key: cord-1006719-xwlpb3db authors: Rodriguez, L.; Pekkarinen, P.; Tadepally, L. K.; Tan, Z.; Rosat Consiglio, C.; Pou, C.; Chen, Y.; Habimana Mugabo, C.; Nguyen Quoc, A.; Nowlan, K.; Strandin, T.; Levanov, L.; Mikes, J.; Wang, J.; Kantele, A.; Hepojoki, J.; Vapalahti, O.; Heinonen, S.; Kekalainen, E.; Brodin, P. title: Systems-level immunomonitoring from acute to recovery phase of severe COVID-19 date: 2020-06-05 journal: nan DOI: 10.1101/2020.06.03.20121582 sha: 697f7dd7a4e176a8888388943f1a27687a0ec534 doc_id: 1006719 cord_uid: xwlpb3db The immune response to SARS-CoV2 is under intense investigation, but not fully understood att this moment. Severe disease is characterized by vigorous inflammatory responses in the lung, often with a sudden onset after 5-7 days of stable disease. Efforts to modulate this hyperinflammation and the associated acute respiratory distress syndrome, rely on the unraveling of the immune cell interactions and cytokines that drive such responses. Systems-level analyses are required to simultaneously capture all immune cell populations and the many protein mediators by which cells communicate. Since every patient analyzed will be captured at different stages of his or her infection, longitudinal monitoring of the immune response is critical. Here we report on a systems-level blood immunomonitoring study of 39 adult patients, hospitalized with severe COVID-19 and followed with up to 14 blood samples from acute to recovery phases of the disease. We describe an IFNg-Eosinophil axis activated prior to lung hyperinflammation and changes in cell-cell coregulation during different stages of the disease. We also map an immune trajectory during recovery that is shared among patients with severe COVID-19. exclusively with mild disease (Brodin, 2020) , while the elderly, over 70 years of 96 age are much more likely to develop severe COVID-19. Males and females are 97 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted June 5, 2020. pregnant women do not seem to be more likely to develop severe disease and 102 this is also true for patients with various forms of immunodeficiency. One likely 103 reason for these observations is that severe disease is associated with exuberant 104 immune responses and a skewed immune regulation against pro-inflammatory 105 responses in pregnancy and T-cell deficiencies in transplan patients make such 106 hyperinflammatory responses less likely. To treat hyperinflammation in severe 107 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted June 5, 2020. . https://doi.org/10.1101/2020.06.03.20121582 doi: medRxiv preprint protein mediators used to orchestrate their response. To this end, we have 109 perfomed systems-level analyses of the immune system in blood from 39 110 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted June 5, 2020. Clinical measurements were taken from acute and recovered patients including 138 body temperature, white blood cell counts and lymphocyte counts. Milder cases 139 of COVID-19 showed lower body temperatures as well faster normalization of 140 body temperatures as compared to severe cases who fluctuated more over time 141 ( Fig. 2A) . The white blood cell (WBC) counts gave a possible correlate to the 142 stage of the infection. High WBC counts are often a reflection of acute 143 inflammation and immune responses and in severe patients we observed more 144 fluctuating levels of WBC over time (Fig. 2B) . Also, there were no signs of 145 secondary bacterial infection in the patients in this cohort. Lymphopenia is one of 146 the common features of COVID-19 and the degree of lymphopenia predict 147 disease severity (Huang et al., 2020). Lymphocyte counts were measured and 148 found that milder cases recuperated their lymphopenia faster than severe cases 149 (Fig. 2C) . This is in line with other previous reports (Lagunas-Rangel, 2020). 150 Plasma protein levels were observed and compared among acute and 151 recovered phases and map immune dynamics of severe COVID-19 ( Fig. 2D-G) . 152 Pro-inflammatory cytokines like IL-6 and IFNg predict disease severity. A 153 decreasing trend is observed in IFNg and IL-6 from early admission to the 154 hospital through recovery during the weeks of the study ( Fig. 2D and 2F , 155 respectively). Similarly, DDX58, the Innate Immune Response Receptor also 156 called RIG-I, and the monocyte chemoattractant protein MCP-3, are elevated 157 during acute disease and decrease during recovery ( Fig. 2E and 2G proportions of 57 immune cell populations over time in the 39 patients (Fig. 3) . 165 We confirm the overrepresentation of Neutrophils over Lymphocytes during acute 166 infection, that is slowly reversed during the recovery phase (Fig. 3) . This is in line 167 with reports suggesting the Neutrophil-to-Lymphocyte ratio (NLR) and degree of 168 lymphopenia are predictive of disease severeity in COVID-19 (Lagunas-Rangel, 169 2020). The plasmablast response is clear and occurs during the first week after 170 admission in these patients (Fig. 3) . The recovery of T-cells after the initial 171 lymphopenia occurs over the following 2-3 weeks and is dominated by CD127 172 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted June 5, 2020. . https://doi.org/10.1101/2020.06.03.20121582 doi: medRxiv preprint expressing effector and central memory CD4 + T-cells, as well as CD57-173 expressing and differentiated memory CD8 + T-cells (Fig. 3) . Also, all Dendritic 174 cell (DC) subsets increased from acute to recovery phases, CD1c + DCs, CD11c + 175 DCs and plasmacytoid DCs (pDC) (Fig. 3) . Also, despite a clear reduction in 176 relative abundance of neutrophils over time, the other granulocyte subsets, 177 Basophils and Eosinophils increased clearly from acute to recovery phases (Fig. 178 3), and both of these were among the most dynamic cell populations during 179 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted June 5, 2020. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted June 5, 2020. cohort we used the same PAGA approach as described above. We find a clear 219 Plasmablasts response early after admission (Fig. 5A) . The CD4 + T-cell 220 binned the samples into four phases from acute disease to recovery phase (Fig. 240 6A) . We find that the first phase (day 0-4) is dominated by an inverse correlation 241 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted June 5, 2020. . https://doi.org/10.1101/2020.06.03.20121582 doi: medRxiv preprint between neutrophils and a number of myeloid and lymphoid cell types, as 242 reflected in the elevated NLR, associated with severe disease (Lagunas-Rangel, 243 2020) (Fig. 6A) . The following phase (day 6-8) is characterized by a strong 244 coordinated Plasmablast, B-cell and abT-cell module, and this is inversely 245 correlated with a strong Treg and CD11c + DC module (Fig. 6A) . From day 9 246 onwards a change is apparent with a shift towards a coregulated module 247 involving Eosinophils, pDCs, CD11c + DCs, with CD8 + T-cells. This module is 248 largely maintained in the recovered patients, possibly reflecting a more normal 249 cell-cell relationship (Fig. 6A) . anti-CoV2 IgG responses (Fig. 6C) . In contrast the Neutrophil-recruiting 267 chemokine CXCL6 are positively associated with anti-CoV2 IgG responses and 268 so was the fraction of circulating Basophils (Fig. 6C) . It is known that basophils 269 are able to bind antigens on their surface and potentiate humoral immune and severe COVID-19 (Fig. 3) , our data collectively suggest that the degree of 272 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted June 5, 2020. . https://doi.org/10.1101/2020.06.03.20121582 doi: medRxiv preprint basophil depletion might influence the efficacy of IgG-responses to SARS-CoV2. 273 through the production of either IL-4 or IL-6, but levels of the latter were found to 275 inversely associated with antibody responses (Fig. 6C) so it more likely that CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted June 5, 2020. . https://doi.org/10.1101/2020.06.03.20121582 doi: medRxiv preprint . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted June 5, 2020. . https://doi.org/10.1101/2020.06.03.20121582 doi: medRxiv preprint Since none of the patients in this cohort were treated with immunomodulatory 285 agents, and have recovered with supportive care alone, we reasoned that a 286 deeper investigation into the immunological changes during recovery from severe We found ten latent factors that explained the variance in the combined dataset 297 (Fig. 7A) , and out of those, latent factor 2 was associated with the transition from 298 acute to recovery phases of disease (Fig. 7B) . There were no clear differences 299 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted June 5, 2020. . https://doi.org/10.1101/2020.06.03.20121582 doi: medRxiv preprint among intensive care unit (ICU) or non-ICU ward patients and the levels of latent 300 factor 2 were highest in the samples taken from recovered patients (Fig. 7B) . To 301 understand the biology of immune recovery we assessed the underlying features 302 contributing to latent factor 2. The plasma proteins changing the most all 303 decreased during recovery and most prominent were IL-6, MCP3, KRT19 304 (Keratin19), CXCL10, AREG and IFNg (Fig. 7C) . Conversely the cells that 305 changed the most during recovery were classical and non-classical monocytes, 306 CD56 dim NK cells, Eosinophils and gdT-cells, all increasing in relative proportions 307 during recovery (Fig. 7D) . This shared, integrated trajectory reveal markers most . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted June 5, 2020. is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted June 5, 2020. is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted June 5, 2020. Homogeneous antibody-based proximity extension assays provide sensitive and 548 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted June 5, 2020. is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted June 5, 2020. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted June 5, 2020. . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted June 5, 2020. . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted June 5, 2020. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted June 5, 2020. . https://doi.org/10.1101/2020.06.03.20121582 doi: medRxiv preprint Vascular Endothelialitis, Thrombosis, and Angiogenesis in Covid-19 A serological assay to detect SARS-CoV-2 seroconversion in humans Multi-Omics Factor Analysis-a 485 framework for unsupervised integration of multi-omics data sets Imbalanced host 489 response to SARS-CoV-2 drives development of COVID-19 The biology of the cell -insights from mass cytometry Why is COVID-19 so mild in children Human immune system variation Human Immune System Is Largely Driven by Non-Heritable Influences A Call for Blood-In Human 499 A global effort to define the 501 human genetics of protective immunity to SARS-CoV-2 infection centrifugation for future use and not intended for this study) were analyzed using 693 a multiplex proximity extension assay Each well contains 696 96 pairs of DNA-labeled antibody probes. Samples were incubated in the presence 697 of proximity antibody pairs tagged with DNA reporter molecules. When the 698 antibody pair bounds to their corresponding antigens, the corresponding DNA tails 699 form an amplicon by proximity extension Detection of anti-SARS-CoV-2 antibody response Antibodies against SARS-CoV-2 were measured using indirect 704 immunofluorescence assay (IFA) and enzyme-linked immunosorbent assay 705 (ELISA) using SARS-CoV-2 receptor-binding domain (RBD) as the antigen. The 706 IFA was conducted as described (Haveri et al., 2020). The RBD ELISA was done 707 following a recently published protocol