Enhanced eosinophilic inflammation associated with antibody and 1 complement-mediated pneumonic insults in severe COVID-19 2 3

3 Dong-Min Kim1,13,*, Jun-Won Seo1,13, Yuri Kim2,3,4,13, Jooyeon Lee5, 13, Uni Park2,3,13, 4 Na-Young Ha2,3,4,13, Jaemoon Koh6,13, Hyoree Park2,3, Jae-Won Lee2,3, Hyo-Jin Ro2,3, Na Ra Yun1, 5 Da Young Kim1, Sung Ho Yoon1, Yong Sub Na1, Do Sik Moon1, Sung-Chul Lim7, 6 Choon-Mee Kim8, Kyeongseok Jeon2,3, Jun-Gu Kang9, Hyeongseok Jeong5, Jungok Kim10, 7 Shinhyea Cheon5, Kyung Mok Sohn5, Jae Youg Moon10, Sungmin Kym10, Myung-Shin Lee11, 8 Hyun-Je Kim12, Woong-Yang Park12, Hyun-Woo Shin3, Hye-Young Kim3, Chung-Hyun Cho3, 9 Yoon Kyung Jeon6, Yeon-Sook Kim5,*, and Nam-Hyuk Cho2,3,4,* 10


Introduction
Severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2) 1 has been rapidly spreading  systemic inflammation, as indicated by the levels of C-reactive proteins (CRP) in plasma. 8 The 110 kinetics of viral loads in upper (nasopharyngeal and throat swabs) and lower (sputa and BALFs) 111 respiratory tracts were not significantly different between severe and mild groups (Fig. 1a), as 112 previously reported. 9,10 However, the levels of CRP in plasma were more significantly elevated 113 in severe patients, especially during the first 20 days after symptom onset and peaking around 114 day 10 (D10) (Fig. 1b). Therefore, viral loads measured in respiratory secretions may not be 115 significantly associated with disease severity and systemic inflammation in COVID-19 patients. 116 To investigate the potential causative factors driving severe pulmonary inflammation during the  total leukocytes), monocytes/macrophages (37.5 ± 32.7%), and a few lymphocytes (11.6 ± 124 12.2%). We also observed 48.9% (22/45) of the respiratory specimens included eosinophils (4.3 125 ± 7.3%) ( Fig. 2a and Supplementary Table S2). When we compared the level of each 126 inflammatory cell type between mild and severe patients, the relative proportion of all the cell 127 types were not significantly different between the groups, although eosinophils were slightly 128 higher in the mild group (6.0 ± 9.4%) than severe patients (2.8 ± 4.6%, p = 0.18) (Supplementary 129   Table S2).

130
In order to characterize the immune cell population in detail and compare them with systemic 131 leukocytes in COVID-19 patients, five paired BALFs and blood leukocytes collected from three 132 7 severe patients at the indicated day after symptom onset were directly applied for flow 133 cytometric analysis. We identified the relative proportions of myeloid cells, including CD14 + 134 monocytes/macrophages and SSC High /CD14 -PMNs, among BALFs' and blood leukocytes, as 135 well as CD3 + T cells, CD20 + B cells, CD3 + /CD56natural killer (NK) cells, and CD3 + /CD56 +

136
NKT-like (denoted as NKT hereafter) cells, 11 based on expression levels of the indicated surface 137 markers (Fig. 2b, Supplementary Fig. S2 -S5). In addition, PMNs were further defined by 138 assessing the surface expression of CD16 and CD24. 12 T cells and NKT cells were also 139 characterized by measuring CD4 and CD8 surface expression. As we observed in respiratory 140 specimens by H&E staining, PMNs were the predominant inflammatory cells, ranging from 141 45.5% to 79.7% of lung-infiltrating leukocytes ( Fig. 2b and Supplementary Fig. S3 Supplementary Fig. S4). NK and NKT cells represented 2.3 ~ 9.6% and 3.7 ~ 11.5% of 150 pulmonary lymphocytes, respectively. The ratio of CD8 + and CD4 + T cell in the respiratory 151 specimens fluctuated (1.6 at D14, 0.6 at D20, 1.1 at D27, 1.0 at D30, and 1.9 at D39), but CD8 + 152 T cells generally predominated, suggesting a potential role of cytotoxic T cell responses in the

156
To assess the kinetic changes of immune cell types in respiratory specimens of COVID-19 157 patients, the relative proportion of each cell type was measured over time after symptom onset by 158 H&E staining and flow cytometry ( Fig. 2c and d). Despite individual variations and fluctuations 159 among the specimens, PMNs, primarily neutrophils, were sustained in respiratory specimens 160 from severe cases, but rapidly declined in mild patients during the first 20 days after symptom 161 onset. Monocytes/macrophages also gradually decreased after an initial peak around D10 after 162 symptom onset in both mild and severe groups (Fig. 2c), whereas lymphocytes gradually 163 increased with more rapid response in mild patients than the severe group (Fig. 2d). Interestingly, 164 eosinophils were detectable in 55% (11/20) of specimens from mild patients during the first 10 165 days after symptom onset, whereas their infiltration was rather delayed and peaked during D10-166 20 in severe cases (Fig. 2d). Eosinophil-positive rate in severe patients' sample was 64.0% 167 (16/25) when assessed by H&E and flow cytometry.

168
Since we observed pulmonary infiltration of PMNs, including neutrophils and eosinophils, we 169 measured the levels of inflammatory mediators derived from neutrophils (lipocalin-2, LCN; 170 calprotectin, CAPL) 13,14 and eosinophils (eosinophil-derived neurotoxin, EDN; eosinophilic 171 cationic protein, ECP) 15,16 in the respiratory specimens to assess innate cellular activation. LCN 172 and CALP were detected in specimens from both mild and severe patients (Fig. 3a). The relative 173 levels of both mediators were not significantly different between mild (mean ± S.D.: 977.8 ± 174 1366.4 ng/ml and 155.3 ± 478.7 mg/ml for LCN and CALP, respectively) and severe groups 175 (905.6 ± 1741.7 ng/ml and 44.7 ± 91.1 mg/ml for LCN and CALP, respectively), although 176 neutrophil responses were more sustained in severe cases (Fig. 3a). In contrast, levels of EDN 177 and ECP were generally higher in the respiratory specimens from severe patients (mean ± S.D.: 937.5 ± 1244.7 pg/ml and 60.5 ± 77.5 ng/ml for EDN and ECP, respectively) than mild cases 179 (231.8 ± 375.6 pg/ml and 32.5 ± 62.3 ng/ml for EDN and ECP, respectively) (Fig. 3b). Of note, 180 the levels of ECP were significantly higher in severe cases than the mild group (p = 0.0187). In 181 addition, the level of mast cell tryptase (MCT) derived from mast cells upon activation was 182 approximately 3.5 times higher in the severe group (31.1 ± 42.3 ng/ml) than mild cases (8.8 ± 4.4 183 ng/ml), although the difference was not statistically significant (p = 0.1141) (Fig. 3c). We also  In addition, cytotoxic activity of CTLs and NK cells was not significantly different between mild 193 and severe groups, when assessed by measuring granzyme A in respiratory specimens (Fig. 3e).

194
Therefore, these results highlight the significant contribution of enhanced and sustained inflammatory markers, IL-6 (mean ± S.D.: 2.8 ± 8.1 vs. 9.1 ± 14.1 ng/ml for mild and severe 217 group, respectively), TGF-β (0.9 ± 1.9 vs. 3.2 ± 4.0 ng/ml for mild and severe group, 218 respectively), and ECP (as noted above), were significantly higher in severe patients than the 219 mild group (Supplementary Table S3). Systemic elevation of IL-6 is known to be a hallmark of 220 respiratory failure and cytokine release syndrome in severe COVID-19, 22,23 and our data 221 confirms a significant association of this inflammatory cytokine with disease severity even in the 222 respiratory environment of COVID-19 patients. In addition, IL-6 also showed broad and 223 11 significant association with pro-inflammatory cytokines, as well as type 2 cytokines. ECP is 224 significantly correlated with proinflammatory cytokines, including IL-6 and TNF-α, in addition 225 to type 2 cytokines, such as IL-5 and IL-13 (Fig. 4a), suggesting a potential association of   concentrations of C3a and C5a. As observed in respiratory specimens (Fig. 5b), levels of C3a and 269 C5a were generally higher in plasma from severe patients (7.9 ± 1.3 µg/ml and 50.5 ± 20.3 ng/ml 270 for C3a and C5a, respectively) than in mild group (1.2 ± 2.0 µg/ml and 38.9 ± 30.1 ng/ml for 271 C3a and C5a, respectively) (Fig. 6b). The difference in C5a levels between mild and severe 272 groups was statistically significant, as previously reported. 28 In addition, we observed a 273 significant positive correlation of C5a levels with plasma IgG1 and IgG3 levels (Fig. 6c). These succumbing to death at D59 (Fig. 7). Viral loads in respiratory specimens gradually declined 286 after symptom onset and were barely detected after D30 (Fig. 7a), as observed in other surviving 287 cases (Fig. 1). Even though plasma CRP levels peaked at D13 and decreased thereafter, the levels  Based on our observations of extensive kinetic analyses using respiratory specimens, we propose 335 that sustained eosinophilic inflammation is followed by Th2-biased adaptive immune responses, consistently observed in severe cases ( Fig. 2 and 3b). Interestingly, rapid eosinophilic infiltration 345 into infected lungs was often observed in mild patients within 10 days after symptom onset, 346 whereas eosinophil infiltration is delayed but prolonged and eosinophilic inflammation is are not necessarily critical for antiviral immunity, since viral copy numbers in respiratory 397 secretions declined to baseline during peak antibody responses (D20-30) and there was no 398 significant difference in viral loads between mild and severe patients (Fig. 1a). In addition, showed prominent increase in hallmark inflammation scores when compared to those of healthy 419 control group (Fig. 9b). Analysis of scRNA seq data sets from mononuclear phagocytes 420 (monocytes, macrophages, and dendritic cells) and neutrophils showed significantly higher levels 421 of hallmark gene set scores for both FcγR signaling and complement activation in severe group 422 than those of mild cases (Fig. 9c). These results clearly indicate a significant role of FcγR       Analysis of single cell RNA seq datasets of BALFs from COVID-19 patients. 552 We collected BALF single cell RNA seq (scRNA-seq) datasets from the Gene Expression      Table S3. Raw data of 28 cytokines, chemokines, and inflammatory mediators measured in 906 respiratory specimens collected from COVID-19 patients.