Distinct lung-homing receptor expression and activation profiles on NK cell and T cell subsets in COVID-19 and influenza

Respiratory viral infections with SARS-CoV-2 or influenza viruses commonly induce a strong infiltration of immune cells into the lung, with potential detrimental effects on the integrity of the lung tissue. Despite comprising the largest fractions of circulating lymphocytes in the lung, little is known about how blood natural killer (NK) cells and T cell subsets are equipped for lung-homing in COVID-19 and influenza. Using 28-colour flow cytometry and re-analysis of published RNA-seq datasets, we provide a detailed comparative analysis of NK cells and T cells in peripheral blood from moderately sick COVID-19 and influenza patients, focusing on the expression of chemokine receptors known to be involved in leukocyte recruitment to the lung. The results reveal a predominant role for CXCR3, CXCR6, and CCR5 in COVID-19 and influenza patients, mirrored by scRNA-seq signatures in peripheral blood and bronchoalveolar lavage from publicly available datasets. NK cells and T cells expressing lung-homing receptors displayed stronger phenotypic signs of activation as compared to cells lacking lung-homing receptors, and activation was overall stronger in influenza as compared to COVID-19. Together, our results indicate migration of functionally competent CXCR3+, CXCR6+, and/or CCR5+ NK cells and T cells to the lungs in moderate COVID-19 and influenza patients, identifying potential common targets for future therapeutic interventions in respiratory viral infections. Author summary The composition of in particular CXCR3+ and/or CXCR6+ NK cells and T cells is altered in peripheral blood upon infection with SARS-CoV-2 or influenza virus in patients with moderate disease. Lung-homing receptor-expression is biased towards phenotypically activated NK cells and T cells, suggesting a functional role for these cells co-expressing in particular CXCR3 and/or CXCR6 upon homing towards the lung.


Abstract 22
Respiratory viral infections with SARS-CoV-2 or influenza viruses commonly induce a strong 23 infiltration of immune cells into the lung, with potential detrimental effects on the integrity of 24 the lung tissue. Despite comprising the largest fractions of circulating lymphocytes in the lung, 25 little is known about how blood natural killer (NK) cells and T cell subsets are equipped for 26 lung-homing in COVID-19 and influenza. Using 28-colour flow cytometry and re-analysis of 27 published RNA-seq datasets, we provide a detailed comparative analysis of NK cells and T 28 cells in peripheral blood from moderately sick COVID-19 and influenza patients, focusing on 29 the expression of chemokine receptors known to be involved in leukocyte recruitment to the 30 lung. The results reveal a predominant role for CXCR3, CXCR6, and CCR5 in COVID-19 and 31 influenza patients, mirrored by scRNA-seq signatures in peripheral blood and bronchoalveolar 32 lavage from publicly available datasets. NK cells and T cells expressing lung-homing receptors 33 displayed stronger phenotypic signs of activation as compared to cells lacking lung-homing 34 receptors, and activation was overall stronger in influenza as compared to  Together, our results indicate migration of functionally competent CXCR3 + , CXCR6 + , and/or 36

CCR5 + NK cells and T cells to the lungs in moderate COVID-19 and influenza patients, 37
identifying potential common targets for future therapeutic interventions in respiratory viral 38 infections. 39

Introduction 50
The ongoing pandemic of coronavirus disease 19 , caused by the novel severe 51 acute respiratory syndrome coronavirus 2 (SARS-CoV-2), highlights the need for a better 52 understanding of respiratory viral infections which have the potential to cause recurrent 53 epidemics or pandemics. In addition to SARS-CoV-2, this also includes influenza virus, 54 respiratory syncytial virus (RSV), SARS-CoV, and the Middle East Respiratory Syndrome 55 (MERS)-CoV. Future disease outbreaks with novel viruses affecting the airways are to be 56 expected and prepared for. 57 During acute infections with respiratory viruses such as with SARS-CoV-2 or influenza, 58 specific chemokines mediating leukocyte recruitment are increased in the lung and 59 bronchoalveolar lavage (BAL) fluid. These chemokines include CCL2, CCL3, CCL20, 60 CXCL1, CXCL3, CXCL10, and IL8, attracting cells expressing chemokine receptors such as 61 CCR2, CCR5, CXCR3, and CXCR6 (1-3). In patients suffering from severe COVID-19, recent 62 reports suggest exacerbated lung tissue damage and epithelial cell death resulting from 63 hyperactivated immune cells, such as inflammatory macrophages (1), natural killer (NK) cells 64 (4) and/or T cells (1,5). While moderate disease in COVID-19 and in influenza patients is per 65 definition not fatal, patients may still require hospitalization and/or experience persistent long-66 term symptoms such as fatigue, respiratory problems, loss of taste or smell, headache, and 67 diarrhea. Since lung-homing cytotoxic lymphocytes are likely involved in lung pathology 68 during acute infection, a better understanding of their major homing mechanisms will help in 69 developing and improving treatment strategies in COVID-19 and other respiratory viral 70 infections. 71 In this study, we investigated the composition of NK cell and T cell subsets which are 72 equipped with lung-homing properties in the peripheral blood of patients suffering from 73 moderate COVID-19 or influenza, and in healthy controls. In addition to analyses by 28-colour 74 flow cytometry, we assessed transcript expression in NK cells and T cells using three publicly available single cell (sc)RNA-seq datasets of cells from peripheral blood and bronchoalveolar 76 lavage, respectively (6-8). Our data indicate a universal role for CXCR3-mediated lung-homing 77

of NK cells and T cells in COVID-19 and influenza and an additional role for recruitment via 78
CXCR6 and CCR5. 79 Together, we provide an extensive characterization of the lung-homing potential of 80 functional NK cells and T cells in homeostasis and acute respiratory viral infections such as 81 COVID-19 and influenza. The present results are of relevance for the understanding of the 82 disease progression and for identifying target molecules to improve future therapeutic treatment 83 strategies. 84 7 CXCR6 expression in COVID-19 and influenza patients, with T cells being more strongly 111

affected. 112
In order to compare the present phenotypic results, with those of other data collections, 113 we analyzed two publicly available scRNA-seq datasets from peripheral blood from a different 114 cohort of SARS-CoV-2-infected patients with moderate COVID-19 disease (6) (Fig. 1k) and 115 from patients infected with either SARS-CoV-2 or influenza A virus (IAV) (7) (Fig. 1l), 116 respectively. In order to allow a fair comparison of the RNA-seq data to the present dataset, we 117 selectively analyzed data from patients with similar clinical characteristics. This analysis 118 revealed differences of chemokine receptor expression between NK cells and T cells at the 119 mRNA level in peripheral blood. While CXCR2 was largely confined to NK cells, CCR2 and 120 CCR5 dominated in T cells, both in healthy controls and in COVID-19 patients ( Fig. 1k) (6). 121 Lower average expression of CXCR3 was found in all three lymphocyte subsets (NK cells,122 CD4 + T cells, CD8 + T cells) in COVID-19 patients as compared to healthy controls (Fig. 1k). 123 However, in contrast to protein data, CXCR6, CCR2, and CCR5 were strongly increased both 124 in frequency and in mean intensity in T cells in COVID-19 patients as compared to healthy 125 controls (Fig. 1k), suggesting post-transcriptional regulation of these chemokine receptors. 126 Finally, direct comparative analysis of transcript expression of CXCR2, CXCR3, CXCR6, and 127 CCR5 revealed stronger loss of CXCR2, CXCR3 and CXCR6 in NK cells from influenza 128 patients as compared to COVID-19 patients as well as a trend towards upregulation of CCR5 129 ( Fig. 1l) (7), in line with the similarly observed trend at the protein level (Fig. 1d, e) and 130 suggesting a role for these chemokine receptors particularly in influenza infection. While 131 CXCR3, CXCR6, CCR2, and CCR5 are predominantly expressed on CD56 bright CD16 -NK cells 132 in peripheral blood from healthy controls and patients with COVID-19 or influenza, 133 respectively, CXCR2 is strongly expressed on CD56 dim CD16 + blood NK cells (Fig. S2d, e). In 134 COVID-19 patients, CXCR2 + CD56 dim CD16 + NK cells were mainly lost within the least 135 differentiated NKG2A + CD57subset (Fig. S2f). On T cells, CXCR2 expression was overall 136 low, without significant differences between healthy controls and COVID-19 patients (Fig. 137 S2g,h). 138 Together, the present phenotypic analyses indicate that CXCR3 is a common lung-139 homing receptor for both NK cells and T cell subsets during acute infection with COVID-19 or 140 influenza, despite major differences for CXCR6, CCR2, and CCR5 between the two diseases at 141 transcriptional level. Furthermore, CXCR6 + NK and T cells were strongly affected in influenza 142 but not in COVID-19 patients, indicating potential differences in homing capacities in the two 143 We and others previously demonstrated an activated phenotype in peripheral blood NK 148 cells and T cells in respectively (4,9,10). Here, we aimed at 149 identifying NK and T cell activation markers in relation to expression of lung-homing receptors 150 as identified by boolean gating, combining cells expressing either CXCR3, CXCR6, CCR2, or 151 CCR5 (Fig. S1b, c). Expression of activation markers such as CD69, CD38, Ki67, and NKG2D 152 was elevated on NK cells in moderate COVID-19 patients (Fig. 2a, b), with strongest increases 153 detected for CD69 and Ki67 (Fig. 2b,c). In detail, induction of CD69 expression was highest 154 on CD56 bright CD16 -NK cells co-expressing lung-homing receptors, while Ki67 induction was 155 highest in corresponding CD56 dim CD16 + NK cells, particularly in those co-expressing lung-156 homing receptors (Fig. 2c). Due to very low numbers of CD56 bright CD16 -NK cells lacking any 157 relevant chemokine receptor in healthy controls, no comparisons could be performed for this 158 subset. In contrast to NK cells from COVID-19 patients, upregulation of CD69 was highest in 159 CD56 dim CD16 + NK cells in influenza patients, irrespective of co-expression of lung-homing 160 receptors (Fig. 2d, e, f). Expression and upregulation of CD38 was similar in COVID-19 and 161 influenza patients (Fig 2c, f). While these data indicate fundamental differences in activation 162 patterns for NK cells in COVID-19 and influenza patients, other phenotypic characteristics 163 remained stable between the two diseases ( Fig. S3a, b), despite induction of granzymes and 164 perforin observed in particular in chemokine receptor-positive CD56 bright CD16 -NK cells (Fig.  165 S3c-f). In the latter subset, upregulation of perforin expression was three times higher in 166 influenza patients as compared to COVID-19 patients (Fig. S3e, f), indicating stronger 167 activation of blood NK cells in influenza. 168 As indicated by transcript level, CXCR2 might have an additional role in lung-homing 169 for NK cells. Since activation affected both CD56 dim and CD56 bright NK cells in COVID-19 170 patients, we next sought to determine changes in CXCR2 + expression (Fig. 2g,h). In this regard, 171 expression of CXCR2 was higher on CD56 dim CD16 + NK cells lacking other lung-homing 172 receptors (Fig. 2g). The frequency of this CXCR2 + CD56 dim CD16 + NK cell subset was 173 significantly reduced in COVID-19 patients as compared to healthy controls (Fig. 2h), 174 indicating an alternative migration mechanism for CXCR2 + CD56 dim CD16 + NK cells to the lung 175 in COVID-19 patients. 176 Together, our data show that the activation patterns differ for NK cell subsets in 177 COVID-19 and influenza patients, respectively, with generally stronger activation of blood NK 178 cells in influenza as compared to moderate COVID-19 patients. Furthermore, the data suggest 179 that CXCR2 might act as an alternative lung-homing receptor for CD56 dim CD16 + NK cells 180 lacking other lung-homing receptors. 181 182

Biased activation of T cells co-expressing lung-homing receptors in COVID-19 and 183
influenza 184 Since NK cell subset activation in COVID-19 and influenza is associated with 185 chemokine receptor expression, we next determined whether T cells expressing lung-homing 186 receptors displayed an equivalent phenotypic bias towards stronger activation of chemokine 187 receptor-positive T cells (Fig. 3). Similar to NK cells, a larger proportion of CD4 + and CD8 + T 188 cells expressed CD69, both in COVID-19 (Fig. 3a, c) and influenza (Fig. 3b, c) compared to 189 healthy controls. While in COVID-19 patients CD69 upregulation was biased towards CD8 + T 190 cells co-expressing lung-homing receptors (Fig. 3a, c), upregulation was overall higher and 191 more uniform between the T cell subsets in influenza (Fig. 3b, c). Furthermore, CD38 was 192 strongly upregulated on CD8 + T cells in influenza but not COVID-19 patients (Fig. 3c). Finally, 193 expression of Ki67 was largely confined to chemokine receptor-positive T cells, both in healthy 194 controls and in COVID-19 patients (Fig. 3a). Upregulation of Ki67 was strongest in chemokine 195 receptor-positive CD8 + T cells (Fig. 3c), which is in line with strongest upregulation in 196 CD56 dim CD16 + chemokine receptor-positive NK cells (Fig. 2c), indicating a particular 197 activation of cytotoxic lymphocytes expressing lung-homing receptors in  In comparison to NK cells where upregulation of granzymes and perforin was more 199 uniform between COVID-19 and influenza patients (Fig. S3), differences were more distinct 200 for T cells (Fig. 3d-f). As to be expected, expression of cytotoxic effector molecules was to a 201 large extent contained to cytotoxic CD8 + T cells (Fig. 3d, f), although some expression was also 202 observed in CD4 + T cells both in COVID-19 and influenza patients, respectively, with a 203 particular bias towards chemokine receptor-negative CD4 + T cells (Fig. 3e, f). Expression of 204 granzyme B and perforin was highest in chemokine receptor-negative CD8 + T cells in healthy 205 controls as well as in COVID-19 and influenza patients (Fig. 3d, e). Importantly, however, 206 lung-homing receptor-positive T cells displayed the highest increase compared to the same 207 subset in healthy controls (Fig. 3f). 208 Together, these data indicate an overall stronger in vivo priming of T cells and NK cells 209 in influenza patients as compared to moderate COVID-19 patients and also suggest a specific 210 role for cytotoxic lymphocytes co-expressing lung-homing receptors. 211 212 Distinct CXCR3-and CXCR6-mediated accumulation of phenotypically primed NK cells 213

Activation of cytotoxic NK cells and CD8 + T cells in COVID-19 and influenza patients 215
was associated with high expression of effector molecules (Fig. S3, Fig. 3). Stratification of 216 cells based on expression of chemokine receptors, granzyme A, granzyme B, and perforin 217 revealed distinct co-expression patterns between CD56 bright CD16and CD56 dim CD16 + NK cells 218 and CD8 + T cells as well as between COVID-19 and influenza (Fig. 4a). Shared between the 219 different subsets and between COVID-19 and influenza was an increase of all effector 220 molecules particularly on chemokine receptor-positive cells. Increased expression of effector 221 molecules could also be confirmed at transcriptional level both for NK cells (Fig. 4b) and CD8 + 222 T cells (Fig. 4c) in peripheral blood of COVID-19 patients as compared to healthy controls (6). 223 Interestingly, levels of effector molecules gene transcripts were nearly identical between 224 ventilated (severe) and non-ventilated (moderate) COVID-19 patients, both in NK cells (Fig. 225 4b) and CD8 + T cells (Fig. 4c), indicating that similar NK and T cell activation is similar in 226 patients with moderate and severe disease. 227 The loss of NK cells and T cells expressing lung-homing receptors in the peripheral 228 blood of COVID-19 patients suggests migration of the respective cells to the infected lung 229 tissue. In order to identify characteristics of NK cells and T cells in the lung, we next used a 230 publicly available scRNAseq dataset from BAL fluid cells from COVID-19 patients (8)  of a large number of chemokines were found in patients with severe disease, while moderate 234 patients were mainly distinguished from healthy patients by a significantly upregulated 235 expression of CXCL10 and CCL5, in addition to increased levels of CXCL11 and CXCL16 (Fig.  236 4d), highlighting the role for CXCR3, CXCR6, and CCR5 in moderate COVID-19. In line with 237 these results, transcripts for CXCR3, CXCR6, and CCR5 were highly enriched in NK cells as 238 well as CD4 + and CD8 + T cells in COVID-19 patients with moderate disease (Fig. 4e). 239 Although it is possible that some of the cells in BAL fluid are comprised of tissue-resident NK 240 cells and memory T cells which express high levels of CXCR3 and CXCR6 at the 241 transcriptional and protein levels (11,12), our data strongly suggest infiltration of NK cells and 242 T cells from peripheral blood into the lung in COVID-19 patients. Since NK cells and CD8 + T 243 cells from peripheral blood displayed upregulated levels of effector molecules, these cells might human lung tissue explants infected with SARS-CoV-2 (16), and in the lungs of mice infected 288 lung-homing upon respiratory viral infection and, interestingly, also in non-viral lung tissue 290 injury (24). Elevated CXCL10 levels in BAL was associated with longer duration of mechanical 291 ventilation in . In mice, CXCR3-deficiency rescued CCR5-deficient 292 mice from IAV-induced mortality (25). Other murine IAV infection models demonstrated a 293 role for CXCR3 and CCR5 for NK cell lung-homing and showed NK cell accumulation in the 294 lung was not due to proliferation or apoptosis (23). For CD8 + T cells, murine models 295 demonstrated that virus-specific T cells express CXCR3 and migrate to CXCR3 ligands in vitro 296 (26). Furthermore, CCR5 is required for recruitment of memory CD8 + T cells to IAV-infected 297 epithelium and is rapidly upregulated on the surface of memory CD8 + T cells upon viral 298 challenge (26). Interestingly, these mouse models also revealed that CCR5 is required for 299 circulating CD8 + memory T cells to migrate to respiratory airways but not lung parenchyma 300 during virus challenge (26), indicating potential distinct migration patterns depending on 301 chemokine receptor expression. In addition to CCR5 and CXCR3, CXCR6 has been suggested 302 to be of importance for recruitment of resident memory T cells to the airways both in mice (12)  303 and in moderate . 304 The relevance of the CXCR3:CXCL10 axis for lung tissue-homing of cytotoxic immune 305 cells such as NK cells and CD8 + T cells might be of interest for future approaches of 306 intervention. Antibody-mediated targeting of CXCL10 improved survival of H1N1-infected 307 mice (27), revealing a novel approach for immunotherapy in patients with severe respiratory 308 viral infections. A thorough review recently summarized relevant known factors and cells 309 involved in lung-homing during infection with SARS-CoV-2 and influenza, emphasizing the 310 role of circulating NK cells not only in terms of their cytotoxic potential but also in potentially 311 facilitating the recruitment of other cell types such as neutrophils (28). The potential 312 immunoregulatory roles of NK cells in the human lung in health and disease however remains 313 to be studied further. 314

Despite the parallels between SARS-CoV-2 and influenza virus infection, both diseases 315
differ at an immunological level including that SARS-CoV-2 does not infect NK cells or other 316 mononuclear blood cells due to the lack of ACE2 expression (29), while influenza virus can 317 infect NK cells (30). In contrast to influenza virus, SARS-CoV-2 can spread to other organs if 318 not cleared efficiently from the respiratory tract (31), and the viruses induce different antiviral 319 responses in lung epithelial cells (32). 320 Here, we demonstrate common lung-homing potential of circulating NK cells and CD8 + 321 T cells in SARS-CoV-2 and influenza patients. A shortcoming of our study is a low number of 322 patients, impeding a detailed stratification by clinical or other parameters. However, the present 323 cohort was limited to clinically well-defined patients with moderate disease. Future studies will 324 assess how acute respiratory viral infection with SARS-CoV-2, influenza viruses, and other 325 viruses affects the landscape of activated NK cells and T cells in the lung. 326 Together, our results strongly implicate an importance of CXCR3 as a lung-homing 327 receptor in respiratory viral infections such as SARS-CoV-2 and influenza virus in humans. 328 The results also reveal a role for other receptors such as CXCR6, CCR5 on CD56 bright CD16 -329 NK cells, as well as CXCR2 on CD56 dim CD16 + NK cells, as potential alternative receptors of 330 importance. A better understanding of how these chemokine receptors are affecting disease 331 progression might help to develop future immunotherapeutic interventions in patients that 332 developed disease in current or future epidemics or pandemics with respiratory viral infections. 333