Lipid Storm Within the Lungs of Severe COVID-19 Patients

Home , Leukotriene B4, Omega oxidation

medRxiv preprint doi: https://doi.org/10.1101/2020.12.04.20242115; this version posted December 7, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission. 1 1 2 Lipid storm within the lungs of severe COVID-19 patients: Extensive levels of cyclooxygenase 3 and lipoxygenase-derived inflammatory metabolites 4 5 *Anne-Sophie Archambault1,2, *Younes Zaid3,4, Volatiana Rakotoarivelo1,2, Étienne Doré5,6, Isabelle 6 Dubuc5, Cyril Martin1,2, Youssef Amar7, Amine Cheikh4, Hakima Fares4, Amine El Hassani4, Youssef 7 Tijani8, Michel Laviolette1, Éric Boilard5,6,9, #Louis Flamand5,9 and #Nicolas Flamand1,2. 8 9 *ASA and YZ equally contributed to this work. 10 #LF and NF equally contributed to this work 11 12 1Centre de recherche de l’Institut universitaire de cardiologie et de pneumologie de Québec, Faculté de 13 médecine, Département de médecine, Université Laval, Québec City, QC, Canada 14 2Canada Research Excellence Chair in the Microbiome-Endocannabinoidome Axis in Metabolic Health, 15 Université Laval, Québec, QC, Canada. 16 3Biology Department, Faculty of Sciences, Mohammed V University, Rabat, Morocco 17 4Cheikh Zaïd Hospital, Abulcasis University of Health Sciences, Rabat, Morocco 18 5Centre de Recherche du Centre Hospitalier Universitaire de Québec- Université Laval, Canada; 19 6Centre de Recherche Arthrite, Université Laval, Québec, Canada 20 7Moroccan Foundation for Advanced Science, Innovation & Research (MAScIR), Rabat, Morocco 21 8Faculty of Medicine, Mohammed VI University of Health Sciences, Casablanca, Morocco 22 9Département de microbiologie-infectiologie et d'immunologie, Université Laval, QC, Canada 23 24 25 Corresponding authors: 26 Nicolas Flamand, PhD Louis Flamand, PhD 27 Centre de recherche de l’IUCPQ Centre de recherche du CHU-CHUL 28 2725 chemin Ste-Foy, Office A3140 2705 Laurier boulevard, Office T1-64 29 Québec City, QC G1V 4G5, Canada Québec City, QC G1V 4G2, Canada 30 Tel: 1-418-656-8711 ext.3733 Tel: 1-418-525-4444 ext.46164 31 Email: [email protected] Email: [email protected] 32 33 34 Funding: This work was supported by the Cheikh Zaid Foundation awarded to YZ and grants from the 35 Canadian New Frontier Research Fund (NFRN-2019-00004) awarded to LF (PI), EB (co-PI) and NF (co- 36 PI) and the Natural Sciences and Engineering Research Council of Canada awarded to NF. ASA received 37 a doctoral award from the Canadian Institutes of Health Research. ED and EB are respectfully recipient 38 of doctoral and senior awards from the Fonds de Recherche du Québec en Santé (FRQS).

NOTE: This preprint reports new research that has not been certified by peer review and should not be used to guide clinical practice. medRxiv preprint doi: https://doi.org/10.1101/2020.12.04.20242115; this version posted December 7, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission. 2 39 ABSTRACT 40 41 BACKGROUND. Severe Acute Respiratory Syndrome coronavirus 2 (SARS-CoV-2) is the infectious 42 agent responsible for Coronavirus disease 2019 (COVID-19). While SARS-CoV-2 infections are often 43 benign, there are also severe COVID-19 cases, characterized by severe bilobar pneumonia that can 44 decompensate to an acute respiratory distress syndrome, notably characterized by increased 45 inflammation and a cytokine storm. While there is no cure against severe COVID-19 cases, some 46 treatments significantly decrease the severity of the disease, notably aspirin and dexamethasone, which 47 both directly or indirectly target the biosynthesis (and effects) of numerous bioactive lipids. 48 49 OBJECTIVE. Our working hypothesis was that severe COVID-19 cases necessitating mechanical 50 ventilation were characterized by increased bioactive lipid levels modulating lung inflammation. We thus 51 quantitated several lung bioactive lipids using liquid chromatography combined to tandem mass 52 spectrometry. 53 54 RESULTS. We performed an exhaustive assessment of the lipid content of bronchoalveolar lavages 55 from 25 healthy controls and 33 COVID-19 patients necessitating mechanical ventilation. Severe 56 COVID-19 patients were characterized by increased fatty acid levels as well as an accompanying 57 inflammatory lipid storm. As such, most quantified bioactive lipids were heavily increased. There was a

58 predominance of cyclooxygenase metabolites, notably TXB2 >> PGE2 ~ 12-HHTrE > PGD2.

59 Leukotrienes were also increased, notably LTB4, 20-COOH-LTB4, LTE4, and eoxin E4. 15-lipoxygenase 60 metabolites derived from linoleic, arachidonic, eicosapentaenoic and docosahexaenoic acids were also

61 increased. Finally, yet importantly, specialized pro-resolving mediators, notably lipoxin A4 and the D- 62 series resolvins, were also found at important levels, underscoring that the lipid storm occurring in severe 63 SARS-CoV-2 infections involves pro- and anti-inflammatory lipids. 64 65 CONCLUSIONS. Our data unmask the important lipid storm occurring in the lungs of patients afflicted 66 with severe COVID-19. We discuss which clinically available drugs could be helpful at modulating the 67 lipidome we observed in the hope of minimizing the deleterious effects of pro-inflammatory lipids and 68 enhancing the effects of anti-inflammatory and/or pro-resolving lipids. 69 medRxiv preprint doi: https://doi.org/10.1101/2020.12.04.20242115; this version posted December 7, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission. 3 70 Abbreviations 71 2-AG, 2-arachidonoyl-glycerol; AA, arachidonic acid; AEA, N-arachidonoyl-ethanolamine; ALT, 72 alanine transaminase; AST, aspartate transaminase;BAL, bronchoalveolar lavage; COVID-19, 73 coronavirus disease 2019; COX, cyclooxygenase; CRP, C-reactive protein; DGLA, dihomo-γ-linolenic 74 acid; DHA, docosahexaenoic acid; DPA, docosapentaenoic acid; EPA, eicosapentaenoic acid; EX, 75 eoxin; HETE, hydroxyeicosatetraenoic acid; ICU, intensive care unit; IL, interleukin; LA, linoleic acid; 76 LDH, lactate dehydrogenase; LOX, lipoxygenase; LT, leukotriene; PG, prostaglandin; Rv, resolvin; 77 SARS-CoV-2, severe acute respiratory syndrome coronavirus 2; SPM, specialized proresolving 78 mediator; TX, thromboxane. 79 medRxiv preprint doi: https://doi.org/10.1101/2020.12.04.20242115; this version posted December 7, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission. 4 80 INTRODUCTION 81 Coronavirus disease 2019 (COVID-19) is an infectious disease caused by the Severe Acute Respiratory 82 Syndrome coronavirus 2 (SARS-CoV-2) (1). The clinical manifestations of the disease are often mild 83 but severe cases can lead to bilobar pneumonia with hyperactivation of the inflammatory cascade and 84 progression to acute respiratory distress syndrome (ARDS), necessitating mechanical ventilation (2, 3). 85 SARS-CoV-2 spreads predominantly from respiratory droplets of infected individuals to mucosal 86 epithelial cells in the upper airways and oral cavity (4). It infects cells via its homotrimeric spike protein 87 binding to host-cell expressing angiotensin-converting enzyme-2 (ACE2) receptor in a protease- 88 dependent manner (5). 89 90 Autopsies from severe COVID-19 patients unmasked a heterogeneous disease that can be characterized 91 by diffuse alveolar damage, endothelial damage, thrombosis of small and medium vessels, pulmonary 92 embolism, and inflammatory cell infiltration, sometimes more lymphocytic, sometimes more 93 granulocytic (3, 6-8). An uncontrolled systemic inflammatory response known as the cytokine storm also 94 occurs. This cytokine storm, which is the consequence of an important release of pro-inflammatory 95 cytokines, such as tumor necrosis factor-α, interleukin (IL)-1β, IL-6, IL-7 and granulocyte-colony 96 stimulating factor, is suggested as an important contributor of SARS-CoV-2 lethality (8, 9). However, 97 the heterogeneity of the disease indicates that, aside cytokines (and chemokines), other immunological 98 effectors contribute to the sustained inflammatory responses observed in the most severe forms of 99 COVID-19. 100 101 Among the numerous soluble effectors involved in the infectious/inflammatory response, bioactive lipids 102 such as eicosanoids likely promote their fair share of deleterious effects, notably by enhancing leukocyte 103 recruitment and activation, by participating in the exudate formation and by stimulating platelet 104 aggregation and thrombus formation. In contrast, specialized pro-resolving mediators (SPMs), which 105 mainly consist of docosanoids, could dampen the inflammatory response and even promote its resolution 106 (10). While therapeutic approaches targeting the biosynthesis or effects of inflammatory lipids are readily 107 available, the levels of those lipids in the lungs during severe COVID-19 remain unknown. 108 109 Herein, we performed a detailed lipidomic analysis of bronchoalveolar lavage (BAL) fluids from healthy 110 volunteers and severe COVID-19 patients. We provide evidence that severe COVID-19 patients are 111 characterized by robust levels of lipids arising from the cyclooxygenase (COX) and lipoxygenase (LOX) medRxiv preprint doi: https://doi.org/10.1101/2020.12.04.20242115; this version posted December 7, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission. 5 112 pathways as well increased SPM levels. Our study further highlights that few associations exist between 113 clinical parameters and bioactive lipids present in the primary site of disease, the lungs. medRxiv preprint doi: https://doi.org/10.1101/2020.12.04.20242115; this version posted December 7, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission. 6 114 MATERIAL AND METHODS 115 116 Materials 117 All lipids were obtained from Cayman Chemical (Ann Arbor, MI, USA) unless indicated otherwise. 118 DMSO and C17:1-LPA were purchased from Sigma-Aldrich (St-Louis, MO, USA). DMSO, ammonium

119 acetate, acetic acid, LC-MS-grade MeOH, MeCN and H2O were purchased from Fisher Scientific 120 (Ottawa, Canada). Cytokine/chemokine bead arrays were purchased from Becton Dickinson (BD). 121 122 Ethics 123 This study was approved by the Local Ethics Committee of Cheikh Zaid Hospital (Project: 124 CEFCZ/PR/2020- PR04), Rabat, Morocco and complies with the Declaration of Helsinki and all subjects 125 signed a consent form. The analysis of bioactive lipids in the BAL of healthy donors and severe COVID- 126 19 patients was approved by the Comité d’éthique de la recherche de l’Institut universitaire de cardiologie 127 et de pneumologie de Québec – Université Laval and the Comité d’éthique de la recherche du CHU de 128 Québec-Université Laval. 129 130 Clinical criteria for subject selection. Non-smoking healthy subjects taking no other medication than 131 anovulants, without any documented acute or chronic inflammatory disease and without recent (8 weeks) 132 airway infection were recruited for bronchoalveolar lavages at Institut universitaire de cardiologie et 133 pneumologie de Québec (Québec City, Canada). Only BALs in which the cell content in alveolar 134 macrophages was greater than 95%, with lymphocytes as the other main cell type, were kept for further 135 analyses. Severe COVID-19 patients were enrolled based on the inclusion criteria for intubation and the 136 need of mechanical ventilation support at Cheikh Zaid Hospital (Rabat, Morocco). Following intubation 137 and the initiation of mechanical ventilation, blood sampling was done to assess the clinical variables 138 shown in Table 1. Patients were ventilated in the prone decubitus position with a tidal volume of 6 ml/kg

139 and a PEEP varying between 8 – 14 cm H2O. The FiO2 was adjusted between 0.6 – 1.0 to obtain a SpO2 140 ≥ 92%. 141 142 Bronchoalveolar lavages. BAL fluids from healthy volunteers were obtained as follows: One 50 ml 143 bolus of sterile 0.9% saline was injected in a sub-segmental bronchi of the right middle lobe. The obtained 144 lavages were centrifuged (350 × g, 10 minutes) to pellet cells and the supernatants were immediately 145 frozen (-80º Celsius) until further processing. A 5 ml aliquot of the samples was thawed then reduced to 146 ~500 µl using a stream of nitrogen. BAL fluids from severe COVID-19 patients were obtained less than medRxiv preprint doi: https://doi.org/10.1101/2020.12.04.20242115; this version posted December 7, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission. 7 147 2 hrs following intubation and as follows: a total of 100 ml of sterile 0.9% saline (2 boli of 50 ml) was 148 injected in a sub-segmental bronchi of the right middle lobe. The obtained lavages were pooled and 149 centrifuged (320 × g, 15 minutes) to pellet cells. Supernatants were next concentrated as for the BALs of 150 healthy donors using a stream of nitrogen and immediately frozen (-80º Celsius) until further processing. 151 All of the evaporated samples were processed for LC-MS/MS analysis as described below. 152 153 LC-MS/MS analyses. Samples (500 µl) were denatured by adding 500 µl of LC-MS grade MeOH 154 containing the internal standards then warmed at 60º Celsius for 30 minutes to inactivate (or not) SARS- 155 CoV-2. Samples were then denatured overnight (-20º Celsius). The morning after, samples were warmed 156 to room temperature and centrifuged at 10,000 × g to remove the denaturated proteins. The obtained 157 supernatants were diluted with water containing 0.01% acetic acid to obtain a MeOH concentration of 158 10%, then acidified to pH 3 with acetic acid. Lipids were extracted by solid phase extraction (SPE) using 159 Strata-X cartridges (Polymeric Reversed Phase, 60 mg/1 ml, Phenomenex, Torrance, CA, USA) as 160 described before (11). In brief, columns were preconditioned with 2 ml MeOH containing 0.01% acetic

161 acid followed by 2 ml LC-MS grade H2O containing 0.01% acetic acid. After sample loading, the

162 columns were washed with 2 ml LC-MS grade H2O containing 0.01% acetic acid. Lipids were then eluted 163 from the cartridges with 1 ml MeOH containing 0.01% acetic acid. Eluates were dried under a stream of

164 nitrogen and reconstituted with 25 µl of solvent A (H2O + 1 mM NH4 + 0.05% acetic acid) and 25 µl of

165 solvent B (MeCN/H2O, 95/5, v/v + 1 mM NH4 + 0.05% acetic acid). A volume of 40 µl of the resulting 166 mixture was injected onto a RP‐HPLC column (Kinetex C8, 150 × 2.1 mm, 2.6 µm, Phenomenex). 167 Samples were eluted at a flow rate of 400 µl/min with a linear gradient of 10% solvent B that increased 168 to 35% in 2 min, up to 75% in 10 min, from 75% to 95% in 0.1 min, and held at 98% for 5 min before 169 re‐equilibration to 10% solvent B for 2 min. The HPLC system was directly interfaced into the 170 electrospray source of a Shimadzu 8050 triple quadrupole mass spectrometer and mass spectrometric 171 analyses were performed in the positive (+) or the negative (−) ion mode using multiple reaction 172 monitoring for the specific mass transitions of each lipid (Table 2). Quantification of each compound 173 was done using internal standards and calibration curves that were extracted by solid-phase extraction as 174 described above. 175 176 Statistical analysis 177 Statistical analyses were done using the GraphPad Prism 9 software. P values < 0.05 were considered 178 significant. 179 medRxiv preprint doi: https://doi.org/10.1101/2020.12.04.20242115; this version posted December 7, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission. 8 180 RESULTS 181 While the circulating (plasmatic) cytokines’ profile observed in severe COVID-19 patients is relatively 182 well-defined (8, 12-14), a much more limited number of studies characterizing the cytokine storm in the 183 lungs of severe COVID-19 patients were available (13, 15, 16). In fact, to the best of our knowledge, no 184 study simultaneously quantitated numerous soluble mediators in BAL fluids. We thus performed their 185 lipidomic analysis in order to define whether an inflammatory lipid storm was present in the lungs of 186 severe COVID-19 patients. Samples were compared to those obtained from healthy subjects. As a 187 starting point, we investigated eicosanoids derived from the cyclooxygenase (COX) pathway. Our 188 working hypothesis was that some COX metabolites would be increased, considering that IL-1β levels 189 are elevated in the BALs of some COVID-19 patients (16) and that IL-1β is a potent inducer of 190 cyclooxygenase (COX)-2 expression (17-19). The only COX metabolites that were detected in the BALs

191 of healthy subjects were PGD2 (16 donors out of 25) and 12-HHTrE (2 donors out of 25). In contrast, 192 BALs from severe COVID-19 patients had a significant increase of all COX metabolites investigated

193 (figure 1). Most samples had detectable levels of PGD2 (27 of 33 donors) and the PGI2 metabolite 6-

194 keto-PGF1α (16 of 33 donors), while all samples contained PGF2α, important levels of PGE2, the inactive

195 TXA2 metabolite TXB2, and the thromboxane synthase metabolite 12-HHTrE (figure 1A). Although 196 there was a significant increase in all metabolites, the most increased COX-derived metabolites were

197 TXB2 >> PGE2 ~ 12-HHTrE > PGD2. We next addressed if the different COX metabolites correlated

198 with each other. The levels of TXB2 strongly correlated with those of 12-HHTrE, PGD2, PGE2 and PGF2α 199 (figure 1B-E). Of note, few correlations were found between COX metabolites and the clinical

200 parameters shown in Table1. To that end, PGF2α and 12-HHTrE weakly but significantly correlated 201 negatively with AST (figure 7), 202 203 Leukotrienes are also potent bioactive lipids previously documented to regulate the inflammatory 204 responses in the airway/lung in diseases, notably ARDS (20, 21). As such, their involvement in the 205 pathogenesis of COVID-19 has recently been postulated even though their levels were unknown (22). 206 We thus quantitated the levels of leukotrienes derived from the 5-lipoxygenase pathway as well as those 207 derived from the 15-lipoxygenase pathway, the 14,15-leukotrienes, which are now classified as eoxins

208 (23). Leukotrienes were not detected in the BAL fluids of healthy volunteers, with the exception of LTB4, 209 which was detected in 4 of 25 healthy subjects. In contrast, COVID-19 patients were characterized by

210 increased levels of LTB4 and its CYP4F3A metabolite 20-COOH-LTB4 (figure 2A) as well as other

211 minor metabolites (20-OH-LTB4 and 12-oxo-LTB4). Of note, the levels of LTB4 found in the BAL fluids 212 of severe COVID-19 patients are comparable to those found in BAL fluids of ARDS patients (16). As medRxiv preprint doi: https://doi.org/10.1101/2020.12.04.20242115; this version posted December 7, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission. 9

213 expected, there was a strong correlation between the levels of LTB4 and those of 20-COOH-LTB4 (figure

214 2B). Importantly, the ratio between the sum of 20-OH- and 20-COOH-LTB4 to LTB4 ((20-OH- + 20-

215 COOH-LTB4)/LTB4) was 2.843 +/- 0.532 (mean +/- SEM), indicating an increased omega oxidation of

216 LTB4. The levels of LTB4 strongly correlated with those of TXB2 (figure 2C). The cysteinyl-leukotrienes

217 LTE4 and EXE4 were also found in most severe COVID-19 patients (figure 2D). Surprisingly, the levels

218 of EXE4 trended to be higher than those of LTE4. Furthermore, there was a trend toward a negative

219 correlation between LTE4 and EXE4 levels although it did not reach statistical significance (figure 2E). 220 This suggests that some COVID-19 patients are more prone to biosynthesize the 15-lipoxygnease-

221 derived EXC4 while others are more prone to biosynthesize the 5-lipoxygenase-derived LTC4. Of note,

222 LTB4 levels correlated with those of EXE4 but not with those of LTE4 (figure 2F,G). Again, a limited 223 number of correlations between the different leukotrienes and the clinical parameters from Table 1 were 224 found (figure 7), supporting the concept that the levels of leukotrienes in the lungs are not reflected in 225 the blood by the most recognized COVID-19-related biomarkers. 226 227 The prostaglandins and leukotrienes data (figures 1 and 2) support the idea that an increase in both 228 arachidonic acid (AA) release and metabolism occurs in severe COVID-19 patients. As such, we 229 investigated the levels of some fatty acids, which are the precursors of oxidized linoleic acid (LA) 230 mediators (OXLAMs), eicosanoids and docosanoids. To that end, the levels of LA, AA, 231 docosapentaenoic acid n-3 (DPAn-3) and docosahexaenoic acid (DHA) were all significantly increased 232 in the BALs of severe COVID-19 patients. In contrast, the levels of eicosapentaenoic acid (EPA) did not 233 change (figure 3). Of note, all tested BAL fatty acids negatively correlated with AST and ALT but 234 positively correlated with the circulating platelet-to-lymphocyte ratio (figure 7). 235 236 The increased fatty acid levels (figure 3) and the increased in the 15-lipoxygenase-derived eoxins (figure 237 2D) led us to investigate if other fatty acid-related biosynthetic pathways were also enhanced in SARS- 238 CoV-2-infected patients. To that end, we also quantitated additional AA-derived metabolites derived 239 from the 12- and 15-lipoxygenases. The levels of 12-HETE and 15-HETE were also significantly 240 increased in the BAL fluids of severe COVID-19 patients, although not to the same extent than the 5- 241 lipoxygenase metabolite 5-HETE (figure 4A), indicating that AA metabolism via the 15-lipoxygenase 242 (and possibly the 12-lipoxygenase) pathway is increased. In addition to AA, the 15-lipoxygenase 243 pathway can metabolize numerous polyunsaturated fatty acids having a 1Z,4Z-pentadiene motif near 244 their omega end, notably LA, EPA, DPAn-3 and DHA. We thus also assessed the levels of their most 245 documented 15-lipoxygenase metabolites and found significant increases in 13-HODE (derived from medRxiv preprint doi: https://doi.org/10.1101/2020.12.04.20242115; this version posted December 7, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission. 10 246 LA), 15-HETrE (derived from dihomo-γ-linolenic acid, DGLA), 12-, 15-, and 18-HEPE, 17-HDPA(n- 247 3) as well as 14- and 17-HDHA (figure 4B,C). Dual lipoxygenases metabolites from AA were also 248 significantly increased (figure 4D). Of note, the levels of 15-HETE strongly correlated with those of 13- 249 HODE (figure 4E) supporting an increased 15-lipoxygenase activity in the lungs of severe COVID-19

250 patients. Furthermore, 15-HETE levels also correlated with those of TXB2 (figure 4F). Finally yet 251 importantly, we also detected significant increases in 9-HODE, 8-HETE, and 11-HETE, which arise from 252 the non-enzymatic oxidation of LA and AA, pointing to an increased oxidative bust in the lungs of severe 253 COVID-19 patients (figure 4A,B). 254 255 14- and 17-HDHA, which are increased in the BALs of severe COVID-19 patients (figure 4C), are often 256 regarded as good indicators of D-series resolvin and maresin levels, which are part of the ever-expanding 257 class of specialized pro-resolving mediators (SPMs). SPMs are usually more complicated to quantitate 258 due to their very low concentrations (10). Given their documented pro-resolving actions and anti- 259 microbial activities, SPMs were recently postulated as being a potential approach for diminishing the 260 inflammatory burden of COVID-19 (24). We thus investigated the levels of some commercially available 261 SPMs. All SPMs were below detection limit in the BAL samples of healthy volunteers we analyzed. In 262 contrast, most investigated SPMs, aside from RvD3, were detected in the BAL fluids of severe COVID-

263 19 patients. LXA4 was the most prominent one, followed by RvD4, RvD5, RvD2, RVD1 and PDX 264 (figure 5A). RvD5 levels correlated with those of RvD4 (figure 5B) but not those of DHA (figure 5C) 265 indicating that the same patients are responsible for the increased D-series resolvins, which is mostly due 266 to increased 15-lipoxygenase activity rather than an increase in DHA availability. The levels of RvD5

267 also correlated with those of TXB2 (figure 5D). Again, there was no correlation between RvD5 with the

268 clinical parameters from table 1. Unsurprisingly, the levels of LXA4 correlated with those of 5-HETE

269 and 15-HETE (figure 5E,F), given that LXA4 biosynthesis requires both the 5- and the 15-lipoxygenase

270 pathways. As for RvD5, the levels of LXA4 also correlated with those of TXB2 and the other 271 prostaglandins (figure 6). Despite our good sensitivity (see table 2), we did not detect the presence or 272 RvE1, Maresin-1 and -2. 273 274 When assessing the lipidome for which we could quantitate a given lipid mediator in at least 50% of 275 severe COVID-19 patients, we can appreciate that most lipids consistently correlated with each other 276 with few exceptions (figure 6). Indeed, Fatty acids levels were not a key determinant for the levels of 277 their metabolites, indicating that downstream enzymes (cyclooxygenases, lipoxygenases) are likely more 278 important in that respect. Furthermore, and in agreement with their almost significant negative medRxiv preprint doi: https://doi.org/10.1101/2020.12.04.20242115; this version posted December 7, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission. 11

279 association (figure 2E), EXE4 and LTE4 correlated with different bioactive lipid subsets, LTE4 positively

280 correlating with fatty acid and 15-HETrE levels while EXE4 mostly correlating with other 15- 281 lipoxygenase metabolites. medRxiv preprint doi: https://doi.org/10.1101/2020.12.04.20242115; this version posted December 7, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission. 12 282 DISCUSSION 283 284 Bioactive lipids are recognized pharmaceutical targets for the treatment of numerous inflammatory 285 diseases and lipid modulators have been utilized as such, in the last fifty years, with great success. It was 286 thus imperative, at least to us, to define whether bioactive lipid levels were modulated in the context of 287 severe COVID-19. 288 289 Herein, we provide clear evidences that the BALs of severe COVID-19 patients contain large amounts 290 of bioactive lipids, which likely participate in the deleterious inflammatory response found in severe 291 cases. More specifically, our data indicate that the BALs of severe COVID-19 patients are characterized

292 by 1) an increase in COX metabolites, notably the TXA2 metabolite TXB2; 2) an increase in LTB4 and 293 its metabolites; 3) an increase in CysLT metabolites; 4) an increase of 15-lipoxygenases metabolites 294 derived from LA, AA, EPA, DPA and DHA; 5) an increase in SPMs derived from AA and DHA; and 6) 295 a limited number of correlations between BAL lipids vs. blood markers and/or clinical parameters. 296 297 Cyclooxygenase metabolites were heavily increased in the BALs of severe COVID-19 patients (figure 298 1). This was somewhat expected, notably because of the increased levels of IL-1β described before (16),

299 the latter being a good COX-2 inducer. In the lungs, COX metabolites play multiple roles. TXA2 300 promotes bronchoconstriction and activate inflammatory cells and platelets, which might be consistent 301 with the reported enhanced platelet activation and increased thrombosis in COVID-19 (14, 25, 26). On

302 the other hand, PGD2 acting via the DP2 receptor promotes the recruitment and activation of eosinophils,

303 basophils, mastocytes and innate lymphoid cells (27-30). In contrast, PGE2 is usually regarded as anti- 304 inflammatory in the lungs, notably by downregulating the pro-inflammatory functions of myeloid

305 leukocytes such as neutrophils and macrophages, notably by activating the EP2 and EP4 receptors, 306 ultimately leading to increased cyclic AMP levels (31). Therapies targeting the prostaglandin pathway 307 are numerous and usually effective at controlling inflammation. Dexamethasone, a steroid improving 308 COVID-19 symptoms and diminishing mortality (32-34), limits COX-2 expression (35-37) and might 309 thus improve COVID-19 severity, at least in part, by diminishing the biosynthesis of prostaglandins and

310 possibly TXA2. Furthermore, nonsteroidal anti-inflammatory drugs (NSAIDs) such as aspirin and 311 ibuprofen are recognized inhibitor of cyclooxygenase activity. Early at the beginning of the pandemic, it 312 was postulated that NSAIDs could worsen the severity of COVID-19 (38). However, a Danish study 313 indicated that ibuprofen was not linked to severe adverse outcome (39). Furthermore, aspirin was recently 314 shown to significantly diminish ICU admissions and the need of mechanical ventilation in an medRxiv preprint doi: https://doi.org/10.1101/2020.12.04.20242115; this version posted December 7, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission. 13 315 observational, retrospective study (40). Altogether, our data, combined with the recent studies on their 316 safety in a context of COVID-19, support the hypothesis that dexamethasone and NSAIDs might 317 diminish the inflammatory status of severe COVID-19 patients by diminishing the

318 prostaglandins/thromboxane storm in the lungs. While TXA2 and PGD2 might have deleterious effects

319 in the lungs, PGE2 and PGI2 might have the opposite effect, notably by inhibiting leukocyte functions

320 (PGE2) and/or promoting the vasodilation of the microvasculature (PGI2). To that end, testing whether

321 blocking the deleterious effects of PGD2 and TXA2 with the dual DP2/TP antagonist Ramotraban might 322 be beneficial in this particular context, as proposed recently (41). 323

324 Prostaglandins and TXA2, as assessed by quantitating TXB2 levels, were not the only eicosanoids that 325 were increased in the BALs of severe COVID-19 patients. Indeed, leukotrienes were also increased,

326 notably LTB4, LTE4, and EXE4 (figure 2). This indicates that in addition of the deleterious effects that

327 some COX metabolites might have, those linked to LTB4 and CysLTs might also impair lung functions

328 and participate in the inflammatory burden in the lungs of severe COVID-19 patients (22). While LTB4 329 has previously been shown to stimulate host defense (41), notably during viral infections, the

330 w-LTB4/LTB4 ratio of 2.84 we observed is more reminiscent of what is found in cystic fibrosis and 331 severely burned patients, two conditions in which infections are not well controlled (42-45).

332 Conceptually, this might diminish the effectiveness of LTB4 at promoting its host-defense related 333 functions, as we recently published (46). Interestingly, there was a trend toward an inverse correlation

334 between LTE4 and EXE4, indicating that these cysteinyl-LTs are likely coming from different cellular 335 sources. This was supported by the differential and significant correlations between the different lipids

336 we observed, LTE4 and EXE4 differentially correlating with distinct lipids (figure 6). To that end, EXE4 337 levels in the BALs positively and significantly correlated with blood eosinophils counts. While there was

338 a strong trend between EXE4 and BAL eosinophils counts the latter correlation had a p value of 0.0536. 339 Altogether, this suggests that the presence and/or biosynthesis of eoxins is, in part, linked to eosinophils. 340 A therapeutic approach to target LTs from the 5-lipoxygenase pathway could be the use of Zileuton, 341 which is commercially available for the management of asthma and has a good safety profile (47). 342 343 Another pharmaceutical approach that could simultaneously diminish the levels of COX and 344 5-lipoxygenase-derived eicosanoids could be the use of phosphodiesterase 4 inhibitors such as 345 Roflumilast. While there is no evidence of phosphodiesterase 4 inhibitors’ efficacy in the management 346 of SARS-CoV-2-infected people (48), they might diminish inflammation by preventing the degradation 347 of cyclic AMP, increasing intracellular cAMP concentrations and decreasing the release of AA, the medRxiv preprint doi: https://doi.org/10.1101/2020.12.04.20242115; this version posted December 7, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission. 14 348 precursor of prostaglandins and leukotrienes. Furthermore, cAMP-elevating agents such as adenosine

349 and PGE2 are recognized anti-inflammatory autacoids that inhibit numerous leukocyte functions such as 350 eicosanoid biosynthesis, oxidative burst, calcium mobilization and migration (49-53). While

351 phosphodiesterase 4 inhibitors might diminish the levels of PGE2 by diminishing AA levels, they would

352 nonetheless amplify the inhibitory effects of the remaining PGE2, as recently demonstrated in cellulo 353 (54). 354

355 While it is largely accepted that the group IVA cytosolic phospholipase A2 mediates most of AA release 356 in activated human leukocytes (e.g. in (55)), the increased levels of LA, EPA and DHA support the idea

357 that other phospholipases A2 are also heavily involved in the lipidome we have quantitated. Those

358 additional phospholipases likely belong to the superfamily of secreted phospholipases A2, which can 359 hydrolyze phospholipids containing fatty acids other than AA (56, 57) and are reported to contribute to 360 lung inflammation (58-61). However, and assuming that a large part of eicosanoids is the consequence

361 of the group IVA cytosolic phospholipase A2, our findings do not allow us to pinpoint which other 362 phospholipases are involved in the release of LA, AA, EPA, DPA and DHA. 363 364 The metabolism of EPA, DPA and DHA, which are omega-3 fatty acids, often results in the biosynthesis 365 of SPMs, most of which are anti-inflammatory and pro-resolving lipids. Their presence in high levels 366 was somewhat surprising as they are generally thought to appear during the resolution of inflammation, 367 after the biosynthesis of prostaglandins and leukotrienes has occurred and following a class-switch from 368 pro-inflammatory to pro-resolving lipids (62). However, our data indicate that prostaglandins, 369 leukotrienes and SPMs can certainly co-exist and participate in the inflammatory cascade during the 370 acute phase of inflammation/infection in which resolution has not been fully engaged. Moreover, given 371 their strong documented anti-inflammatory effects and their host-defense boosting functions, they could 372 be viewed as potential mediators to enhance to limit the infection. To that end, it was recently proposed 373 that dexamethasone could participate in the upregulation of SPM biosynthesis and effects in moderate to 374 severe COVID-19 patients, although this needs to be confirmed (63). While omega-3 fatty acids, notably 375 purified preparation enriched in EPA and DHA are commercially available as dietary supplements, we 376 did not see a correlation between SPMs and the levels of EPA or DHA, indicating that SPMs levels are 377 likely more dependent on biosynthetic enzymes other than phospholipases such as lipoxygenases than 378 EPA or DHA availability. As of today, no resolution pharmacology treatment has been approved by 379 governmental health agencies yet. However, some putative E- and D-series resolvin precursors, notably 380 18-HEPE and 17-HDHA, are available as dietary supplements. medRxiv preprint doi: https://doi.org/10.1101/2020.12.04.20242115; this version posted December 7, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission. 15 381 382 In addition to cyclooxygenase and lipoxygenase derivatives, other bioactive lipids are recognized, at least 383 in mice, to downregulate the immune response. This is notably the case of the endocannabinoids N- 384 arachidonoyl-ethanolamine (AEA) and 2-arachidonoyl-glycerol (2-AG), which are AA-derived lipids

385 (64). The anti-inflammatory effects of AEA and 2-AG are mostly related to the activation of the CB2 386 receptor, which is heavily expressed by leukocytes (65). In addition to 2-AG and AEA, their congeners 387 from the monacyl-glycerols and N-acyl-ethanolamines and N-acyl-aminoacids families, now viewed as 388 the endocannabinoidome, could potentially regulate the inflammatory response during SARS-CoV-2 389 infections, notably by activating other anti-inflammatory receptors (66). It will thus be crucial to decipher

390 whether the endocannabinoidome is modulated during COVID-19 and if a potential target such as CB2 391 receptor agonists, or endocannabinoid hydrolase inhibitors could be helpful in the treatment of COVID- 392 19 or other coronavirus-mediated infections. 393 394 Nonetheless the expanded definition of a dysregulated lipidome we unravel in SARS-CoV-2 infections 395 requiring mechanical ventilation, our study has limitations. 1) BALs from healthy controls are from 396 younger individuals than those from severe COVID-19 patients. This could have an impact on some

397 mediators, notably PGD2 levels, which are increased in aged mice and worsen SARS-CoV infections 398 (67); 2) The procedure to obtain BAL fluids is slightly different between healthy controls and severe 399 COVID-19 patients. Indeed, while the volume of saline injected to healthy people was 50 ml, that of 400 severe COVID-19 patients was 100 ml. The possible outcome of this discrepancy could be and 401 underestimation of lipid levels in one group vs. the other. While some might argue that the first saline 402 bolus contains more lipids than the following one(s), other might argue that lipid levels are consistent in 403 each bolus. Given that, in our hands, the number of BAL cells is usually greater in the first bolus than in 404 the subsequent ones (when multiple boli are utilized), we are confident that the BAL fluid from the first 405 bolus likely contains more lipids than the second one. This would thus lead to a slight underestimation 406 of lipids in the COVID-19 group. 3) Unfortunately, the leukocyte counts in the BALs of severe COVID- 407 19 patients did not include the number of alveolar macrophages and it is thus impossible to determine a 408 clear differential count of leukocytes in those samples. Furthermore, the eosinophil counts were 409 somewhat limited by the apparatus as it could estimate the counts to +/- 105/ml cells, which likely has an

410 impact on the different correlations involving eosinophils, notably with EXE4; 4) We could not compare 411 the levels of lipids in the BALs of severe COVID-19 patients to those found in the blood. This appears 412 important given the limited number of correlations found between BAL lipids and other circulating 413 markers such as CRP and D-dimers. To that end, a recent study indicated differences in some of the medRxiv preprint doi: https://doi.org/10.1101/2020.12.04.20242115; this version posted December 7, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission. 16 414 herein investigated bioactive lipids in the circulation (68). It will thus be important to pursue our 415 investigations and determine whether there is a transposition of the BAL lipidome to the circulation. 416 417 In conclusion, our data unmask that in addition to the recognized cytokine storm previously documented, 418 lung fluids of SARS-CoV-2-infected patients indicate that a lipid storm also occurs. As highlighted 419 above, the rich library of governmental health agencies-approved lipid modulators might provide 420 beneficial and usually low-cost add-ons to the current therapeutic arsenal utilized to diminish the severity 421 of COVID-19 and possibly additional coronavirus-mediated infections. 422 423 424 425 426 ACKNOWLEDGEMENTS 427 ASA, AC, VR, ML, and NF are members of the Quebec Respiratory Health research Network. medRxiv preprint doi: https://doi.org/10.1101/2020.12.04.20242115; this version posted December 7, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission. 17 428 REFERENCES 429 430 431 432 1. Wu F, Zhao S, Yu B, Chen YM, Wang W, Song ZG, Hu Y, Tao ZW, Tian JH, Pei YY, Yuan ML, 433 Zhang YL, Dai FH, Liu Y, Wang QM, Zheng JJ, Xu L, Holmes EC, Zhang YZ. A new 434 coronavirus associated with human respiratory disease in China. Nature 2020; 579: 265-269. 435 2. Verdoni L, Mazza A, Gervasoni A, Martelli L, Ruggeri M, Ciuffreda M, Bonanomi E, D'Antiga L. 436 An outbreak of severe Kawasaki-like disease at the Italian epicentre of the SARS-CoV-2 437 epidemic: an observational cohort study. Lancet 2020; 395: 1771-1778. 438 3. Huang C, Wang Y, Li X, Ren L, Zhao J, Hu Y, Zhang L, Fan G, Xu J, Gu X, Cheng Z, Yu T, Xia J, 439 Wei Y, Wu W, Xie X, Yin W, Li H, Liu M, Xiao Y, Gao H, Guo L, Xie J, Wang G, Jiang R, 440 Gao Z, Jin Q, Wang J, Cao B. Clinical features of patients infected with 2019 novel coronavirus 441 in Wuhan, China. Lancet 2020; 395: 497-506. 442 4. Zhang XSD, C. SARS-CoV-2 and Health Care Worker Protection in Low-Risk Settings: a Review of 443 Modes of Transmission and a Novel Airborne Model Involving Inhalable Particles. Clinical 444 Microbiology Reviews 2020; 34. 445 5. Oberfeld B, Achanta A, Carpenter K, Chen P, Gilette NM, Langat P, Said JT, Schiff AE, Zhou AS, 446 Barczak AK, Pillai S. SnapShot: COVID-19. Cell 2020; 181: 954-954 e951. 447 6. Damiani S, Fiorentino M, De Palma A, Foschini MP, Lazzarotto T, Gabrielli L, Viale PL, Attard L, 448 Riefolo M, D'Errico A. Pathological Post Mortem Findings in Lungs Infected With Sars-Cov 2. 449 J Pathol 2020. 450 7. Barton LM, Duval EJ, Stroberg E, Ghosh S, Mukhopadhyay S. COVID-19 Autopsies, Oklahoma, 451 USA. Am J Clin Pathol 2020; 153: 725-733. 452 8. Zhang H, Zhou P, Wei Y, Yue H, Wang Y, Hu M, Zhang S, Cao T, Yang C, Li M, Guo G, Chen X, 453 Chen Y, Lei M, Liu H, Zhao J, Peng P, Wang CY, Du R. Histopathologic Changes and SARS- 454 CoV-2 Immunostaining in the Lung of a Patient With COVID-19. Ann Intern Med 2020; 172: 455 629-632. 456 9. Amigues I, Pearlman AH, Patel A, Reid P, Robinson PC, Sinha R, Kim AH, Youngstein T, 457 Jayatilleke A, Konig MF. Coronavirus disease 2019: investigational therapies in the prevention 458 and treatment of hyperinflammation. Expert Rev Clin Immunol 2020. 459 10. Serhan CN, Levy BD. Resolvins in inflammation: emergence of the pro-resolving superfamily of 460 mediators. J Clin Invest 2018; 128: 2657-2669. 461 11. Archambault AS, Turcotte C, Martin C, Provost V, Larose MC, Laprise C, Chakir J, Bissonnette E, 462 Laviolette M, Bosse Y, Flamand N. Comparison of eight 15-lipoxygenase (LO) inhibitors on 463 the biosynthesis of 15-LO metabolites by human neutrophils and eosinophils. PLoS One 2018; 464 13: e0202424. 465 12. Del Valle DM, Kim-Schulze S, Huang HH, Beckmann ND, Nirenberg S, Wang B, Lavin Y, Swartz 466 TH, Madduri D, Stock A, Marron TU, Xie H, Patel M, Tuballes K, Van Oekelen O, Rahman A, 467 Kovatch P, Aberg JA, Schadt E, Jagannath S, Mazumdar M, Charney AW, Firpo-Betancourt A, 468 Mendu DR, Jhang J, Reich D, Sigel K, Cordon-Cardo C, Feldmann M, Parekh S, Merad M, 469 Gnjatic S. An inflammatory cytokine signature predicts COVID-19 severity and survival. Nat 470 Med 2020. 471 13. Tang Y, Liu J, Zhang D, Xu Z, Ji J, Wen C. Cytokine Storm in COVID-19: The Current Evidence 472 and Treatment Strategies. Front Immunol 2020; 11: 1708. 473 14. Zaid Y, Puhm F, Allaeys I, Naya A, Oudghiri M, Khalki L, Limami Y, Zaid N, Sadki K, Ben El 474 Haj R, Mahir W, Belayachi L, Belefquih B, Benouda A, Cheikh A, Langlois MA, Cherrah Y, 475 Flamand L, Guessous F, Boilard E. Platelets Can Associate with SARS-Cov-2 RNA and Are 476 Hyperactivated in COVID-19. Circ Res 2020. medRxiv preprint doi: https://doi.org/10.1101/2020.12.04.20242115; this version posted December 7, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission. 18 477 15. Xiong Y, Liu Y, Cao L, Wang D, Guo M, Jiang A, Guo D, Hu W, Yang J, Tang Z, Wu H, Lin Y, 478 Zhang M, Zhang Q, Shi M, Liu Y, Zhou Y, Lan K, Chen Y. Transcriptomic characteristics of 479 bronchoalveolar lavage fluid and peripheral blood mononuclear cells in COVID-19 patients. 480 Emerg Microbes Infect 2020; 9: 761-770. 481 16. Liao M, Liu Y, Yuan J, Wen Y, Xu G, Zhao J, Cheng L, Li J, Wang X, Wang F, Liu L, Amit I, 482 Zhang S, Zhang Z. Single-cell landscape of bronchoalveolar immune cells in patients with 483 COVID-19. Nat Med 2020; 26: 842-844. 484 17. Miller DB, Munster D, Wasvary JS, Simke JP, Peppard JV, Bowen BR, Marshall PJ. The 485 heterologous expression and characterization of human prostaglandin G/H synthase-2 (COX-2). 486 Biochem Biophys Res Commun 1994; 201: 356-362. 487 18. Newman SP, Flower RJ, Croxtall JD. Dexamethasone suppression of IL-1b-induced 488 cyclooxygenase 2 expression is not mediated by lipocortin-1 in A549 cells. Biochem Biophys 489 Res Commun 1994; 202: 931-939. 490 19. Mitchell JA, Belvisi MG, Akarasereenont P, Robbins RA, Kwon OJ, Croxtall J, Barnes PJ, Vane 491 JR. Induction of cyclo-oxygenase-2 by cytokines in human pulmonary epithelial cells: 492 regulation by dexamethasone. Br J Pharmacol 1994; 113: 1008-1014. 493 20. Stephenson AH, Lonigro AJ, Hyers TM, Webster RO, Fowler AA. Increased concentrations of 494 leukotrienes in bronchoalveolar lavage fluid of patients with ARDS or at risk for ARDS. Am 495 Rev Respir Dis 1988; 138: 714-719. 496 21. Ratnoff WD, Matthay MA, Wong MY, Ito Y, Vu KH, Wiener-Kronish J, Goetzl EJ. 497 Sulfidopeptide-leukotriene peptidases in pulmonary edema fluid from patients with the adult 498 respiratory distress syndrome. J Clin Immunol 1988; 8: 250-258. 499 22. Funk CD, Ardakani A. A Novel Strategy to Mitigate the Hyperinflammatory Response to COVID- 500 19 by Targeting Leukotrienes. Front Pharmacol 2020; 11: 1214. 501 23. Feltenmark S, Gautam N, Brunnstrom A, Griffiths W, Backman L, Edenius C, Lindbom L, 502 Bjorkholm M, Claesson HE. Eoxins are proinflammatory arachidonic acid metabolites 503 produced via the 15-lipoxygenase-1 pathway in human eosinophils and mast cells. Proc Natl 504 Acad Sci U S A 2008; 105: 680-685. 505 24. Panigrahy D, Gilligan MM, Huang S, Gartung A, Cortes-Puch I, Sime PJ, Phipps RP, Serhan CN, 506 Hammock BD. Inflammation resolution: a dual-pronged approach to averting cytokine storms 507 in COVID-19? Cancer Metastasis Rev 2020; 39: 337-340. 508 25. Manne BK, Denorme F, Middleton EA, Portier I, Rowley JW, Stubben C, Petrey AC, Tolley ND, 509 Guo L, Cody M, Weyrich AS, Yost CC, Rondina MT, Campbell RA. Platelet gene expression 510 and function in patients with COVID-19. Blood 2020; 136: 1317-1329. 511 26. Hottz ED, Azevedo-Quintanilha IG, Palhinha L, Teixeira L, Barreto EA, Pao CRR, Righy C, 512 Franco S, Souza TML, Kurtz P, Bozza FA, Bozza PT. Platelet activation and platelet-monocyte 513 aggregate formation trigger tissue factor expression in patients with severe COVID-19. Blood 514 2020; 136: 1330-1341. 515 27. Nagata K, Hirai H, Tanaka K, Ogawa K, Aso T, Sugamura K, Nakamura M, Takano S. CRTH2, an 516 orphan receptor of T-helper-2-cells, is expressed on basophils and eosinophils and responds to 517 mast cell-derived factor(s). FEBS Lett 1999; 459: 195-199. 518 28. Monneret G, Gravel S, Diamond M, Rokach J, Powell WS. Prostaglandin D2 is a potent 519 chemoattractant for human eosinophils that acts via a novel DP receptor. Blood 2001; 98: 1942- 520 1948. 521 29. Hirai H, Tanaka K, Yoshie O, Ogawa K, Kenmotsu K, Takamori Y, Ichimasa M, Sugamura K, 522 Nakamura M, Takano S, Nagata K. Prostaglandin D2 selectively induces chemotaxis in T helper 523 type 2 cells, eosinophils, and basophils via seven-transmembrane receptor CRTH2. J Exp Med 524 2001; 193: 255-261. medRxiv preprint doi: https://doi.org/10.1101/2020.12.04.20242115; this version posted December 7, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission. 19

525 30. Chang JE, Doherty TA, Baum R, Broide D. Prostaglandin D2 regulates human type 2 innate 526 lymphoid cell chemotaxis. J Allergy Clin Immunol 2014; 133: 899-901 e893. 527 31. Medeiros A, Peres-Buzalaf C, Fortino Verdan F, Serezani CH. Prostaglandin E2 and the 528 suppression of phagocyte innate immune responses in different organs. Mediators Inflamm 529 2012; 2012: 327568. 530 32. Group RC, Horby P, Lim WS, Emberson JR, Mafham M, Bell JL, Linsell L, Staplin N, Brightling 531 C, Ustianowski A, Elmahi E, Prudon B, Green C, Felton T, Chadwick D, Rege K, Fegan C, 532 Chappell LC, Faust SN, Jaki T, Jeffery K, Montgomery A, Rowan K, Juszczak E, Baillie JK, 533 Haynes R, Landray MJ. Dexamethasone in Hospitalized Patients with Covid-19 - Preliminary 534 Report. N Engl J Med 2020. 535 33. Tomazini BM, Maia IS, Cavalcanti AB, Berwanger O, Rosa RG, Veiga VC, Avezum A, Lopes RD, 536 Bueno FR, Silva M, Baldassare FP, Costa ELV, Moura RAB, Honorato MO, Costa AN, 537 Damiani LP, Lisboa T, Kawano-Dourado L, Zampieri FG, Olivato GB, Righy C, Amendola 538 CP, Roepke RML, Freitas DHM, Forte DN, Freitas FGR, Fernandes CCF, Melro LMG, Junior 539 GFS, Morais DC, Zung S, Machado FR, Azevedo LCP, Investigators CC-BI. Effect of 540 Dexamethasone on Days Alive and Ventilator-Free in Patients With Moderate or Severe Acute 541 Respiratory Distress Syndrome and COVID-19: The CoDEX Randomized Clinical Trial. JAMA 542 2020; 324: 1307-1316. 543 34. Group WHOREAfC-TW, Sterne JAC, Murthy S, Diaz JV, Slutsky AS, Villar J, Angus DC, 544 Annane D, Azevedo LCP, Berwanger O, Cavalcanti AB, Dequin PF, Du B, Emberson J, Fisher 545 D, Giraudeau B, Gordon AC, Granholm A, Green C, Haynes R, Heming N, Higgins JPT, Horby 546 P, Juni P, Landray MJ, Le Gouge A, Leclerc M, Lim WS, Machado FR, McArthur C, Meziani 547 F, Moller MH, Perner A, Petersen MW, Savovic J, Tomazini B, Veiga VC, Webb S, Marshall 548 JC. Association Between Administration of Systemic Corticosteroids and Mortality Among 549 Critically Ill Patients With COVID-19: A Meta-analysis. JAMA 2020; 324: 1330-1341. 550 35. Crofford LJ, Wilder RL, Ristimaki AP, Sano H, Remmers EF, Epps HR, Hla T. Cyclooxygenase-1 551 and -2 expression in rheumatoid synovial tissues. Effects of interleukin-1b, phorbol ester, and 552 corticosteroids. J Clin Invest 1994; 93: 1095-1101. 553 36. Masferrer JL, Reddy ST, Zweifel BS, Seibert K, Needleman P, Gilbert RS, Herschman HR. In vivo 554 glucocorticoids regulate cyclooxygenase-2 but not cyclooxygenase-1 in peritoneal 555 macrophages. J Pharmacol Exp Ther 1994; 270: 1340-1344. 556 37. Wilborn J, Crofford LJ, Burdick MD, Kunkel SL, Strieter RM, Peters-Golden M. Cultured lung 557 fibroblasts isolated from patients with idiopathic pulmonary fibrosis have a diminished capacity 558 to synthesize prostaglandin E2 and to express cyclooxygenase-2. J Clin Invest 1995; 95: 1861- 559 1868. 560 38. Little P. Non-steroidal anti-inflammatory drugs and covid-19. BMJ 2020; 368: m1185. 561 39. Lund LC, Kristensen KB, Reilev M, Christensen S, Thomsen RW, Christiansen CF, Stovring H, 562 Johansen NB, Brun NC, Hallas J, Pottegard A. Adverse outcomes and mortality in users of non- 563 steroidal anti-inflammatory drugs who tested positive for SARS-CoV-2: A Danish nationwide 564 cohort study. PLoS Med 2020; 17: e1003308. 565 40. Chow JH, Khanna AK, Kethireddy S, Yamane D, Levine A, Jackson AM, McCurdy MT, Tabatabai 566 A, Kumar G, Park P, Benjenk I, Menaker J, Ahmed N, Glidewell E, Presutto E, Cain S, 567 Haridasa N, Field W, Fowler JG, Trinh D, Johnson KN, Kaur A, Lee A, Sebastian K, Ulrich A, 568 Pena S, Carpenter R, Sudhakar S, Uppal P, Fedeles BT, Sachs A, Dahbour L, Teeter W, Tanaka 569 K, Galvagno SM, Herr DL, Scalea TM, Mazzeffi MA. Aspirin Use is Associated with 570 Decreased Mechanical Ventilation, ICU Admission, and In-Hospital Mortality in Hospitalized 571 Patients with COVID-19. Anesth Analg 2020. 572 41. Gupta A, Kalantar-Zadeh K, Reddy ST. Ramatroban as a Novel Immunotherapy for COVID-19. J 573 Mol Genet Med 2020; 14. medRxiv preprint doi: https://doi.org/10.1101/2020.12.04.20242115; this version posted December 7, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission. 20 574 42. Yang J, Eiserich JP, Cross CE, Morrissey BM, Hammock BD. Metabolomic profiling of regulatory 575 lipid mediators in sputum from adult cystic fibrosis patients. Free Radic Biol Med 2012; 53: 576 160-171. 577 43. Saak A, Schonfeld W, Knoller J, Steinkamp G, von der Hardt H, Konig W. Generation and 578 metabolism of leukotrienes in granulocytes of patients with cystic fibrosis. Int Arch Allergy 579 Appl Immunol 1990; 93: 227-236. 580 44. Knoller J, Schonfeld W, Koller M, Hensler T, Konig W. Arachidonic acid metabolites from 581 polymorphonuclear leukocytes of healthy donors, severely burned patients and children with 582 cystic fibrosis--routine monitoring by high-performance liquid chromatography. J Chromatogr 583 1988; 427: 199-208. 584 45. Brom J, Konig W, Koller M, Gross-Weege W, Erbs G, Muller F. Metabolism of leukotriene B4 by 585 polymorphonuclear granulocytes of severely burned patients. Prostaglandins Leukot Med 1987; 586 27: 209-225. 587 46. Archambault AS, Poirier S, Lefebvre JS, Robichaud PP, Larose MC, Turcotte C, Martin C, Provost 588 V, Boudreau LH, McDonald PP, Laviolette M, Surette ME, Flamand N. 20-Hydroxy- and 20- 589 carboxy-leukotriene (LT) B4 downregulate LTB4-mediated responses of human neutrophils and 590 eosinophils. J Leukoc Biol 2019; 105: 1131-1142. 591 47. Bouchette D, Preuss CV. Zileuton. StatPearls. Treasure Island (FL); 2020. 592 48. Halpin DMG, Criner GJ, Papi A, Singh D, Anzueto A, Martinez FJ, Agusti AA, Vogelmeier CF, 593 Committee GS. Global Initiative for the Diagnosis, Management, and Prevention of Chronic 594 Obstructive Lung Disease: The 2020 GOLD Science Committee Report on COVID-19 & 595 COPD. Am J Respir Crit Care Med 2020. 596 49. Ham EA, Soderman DD, Zanetti ME, Dougherty HW, McCauley E, Kuehl FA, Jr. Inhibition by 597 prostaglandins of leukotriene B4 release from activated neutrophils. Proc Natl Acad Sci U S A 598 1983; 80: 4349-4353. 599 50. Fonteh AN, Winkler JD, Torphy TJ, Heravi J, Undem BJ, Chilton FH. Influence of isoproterenol 600 and phosphodiesterase inhibitors on platelet-activating factor biosynthesis in the human 601 neutrophil. J Immunol 1993; 151: 339-350. 602 51. Denis D, Riendeau D. Phosphodiesterase 4-dependent regulation of cyclic AMP levels and 603 leukotriene B4 biosynthesis in human polymorphonuclear leukocytes. Eur J Pharmacol 1999; 604 367: 343-350. 605 52. Flamand N, Boudreault S, Picard S, Austin M, Surette ME, Plante H, Krump E, Vallee MJ, Gilbert 606 C, Naccache P, Laviolette M, Borgeat P. Adenosine, a potent natural suppressor of arachidonic 607 acid release and leukotriene biosynthesis in human neutrophils. Am J Respir Crit Care Med 608 2000; 161: S88-94. 609 53. Flamand N, Plante H, Picard S, Laviolette M, Borgeat P. Histamine-induced inhibition of 610 leukotriene biosynthesis in human neutrophils: involvement of the H2 receptor and cAMP. Br J 611 Pharmacol 2004; 141: 552-561. 612 54. Turcotte C, Zarini S, Jean S, Martin C, Murphy RC, Marsolais D, Laviolette M, Blanchet MR, 613 Flamand N. The Endocannabinoid Metabolite Prostaglandin E2 (PGE2)-Glycerol Inhibits 614 Human Neutrophil Functions: Involvement of Its Hydrolysis into PGE2 and EP Receptors. J 615 Immunol 2017; 198: 3255-3263. 616 55. Flamand N, Picard S, Lemieux L, Pouliot M, Bourgoin SG, Borgeat P. Effects of pyrrophenone, an 617 inhibitor of group IVA phospholipase A2, on eicosanoid and PAF biosynthesis in human 618 neutrophils. Br J Pharmacol 2006; 149: 385-392. 619 56. Hanasaki K, Ono T, Saiga A, Morioka Y, Ikeda M, Kawamoto K, Higashino K, Nakano K, 620 Yamada K, Ishizaki J, Arita H. Purified group X secretory phospholipase A2 induced prominent 621 release of arachidonic acid from human myeloid leukemia cells. J Biol Chem 1999; 274: 34203- 622 34211. medRxiv preprint doi: https://doi.org/10.1101/2020.12.04.20242115; this version posted December 7, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission. 21 623 57. Murakami M, Masuda S, Shimbara S, Bezzine S, Lazdunski M, Lambeau G, Gelb MH, Matsukura 624 S, Kokubu F, Adachi M, Kudo I. Cellular arachidonate-releasing function of novel classes of 625 secretory phospholipase A2s (groups III and XII). J Biol Chem 2003; 278: 10657-10667. 626 58. Nolin JD, Murphy RC, Gelb MH, Altemeier WA, Henderson WR, Jr., Hallstrand TS. Function of 627 secreted phospholipase A2 group-X in asthma and allergic disease. Biochim Biophys Acta Mol 628 Cell Biol Lipids 2019; 1864: 827-837. 629 59. Hallstrand TS, Lai Y, Altemeier WA, Appel CL, Johnson B, Frevert CW, Hudkins KL, Bollinger 630 JG, Woodruff PG, Hyde DM, Henderson WR, Jr., Gelb MH. Regulation and function of 631 epithelial secreted phospholipase A2 group X in asthma. Am J Respir Crit Care Med 2013; 188: 632 42-50. 633 60. Arbibe L, Koumanov K, Vial D, Rougeot C, Faure G, Havet N, Longacre S, Vargaftig BB, Bereziat 634 G, Voelker DR, Wolf C, Touqui L. Generation of lyso-phospholipids from surfactant in acute 635 lung injury is mediated by type-II phospholipase A2 and inhibited by a direct surfactant protein 636 A-phospholipase A2 protein interaction. J Clin Invest 1998; 102: 1152-1160. 637 61. Ohtsuki M, Taketomi Y, Arata S, Masuda S, Ishikawa Y, Ishii T, Takanezawa Y, Aoki J, Arai H, 638 Yamamoto K, Kudo I, Murakami M. Transgenic expression of group V, but not group X, 639 secreted phospholipase A2 in mice leads to neonatal lethality because of lung dysfunction. J 640 Biol Chem 2006; 281: 36420-36433. 641 62. Buckley CD, Gilroy DW, Serhan CN. Proresolving lipid mediators and mechanisms in the 642 resolution of acute inflammation. Immunity 2014; 40: 315-327. 643 63. Andreakos E, Papadaki M, Serhan CN. Dexamethasone, pro-resolving lipid mediators and 644 resolution of inflammation in COVID-19. Allergy 2020. 645 64. Turcotte C, Chouinard F, Lefebvre JS, Flamand N. Regulation of inflammation by cannabinoids, 646 the endocannabinoids 2-arachidonoyl-glycerol and arachidonoyl-ethanolamide, and their 647 metabolites. J Leukoc Biol 2015; 97: 1049-1070. 648 65. Turcotte C, Blanchet MR, Laviolette M, Flamand N. The CB2 receptor and its role as a regulator of 649 inflammation. Cell Mol Life Sci 2016; 73: 4449-4470. 650 66. Di Marzo V. New approaches and challenges to targeting the endocannabinoid system. Nat Rev 651 Drug Discov 2018; 17: 623-639. 652 67. Zhao J, Zhao J, Legge K, Perlman S. Age-related increases in PGD2 expression impair respiratory 653 DC migration, resulting in diminished T cell responses upon respiratory virus infection in mice. 654 J Clin Invest 2011; 121: 4921-4930. 655 68. Schwarz B, Sharma L, Roberts L, Peng X, Bermejo S, Leighton I, Massana AC, Farhadian S, Ko A, 656 DelaCruz C, Bosio CM. Severe SARS-CoV-2 infection in humans is defined by a shift in the 657 serum lipidome resulting in dysregulation of eicosanoid immune mediators. medRxiv 2020. 658 659 medRxiv preprint doi: https://doi.org/10.1101/2020.12.04.20242115; this version posted December 7, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission. 22 660 Table 1. Clinical characteristics of BAL donors 661 Severe COVID-19 Healthy controls N 33 25 Women/Men 16/17 16/9 Death (Y/N) 2/31 N/A SARS-CoV-2 detection in BAL (Ct) 25.303 ± 0.906 N/A Age 58.455 ± 3.175 26.130 ± 1.030 Weight (kg) 73.097 ± 2.873 N/A Hospital stay (days) 20.545 ± 2.014 N/A Days before BAL 3.485 ± 0.763 N/A D-Dimers (mg/ml) 1.210 ± 0.129 N/A CRP (mg/l) 20.652 ± 2.525 N/A ALT (U/l) 32.582 ± 2.144 N/A AST (U/l) 35.145 ± 2.978 N/A LDH (U/l) 616.152 ± 41.175 N/A Hemoglobin (g/l) 123.152 ± 2.365 N/A Blood Monocytes (million/ml) 0.327 ± 0.034 N/A Blood Neutrophils (million/ml) 2.855 ± 0.194 N/A Blood Eosinophils (million/ml) 0.026 ± 0.003 N/A Blood Platelets (million/ml) 181.848 ± 13.397 N/A Blood Lymphocytes (million/ml) 1.127 ± 0.127 N/A Platelet to Lymphocyte ratio 223.893 ± 34.934 N/A BAL Neutrophils (million/ml) 24.694 ± 2.152 N/A BAL Eosinophils (million/ml) 0.197 ± 0.077 N/A BAL Lymphocytes (million/ml) 21.830 ± 2.393 N/A 662 Data are presented as the mean ± SEM. ALT, alanine transaminase; AST, aspartate transaminase; 663 CRP, C-reactive protein; LDH, Lactate dehydrogenase; N/A, not available; 664 medRxiv preprint doi: https://doi.org/10.1101/2020.12.04.20242115; this version posted December 7, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission. 23

665 Table 2. Mass transition of the investigated lipid mediators. 666 Retention time LLOQ Compounds ISTD used Q1 > Q3 (min) (fmol)

TXB2-D4 373.10>199.35 4,083

TXB2 TXB2-D4 369.20>195.30 4,090 250

PGE2-D4 355.20>275.35 4,627

6-keto PGF1α PGE2-D4 369.30>163.10 3,493 250 PGF3α PGE2-D4 351.30>193.30 3,954 50 PGE3 PGE2-D4 349.30>269.35 4,160 25 PGF2α PGE2-D4 353.40>309.30 4,412 250 PGE1 PGE2-D4 353.30>273.40 4,770 50 PGE2 PGE2-D4 351.20>271.15 4,627 50 PGD2 PGE2-D4 351.30>271.20 4,847 50 1a,1b-dihomo-PGF2α PGE2-D4 381.20>337.40 5,428 50

RvD2-D5 380.40>175.20 4,685

RvE1 RvD2-D5 349.30>195.20 3,529 250 RvD3 RvD2-D5 375.40>147.20 4,689 250 RvD2 RvD2-D5 375.40>175.20 4,780 500 RvD1 RvD2-D5 375.40>141.10 5,126 500 RvD4 RvD2-D5 375.40>101.05 5,588 250 Maresin 1 RvD2-D5 359.40>177.25 6,684 500 PDX RvD2-D5 359.30>153.15 6,877 25 RvD5 RvD2-D5 359.40>199.25 6,899 250 Maresin 2 RvD2-D5 359.40>221.05 7,433 250

LTC4-D5 629.30>272.20 4,999

EXD4 LTC4-D5 495.30>176.90 4,918 250 EXC4 LTC4-D5 624.30>272.25 5,025 500 LTC4 LTC4-D5 624.30>272.20 5,065 500 LXA4 LTB4-D4 351.50>115.05 5,051 250 EXE4 LTC4-D5 438.30>351.10 5,263 700 LTD4 LTC4-D5 495.30>176.85 5,359 250 LTE4 LTC4-D5 438.30>333.20 5,677 500

LTB4-D4 339.30>197.20 6,742 20-COOH-LTB4 LTB4-D4 365.00>347.25 3,549 250 20-OH-LTB4 LTB4-D4 351.10>195.15 3,670 50 LTB5 LTB4-D4 333.40>195.25 5,992 250 LTB4 LTB4-D4 335.30>195.25 6,765 25 12-oxo-LTB4 LTB4-D4 333.40>179.20 7,399 250 LTB3 LTB4-D4 337.40>195.20 7,523 250

13-HODE-D4 299.10>198.15 9,196 13(S)-HOTrE 13-HODE-D4 292.90>195.25 8,474 250 9-HODE 13-HODE-D4 295.30>171.35 9,237 25 medRxiv preprint doi: https://doi.org/10.1101/2020.12.04.20242115; this version posted December 7, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission. 24

13-HODE 13-HODE-D4 295.50>195.30 9,238 25 13-KODE 13-HODE-D4 295.20>277.30 9,584 25

15-HETE-D8 327.20>226.20 9,503

8,15-DiHETE 15-HETE-D8 335.40>317.30 6,480 250 5(S),15(S)-DiHETE 15-HETE-D8 335.20>255.30 6,630 250 5(S),12(S)-DiHETE 15-HETE-D8 334.99>195.10 7,023 250 12(S)-HHTrE 15-HETE-D8 279.20>179.25 7,699 250 14(15)-DIHET 15-HETE-D8 337.40>207.30 7,712 25 5(S),6(R)-DiHETE 15-HETE-D8 335.00>145.10 8,020 250 18-HEPE 15-HETE-D8 317.30>259.40 8,497 250 15-HEPE 15-HETE-D8 317.40>219.25 8,840 250 12-HEPE 15-HETE-D8 317.30>179.35 9,015 250 15-HETE 15-HETE-D8 319.40>219.30 9,574 250 15-KETE 15-HETE-D8 317.20>113.00 9,707 25 17-HDHA 15-HETE-D8 343.50>281.30 9,719 500 14-HDHA 15-HETE-D8 343.50>281.30 9,878 250 11-HETE 15-HETE-D8 319.30>167.35 9,770 25 8-HETE 15-HETE-D8 319.30>155.25 9,841 50 12-HETE 15-HETE-D8 319.10>179.25 9,924 250 5-HETE 15-HETE-D8 319.30>115.05 10,008 250 17-HDPA 15-HETE-D8 345.20>327.10 10,086 250 15-HETrE 15-HETE-D8 321.50>303.40 10,095 25 14(15)-EET 15-HETE-D8 319.40>219.25 10,623 500

EPA-D5 306.20>262.20 12,128 EPA EPA-D5 301.20>257.30 12,156 25

DHA-D5 332.20>288.20 12,889

DHA DHA-D5 327.50>283.30 12,919 50

DPA(n-3)-D5 334.40>290.35 13,178

DPA(n-3) DPA(n-3)-D5 329.50>285.15 13,191 25

AA-D8 311.20>267.25 13,006

AA AA-D8 303.20>259.30 13,052 250 LA AA-D8 279.40>261.35 13,079 2500 667 668 669 medRxiv preprint doi: https://doi.org/10.1101/2020.12.04.20242115; this version posted December 7, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission. 25 670

671 672 Figure 1. Levels of cyclooxygenase-derived metabolites in the BAL fluids of healthy and severe 673 COVID-19 patients. BALs were obtained and processed as described in Methods. A) Results are from 674 the BALs from 25 healthy subjects and 33 severe COVID-19 patients. P values were obtained by 675 performing a Mann-Whitney test: **** = p < 0.0001. B-E) Correlation tests were performed with 676 COVID-19 patients by using the non-parametric Spearman’s rank. P<0.05 was the threshold of 677 significance. The linear regression is represented with a solid line and the best-fit lines of 95% confidence 678 are represented with dotted lines. 679 680 681 682 683 medRxiv preprint doi: https://doi.org/10.1101/2020.12.04.20242115; this version posted December 7, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission. 26 684

685 686 Figure 2. Comparison of leukotriene and eoxin levels in the BAL fluids of healthy subjects and 687 severe COVID-19 patients. A,D) BALs were obtained and processed as described in methods. Lipids 688 then were extracted and quantitated by LC-MS/MS. Results are from the BALs from 25 healthy subjects 689 and 33 severe COVID19 patients. P values were obtained by performing a Mann-Whitney test: * p < 690 0.05; *** p < 0.001; **** p < 0.0001 B,C,E-G) Correlation tests were performed with severe COVID- 691 19 patients by using the non-parametric Spearman’s rank. P<0.05 was the threshold of significance. The 692 linear regression is represented with a solid line and the best-fit lines of 95% confidence are represented 693 with dotted lines. 694 695 medRxiv preprint doi: https://doi.org/10.1101/2020.12.04.20242115; this version posted December 7, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission. 27

696 Figure 3. Fatty acid levels in the BAL fluids of healthy subjects and severe COVID-19 patients. 697 BALs were obtained and processed as described in Methods. Lipids then were extracted and quantitated 698 by LC-MS/MS. Results are from the BALs from 25 healthy subjects and 33 severe COVID19 patients. 699 P values were obtained by performing a Mann-Whitney test: **** p < 0.0001. 700 medRxiv preprint doi: https://doi.org/10.1101/2020.12.04.20242115; this version posted December 7, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission. 28

701 702 Figure 4. Comparison of 12- and 15-lipoxygenase-derived oxylipins and diHETEs levels in the 703 BALs of healthy subjects and severe COVID-19 patients. A-D) BALs were obtained and processed 704 as described in Methods. Lipids then were extracted and quantitated by LC-MS/MS. Results are from the 705 BALs from 25 healthy subjects and 33 Severe COVID-19 patients. P values were obtained by performing 706 a Mann-Whitney test: **** p < 0.0001. 707 708 709 medRxiv preprint doi: https://doi.org/10.1101/2020.12.04.20242115; this version posted December 7, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission. 29

710 711 712 Figure 5. Comparison of specialized pro-resolving lipid mediator levels in the BAL fluids of healthy 713 subjects and severe COVID19 patients. A) BALs were obtained and processed as described in 714 Methods. Lipids then were extracted and quantitated by LC-MS/MS. Results are from the BALs from 25 715 healthy subjects and 33 severe COVID-19 patients. . P values were obtained by performing a Mann- 716 Whitney test: * p < 0.05; ** p < 0.01; **** p < 0.0001. B-G) Correlation tests were performed with 717 severe COVID-19 patients by using the non-parametric Spearman’s rank. P<0.05 was the threshold of 718 significance. The linear regression is represented with a solid line and the best-fit lines of 95% confidence 719 are represented with dotted lines. medRxiv preprint doi: https://doi.org/10.1101/2020.12.04.20242115; this version posted December 7, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission.

720 Figure 6. Correlation between selected bioactive lipids in the BALs of severe COVID-19 patients. Correlation tests were performed by 721 using the non-parametric Spearman’s rank using the Graphpad Prism Software. P<0.05 was the threshold of significance. Positive correlations 722 are shown using shades of green. medRxiv preprint doi: https://doi.org/10.1101/2020.12.04.20242115; this version posted December 7, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission. 31 723

724 Figure 7. Correlation between bioactive lipids in the BALs of severe COVID-19 patients and 725 clinical parameters. Correlation tests were performed by using the non-parametric Spearman’s rank 726 using the Graphpad Prism Software. P<0.05 was the threshold of significance. Positive correlations are 727 shown using shades of green and negative correlation are shown using shades of red.