The emergence of goblet inflammatory or ITGB6hi nasal progenitor cells determines age-associated SARS-CoV-2 pathogenesis

Children infected with SARS-CoV-2 rarely progress to respiratory failure, but the risk of mortality in infected people over 85 years of age remains high, despite vaccination and improving treatment options. Here, we take a comprehensive, multidisciplinary approach to investigate differences in the cellular landscape and function of paediatric (<11y), adult (30- 50y) and elderly (>70y) nasal epithelial cells experimentally infected with SARS-CoV-2. Our data reveal that nasal epithelial cell subtypes show different tropism to SARS-CoV-2, correlating with age, ACE2 and TMPRSS2 expression. Ciliated cells are a viral replication centre across all age groups, but a distinct goblet inflammatory subtype emerges in infected paediatric cultures, identifiable by high expression of interferon stimulated genes and truncated viral genomes. In contrast, infected elderly cultures show a proportional increase in ITGB6hi progenitors, which facilitate viral spread and are associated with dysfunctional epithelial repair pathways. Graphical Abstract


Introduction
Chronological age remains the single greatest risk factor for COVID-19 mortality, despite the availability of effective vaccines. Children infected with severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) rarely develop moderate or severe disease [1][2][3][4] whilst the risk of mortality in infected people over 85 years old is currently as high as 1 in 10 [5][6][7] . The nasal epithelium is the first line of defence against inhaled pathogens and nasal epithelial cells (NECs) are the primary target of SARS-CoV-2 infection [8][9][10][11] . Infection of upper airway cells [12][13][14] can progress distally, leading to diffuse alveolar injury and long-term complications such as the development of lung fibrosis 15,16 .
It was initially thought that increased COVID-19 severity in the elderly may be due to greater availability of the angiotensin converting enzyme 2 (ACE2) receptor and the host transmembrane serine protease 2 (TMPRSS2) that aid viral entry to NECs compared to children. However, the differences in the expression of these host factors between children and adults are uncertain [17][18][19][20] . Current evidence suggests that children at least in part are protected from infection by a pre-activated antiviral interferon (IFN) state in the upper airway epithelium 19,[21][22][23] . Whilst this may partially explain why children are protected, it does not fully explain the gradient of disease risk seen among adults with increasing age. Understanding epithelial infection and repair in response to SARS-CoV-2 infection has the potential to provide novel therapeutic strategies not only to COVID-19, but also to any future emerging respiratory virus threats.
Here, we investigated the effects of early SARS-CoV-2 infection on human NECs from children (0-11 years old), adults (30-50 years old) and, for the first time, elderly (>70 years old) individuals. NECs were differentiated in air-liquid interface (ALI) cultures. ALI cultures were mock-or SARS-CoV-2-infected for up to 3 days to examine cell-intrinsic differences in function, viral replication, gene and protein expression (Fig. 1a). We reveal age-specific responses, with a strong IFN response in infected paediatric goblet inflammatory cells, and the appearance of elderly basaloid-like cells that sustain viral replication and are associated with fibrotic signalling pathways.

Cellular landscape of paediatric, adult and elderly-derived nasal epithelial cultures
We first investigated differences in the cellular landscape of nasal epithelial cultures with age. In total, we generated a single-cell RNA sequencing (scRNAseq) dataset of 139,598 cells, identifying 24 epithelial cell types or states (Fig. 1b) based upon key canonical marker genes, including the newly described transit epithelial cells (Yoshida et al., 2022) 19 (Extended Data Fig. 1a,b,c for quality control). We distinguish three broad cell domains: Interestingly, in healthy control cultures (non-infected), the proportions of many epithelial cell populations varied with age. Using our cell domain categories, we found that adult and elderly culture datasets contained a greater abundance of basal/progenitor (KRT5 hi ) subtypes (Fig. 1c), 36.5% and 35.3% respectively, compared to 13.8% in the paediatric cultures. This age-associated difference in basal cell numbers was also seen in an agematched healthy subset of our previous in vivo nasal epithelial dataset 19 (Fig. 1c), using label transfer of our ALI-model cell annotation (Extended Fig. 1d,e). Adult and elderly cultures contained the greatest proportion of Basal 1 and Basal 2 cells that were almost absent in paediatric cultures (Extended Fig. 1g). Despite this difference in the number of basal cells, all age-groups displayed similar apical differentiation, including mucus (MUC5AC) production and similar levels of cilia ( Fig. 1d and Extended Fig. 2a), with no significant difference in ciliary beat frequency (CBF) (Extended Fig. 2b) and no change in cellular motility (measured as the time taken for cells to cover a scratch) with age ( Extended   Fig. 2c). However, ALI cultures from elderly were thicker (mean±SD, 40±18 µm) (Extended Fig. 1k) than paediatric cultures (20±10 µm; p=0.02) with a distinct spiral morphology typical of ALI cultures (Extended Fig. 2e), which was absent in the paediatric cultures. This did not affect the integrity of the epithelial barrier as the trans-epithelial resistance (TEER) was comparable between age groups ranging from mean of 640-450 Ω .cm 2 (Extended Fig. 2f).
The most striking difference in the paediatric cultures was the greater abundance of goblet cells, particularly the Goblet 2 subsets (Extended Fig. 2g), which appear to represent a shift in cell state from Secretory cells (higher KRT5) identified in adult and elderly cultures, to Goblet cells which are higher in BPIFA1 (see similarity in markers gene expression in Extended Fig. 1d). Importantly, whilst there was no overall difference in the protein levels of SARS-CoV-2 entry factors in the cultures with age (Fig. 1e), the goblet cell types in the paediatric cultures showed highest expression of TMPRSS2 (Fig. 1f) which is necessary for SARS-CoV-2 spike protein priming 26 , and ACE2, which contains the SARS-CoV-2 spike protein binding site required for viral entry. The highest expression of these markers in the adult and elderly cultures is found in Secretory and Basal 2 cell types (Fig. 1f). This suggests a shift in susceptibility to viral infection from goblet to secretory cell types with age.

Increased infectious virus production in SARS-CoV-2 infected elderly cultures
To determine differences in viral replication between age groups, ALI cultures were infected with an early-lineage SARS-CoV-2 isolate (hCoV-19/England/2/2020; 4x10 4 pfu/well (~multiplicity of infection (MOI) of 0.01 pfu/cell)) for up to 72 hours (h) post infection (p.i). Our preliminary experiments (Extended Fig. 3a,b) indicated that over a 5-day infection period, SARS-CoV-2 replication peaked at 72h p.i; therefore, all subsequent investigations were completed prior to this time point. Infectious SARS-CoV-2 particles were measured by performing plaque assays on combined apical washes and cell lysates at 4-, 24-and 72h p.i.
We found no significant differences in the level of viral transcripts between age groups at any time point (Fig. 1g). However, we found that at 24h p.i SARS-CoV-2 viral reads were detected across more epithelial cell subtypes in adult and elderly cultures (7/24 and 11/24 cell types, respectively) compared to the paediatric cultures, in which viral reads were detected in only 3/24 cell subtypes ( Fig. 1h and Extended Fig. 3c). At 72h p.i the number of infected cell types increases in all age groups (Fig. 1h). This is supported by immunofluorescent analysis at 72h p.i showing greater viral spread (measured as %dsRNA+ signal coverage) in elderly (mean±SD 16.1%±9.5) compared to paediatric cultures (mean±SD 3.8%±3.1) (Fig. 1i). Of the cell types expressing viral reads at 72h p.i, we found that Ciliated 2 and Transit epi 2 cells had the highest proportion of viral reads irrespective of age (Fig. 1h). The transit epi cell subtype was characterised by the presence of basal (KRT5), goblet (MUC5AC) and ciliated (FOXJ1) markers (Extended Fig. 1d) suggesting the potential for both mucus and cilia production. Strikingly, goblet cell types appeared more infected in paediatric cultures, while adult and elderly cultures showed highest viral reads in secretory cell types ( Fig. 1h and Extended Fig. 3c,d,e). Transmission electron microscopy (TEM) demonstrated the presence of viral particles (red) in cells possessing both mucincontaining secretory granules and cilia (Fig. 1j, k).

Emergence of paediatric goblet 2 inflammatory cells in response to SARS-CoV-2, whilst infected elderly epithelium promotes basal cell proliferation.
We next profiled the phenotypic effects of infection on epithelial cells, using a diverse combination of live cell microscopy, immunofluorescence staining, proteomics, and gene expression analysis, and compared these across the age groups.
We also found a significant increase in epithelial escape from pseudostratified culture (cell protrusion) in elderly cultures compared to paediatric at 72h p.i with SARS-CoV-2 (mean±SD, 0.71±0.76 compared to 10.7±7.8; p<0.01, n=P7, E6) (Fig. 2e); in some instances these cells were heavily burdened with viral particles (Fig. 2f) and co-labelled for SARS-CoV-2 spike protein (Fig. 2g). Some protruded cells were shown to completely detach from the pseudostratified epithelium on the apical surface of the culture (Fig. 2h,   Extended Fig. 4e).
In terms of ultrastructural changes following infection, we observed mild abnormalities across all age groups, such as endocytosis of cilia basal bodies and sloughing of ciliated cells at 72h p.i using TEM (Extended Fig. 4f). However, this did not result in a significant decrease of ciliated cells (a-tubulin protein expression), changes in the ciliary beat frequency or entry factor protein expression over the 72h p.i study period (p>0.05) (Extended Fig.   5a,b,c,d). These data suggest that, although ciliated cells are a target of SARS-CoV-2 infection, there is no substantial loss of motile cilia within 72h p.i in our model. Proteomic data, including SARS-CoV-2 entry factor expression and mass spectrometry of apical supernatants and cytokine levels are shown in Extended Fig. 5e.
Using Milo 27 , a tool to test for differential cell state abundance, we investigated how the cellular landscape of the nasal epithelium changed following infection and whether this differed with age. In paediatric cultures, the airway epithelial cell type composition showed trends of decreasing basal, secretory and goblet cell neighbourhoods (blue) whilst Transit epi 2 and a population of terminally differentiated goblet cells neighbourhoods increased in frequency (Fig. 2i). The most striking cellular change following infection was the emergence of Goblet 2 inflammatory cells that were absent from mock-infected paediatric (but not adult or elderly) cultures (Fig. 2i, and Extended Fig. 5f,g). The Goblet 2 inflammatory cell type is strongly associated with type I IFN signalling, with higher levels of CXCL10, IFIT1 and IFIT3 than other goblet cell subtypes (Extended Fig. 1d). Whilst goblet inflammatory cells have previously been seen in vivo, it is interesting that this inflammatory phenotype is cell intrinsic and independent of immune cells that are not present in our culture system. We later (see next section) explore the impact of this on viral replication and spread.
Infected elderly epithelial cell cultures displayed an increase in basal (KRT5 hi ) cell neighbourhoods with SARS-CoV-2 infection compared to mock suggesting an elderlyspecific progenitor cell mobilisation (proliferation) following SARS-CoV-2 infection (Fig. 2j, adult dataset shown in Extended Fig. 5h). We noted a particular expansion of Basaloid-like 2 cell neighbourhoods (Fig. 2j). These recently identified cells are poorly described and characterised by markers associated with tissue injury and fibrosis (ITGB6, ITGB1, ITGAV, Fig. 1d). In healthy epithelial tissue, including skin and lung, integrin beta 6 mRNA is virtually undetectable 28 , but its expression is considerably upregulated during wound healing 29 , tumorigenesis and the development of fibrosis 30 . The presence of ITGB6+ cells are of major interest as they may be involved in exacerbation of disease in the elderly.
To determine the differentiation pathways of these cell types, we used pseudotime trajectory inference and predicted three terminally differentiated end points, including Goblet 2 inflammatory (paediatric) (Fig. 2k,l), Basaloid-like 2 (elderly) (Fig. 2m,n) and Ciliated 1 cells (see also Extended Fig. 5i). Alignment of estimated pseudotimes showed that immediate precursors for the Goblet 2 inflammatory cells are the Goblet 2 PLAU+ cells (Fig. 2l), while immediate precursors for the Basaloid-like 2 cells are the Basal|EMT2 cells (Fig. 2n).

Interferon dominates the SARS-CoV-2 response in paediatric cultures leading to truncated viral genomes and fewer infectious viruses.
As discussed above, SARS-CoV-2 infection of paediatric epithelial cells led to the emergence of a Goblet 2 inflammatory cell type that was absent in the mock-infected cultures and found in very few numbers in infected older age groups (proportion paediatric = 1455/1578, adult = 90/1578, elderly = 33/1578) (Extended Fig. 1g). The Goblet 2 inflammatory cell type has high levels of interferon-stimulated genes (ISGs), particularly ISG15, MX1, IF16, IFIT3, OAS1, OAS2, ICAM1, CXCL10 (Fig. 3a) and is strongly associated with type I IFN signalling ( Fig. 3b and Extended Fig 6a and

see Gene Set
Enrichment Analysis (GSEA) score in Extended Fig. 6b), shown to reduce COVID-19 severity 21,22,31 . We also applied GSEA to the apical secretome and found biological processes including humoral immune response and activation of immune response (Extended Fig. 6c,d). Counterintuitively, in paediatric derived cultures, viral reads were high within the Goblet 2 inflammatory and ciliated cells (Fig. 1h); indeed we show colocalisation of MX1 protein with SARS-CoV-2 spike protein (Fig. 3c). This suggests that in paediatric cultures the goblet cell types act as a primary cell target for SARS-CoV-2 infection. In support of this, we found the apparent precursor of the Goblet 2 inflammatory subtype, the Goblet 2 PLAU+ cells (Fig. 2l) displayed the highest co-expression levels of the serine protease TMPRSS2 and ACE2, that are required for SARS-CoV-2 entry compared to other cell types in the paediatric dataset (Fig. 3d). Following SARS-CoV-2 infection, the Goblet 2 inflammatory subtype also displays high levels of these entry factors across the dataset (Fig.   3d), indicating that this feature is not lost following differentiation. The gene expression profiles of these subtypes showed that viral replication and the interferon response are distinctive features of the Goblet 2 inflammatory subtype in these cultures (Fig. 3e).
To determine why, despite the high viral reads in paediatric cells, these cultures produce less infectious virions compared to elderly cultures (Fig. 1m), we looked at the distribution of RNA reads over the SARS-CoV-2 genome within the Goblet 2 inflammatory and ciliated cell types found in paediatric cultures. Here, we found that viral transcription within the Ciliated 2 cells, which display similar levels of viral reads across the age groups (Fig. 1h), showed high viral read expression towards the 3', underlying that ciliated cells are active sources of viral replication (Fig. 3f). However, in the paediatric Goblet 2 inflammatory cells, viral reads were uniquely highest near the 5' and not towards the 3', indicating these cells do not replicate virus successfully ( Fig. 3f and Extended Fig. 6e,f,g). We confirmed that this bias towards the 3' end is not a technical artefact due to introduction of the spike-in primer to increase detection of viral reads, as SARS-CoV-2 reads were successfully amplified without biassing viral distribution (Extended Fig. 6h,i,j). Using deep viral sequencing, we found a greater (p=0.042) quantity of non-canonical subgenomic SARS-CoV-2 RNAs (sgRNA) in paediatric and adult samples compared to elderly (Fig. 3g), which appeared to be predominantly driven by spike and ORF7a sgRNA (Fig. 3h). Whilst canonical sgRNA act as templates for mRNA during SARS-CoV-2 replication, noncanonical sgRNA can result in defective viral genomes (DVGs). These are by-products of the internal recombination processes that form part of normal viral replication. DVGs in other RNA viruses have been associated with antiviral immunity and increased interferon production 32,33 . In support of this, we also found greater non-canonical (based on the leader sequence at the 5' end) low frequency and fixed mutations in viral genomes produced by paediatric compared to elderly cultures ( Fig. 3i (Fig. 1m), these data suggest that Goblet 2 inflammatory cells may be directing DVG production, further enhancing the antiviral response in SARS-CoV-2 infected paediatric epithelium (Fig. 3k). This is further supported by our ultrastructural (TEM) observation that less viral particles are detected in goblet cells in paediatric cultures compared to neighbouring ciliated cells which were heavily burdened with virus, some of which appear to be within secretory mucin granules ( Fig. 3l and Extended Fig. 7).

SARS-CoV-2 infected elderly cultures contain Basaloid-like 2 cells expressing profibrotic and EMT gene signatures.
As discussed above, SARS-CoV-2 infection of elderly cultures resulted in a large compositional cell change in the KRT5 hi population, most notably the Basaloid-like 2 cells, which conversely decrease in number with infection in paediatric and adult cultures (Fig. 4a).
We also show that the Basaloid-like 2 cells are a terminally differentiated in vitro KRT5 hi population ( Fig. 2n) and that their differentiation is characterised by gene expression of profibrotic factors and markers associated with EMT, including ITGB6, VIM, FN1, ITGB1, ITGAV, and TGFB ( Fig. 4b and Extended Fig. 1d). ITGAV and ITGB6 and TMPRSS2 were also found to be increased in abundance as proteins in the supernatant of the SARS-CoV-2 infected elderly cultures ( Fig. 4c and Extended Fig. 8a). ITGB6 and ITGAV are glycoproteins which span the plasma membrane. Therefore their presence in the supernatant/secretome is likely to be from shed cells, exosomes or cellular debris. Vimentin was not found to be upregulated in the secretome, but was found to be significantly upregulated in Elderly cell lysates at 72h p.i with SARS-CoV-2 compared to mock (n=9; p<0.05) ( Fig. 4d and Extended Fig. 8b).
To investigate whether these ITGB6+, VIM+ and KRT5+ expressing cells were permissive to infection and could potentiate viral spread, we analysed cultures by immunofluorescence microscopy, in which we observed ITGB6 protein localisation with SARS-CoV-2 S protein ( Fig. 4e and Extended Fig. 8c,d) and the formation of a characteristic vimentin cage structure around SARS-CoV-2 S protein ( Fig. 4f and Extended Fig. 8e,f,g) 34  Interestingly, these Basaloid-like 2 cells show a low level of SARS-CoV-2 transcription (Fig.   1h). Therefore, it is likely that infected and damaged Basaloid-like 2 cells may be shed from the epithelium into the secretome (as hypothesised in Fig. 4h), supporting our evidence of increased ITGB6 protein (Fig. 4c) and increased number of protruding cells found in elderly (Fig 2e).

Dysfunctional repair and ITGB6 hi expression enhances viral replication in NECs.
To establish the role of the Basaloid-like 2 cells on SARS-CoV-2 pathogenesis we first applied GSEA to the differentially expressed genes in this cell type. The key biological processes associated with this cell type were extracellular matrix and structure reorganisation, response to wounding and a number of migration processes (Fig. 5a). Such processes may facilitate viral spread, metastasis and fibrogenic remodelling [35][36][37] .
Furthermore, Basaloid-like 2 cells showed upregulation of alternative SARS-CoV-2 entry receptors CTSL, FURIN, NRP1 and NRP2 (Extended Fig. 9a for comparison to other cell types), suggesting they could also be potential targets for infection and spread. This further highlights the upregulation of this cell type as a culprit for poor disease prognosis.
To investigate whether the Basaloid-like 2 cells gene signature can facilitate viral spread, we stimulated epithelial repair pathways, including mobility, using a wound healing assay (Fig.   5b). This assay increased expression of the Basaloid-like 2 cell type markers around the site of the wound, including increased expression of KRT5 protein (Fig. 5c, Fig. 9h).
Importantly, we found that stimulating wound repair correlated with an increase in SARS-

CoV-2 infection. The percentage coverage of dsRNA+ve cells (a marker of replicating virus)
was increased in wounded cultures (mean±SD, 4.09±3.61% to 9.69±9.04%; p=0.03, n=5) ( Fig. 5h and Extended Fig. 9i), particularly around the site of the wound (Fig. 5i,j). We also found that wounding cultures led to an increase in infectious viral particle production from donors that produced low levels of infectious particles (<10 4 pfu/donor 72h p.i) in the absence of wounding (Fig 5k).

Discussion
In this multi-disciplinary, comprehensive investigation of the early cellular response to SARS-CoV-2 infection of human NECs, we describe several novel age-associated mechanistic differences driving COVID-19 pathogenesis in the elderly. Our key findings are: • SARS-CoV-2 infected paediatric cultures induce an early, strong type I interferon response emerging from infected Goblet 2 inflammatory cells resulting in truncated viral genomes.
• SARS-CoV-2 infected elderly cultures produce more infectious virus across more epithelial cell subtypes compared to paediatric cultures.
• SARS-CoV-2 infected elderly cultures become thinner, leakier with increased cell shedding compared to controls and paediatric cultures. This leads to an increase in migrating KRT5+ and ITGB6+ Basaloid-like 2 cells with a gene signature associated with wound repair.
• Wounding cultures prior to infection stimulates expression of the Basaloid-like 2 cell gene signature (KRT5+ and ITGB6+). This leads to enhanced viral spread and increased infectious viral yield.

In vitro model and relevance to in vivo datasets
The Recent scRNAseq work found a stereotypical increase in upper airway progenitor basal cell types with age 21,38,39 . This finding was supported by our in vivo dataset of nasal brushings from healthy adults and paediatric donors 19 , which was further annotated in accordance to our ALI model cell type predictions. In the current study, we also found a greater abundance of basal (KRT5 hi ) subtypes in adult compared to paediatric cultures. One of the strengths of our in vitro approach is the ability to detect cell-intrinsic differences with age, without confounding factors of intra-individual variations in host immunity. The ability of nasal epithelial cells to recapitulate age-associated differences in vitro strongly suggests that these age effects are wired into the epigenome of epithelial cells.
The most striking differences we found across age groups was within the SCGB1A1 hi cell domain, which shifted from the goblet cell state in paediatric cultures to a secretory cell state with age. This is important as these cells expressed the highest amounts of SARS-CoV-2 entry factors (ACE2 and TMPRSS2), and may indicate a shift in viral susceptibility with age.
Indeed this was proved to be the case as, excluding ciliated cells, the Goblet 2 and Secretory cells were the primary target of SARS-CoV-2 infection in paediatric and elderly cultures, respectively, containing the highest proportion of these cells expressing viral reads after 72h p.i.

Cellular innate immunity in paediatric airways
Overall, fewer paediatric cell types expressed viral reads compared to those in the adult and elderly cultures, particularly at 24h p.i. This finding is supported by similar work where adult whilst discrepancies between viral RNA and infectious viral load have also been reported [45][46][47] . Some animal challenge experiments also support the conclusion that infectious viral production is higher with age 48 .

Damage and repair in elderly airways
Strikingly, SARS-CoV-2 infection of elderly NECs resulted in epithelial damage and early signs of epithelial repair, including cell migration and proliferation of basal NECs to repopulate damaged areas; events that were not detected in cultures derived from younger age groups. We also detected an increase in ITGAV, ITGB6 and VIM proteins which were attributed to the emergence of Basaloid-like 2 cells. ITGB6 encodes the beta 6 integrin subunit and exclusively partners with the alpha v subunit to form a heterodimeric α vβ6 integrin 49 . This complex is expressed exclusively on epithelial cells but is virtually absent or expressed at very low levels in normal healthy adult epithelium 28,50 . It is highly upregulated in response to epithelial injury 51 and is associated with disease progression in the setting of fibrosis and epithelial cancers and is therefore being actively pursued as a major drug target.
Integrins also play a direct role in extracellular signalling and modulating the expression of a number of cytokines and chemokines 52,53 , including and most notably, binding to and activating the potent cytokine TGF-β1, which has been widely implicated in the development of fibrosis. This axis has also been implicated in triggering EMT 51 , a process with distinct pathological roles in wound healing, tissue regeneration and organ fibrosis and cancer 54 . We therefore hypothesised that SARS-CoV-2 infected NECs undergo reprogramming by these mechanisms in an age-dependent manner and these processes contribute to COVID-19 pathogenesis by delaying disease resolution and enhancing viral spread.

Disease pathogenesis
Dysregulated epithelial repair processes have previously been identified in the pathogenesis of respiratory infections 55 . We hypothesise that the emergence of the Basaloid-like 2 cell type in SARS-CoV-2 infected elderly NECs drives epithelial-mesenchymal transition (EMT) repair pathways. EMT is a key developmental pathway, which allows for terminally differentiated epithelial cells to dedifferentiate and acquire a mesenchymal-like identity. 56 have been previously shown to induce EMT in cell lines and airway epithelia 57,58,59 . However, its role within viral infection, particularly in regards to age, is poorly understood. We show that SARS-CoV-2 infection of elderly cultures results in flattening of epithelial tissue, a decrease in transepithelial resistance, and increase in cell shedding, which are functional markers of the EMT process 60 and crucial steps in disease progression [61][62][63][64][65][66] .

Viral spread
One explanation for increased viral spread in elderly samples could be the direct interaction of the virus with ITGB6, which forms part of the caveolae, a special subcellular structure on the plasma membrane critical to the internalisation of various viruses and respiratory pathogens 51,[67][68][69][70] . Recently, a number of in vitro studies have suggested that SARS-CoV-2 spike protein interacts directly with integrins 71,72 , and suggest that they may serve as a viral entry route into non-ACE2 expressing cells, further potentiating infection in the elderly.
In summary, we have shown that SARS-CoV-2 exhibits differential tropism for nasal     Fig. 3a).

Viral genome read coverage
To visualise the viral read coverage along the viral genome we used the 10X Genomics cellranger barcoded binary alignment map (BAM) files for every sample. We filtered the BAM files to only retain reads mapping to the viral genome using the bedtools intersect tool [52].
We converted the BAM files into sequence alignment map (SAM) files to filter out cells that Analysis was performed using the FCAP software v3.0 (BD Biosciences).

Immunofluorescence confocal microscopy
For immunofluorescence confocal imaging, ALI cultures were fixed for microscopy using 4% ( with Nikon NIS-Elements analysis module. Imaris software (Bitplane, Oxford Instruments; version 9.5/9.6) was employed for 3D rendering of Immunofluorescence images.

Transmission electron microscopy
Cultured NECs that were either SARS-CoV-2 infected or non-infected were fixed with 4% Engine.

Sample preparation for single cell RNA sequencing
The ALI wells were processed using an adapted cold-active protease single cell dissociation protocol 76 , as described below, based upon that previously used 19

Library generation and sequencing
The

Single-cell RNA-seq computational pipelines, processing and analysis
The single-cell data were mapped to a GRCh38 ENSEMBL 93 derived reference, When examining viral load per cell type, we first removed ambient RNA by SoupX 77 85 . Leiden clustering with a resolution of 1 was used to separate broad cell types (basal, goblet, secretory). For each broad cell type, clustering was then repeated, starting from highly variable gene discovery to achieve a higher resolution and a more accurate separation of refined cell types. Annotation was first performed automatically using Celltypist 86 model built on the in vivo dataset of nasal airways brushes 19 and secondly using manual inspection of each of the clusters and further doing manual annotation using known airway epithelial marker genes.

Developmental trajectory inference
Pseudotime inference was performed on the whole object or the basal/goblet compartment

Differential abundance analysis
To determine cell states that are enriched in the SARS-CoV-2 versus mock conditions for the different age groups, we use the Milo framework for differential abundance analysis using cell neighbourhoods 27

Gene set enrichment analysis
Wilcoxon rank-sum test was performed to determine differentially expressed genes (DEGs) between clusters using scanpy.tl.rank_genes_groups() function. DEGs were further analysed using gene set enrichment analysis via ShinyGO 90 .

In vivo sub-analysis
Sex-and age-matched healthy adults and paediatric airway samples (n=10 total) were subsetted from our previous dataset 19 for purposes of label transfer of the in vitro cell annotation using CellTypist as described above. Selected sample IDs from the in-vivo dataset are shown in Table 3. These were selected to match the mean age and range and sex of the current study as the sample collection and processing was conducted in parallel between studies.

Conflict of interest statement
In the past three years, SAT has received remuneration for consulting and Scientific   SARS-CoV-2 (white arrows); Cilia (green arrows): secretory mucin granule (blue arrows) and with viral particles false coloured with red to aid visualisation. (l) SARS-CoV-2 protein abundance in apical fluid (extracellular) and cell lysates (intracellular) from SARS-CoV-2 infected NECs for 72h p.i as determined by mass spectrometry. Data shown as mean abundance of protein (dot size) and mean fold change in protein abundance per donor from mock infected NECs (colour scale) (n= 5P, 5A, 5E). (m) Infectious viral titres in combined cell lysate and apical fluid of SARS-CoV-2 nasal epithelial cells from paediatric (P), adult (A) and elderly (E) donors as determined by plaque assays (n= 13P, 8A, 8E). Subject to twoway ANOVA with Tukey's multiple comparisons test. showing the log fold changes observed when comparing SARS-CoV-2 infected versus mock conditions in the paediatric (i) and elderly (j), with a significant enrichment of Goblet 2 inflammatory cells and Basaloid-like 2 cells respectively observed with infection (colour grey is non-significant, colour red is significantly increased, colour blue is significantly decreased at FDR 10%). In the UMAP nodes are neighbourhoods, coloured by their log fold change when comparing SARS-CoV-2 infected versus mock conditions. Node sizes correspond to the number of cells in a neighbourhood and node layout is determined by the position of the neighbourhood index cell in the UMAP. Beeswarm plot shows the distribution of log-fold change across annotated cell clusters when comparing Paediatric-SARS versus Paediatricmock groups. DA neighbourhoods at FDR 10% are coloured with increased log fold change in red and decreased log fold change in blue.       The difference in wound closure per hour between mock and SARS-CoV-2 infected from the same donor (n= P8, A5, E4), subject to one-way ANOVA with Tukey's multiple comparison test. (i) dsRNA coverage for NECs irrespective of age group at 72h p.i. Determined by percentage area covered with dsRNA signal (yellow) from maximum intensity projections of fixed NECs. Subject to ratio paired t test (n=5). (j) Representative immunofluorescence images of Basaloid-like 2 cell markers ITGB6 (cyan), KRT5 (white), dsRNA (yellow) and F-actin (phalloidin; magenta) in SARS-CoV-2 infected NECs. Maximum intensity projection images from wounded cultures after 24h, shown both as maximal projections (top) and as an orthogonal view (bottom). KRT5 (white) is omitted from composite images, so that overlap of ITGB6 (cyan) and dsRNA (yellow) is apparent (white). Scale differs and scale bars are given on each image. (k) Infectious viral titres at 72h p.i in combined cell lysate and apical fluid of SARS-CoV-2 nasal epithelial cells from non-wounded (-) and wounded (+) donors that were previously shown to propagate low levels of infectious particles (<10,000 pfu/donor at 72h p.i.). Infectious viral load in combined apical and cell lysates (pfu/donor) were determined by plaque assays, representative plaque assay wells are shown (bottom). Subject to ration paired t test (n=6).   Tubulin+ve signal (cyan) from maximum intensity projections of fixed NECs using threshold analysis (red) in ImageJ. Summary of cilia coverage (right) (n= P5, A4, E5). Subject to one-way ANOVA with Tukey's multiple comparison test. For 72h p.i (mock and SARS-CoV-2 conditions) NECs were compared between age groups, looking at (b) cilia beat frequency (Hz) (n= P8, A8, E8); (c) α -Tubulin protein expression (n= P11, A8, E7) and (d) SARS-CoV-2 entry factor protein expression (ACE2, TMPRSS2 and short ACE2). Protein levels were determined via Western blot and normalised to GAPDH (n= P6-8, A5-9, E5-8). (e) Volcano plot of the apical secretome showing differential expressed proteins per age group between mock and SARS-CoV-2 infected cultures that were unique (highly expressed) to each age group (as shown in colour legend). Blue text highlights those that are highly expressed in mock compared to SARS-CoV-2 infection conditions and black text enriched with infection. (i) Representative Immunofluorescent images from 72h p.i NECs with SARS-CoV-2 without (top) and with (bottom) wounding stained for ITGB6 (cyan) and dsRNA (yellow). Percentage area covered (right) with ITGB6+ve signal or dsRNA+ve signal from maximum intensity projections of fixed NECs using threshold analysis (red) in ImageJ, the percentage coverage is given at the bottom right of each image.