key: cord-0991853-x1v5lczw
authors: Levy, Y.; Wiedemann, A.; Hejblum, B. P.; Durand, M.; Lefebvre, C.; Surenaud, M.; Lacabaratz, C.; Perreau, M.; Foucat, E.; Dechenaud, M.; Tisserand, P.; Blengio, F.; Hivert, B.; Gautier, M.; Cervantes, M.; Bachelet, D.; Laouenan, C.; Bouadma, l.; Timsit, J.-F.; Yazdanpanah, Y.; Pantaleo, G.; Hocini, H.; Thiebaut, R.; Study Group, F. C.
title: CD177, a specific marker of neutrophil activation, is a hallmark of COVID-19 severity and death
date: 2020-12-14
journal: nan
DOI: 10.1101/2020.12.12.20246934
sha: 92efe4312fd123825a1b9480d3c0819795c267dd
doc_id: 991853
cord_uid: x1v5lczw

COVID-19 SARS-CoV-2 infection exhibits wide inter-individual clinical variability, from silent infection to severe disease and death. The identification of high-risk patients is a continuing challenge in routine care. We aimed to identify factors that influence clinical worsening. We analyzed 52 cell populations, 71 analytes, and RNA-seq gene expression in the blood of severe patients from the French COVID-19 study upon hospitalization (n = 61). COVID-19 patients showed severe abnormalities of 27 cell populations relative to healthy donors (HDs). Forty-two cytokines, neutrophil chemo-attractants, and inflammatory components were elevated in COVID-19 patients. Supervised gene expression analyses showed differential expression of genes for neutrophil activation, interferon signaling, T- and B-cell receptors, EIF2 signaling, and ICOS-ICOSL pathways in COVID-19 patients. Unsupervised analysis confirmed the prominent role of neutrophil activation, with a high abundance of CD177, a specific neutrophil activation marker. CD177 was the most highly differentially-expressed gene contributing to the clustering of severe patients and its abundance correlated with CD177 protein serum levels. CD177 levels were higher in COVID-19 patients from both the French and confirmatory Swiss cohort (n = 203) than in HDs (P< 0.01) and in ICU than non-ICU patients (P< 0.001), correlating with the time to symptoms onset (P = 0.002). Longitudinal measurements showed sustained levels of serum CD177 to discriminate between patients with the worst prognosis, leading to death, and those who recovered (P = 0.01). These results highlight neutrophil activation as a hallmark of severe disease and CD177 assessment as a reliable prognostic marker for routine care.

The Coronavirus Disease 2019 (COVID-19) pandemic is caused by a newly described highly 63 pathogenic beta coronavirus, Severe Acute Respiratory Syndrome coronavirus 2 (SARS-64

CoV-2) 1,2 . COVID-19 consists of a spectrum of clinical symptoms that range from mild upper 65 respiratory tract disease in most cases to severe disease that affects approximately 15% of However, these responses are dynamic, changing rapidly during the clinical course of the 80 disease, which may explain the high variability of the immunological spectrum described 81 14, 21, 22 . This makes it difficult to deduce a unique profile of the pathophysiology of this 82 infection, which is still undetermined. Furthermore, the high amplitude of the signals 83 generated by the inflammation associated with the disease may hide other pathways that are 84

From a clinical standpoint, clinicians face the daily challenge of predicting worsening patients 86 due to the peculiar clinical course of severe COVID-19, characterized by a sudden 87 deterioration of the clinical condition 7 to 8 days after the onset of symptoms. Determination 88 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 this this version posted December 14, 2020. ; https://doi.org/10.1101/2020.12.12.20246934 doi: medRxiv preprint of the onset of the pathological process once infection has been established in a patient with 89 a severe stage of infection is highly imprecise because of the possible pauci-or 90 asymptomatic phase of the infection, as well as the low specificity of self-limited "flu" illness. 91

We used a systems immunology approach to identify host factors that were significantly 92 associated with the time to illness onset, severity of the disease (ICU or transfer to ICU), and 93 mortality of COVID-19 patients enrolled in the multicentric French COVID cohort 23 . In 94 addition to the depletion of T cells and mobilization of B cells, neutrophil activation, and 95 severe inflammation, we show upregulation of CD177 gene expression and protein levels in 96 the blood of COVID-19 patients in both the COVID-19 cohort and a "confirmatory" cohort, 97

i.e., Swiss cohort, relative to healthy subjects. CD177, a neutrophil activation marker, 98 characterized critically ill patients and marked disease progression and death. Our finding 99 highlights the major role of neutrophil activation through CD177 over-expression in the critical 100 clinical transition point in the trajectory of COVID-19 patients. 101 102 103 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 this this version posted December 14, 2020. ; https://doi.org/10.1101/2020.12.12.20246934 doi: medRxiv preprint

Patient characteristics from the French COVID cohort enrolled in this analysis are shown in 106 We not only confirmed previously reported abnormalities but also revealed new 118 immunological features of COVID-19 patients (supplementary Figure 1) . The COVID-19 119 patients showed a significant reduction in the frequency of total CD3 + T cells and CD8 + T 120 cells relative to HDs, as previously reported 24,25 , that expressed an activated phenotype 121 (CD38 + HLA-DR + ) ( Figure 1A ). The COVID-19 patients also showed lower frequencies of 122 resting memory B cells contrasting with higher frequencies of activated memory B cells and 123 exhausted B cells ( Figure 1B ). As previously reported 10,22 , the proportion of plasmablasts 124 was markedly higher in COVID-19 patients (median [Q1-Q3]: 10.85% [3-23]) than HDs 125 (0.76% [0.4-0.8]) (P < 0.001). Total NK-cell frequencies, more precisely those of the 126 CD56 bright and CD56 dim CD57 -NK-cell subpopulations, were lower than in HDs (P = 0.017, P < 127 0.001, and P = 0.004, respectively) ( Figure 1C) 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 this this version posted December 14, 2020. ; https://doi.org/10.1101/2020.12.12.20246934 doi: medRxiv preprint 7 0.001 compared to HD) ( Figure 1C ). In addition, COVID 19-patients showed significantly 131 smaller classical (CD14 + CD16 -), intermediate (CD14 + CD16 + ), and non-classical (CD14 -132 CD16 + ) monocyte subpopulations than HDs (P = 0.013, P = 0.017, P < 0.001, respectively) 133 ( Figure 1D ). Interestingly, COVID-19 patients tended to exhibit a higher frequency of γδ T 134 cells than HDs (median 10.4% [7.5-16.1] vs 7.3% [6-10] in HDs; P = 0.068) ( Figure 1E ), with 135 a significant proportion of γδ T cells showing higher expression of the activation marker 136 CD16 (P = 0.01) and lower expression of the inhibitory receptor NKG2A (P < 0.001) than 137

HDs ( Figure 1E ). Finally, we observed markedly smaller frequencies of dendritic cells (DCs) 138 for all populations studied (pre-DC, plasmacytoid DC (pDC), and conventional DC (cDC1 and 139 cDC2) in COVID-19 patients than in HDs (P < 0.001, for all comparisons) ( Figure 1F ). interferon MIG/CXCL9, IL-8, IL-9). Interestingly, we found higher levels of midkine, a marker 154 usually not detectable in the serum, which enhances the recruitment and migration of 155 inflammatory cells and contributes to tissue damage 26 . In parallel, Granzyme B and IL-21 156 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 this this version posted December 14, 2020. ; https://doi.org/10.1101/2020.12.12.20246934 doi: medRxiv preprint 8 levels were significantly lower in COVID-19 patients than HDs (P = 0.007 and P = 0.004, 157 respectively) (Supplementary Figure 2) . 158

The comparison of gene abundance in whole blood between COVID-19 patients (n = 44) and 160

HDs (n = 10) showed 4,079 differentially expressed genes (DEG) with an absolute fold 161 change ≥ 1.5, including 1,904 that were upregulated and 2,175 that were downregulated 162 ( Figure 3A ). The main pathways associated with the DEG correspond to the immune 163 response, including neutrophil and interferon signaling, T and B cell receptor responses, 164 metabolism, protein synthesis, and regulators of the EiF2 and mammalian target of 165 rapamycin (mTOR) signaling pathways ( Figure 3A ). Although several of these pathways 166 involved multiple cell types, analysis of the neutrophil pathway showed higher abundance of 167 genes mainly related to neutrophil activation, their interaction with endothelial cells, and 168 migration ( Figure 3B ). Among the most highly expressed genes, this signature included 169 CD177, a specific marker of neutrophil adhesion to the endothelium and transmigration 27 , 170 HP (Haptoglobin), a marker of granulocyte differentiation and released by neutrophils in 171 response to activation 28 , VNN1 (hematopoietic cell trafficking), GPR84 (neutrophil 172 chemotaxis), MMP9 (neutrophil activation and migration), and S100A8 and S100A12 173 (neutrophil recruitment, chemotaxis, and migration). The S100A12 protein is produced 174 predominantly by neutrophils and is involved in inflammation and the upregulation of vascular 175 endothelial cell adhesion molecules 29 ( Figure 3B) . 176

In parallel, we observed a higher abundance of several interferon stimulating genes (ISG) 177 genes was lower ( Figure 3E ). We observed severe dysregulation of T-cell function that 181 involved inhibition of serine/threonine kinase PKCθ signaling (z-score = -4.46) (data not 182 shown), as well as the inducible T-cell co-stimulator/ICOSL axis (z-score = -4.5) 183 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 this this version posted December 14, 2020. ; https://doi.org/10.1101/2020.12.12.20246934 doi: medRxiv preprint (Supplementary Figure 3) . In contrast to the results for T cells, the peripheral expansion of 184

Receptor (BCR) ( Figure 3F and Supplementary Table 2 ). 186

We also observed genes belonging to several crucial pathways and biological processes that 187

had not been previously reported to characterize COVID-19 patients to be underrepresented. Detailed patient characteristics according to group are presented in supplementary Table 1 . 200

Among a large set of clinical and biological characteristics, the analysis showed the 201 differential clustering to not be explained by the severity of the disease. Indeed, the median 202 Sequential Organ Failure Assessment (SOFA) score and Simplified Acute Physiology Score 203 (SAPS2), which include a large number of physiological variables 32,33 and evaluate the 204 clinical severity of the disease (a high score is associated with a worse prognosis) of the 205 COVID-19 patients, were 6 [4-7] and 36 [28-53], respectively, with no significant differences 206 between groups. Nevertheless, we observed a significant difference in the median time to 207 symptoms onset at admission, which ranged from 7 [6 -11] days for patients in group 1 to 11 208

[10-14] and 13 [9-14] days for patients in groups 2 and 3, respectively (P = 0.04, Kruskal-209 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 this this version posted December 14, 2020. ; https://doi.org/10.1101/2020.12.12.20246934 doi: medRxiv preprint Wallis test). Finally, group 1 which was the closest to HDs in terms of gene profile, consisted 210 of patients in the early days of the disease (Supplementary Table 1 ). 211

Analysis of the genes contributing to the differences between groups confirmed and 212 extended the findings described above ( (81%), and EiF2 signaling (92%) were also highly represented. The interferon signaling 219 genes, such as IFI44L, IFIT2, and IRF8, a regulator of type I Interferon (α, β), were 220 significantly more abundant at earlier stages (in patients from group 2) and tended to be less 221 abundant in group 3, at more advanced stages of the disease. Finally, the abundance of 222 genes belonging to T-cell pathways (TCR, iCOS-iCOSL signaling) or mTOR and EIF2 223 signaling was lower in group 3, that is to say, those who were analyzed after a longer time to 224 symptoms onset. The findings described above highlight the heterogeneity of COVID-19 225

Integrative analysis of all biomarkers reveals the major contribution of CD177 in the 227 clinical outcome of COVID-19 patients. 228

We performed an integrative analysis using all available data to disentangle the relative 229 contribution of the various markers at the scale of every patient. We thus pooled the data for 230 29,302 genes from whole blood RNA-seq, cell phenotypes (52 types), and cytokines (71 231 analytes) using the recently described MOFA approach 34 , which is a statistical framework for 232 dimension reduction adapted to the multi-omics context. The data are reduced to 233 components that are linear combinations of variables explaining inter-patient variability 234 across the three biological measurement modalities. The first component, that we called our 235 integrative score, discriminated between the three groups of COVID-19 patients and HDs 236 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 this this version posted December 14, 2020. ; https://doi.org/10.1101/2020.12.12.20246934 doi: medRxiv preprint ( Figure 5A ), although it only explained a portion of the variability within each of the three 237 types of markers (14% of gene expression, 14% of cell phenotypes, and 5% of cytokines). 238

The main contributors for the cell phenotype were the significantly lower frequency of cDC2 239 and T cells and, marginally, the higher number of plasmablasts and CD16 + γδ T cells in 240 COVID-19 patients ( Figure 5B ). The contribution of soluble factors was marked by higher 241 levels of soluble CD163 (sCD163), a marker of polarized M2 macrophages involved in tissue 242 repair 35 , in more advanced COVID-19 groups ( Figure 5C ). Indeed, CD163 gene expression 243 was also significantly higher in COVID-19 patients than HDs (log2 fold change = +1.55; FDR 244 = 4.79 10 -2 ). sCD163 has also been reported to be a marker of disease severity in critically ill 245 patients with various inflammatory or infectious conditions 36 . Interestingly, the genes that 246

contribute the most to the synthesis of this factor were part of the neutrophil module (CD177, 247 

Given the contribution of the neutrophil activation pathway in the clustering of COVID-19 256 patients, we sought neutrophil-activation features that could act as possible reliable markers 257 of disease evolution. We focused on CD177 because: i) it is a neutrophil-specific marker 258 representative of neutrophil activation, ii) it was the most highly differentially expressed gene 259 in patients, and iii) the protein can be measured in the serum, making its use as a marker 260 clinically applicable. Thus, we used an ELISA to quantify CD177 in the serum of 203 COVID-261 19 patients (115 patients from the French cohort and 88 patients from the Swiss COVID-19 262 cohort that we used as "a confirmatory" cohort, patient characteristics are described in Table  263 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 this this version posted December 14, 2020. ; https://doi.org/10.1101/2020.12.12.20246934 doi: medRxiv preprint 2), 21% of whom the measurements were repeated (from 2 to 10 measurements per 264 individual). First, we confirmed the significantly higher median serum protein level in the 265 global cohort of COVID-19 patients (4.5; [2.2-7.4]) relative to that of 16 HDs (2.2 [0.9-4.2]) (P 266 = 0.015, Wilcoxon test) ( Figure 6A ). Second, we found a robust agreement between CD177 267 gene expression measured by RNA-seq and CD177 protein levels measured by ELISA 268 (intraclass correlation coefficient 0.88), ( Figure 6B ). 269

Then, we examined the association of clinical characteristics and outcomes with serum 270 CD177 concentration at the time of admission. The serum CD177 concentration was 271 positively associated with the time to symptoms onset (P = 0.0026) ( Figure 6C ) and was 272 higher for patients hospitalized in an ICU (6.0 ng/ml [3.5-9.4] vs 3.3 ng/ml [1.5-5.6], P < 273 0.001) ( Figure 6D ). The association between serum CD177 levels and hospitalization in an 274 ICU was independent of the usual risk factors, such as age, sex, chronic cardiac or 275 pulmonary diseases, or diabetes (multivariable logistic regression, adjusted odds ratio 1.14 276 per unit increase, P < 0.001). We observed a trend towards a positive association with the 277 SOFA and SAPS2 risk scores that was not statistically significant (P = 0.17 and P = 0.074, 278 respectively) (supplementary Figure 5A and B). CD177 levels were not associated with other 279

conditions that contribute to a high risk of severe disease, such as diabetes (P = 0.632), 280 chronic cardiac disease (P = 0.833), chronic pulmonary disease (P = 0.478), or age of the 281 patient (P = 0.83) (data not shown). 282

We then examined the dynamics of the CD177 concentration in 172 COVID-19 patients, with 283 longitudinal serum samples, using all available measurements ( Figure 6E ). At the first 284 measurement, the average concentration of CD177 was not significantly different between 285 the patients who died and those who recovered (5.93 vs 5.06, P= 0.26, Wald test). However, 286 CD177 levels decreased significantly in those who recovered (-0.22 ng/mL/day, 95% CI -287 0.307; -0.139), whereas it was stable in those who died (+0.10 ng/mL/day, 95% CI +0.014; 288 +0.192) and the difference between the two groups was statistically significant (Wald test for 289 interaction, P = 0.010). These results show that the stability of CD177 protein levels in severe 290 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 this this version posted December 14, 2020. ; https://doi.org/10.1101/2020.12.12.20246934 doi: medRxiv preprint 13 COVID-19 patients during the course of the disease is a hallmark of a worse prognosis, 291 leading to death. 292 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 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 this this version posted December 14, 2020. ; https://doi.org/10.1101/2020.12.12.20246934 doi: medRxiv preprint

The neutrophil activation signature is a specific feature of the homing of activated neutrophils 319 toward infected lung tissue in acute lung injury 50 , followed by the initiation of aggressive 320 responses and the release of neutrophil extracellular traps (NETs), leading to an oxidative 321 burst and the initiation of thrombus formation 51 . Previous studies have reported elevated 322 levels of circulating NETs in COVID-19 disease 52 . Consistent with this finding and extending 323 these data, we showed the differential expression of NET-related genes 41,47,48 (S100A8, 324 S100A9, and S100A12), confirming the recently described elevated expression of 325 calprotectin (heterodimer of S100A8 and A9) in severe COVID-19 patients 13 . The 326 association of neutrophil activation signature with COVID-19 severity has also been 327 . The frequency of γδ T cells, a subpopulation of CD3 + T cells that were first described in 345 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 this this version posted December 14, 2020. ; https://doi.org/10.1101/2020.12.12.20246934 doi: medRxiv preprint the lung 56 and that play critical roles in anti-viral immune responses, tissue healing, and 346 epithelial cell maintenance 57 , was elevated and they expressed an activation marker (CD16) 347 and low levels of the inhibitory receptor NKG2A, suggesting possible killing capacity. In 348 accordance with our observation that EiF2 signaling is significantly inhibited in COVID-19 349 patients, recent studies have shown that coronaviruses encode ISR antagonists, which act reported to be the most highly represented in S-specific SARS-CoV-2 sequences 62 . We also 364 found enrichment of VH3-33, previously described in a set of clonally related anti-SARS-365

CoV-2 receptor-binding domain antibodies 63 . 366

Globally, these results show that the defense against SARS-CoV-2 following pathogen 367 recognition triggers a fine-tuned program that not only includes the production of antiviral 368 (Interferon signaling) and pro-inflammatory cytokines but also signals the cessation of the 369 response and a strong disturbance of adaptive immunity. 370

The same pathways (immune and stress responses through EiF2 signaling, neutrophil and 371

Interferon signaling, T-and B-cell receptor responses, and mTOR pathways) contributed to 372 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 this this version posted December 14, 2020. ; https://doi.org/10.1101/2020.12.12.20246934 doi: medRxiv preprint the ability to discriminate between three groups of severe COVID-19 patients in an 373 unsupervised analysis. One limitation of our study was that we did not repeat the RNA-seq 374 analyses in these specific groups of patients. Nonetheless, it is noteworthy that these groups 375 prognosis. In addition, they suggest that therapies aiming to control neutrophil activation and 392 chemotaxis should be considered for the treatment of hospitalized patients. 393

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 this this version posted December 14, 2020. ; https://doi.org/10.1101/2020.12.12.20246934 doi: medRxiv preprint

We enrolled a subgroup of COVID19 patients of the prospective French COVID cohort 396 (registered at clinicaltrials.gov NCT04262921) in this immunological study which is part of the 397 cohort main objectives. Ethics approval was given on February 5th by the French Ethics 398 Table 3 . 411

In total, 71 analytes were quantified in heat-inactivated serum samples by multiplex magnetic 413 bead assays or ELISA. Serum samples from five healthy donors were also assayed as 414 controls. The following kits were used according to the manufacturers' recommendations: 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 this this version posted December 14, 2020. ; https://doi.org/10.1101/2020.12.12.20246934 doi: medRxiv preprint 

Immune-cell phenotyping was performed using an LSR Fortessa 4-laser (488, 640, 561, and 433 405 nm) flow cytometer (BD Biosciences) and Diva software version 6.2. FlowJo software 434 version 9.9.6 (Tree Star Inc.) was used for data analysis. CD4 + and CD8 + T cells were 435 analyzed for CD45RA and CCR7 expression to identify the naive, memory, and effector cell 436 subsets for co-expression of activation (HLA-DR and CD38) and exhaustion/senescence 437 (CD57and PD1) markers. CD19 + B-cell subsets were analyzed for the markers CD21 and 438 CD27. ASC (plasmablasts) were identified as CD19 + cells expressing CD38 and CD27. We 439 used CD16, CD56, and Ki57 to identify NK-cell subsets. γδ T cells were identified using an 440 anti-TCR γδ antibody. HLA-DR, CD33, CD45RA, CD123, CD141, and CD1c were used to 441 identify dendritic cell (DC) subsets, as previously described 68 442

Total RNA was purified from whole blood using the Tempus™ Spin RNA Isolation Kit 444 (ThermoFisher Scientific). RNA was quantified using the Quant-iT RiboGreen RNA Assay Kit 445 (Thermo Fisher Scientific) and quality control performed on a Bioanalyzer (Agilent). Globin 446 mRNA was depleted using Globinclear Kit (Invitrogen) prior to mRNA library preparation with 447 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 analysis, a Z-score ≥ 2 was defined as the threshold for significant activation, whereas a Z-468 score ≤ −2 was defined as the threshold for significant inhibition. 469

The integrative analysis of the three types of biological data (RNA-Seq, cell phenotypes, perpetuity. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in 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 this this version posted December 14, 2020. ; https://doi.org/10.1101/2020.12.12.20246934 doi: medRxiv preprint 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 this this version posted December 14, 2020. ; https://doi.org/10.1101/2020.12.12.20246934 doi: medRxiv preprint to IFN and interleukin responses, respectively. E Main TCRV T-cell repertoire DEG F Main 711 B-cell IGHV repertoire DEG. Red symbols represent overabundant genes in COVID-19 712 relative to HD, green symbols represent underabundant genes. 713 The integrative score corresponds to the first factor of the analysis and allows the ordering of 728 individuals along an axis centered at 0. Individuals with an opposite sign for the factor 729 therefore have opposite characteristics. 730 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 this this version posted December 14, 2020. ; https://doi.org/10.1101/2020.12.12.20246934 doi: medRxiv preprint 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 this this version posted December 14, 2020. ; https://doi.org/10.1101/2020.12.12.20246934 doi: medRxiv preprint 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 this this version posted December 14, 2020. ; https://doi.org/10.1101/2020.12.12.20246934 doi: medRxiv preprint 

COVID-19 disease progression

Elevated Calprotectin and Abnormal Myeloid Cell Subsets Discriminate 545 Severe from Mild COVID-19

Longitudinal analyses reveal immunological misfiring in severe COVID-547 19

A Dynamic Immune Response Shapes COVID-19 Progression

Complex Immune Dysregulation in COVID-19

Patients with Severe Respiratory Failure

Clinical features of patients infected with 2019 novel coronavirus in

Heightened Innate Immune Responses in the Respiratory Tract of

COVID-19 Patients

The clinical implication of dynamic neutrophil to lymphocyte ratio and D-558 dimer in COVID-19: A retrospective study in Suzhou China

Epidemiological and clinical characteristics of 99 cases of 2019 novel 561 coronavirus pneumonia in Wuhan, China: a descriptive study

Systems biological assessment of immunity to mild versus 564 severe COVID-19 infection in humans

Immune Alterations in a Patient with SARS-CoV-2-Related Acute

Respiratory Distress Syndrome

Kawasaki-like disease: emerging complication during the 623 COVID-19 pandemic

A systems analysis identifies 625 a feedforward inflammatory circuit leading to lethal influenza infection

Excessive neutrophils and neutrophil extracellular traps contribute 628 to acute lung injury of influenza pneumonitis

Gene Expression Signatures Diagnose Influenza and Other 630

Symptomatic Respiratory Viral Infections in Humans

Acute Respiratory Distress Syndrome Neutrophils Have a Distinct 633 Phenotype and Are Resistant to Phosphoinositide 3-Kinase Inhibition

Tissue factor-positive neutrophils bind to injured endothelial wall 636 and initiate thrombus formation

Targeting potential drivers of COVID-19: Neutrophil extracellular 638 traps

Interleukin-32: a cytokine 641 and inducer of TNFalpha

The IL-1 family 643 cytokines and receptors in autoimmune diseases

Resident pulmonary lymphocytes expressing the 645 gamma/delta T-cell receptor

Lung-resident gammadelta T cells and their roles in lung diseases

Middle East Respiratory Coronavirus Accessory Protein 4a Inhibits 649 PKR-Mediated Antiviral Stress Responses

Inhibition of the integrated stress response by viral proteins that 651 block p-eIF2-eIF2B association

Sources of Type I Interferons in Infectious Immunity: Plasmacytoid 653

Dendritic Cells Not Always in the Driver's Seat

Regulatory effects of mTORC2 complexes in type I IFN signaling and in 655 the generation of IFN responses

Potent neutralizing antibodies from COVID-19 patients define 658 multiple targets of vulnerability

Structures of Human Antibodies Bound to SARS-CoV-2 Spike

Reveal Common Epitopes and Recurrent Features of Antibodies

A dynamic COVID-19 immune signature includes associations with 663 poor prognosis

Autoantibodies against type I IFNs in patients with life-threatening

Corticosteroids in COVID-19 ARDS

Association Between Administration of Systemic Corticosteroids 669 and Mortality Among Critically Ill Patients With COVID-19

Mapping the human DC lineage through the integration of high-671 dimensional techniques

Salmon provides fast 673 and bias-aware quantification of transcript expression

Mapping identifiers for the 676 integration of genomic datasets with the R/Bioconductor package biomaRt

the linear regression line and the grey area the 95% prediction confidence interval. C 738 Association between CD177 serum concentration and time from symptoms onset

This association was tested using Spearman correlation tests. The blue line represents the 740 linear regression line and the grey area the 95% confidence interval. D Measurement of 741 CD177 serum concentration in patients hospitalized in an intensive care unit (ICU) or not (n = 742 196). Wilcoxon rank tests were used

The lower and upper boundaries of the box represent the 25% and 75% percentiles

Change of CD177 concentration over time according to the occurrence of death for 172

COVID-19 patients and a total of 248 measurements. Predictions were calculated using a 746 mixed effect models for longitudinal data

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 this this version posted December 14

associated with CD177 protein concentration were performed using non parametric Wilcoxon 475 test or Spearman correlation coefficient when appropriate. To look at the independent 476 association of CD177 with ICU, a logistic regression for the prediction of hospitalization in 477 ICU adjusted for age, sex, chronic cardiac disease, chronic pulmonary disease, diabetes was 478 fitted. The analysis of repeated measurements of CD177 over time was done by using a 479 linear mixed effect model adjusted for time from hospitalization and an interaction with 480 survival outcome (death or recovery). The model included a random intercept and a random 481 slope with an unstructured matrix for variance parameters. Predictions of marginal 482 trajectories were performed. All analyses, if not stated otherwise, were performed using R 483 software version 3.6.3 (R Core Team (2020)). R: A language and environment for statistical 484 computing. R Foundation for Statistical Computing, Vienna, Austria. URL: https://www.R-485 project.org/ 486

RNA sequencing data that support the findings of this study will be deposited in the Gene 488Expression Omnibus (GEO) repository. Other data will be provided as Source data files. 489 490 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 

None of the authors has any conflict of interest to declare. 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 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 this this version posted December 14, 2020. ; https://doi.org/10.1101/2020.12.12.20246934 doi: medRxiv preprint 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 this this version posted December 14, 2020. ; https://doi.org/10.1101/2020.12.12.20246934 doi: medRxiv preprint 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 this this version posted December 14, 2020. ; https://doi.org/10.1101/2020.12.12.20246934 doi: medRxiv preprint 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 perpetuity.preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint inThe copyright holder for this this version posted December 14, 2020. ; https://doi.org/10.1101/2020.12.12.20246934 doi: medRxiv preprint