T cell activation, highly armed cytotoxic cells and a sharp shift in monocytes CD300 receptors expression is characteristic of patients with severe COVID-19

COVID-19 manifests with a wide diversity of clinical phenotypes characterized by dysfunctional and exaggerated host immune responses. Many results have been described on the status of the immune system of patients infected with SARS-CoV-2, but there are still aspects that have not been fully characterized. In this study, we have analyzed a cohort of patients with mild, moderate and severe disease. We performed flow cytometric studies and correlated the data with the clinical features and clinical laboratory values of patients. Both conventional and unsupervised data analyses concluded that patients with severe disease are characterized, among others, by a higher state of activation in all T cell subsets, higher expression of perforin and granzyme B in cytotoxic cells, expansion of adaptive NK cells and the accumulation of activated and immature dysfunctional monocytes which are identified by a low expression of HLA-DR and an intriguing abrupt change in the expression pattern of CD300 receptors. More importantly, correlation analysis showed a strong association between the alterations in the immune cells and the clinical signs of severity. These results indicate that patients with severe COVID-19 have a broad perturbation of their immune system, and they will help to understand the immunopathogenesis of severe COVID-19 as well as could be of special value for physicians to decide which specific therapeutic options are most effective for their patients.

for its therapeutic implications, but also to better understand the immunopathology of 150 the disease. Therefore, it is essential to entirely define the immune response 151 characteristics related to disease features and determine at which stage of the disease 152 specific therapeutic options may be most effective. 153 We have characterized lymphocytes (T, B and NK cells) and monocytes of patients with   169 Our aim was to evaluate the impact of acute SARS-CoV-2 infection in circulating 170 leukocytes. To this end, we performed a cross-sectional study. Forty four patients with 171 COVID-19 disease were recruited for the study. To correlate laboratory findings, 172 including frequencies and phenotype of circulating leukocytes and the severity of the  Table S1 and Table S2. Inclusion and exclusion criteria were followed to guarantee the 177 homogeneity of the cohort, including age, gender, severity of the disease and time from 178 the onset of symptoms to sample collection. In addition, twelve healthy controls (HC) 179 were included in the study.

180
No significant differences were found between COVID-19 patients and HC in relation 181 to age (median ages of 64 and 59.5, respectively). There were also no significant 182 differences between the three groups of patients (severe, moderate and mild) in relation 183 to the number of days from the appearance of symptoms and the sample collection:  (Fig. 1B). Specifically, 26% of patients exhibited IL-6 192 levels above the normal range (>40 pg/mL). Interestingly, all HC had IL-6 levels below 193 the limit of detection (<3 pg/mL), while 69% of patients had >3 pg/mL of IL-6. On the 194 other hand, 66% of patients exhibited CRP levels above the normal range (>11 mg/L) 195 and 57% of patients exhibited ferritin levels above the normal range (>300 ng/mL) (Fig. 196 1B). Furthermore, although white blood cell (WBC) counts were mostly normal in mild 197 and moderate COVID-19 patients, some moderate and severe patients exhibited high 198 WBC counts (Fig. 1C). Also, and in accordance with the literature (Hadjadj et al., 2020;199 Huang et al., 2020), we observed frequencies and absolute numbers of lymphocytes 200 8 below the normal values, and frequencies and absolute number of neutrophils above the 201 normal values associated with the severity of the disease (Fig. 1C). Finally, increased 202 levels of IL-6 (>40 pg/mL), CRP (>11 mg/L), ferritin (>300 ng/mL), fibrinogen (>400 203 mg/dL) and D-dimer (>500 ng/mL) and lower levels of hemoglobin (<13 g/dL) were 204 observed mostly in moderate and severe patients (Fig. 1D).

205
To examine potential associations between these general laboratory values and other 206 clinical features, we performed correlation analysis (Fig. 1E). The analysis revealed

SARS-CoV-2 infection is associated with activated CD4 T cells subsets expressing
216 higher levels of PD-1 and perforin 217 We next performed a detailed multiparametric flow cytometry analysis to further 218 investigate circulating leukocytes status in COVID-19 patients (see gating strategy for 219 each cell population in Fig. S1). Given the important role of T cells in the defense 220 against viral infections and in the establishment of an immunological memory, as well 221 as in the immunopathology and damage that may occur, we studied T cell 222 subpopulations. We did not observe significant differences in the frequency of the major 223 T cells subsets, i.e. CD4, CD8 and double negative (DN) neither in the CD4/CD8 ratio 224 between the patients and compared with the HC (Fig. S2). Four major CD4 T cell  memory. But also, highly differentiated CD8 T cells have been suggested to induce 301 damage in SARS-CoV-2 infected lungs in an antigen-independent manner (AN and 302 DW, 2020). Therefore, we next examined the four major subpopulations (naïve, 303 memory, effector-memory and TEMRA). We observed no significant differences in the 304 frequencies of naïve, memory and TEMRA subsets between HC and COVID-19 305 patients. Nevertheless, the frequency of CD8 effector-memory cells was significantly 306 higher in patients with severe disease (Fig. 3A). 307 Then, we determined the activation status of CD8 T cells. We observed that COVID-19 308 patients exhibited a significant expansion of activated (CD38+HLA-DR+) cells,  In contrast to the CD4 T cells, we did not observe a correlation between CD38+HLA-321 DR+ cells and PD-1+ cells in COVID-19 patients (Fig. S5C), probably suggesting that 322 PD-1 expression on CD8 T cells is more a marker of exhaustion than of activation.

323
Nevertheless, more studies are required to confirm this statement.  When we looked at the total CD8 T cell population we observed a significant increase 328 in the frequency of cells containing perforin in patients with severe disease (Fig. S5D).   CD300c and CD300e on monocytes from patients with COVID-19 (Fig. 4D). Results 385 showed a differential expression of these receptors between HC and patients. Very   with COVID-19 patients (Fig. 6A). 445 Next, we analyzed the expression of perforin and granzyme B in NK cells (Fig. 6B).

446
Results showed that both CD56 bright and CD56 dim NK cells from COVID-19 patients  (Fig. 6C). Nevertheless, the increased expression of perforin and granzyme B that 458 associated with the severity of the disease was also evident in each of the four subsets 459 (Fig. S9A). 460 We also performed high-dimensional mapping of the eight parameter flow cytometry 461 data using tSNE representation and it was evident that some regions were preferentially 462 found in COVID-19 patients when compared with HC (Fig. 6D). To gain more insight 463 into the NK cell alterations observed in COVID-19, we used the FlowSOM clustering 464 tool and compared the expression of the eight markers to define 16 populations (Fig. 465 6E). Using this approach, we were able to identify some populations that were 466 differentially expressed between COVID-19 patients and HC (Fig. 6F and S9B). Pop6, are more than 5% of circulating CD56 dim cells. In addition, we have also determined 504 that there is an expansion of adaptive FcRγ-NK cells when they represent more than 505 7% of the CD56 dim cells. Results in Fig. 7C shows that CMV-seronegative individuals 506 do not have expansions of adaptive NK cells, except for one patient with moderate 507 disease, in which the NKG2C+CD57+ cells represented more than 5%, and another 508 patient with severe disease, in which the FcRγ-cells were more than 7%. When only the 509 CMV-seropositive individuals were taken into account, we observed a significant 510 expansion of adaptive NK cells in patients with moderate and severe COVID-19, which 511 was more pronounced when NKG2C+CD57+ cells were taken into account instead of 512 FcRγ-cells (Fig. 7C). 513 Finally, we performed high-dimensional mapping using tSNE representation and we 514 observed that some regions were preferentially found in CMV-seropositive individuals 515 compared with CMV-seronegative donors (Fig. 7D). Then, to better understand the 516 differences in NK cell subsets between CMV-seropositive and CMV-seronegative 517 individuals we used the FlowSOM clustering tool and compared the expression of seven 518 markers to define 8 populations (Fig. 7E). Using this approach, we were able to identify 519 some populations that were differentially expressed between the two groups of donors 520 ( Fig. 7F and S10C). Specifically, the adaptive NK cells Pop6 and Pop7 were 521 characterized by the phenotype NKG2C+FcRγ-, and while Pop6 was CD57+, Pop7 was

Statistical analysis reveals the relationships between circulating T cells, NK cells 527
and monocytes with disease severity in COVID-19 patients. 528 We first performed a bivariate analysis of 203 clinical laboratory and flow cytometry 529 variables (Table S3). We selected the statistically significant variables for a multivariate 530 analysis. Then, to reduce the number of variables to include in the multivariate analysis 531 we performed a principal component analysis (PCA) (Fig. 8A). Components 1 to 4 532 explained around 73.7% of the variance, and components 1 and 2 explained around 533 60.8% of the variance (Fig. S11A). In Figure S11B  different between patients with mild and moderate disease (Fig. 8B, upper panel). On 545 the other hand, when we compared patients with moderate disease with those with a 546 mild and severe disease, we could see that component 1 was significantly different 547 between patients with a moderate and severe disease and, as expected, component 2 was 548 different between patients with moderate and mild disease (Fig. 8B, lower panel). 549 Next, we performed correlation analysis. A different correlogram pattern was observed 550 between HC and patients groups when we looked at the correlation between the 551 significant flow cytometry variables (Fig. S12). Then, the analysis was performed to    Table S1 and S2). All donors   Table S4).    test.

808
Bivariate analyses were performed (Table S3). First, using the Shapiro-Wilks normality 809 test we determined if variables followed a normal distribution. If they did, the average Whitney U test otherwise. To take into account multiple comparisons, we also presented 814 the adjusted p-values using the Benjamini & Hochberg test (Table S3)