key: cord-0694986-iuctzaw3
authors: Kvedaraite, E.; Hertwig, L.; Sinha, I.; Ponzetta, A.; Hed Myrberg, I.; Lourda, M.; Dzidic, M.; Akber, M.; Klingstrom, J.; Folkesson, E.; Muvva, R.; Chen, P.; Brighenti, S.; Norrby-Teglund, A.; Eriksson, L. I.; Rooyackers, O.; Aleman, S.; Stralin, K.; Ljunggren, H.-G.; Ginhoux, F.; Bjorkstrom, N.; Henter, J.-I.; Svensson, M.
title: Perturbations in the mononuclear phagocyte landscape associated with COVID-19 disease severity
date: 2020-08-31
journal: nan
DOI: 10.1101/2020.08.25.20181404
sha: 25372462ae5ad3736e43893a9600a61f3b234502
doc_id: 694986
cord_uid: iuctzaw3

Monocytes and dendritic cells are crucial mediators of innate and adaptive immune responses during viral infection, but misdirected responses by these cells might contribute to immunopathology. A comprehensive map of the mononuclear phagocyte (MNP) landscape during SARS-CoV-2 infection and concomitant COVID-19 disease is lacking. We performed 25-color flow cytometry-analysis focusing on MNP lineages in SARS-CoV-2 infected patients with moderate and severe COVID-19. While redistribution of monocytes towards intermediate subset and decrease in circulating DCs occurred in response to infection, severe disease associated with appearance of Mo-MDSC-like cells and a higher frequency of pre-DC2. Furthermore, phenotypic alterations in MNPs, and their late precursors, were cell-lineage specific and in select cases associated with severe disease. Finally, unsupervised analysis revealed that the MNP profile, alone, could identify a cluster of COVID-19 non-survivors. This study provides a reference for the MNP response to clinical SARS-CoV-2 infection and unravel myeloid dysregulation associated with severe COVID-19.

The current COVID-19 pandemic has claimed more than 800.000 lives worldwide during its first 8 months in 2020. Clinical presentation of COVID-19 can vary from asymptomatic to lifethreatening acute respiratory distress syndrome and multiple organ failure. Additional potentially lethal complications include profound coagulation abnormalities associated with systemic thrombogenicity combined with a hyperinflammatory state 1, 2 . Despite a steadily growing body of information regarding the host immune response to SARS-CoV-2 infection and the pathophysiology behind COVID-19, it is still unclear why certain patients enter the detrimental courses of the disease while others merely present with mild or no symptoms 2 .

Thus, there is an urgent need to characterize in depth the immunological and inflammatory aspects of SARS-CoV-2 infection and ensuing COVID-19 disease.

Mononuclear phagocytes (MNPs) in peripheral blood comprise of dendritic cells (DCs) and monocytes, with central roles in orchestrating induction of both innate and adaptive immune responses. Monocytes and DCs are at the front line during an infection, able to recognize, process, and present antigens to immune cells and at the same time produce cytokines and regulate immune responses 3, 4 . These cells are divided in subsets with respect to their ontogeny and specialized functions. The monocytes are rapidly recruited to the site of infection, and can be divided into three major subsets; i.e., classical, intermediate, and non-classical [5] [6] [7] .

Monocytes provide both proinflammatory and/or resolving activities as a supplement to other immune cells present at the site of infection 8, 9 . DCs are also a heterogeneous group of cells divided into conventional DCs (cDCs) and type I interferon-(IFN-) producing plasmacytoid DCs (pDCs) 10, 11 . Among cDCs, cDC1s are specialized in cross-presenting antigens to CD8+ T cells while cDC2s initiate T helper cell responses, both essential for successful viral clearance 12 . cDC1s constitute a discrete population of cells identified based on highly specific markers such as CLEC9A. cDC2s are more heterogeneous as evident by the functionally distinct CD5+ DC2 and CD5-DC3 subsets, among which inflammatory DC3s, positive for the classically monocyte restricted marker CD14 are found [13] [14] [15] [16] . In addition, circulating DC precursors, described as pre-DC 17 or AS DC 18 , have recently been discovered.

Except for having a central role in the defence against infections, misdirected MNP responses might also contribute to immunopathology. Indeed, the importance of monocytes, monocytederived cells and DCs in COVID-19 pathogenesis is emerging [19] [20] [21] [22] [23] [24] [25] [26] [27] is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted August 31, 2020. . https://doi.org/10.1101/2020.08.25.20181404 doi: medRxiv preprint

To study the immune profile of MNPs in SARS-CoV-2 infection, hospitalized patients with ongoing moderate and severe COVID-19 disease were recruited based on strict inclusion and exclusion criteria early in their disease course (Fig. 1A , Supp. Material and Methods, Supp. Table 2 ). There was no difference in time from onset of symptoms until hospital admission, nor until study sampling, in the two severity groups (Supp. Table 2 ). An integrative analysis approach was taken where the detailed phenotype of distinct MNP lineages was analyzed in relation to the clinical disease status, combining supervised and unsupervised strategies, with the aim to comprehensively chart the response of circulating MNPs in response to SARS-CoV-2 infection and severity of COVID-19 disease (Fig. 1B) . As immune homeostasis is significantly disrupted in COVID-19, canonical lineage markers, such as HLA-DR used to identify discrete subsets of MNPs, are altered in expression 19 . As a point of departure, we designed an MNP-focused 25-color flow cytometry panel (Fig. 1C) . After exclusion of granulocytes (CD15+), NK cells (CD7+), ILCs (CD7+), B cells (CD19+), T cells (CD3+), circulating early progenitors (CD34+), basophils (FCER1A+HLA-DR-), and plasma cells (CD38+CD45RA+CD19low), the total MNPs were identified among the cells defined as CD88+ and/or CD116+ (Fig. 1E, D) . The MNP identification-strategy allowed clear visualization of DC1, pre-DC and pre-DC2, cDC2, CD5+ DC2, and three subsets of DC3 (CD163-CD14-, CD163-CD14+, CD163+CD14+) ( Fig. 2A) . This approach was further validated by assessing DC subset-specific markers (Fig. 2B) . Next, analysis of absolute numbers of the identified DC subsets revealed that all subsets were decreased in SARS-CoV-2 infected patients (Fig. 2C) . In relation to the drastic loss in patients compared to controls, only minor differences where noted between severity groups with pDC and pre-DC loss, if anything, even more profound in the severe COVID-19 patients (Fig. 2D) , and with no major differences in-between the different DC sublineages (Fig. 2E) . Thus, declining circulating cDCs, their late progenitors, and pDCs is a response feature in SARS-CoV-2 infection.

Next, a detailed phenotypic mapping of each DC subset was performed. As a starting-point, is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted August 31, 2020. . https://doi.org/10.1101/2020.08.25.20181404 doi: medRxiv preprint development (Fig. 3M) . Finally, levels of the maturation markers HLA-DR and CD86, the ectoenzyme CD38, the inhibitory receptor CD200R, the GM-CSF receptor (CD116), the IL-6R (CD126), thrombomodulin (CD141), and a possible SARS-CoV-2 spike protein receptor CD147 were assessed across the DC subsets (Fig. 3Q ). Higher levels of CD38 were detected specifically in moderate COVID-19 patients, while the maturation markers HLA-DR and CD86

were decreased across all DC subsets (with an exception for DC1s) in severe disease (Fig. 3Q ).

Finally, in severe disease, all subsets of the cDC2 linage presented with lower levels of CD200R while DC1s specifically downregulated the IL-6R (Fig. 3Q) . Taken together, this analysis suggests lineage specific changes and affected developmental phenotype in DC subsets of COVID-19 patients that occur either in response to SARS-CoV-2 infection or specifically in patients with severe COVID-19 disease.

Next, we focused on monocytes and their subsets (Fig. 4A ). As expected, three major populations of monocytes were identified based on CD14 and CD16 expression; i.e., CD14+CD16-cMonos, CD14+CD16+ iMonos, and CD14lowCD16++ ncMonos (Fig. 4A ).

UMAP-analysis verified the present gating strategy and demonstrated separation between monocytes and DCs (Fig. 4A ). This was also the case for DC3s, which in many aspects are similar to monocytes in phenotype, especially due to their expression of CD14 (Fig. 4A ). In accordance with previous publications, iMonos expressed the highest levels of HLA-DR while cMonos displayed relatively high levels of CCR2 (Fig. 4B) . Assessment of absolute counts of monocyte populations in COVID-19 patients and controls revealed no change in cMono numbers, significantly increased iMono numbers in response to infection, especially in moderate patients, and declining numbers of ncMonos in all patients (Fig. 4C) . A similar pattern was observed when monocyte subset frequencies were assessed (Fig. 4D) . While increased . CC-BY-NC-ND 4.0 International license It is made available under a perpetuity.

is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted August 31, 2020. . https://doi.org/10.1101/2020.08.25.20181404 doi: medRxiv preprint CD38 expression across monocyte subsets was a general feature unrelated to disease severity, higher levels of CCR2 on ncMonos and iMonos, as well as lower levels of HLA-DR and CD86 with increased expression of CD163 in all monocyte subsets were found in severe COVID-19 ( Fig. 4E) . Furthermore, thrombomodulin (CD141) expression was increased in severe COVID-19 in ncMonos and iMonos while it was also increased in moderate patients in cMonos (Fig.   4E ). In summary, this shows redistribution and an immature phenotype of the monocyte compartment in COVID-19 that is linked to disease severity.

To analyze the global monocyte landscape in an unbiased manner and integrate the contribution of all tested markers, we next re-clustered all monocytes revealing high heterogeneity of cells that fell into the cMono category (Fig. 4F) . Indeed, cMonos stratified into 15 Phenograph clusters whereas only one or two clusters were found for iMono and ncMono (Fig. 4F , Supp. In addition, the Mo-MDSC-like cluster 9 expressed high levels of c-KIT and CD163 (Fig. 4H) . is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted August 31, 2020. . https://doi.org/10.1101/2020.08.25.20181404 doi: medRxiv preprint Furthermore, a positive correlation was found between cMono Phenograph clusters specific to severe COVID-19 disease (clusters 2, 9, 10, 18) and the soluble immunosuppressive factors IL-10, TGF-b, and VEGFA as well as with AREG (Fig. 4I , J), known to be involved in tolerance and tissue repair 31 . In contrast, these clusters showed a negative association with stem cell factors important for myeloid cell development and differentiation (Fig. 4I) 

The patients all presented with elevated inflammatory markers but clinical hyperinflammation and signs of coagulation disturbances were most pronounced in the severe COVID-19 patients is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted August 31, 2020. . https://doi.org/10.1101/2020.08.25.20181404 doi: medRxiv preprint PCR+, measured at study sampling), while the majority of seroconverted (SARS-CoV-2 IgG+, measured at study sampling) patients were critically ill (74%, 14/19) (Fig. 5C ). Significantly higher numbers of pre-DC2 as well as of inflammatory DC3 (CD163+CD14+ DC3 and CD163+CD14-DC3) were found in patients without ongoing viremia (serum PCR-) and the same DC3 populations were elevated in seroconverted patients (Fig. 5C, Supp. Fig. 3B ).

Following a similar trend, ncMonos were also present at higher numbers in seroconverted patients (Fig. 5C, Supp. Fig. 3B ). As expected, patients that had seroconverted (IgG+) were sampled slightly later in their disease course (Fig. 5D ). Absolute numbers of the inflammatory Finally, to get an overall view of the MNP landscape in relation to COVID-19 disease severity, we performed a principal component analysis of the 108 MNP parameters measured. Here, COVID-19 patients clearly separated from healthy controls and within the patient groups, moderate and severe additionally clustered apart (Fig. 5F ). The MNP parameters that contributed highly to separate the groups were: higher DC counts and Mono cluster 4 (non- is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted August 31, 2020. . https://doi.org/10.1101/2020.08.25.20181404 doi: medRxiv preprint DC progenitors, and higher frequencies of pre-DC2s in severe patients (Fig. 5F , Supp. Fig. 3D ).

Dimensionality reduction using UMAP and Phenograph was performed using the same MNP variables, revealing 3 major clusters, where cluster 1 corresponded to healthy controls, cluster 2 corresponded to all moderate and some severe COVID-19 patients, and cluster 3 corresponded to severe COVID-19 patients only (Fig. 5G ). All patients with fatal outcome were found in cluster 3 and these further had a high oxygen requirement, showed higher levels of LDH (peak measurement), high neutrophil/lymphocyte ratio, and elevated ferritin (measured within 24 hours from MNP profiling) (Fig. 5G , Supp. Fig. 3E ). To visualize the key MNP profile components specific for each cluster, 44 of the most significant MNP parameters analyzed so far were selected and subjected to hierarchical clustering. Monocyte clusters 9 and 10, corresponding to Mo-MDSC-like cells, were specific to patient cluster 3 with the nonsurvivors. These parameters also clustered together with CD163 expression on all monocyte subsets, CCR2 expression on iMonos, and c-KIT in total pre-DCs. For moderate disease, higher expression of CD38 in all DC lineages, CD141 on cMono, and higher levels of monocyte cluster 18 (iMono, the only cluster expressing high levels of CD200R among the patient specific clusters), were the determining parameters. A set of markers were lost in severe patients including HLA-DR and CD86 in both monocytes and DCs, and CD200R in DC subsets ( Fig.   5H ). Moreover, the CD200R decrease in non-survivors was restricted to the cDC2 lineage, with the most significant differences seen in progenitors (pre-DCs and pre-DC2) (Supp. Fig. 3F ).

Finally, to address if this was entirely dependent on severity status rather than outcome, CD200R levels were assessed within severe patients only. A similar pattern was observed, suggesting that CD200R is diminished in the cDC2 lineage and is related not only to disease severity but also to fatal outcome (Supp. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted August 31, 2020. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted August 31, 2020. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted August 31, 2020. . https://doi.org/10.1101/2020.08.25.20181404 doi: medRxiv preprint myeloid-derived cells 39,40 . Moreover, it has been suggested that CD200R signalling plays an important role in T cell priming during viral infection 41 , but remains to be studied in further detail in COVID19. Our findings suggest that in severe COVID-19, DCs, and in particular the cDC2 lineage, are immature with alterations in the developmental phenotype, lack upregulation of the activation marker CD38 that is seen in moderate patients, and loose the inhibitory receptor CD200R, all possibly contributing to inefficient regulation of pro-inflammatory conditions in severe COVID-19.

In line with previous reports, we observed an expansion of intermediate CD14+CD16+ monocytes in COVID-19 patients 21, 42 . This appeared to be a general feature of clinical SARS-CoV-2 infection. In addition, all monocyte subsets responded with increased CD38 expression, that in contrast to DCs, was independent on disease severity. It is plausible that iMonos, is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted August 31, 2020. . https://doi.org/10.1101/2020.08.25.20181404 doi: medRxiv preprint being better equipped to clear virus through the IFN-response, contribute to the immunopathogenesis of COVID-19. In support of this, we found a negative correlation between Mo-MDSC and the growth factors FLT3L and SCF, which are required for myeloid differentiation. This coincided with a positive correlation between Mo-MDSC and HGF, known to mediate Mo-MDSC expansion 32 (Fig. 5I ). This would fit with the model of sequential transition, suggesting that a certain proportion of cMonos eventually become iMonos, followed by their development into ncMonos 45 . While the elevated CCR2 on ncMonos found here may present an explanation for their more pronounced disappearance from the circulation in severe COVID-19, a sequential transition model might also explain the decrease of ncMonos detected in COVID-19 patients by us (Fig. 4C , D) and others 20, 27, 35, 42 .

Hyperinflammation in COVID-19 is characterized by high systemic cytokine levels combined with coagulation abnormalities and thromboembolism [46] [47] [48] , also indicated in our study by increased levels of fibrinogen and D-dimer. Increased levels of thrombomodulin (CD141), a cofactor for thrombin reducing blood coagulation, possibly possessing an anticoagulant potential, were observed on cMonos in both moderate and severe disease, and on iMonos and ncMonos in severe COVID-19. While the role of the extracellular metalloprotease inducer CD147 that was discussed to be a viral spike protein receptor remains to be determined, it is also a potential host factor in infection-mediated coagulation 49 . In the present study, elevated levels of CD147 on iMonos and ncMonos was a general feature of all patients, while increased levels of CD147 on pDCs, pre-DCs and pre-DC2s were specific to severe COVID-19. In this context it is interesting to note that DC progenitors can themselves be susceptible to viral insult 50 . It remains to be addressed what causes CD147 upregulation in COVID-19 and how the MNP system may participate in dysregulated coagulation, but it has previously been shown that CD147 support platelet-monocyte interactions promoting vascular inflammation 51 . In addition, is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted August 31, 2020. . https://doi.org/10.1101/2020.08.25.20181404 doi: medRxiv preprint CD147 is upregulated upon TNF and IFN-γ stimulation, an effect strongly potentiated by bacterial stimuli 52 . Of note, the two pulmonary embolism cases in our cohort occurred in severe COVID-19 patients suffering from superinfection with positive blood cultures.

In summary, we here provide a comprehensive mapping of the MNP landscape in COVID-19. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted August 31, 2020. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted August 31, 2020. 

In order to determine the absolute leucocyte numbers, trucount staining was performed. Briefly, is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted August 31, 2020. . https://doi.org/10.1101/2020.08.25.20181404 doi: medRxiv preprint fixed in PBS containing 1% PFA for 2 hours in order to inactivate the virus. Cells were acquired on a FACSymphony A5 instrument (BD Biosciences), equipped with UV (355nm), violet (405 nm), blue (488 nm), yellow/green (561 nm), and red laser (637 nm). Details of antibodies used as well as information on filters for cytometer configuration are provided in Table 2 . 

Serum levels of soluble factors were analyzed at the day of the sampling using proximity extension assay, based on real-time PCR quantification of pair-wise binding of oligonucleotidelabeled target antibodies (Olink Proteomics, Uppsala, Sweden; sensitivity, specificity, dynamic range, repeatability, reproducibility and scalability validation documents are available at www.olink.com/downloads).

Seurat version 3 was used to re-analyze single cell data. In brief, scRNA-seq from previously is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted August 31, 2020. . https://doi.org/10.1101/2020.08.25.20181404 doi: medRxiv preprint data matrix followed by PCA using top 2000 most variable genes. Top 50 principal components were used to visualize the cells in UMAP, and 'bimod' test was used to detect differentially expressed genes. First, we selected myeloid DC clusters 22, 25, 27 (CLEC9A, CADM1, CD1c, FCER1A), pDC cluster 28 (LILRA4, IL3RA), and other myeloid clusters 0, 1, 2, 3, 4, 5, 6, 9, 10, 11, 12, 13, 21, 23, 29 (CD14, CD68), after exclusion of B cells, plasma cells, epithelium, NK cells, and T cells (Supp. Fig. 1A-F) . After re-running total myeloid cells through the dimensionality reduction pipeline (Supp. Fig. 1G-K) , neutrophils were excluded (Supp. Fig.   1F , G). To identify DCs, DC-containing clusters 11, 27, and 29 from the myeloid cell UMAP were re-clustered (Supp. 1L-P). As cluster 4 contained a comparable number of cells in severe and moderate cohorts, and cluster 7 in severe and healthy, the DEGs in those clusters were calculated between the severe patients and the respective cohort and subjected to pathway analysis. Ingenuity Pathway Analysis (IPA) was performed to study pathways (Content version: 51963813; Release Date: 2020-03-11; Ingenuity Systems). Predicted upregulation and downregulation of pathways was based on Z-score, where positive score implied upregulation and negative score implied downregulation.

PCA plots were created using 'prcomp' function and 'ggbiplot' package in R and correlation mapping was performed using 'corrplot' package in R. The color of the circles indicated positive (red) and negative (blue) correlations, color intensity represented correlation strength as measured by the Pearson's correlation coefficient. The correlation matrix was reordered using "hclust" for hierarchical clustering order. Significance tests were performed to produce p-values and confidence intervals for each pair of input features. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted August 31, 2020. . https://doi.org/10.1101/2020.08.25.20181404 doi: medRxiv preprint

Differences between two groups were evaluated using Mann-Whitney test, and Kruskal-Wallis test with Dunn's multiple comparisons test were used to evaluate differences among the three groups in all the analysis. Friedman test was used in Fig. 2E to evaluate differences among paired variables. Spearman test was used for correlations of non-normally distributed data and Pearson test was used for normally distributed data. Significance for pathways analyses in Fig.   3E , 3H was defined by the IPA software (Ingenuity Systems). Significance level: *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted August 31, 2020. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted August 31, 2020. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted August 31, 2020. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted August 31, 2020. . https://doi.org/10.1101/2020.08.25.20181404 doi: medRxiv preprint is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted August 31, 2020. . https://doi.org/10.1101/2020.08.25.20181404 doi: medRxiv preprint is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted August 31, 2020. . https://doi.org/10.1101/2020.08.25.20181404 doi: medRxiv preprint (A) All DCs from patients and controls concatenated in a UMAP, as described in Fig Heatmap showing marker expression in the three cohorts, across the DC subsets, only samples with more than 10 cells included. Statistical evaluation was made separately in each indicated subset by comparing the MFI in the three cohorts; significance for healthy to moderate and healthy to severe comparisons is indicated by * and for moderate to severe comparison is indicated by #. Statistical evaluation using Spearman test for correlation, Kruskal-Wallis test and Dunn's multiple comparisons test for all other analysis. Samples with less than 10 cells in any of the DC subsets annotated in (Q) were excluded from the analyses. Significance level: *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted August 31, 2020. . https://doi.org/10.1101/2020.08.25.20181404 doi: medRxiv preprint is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted August 31, 2020. . https://doi.org/10.1101/2020.08.25.20181404 doi: medRxiv preprint controls concatenated in a UMAP with color-coded manually gated cell subsets (left), rerun on monocytes presented in a UMAP with color-coded Phenograph clusters (right). (G) Color-coded manual gates (upper panel) and indicated Phenograph clusters (lower panel) in monocyte UMAP of concatenated samples. (H) Expression of markers in selected significantly differential Phenograph clusters among the three cohorts (see also Supp. Fig.  2 ). (I) Heatmap of correlation between Phenograph clusters 1-18 and soluble factors in serum. (J) Correlation between soluble serum AREG and % of Phenograph clusters 9 and 10 in total monocytes. (K) MDSC signature genes in lung MNP compartment differentially expressed between severe COVID-19 patients and controls, presented in the three cohorts; data re-analyzed from previously published report 22 , as described in "Single cell analysis and correlation plots". Statistical evaluation using Kruskal-Wallis test and Dunn's multiple comparisons test for comparison between the three cohorts and Spearman test for correlations, 'bimod' test for differentially expressed genes. Significance level: *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001 is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted August 31, 2020. . https://doi.org/10.1101/2020.08.25.20181404 doi: medRxiv preprint is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted August 31, 2020. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted August 31, 2020. . https://doi.org/10.1101/2020.08.25.20181404 doi: medRxiv preprint is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted August 31, 2020. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted August 31, 2020. . https://doi.org/10.1101/2020.08.25.20181404 doi: medRxiv preprint

To address general features of the clinical biomarker pattern of COVID-19 we investigated parameters measured in the clinic as a part of routine monitoring, including inflammatory, hematological, biochemical and coagulation related parameters, and compared them between the two cohorts of COVID-19 patients, reference values indicated in grey ( Figure 6A ). As expected, almost all patients had elevated levels of almost all inflammatory parameters previously associated with COVID-19, such as CRP, ferritin, as well as the cytokine levels (IL-10, IL-6, IL-1β, TNF), and differences in IL-6 levels was detected between the two study cohorts with higher levels in the critically ill patients. Coagulation parameters were also affected, and higher levels of D-dimer were detected in the severe patients, who also had elevated levels of fibrinogen (85%, 11/13). Biochemical status revealed higher levels of lactate dehydrogenase (LDH) and myoglobin in critically ill patients and hypoalbuminemia in all, with lower levels in severe disease. From the hematological perspective, severe patients had higher levels of WBC, higher neutrophil counts, lower lymphocytes and subsequently higher neutrophil to lymphocyte ratio, with no striking differences in platelets or monocytes ( Figure   6A ). Radiologically, all patients examined showed bilateral infiltrates on chest x-ray; therapeutically, the majority received anticoagulants (93%, 25/27) and more than half broadspectrum antibiotics (56%, 15/27). From a HLH perspective, all patients had fever at admission (100%, 27/27) and elevated ferritin levels (100%, 26/26, data not available for 1 patient), but no suspicion from the hematological parameters available, and in line rather elevated than low fibrinogen levels. Four patients in the severe group had superinfections (24% of severe patients, 4/17), in two of whom pulmonary embolism was detected, representing all detected cases of pulmonary embolism. To further address clinical laboratory status in relation to each other clinical parameters and outcome, integrative correlation mapping was performed ( Figure 6B ). is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted August 31, 2020. This showed cytokines (IL-10, TNF, IL-6), LDH, neutrophil and D-dimer levels clustering together with fatal outcome, peak oxygen need and viremia. Of note, levels of monocytes showed a positive correlation with body mass index (BMI), a known COVID-19 risk factor, and monocytes were the only clinically available cell type correlating with days from symptom debut, indicating the myeloid mononuclear cell importance in disease development timeline.

We also investigated symptom panorama at admission, and found that in addition to fever, all patients had dyspnea (100%, 27/27), most had cough (85%, 23/27), few had GI related symptoms (15%, 4/27), while body ache showed higher variation (44%, 12/27), and correlated with thromboembolism, days since symptom debut and, intriguingly, levels of monocytes ( Figure 6B ). Importantly, a similar pattern was observed when integrative correlation mapping was performed including clinical parameters available 24 hours within MNP profiling (Supplementary Figure 4A) . Further details on clinical and laboratory findings are presented in Supplementary Table 1 is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted August 31, 2020. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted August 31, 2020. . https://doi.org/10.1101/2020.08.25.20181404 doi: medRxiv preprint

SUPPLEMENTARY FIGURE 1: sc-RNAseq re-analysis (related to Fig. 3, Fig. 4 

. CC-BY-NC-ND 4.0 International license It is made available under a perpetuity.

is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted August 31, 2020. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted August 31, 2020. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted August 31, 2020. . https://doi.org/10.1101/2020.08.25.20181404 doi: medRxiv preprint

COVID-19: consider cytokine storm syndromes and immunosuppression

Clinical features of patients infected with 2019 novel coronavirus in Wuhan

Dendritic cells, monocytes and macrophages: a unified nomenclature based on ontogeny

Dendritic cell functions: Learning from microbial evasion strategies

Gene expression profiling reveals the defining features of the classical, intermediate, and nonclassical human monocyte subsets

Comparison of gene expression profiles between human and mouse monocyte subsets

Blood monocytes consist of two principal subsets with distinct migratory properties

A close encounter of the third kind: monocytederived cells

Monocyte-Derived Cells in Tissue-Resident Memory T Cell Formation

Distinct progenitor lineages contribute to the heterogeneity of plasmacytoid dendritic cells

Plasmacytoid dendritic cells develop from Ly6D+ lymphoid progenitors distinct from the myeloid lineage

Human dendritic cell subsets: an update

Human inflammatory dendritic cells induce Th17 cell differentiation

Single-Cell Analysis of Human Mononuclear Phagocytes Reveals Subset-Defining Markers and Identifies Circulating Inflammatory Dendritic Cells

Transcriptional and Functional Analysis of CD1c+ Human Dendritic Cells Identifies a CD163+ Subset Priming CD8+CD103+ T Cells

Differential IRF8 Transcription Factor Requirement Defines Two Pathways of Dendritic Cell Development in Humans

Mapping the human DC lineage through the integration of highdimensional techniques

Single-cell RNA-seq reveals new types of human blood dendritic cells, monocytes, and progenitors

Complex Immune Dysregulation in COVID-19 Patients with Severe Respiratory Failure

A single-cell atlas of the peripheral immune response in patients with severe COVID-19

Pathological inflammation in patients with COVID-19: a key role for monocytes and macrophages

Decrease of nonclassical and intermediate monocyte subsets in severe acute SARS-CoV-2 infection. Cytometry A cyto

Imbalanced Host Response to SARS-CoV-2 Drives Development of COVID-19

Expansion of myeloid-derived suppressor cells in patients with severe coronavirus disease (COVID-19)

The fate and lifespan of human monocyte subsets in steady state and systemic inflammation

Tocilizumab for the treatment of severe COVID-19 pneumonia with hyperinflammatory syndrome and acute respiratory failure: A single center study of 100 patients in Brescia

Clinical features of patients infected with 2019 novel coronavirus in Wuhan

COVID-19_ Review on latest available drugs and therapies against SARS-CoV-2. Coagulation and inflammation cross-talking

Protective vascular coagulation in response to bacterial infection of the kidney is regulated by bacterial lipid A and host CD147

Constitutive Siglec-1 expression confers susceptibility to HIV-1 infection of human dendritic cell precursors

EMMPRIN (CD147/basigin) mediates platelet-monocyte interactions in vivo and augments monocyte recruitment to the vascular wall

A role for membrane-bound CD147 in NOD2-mediated recognition of bacterial cytoinvasion

The Third International Consensus Definitions for Sepsis and Septic Shock (Sepsis-3)

Remdesivir for the Treatment of Covid-19 -Preliminary Report

After quality control, scRNA-seq data reanalyzed using the Seurat version 3 (V3). (C) Expression plots of key markers used to discriminate between cells to be included and excluded in subsequent analysis. (D) Distribution of cells from COVID-19 patients and healthy controls among clusters generated in the whole data set. (E) Distribution of cells from moderate and severe COVID-19 patients and healthy controls among clusters

Distribution of cells from each sample among clusters generated in the whole data set

Expression plots of key markers used to discriminate between MNP and neutrophil clusters. (I) Distribution of cells from COVID-19 patients and healthy controls in MNP clusters. (J) Distribution of cells from moderate and severe COVID-19 patients and healthy controls in MNP clusters. (K) Distribution of cells from each sample in MNP clusters. (L) Clusters containing DCs selected and reanalyzed. (M) Distribution of cells from COVID-19 patients and healthy controls in DC clusters. (N) Distribution of cells from moderate and severe COVID-19 patients and healthy controls in DC clusters. (O) Distribution of cells from each sample in DC clusters. (P) Key genes of pDC and DC1 in cluster 4 (green) and 7 (red), respectively. (Q) Genes in pDCs differentially expressed between severe and moderate COVID-19 patients

24 hours from the MNP profiling; the color of the circles indicated positive (red) and negative (blue) correlations, color intensity represented correlation strength as measured by the Pearson's correlation coefficient. (B) Differences in absolute numbers of MNP populations between SARS-CoV-2 PCR+ and SARS-CoV-2 PCR-patients, and between SARS-CoV-2 IgG-and SARS-CoV-2 IgG+ patients. (C) Correlation of days since symptom debut until the sample collection and absolute numbers of cells within MNP populations. (D) Principal component analysis of 108 MNP parameters in the three cohorts, parameters contributing most to the separation highlighted; each dot represents controls (white), moderate COVID-19 (light green), and severe COVID-19 (dark green) patient; to better visualize marker names, a plot with smaller dots is provided (right). (E) Differences in LD, D-dimer, Ferritin, neutrophils/lymphocyte ratio at peak or 24h within MNP profiling (lower row) between patient cluster 2 and 3. (F) Differences in CD200R MFI between survivors and nonsurvivors in all patients (left) and in severe patients only (right). Statistical evaluation using Mann-Whitney for comparison between the two groups, Spearman test for correlations of non-normally distributed data and Pearson test for correlations of normally distributed data