Appendix 498 All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for thisthis version posted January 15, 2021. ; https://doi.org/10.1101/2021.01.07.20248970doi: medRxiv preprint Page 22 of 84 III). Besides, we detected other DEGs in biopsy samples from Covid -19 patients 499 (glucocorticoid-treated versus non -treated) associated with AA, ACh, interferon 500 pathways, and Covid-19 biomarkers (Supplementary Appendix III). 501

502 A consensus is building around the fatal effects of SARS-CoV-2 infection, which is 503 increasingly believed to cause death as a result of systemic hyperinflammation and multi-504 organ collapse32, secondary to systemic cytokine storms 33–35. However, few studies have 505 compared pulmonary and systemic inflammation 36 or considered the contribution of 506 eicosanoids and neurotransmitters to these effects. In our study, we hypothesised that, in 507 Covid-19, pulmonary cells and leukocytes, in addition to cytokines, release eicosanoids 508 and ACh, me diating local and systemic manifestations. In patients with severe/critical 509 SARS-CoV-2 infection, we found that ACh, 5 -HETE, 11 -HETE, and cytokines were 510 more abundant in the lung than under systemic conditions. In contrast, only the levels of 511 AA were found to be higher in the circulation than in the BAL fluid. Interestingly, in 512 patients with severe/critical disease who were treated with corticosteroids, only ACh was 513 inhibited. 514 In our study, we compared the lung and systemic responses in association with the 515 lung transcriptome and demonstrated a robust correlation between lipid mediators, 516 neurotransmitters, and their receptors in SARS -CoV-2 infection. However, contrary to 517 what has been suggested by previous studies 37,38, only small amounts of eicosanoids were 518 found in the plasma of patients with severe/critical disease, with significant differences 519 observed only in AA, 5 -HETE, and 11 -HETE. Interestingly, the levels of 5 -HETE and 520 11-HETE were found to be remarkably higher in BAL, as well as AA in plasma, 521 suggesting that AA and its metabolites mediate responses to Covid-19. 5-HETE induces 522 All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for thisthis version posted January 15, 2021. ; https://doi.org/10.1101/2021.01.07.20248970doi: medRxiv preprint Page 23 of 84 neutrophil recruitment 39, pulmonary oedema 40, and ACh release 30. It is releas ed by 523 human neutrophils, and its esterified form promotes IL -8 secretion 31. Unlike the 524 esterified form 31, free-5-HETE did not inhibit NETs formation, a key event in Covid-19 525 41. Remarkably, 11-HETE originating from monocytes/macrophages 42, endothelial cells 526 43, and platelets 44 is induced by hypoxia 45, IL-1 46, contributes to ACh functions 43, and 527 inhibits insulin release 47. The involvement of AA was confirmed by bioinformatic 528 analyses that identified the expression or upregulation of genes related to AA metabolism 529 and eicosanoid receptors in some lung biopsies. These included the OXER1 gene, which 530 encodes a receptor for AA and 5 -HETE 48 which med iates neutrophil 531 activation/recruitment 49,50. Our metabolomic analysis also showed an increase in the 532 plasma AA and linoleic acid levels, similar to a plasma lipidome performed by Schwarz 533 et al., suggesting a strong correlation between lipid mediators and Covid -19 severity 51. 534 Accordingly, the ELOVL2 gene, which is involved in linoleic acid metabolism and AA 535 synthesis 52, was upregulated in lung biopsies. 536 The RNA expression of ALOX5, an enzyme that participates in lipid mediator 537 production, is upregulated in some immune cell types from severe Covid -19 patients 53, 538 and has been detected in the lung biopsies of deceased Covid -19 patients 18. In contrast 539 to our lung transcriptome re-analysis findings, the expression of cytochrome p450 (CYP) 540 enzymes, which are also involved in lipid mediator gen eration, was not detected in the 541 peripheral blood mononuclear cells (PBMC) transcriptome re -analysis from severe 542 Covid-19 patients 53. One possible explanation for this contradiction is that Covid-19 is a 543 heterogeneous illness composed of distinct tissue gene expression associated with Covid-544 19 severity and different immunopathological profiles in inf ected tissues 18,54. This 545 immunological heterogeneity could be related to two types of evolution of severe Covid-546 19 patients, one associated with a high viral load and susceptibility to SARS -CoV-2 547 All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for thisthis version posted January 15, 2021. ; https://doi.org/10.1101/2021.01.07.20248970doi: medRxiv preprint Page 24 of 84 infection, a shorter hospitalisation time, and exudative diffuse alveolar damage, and 548 another associated with a low or undetectable viral load, a mixed lung histopathological 549 profile, and a longer hospitalisation time associated with non -homeostatic pulmonary 550 inflammation 18,54,55. In this context, we found that the expression levels of some lipid 551 mediators and ACh and AA pathway genes varied in the plasma and BAL samples of 552 severe/critical patients, which were found to be preferentially activated in the lungs of 553 deceased Covid -19 patients with low viral loads, long hospitalisation times, and 554 damaging lung inflammation. 555 Nevertheless, it remains unclear whether AA and linoleic acid contribute to host 556 protection 56 and tissue damage, or whether they represent a viral escape mechanism. AA 557 is known to directly interact with the virus, reducing its viability and inducing membrane 558 disturbances, disfavouring SARS-CoV-2 entry 57–59. On the other hand, reduced plasma 559 AA levels may be associated with lung injury and poor outco mes in Covid-19 infection 560 60. Within this context, Shen and et al. found lower concentrations of AA in survivors of 561 severe Covid-19 infection 37. However, in the present study, the levels of AA were found 562 to be increased in the plasma of patients who died from severe Covid-19, suggesting that 563 the potential benefits of AA are overcome by a higher production of its proinflammatory 564 metabolites, 5-HETE and 11-HETE. Interestingly, in our cohort, AA, 5 -HETE, and 11-565 HETE were not altered in the plasma or BAL fluid of patients with severe/critical Covid-566 19 infection who had received glucocorticoid treatment. This may be explained by the 567 fact that AA, in addition to calcium -activated-PLA2 61 also originates from 568 glucocorticoid-insensitive phospholipase 62, adipocyte destruction 63, linoleic acid 64, or 569 the degradation of endocannabinoids by FAAH 65, a macrophage enzyme detected in 570 SARS-CoV-2-infected patients 66. Notably, some CVL Covid -19 patients, regardless of 571 whether they were treated or not with glucocorticoids, showed transcript expression of 572 All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for thisthis version posted January 15, 2021. ; https://doi.org/10.1101/2021.01.07.20248970doi: medRxiv preprint Page 25 of 84 MGLL and NAAA, as well as an up-regulation of FAAH and all of the enzymes involved 573 in the production of AA from endocannabinoid 67. In addition, we detected other DEGs 574 in biopsy samples (glucocorticoid-treated versus non-treated) associated with AA, ACh, 575 interferon pathways, and Covid -19 biomarkers. Glucocorticoids have been used widely 576 to reduce the morbidity and mortality rates of Covid -19 patients, but have not been 577 effective for all patients 68. The mortality rate for glucocorticoid -treated hospitalised 578 Covid-19 patients from our cohort was higher than that reported in the RECOVERY trial 579 and similar to that described in the CoDEX trial conducted in Brazil 69. High mortality 580 rates could be associated with several factors, including a low mean PaO2:FiO2 ratio and 581 overloaded public health syst ems in countries with limited resources, such as Brazil 69 582 and as well as glucocorticoid dose, initiation, and duration of therapy 70,71. In addition, 583 the precise threshold at which a patient should be treated with glucocorticoids, that is to 584 avoid the manifestation of adverse effects associated with comorbidities and inefficient 585 clearance of SARS-CoV-2, remains unclear 72. However, a delayed start of glucocorticoid 586 therapy could result in a lack of response due to patients reaching a point of no return, as 587 suggested previously b y our research group with regards to scorpion poisoning, which 588 triggers a sterile inflammatory process 3. On the other hand, considering that Covid-19 is 589 a non-sterile hyperinflammatory disease, the administ ration of glucocorticoids is more 590 effective after the initial phase in which hospitalised patients have low or undetectable 591 viral loads 18,68. Our data suggest that SARS-CoV-2 infection activates the glucocorticoid-592 insensitive release of AA metabolites associated with Covid -19 severity. Hence, lipid 593 mediator production pathways could be important molecular targets for Covid -19 594 treatment, as suggested by other researchers 53. Accordingly, we suggested that, in order 595 to improve the benefits of glucocorticoid therapy in hospitalized patients, who show 596 All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for thisthis version posted January 15, 2021. ; https://doi.org/10.1101/2021.01.07.20248970doi: medRxiv preprint Page 26 of 84 lower or undetectable viral loads, patients should be treated as soon as possible in 597 combination with AA metabolism inhibitors. 598 CD36 is expressed in several cells 73,74, where it induces Ca ++ mobilisation, cell 599 signalling, LTB4 production, and cellular fatty acid uptake 75–78. The maintenance of high 600 free-AA levels in patients with severe/critical glucocorticoid -treated disease may result 601 from the reduction of CD36 on macrophage membranes. Intriguingly, CD36 is a gustatory 602 lipid sensor 79, whose deficit in cell membranes may account for the symptoms of ageusia, 603 anosmia, diarrhoea, hyperglycaemia, platelet aggregation, and cardiovascular 604 disturbances experienced by Covid -19 patients 75–77. CD36 is a substrate of matrix 605 metalloproteinase-9 (MMP -9) and disintegrin metalloproteinase domain -containing 606 protein 17 (ADAM17) 80,81. Decreased CD36 expression in Covid-19 patients most likely 607

from the action of these enzymes. MMP-9 is produced by neutrophils 82, induced 608 by TNF-α 83, unaffected by glucocorticoids, and associated with respiratory syndrome in 609 Covid-19 84. ADAM17 has been described as a target for Covid-19 treatment 85. 610 In agreement with previous reports 1,34,86,87, in the present study, high 611 concentrations of IL-6, IL-8, and IL-10 were detected, along with insignificant levels of 612 IL-1 in plasma. In contrast, the inflammatory cytokine TNF -α was found at the lowest 613 concentration in BAL fluid, and insignificant levels were detected in the blood. The low 614 levels of TNF -α production are correlated with the decreased counts of intermediate 615 monocytes in the blood and BAL, which are major sources of this cytokine 88. As 616 expected, in the BAL from patients with severe/critical Covid -19, the levels of IL-8, IL-617 6, IL-1, TNF-α, and IL -10 were significantly higher compared with the plasma of all 618 participants, or when paired with patients’ own plasma. Notably, in BAL, the most 619 abundant cytokine was IL -8, which is produced by neutrophils 89, lung macrophages 90, 620 and bronchial epithelial cells 10, and is induced by 5-HETE 31 and ACh 10. Interestingly, 621 All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for thisthis version posted January 15, 2021. ; https://doi.org/10.1101/2021.01.07.20248970doi: medRxiv preprint Page 27 of 84 BAL fluid from non-Covid-19 patients also presented a high concentration of cytokines 622 that did not differ from that of patients with Covid-19. This suggests that cytokine storms 623 are not a key differential factor in this disease. Contrary to our expecta tions, 624 glucocorticoids did not alter cytokine production (Table S7). As a result, we suggest that, 625 regardless of whether Covid-19 is treated with glucocorticoids, cross-talk occurs between 626 cytokines, lipid mediators, and ACh, as previously reported 2,3,91. IL-1 induces 11-HETE 627 46; arterial relaxation induced by ACh is mediated by 11 -HETE and is inhibited by 628 indomethacin 43. In our studies on scorpion envenomation, ACh release was found to be 629 mediated by PGE 2-induced by IL -1 and inhibited by indomethacin 3. In the present 630 study, comparing AA, 5 -HETE, 11 -HETE, cytokine, and receptor expression 631 demonstrated the absence of glucocorticoid effects. These results, in addition to the 632

of other studies from our laboratory 3, suggest that glucocorticoids did not have an 633 effect in our cohort. This was most likely due to the fact that treatment was started late, 634 namely after inflammation has been triggered by infection. In this context, Covid -19 635 patients could had already reached the point of no return, similar to w hat has been 636 previously described in scorpion envenomation 3. 637 As predicted by Virgilis and Giovani 92, neurotransmitters are produced in Covid -638 19. ACh induces mucus secretion, bronchoconstriction, lu ng inflammation and 639 remodelling 9, cardiac dysfunction in scorpionism 3, NETs formation 93, IL-8 release 10, 640 thrombosis 94, and obesity-related severity 95. In addition to the nervous system 4, ACh is 641 also produced by pulmonary vessels 96, airway epithelial cells 6, and immune cells 5. When 642 binding to anti-inflammatory neural nicotinic receptors induces AA release 97, it inhibits 643 nicotine receptors 97, which are highly expressed in lungs 98. In our cohort, a correlation 644 between the AA and ACh levels and the severity of Covid -19 infection was observed, 645 which was reinforced by the results of our bioinformatics study. Our lung transcriptome 646 All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for thisthis version posted January 15, 2021. ; https://doi.org/10.1101/2021.01.07.20248970doi: medRxiv preprint Page 28 of 84 re-analyses showed that ACh release cycle genes were activated, including acetylcholine-647 synaptic release (synaptotagmin 1) and neuronal choline transporter (Solute Carrier 648 Family 5 Member 7, SLC5A7). Accordingly, SLC5A7 gene expression was higher in the 649 lung of patients who died from Covid-19 infection than patients who survived 99. Because 650 this gene mediates the translocation of choline into lung epithelial cells 100 and 651 macrophages 101, the production of ACh in BAL may also depend on non-neuronal cells. 652 Indeed, the ACh repressor gene in non-neural cells (RE1 Silencing Transcription Factor) 653 is reduced in the lungs of patients who died of Covid -19 99. The activation of T -654 lymphocyte EP4 induces the release of ACh 102, suggesting that AA metabolites 655 contribute to the release of ACh from lung and immune cells. As expected, ACh 656 production in patients with sev ere/critical disease was found to be inhibited by 657 glucocorticoids, since this drug blocks ACh production by lung epithelial cells 103–105. 658 Patients with Covid -19 (mostly methylprednisolone -treated patients) showed lower 659 plasma levels of choline and higher plasma levels of phosphocholine 37. In addition, a 660 distinct pattern of ACh receptor mRNA expression was found in patients who died of 661 Covid-19 (mainly in the CVL -lengthy hospital stay group). This profile is characterised 662 by low levels of expression of the nicotinic receptor encoded by cholinergic receptor 663 nicotinic alpha 7 subunit gene (CHRNA7) and increased levels of expression of t wo 664 nicotinic and one muscarinic receptor in deceased -CVL compared to deceased non -665 Covid-19 patients. These proteins are encoded by the cholinergic nicotinic alpha 3 subunit 666 gene (CHRNA3), the cholinergic nicotinic receptor 5 subunit gene (CHRNA5), and the 667 muscarinic cholinergic receptor 3 gene (CHRM3), respectively. These three upregulated 668 genes were also co -expressed and associated with pulmonary inflammatory disorders 669 9,106–110. Notably, SARS-CoV-2 spike glycoprotein interacts with the α7 nicotinic 670 All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for thisthis version posted January 15, 2021. ; https://doi.org/10.1101/2021.01.07.20248970doi: medRxiv preprint Page 29 of 84 acetylcholine receptor, which may compromise the cholinergic anti -inflammatory 671 pathway111. 672 The ACh-nicotinic receptor, encoded by CHRNA7, is expressed in various lung 673 cells and macrophages 112. It dis plays extensive anti -inflammatory activities, including 674 the inhibition of pro-inflammatory cytokine production 113, the recruitment of neutrophils 675 114, and the reduction of CD14 expression in human monocytes 115. The increased degrees 676 of inflammation in the BAL fluid of patients with severe/critical Covid -19 may be 677 associated with the lower levels of anti -inflammatory CHRNA7 expression. However, a 678 reduction in monocyte CD14 may result from high levels of IL-6 production 116, since the 679 levels of CHRNA7 expression were very low. Interestingly, the ACh -M3 receptor in 680 immune cells has been found to be up -regulated by ACh, which mediates its pro -681 inflammatory actions, including the production of IL-8 and the recruitment of neutrophils 682 117. Furthermore, the interaction between ACh and its receptor triggers the AA -derived 683 release of eicosanoids, including 5-HETE 97,118,119. Remarkably, vitamin D modulates the 684 ACh-M3 receptor 120, which may explain its beneficial effects in the prevention of Covid-685 19 121. Based on the anti -inflammatory effects of nicotinic receptors and the 686 downregulation of the SARS-CoV-2 receptor ACE2 promoted by nicotine, therapies 687 involving nicotinic receptors have been proposed to treat Covid -19 infection 122,123. In 688 fact, based on these findings, an acetylcholinesterase inhibitor 124,125 and a nicotinic 689 receptor agonist have been tested 126,127. However, recent studies have demonstrated that 690 nicotine and smoking increase ACE2 receptor density 128,129, and that the potential 691 beneficial effects of nicotine were restricted to a small group of individuals 130. Our 692 findings provide a warning that precautions should be taken when considering the 693 therapeutic use of nicotinic agonists, since patients with severe/critical Covid -19 were 694 found to release high amounts of ACh, in addition to showing increased levels of M3 -695 All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for thisthis version posted January 15, 2021. ; https://doi.org/10.1101/2021.01.07.20248970doi: medRxiv preprint Page 30 of 84 receptor expression. Due to the antiviral and anti-inflammatory effects of AA, it has been 696 proposed to treat Covid -19 patients 60. Nevertheless, AA may be not be an adequate 697 therapeutic target, based on both our findings and previous studies 131, which have shown 698 that AA can favour hyperinflammation and lethality in Covid-19 infection. In conclusion, 699 ACh, AA, 5-HETE, and 11-HETE mediate the innate immune response to SARS-CoV-2 700 and may define the outcome of infection. 701 Covid-19 severity can be associated with disruption of homeostatic lipidome a nd 702 metabolic alterations and this phenomenon may be influenced by comorbidities/risk 703 factors related to the infection. Our findings demonstrated that (i) high plasma levels of 704 ACh and lipid mediators positively correlated with Covid -19 severity; and (ii) o nly 705 hypertension and/or age were confounding variables for analyzing the association of high 706 plasma levels of AA, 5-HETE, and ACh with disease severity (Table S7). Similarly, the 707 correlation between altered plasma profile of lipid mediators and Covid -19 severity is 708 associated with selective comorbidities – mainly high body mass index (BMI) – but 709 poorly associated with gender, advanced age, and diabetes. However, the correlation 710 between altered AA and/or 5 -HETE levels and disease severity is associated with male 711 gender, hypertension, and heart disease, but not BMI53. 712 We also examined whether glucocorticoid -therapy interfered with the potential 713 effect of some confounding variables on the correlation between high ACh levels and 714 Covid-19 severity in severe/critical patients. Despite the relatively underrepresented 715 samples from non-glucocorticoid treated patients, no confounding variable significantly 716 affected the correlation between plasma and BAL ACh levels and Covid -19 severity in 717 severe/critical patients treated or not with glucocorticoids (Table S8). The altered plasma 718 and BAL levels of the lipid mediators AA, 5 -HETE, and 11 -HETE in severe/critical 719 patients were associated with the disease severity but not with the confounding variables 720 All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for thisthis version posted January 15, 2021. ; https://doi.org/10.1101/2021.01.07.20248970doi: medRxiv preprint Page 31 of 84 tested (Table S9). Altogether, the findings here reported suggest that age and/or 721 hypertension are significant confounding variables for analyzing the association between 722 increased levels of cholinergic and lipid mediators and Covid -19 severity. Also, the 723 confounding variables tested probably did not modify the inhibitory action of 724 glucocorticoids on plasma and BAL ACh levels in severe/critical patients. 725 To the best of our knowledge, this study is the first to demonstrate that the lung 726 inflammatory process and poor outcomes of patients with Covid -19 infection are 727 associated with high level s of lipid mediators and ACh, produced via a partially 728 glucocorticoid-insensitive pathway. Glucocorticoid therapy was found to lower only the 729 levels of ACh. Thus, to improve the benefits of glucocorticoid therapy, we suggest that 730 treatment in hospitalised patients be started early and be preferentially administered to 731 patients with low or undetectable viral loads and harmful lung inflammation in 732 combination with AA metabolism inhibitors. 733

734 We demonstrate for the first time that the lung inflammato ry process and worse 735 outcomes in Covid-19 are associated with lipid mediators and ACh, which are produced 736 through a partially glucocorticoid -insensitive pathway . To improve the benefits of 737 glucocorticoid therapy, we suggest that it should be started early in severe/critical 738 hospitalized patients and in combination with AA metabolism inhibitors. 739 740 741 742 All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for thisthis version posted January 15, 2021. ; https://doi.org/10.1101/2021.01.07.20248970doi: medRxiv preprint Page 32 of 84

743 1. Huang C, Wang Y, Li X, et al. Clinical features of patients infected with 2019 744 novel coronavirus in Wuhan, China. Lancet 2020;395(10223):497–506. 745 2. Esser-von Bieren J. Immune-regulation and -functions of eicosanoid lipid 746 mediators. Biol. Chem. 2017; 747 3. Reis M, Rodrigues F, Lautherbach N, et al. Interleukin-1 receptor-induced PGE2 748 production controls acetylcholine-mediated cardiac dysfunction and mortality 749 during scorpion envenomation. Nat Commun 2020;11(1). 750 4. McGovern AE, Mazzone SB. Neural regulation of inflammation in the airways 751 and lungs. Auton Neurosci Basic Clin 2014; 752 5. Wessler I, Kirkpatrick CJ. Cholinergic signaling controls immune functions and 753 promotes homeostasis. Int. Immunopharmacol. 2020; 754 6. Proskocil BJ, Sekhon HS, Jia Y, et al. Acetylcholine is an autocrine or paracrine 755 hormone synthesized and secreted by airway bronchial epithelial cells. 756 Endocrinology 2004; 757 7. Chang EH, Chavan SS, Pavlov VA. Cholinergic control of inflammation, 758 metabolic dysfunction, and cognitive impairment in obesity-associated disorders: 759 Mechanisms and novel therapeutic opportunities. Front. Neurosci. 2019; 760 8. Roy A, Guatimosim S, Prado VF, Gros R, Prado MAM. Cholinergic activity as a 761 new target in diseases of the heart. Mol Med 2014; 762 9. Kistemaker LEM, Gosens R. Acetylcholine beyond bronchoconstriction: roles in 763 inflammation and remodeling. Trends Pharmacol Sci 2015;36(3):164–71. 764 10. Profita M, Bonanno A, Siena L, et al. Acetylcholine mediates the release of IL-8 765 in human bronchial epithelial cells by a NFkB/ERK-dependent mechanism. Eur J 766 Pharmacol 2008; 767 All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for thisthis version posted January 15, 2021. ; https://doi.org/10.1101/2021.01.07.20248970doi: medRxiv preprint Page 33 of 84 11. Shields MD, Riedler J. Bronchoalveolar lavage and tracheal aspirate for 768 assessing airway inflammation in children. Am J Respir Crit Care Med 769 2000;162(2 II). 770 12. Schindelin J, Rueden CT, Hiner MC, Eliceiri KW. The ImageJ ecosystem: An 771 open platform for biomedical image analysis. Mol Reprod Dev 2015;82(7–772 8):518–29. 773 13. Sorgi CA, Peti APF, Petta T, et al. Data descriptor: Comprehensive high-774 resolution multiple-reaction monitoring mass spectrometry for targeted 775 eicosanoid assays. Sci Data 2018; 776 14. Chambers MC, MacLean B, Burke R, et al. A cross-platform toolkit for mass 777 spectrometry and proteomics. Nat. Biotechnol. 2012; 778 15. Yu T, Park Y, Johnson JM, Jones DP. apLCMS-adaptive processing of high-779 resolution LC/MS data. Bioinformatics 2009; 780 16. Li S, Park Y, Duraisingham S, et al. Predicting Network Activity from High 781 Throughput Metabolomics. PLoS Comput Biol 2013; 782 17. Kuri-Cervantes L, Pampena MB, Meng W, et al. Comprehensive mapping of 783 immune perturbations associated with severe COVID-19. Sci Immunol 2020; 784 18. Desai N, Neyaz A, Szabolcs A, et al. Temporal and spatial heterogeneity of host 785 response to SARS-CoV-2 pulmonary infection. Nat Commun [Internet] 786 2020;11(1):6319. Available from: http://www.nature.com/articles/s41467-020-787 20139-7 788 19. Edgar R, Domrachev M, Lash AE. Gene Expression Omnibus: NCBI gene 789 expression and hybridization array data repository. Nucleic Acids Res 2002; 790 20. Russo PST, Ferreira GR, Cardozo LE, et al. CEMiTool: A Bioconductor package 791 for performing comprehensive modular co-expression analyses. BMC 792 All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for thisthis version posted January 15, 2021. ; https://doi.org/10.1101/2021.01.07.20248970doi: medRxiv preprint Page 34 of 84 Bioinformatics 2018; 793 21. Jassal B, Matthews L, Viteri G, et al. The reactome pathway knowledgebase. 794 Nucleic Acids Res 2020; 795 22. Love MI, Huber W, Anders S. Moderated estimation of fold change and 796 dispersion for RNA-seq data with DESeq2. Genome Biol 2014; 797 23. Benjamini Y, Hochberg Y. Controlling the False Discovery Rate: A Practical and 798 Powerful Approach to Multiple Testing. J R Stat Soc Ser B 1995; 799 24. Stark C, Breitkreutz BJ, Reguly T, Boucher L, Breitkreutz A, Tyers M. 800 BioGRID: a general repository for interaction datasets. Nucleic Acids Res 2006; 801 25. Csardi G, Nepusz T. The igraph software package for complex network research. 802 InterJournal Complex Syst 2006; 803 26. Maintainer MB, Bojanowski M. Package “intergraph” Type Package Title 804 Coercion Routines for Network Data Objects. 2016; 805 27. Tyner S, Briatte F, Hofmann H. Network visualization with ggplot2. R J 806 2017;9(1):27–59. 807 28. Taylor R. Interpretation of the Correlation Coefficient: A Basic Review. J 808 Diagnostic Med Sonogr 1990; 809 29. Abreu-Filho PG, Tarragô AM, Costa AG, et al. Plasma Eicosanoid Profile in 810 Plasmodium vivax Malaria: Clinical Analysis and Impacts of Self-Medication. 811 Front Immunol 2019; 812 30. Fukunaga Y, Mine Y, Yoshikawa S, Takeuchi T, Hata F, Yagasaki O. Role of 813 prostacyclin in acetylcholine release from myenteric plexus of guinea-pig ileum. 814 Eur J Pharmacol 1993;233(2–3):237–42. 815 31. Clark SR, Guy CJ, Scurr MJ, et al. Esterified eicosanoids are acutely generated 816 by 5-lipoxygenase in primary human neutrophils and in human and murine 817 All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for thisthis version posted January 15, 2021. ; https://doi.org/10.1101/2021.01.07.20248970doi: medRxiv preprint Page 35 of 84 infection. Blood 2011; 818 32. Chen G, Wu D, Guo W, et al. Clinical and immunological features of severe and 819 moderate coronavirus disease 2019. J Clin Invest 2020;130(5):2620–9. 820 33. Knight DS, Kotecha T, Razvi Y, et al. COVID-19: Myocardial injury in 821 survivors. Circulation. 2020; 822 34. Del Valle DM, Kim-Schulze S, Huang HH, et al. An inflammatory cytokine 823 signature predicts COVID-19 severity and survival. Nat Med 2020; 824 35. Mehta P, McAuley DF, Brown M, Sanchez E, Tattersall RS, Manson JJ. COVID-825 19: consider cytokine storm syndromes and immunosuppression. Lancet. 2020; 826 36. Polidoro RB, Hagan RS, de Santis Santiago R, Schmidt NW. Overview: 827 Systemic Inflammatory Response Derived From Lung Injury Caused by SARS-828 CoV-2 Infection Explains Severe Outcomes in COVID-19. Front. Immunol. 829 2020; 830 37. Shen B, Yi X, Sun Y, et al. Proteomic and Metabolomic Characterization of 831 COVID-19 Patient Sera. Cell 2020; 832 38. Delafiori J, Claudio Navarro L, Focaccia Siciliano R, et al. Covid-19 automated 833 diagnosis and risk assessment through Metabolomics and Machine-Learning. 834 2020. 835 39. Bittleman DB, Casale TB. 5-Hydroxyeicosatetraenoic acid (HETE)-induced 836 neutrophil transcellular migration is dependent upon enantiomeric structure. Am 837 J Respir Cell Mol Biol 1995;12(3):260–7. 838 40. Harder DR. Pressure-induced myogenic activation of cat cerebral arteries is 839 dependent on intact endothelium. Circ Res 1987; 840 41. Barnes BJ, Adrover JM, Baxter-Stoltzfus A, et al. Targeting potential drivers of 841 COVID-19: Neutrophil extracellular traps. J. Exp. Med. 2020; 842 All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for thisthis version posted January 15, 2021. ; https://doi.org/10.1101/2021.01.07.20248970doi: medRxiv preprint Page 36 of 84 42. Kita Y, Takahashi T, Uozumi N, Nallan L, Gelb MH, Shimizu T. Pathway-843 oriented profiling of lipid mediators in macrophages. Biochem Biophys Res 844 Commun 2005; 845 43. Gauthier KM, Goldman DH, Aggarwal NT, Chawengsub Y, Falck JR, Campbell 846 WB. Role of arachidonic acid lipoxygenase metabolites in acetylcholine-induced 847 relaxations of mouse arteries. Am J Physiol - Hear Circ Physiol 2011; 848 44. Rauzi F, Kirkby NS, Edin ML, et al. Aspirin inhibits the production of 849 proangiogenic 15(S)-HETE by platelet cyclooxygenase-1. FASEB J 2016; 850 45. Berna N, Arnould T, Remacle J, Michiels C. Hypoxia-induced increase in 851 intracellular calcium concentration in endothelial cells: Role of the Na+-glucose 852 cotransporter. J Cell Biochem 2002; 853 46. López S, Vila L, Breviario F, de Castellarnau C. Interleukin-1 increases 15-854 hydroxyeicosatetraenoic acid formation in cultured human endothelial cells. 855 Biochim Biophys Acta (BBA)/Lipids Lipid Metab 1993; 856 47. Metz SA, Murphy RC, Fujimoto W, . Effects on glucose-induced insulin 857 secretion of lipoxygenase-derived metabolites of arachidonic acid. Diabetes 858 1984;33(2):119–24. 859 48. Hosoi T, Koguchi Y, Sugikawa E, et al. Identification of a novel human 860 eicosanoid receptor coupled to Gi/o. J Biol Chem 2002; 861 49. Powell WS, Rokach J, ., . The eosinophil chemoattractant 5-oxo-ETE and the 862 OXE receptor. Prog. Lipid Res. 2013; 863 50. Powell WS, Rokach J. Targeting the OXE receptor as a potential novel therapy 864 for asthma. Biochem Pharmacol 2020;179:113930. 865 51. Schwarz B, Sharma L, Roberts L, et al. Cutting Edge: Severe SARS-CoV-2 866 Infection in Humans Is Defined by a Shift in the Serum Lipidome, Resulting in 867 All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for thisthis version posted January 15, 2021. ; https://doi.org/10.1101/2021.01.07.20248970doi: medRxiv preprint Page 37 of 84 Dysregulation of Eicosanoid Immune Mediators. J Immunol [Internet] 868 2020;ji2001025. Available from: 869 http://www.jimmunol.org/content/early/2020/12/04/jimmunol.2001025.abstract 870 52. Hanna VS, Hafez EAA. Synopsis of arachidonic acid metabolism: A review. J. 871 Adv. Res. 2018; 872 53. Schwarz B, Sharma L, Roberts L, et al. Severe SARS-CoV-2 infection in humans 873 is defined by a shift in the serum lipidome resulting in dysregulation of 874 eicosanoid immune mediators. Res Sq 2020; 875 54. Nienhold R, Ciani Y, Koelzer VH, et al. Two distinct immunopathological 876 profiles in autopsy lungs of COVID-19. Nat Commun [Internet] 877 2020;11(1):5086. Available from: https://doi.org/10.1038/s41467-020-18854-2 878 55. Pairo-Castineira E, Clohisey S, Klaric L, et al. Genetic mechanisms of critical 879 illness in Covid-19. Nature [Internet] 2020;Available from: 880 https://doi.org/10.1038/s41586-020-03065-y 881 56. Taha AY. Linoleic acid–good or bad for the brain? npj Sci. Food. 2020; 882 57. Kohn A, Gitelman J, Inbar M. Unsaturated free fatty acids inactivate animal 883 enveloped viruses. Arch Virol 1980; 884 58. Das UN. Arachidonic acid and other unsaturated fatty acids and some of their 885 metabolites function as endogenous antimicrobial molecules: A review. J. Adv. 886 Res. 2018; 887 59. Chandrasekharan JA, Sharma-Walia N. Arachidonic acid derived lipid mediators 888 influence Kaposi’s sarcoma-associated herpesvirus infection and pathogenesis. 889 Front. Microbiol. 2019; 890 60. Das UN. Bioactive Lipids in COVID-19-Further Evidence. Arch Med Res 2020; 891 61. Nakano T, Ohara O, Teraoka H, Arita H. Glucocorticoids suppress group II 892 All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for thisthis version posted January 15, 2021. ; https://doi.org/10.1101/2021.01.07.20248970doi: medRxiv preprint Page 38 of 84 phospholipase A2 production by blocking mRNA synthesis and post-893 transcriptional expression. J Biol Chem 1990; 894 62. Kobza Black A, Greaves M, Hensby C. The effect of systemic prednisolone on 895 arachidonic acid, and prostaglandin E2 and F2 alpha levels in human cutaneous 896 inflammation. Br J Clin Pharmacol 1982; 897 63. Hu X, Cifarelli V, Sun S, Kuda O, Abumrad NA, Su X. Major role of adipocyte 898 prostaglandin E2 in lipolysis-induced macrophage recruitment. J Lipid Res 2016; 899 64. Alzoghaibi MA, Walsh SW, Willey A, Yager DR, Fowler AA, Graham MF. 900 Linoleic acid induces interleukin-8 production by Crohn’s human intestinal 901 smooth muscle cells via arachidonic acid metabolites. Am J Physiol - 902 Gastrointest Liver Physiol 2004; 903 65. Ahn K, McKinney MK, Cravatt BF. Enzymatic pathways that regulate 904 endocannabinoid signaling in the nervous system. Chem Rev 2008;108(5):1687–905 707. 906 66. Grant RA, Morales-Nebreda L, Markov NS, et al. Alveolitis in severe SARS-907 CoV-2 pneumonia is driven by self-sustaining circuits between infected alveolar 908 macrophages and T cells. bioRxiv 2020; 909 67. Malcher-Lopes R, Franco A, Tasker JG. Glucocorticoids shift arachidonic acid 910 metabolism toward endocannabinoid synthesis: A non-genomic anti-911 inflammatory switch. Eur. J. Pharmacol. 2008; 912 68. The RECOVERY Collaborative G. Dexamethasone in Hospitalized Patients with 913 Covid-19 — Preliminary Report. N Engl J Med 2020; 914 69. Tomazini BM, Maia IS, Cavalcanti AB, et al. Effect of Dexamethasone on Days 915 Alive and Ventilator-Free in Patients with Moderate or Severe Acute Respiratory 916 Distress Syndrome and COVID-19: The CoDEX Randomized Clinical Trial. 917 All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for thisthis version posted January 15, 2021. ; https://doi.org/10.1101/2021.01.07.20248970doi: medRxiv preprint Page 39 of 84 JAMA - J Am Med Assoc 2020; 918 70. Tang Y, Liu J, Zhang D, Xu Z, Ji J, Wen C. Cytokine Storm in COVID-19: The 919 Current Evidence and Treatment Strategies. Front. Immunol. 2020; 920 71. Ye Q, Wang B, Mao J, . The pathogenesis and treatment of the ‘Cytokine 921 Storm’’ in COVID-19.’ J. Infect. 2020; 922 72. Prescott HC, Rice TW. Corticosteroids in COVID-19 ARDS: Evidence and Hope 923 during the Pandemic. JAMA - J. Am. Med. Assoc. 2020; 924 73. Noushmehr H, D’Amico E, Farilla L, et al. Fatty acid translocase (FAT/CD36) is 925 localized on insulin-containing granules in human pancreatic β-cells and 926 mediates fatty acid effects on insulin secretion. Diabetes 2005; 927 74. Glatz JFC, Luiken J. From fat to FAT (CD36/SR-B2): Understanding the 928 regulation of cellular fatty acid uptake. Biochimie. 2017; 929 75. Febbraio M, Hajjar DP, Silverstein RL. CD36: A class B scavenger receptor 930 involved in angiogenesis, atherosclerosis, inflammation, and lipid metabolism. J. 931 Clin. Invest. 2001; 932 76. Glatz JFC, Luiken JJFP, Bonen A. Membrane fatty acid transporters as regulators 933 of lipid metabolism: Implications for metabolic disease. Physiol. Rev. 2010; 934 77. Pepino MY, Kuda O, Samovski D, Abumrad NA. Structure-function of CD36 935 and importance of fatty acid signal transduction in fat metabolism. Annu. Rev. 936 Nutr. 2014; 937 78. Zoccal KF, Gardinassi LG, Sorgi CA, et al. CD36 shunts eicosanoid metabolism 938 to repress CD14 Licensed interleukin-1β release and inflammation. Front 939 Immunol 2018; 940 79. Khan NA, Besnard P. Oro-sensory perception of dietary lipids: New insights into 941 the fat taste transduction. Biochim. Biophys. Acta - Mol. Cell Biol. Lipids. 2009; 942 All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for thisthis version posted January 15, 2021. ; https://doi.org/10.1101/2021.01.07.20248970doi: medRxiv preprint Page 40 of 84 80. Deleon-Pennell KY, Tian Y, Zhang B, et al. CD36 Is a Matrix Metalloproteinase-943 9 Substrate That Stimulates Neutrophil Apoptosis and Removal during Cardiac 944 Remodeling. Circ Cardiovasc Genet 2016; 945 81. Novak ML, Thorp EB. Shedding light on impaired efferocytosis and 946 nonresolving inflammation. Circ. Res. 2013; 947 82. Cundall M, Sun Y, Miranda C, Trudeau JB, Barnes S, Wenzel SE. Neutrophil-948 derived matrix metalloproteinase-9 is increased in severe asthma and poorly 949 inhibited by glucocorticoids. J Allergy Clin Immunol 2003; 950 83. Hozumi A, Nishimura Y, Nishiuma T, Kotani Y, Yokoyama M. Induction of 951 MMP-9 in normal human bronchial epithelial cells by TNF-α via NF-κB-952 mediated pathway. Am J Physiol - Lung Cell Mol Physiol 2001; 953 84. Ueland T, Holter JC, Holten AR, et al. Distinct and early increase in circulating 954 MMP-9 in COVID-19 patients with respiratory failure: MMP-9 and respiratory 955 failure in COVID-19. J. Infect. 2020; 956 85. Palau V, Riera M, Soler MJ. ADAM17 inhibition may exert a protective effect 957 on COVID-19. Nephrol Dial Transplant 2020; 958 86. Laing AG, Lorenc A, del Molino del Barrio I, et al. A dynamic COVID-19 959 immune signature includes associations with poor prognosis. Nat Med 2020; 960 87. Zhao Y, Qin L, Zhang P, et al. Longitudinal COVID-19 profiling associates IL-961 1RA and IL-10 with disease severity and RANTES with mild disease. JCI Insight 962 2020; 963 88. Belge K-U, Dayyani F, Horelt A, et al. The Proinflammatory CD14 + CD16 + 964 DR ++ Monocytes Are a Major Source of TNF . J Immunol 2002; 965 89. Kunkel SL, Standiford T, Kasahara K, Strieter RM. Interleukin-8 (IL-8): The 966 major neutrophil chemotactic factor in the lung. Exp Lung Res 1991; 967 All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for thisthis version posted January 15, 2021. ; https://doi.org/10.1101/2021.01.07.20248970doi: medRxiv preprint Page 41 of 84 90. Kunkel TA, Bebenek K, McClary J. Efficient site-directed mutagenesis using 968 uracil-containing DNA. Methods Enzymol 1991; 969 91. Zoccal KF, Sorgi CA, Hori JI, et al. Opposing roles of LTB4 and PGE2 in 970 regulating the inflammasome-dependent scorpion venom-induced mortality. Nat 971 Commun 2016; 972 92. De Virgiliis F, Di Giovanni S. Lung innervation in the eye of a cytokine storm: 973 neuroimmune interactions and COVID-19. Nat Rev Neurol 2020; 974 93. Carmona-Rivera C, Purmalek MM, Moore E, et al. A role for muscarinic 975 receptors in neutrophil extracellular trap formation and levamisole-induced 976 autoimmunity. JCI Insight 2017; 977 94. do Espírito Santo DA, Lemos ACB, Miranda CH, . In vivo demonstration of 978 microvascular thrombosis in severe COVID-19. J Thromb Thrombolysis 2020; 979 95. Supuran CT, Di Fiore A, De Simone G. Carbonic anhydrase inhibitors as 980 emerging drugs for the treatment of obesity. Expert Opin. Emerg. Drugs. 2008; 981 96. Cavallotti C, Bruzzone P, Mancone M, Leali FMT. Distribution of 982 acetylcholinesterase and cholineacetyl-transferase activities in coronary vessels 983 of younger and older adults. Geriatr Gerontol Int 2004; 984 97. Vijayaraghavan S, Huang B, Blumenthal EM, Berg DK. Arachidonic acid as a 985 possible negative feedback inhibitor of nicotinic acetylcholine receptors on 986 neurons. J Neurosci 1995; 987 98. Mabley J, Gordon S, Pacher P. Nicotine exerts an anti-inflammatory effect in a 988 murine model of acute lung injury. Inflammation 2011; 989 99. Blanco-Melo D, Nilsson-Payant BE, Liu WC, et al. Imbalanced Host Response to 990 SARS-CoV-2 Drives Development of COVID-19. Cell 2020; 991 100. Horiguchi K, Horiguchi S, Yamashita N, et al. Expression of SLURP-1, an 992 All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for thisthis version posted January 15, 2021. ; https://doi.org/10.1101/2021.01.07.20248970doi: medRxiv preprint Page 42 of 84 endogenous α7 nicotinic acetylcholine receptor allosteric ligand, in murine 993 bronchial epithelial cells. J Neurosci Res 2009; 994 101. Schirmer SU, Eckhardt I, Lau H, et al. The cholinergic system in rat testis is of 995 non-neuronal origin. Reproduction 2011; 996 102. Suenaga A, Fujii T, Ogawa H, et al. Up-regulation of lymphocytic cholinergic 997 activity by ONO-4819, a selective prostaglandin EP4 receptor agonist, in MOLT-998 3 human leukemic T cells. Vascul Pharmacol 2004; 999 103. Reinheimer T, Münch M, Bittinger F, Racké K, Kirkpatrick CJ, Wessler I. 1000 Glucocorticoids mediate reduction of epithelial acetylcholine content in the 1001 airways of rats and humans. Eur J Pharmacol 1998; 1002 104. Gross I, Ballard PL, Ballard RA, Jones CT, Wilson CM. Corticosteroid 1003 stimulation of phosphatidylcholine synthesis in cultured fetal rabbit lung: 1004 Evidence for de novo protein synthesis mediated by glucocorticoid receptors. 1005 Endocrinology 1983; 1006 105. Nakamura T, Fujiwara R, Ishiguro N, et al. Involvement of choline transporter-1007 like proteins, CTL1 and CTL2, in glucocorticoid-induced acceleration of 1008 phosphatidylcholine synthesis via increased choline uptake. Biol Pharm Bull 1009 2010; 1010 106. Wilk JB, Shrine NRG, Loehr LR, et al. Genome-wide association studies identify 1011 CHRNA5/3 and HTR4 in the development of airflow obstruction. Am J Respir 1012 Crit Care Med 2012; 1013 107. Lam DCL, Luo SY, Fu KH, et al. Nicotinic acetylcholine receptor expression in 1014 human airway correlates with lung function. Am J Physiol - Lung Cell Mol 1015 Physiol 2016; 1016 108. Oenema TA, Kolahian S, Nanninga JE, et al. Pro-inflammatory mechanisms of 1017 All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for thisthis version posted January 15, 2021. ; https://doi.org/10.1101/2021.01.07.20248970doi: medRxiv preprint Page 43 of 84 muscarinic receptor stimulation in airway smooth muscle. Respir Res 2010; 1018 109. Yamada M, Ichinose M. The cholinergic pathways in inflammation: A potential 1019 pharmacotherapeutic target for COPD. Front. Pharmacol. 2018; 1020 110. Fadason T, Schierding W, Lumley T, O’Sullivan JM. Chromatin interactions and 1021 expression quantitative trait loci reveal genetic drivers of multimorbidities. Nat 1022 Commun 2018;9(1):5198. 1023 111. Lagoumintzis G, Chasapis CT, Alexandris N, et al. COVID-19 and Cholinergic 1024 Anti-inflammatory Pathway: In silico Identification of an 1025 Interaction between α7 Nicotinic Acetylcholine Receptor and the Cryptic 1026 Epitopes of SARS-CoV and SARS-CoV-2 Spike Glycoproteins. bioRxiv 1027 [Internet] 2020;2020.08.20.259747. Available from: 1028 http://biorxiv.org/content/early/2020/08/21/2020.08.20.259747.abstract 1029 112. Gahring LC, Myers EJ, Dunn DM, Weiss RB, Rogers SW. Nicotinic alpha 7 1030 receptor expression and modulation of the lung epithelial response to 1031 lipopolysaccharide. PLoS One 2017; 1032 113. Wang H, Yu M, Ochani M, et al. Nicotinic acetylcholine receptor α7 subunit is 1033 an essential regulator of inflammation. Nature 2003;421(6921):384–8. 1034 114. Pinheiro NM, Santana FPR, Almeida RR, et al. Acute lung injury is reduced by 1035 the a7nAChR agonist PNU-282987 through changes in the macrophage profile. 1036 FASEB J 2017; 1037 115. Hamano R, Takahashi HK, Iwagaki H, Yoshino T, Nishibori M, Tanaka N. 1038 Stimulation of α7 nicotinic acetylcholine receptor inhibits CD14 and the toll-like 1039 receptor 4 expression in human monocytes. Shock 2006; 1040 116. Hasday JD, Dubin W, Mongovin S, et al. Bronchoalveolar macrophage CD14 1041 expression: Shift between membrane- associated and soluble pools. Am J Physiol 1042 All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for thisthis version posted January 15, 2021. ; https://doi.org/10.1101/2021.01.07.20248970doi: medRxiv preprint Page 44 of 84 - Lung Cell Mol Physiol 1997; 1043 117. Gori S, Alcain J, Vanzulli S, et al. Acetylcholine-treated murine dendritic cells 1044 promote inflammatory lung injury. PLoS One 2019; 1045 118. Edwards JM, McCarthy CG, Wenceslau CF. The Obligatory Role of the 1046 Acetylcholine-Induced Endothelium-Dependent Contraction in Hypertension: 1047 Can Arachidonic Acid Resolve this Inflammation? Curr Pharm Des 2020; 1048 119. Radu BM, Osculati AMM, Suku E, et al. All muscarinic acetylcholine receptors 1049 (M1-M5) are expressed in murine brain microvascular endothelium. Sci Rep 1050 2017; 1051 120. Kumar PT, Antony S, Nandhu MS, Sadanandan J, Naijil G, Paulose CS. Vitamin 1052 D3 restores altered cholinergic and insulin receptor expression in the cerebral 1053 cortex and muscarinic M3 receptor expression in pancreatic islets of 1054 streptozotocin induced diabetic rats. J Nutr Biochem 2011; 1055 121. Ali N. Role of vitamin D in preventing of COVID-19 infection, progression and 1056 severity. J. Infect. Public Health. 2020; 1057 122. CHANGEUX jean-pierre, Amoura Z, Rey F, Miyara M. A nicotinic hypothesis 1058 for Covid-19 with preventive and therapeutic implications. Qeios 2020; 1059 123. Oakes JM, Fuchs RM, Gardner JD, Lazartigues E, Yue X. Nicotine and the 1060 renin-angiotensin system. Am. J. Physiol. - Regul. Integr. Comp. Physiol. 2018; 1061 124. Nct. Pyridostigmine in Severe SARS-CoV-2 Infection. 1062 https://clinicaltrials.gov/show/NCT04343963 2020; 1063 125. Ahmed H O. COVID-19: Targeting the cytokine storm via cholinergic anti-1064 inflammatory (Pyridostigmine). Int J Clin Virol 2020; 1065 126. Gonzalez-Rubio J, Navarro-Lopez C, Lopez-Najera E, et al. Cytokine Release 1066 Syndrome (CRS) and Nicotine in COVID-19 Patients: Trying to Calm the Storm. 1067 All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for thisthis version posted January 15, 2021. ; https://doi.org/10.1101/2021.01.07.20248970doi: medRxiv preprint Page 45 of 84 Front Immunol 2020; 1068 127. ahir AMOURA FT. Efficacy of Nicotine in Preventing COVID-19 Infection in 1069 Caregivers ( NICOVID-PREV ). Natl Libr Med 2020;2:1–8. 1070 128. Smith JC, Sausville EL, Girish V, et al. Cigarette Smoke Exposure and 1071 Inflammatory Signaling Increase the Expression of the SARS-CoV-2 Receptor 1072 ACE2 in the Respiratory Tract. Dev Cell 2020; 1073 129. Leung JM, Leung JM, Yang CX, Sin DD, Sin DD. COVID-19 and nicotine as a 1074 mediator of ACE-2. Eur. Respir. J. 2020; 1075 130. Zureik M, Baricault B, Vabre C, et al. Nicotine-replacement therapy as a 1076 surrogate of smoking and the risk of hospitalization with Covid-19 and all-cause 1077 mortality a nationwide observational cohort study in France. medrxiv 2020; 1078 131. Margină D, Ungurianu A, Purdel C, et al. Chronic inflammation in the context of 1079 everyday life: Dietary changes as mitigating factors. Int. J. Environ. Res. Public 1080 Health. 2020; 1081 132. Hadjadj J, Yatim N, Barnabei L, et al. Impaired type I interferon activity and 1082 inflammatory responses in severe COVID-19 patients. Science (80- ) 2020; 1083 133. Ye G, Pan Z, Pan Y, et al. Clinical characteristics of severe acute respiratory 1084 syndrome coronavirus 2 reactivation. J Infect 2020; 1085 134. Xu XW, Wu XX, Jiang XG, et al. Clinical findings in a group of patients infected 1086 with the 2019 novel coronavirus (SARS-Cov-2) outside of Wuhan, China: 1087 Retrospective case series. BMJ 2020; 1088 135. Grasselli G, Zangrillo A, Zanella A, et al. Baseline Characteristics and Outcomes 1089 of 1591 Patients Infected with SARS-CoV-2 Admitted to ICUs of the Lombardy 1090 Region, Italy. JAMA - J Am Med Assoc 2020; 1091 136. Marshall JC, Murthy S, Diaz J, et al. A minimal common outcome measure set 1092 All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for thisthis version posted January 15, 2021. ; https://doi.org/10.1101/2021.01.07.20248970doi: medRxiv preprint Page 46 of 84 for COVID-19 clinical research. Lancet Infect. Dis. 2020; 1093 137. Office WHOEMR. Updated clinical management guideline for COVID-19. 1094 Wkly. Epidemiol. Monit. 2020; 1095 138. Rauch A, Labreuche J, Lassalle F, et al. Coagulation biomarkers are independent 1096 predictors of increased oxygen requirements in COVID-19. J Thromb Haemost 1097 2020; 1098 139. Sahu BR, Kampa RK, Padhi A, Panda AK. C-reactive protein: A promising 1099 biomarker for poor prognosis in COVID-19 infection. Clin Chim Acta 2020; 1100 140. Richardson S, Hirsch JS, Narasimhan M, et al. Presenting Characteristics, 1101 Comorbidities, and Outcomes Among 5700 Patients Hospitalized With COVID-1102 19 in the New York City Area. JAMA [Internet] 2020;323(20):2052–9. Available 1103 from: https://doi.org/10.1001/jama.2020.6775 1104 141. Thomas T, Stefanoni D, Reisz J, et al. COVID-19 infection results in alterations 1105 of the kynurenine pathway and fatty acid metabolism that correlate with IL-6 1106 levels and renal status. medRxiv Prepr Serv Heal Sci 2020; 1107 142. Moolamalla STR, Chauhan R, Priyakumar D, Vinod PK. Host metabolic 1108 reprogramming in response to SARS-Cov-2 infection. bioRxiv 2020; 1109 143. Soliman S, Faris MAIE, Ratemi Z, Halwani R. Switching Host Metabolism as an 1110 Approach to Dampen SARS-CoV-2 Infection. Ann. Nutr. Metab. 2020; 1111 144. Frankenberger M, Sternsdorf T, Pechumer H, Pforte A, Ziegler-Heitbrock HWL. 1112 Differential cytokine expression in human blood monocyte subpopulations: A 1113 polymerase chain reaction analysis. Blood 1996; 1114 145. Giamarellos-Bourboulis EJ, Netea MG, Rovina N, et al. Complex Immune 1115 Dysregulation in COVID-19 Patients with Severe Respiratory Failure. Cell Host 1116 Microbe 2020; 1117 All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for thisthis version posted January 15, 2021. ; https://doi.org/10.1101/2021.01.07.20248970doi: medRxiv preprint Page 47 of 84 146. Lamontagne F, Brower R, Meade M. Corticosteroid therapy in acute respiratory 1118 distress syndrome. CMAJ. 2013; 1119 147. Russell CD, Millar JE, Baillie JK. Clinical evidence does not support 1120 corticosteroid treatment for 2019-nCoV lung injury. Lancet. 2020; 1121 1122

1123 The authors thankfully acknowledge Innovation and Technology Park (Supera), the 1124 healthy-participants joining as controls and the positive COVID-19 patients as well as 1125 their families. We grieve for all patients who lost their lives as a result of this pandemic, 1126 including those who provided us with samples to be able to answer scientific questions 1127 and contribute to humanity's eradication of this disease. We also thank Fabiana Rossetto 1128 de Moraes, B.Sc., for the cytometry analysis, Caroline Fontanari, M.Sc., for laboratory 1129 and technical support, the ICU team, and all hospital professionals, especially the 1130 technicians, nurses, physiotherapists, and biomedical personnel, who collaborated on this 1131 work through Hospital Santa Casa de Misericórdia in Ribeirão Preto and Hospital São 1132 Paulo in Ribeirão Preto. We are grateful for the indispensable contribution of the Ribeirão 1133 Preto Municipal Health Department and the employees of the Serviço de Análises 1134 Clínicas (SAC) of the Faculdade de Ciências Farmacêuticas de Ribeirão Preto, USP. We 1135 also thank Professors Victor Hugo Aquino Quintana, Ph.D., Márcia Regina von Zeska 1136 Kress, Ph.D., and Marcia Eliana da Silva Ferreira, Ph.D. for sharing the BSL2 viral 1137 laboratory. 1138 1139 1140 All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for thisthis version posted January 15, 2021. ; https://doi.org/10.1101/2021.01.07.20248970doi: medRxiv preprint Page 48 of 84 Additional Information 1141 Extended data available in Supplementary Appendixes: 1142 • Supplementary Appendix I - Pathways and list of corresponding genes related to 1143 acetylcholine and arachidonic acid observed in Reactome pathways and Covid-19 1144 biomarkers. 1145 1146 • Supplementary Appendix II - Report of CEMITool results for the gene co -1147 expression modular analysis of lung samples from biopsies of Covid-19 and non-1148 Covid-19 patients. 1149 1150 • Supplementary Appendix III - Differential gene expression analysis results 1151 between Covid -19-death groups (CV, NCV, CVL and CVH) and Covid -19 1152 samples from patients who underwent glucocorticoid treatment (GC) and not 1153 (non-GC). 1154 1155 • Supplementary Appendix IV - Betweenness and degree values o f genes from 1156 biological network constructed based on the principal co-expression module M1. 1157 1158 1159 Funding 1160 Fundação de Amparo a Pesquisa do Estado de São Paulo (FAPESP): 1161 #2020/05207-6, #2014/07125 -6 and #2015/00658 -1 for L.H.F.; #2020/08534-8 for 1162 M.M.P.; #2018/22667-0 for C.O.S.S.; #2020/05270-0 for V.D.B. Additional support was 1163 All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for thisthis version posted January 15, 2021. ; https://doi.org/10.1101/2021.01.07.20248970doi: medRxiv preprint Page 49 of 84 provided by the National Council for Scientific and Technological Development (CNPq), 1164 the Coordination for the Improvement of Higher Educational Personnel (CAPES-Finance 1165 Code 001)), and Pró -Reitora de Pesquisa da Universidade de São Paulo, grant -USP-1166 VIDA. 1167 Conflict of Interest Statement 1168 The authors declare that this research was performed without conflicts of interest 1169 or commercial or financial gains. 1170 Legends of main figures 1171 Figure 1. Metabolomic and lipidomic analysis revealed increased levels of 1172 circulating fatty acids, arachidonic acid (AA), 5 -HETE and 11 -HETE in patients 1173 with Covid-19. 1174 Plasma samples were collected from healthy-participants (n=20) and patients with 1175 asymptomatic-to-mild (n=10), moderate (n=12), severe (n=16), or critical (n=13) disease. 1176 (A) A Manhattan plot for the differential abundance of metabolite features in different 1177 groups of patients with Covid-19 and healthy participants (false discovery rate (FDR) of 1178 595, adjusted P<0.05, above the dashed line) and (B, left panel) metabolic pathway 1179 enrichment of significant metabolite features based on data from untargeted mass 1180 spectrometry. Differential abundance was calculated using the limma package for R, and 1181 FDR was controlled using the Benjamini-Hochberg method. Mummichog software v2.3.3 1182 was used for pathway enrichment analysis. (B, right panel) Schematic illustration of the 1183 metabolic pathways involved in the production of hydroxyeicosate traenoic acids 1184 (HETEs), such as 5 -HETE and 11 -HETE, from arachidonic (AA) and linoleic acid 1185 metabolism, which discriminate the severity of Covid -19. (C-E) Plasma metabolomics 1186 All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for thisthis version posted January 15, 2021. ; https://doi.org/10.1101/2021.01.07.20248970doi: medRxiv preprint Page 50 of 84 indicating an increase in free fatty acids in Covid -19. Plasma obtained from heal thy 1187 participants (n=18) and patients with asymptomatic -to-mild (n=10), moderate (n=12), 1188 severe (n=14), or critical (n=16) disease (F-H) was subjected to lipidomics analysis using 1189 targeted mass spectrometry, confirming the elevation of AA, 5 -HETE and 11 -HETE, 1190 according to the severity of Covid-19. Data are expressed as the mean ± SEM. Differences 1191 in (C -H) were considered significant at P<0.05 according to Kruskal –Wallis analysis 1192 followed by Dunn’s posttest, and specific P-values are shown in each figure. 1193 1194 Figure 2. Systemic markers of inflammation determined Covid-19 severity. 1195 (A-F) Total and differential leukocyte counts in peripheral blood from healthy-1196 participants (n=17) and from patients with asymptomatic -to-mild (n=10), moderate 1197 (n=12), severe (n=16) or critical (n=13) disease showed that Covid -19 modified the 1198 numbers of distinct leukocyte populations. (G -H) Flow cytometry analysis of 1199 CD14+HLA-DR circulating monocytes by t -distributed stochastic neighbour embedding 1200 indicated reduced (G) CD14 and (H) CD36 mean fluorescence intensity in asymptomatic-1201 to-mild (n=16), moderate (n=15), severe (n=35), and critical (n=20) Covid -19 patients 1202 compared to healthy participants (n=12). (I) Soluble CD14 (sCD14), as determined by 1203 ELISA) in healthy participants ( n=12), asymptomatic-to-mild (n=12), moderate (n=13), 1204 severe (n=15), and critical (n=15) patients, showed increases only in samples from critical 1205 Covid-19 patients. (J-L) Classical, intermediate, and nonclassical monocytes determined 1206 by flow cytometry analyses of CD14, CD16 and HLA-DR expression in blood cells from 1207 healthy participants (n=12), asymptomatic -to-mild (n=16), moderate (n=15), severe 1208 (n=35), and critical (n=20) Covid -19 patients revealed a decrease in intermediate 1209 monocytes in Covid-19. (M-Q) Cytokines (IL-8, IL-6, IL-1, TNF and IL-10) quantified 1210 All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for thisthis version posted January 15, 2021. ; https://doi.org/10.1101/2021.01.07.20248970doi: medRxiv preprint Page 51 of 84 in plasma using a Cytometric Bead Array (CBA) of healthy volunteers (n=35), 1211 asymptomatic-to-mild (n=29), moderate (n=35), severe (n=42), and critical (n=21) 1212 patients demonstrated a distinct cytokine prof ile according to disease severity. Data are 1213 expressed as the mean ± SEM, and differences between groups were calculated using 1214 Kruskal–Wallis with Dunn’s multiple comparison post -tests. The specific P -values are 1215 displayed in each figure. Differences were considered significant at P<0.05. 1216 1217 Figure 3. Altered production of lipid mediators and cellular infiltrates in the lungs 1218 drove the local response to SARS-CoV-2 infection. 1219 BAL was collected from intubated patients with severe/critical confirmed Covid-1220 19 diagnosis and from intubated patients without SARS -CoV-2 infection, which are 1221 labelled as Covid -19 and non -Covid-19, respectively. Data from untargeted mass 1222 spectrometry demonstrated (A) differential abundance of metabolite features comparing 1223 Covid-19 and non-Covid-19 participants (Manhattan plot, 595 at FDR adjusted P<0.05, 1224 above the dashed line ) and (B) metabolic pathway enrichment of significant metabolite 1225 features. Differential abundance was calculated using the limma package for R, and the 1226 false discovery rate was controlled using the Benjamini -Hochberg method; Covid -19 1227 (n=26) and non -Covid-19 (n=12). Mummichog software v2.3.3 was used for lipid 1228 pathway enrichment analysis . (C) Lipid mediators derived from arachidonic acid (AA) 1229 metabolism by lipoxygenase (LOX) or cyclooxygenase (COX) in both Covid-19 (n=19) 1230 and non-Covid-19 (n=11) samples, as determined by target mass spectrometry, showed a 1231 significant increase in 5-HETE and 11-HETE in Covid-19. (D) Total leukocytes in BAL 1232 fluid (left panel) and a representative image of Romanowsky staining in these cells (right 1233 panel) used for (E) identification of specific cellular populations in Covid-19 (n=23) and 1234 All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for thisthis version posted January 15, 2021. ; https://doi.org/10.1101/2021.01.07.20248970doi: medRxiv preprint Page 52 of 84 non-Covid-19 (n=11) patients showed a similar profile of infiltrating cells. (F) Cytokine 1235 concentrations in BAL from intubated Covid -19 patients (n=23) and non -Covid-19 1236 participants (n=11) also revealed a similar profile. (G, H) Classical and intermediate 1237 monocytes determined by flow cytometry analyses of CD14, CD16 and HLA -DR 1238 expression in BAL cells from Covid -19 (n=10) and non -Covid-19 (n=11) patients 1239 demonstrate that both populations are decreased in Covid -19. (I) CD14 and (J) CD36 1240 mean fluorescence intensity (MFI) of CD14 +HLA-DR gated BAL monocytes is 1241 decreased in Covid-19 (n=10) compared to non-Covid-19 (n=11) patients. 5-HPETE: 5-1242 hydroperoxyeicosatetraenoic acid, LTA4: leukotriene A4, LTA4 hydrolase: leukotriene A4 1243 hydrolase, LTB 4: leukotriene B 4, 6 -trans LTB 4: 6 -trans le ukotriene B 4, 5 -HETE: 5 -1244 hydroxyeicosatetraenoic acid, 11 -HETE: 11 -hydroxyeicosatetraenoic acid, 12 -HETE: 1245 12-hydroxyeicosatetraenoic acid, 15 -HETE: 15 -hydroxyeicosatetraenoic acid, 5 -oxo-1246 HETE: 5 -oxoeicosatetraenoic acid, 12 -oxo-ETE: 12 -oxoeicosatetraenoic a cid, 15 -oxo-1247 HETE: 15-oxoeicosatetraenoic acid, PGH 2: prostaglandin H 2, PGE 2: prostaglandin E 2, 1248 PGD2: prostaglandin D 2, TXB 2: thromboxane B, TX synthase: thromboxane. Data are 1249 expressed as the mean ± SEM. Differences between groups were calculated using the 1250 Mann–Whitney test, and specific P -values are shown in the figure. Differences were 1251 considered significant at P<0.05. 1252 1253 Figure 4. Severe and critical phases of Covid-19 correlated with increased systemic 1254 and lung acetylcholine and lipid mediators, but only ACh was diminished by 1255 glucocorticoids 1256 Comparison of systemic and local lung responses was analysed using blood and 1257 BAL samples from patients with severe/critical Covid -19 and showed that among (A) 1258 All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for thisthis version posted January 15, 2021. ; https://doi.org/10.1101/2021.01.07.20248970doi: medRxiv preprint Page 53 of 84 cytokines (plasma n=9; BAL n=9), (B) AA (plasma n=29; BAL n=19), (C) 5 -HETE 1259 (plasma n=29; BAL n=19), and (D) 11-HETE (plasma n=29; BAL n=19), levels of only 1260 AA were higher in blood. (E) The ACh concentration (pmol.mL-1) in heparinized plasma 1261 from healthy-participants (n=6) compared to plasma from SARS-CoV-2-infected patients 1262 not treated with glucocorticoids (non -GC) classified as having asymptomatic -to-mild 1263 (n=5), moderate (n=9), severe (n=7), and critical (n=7) disease showed an increase based 1264 on disease severity. (F) ACh in the plasma of patients with Covid -19 at severe/critical 1265 stages of the disease and treated (GC, n=18) or not (non-GC, n=14) with glucocorticoids 1266 shows decreased release in response to the treatment. (G) Comparison of ACh in BAL 1267 from SARS-CoV-2-infected severe/critical patients treated (CG, n=17 ) or not (non -GC, 1268 n=3) with glucocorticoids confirmed inhibition of the neurotransmitter by the treatment. 1269 (H) Comparison of neutrophil numbers from blood (n=14) and BAL (n=14) of patients 1270 with severe/critical disease demonstrated increased neutrophil infi ltration in the lungs. 1271 Data are expressed as the mean ± SEM. Differences were considered significant at P<0.05 1272 according to (A) and (E) Kruskal –Wallis tests followed by Dunn’s posttest. (B, C, D, F 1273 and G) Student’s t-tests, Mann–Whitney, (H) Wilcoxon matched -pairs signed rank test. 1274 Concentrations of ACh in the blood of patients with severe and critical disease and not 1275 treated with glucocorticoids were the same, as shown in panels (E) and (F). Blood and 1276 BAL were collected during hospitalization, on average 6 to 17 days after admission, and 1277 glucocorticoids (methylprednisolone; range 40 to 500 mg/kg/day, or dexamethasone; 1278 range 1.5 to 6.0 mg/kg/day) were administered intravenously. Interaction networks 1279 between various pai rs of mediators quantified in (I) blood (n=151) or (J) BAL (n=32) 1280 from our Covid-19 cohort, as constructed using the open access software Cytoscape v3.3 1281 (Cytoscape Consortium, San Diego, CA) and Spearman tests and showing significant 1282 correlations depicted by different lines ( r and P values described in section 7.1). (K -P) 1283 All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for thisthis version posted January 15, 2021. ; https://doi.org/10.1101/2021.01.07.20248970doi: medRxiv preprint Page 54 of 84 Venn diagram constructed according to the online tool Draw Venn Diagram 1284 (http://bioinformatics.psb.ugent.be/webtools/Venn/) illustrates high -producing patients 1285 in severe and critical stages of disease, with (GC) or without (non-GC) treatment. (K-N) 1286 Plasma from patients producing high levels of AA, 5-HETE, 11-HETE, ACh and sCD14 1287 in GC (n=49) or non-GC (n=17) groups. (O, P) BAL from patients producing high levels 1288 of AA, 5-HETE, 11-HETE, and ACh in GC (n=23) or non-GC (n=4) groups. For analysis, 1289 the global median between controls and patients was considered as the cut-off point. (Q) 1290 Summary data of the values used for construction of the Venn diagrams. The table shows 1291 the absolute number of patients present at each stage of the disease and whether 1292 glucocorticoids were used for each molecule analysed. The percentage of individuals 1293 present in each experimental condition is highlighted in brackets. 1294 1295 Figure 5. Re-analysis of transcriptome data reinforced the correlation between 1296 acetylcholine and arachidonic acid to Covid-19 severity and mortality. 1297 (A) Gene coexpression profile from CVL, CVH, and NCV samples in principal 1298 module M1 (1047 genes, red lines) and mean gene expression (black line) (n=51). Genes 1299 identified as being associated with AA and ACh were only found in module M1; ten are 1300 related to cholinergic receptors and the ACh release cycle and eight to AA metabolism 1301 along with several biomarkers related to Covid-19 severity (Supplementary Appendix II; 1302 Table S4). (B) Heatmap of normalized gene expression related to the ACh and AA 1303 pathways and Covid-19 biomarkers (n=51). Almost all coexpressed genes of the ACh and 1304 AA pathways were upregulated in the CVL versus CVH group, including twelve 1305 cholinergic receptors, six ACh release cycle genes, eighteen AA production and 1306 metabolism pathway genes, and nine eicosanoid receptors (Figure S6A ; Supplementary 1307 All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for thisthis version posted January 15, 2021. ; https://doi.org/10.1101/2021.01.07.20248970doi: medRxiv preprint Page 55 of 84 Appendix III). Approximately 36% of these genes were also upregulated in the CVL 1308 versus NCV group but not in the CV versus NCV group (n=51); among them, elongation 1309 of very long-chain fatty acid protein 2 (ELOVL2) is involved in linoleic acid metabolism 1310 (Figure S6A ; Supplementary Appendix III). Cases 3, 9, and 11 were glucocorticoid -1311 treated patients. (C) First-order gene interaction network of the M1 module containing 1312 differentially expressed genes related to ACh (brown), AA (green), Covid-19 biomarkers 1313 (red), betweenness hubs (blue), and albumin genes (magenta – Covid-19 biomarker and 1314 hub). Ten DEGs related to ACh and ten DEGs of AA populated the biological network 1315 and are connected to thirteen genes involved in the pathophysiology of Covid -19 and 1316 CD36. ACh- and AA-related genes showed relatively low values of centrality metrics and 1317 may be under the action of some hub genes, such as oestrogen receptor II (ESR2), insulin-1318 like growth factor II mRNA binding protein (IGF2BP1), albumin (ALB), and others 1319 (Supplementary Appendix IV). (D) Plasma and BAL levels of ACh (n=52) . ACh levels 1320 in BAL fluid exhibited a tendency towards increase compared to plasma samples from 1321 deceased or discharged patients . (E) AA (n= 48) and AA concentrations were higher in 1322 plasma than in BAL in both patient groups. (F) 5-HETE (n= 48) and (G) 11-HETE (n=47) 1323 in several/critical patients according to outcome (discharged or deceased). 5-HETE and 1324 11-HETE levels were only significantly higher in the BAL of deceased patients. (H) 1325 Transcription levels of choliner gic muscarinic M3 receptor (CHRM3) and cholinergic 1326 receptor nicotinic alpha 7 (CHRNA7) (n=51) and (I) oxoeicosanoid receptor 1 (OXER1) 1327 (n=51). A unique gene expression profile in lung samples from Covid -19 patients who 1328 had died was characterized by high le vels of pro -inflammatory cholinergic receptors 1329 (CHRM3, CHRNA3, and CHRNA5) 9,110, low levels of anti -inflammatory cholinergic 1330 receptor (CHRNA7) 113, and high expression of oxoeicosanoid receptor 1 (OXER1) 50 1331 which mediates the pro -inflammatory effects of eicosanoids (Figures S6B , S6C). Other 1332 All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for thisthis version posted January 15, 2021. ; https://doi.org/10.1101/2021.01.07.20248970doi: medRxiv preprint Page 56 of 84 eicosanoid and cholinergic receptors were also differentially expressed in CVL patients 1333 compared to the CVH or NCV group (Figures S6A, S6B, S6C). Finally, transcriptome 1334 analysis of fifteen Covid -19 and five non -Covid-19 samples indicated a correlation 1335 between ACh and AA genes and Covid -19 severity in some CVL patients. Significant 1336 differences in BAL/plasma level s of mediators were set at P< 0.05 according to the 1337 Kruskal-Wallis test followed by Dunn’s posttest. *present in module M1. Abbreviations: 1338 Covid-19 low viral load (CVL) - Covid-19 high viral load (CVH) - non-Covid-19 (NCV). 1339 All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for thisthis version posted January 15, 2021. ; https://doi.org/10.1101/2021.01.07.20248970doi: medRxiv preprint Page 57 of 84 Figure 1 1340 All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for thisthis version posted January 15, 2021. ; https://doi.org/10.1101/2021.01.07.20248970doi: medRxiv preprint Page 58 of 84 Figure 2 1341 All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for thisthis version posted January 15, 2021. ; https://doi.org/10.1101/2021.01.07.20248970doi: medRxiv preprint Page 59 of 84 Figure 3 1342 All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for thisthis version posted January 15, 2021. ; https://doi.org/10.1101/2021.01.07.20248970doi: medRxiv preprint Page 60 of 84 Figure 4 1343 All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for thisthis version posted January 15, 2021. ; https://doi.org/10.1101/2021.01.07.20248970doi: medRxiv preprint Page 61 of 84 Figure 5 1344 All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for thisthis version posted January 15, 2021. ; https://doi.org/10.1101/2021.01.07.20248970doi: medRxiv preprint Page 62 of 84 Supplementary Tables 1345 Table S1 – Classification of study participants 1346 1347 The participants were classified into five clinical groups (G1–G5): healthy participants (G1) and 1348 Covid-19 patients (G2 to G5), based on the severity of disease, clinical parameters, patient’s 1349 management, and laboratory findings, following the recommendations from WHO 132–137. These 1350 classifications were used to define the scale of the clinical progression of patients. The inclusion 1351 criteria were as follows: (i) signed informed consent form; (ii) fit into one of the five clinical 1352 groups; (iii) healthy participants must be negative f or SARS -CoV-2 nucleic acid; (iv) 1353 asymptomatic or symptomatic participants must be positive for SARS-CoV-2 nucleic acid and/or 1354 anti-SARS-CoV-2 antibody; (v) age ≥12 years. Participants from groups G1, G2, and G5 were 16 1355 All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for thisthis version posted January 15, 2021. ; https://doi.org/10.1101/2021.01.07.20248970doi: medRxiv preprint Page 63 of 84 years old or older, while the partici pants from groups G3 and G4 were 12 years old or older. 1356 Pregnancy was the only exclusion criterion for healthy participants (G1). Abbreviations: FiO 2, 1357 fraction of inspired oxygen; PaO2, partial pressure of oxygen. 1358 1359 All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for thisthis version posted January 15, 2021. ; https://doi.org/10.1101/2021.01.07.20248970doi: medRxiv preprint Page 64 of 84 Table S2 – Data of demographic, clinical, and blood findings (n=229) 1360 1361 All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for thisthis version posted January 15, 2021. ; https://doi.org/10.1101/2021.01.07.20248970doi: medRxiv preprint Page 65 of 84 1362 Abbreviations: N, number of participants; SD, standard deviation; IQR, interquartile; RNL, ratio between neutrophils and lymp hocytes; N 1363 (%); TTPa, activated partial thromboplastin time; TP, prothrombin time; INR, international normalised ratio. §Comparison of the control 1364 group (healthy participants) with all patients. The values were compared using the χ2 test and one -way analysis of variance (ANOVA), 1365 Mann-Whitney test, and nonparametric t-tests for continuous variables. p<0.05 was considered statistically significant. 1366 All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for thisthis version posted January 15, 2021. ; https://doi.org/10.1101/2021.01.07.20248970doi: medRxiv preprint Page 66 of 84 Table S3 - Demographic, clinical characteristics and blood findings data from 1367 severe/critical participants of which bronchoalveolar lavage (BAL) were collected 1368 (n=45) 1369 1370 All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for thisthis version posted January 15, 2021. ; https://doi.org/10.1101/2021.01.07.20248970doi: medRxiv preprint Page 67 of 84 Abbreviations: N, number or values; SD, standard deviation; IQR, minimum and maximum 1371 values; RNL, ratio between neutrophils and lymphocytes; N (%); TTPa, activated partial 1372 thromboplastin time; TP, prothrombin time; INR, international normalised ratio. §Comparison of 1373 the Covid-19 negative patients group with Covid-19 positive patients. The values were compared 1374 using the χ2 test and one -way analysis of variance (ANOVA), Mann -Whitney test, and 1375 nonparametric t-tests for continuous variables. p<0.05 was considered statistically significant. 1376 1377 All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for thisthis version posted January 15, 2021. ; https://doi.org/10.1101/2021.01.07.20248970doi: medRxiv preprint Page 68 of 84 Table S4 – List of genes related to acetylcholine and arachidonic acid 1378 pathways and Covid-19 biomarkers founded in co-expression module M1 1379 1380 Co-expression analysis generated five different co-expression gene modules (M1 to 1381 M5) and only in M1 we identified genes that encoding proteins associated to 1382 cholinergic receptors (muscarinic and nicotinic), ACh release cycle, and AA 1383 metabolism. In additio n, the M1 module also contains some biomarkers related to 1384 All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for thisthis version posted January 15, 2021. ; https://doi.org/10.1101/2021.01.07.20248970doi: medRxiv preprint Page 69 of 84 Covid-19 severity (albumin, C -reactive protein, IL -8, fibrinogen, and IL1β) 1385 21,34,138,139. 1386 Table S5 - Treatment of severe and critical Covid -19 patients with glucocorticoids 1387 does not alter the levels of plasmatic cytokines and circulating neutrophils and 1388 lymphocytes 1389 1390 Abbreviations: N, number of participants; SEM, standard error of the mean; IQR, interquartile; 1391 RNL, ratio between neutrophils and lymphocytes. §Comparison of the severe non-GC group with 1392 the severe GC group, and the critical non -GC group with the critical GC group. The values were 1393 compared using the χ2 test, one -way analysis of variance (ANOVA), Mann -Whitney test, and 1394 nonparametric t-tests for continuous variables. p<0.05 was considered statistically significant. 1395 As shown in the table, glucocorticoid therapy in severe Covid -19 patients did not show a 1396 statistically significant influence on the plasma levels of the cytokines evaluated in our study. The 1397 use of glucocorti coids and their impact on the production of inflammatory molecules is closely 1398 associated with several factors, such as the dose, duration, and time of initiation of therapy, since, 1399 All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for thisthis version posted January 15, 2021. ; https://doi.org/10.1101/2021.01.07.20248970doi: medRxiv preprint Page 70 of 84 in the critical stages of Covid -19, no beneficial effects of glucocorticoid s have been observed in 1400 the inflammation control and patient outcome 70,71. 1401 Table S6 - Impact of glucocorticoid therapy in the outcome of hospitalized Covid -1402 19 patients 1403 1404 Abbreviations: N, number of participants (%). 1405 Hospitalised Covid-19 patients from the moderate, severe, and critical groups were 1406 administered glucocorticoid (GC) therapy or not (non-GC) (methylprednisolone; range 40 to 500 1407 mg/kg/day, or dexamethasone; range 1.5 to 6.0 mg/kg/day, by intravenous route). Most 1408 hospitalised patients were treated with glucocorticoids (moderate [69.6%], severe [87.0%], and 1409 critical [73.5%]). All moderate non -GC-patients were discharged, while GC patients had a 1410 mortality rate of 18.8%. In addition, the mortality rates for severe -GC (40.4%) and critical-GC 1411 (86.1%) patients were higher than those for moderate -CG patients. The discharge rates of non -1412 GC patients vary according to the clinical categorization (100.0%, 57.1%, and 30.8% for 1413 moderate, severe, and critical, respectively). Hospitalis ed Covid -19 patients from our cohort 1414 showed fewer benefits of glucocorticoid therapy than those reported in the RECOVERY trial 68, 1415 but similar benefits to those described in the CoDEX trial conducted in Brazil 69. This reduction 1416 of glucocorticoid benefits could be associated with several factors, including a low mean 1417 PaO2:FiO2 ratio and an overload of the health systems of countries with limited resources, such 1418 as Brazil 69, as well as glucocorticoid dose, initiation, and duration of therapy 70,71. 1419 1420 All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for thisthis version posted January 15, 2021. ; https://doi.org/10.1101/2021.01.07.20248970doi: medRxiv preprint Page 71 of 84 Table S7 – Analysis of the potential interference of Covid-19-confounding variables 1421 on the correlation between high plasma levels of cholinergic and lipid mediators and 1422 Covid-19 severity. 1423 1424 We analysed some confounding variables associated with Covid -19, such as comorbidities (Diabetes 1425 mellitus, hypertension, and obesity) and risk factors (advance age and gender) 51,140, using data from 1426 participants whose plasma levels of lipid mediators (AA, 5-HETE, and 11-HETE) and ACh were measured. 1427 Lipid mediators data came from Covid -19 patients and heathy participants presented in Figure 1 (Panels: 1428 F, G, and H), while ACh data came from glucocorticoid-non-treated Covid-19 patients presented in Figure 1429 4E. Age and hypertension or only age are potential confounding variables on the correlation between high 1430 plasma levels of lipid mediators or ACh and Covid-19 severity, because (i) age had significant Spearman’s 1431 correlation with AA (r=0.36; P=0.0042), 5 -HETE (r=0.36; P=0.0042), and ACh (r=0.37; P =0.0485); and 1432 (ii) hypertensive patients had significantly increased plasma levels of AA (p=0.0062 - Mann-Whitney test). 1433 §Kruskal-Wallis or Fisher’s tests were used to compare the differences between all clinical categories. Data 1434 are expressed as median (IQR - interquartile range) or number (% - percentages), and P < 0.05 was 1435 considered statistically significant. 1436 All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for thisthis version posted January 15, 2021. ; https://doi.org/10.1101/2021.01.07.20248970doi: medRxiv preprint Page 72 of 84 Table S8 - Analysis of the potential interference of Covid -19-confounding 1437 variables on the correlation between high ACh levels in severe/critical 1438 patients, treated or not with glucocorticoids, and Covid-19 severity. 1439 1440 Here we used the same confounding variables associated with Covid-19 described on Table 1441 S7. This analysis was based on plasma and BAL cholinergic mediator (ACh) data from 1442 severe/critical Covid -19 patients, treated (GC) or not (non -GC) with glucocorticoids, and 1443 reported in Figure 4 (Panels: F and G). None of the confounding variables tested had significant 1444 potential to interfere with the glucocorticoid-therapy ability to reduce the high plasma and BAL 1445 ACh levels in severe/critical patients. §Mann-Whitney or Fisher’s tests were used to compare 1446 the differences between severe and critical clinical categorie s. Data are expressed as median 1447 (IQR - interquartile range) or number (% - percentages), and P < 0.05 was considered 1448 statistically significant. 1449 All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for thisthis version posted January 15, 2021. ; https://doi.org/10.1101/2021.01.07.20248970doi: medRxiv preprint Page 73 of 84 Table S9 - Impact of Covid -19 confounding variables on the c omparative analysis 1450 between high levels of lipid mediators in plasma and BAL samples from severe and 1451 critical patients. 1452 1453 We used the same confounding variables associated with Covid-19 reported on Table S7. This 1454 analysis was based on plasma and BAL levels of lipid mediators (AA, 5 -HETE, and 11-HETE) 1455 from severe/critical Covid -19 patients reported in Figure 4 (Panels: B, C, a nd D). None of the 1456 confounding variables tested had significant potential to interfere with the comparison between 1457 altered plasm and BAL levels of lipid mediators in severe/critical patients. §Mann-Whitney, chi-1458 square, or Fisher’s tests were used to compare the differences between severe and critical patients. 1459 Data are expressed as median (IQR - interquartile range) or number (% - percentages), and P < 1460 0.05 was considered statistically significant. 1461 All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for thisthis version posted January 15, 2021. ; https://doi.org/10.1101/2021.01.07.20248970doi: medRxiv preprint Page 74 of 84 Supplementary Figures 1462 Figure S1 – Metabolic signatures of plasma and BAL samples from patients 1463 infected or not with SARS-CoV-2. 1464 1465 Hierarchical clustering based on different characteristics of abundant metabolites 1466 (ANOVA-FDR <0.1) between controls and patients positive for SARS-CoV-2 at different stages 1467 of the disease in plasma (A) and non -Covid-19 versus Covid -19 BAL patients (B). In (A), the 1468 clinical classification of Covid -19 was: healthy participants (n=20), asymptomatic -to-mild 1469 (n=10), moderate (n=12), severe (n=16), and critical (n=13) groups. In (B), the same analysis was 1470 performed with data from BAL samples of non -Covid-19 (n=12) and Covid-19 (n=26) patients. 1471 The detection levels of metabolites were defined using Z-score normalisation. 1472 The analysis of hierarchical clustering (A) shows highly different metabolomic profiles 1473 in the plasma for each group of individuals according to disease severity. BAL analysis (B) of 1474 samples from Covid -19 patients demonstrated an increase in the abundance of metabolites 1475 compared to non-Covid-19 individuals. The results indicated that SARS-CoV-2 infection induces 1476 All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for thisthis version posted January 15, 2021. ; https://doi.org/10.1101/2021.01.07.20248970doi: medRxiv preprint Page 75 of 84 changes in the metabolic profile of humans. Our data are in agreement with previous results 1477 showing virus -induced metabolic reprogramming in the host. The increased abundance of 1478 metabolites and their pathways, such as free fatty acids and amino acids, was correlated to the 1479 increase in viral proliferation, since these biomolecules can act as building blocks and fuel for 1480 this process 141–143. 1481 All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for thisthis version posted January 15, 2021. ; https://doi.org/10.1101/2021.01.07.20248970doi: medRxiv preprint Page 76 of 84 Figure S2 - Gating strategy used for flow cytometry analysis of monocyte subsets 1482 and CD14/CD36 expression 1483 1484 1485 In (A), dot plots show a representative gating strategy for the analysis of FSC-H/FSC-A, 1486 SSC-A/FVS-620 (viable cells), and SSC-A/FSC-A, followed by CD14+HLA-DR+ monocytes and 1487 the classical (CD14 highCD16-), intermediate (CD14 +CD16+) or non -classical CD14 lowCD16+ 1488 subsets, with subsequent CD36 mean fluorescence intensity (MFI) in whole blood (upper panel) 1489 and bronchoalveolar lavage fluid (BAL, bottom panel). In (B), a violin plot shows the frequency 1490 of non-classical monocytes from BAL of non-Covid-19 and Covid-19 patients. In (C), CD36 MFI 1491 in classical, intermediate, and non -classical monocytes from BAL of non -Covid-19 and Covid-1492 19 samples. Differences between groups were calculated using the Mann-Whitney test. 1493 All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for thisthis version posted January 15, 2021. ; https://doi.org/10.1101/2021.01.07.20248970doi: medRxiv preprint Page 77 of 84 The membrane markers and gating strategies used to defi ne the different monocyte 1494 subpopulations in the blood and BAL samples were adapted from a previous publication (ref). In 1495 BAL, there was a tendency to reduce non-classical monocytes in Covid-19 patients compared to 1496 non-Covid-19 patients (Figure 2SB), as wel l as a decrease in CD36 expression in classical and 1497 intermediate monocytes (Figure 2SC), although the difference was not statistically significant. 1498 Although not significantly, this reduction may have a biological importance. These data are 1499 similar to the profile observed in the blood monocytes of Covid-19 patients (Figure 3J). 1500 All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for thisthis version posted January 15, 2021. ; https://doi.org/10.1101/2021.01.07.20248970doi: medRxiv preprint Page 78 of 84 Figure S3 - Determination of monocyte subsets in the blood and BAL samples from 1501 non-Covid-19 and Covid-19 patients 1502 1503 1504 1505 1506 Peripheral blood from healthy participants (n=12), and asympto matic-to-mild 1507 (n=15), moderate (n=15), severe (n=35), or critical (n=20) patients, and BAL from Covid-1508 19 (n=10) and non -Covid-19 (n=11) participants were collected and used for the 1509 characterisation of monocyte subpopulations by flow cytometry and t -distributed 1510 stochastic neighbour embedding (t -SNE) analysis. (A) Frequency of CD14 +HLA-DR+ 1511 cells. (B) t -SNE evaluation of classical (CD14 highCD16-), intermediate (CD14 +CD16+), 1512 All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for thisthis version posted January 15, 2021. ; https://doi.org/10.1101/2021.01.07.20248970doi: medRxiv preprint Page 79 of 84 and non -classical CD14 lowCD16+ blood monocytes in Covid -19 patients, categorised 1513 according to disease severity. (C) Colour mapping t -SNE and violin plot showing the 1514 mean fluorescence intensity (MFI) of HLA -DR expression in the circulating monocytes 1515 of Covid-19 subjects. In (D) and (E), samples of BAL from non-Covid-19 and Covid-19 1516 patients were evaluated to determine the frequency of CD14+HLA-DR+ cells and 1517 classical, intermediate, or non -classical monocyte subsets, respectively. Differences 1518 among groups were calculated using the Kruskal -Wallis test with Du nn’s multiple 1519 comparison post-test, and the corresponding values are indicated in the figures. 1520 In comparison to healthy participants (Figure S3A, S3B, and 3SC), a significant 1521 reduction was observed in the intermediate monocyte popul ation, as well as in the 1522 expression of HLA -DR molecules in the blood of Covid -19 patients, depending on 1523 disease severity. The diminishment of HLA-DR in monocytes correlated with TNF values 1524 obtained in the blood of patients positive for SARS-CoV-2 (Figure 2P; principal article), 1525 since monocytes with a pro-inflammatory profile are the main producers of this cytokine 1526 144. Similarly, reduced intermediate monocytes were found in the BAL of patients positive 1527 for Covid-19 (Figure S3E). Interestingly, these patients also had a significant increase in 1528 the production of IL-8, IL-10, and IL-6 cytokines in the blood and BAL, as shown before 1529 (Figure 2M, 2N, and 2Q and Figure 4A; principal article). The high production of 1530 cytokines, especially IL-6, in patients with Covid-19, has been correlated with a decrease 1531 in the expression of HLA -DR in monocytes, while the therapeutic inhibition of IL -6 re-1532 established the expression of this molecule in those individuals1451533 All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for thisthis version posted January 15, 2021. ; https://doi.org/10.1101/2021.01.07.20248970doi: medRxiv preprint Page 80 of 84 Figure S4 - Biomarker networks in non-Covid-19 and healthy participants 1534 1535 A network of interactions between lipid mediators, ACh , sCD14, and cytokines in (A) 1536 healthy participants (n=39) and (B) BAL non-Covid-19 participants (n=13). Network layouts of 1537 personalised biomarkers were set up to identify the relevant association in healthy and non-Covid-1538 19 participants. Each connecting line denotes a significant correlation between a pair of markers. 1539 Continuous lines represent positive correlations, while dashed lines represent negative 1540 correlations ( p<0.05). The degree of significance is represented by the thickness of the line. 1541 Correlations were determined using Spearman’s test; the values of r and p were used to classify 1542 the connections as weak (r ≤ 0.35, p<0.05), moderate (r = 0.36–0.67, p<0.01), or strong (r ≥ 0.68, 1543 p< 0.001). The absence of a line indicates the non -existence of the relationship. To evaluate the 1544 relationship between levels of lipid mediators, ACh, sCD14, and cytokines in non -Covid-19, a 1545 series of correlation analyses were performed. 1546 When comparing the interactions analysed between the molecules of the healthy and non -1547 Covid-19 groups, we observed that (A) In the blood of healthy individuals, the interactions were 1548 found to occur mainly between cytokines (IL -1β, TNF, IL -8, IL -6, and IL -10) and the lipid 1549 mediator 5 -HETE. (B) In thee BAL samples from hospitalised non -Covid-19 patients, lipid 1550 mediators were found to mediate these interactions and influence the clinical outcomes of this 1551 group of patients. 1552 All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for thisthis version posted January 15, 2021. ; https://doi.org/10.1101/2021.01.07.20248970doi: medRxiv preprint Page 81 of 84 Figure S5 – Treatment of severe and critical SARS-CoV-2-infected patients with 1553 glucocorticoids inhibits ACh release 1554 1555 1556 The production of lipid mediators and the shedding of sCD14 in patients with Covid -19 1557 at the severe and critical stages was not modified by treatment w ith glucocorticoids (GC). 1558 However, a marked reduction was observed in ACh release. Intersection analysis depicted in 1559 Venn diagrams revealed the number of patients producing AA, 5 -HETE, 11 -HETE, ACh, or 1560 sCD14, above the control (healthy participants) values, in the absence (A) or presence (B) of GC 1561 therapy. The table shows the number (n) of plasma or BAL samples from Covid -19 patients at 1562 the severe and critical stages of disease, for each mediator. The percentages shown in parentheses 1563 represent the frequency of patients producing the respective mediator for the corresponding 1564 sample number. 1565 The Venn diagrams show the intersections of AA, 5-HETE, 11-HETE, ACh, and sCD14s 1566 in Covid-19 patients at the severe and critical stages of the disease, in both blood and B AL. No 1567 significant changes were observed in the quantification of lipid mediators or sCD14, comparing 1568 patients treated with or without glucocorticoids (Figure S5A and FigureS5B). Interestingly, 1569 critically ill patients treated with glucocorticoids showed a 20% and 65% reduction in the ACh 1570 All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for thisthis version posted January 15, 2021. ; https://doi.org/10.1101/2021.01.07.20248970doi: medRxiv preprint Page 82 of 84 concentration in both blood and BAL samples, respectively, compared to patients who were not 1571 administered drugs (Figure S5B). While a decrease in free-AA and its metabolites (5-HETE and 1572 11-HETE) has not been reported in the literature, some studies have demonstrated the beneficial 1573 effects of the use of glucocorticoids in other respiratory syndromes, having been found to reduce 1574 the production of AA-derived mediators 72,146. In agreement with our results, there is no evidence 1575 to support corticoid treatment for Covid -19 147. Data related t o the effect of glucocorticoids on 1576 ACh release in viral infections are scarce. However, according to our previous findings on 1577 scorpion envenomation3, glucocorticoid treatment should be initiated early after infection to block 1578 the release of lipid mediators, which could lead to the inhibition of premature ACh release. On 1579 the contrary, the late administration of GC could place patients at a point of no return, with early 1580 ACh release impacting vital organs, dama ged by the negative impacts of this neurotransmitter 1581 before GC administration. Moreover, it should be considered that, in Covid -19, other unknown 1582 mechanisms may control ACh release, in addition to the COX-2 dependent metabolites. 1583 All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for thisthis version posted January 15, 2021. ; https://doi.org/10.1101/2021.01.07.20248970doi: medRxiv preprint Page 83 of 84 Figure S6 - Differential expression and profile of genes involved in AA and ACh 1584 pathways from SARS-CoV-2 deceased patients 1585 1586 1587 1588 Abbreviations: CVL, Covid-19 low viral load; CVH, Covid-19 high viral load; NCV, 1589 non-Covid-19 viral load 9,50,110,113. 1590 As shown in (A), DEGs were upregulated in CVL versus CVH patients and CVL 1591 versus NCV patients. The transcript expression of cholinergic receptors (muscarinic and 1592 nicotinic) (B) and eicosanoid receptors (C) was normalised in the NCV, CVH, and CVL 1593 groups. CVL patients displayed increased pulmonary levels of expression of genes 1594 All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for thisthis version posted January 15, 2021. ; https://doi.org/10.1101/2021.01.07.20248970doi: medRxiv preprint Page 84 of 84 associated with the ACh and AA pathways, encoding for cholinergic receptors, the ACh 1595 release cycle, AA metabolism, and eicosanoid receptors, compared to CVH or NCV 1596 patients. The expression profile of some genes may favour inflammatory events, such as 1597 upregulated cholinergic and eicosanoid receptors (CHRM3, CHRNA3, CHRNA5, and 1598 OXER1) and low levels of the anti -inflammatory cholinergic receptor (CHRNA7) 1599 9,50,110,113. 1600 All rights reserved. No reuse allowed without permission. perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in The copyright holder for thisthis version posted January 15, 2021. ; https://doi.org/10.1101/2021.01.07.20248970doi: medRxiv preprint