Prolonged activation of nasal immune cell populations and development of tissue-resident SARS-CoV-2 specific CD8 T cell responses following COVID-19

The immune system plays a major role in Coronavirus Disease 2019 (COVID-19) pathogenesis, viral clearance and protection against re-infection. Immune cell dynamics during COVID-19 have been extensively documented in peripheral blood, but remain elusive in the respiratory tract. We performed minimally-invasive nasal curettage and mass cytometry to characterize nasal immune cells of COVID-19 patients during and 5-6 weeks after hospitalization. Contrary to observations in blood, no general T cell depletion at the nasal mucosa could be detected. Instead, we observed increased numbers of nasal granulocytes, monocytes, CD11c+ NK cells and exhausted CD4+ T effector memory cells during acute COVID-19 compared to age-matched healthy controls. These pro-inflammatory responses were found associated with viral load, while neutrophils also negatively correlated with oxygen saturation levels. Cell numbers mostly normalized following convalescence, except for persisting CD127+ granulocytes and activated T cells, including CD38+ CD8+ tissue-resident memory T cells. Moreover, we identified SARS-CoV-2 specific CD8+ T cells in the nasal mucosa in convalescent patients. Thus, COVID-19 has both transient and long-term effects on the immune system in the upper airway.


Main 18
Individuals infected with SARS-CoV-2 suffer from a wide range of symptoms, ranging from 19 none to fever, cough and dyspnea and severe acute respiratory distress syndrome, which can 20 culminate in death 1 . The immune cell perturbations during COVID-19 have been described 21 extensively in blood, with changes observed in almost all immune cell populations that often 22 could be linked to disease severity. A global depletion of peripheral blood T cells has been 23 proposed to be a hallmark of COVID-19 2,3 . At the same time, neutrophil numbers are strongly 24 increased and an increased neutrophil to lymphocyte ratio is associated with poor prognosis 25 4 . Natural killer (NK) cells show an activated profile, associating with disease severity 5 . 26 Myeloid cells show an aberrant profile with increased proliferation and altered functionality 6,7 . 27 The B cell compartment is characterized by oligoclonal expansion of plasmablasts and 28 extrafollicular B cells 8,9 . Eosinophils were found to be decreased in blood in most patients at 29 time of hospital admission 10 . However, eosinophils might expand during hospitalization and 30 show upregulated levels of the homing marker CD62L and activation profiles, which was found 31 to precede lung hyperinflammation 11 . 32 Although SARS-CoV-2 mainly replicates in the respiratory tract, and lower respiratory tract 33 complications are major drivers of morbidity and mortality, it is unclear to what extent 34 immunological dynamics observed in blood can be translated to the respiratory tract. Indeed, 35 cytokines and antibodies do not seem to correlate between the nasopharynx and peripheral 36 blood during COVID-19 12 . Several studies already investigated mucosal immune responses 37 using either bronchoalveolar lavage (BAL) 13 or nasopharyngeal and oropharyngeal swabs 14 38 from hospitalized patients, demonstrating increased neutrophil levels and activated alveolar 39 macrophages/monocytes during COVID-19. T cell recruitment to the respiratory tract might be 40 beneficial, as an expansion of CD8 + T cell receptor (TCR) clones in BAL has been reported in 41 moderate compared to critical cases 13 . Another study with severe COVID-19 patients found 42 that increased CD4 + T cells in tracheal aspirates associated with survival 15 . In addition to 43 these findings, several other studies have used nasopharyngeal swabs (NPS) to analyse local 44 responses 16,17 . However, NPS collect cells only very superficially and mainly provide epithelial 45 and luminal infiltrating cells such as neutrophils and monocytes, while other immune cells, 46 such as T cells are incompletely captured. In contrast, BAL samples and tracheal aspirates 47 provide a clear picture of the lower airways 18-21 , but are difficult to collect longitudinally, after 48 recovery, or from healthy controls and patients that do not require intubation. Moreover, 49 studies performed on tissue are usually done in severe patients that died from the infection, 50 which creates a bias in the outcomes. As a result, we still have limited understanding of how 51 COVID-19 affects mucosal immunity 22 . In the current study, we aimed to characterize immune 52 cell dynamics in the upper respiratory tract mucosa during the acute phase and early and later 53 recovery of COVID-19 disease and assess whether alterations persist during convalescence. 54 To this end, we performed a prospective observational cohort study, for which we recruited 55 patients with PCR-confirmed SARS-CoV-2 infection after hospital admission (Figure 1a).

56
Nasal curettage samples from 20 patients during hospitalization with up to 4 samples per 57 patient were collect and analysed in-depth using mass cytometry. The earliest samples were 58 collected 11 days after symptom onset, which corresponded to 2 days after admission, while 59 the latest sample was collected 82 days after onset (61 days after admission there is no increase or decrease of total T cell numbers in the nasal mucosa is in contrast with 95 the general T cell depletion that has been observed in peripheral blood 2,3,6,7,23 . increased during acute infection and ERS compared to convalescent patients and to healthy 101 donors. As expected, subsets of granulocytes were also increased, although with slightly 102 different dynamics during acute infection: CD16 hi neutrophils were elevated during acute stage 103 and to a lesser degree in ERS (median of 121.7x and 12.8x increased compared to healthy 104 donors, respectively). CD16 dim neutrophils were even more strongly increased during acute 105 infection (median 256.8x) and ERS (median 11.9x). Such CD16 dim neutrophils might 106 correspond to activated neutrophils as previously they have been shown to correlate with 107 nasal myeloperoxidase levels and/or they might be related to immature banded neutrophils, 108 recently released from the bone marrow 24,25 . Furthermore, CD16 neg granulocytes, which may 109 in part consist of eosinophils, were increased compared to controls 254.1x and 24.3x during 110 acute stage and ERS, respectively. Although there was no overall change in CD4 + T cell 111 numbers, there was an significant increase (median 18.0x) in effector memory (EM, CCR7 -112 CD45RO + ) CD4 + T cells during the acute phase. And although not statistically significant, CD8 + 113 T effector memory cells re-expressing CD45RA (EMRA: CCR7 -CD45RA + ) also showed a 114 trend towards increased levels during acute infection compared to healthy donors (median 115 10.4x increased during acute infection). These findings agree with reports from peripheral 116 . CC-BY 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 April 22, 2021. ; blood T cell responses during SARS-CoV-2 infection describing that more specific CD4 + T 117 cells are being induced than CD8 + T cells, and that the majority of specific CD8 + T cells are of 118 a T EMRA phenotype 26,27 . Levels of these short-lived effector cells returned more or less to 119 numbers observed in healthy donors during ERS and convalescence. Finally, an additional 120 population that was increased in hospitalized patients (both acute and ERS), compared to 121 convalescent stage and healthy donors, were the CD11c + NK cells, which might indicate a NK 122 cell population with increased interferon-producing capacity and effector function 28 . Thus, 123 dynamic recruitment of various adaptive and innate populations mediating inflammation and 124 antiviral function to the upper respiratory tract was observed during hospitalization with cell 125 type numbers in convalescence that closely resembled levels measured in healthy donors. Of 126 note, we did not observe an increase in nasal B cells in hospitalized patients, which 127 corroborates observations that mucosal antibody levels seem reduced compared to systemic 128 titres in hospitalized patients 12,29 . 129 Next, we analysed whether immune cell types correlated with each other as well as with 130 clinical characteristics during COVID-19 ( Figure 3c). Focusing on samples from hospitalized 131 patient, a clear cluster of subsets with positively correlated frequencies was observed, which 132 included monocyte subsets, granulocyte subsets and CD4 + T EM and CD8 + T EMRA cells.

133
Most of these subsets negatively correlated with days since symptom onset and hospital 134 admission (see also the time-based plots in Supplementary figure 1), indicating that during 135 longer hospitalization the numbers of these cells gradually return to normal levels. Among all 136 cell types, the monocyte subsets, CD16 neg granulocytes and CD16 dim neutrophils correlated 137 positively with viral load (Figure 3c), suggesting that SARS-CoV-2 virus infection dose-138 dependently induced nasal recruitment of various immune subsets. We also correlated cell 139 numbers with measurements of disease severity obtained at the same day as nasal sampling, 140 including breathing rate, oxygen saturation and serum CRP concentrations. This revealed that 141 CD16 hi neutrophils numbers, as well as total numbers of granulocytes, in nasal mucosa 142 negatively correlated with oxygen saturation levels. As expected, oxygen saturation also 143 negatively associated with applied oxygen flow, but not with viral load. This suggests that 144 fluctuations in viral titers are not related to disease worsening, while the enhanced presence 145 of nasal granulocyte populations is, similar to what has been described for granulocyte 146 numbers in peripheral blood. To understand whether factors like sex, co-morbidities and 147 medication were drivers of nasal immune profiles, we performed multi-dimensional scaling 148 using all cell subsets (Supplementary figure 2). Acute patients clearly clustered separately 149 from healthy donors and convalescent patients, with ERS patients intermediate. Therefore, 150 we visualized covariates separately per group (acute, ERS, convalescent, healthy), showing 151 there was no clear clustering based on any of these covariates, although larger sample sizes 152 might be needed to conclusively exclude any such effects. 153 Subsequently, we looked more closely at phenotypic expression profiles on the differentially 154 abundant cell clusters and to what extent these profiles normalized after hospital discharge. 155 Although all monocyte subsets significantly increased during acute infection, patients had 156 relatively more CD163 + and fewer CD163 + CD206 + monocytes/macrophages compared to 157 healthy donors, which normalized during recovery ( Figure 4a). These CD206 + cells are likely 158 fully differentiated tissue-resident macrophages 30 . CD163 + monocytes were also found 159 abundantly in peripheral blood of COVID-19 patients as well as in autopsy lungs and likely 160 represent a recently recruited monocyte population, a hypothesis that was supported by 161 trajectory analysis (Figure 4b) 15 . Of note, we did not observe non-classical or transitional 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 April 22, 2021.

cells in peripheral blood resembling monocyte-derived suppressor cells has been previously 167
reported during severe COVID-19 31,32 . Our results suggest that these cells may rapidly seed 168 the upper airway mucosa where they might further differentiate and acquire a macrophage 169 phenotype. CD163 + CD206 + monocytes/macrophages also expressed elevated levels of the 170 IL-3 receptor CD123 during both acute phase and ERS, while CD163 + and CD163monocyte 171 subsets generally lacked CD123. The expression of CD123 on tissue macrophages has been 172 previously described in patients with histiocytic necrotizing lymphadenitis 33 . The functional 173 consequences of this upregulation of the IL3R on nasal macrophages during COVID-19 174 remain to be elucidated, but this finding further supports the importance of the IL-3/IL3-R axis 175 in COVID-19 34 . 176 We then investigated more closely CD16granulocytes, by further grouping them into 7 sub-177 clusters (Figure 4d). Sub-cluster 1, characterized by increased CD127 expression, was 178 significantly increased in patients during hospitalization (median 28.3%) but also during 179 convalescence (median 22.2%) compared to healthy donors (median 5.8%, Figure 4e,f). It 180 has been shown that engagement of IL-7 with its receptor CD127 on eosinophils leads to 181 increased survival and activation of eosinophils; and in airway allergen challenge in allergic 182 asthmatics, IL-7 levels in BAL strongly correlated with eosinophils 35 . Thus while total numbers 183 of granulocytes returned to similar levels as observed in healthy donors, alterations in their 184 phenotype, and possibly their functionality, remained visible during convalescence. The half-185 life of granulocytes is in the region of hours to days, suggesting either an ongoing recruitment 186 of altered cells or the continued local perturbation after or during entering the respiratory 187 mucosa 35 . 188 The subset of CD4 + EM T cells was also increased during acute infection. To investigate their 189 phenotype more closely, we analysed the expression of markers of activation (CD38, HLA-190 DR), exhaustion (PD1) or inhibition (CTLA-4) on these cells (Figure 5a). CD38 and CTLA4 191 were increased during the acute phase, but normalized at later timepoints, while HLA-DR and 192 PD1 remained expressed at a higher level even during convalescence as compared to healthy 193 donors. This induction of regulatory and inhibitory markers mirrors what has been described 194 in blood and likely reflects attempts of the immune system to restrain excessive activation.

195
Finally, we aimed to assess whether long-term protective CD8 + T cell immunity develops in 196 nasal mucosa to serve as gatekeepers against re-infections. Mouse models have shown that 197 nasal CD8 + tissue-resident memory (TRM) T cells specific for influenza persisted in nasal 198 mucosa following infection and efficiently controlled secondary infections 36 . The majority of 199 nasal CD8 + T cells in our samples highly expressed CD69, which is used to define resident-200 memory T cells 37 , while very few naïve CD8 + T cells were present. Although it is possible that 201 some activated CD8 + T cells upregulate CD69 without being true TRM, the cluster of CD8 + 202 TRM expressed very little KLRG1, which is congruent with a TRM phenotype (Figure 1c) 38 . 203 Sub-clustering of CD8 + TRM showed variable expression of activation markers CD38, HLA-204 DR and Tbet (Figure 5b). Within this embedding, CD8 + TRM from acute phase, ERS, 205 convalescent patients and controls clustered differentially, indicative of altered phenotypes 206 during and following COVID-19 ( Figure 5c). Indeed, sub-cluster frequencies significantly 207 differed between the groups (Figure 5d, e). Sub-cluster 5, marked by expression of HLA-DR, 208 Tbet and CD38, was increased in hospitalized patients, while sub-cluster 3, expressing only 209 HLA-DR and CD38, was higher in convalescent patients compared to healthy donors, as were 210 all CD38 + TRMs. Thus CD8 + TRM had an increased activation profile, which persisted at least 211 5-6 weeks after hospital discharge. To demonstrate antigen specificity, we then attempted to 212 compare the T cell receptor (TCR) repertoire in nasal samples from convalescent patients with 213 SARS-CoV-2 reactive CD8 + and CD4 + T cells isolated from paired peripheral blood 214 (Supplementary figure 4). For one severe, convalescent patient we obtained >10 unique TCRs 215 from both nasal cells and sorted SARS-CoV-2 specific peripheral blood cells (Supplementary 216 . CC-BY 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 April 22, 2021. ; https://doi.org/10.1101/2021.04.19.21255727 doi: medRxiv preprint figure 5). The TCR repertoire in the nasal mucosa of this patient was broad (218 unique TCRs), 217 including one dominant TCR accounting for 12.2% of all TCR reads. This TCR clone was also 218 present in paired peripheral blood SARS-CoV-2 specific CD8 + T cells (4% of reads), but not 219 in sorted CD4 + T cells (Figure 5f). As this sample was collected 2 months after viral clearance 220 and 80.4% of the nasal CD8 + T cells for this patient were of a TRM phenotype (Figure 5g), 221 this indicates that antigen-specific tissue-resident memory was induced. Of note, the number 222 of unique SARS-CoV-2 specific TCRs detected and their overlap between nose and peripheral 223 blood might be underestimated as blood CD8 + T cells were isolated based on reactivity 224 towards structural proteins, while CD8 + T cell reactivity is also directed against non-structural 225 proteins in severe COVID-19 patients [39][40][41] . Taken together, we demonstrated SARS-CoV-2 226 specific CD8 + T cells can persist in the nasal mucosa months after viral clearance, based on 227 their activated phenotype and overlapping TCR clonotype with SARS-CoV-2 reactive CD8 + T 228 cells in blood. This is suggestive of the establishment of local protective immune memory 229 responses that could rapidly control and attenuate re-infections by SARS-CoV-2. oxygen saturation in blood. We show several similarities with reports from blood, including a 237 strong CD4 + T EM cell response marked by markers of activation, but also exhaustion and 238 inhibition, as well as the presence of HLA-DR low monocyte subsets in the nasal mucosa. A 239 striking difference from studies characterizing blood was the absence of a general 240 lymphopenia in the nasal mucosa. We also provided further characterization of cellular 241 subsets infiltrating the mucosa during disease, such as CD11c + NK cells, CD123-expressing 242 differentiated macrophages and CD127 expression on CD16granulocytes. During early and 243 later recovery stages of disease most of the cell numbers progressively returned to levels 244 comparable to age-matched healthy donors. However, several phenotypic changes in nasal 245 immune populations persisted even during recovery. For example CD127-expressing 246 granulocytes remained elevated during convalescence, while also residual increased 247 activation of CD4 + and CD8 + T cells after hospital discharge was observed. Moreover, by 248 comparing nasal TCR clonotypes with antigen-specific cells from blood, we were able to 249 demonstrate that SARS-CoV-2 specific CD8 + T cells can seed the nasal tissue and these We thank all patients and healthy volunteers for taking part in this study. This work was 261 supported by a MKMD-COVID-19 grant (no. 114025007) from ZonMW and Proefdiervrij. This 262 work was also supported by Wake Up To Corona crowdfunding by Leids Universitair Fonds.

263
SPJ is supported by a LUMC Gisela Thier Fellowship. Figure 1 was partly made with 264 . CC-BY 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 . Samples were incubated for 30' at room temperature and then washed with 4mL Cell 317 . CC-BY 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 April 22, 2021. ; Staining Buffer (Fluidigm). Cells were spun 5' at 800xg, supernatant removed and 318 resuspended and combined into 3mL of Perm/Wash. Cells were spun again 5' at 800xg and 319 were resuspended in 45µL Perm/Wash. FcR block (Biolegend, 5µL) and heparin (0.5µL, 320 100U/mL) were added to prevent aspecific binding of antibodies and cells were incubated for 321 20' at room temperature 47 . Then 50µL of antibody cocktail (Table S1) was added, followed by 322 a 45' incubation at room temperature. Cells were then washed twice with 2mL Cell Staining 323 Buffer and spun down for 5' at 800xg. DNA was then stained overnight at 4°C using 1mL Fix

SARS-CoV-2 reactive T-cell isolation 353
PBMCs from convalescent COVID-19 patients were isolated from fresh whole blood using 354 Ficoll-Isopaque and cryopreserved until further use. PBMCs were thawed and immediately 355 used for overnight stimulation assay. For the stimulation assay, 1x10⁶ PBMCs were seeded    CC-BY 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)

. CyTOF analysis of nasal immune cells during and post COVID-19 infection. a)
The copyright holder for this preprint this version posted April 22, 2021. ;

Figure 1
. CC-BY 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.  . CC-BY 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)  . CC-BY 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.  . CC-BY 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 April 22, 2021. ; https://doi.org/10.1101/2021.04.19.21255727 doi: medRxiv preprint . CC-BY 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 April 22, 2021. ; https://doi.org/10.1101/2021.04.19.21255727 doi: medRxiv preprint