A spatial multi-omics atlas of the human lung reveals a novel gland-associated immune niche

doi:10.21203/rs.3.rs-1650134/v1

Download PDF

Article

A spatial multi-omics atlas of the human lung reveals a novel gland-associated immune niche

https://doi.org/10.21203/rs.3.rs-1650134/v1

This work is licensed under a CC BY 4.0 License

Journal Publication

published 21 Dec, 2022

Read the published version in Nature Genetics →

Version 1

posted

You are reading this latest preprint version

To (re)define tissue architecture of the lung and airways at the cellular and molecular level, we profiled five proximal-to-distal locations of healthy human lungs in depth using multi-omic single cell/nuclei and Visium Spatial Transcriptomics. Using computational data integration and analysis, we extend beyond the suspension cell paradigm of lung atlases to date, to define and discover macro and micro-anatomical tissue compartments. We describe novel cell types and states in vascular, stromal and nerve bundle microenvironments. From our spatial transcriptomics, we discover and validate a novel survival niche for IgA plasma cells in the airway submucosal glands (SMG). In this niche we define a supporting role for SMG epithelial cells in mucosal immunity through recruitment and maintenance of IgA plasma, B and CD4 T cells locally at the airway SMG. We identify an immune-supporting role for SMG duct and serous cells with distinct signalling circuits to recruit B cells and IgA plasma cells, promoting longevity and antibody secretion through expression of CCL28, APRIL and IL6. We find high expression of MHC-II in SMG duct and serous cells, which are localised closely with memory CD4 T cells, suggesting local modulation of antigen specific immune responses locally at the glands. This new tissue microenvironment, which we term the “gland-associated immune niche” (GAIN) has major implications for respiratory immunity and infection response. Our single cell and spatial data is available for download and query at lungcellatlas.org.

Lung disease is the third-largest cause of mortality worldwide¹, making a better understanding of the tissue microenvironments and cell types that define lung function imperative. In addition to its main role in gas exchange, the lung has an important barrier function. Whilst other mucosal barrier tissues such as the gut and nasopharynx orchestrate adaptive immunity through well defined mucosa-associated lymphoid tissue (MALT), such secondary lymphoid structures have not been reported in the healthy human lung². The LungMAP and Human Lung Cell Atlas consortia^3,4, have harnessed recent technology advances in single cell and single nucleus RNA sequencing (sc and snRNA-seq)⁵ and generated a number of atlases characterising lung cell types in mice^6–8 and humans in health^1,9−15 and disease^{9–12,14,16–22} to dissect the roles of cell types within the lung.

Current atlases have prioritised parenchyma tissue, but few studies have examined the full tissue depth of the bronchial airways and trachea. Here we carried out deep tissue profiling from deceased organ donors’ healthy lungs and airways, allowing comparison of cell types along the proximal to distal axis of the respiratory tree. We present unbiased spatial transcriptomics as a first for the healthy human lung and airways to contextualise cell types and states within tissue microenvironments, adding a key dimension to the Human Lung Cell Atlas. In total, we profiled the transcriptome of 129,340 single cells and 63,768 single nuclei and performed Visium Spatial Transcriptomics on 20 tissue sections from human trachea, bronchi and upper and lower parenchyma. We make this data available at lungcellatlas.org as an interactive resource for data download, single cell and spatial gene expression analysis as well as automated cell type annotation using the CellTypist algorithm. In our sc/snRNAseq and spatial data we distinguished 79 cell types and states, including 11 novel populations not annotated in previous lung atlas studies. Many of these populations express disease associated genes and, in some cases, are significantly associated with disease in functional GWAS analysis. Our in-depth tissue profiling coupled with spatial genomics reconstructs known micro-anatomical tissue microenvironments in the lungs and airways at full molecular breadth for the first time. Going beyond known units of cellular organisation, we identify a novel immune niche for IgA plasma cells at the airway SMG. We highlight a novel role for SMG epithelial cells in recruitment and maintenance of IgA plasma and B cells and modulation of CD4 T cells responses through MHCII antigen presentation. IgA secretion plays an important role in respiratory immunity and our findings have major implications for lung disease, especially respiratory infections such as COVID-19²³.

A multi-omics transcriptional atlas of the human lung and airway with spatial resolution

To generate a comprehensive representation of the cell types in human lung and airway we applied scRNA-seq and snRNA-seq, VDJ sequencing and spatial transcriptomics to deep tissue samples (Fig. 1a, Tables S1-3). Tissue pieces (1–2 cm³) from healthy deceased transplant organ donors without previous lung disease were collected from the trachea, bronchi at the 1st or 2nd generation airway (Bronchi 2–3), bronchi at the 3rd to 4th generation airway (Bronchi 4), parenchyma from the upper left lobe (UpperPar) and the lower left lobe (LowerPar) to capture deep structures such as cartilage, muscle and the submucosal glands in the airways (Figure S1a).

In total 193,108 high quality transcriptomes were annotated into transcriptionally distinct broad cell type groups: epithelial (secreting, basal, alveolar type 1 and 2 - AT1 & AT2 - pneumocytes, ciliated, and SMG cells), immune (lymphoid, myeloid, B-cells, plasma, mast cells), erythroid, endothelial (vascular- VE, lymphatic - LE) and stromal cells (smooth muscle, fibroblasts, mesothelial and chondrocytes). These cell types were annotated according to widely accepted marker genes and consensus naming from other lung studies including the integrated Human Lung Cell Atlas¹⁴ and LungMAP²⁴ (Fig. 1b, Figure S1b, Supplementary Table 4). We combined the analysis of suspension cells and nuclei with Visium Spatial Transcriptomics (ST) analysis to map the identified cell types back to tissue structures on 20 tissue sections from 5 locations (Figure S2). Whilst Visium ST does not yield single-cell resolution, we annotated distinct micro-anatomical tissue environments (Supplementary Table 5) to assess their transcriptomic make-up and test for co-location of cell types. As expected, we mapped ciliated epithelial cells to the lumen of the airway surrounded by basal cells, and AT1 and AT2 to lung parenchyma (Fig. 1c, d), using the Cell2location²⁵ algorithm. In order to examine the benefits of different protocols and sampling locations, we used a Poisson linear mixed model to assess the relative effects of these variables on cell type composition (see Methods) (Fig. 1e). As expected, different enzymatic treatments enriched for specific cell type groups but had little effect, less than 1% of the total variance, on gene expression (Figure S1h).

Highlighting our comprehensive sampling approach, we transcriptionally defined chondrocytes for the first time in human lungs (ACAN, CHAD, COL9A3, HAPLN1 and CYTL1) (Figure S1c, d) and mapped these cells to the airway cartilage with Visium ST (Fig. 1c, d). Chondrocytes were mostly released using single nuclei sequencing from trachea (Fig. 1e, Figure S1e-g), demonstrating the utility of our multi-omics, multi-location atlasing of the human lung.

Overall, we generated a large scRNA-seq and snRNA-seq dataset with wide representation of distinct cell lineages and matching spatial gene expression data, allowing further exploration of cellular heterogeneity, communication and spatial dynamics to better understand lung function.

Identification of rare fibroblasts with immune recruiting properties

The sequential clustering of lung and airway fibroblasts identified 11 cell types with distinct marker genes (Fig. 2a, Figure S3a, b). We annotated previously described myofibroblasts, mesothelial, adventitial and alveolar fibroblasts ^9,24,26−29, as well as multiple novel and less studied cell types that were not predicted with existing annotations, specifically in the airways (Figure S3c, d, e, Supplementary Table 4). While most of our cell types were isolated from all locations and protocols, there were clear differences in cell type proportions (Fig. 2b, Figure S3b). Adventitial fibroblasts (Fibro-adv) were enriched in the airway samples, while the alveolar fibroblasts (Fibro-alv) were depleted. We further observed a rare cell type recovered only from the single cells, but not from single nuclei. This rare type expressed the chemokines CCL19 and CCL21, also highly expressed in fibroblast reticular cells responsible for T-cell positioning in the lymph nodes, and GREM1 ³⁰, CXCL13 and FDCSP, a follicular dendritic cell (fDC) marker responsible for B-cell positioning ^31,32 (Fig. 2c). These cells mapped to an area of immune cell infiltration on the bronchus with Visium ST, as well as with smFISH methods (Fig. 2d, Figure S4a, b) and were therefore annotated as immune recruiting fibroblasts (IR-fibro). Immune infiltrates were only found in a proportion of the donors and sections, but in keeping with the expected frequency of immune aggregates in donors with and without smoking history³³. In support of IR-fibro similarity to stromal cells from secondary lymphoid organs, the gene signature of germinal centre fibroblast cell types from the Peyer’s Patches of human gut³⁴ also map to the area of the immune infiltrate on lung Visium ST sections (Figure S4b). In conclusion, we describe an IR-fibro population with a likely role in immune cell recruitment to rare immune infiltrates in the healthy bronchus.

Perichondrial and peribronchial fibroblasts reveal disease associations

We next examined two fibroblast populations enriched in the trachea which we name peribronchial (PB-fibro) and perichondrial (PC-fibro) fibroblasts (Fig. 2b) based on their mapping in Visium ST data. For the first time, we provide the human transcriptome for PB-fibro with COL15A1 and ENTPD1 markers and localisation around the airway epithelium (Fig. 2e, Figure S5a). Functional GWAS analysis, that quantifies systematic associations between cell-type specific gene expression and disease-associated SNPs³⁴, demonstrated that PB-fibros are linked to chronic obstructive pulmonary disease (COPD) (Fig. 2f). PB-fibro also express LGR5, as reported in mice^35,36, which suggests the potential for analogous mesenchymal stem cell function in humans (Figure S5b), and share characteristics with the recently described LRG5 + fibroblasts in distal airways¹⁵. A COPD-related increase in WNT-5A downregulates Lgr5 ³⁷, which suggests PB-fibro could be targeted therapeutically to inhibit regenerative failure in COPD specifically in the airways.

The second airway-enriched cell type, PC-fibro, was detected only in the single nuclei data (Fig. 2b), likely due to difficulties in dissociating cartilage. The cell type was located around the chondrocytes within cartilage in our Visium ST (Fig. 2g) and could be identified by the expression of the marker gene COL12A1 in the human protein atlas (Figure S5c). Among the marker genes for PC-fibro, we detected bone development genes LGR4 and LGR6 ³⁸ (Figure S5b). PC-fibro, which expressed standard fibroblast markers, uniquely expressed genes relevant for bone development, placing these as an intermediate cell type in a trajectory from adventitial fibroblasts to chondrocytes (Figure S5d, e). Finally, PC-fibro markers were enriched in genes causing skeletal abnormalities in humans (Figure S5f), including FLNB and FGFR2 that are mutated in rare diseases causing skeletal malformations ^39,40. This cell type could be relevant in bone differentiation, in supporting cartilage functions, and in skeletal abnormalities. In conclusion, we have newly identified perichondrial fibroblasts around the healthy human airway cartilage.

Resolution of four distinct cell types in peripheral nerves from the airway

Finally, in the fibroblast compartment we identify four rare cell clusters within the peripheral nerves in the airway: myelinating Schwann cells (NFASC, NCMAP, MBP, PRX), non-myelinating Schwann cells (nmScwann) (NGFR, SCN7A, CHD2, L1CAM, NCAM1) ^41–44, endoneurial nerve-associated fibroblasts (NAF) (SOX9, OSR2)⁴¹ and perineurial NAF (SLC2A1, ITGA6)^41,45 (Figure S6a). Localisation of these cell types in peripheral nerves is shown by analysis of bulk RNA-seq across tissues (Figure S6d), Visium ST (Fig. 2h, Figure S6e), protein staining (Figure S6f-h) and smFISH (Fig. 2i, Figure S6i). Specifically, we show the expression of perineurial NAF around, and endoneurial NAF along Schwann cells inside the nerve bundle in human airway samples. This further confirms the identity of myelinating and non-myelinating Schwann cells with marker gene set enrichment in myelination and cell adhesion categories respectively (Figure S6b, c). This analysis further reveals EVX1 - a homeobox gene with a role in spinal cord development - as a potential regulator of myelinating Schwann cells. NmSchwann cells expressed some genes involved in Schwann cell development, maintenance (SOX10) and myelination (ERBB3 and LGI4) (Figure S6a, Fig. 2i, Figure S6i). Finally, we observed specific expression of rare peripheral nervous system (PNS) disease genes in these populations (Figure S6j). Thus, our atlas discerns micro-anatomical location within peripheral nerves (Fig. 2j) and pinpoints disease associations, even for very rare cell types. The cell types described here exist in other tissues; therefore this data could have implications outside the lung.

Resolution of the vascular cell types in the systemic and pulmonary circulation

We next focused on the vasculature, which in the lung consists of both systemic circulation, providing oxygen to the tissue, and pulmonary circulation where gas exchange occurs. Uniquely in our dataset, we could distinguish clusters of pulmonary and systemic circulation by the distribution of cell types in the distinct locations we sampled, since pulmonary cells are enriched in parenchyma and systemic cells are enriched in the trachea (Fig. 3d, Figure S7e). We observed previously described endothelial venous (E-Ven-syst, E-Ven-pulm) and capillary (Cap, Car4-Aerocyte) cell types (Fig. 3a, Figure S7a)^46–49. In addition, we now distinguish endothelial arterial cell types (systemic E-Art-syst and pulmonary E-Art-pulm), which have distinct localisation on Visium ST with E-Art-syst enriched in the trachea (Fig. 3a, b). The smooth muscle cells included non-vascular airway smooth muscle cells (ASM), pulmonary perivascular cell types (smooth muscle SM-pulm and pericytes Peri-pulm)⁹ and systemic perivascular cells (smooth muscle SM-Art-syst, pericytes Peri-syst and venous perivascular cells IR-Ven-Peri) (Fig. 3a, c Figure S7b). ASM was mainly captured in single nuclei data (Fig. 3d) enabling the detection of reliable marker genes for non-vascular smooth muscle cell types across tissues (Figure S7c, d).

In addition, we identified another perivascular cell type in the airways, expressing ABCC and ICAM1, but not CSPG4 - in a pattern similar to the postcapillary venous perivascular cells with a role in the homing of immune cells to peripheral lymph nodes^50,51 (Fig. 3c). This cell type expressed known leukocyte-recruiting chemokines (Fig. 3e) and co-localised with a venous endothelial vessel (ACKR1+) in the Visium ST bronchi section (Fig. 3f) and in smFISH microscopy (Fig. 3g, Figure S7f). These cells were therefore annotated as immune recruiting venous perivascular cells (IR-Ven-Peri). The potential lymphocyte recruiting capacity in both venous endothelial and IR-Ven-Peri (Figure S7g) suggests a role in lymphocyte extravasation in the airway veins, similar to the postcapillary venules in peripheral lymph nodes.

In summary, we distinguish between cells of the systemic and pulmonary circulation, and describe a novel immune recruiting postcapillary venous pericyte (IR-Ven-Peri). We provide marker genes for all of these populations, map these cells back into the tissue context and thereby further define the relationship between the endothelial and perivascular cells in the systemic and pulmonary circulation (Fig. 3h).

Identification of duct cells in human airway submucosal glands (SMG)

Further analysis of the airway epithelial compartment (excluding AT1 and AT2 cells) identified expected ciliated, goblet, basal, suprabasal, dividing basal, club, SMG basal, mucous and serous populations along with rare deuterosomal, ionocyte & brush and neuroendocrine cells. In addition to these, we identified a novel population between serous, mucous and club/goblet cells (Fig. 4a, Figure S8a, b) with marker genes overlapping with SMG duct cells in human oesophagus and preferential localisation in trachea^12,52. Thus, we hypothesise that these cells are the columnar non-ciliated duct cells of the human airway, previously only characterised in mice at the single cell level^53–55. smFISH staining for ALDH1A3, MIA and RARRES1 validated localisation at the SMG and distinguished these cells from serous, mucous and other epithelial cells (Fig. 4b, Figure S8e, f). Cell2location integration of single-cell and Visium data could also distinguish distinct locations of SMG duct cells compared to SMG mucous and serous cells, providing orthogonal evidence of the identification of a new, distinct cell type (Figure S8g). In addition, Velocyto analysis suggested that these cells may lie on a trajectory towards the surface epithelial populations (Figure S8c), consistent with a potential role in airway surface regeneration, as has been demonstrated in mice ^56,57.

An additional population was revealed from snRNA-seq, exhibiting both basal epithelium (TP63, KRT14) and muscle (ACTA2, TAGLN, CNN1 positive, but DES negative) markers (Fig. 4a, Figure S8a, j) with localisation around the glands ( Figure S8g-i), consistent with a myoepithelial cell signature previously reported as a small cluster of cells²¹. For this population we identified markers for cell-cell adhesion (FHOD3, LAMA1) as well as for nerve synapse signalling (NTRK2, PLD5). Co-localisation of FHOD3, epithelial marker KRT14 and muscle markers around the glands confirmed the myoepithelial cell signature by smFISH (Figure S8h) and in the Human Protein Atlas (Figure S8i). Interestingly, mouse myoepithelial cells have also been shown to regenerate the surface airway epithelium ⁵⁶. This cell type is not as well described in humans, potentially due to difficulties in dissociating this cell type from the airways consistent with the exclusive recovery of myoepithelial cells from nuclei data. Overall, we provide the human transcriptome for SMG populations and mapped the constituents of the SMG and their spatial location using both spatial transcriptomics and smFISH.

Proportions and gene signatures of airway epithelial cells

Using statistical modelling that accounted for material, donor and dissociation protocol (see Methods), we examined the proportions of airway epithelial cells at the 5 different locations sampled (Figure S8d), with significance given as local true sign rate (LTSR). As expected, club cells were enriched in the parenchyma, whereas SMG epithelial cells were enriched in the trachea. While donor and protocol had little effect on the epithelial cell type proportions, the starting material of cells versus nuclei strongly influenced cell type composition (Figure S8d). SMG epithelial cell capture was increased in nuclei, while the capture of other epithelial cell types was better achieved with the dissociation into single cells (Figure S8d).

Taking advantage of our multi-location data, we compared gene expression for ciliated cells, as one of the most abundant cell types present, across the five locations. We avoided any artefacts due to differing ambient RNA contamination between locations using only snRNA-seq data where samples were pooled across locations for the 10X sequencing reaction (Fig. 1a). Using a linear mixed model⁵⁸ (Methods) we detected 80 differentially expressed genes in ciliated cells from trachea compared to other locations (Local True Sign Rate (LTSR) > 0.9), many of which are upregulated in nasopharyngeal carcinoma including FBXL7, TSHZ2 and RAET1E (Figure S8k)^59–61. We also examined ACE2 expression, a SARS-CoV-2 entry gene, which we found to be highest in ciliated cells from the trachea, and lower in ciliated cells in the distal regions of the lung, where expression of ACE2 is likely to be more relevant in AT2 cells as reported previously (Figure S8l)⁶².

Altogether, we describe transcriptomes for key cell types involved in the SMG structure in the human airways which also functions as an immunological niche which we describe below.

Immune cells in the lung and airways

Myeloid cells show previously undescribed heterogeneity

For the immune compartment, we identified all major populations including myeloid, T&NK, B lineage, mast cells and megakaryocytes which were analysed separately to reveal previously undescribed heterogeneity, especially in the myeloid cells (Fig. 4c, Figure S9a-c, j). We found all major myeloid cell types (DCs, monocytes and macrophages) including many known and novel subsets. Previously identified macrophage subsets included intravascular macrophages (expressing LYVE1 and MAF)^63,64, Macro-AW-CX3CR1 ^63,65−67, Macro-CHIT1 (CHIT1 expressing with roles in asthma, COPD and lung fibrosis)^9,68 and interstitial macrophages (Macro-interstitial expressing chemokines CXCL9, 10 and 11)⁶⁴. We identified a new cluster expressing both monocyte CD14 and macrophage markers, termed Macro-intermediate (Figure S9a). Among alveolar macrophages, two more clusters appeared: dividing cells (Macro-alv-dividing), and a novel cell cluster expressing metallothioneins (Macro-alv-MT) including MT1G, MT1X and MT1F. Metallothioneins have a role in binding and metabolising metal ions⁶⁹, in immunity and stress responses^70,71, and therefore this population may have a function in response to air pollution. Finally we identified a rare novel population of macrophages expressing chemokines including CXCL8, CCL4 and CCL20 and was named Macro-CCL. While the expression of CCL4 was previously identified in interstitial macrophages⁶⁴, the expression of CXCL8 and CCL20 distinguishes this novel subset. Dysregulation of CXCL8 expression is associated with multiple lung conditions including infection, asthma, IPF and COPD⁷² and was identified as a marker for a separate macrophage population in psoriatic skin⁷³.

Overall, we have identified multiple known and novel myeloid populations in the healthy human lungs and airways, many of them expressing specific sets of chemokines, orchestrating the complex lung immune homeostasis.

Different subsets of T & NK cells in the lung and airways

T lymphocytes and natural killer (NK) cells included all major cell types (CD4, CD8, mucosal-associated invariant T (MAIT), NK, NKT, innate lymphoid cells (ILC)) and their subsets (Fig. 4c, Figure S9b). In the CD4 compartment we distinguished naive/central memory (CD4-naive/CM), effector memory/effector (CD4-EM/Effector), regulatory T cells (Treg) and tissue resident memory (CD4-TRM) cells. Within CD8 cells we found gamma-delta T cells (γδT), TRMs (CD8-TRM)⁷⁴ and two distinct clusters analogous to populations found the lung in cross tissue analysis⁷⁵ expressing CX3CR1 and GZMB (CD8-EM/EMRA) and CRTAM and GZMK (CD8-TRM/EM). The CD8-TRM cells, but not other CD8 subsets, nicely localised to airway epithelium in our spatial data as previously reported in the literature (Figure S9d)^76,77. NK subsets included clusters with markers ITGAD/CD11d, LAG3, KLRC3/NKG3E, KLRC2/NKG2C (NK-CD11d), CD16+ (NK-CD16hi) and CD56 bright NK cells (NK-CD56 bright)^78,79. NK cells positive for CD11d are activated in response to viral and other infections in both mouse and human ^80–82 and were previously shown in human blood⁸³. To our knowledge this is the first time this subset has been described in the healthy human lung.

The T and NK cells displayed striking donor-to-donor variability in cell type proportions compared to the myeloid clusters (Figure S9g-i), consistent with higher inter-individual variability in the adaptive immune compartment. The location of origin, material and protocol explained little variation for any of the cell type proportions, including for B cells.

We also obtained TCR VDJ sequencing data that confirmed MAIT cell type annotation (with preferential use of TRAJ33 and TRAV1-2)⁸⁴ and showed low clonal expansion in naive and Treg populations compared to memory and effector subsets (Figure S9e). Lastly we show that, as expected, there was no clonal sharing between individuals, but expanded clones were found in multiple locations of the lung within one donor (Figure S9f).

Co-localisation of IgA plasma cells with the SMG

B cells included naive and memory B cells, and plasma cells that were further annotated into immunoglobulin (Ig) IgG or IgA secreting plasma cells and plasmablasts (Fig. 4c, Figure S9j, k), and this annotation was supported by VDJ-seq data via Scirpy B cell receptor (BCR) isotype analysis. IgA, which is important for mucosal immunity^85,86, was the most frequent isotype in the airway samples, while only the third most abundant in the parenchyma (Fig. 4d, Figure S9l). Interestingly, we observe that proportions of IgA plasma cells, relative to IgG, are increased in COVID-19 patients versus healthy controls in single cell data from nasal and tracheal brush samples (Fig. 4e)⁸⁷. The distinguishing markers for IgA versus IgG plasma cells included CCR10 and B-cell maturation antigen BCMA (TNFRSF17) (Figure S9 k), which are important for plasma cell localisation and survival, respectively^86,88,89.

Using Visium ST, we observed the localisation of IgA plasma cells, but not B or IgG plasma cells, in the airway SMG in trachea and bronchi sections (Fig. 4f). Annotation of micro-anatomical tissue microenvironments across all Visium sections confirmed co-localisation of IgA plasma cells with duct, serous and mucous SMG cells, whilst IgG mapped to immune infiltrates (Fig. 4f). Enrichment of plasma cells to the SMG is confirmed in the Human Protein Atlas which shows an abundance of plasma cells (MZB1+) in the SMG region of the bronchus and the nasopharynx (Figure S10a, b), along with a study from the 1970s that showed localisation of IgA plasma cells in human airway SMG ⁹⁰. We further distinguished enrichment of IgA plasma cells in the non-mucous areas of the glands by annotating exclusively mucous versus other seromucous glands on four FFPE Visium airway sections (Figure S10c).

To dissect this niche at single cell resolution, which is lacking with current ST technologies, we used multiplex IHC to confirm the specific presence of IgA2 plasma cells in the SMG at single cell resolution, while IgG positive cells were present in the airways mostly outside the SMG (Fig. 4g, Figure S10d), consistent with Visium ST (Fig. 4f). We also detected IgD + naive B cells and CD3 + CD4 + T helper cells in the SMG (Fig. 4g), suggesting that IgA plasma cells are supported by a complement of cell types that can orchestrate B cell maturation for IgA secretion directly into the airway mucous. We hypothesise that together these different cell types constitute an immune niche which we term the gland-associated immune niche (GAIN). Understanding the immunological mechanisms at the SMG can help understand disease, as increased plasma cell numbers have been shown in smokers⁹⁰, patients with cystic fibrosis⁹¹, COPD⁹², Kawasaki disease ⁹³ and, as we show here, in COVID-19 infection (Fig. 4e).

Cell-cell interactions and the SMG immune cell niche

To understand co-localisation of B cells, IgA plasma cells and T cells in the SMG (Fig. 4g, f), we explored the molecular mechanisms underpinning the SMG as a potential immune niche. We report that expression of pIgR, which facilitates transcytosis of polymeric Ig across the surface epithelium, was high across all SMG epithelial cells, as was Mucosal Epithelial Chemokine (MEC)/CCL28, known to recruit IgA plasma cells through CCR10 in other mucosal sites (Fig. 5a-d)^88,94. We confirmed expression of CCL28 in SMG by smFISH and IHC (Fig. 5e, Figure S10f), and observed a gradient of expression along the proximal to distal axis, where CCL28 is highest in SMG-Duct and Serous cells of the trachea (Figure S10f, g). Using cell-cell interaction analysis tool CellChat⁹⁵ on cells from the airways we saw that, in addition to the CCL28-CCR10 axis between SMG cells and B plasma cells (combined IgA, IgG and plasmablasts), SMG-Duct cells were predicted to interact with CCR6 on memory and naive B cells and CD4 T cells (combined CD4 subsets, excluding Tregs) through CCL20^96–98 (Fig. 5b-d).

To maintain B-plasma and B-memory cells, A Proliferation Inducing Ligand (APRIL), a factor important for B cell survival, differentiation and class switching, was expressed by SMG-Duct and SMG-Serous cells and predicted to interact with the receptors TACI and BCMA on B cells (Fig. 5f, S10e). smFISH for APRIL in tissue confirmed expression in glands especially in duct and serous cells, correlating with the scRNA-seq results (Fig. 5g, Figure S10e, h, i). APRIL expression was detected at higher levels in SMG areas that contained B cells and IgA plasma cells (Figure S10i). In the colon, APRIL expression can be induced on intestinal epithelial cells and leads to IgA2 class-switch recombination (CSR) in the local tissue environment⁹⁹. We also found that a portion of B memory cells express activation-induced cytidine deaminase (AICDA), suggesting the possibility of local CSR at the SMG through TACI-APRIL signalling (Figure S10j). SMG-Duct and SMG-Serous cells expressed IL-6 in scRNA-seq, which was predicted to interact with IL-6R/IL-6ST on B plasma and a subset of CD4-T cells, the CD4-naive/CM T cells (Fig. 5f, Figure S10e). Using smFISH we confirmed IL-6 expressing gland cells (Fig. 5g). In combination with APRIL, IL-6 induces and supports long lived plasma cells, and is a potent inducer of IgA secretion in IgA committed plasma cells^100,101, and increases IgA secretion in COPD¹⁰². Furthermore, IL-6 has been shown as a required factor for CD4 T cell memory formation, and for overcoming Treg mediated suppression¹⁰³. Salivary gland epithelial cells have been shown to induce CD4 T cell differentiation into Tfh cells in an IL-6 dependent manner¹⁰⁴, which induced proliferation of B cells. In this study, proliferation of B cells was also induced in co-culture only with salivary gland epithelial cells, suggesting additional T-independent mechanisms. IL-6 is upregulated in serum and bronchoalveolar lavage fluid in asthma and COPD patients, suggesting that the balance of IL-6 signalling in the SMG is important for disease^105–107.

CellChat also predicted interactions between HLA genes expressed by SMG epithelial cells, particularly SMG-Duct and SMG-Serous cells, and CD4-T cells (Fig. 5h, i). The expression of HLA-DRA and HLA-DRB1 in SMG-Duct and SMG-Serous cells was comparable to ciliated cells and, as expected, less than professional APCs such as B cells (Fig. 5i). However, HLA-DR protein expression was high in the glands, but not observed at all in the surface epithelium showing discrepancy between transcript and protein levels (Fig. 5j). We additionally found expression of CD40, a co-stimulatory molecule key for APC-T cell interactions, in SMG epithelial cells (Fig. 5i). Interestingly, we observed that CD4 T cells localised to HLA-DR high regions of glands (Fig. 5k), which are likely SMG-Duct and SMG-Serous cells (Figure S10k, l). CD4 T cells in the glands were mostly CD45RO+ (memory) cells with very few CD45RA+ (naive) cells, and could be seen closely interacting with HLA-DR^highSMG-epithelial cells, suggesting that SMG-epithelial cells are modulating CD4 T cells responses through MHCII (Fig. 5k, Figure S10m,n,o). Expression of HLA-DR and CD40 has previously been observed in airway and nasal epithelial cells, resulting in promotion of T cell proliferation in vitro^108–112. Interestingly, CD40 expression has been observed in duct, but not other cells, of the salivary gland and is upregulated in Sjögren's syndrome, a systemic autoimmune exocrinopathy¹¹³. Antigen presentation by SGP^lowMHCII^high epithelial cells in the parenchyma of mice has been shown to localise and regulate CD4 TRM responses, contributing to immune homeostasis¹¹⁴. HLA-DR/MHCIIhigh SMG epithelial cells may have a similar function in the airways, however the precise nature of antigen presentation by SMG epithelial cells is yet to be determined.

In conclusion, we have identified the localisation of IgA plasma cells, naive/memory B cells and T cells at the SMG along with a large number of molecular signalling pathways involved in B lymphocyte recruitment, antigen presentation and both T cell-dependent and -independent B cell maturation and survival. These pathways are known to be functional in secondary lymphoid organs, including within MALT, and we now show their possible involvement in establishing and maintaining the GAIN (Fig. 6) of the lung.

Integrating single cell and single nuclei RNAseq as well as spatial transcriptomics, we provide detailed annotation of 79 cell types/states in the human lung along the tracheobronchial tree and parenchyma. Our access to deep airway samples and combination of modalities distinguished 11 new cell types/states and enabled us to spatially locate cell types within multiple distinct micro-anatomical tissue environments. Our data has contributed to the large integrated HLCA¹⁴ that provides a means of assessing intra-individual variation and examining the effect of clinical covariates. In addition, our in-depth study provides increased cell type resolution and transcriptomic descriptions of airway tissue niches revealing both new cell types and their interactions, as well as new and unexpected emergent properties of cell signalling relationships. We provide the full transcriptome for the first time of human airway chondrocytes, the nerve bundles including two types of Schwann cells and peri- and endoneurial nerve-associated fibroblasts, submucosal gland (SMG) duct cells, enhanced resolution in fibroblast, macrophage and lymphocyte subsets and distinguish between pulmonary versus systemic vasculature and pericytes. Importantly, we highlight potential disease associations for some of these novel populations, such as PC-Fibro for COPD and the association of peripheral nerve cell types with neuropathies. Finally, we discover a novel submucosal Gland Associated Immune Niche (GAIN) with relevance in inflammatory and infectious diseases. We present this data as a resource to the community as open access downloadable files and through our interactive web portal (lungcellatlas.org).

The GAIN defines a novel immunomodulatory role for SMG epithelial cells, which are central to signalling circuits involved in initiating and maintaining lung mucosal immunity. Specifically, we highlight that IgA plasma cells, B cells and CD4 T cells are attracted to the SMG via chemokines secreted by IR-pericytes and SMG-Duct and SMG-Serous cells. The survival, maturation and, potentially, class switching of B-lineage cells are supported by APRIL and IL-6, expressed by SMG-Duct and -Serous cells, providing T cell independent factors. Additionally, SMG epithelial cells have the potential capacity to induce and/or modulate antigen specific T and B cell responses through expression of MHC-II and CD40. Many of the signalling pathways we described have been observed in other tissues and are particularly well described in other mucosal gland tissues such as salivary glands in the oral cavity, and within secondary lymphoid tissues such as the Peyer’s patches of the gut. No such secondary lymphoid structures have been observed within healthy airways, but we postulate that the GAIN microenvironment may be an important site for local induction of immune responses and homeostasis.

This newly defined IgA immune niche is likely to play an important role in common lung diseases as well as respiratory infectious diseases: IgA plasma cells are increased in the airways of COPD¹⁰² and Cystic Fibrosis⁹¹ patients, and pIgR-mediated IgA transport is dysregulated in asthma¹¹⁵ and pulmonary fibrosis¹¹⁶. In COVID-19 patients, early SARS-CoV-2 neutralisation was more closely correlated with IgA than IgM or IgG²³ and here we report higher proportions of IgA plasma cells in the airways of COVID-19 patients, supporting the importance of the GAIN during infection. Nasal vaccines can induce a strong local sIgA response and prevention of viral shedding¹¹⁷ in the respiratory tract (reviewed in ¹¹⁸), yet all currently licenced COVID-19 vaccines are administered intramuscularly¹¹⁹. A better understanding of the GAIN and the pathways that establish this niche are therefore highly relevant for maintaining lung health and providing immunity to respiratory infections such as COVID-19.

Acknowledgements

We thank Jana Eliasova for the graphical illustrations. Cecilia Dominguez Conde supported with a script for TCR clonotype sharing analysis. Martin Prete’s and the Cellgen IT’s computational support has been central for the analysis. We are grateful to the organ donors, their families and the Cambridge Biorepository for Translational Medicine for the gift of human tissue.

K.B.M. and S.A.T. are supported by Wellcome (WT211276/Z/18/Z and Sanger core grant WT206194). E.M. is supported by ESPOD fellowship of EMBL-EBI and Sanger Institute. A.J.O is supported by RESPIRE4 Marie Sklodowska-Curie fellowship (Grant Agreement 847462). K.T.M. is supported by an award from the Chan Zuckerberg Foundation. The project has received funding from the European Union’s Horizon 2020 research and innovation programme under grant agreement No 874656. M.Z.N. acknowledges funding from a MRC Clinician Scientist Fellowship (MR/W00111X/1) and a MRC Rutherford Fellowship (MR/5005579/1). M.Z.N. and K.B.M. have been funded by the Rosetrees Trust (M944) and from Action Medical Research (GN2911). K.B.W. acknowledges funding from University College London, Birkbeck MRC Doctoral Training Programme. This project has been made possible in part by grant 2019-202654 from the Chan Zuckerberg Foundation. This publication is part of the Human Cell Atlas - www.humancellatlas.org/publications.

Author Contributions

K.B.M., E.M. and A.J.O. conceived and designed the experiments; E.M., A.J.O., K.P., A.R.O., J.P.P., C.X., R.E. and N.H. carried out computational analysis; V.K. ran and optimised Cell2Location analysis.; A.W-C. helped with experimental planning, sample management and spatial gene expression; L.M., L.B., A.K., E.P., A.H., and A.O. carried out tissue dissociation and sc and snRNA-seq experiments; M.D., L.T., S.P. and S.F.V. performed Visium Spatial Transcriptomics and RNAScope analysis, supervised by M.P; S.P. provided histology support. N.R., S.P. and A.J.O. carried out IHC and protein staining; P.H. and R.E. contributed to cell types annotation; K.M., N.G., K.S-P. provided human tissue samples; N.K. carried out statistical analysis; M.Y., K.B.W., R.G.H.L. and M.Z.N. shared unpublished data. O.A.B., M.C., O.S., S.A.T. and K.B.M provided funding, discussion and supervision; and E.M., A.J.O., A.W-C. and K.B.M. wrote the manuscript.

Declaration of Interest

In the past three years, SAT has received remuneration for consulting and Scientific Advisory Board Membership from Genentech, Roche, Biogen, GlaxoSmithKline, Foresite Labs and Qiagen. SAT is a co-founder, board member and holds equity in Transition Bio.

Data availability

The processed scRNA-seq, snRNA-seq and Visium ST data is available for browsing and download via our website www.lungcellatlas.org. The dataset is available on the Human Cell Atlas Data Portal here: https://data.humancellatlas.org/explore/projects/957261f7-2bd6-4358-a6ed-24ee080d5cfc. The Visium data is publicly available on ArrayExpress with the accession number E-MTAB-11640: https://www.ebi.ac.uk/arrayexpress/experiments/E-MTAB-11640/. The raw data and accompanying metadata is in European nucleotide Archive (ENA) with the accession PRJEB52292 and BioStudies accession S-SUBS17.

Aim and study design

We aimed to generate a multi-omics view of different regions of the human lung using single cell and nuclei sequencing (10X Genomics droplet sequencing) and spatial analysis (Visium Spatial Transcriptomics and RNAScope in situ hybridisation).

Experimental Methods

Access to human tissue and ethics oversight

Samples were obtained from deceased transplant organ donors by the Cambridge Biorepository for Translational Medicine (CBTM) with informed consent from the donor families and approval from the NRES Committee of East of England – Cambridge South (15/EE/0152). This consent includes generation of open-access genetic sequencing data and publication in open access journals in line with Wellcome Trust policy. CBTM operates in accordance with UK Human Tissue Authority guidelines.

Tissue dissociation and single cell sequencing

Tissue was collected from twelve donors from five lung locations including trachea, bronchi at the 2nd/3rd generation, bronchi at the 4th generation, upper left lobe parenchyma and lower left lobe parenchyma (Figure 1a and Supplementary Tables 1-3). Following collection at the clinic, samples (ranging 1-4cm³) were immediately placed into cold Hypothermasol FRS as per the protocol outlined by Madissoon et al.¹² and transferred to the processing laboratory. Within 12h of cessation of circulation samples were divided into fresh portions of around 1g for immediate dissociation (7 donors, Supplementary Table 1) and pieces for embedding in OCT / freezing in isopentane at -60 to -70 ^oC for later spatial analysis (6 donors, Supplementary Table 3) and nuclei isolation (7 donors, Supplementary Table 2). The majority of samples (n=5) were digested using liberase and trypsin, and CD45 positive cells loaded on 10X as a separate fraction (protocols.io 39ygr7w), although one donor was digested with collagenase D for comparison (protocols.io 34kgquw). Details on which samples received which treatments can be found in Supplementary Table 1 and 2; all protocols are available on protocols.io and linked in the metadata collated in the Human Cell Atlas Data Coordination Platform. Briefly, for the main tissue dissociation protocol used (for 5 donors), a 1g piece of lung tissue was washed with PBS-/- and minced finely with scalpels, before treatment with 13U/ml liberase TL and 0.1mg/ml DNase I for 30 minutes at 37°C with rocking. Cells were filtered through a 70μm strainer, washed with neutralisation media (RPMI + 20% FBS) and pelleted (sample P1). Tissue remaining in the cell strainer was further digested with 0.25% trypsin-EDTA with DNase I for 30 minutes at 37°C with rocking, filtered and washed with neutralisation media. Meanwhile, sample P1 was treated with red blood cell lysis buffer before being separated into CD45 positive and negative fractions using MACS (Miltenyi, as according to the manufacturer’s protocol). The CD45 negative fraction was pooled with cells from trypsin treatment, resulting in two samples for loading on 10X: CD45 positive cells from liberase TL digestion (to enrich for immune cells) and pooled CD45 negative liberase-treated cells with trypsin-treated cells (non-immune fraction). These two cell fractions were resuspended in 0.04% BSA / PBS, counted using a Nucleocounter (Chemometec) and loaded on the 10X Genomics Chromium instrument for single cell sequencing, aiming to capture 5000 cells. The majority of samples were sequenced using 10X Genomics 5’v1 chemistry (5 donors; 2 were donors run using 3’v2 chemistry; Supplementary Table 1) according to the manufacturer’s protocol.

Single nucleus sequencing

Single nuclei were isolated from frozen tissue using a modified version of the protocol described by Krishnaswami et al., 2016¹²⁰ (samples and the protocol used are given in Supplementary Table 3). Frozen tissue sections (8 50 um thick sections) were homogenised using a glass dounce homogeniser (Sigma) in Nuclei Isolation buffer (NIM - Sucrose 0.25 M, MgCl2 0.005 M, KCl 0.025 M, Tris (buffer pH7.4) 0.01 M, DTT 0.001 M and Triton X-100 0.1%) in the presence of Complete protease inhibitors (Roche) and RNAse inhibitors RNasin (Promega) – 0.4 U/ul and Superasin (Invitrogen) 0.2 U/ul). Tissue was homogenised using ~15 strokes with pestle A (clearance 0.0028-0.0047 in.) then pestle B (clearance 0.0008-0.0022 in.). Isolated nuclei were filtered through a 40 uM filter, collected at 2000 x g followed by resuspension in 0.5 ml of storage buffer (PBS containing 4% BSA and RNasin (Promega) – 0.2 U/ul. Nuclei were incubated with NucBlue® (ThermoFisher), 1 drop was added to 0.5 ml nuclei suspension. Nuclei were purified by FACs sorting, stained with Trypan blue and counted using a disposable haemocytometer. Nuclei were diluted to an appropriate concentration and 5,000 nuclei from 5 different samples were pooled and analysed using the 10X 3’ platform. 25,000 nuclei were and loaded onto the 10X Chromium controller using the 3’ v3.1 kit as per the Chromium Single Cell 3ʹ Reagent Kits v3 User Guide, targeting to recover ~3,000 nuclei per sample. Post GEM-RT cleanup, cDNA amplification and 3’ gene expression library construction were carried out according to the appropriate user guide. The resulting libraries were sequenced on the Novaseq platform.

Visium Spatial Transcriptomics

Samples no larger than 0.5cm² were cut from 5 locations of lung (trachea, bronchi at the 2nd/3rd generation, bronchi at the 4th generation, upper left lobe parenchyma and lower left lobe parenchyma) and additional small airway. The majority of the parenchyma was removed from the airway tissues, which were either embedded in OCT and flash frozen in dry-ice cooled isopentane at -55°C or fixed for 24h in 10% neutral buffered formalin and processed into wax as per standard procedures (FFPE). Haematoxylin and eosin staining was used to determine the morphology of tissue blocks, in order to choose those to proceed to Visium. Ten micron sections (fresh frozen samples) or five micron (FFPE) were then cut from the blocks onto Visium slides (10X Genomics) and processed according to the manufacturer’s protocol. Full list of samples, locations, donors and permeabilisation times are shown in Supplementary Table 3 and Supplementary Figure 3. Haematoxylin and eosin images generated during the Visium protocol were captured at 20x magnification on a Hamamatsu Nanozoomer S60 and exported as tiled tiffs for analysis.

Dual-indexed libraries were prepared as in the 10X Genomics protocol, pooled at 2.25nM and sequenced 4 samples per Illumina Novaseq SP flow cell with read lengths 28bp R1, 10bp i7 index, 10bp i5 index, 90bp R2 for fresh frozen samples or 50bp R2 for FFPE.

RNAScope

Tissue blocks for RNAScope in situ hybridisation were chosen based on haematoxylin and eosin staining results as described under ‘Visium’. Ten micron-thick cryosections cut on to superfrost plus slides were processed using the RNAScope 2.5 LS multiplex fluorescent assay (ACD, Bio-Techne) on the Leica BOND RX system (Leica). Fresh frozen lung sections were fixed for 90 minutes with chilled 4% paraformaldehyde, washed twice with PBS and dehydrated through an ethanol series (50%, 70%, 100%, 100% ethanol) before processing according to the manufacturer’s protocol with protease IV treatment. Samples were first tested with RNAScope positive and negative control probes before proceeding to run probes of interest. Slides were stained for dapi (nuclei) and three (or four where indicated) probes of interest, with fluorophores opal 520, opal 570, opal 650 and atto 425 (4-plex only) at between 1:500 - 1:1000 concentration. These were then imaged on a Perkin Elmer Opera Phenix High Content Screening System with water immersion at 20x Magnification.

Tissue preservation and antibody staining

For IBEX staining, the fresh airway tissue was received in cold Hypothermasol, fixed with 1% PFA solution for 24h at 4˚C and transferred to cold 30% sucrose solution overnight, before freezing in OCT. The fixed tissue was sectioned at 10 µm - 30 µm thickness. Iterative staining of human trachea sections was performed as described by Radtke et al.¹²¹. Sections were permeabilised and blocked in 0.1M TRIS, containing 0.1% Triton (Sigma), 1% normal mouse serum, 1% normal goat serum and 1% BSA (R and D). Samples were stained for 2h at RT (primary antibodies) or 1h at RT (secondary antibodies) in a wet chamber with the appropriate antibodies, washed 3 times in PBS and mounted in Fluoromount-G® (Southern Biotech). Images were acquired using a TCS SP8 (Leica, Milton Keynes, UK) inverted confocal microscope. The coverslip was then removed and slides were washed 3 times in PBS to remove any mounting medium. Bleaching of the fluorochromes was achieved using a 1mg/mL solution of lithium borohydride in water (Acros Organics) for 15 minutes at RT. The slides were then washed 3 times in PBS prior to staining with a different set of antibodies as described above. The process was repeated a total of 5 times. Raw imaging data were processed using Imaris (Bitplane) using Hoechst as fiducial for the alignment of subsequent images. The staining set-up and antibody information is in Supplementary Table 6. For human airway samples, co-stained for CCL28 and IgA2, 10 µm thick OCT embedded fresh frozen sections were fixed with cold acetone for 20 minutes, followed by blocking in the buffer described above for 1h at RT. Primary antibodies CCL28 (Atlas Antibodies HPA077434, used 1:25) and IgA2 (same as in Supplementary Table 6, used 1:600) were incubated in blocking buffer for 1h at RT, washed 3 times in PBS and then incubated with goat anti-rabbit IgG AF647 secondary antibody (Abcam ab150079, used 1:1000)for 1h at RT followed by another 3x PBS washes. Lastly, samples were incubated with 5 µg/ml DAPI (Invitrogen) in blocking buffer for 5 minutes before coverslipping with ProLong Gold Antifade Mountant (LifeTechnologies). Slides were imaged using Hamamatsu S60 slide scanner at 40x.

Computational analysis

Mapping of Gene expression libraries

Single-cell and single nuclei RNA-seq gene expression libraries were mapped with Cell Ranger 3.0.2, and Visium libraries were mapped with Space Ranger 1.1.0 from 10x genomics (https://support.10xgenomics.com). Both types of libraries were mapped to an Ensembl 93 based reference (10x-provided GRCh38 reference, version 3.0.0). For nuclei samples, the reference was altered into a pre-mRNA reference as per 10x instructions. TCR/BCR libraries were mapped with Cell Ranger 4.0.0 to the 10x-provided VDJ reference, version 4.0.0.

Spatial mapping of cell types using Visium Spatial Transcriptomics and cell2location

Visium Spatial Transcriptomics (ST) data was used to establish tissue locations of the cell types presented in the paper. This was achieved by applying the Cell2location method to estimate the abundance of each cell type in the Visium slides by integrating the single cell/single nuclei and spatial transcriptomes²⁵. The first step of cell2location workflow is estimation of reference expression signatures of cell types representing gene expression profiles for each annotated cell type and subpopulation (n=85). This step was performed using Negative Binomial regression (from cell2location package) that allows accounting for technology (sc/sn), batch and donor effects. To spatially map immune infiltrate microenvironment we extended the cell type reference by including immune populations involved in the germinal center reaction. This was done by extending the above mentioned set of reference cell type signatures with the signature of these populations which were derived using published human gut data sets that featured good representation of such tissue infiltrating populations ³⁴. In the second step of cell2location workflow these reference signatures are used to estimate absolute spatial abundance of cell types, integrating and normalising data across 11 Visium sections (5 were excluded based on quality control metrics). 10X Visium data was processed as described above to untransformed and unnormalised mRNA counts, filtered to genes shared with scRNA-seq, which were used as input to cell2location with hyperparameters that were determined using information about the tissue and experiment quality:

Expected cell abundance per location , which were estimated by computing average cell count across 20 representative locations derived by manually counting cells in histology images.
Regularisation of within-experiment variation in RNA detection sensitivity , representing more lenient regularisation to account for large technical within-slide variability in total count observed in this data.

The mode was trained until convergence was achieved (40,000 iterations).

To aid interpretation, all locations in the Visium data were annotated with micro-anatomical tissue environment labels using paired histology images. Each spot on each Visium slide that was aligned with the tissue was annotated as “Tissue” (manual annotation in loupe browser) and was used for analysis. Specific tissue environments were annotated as follows: cartilage, glands, multilayer epithelium, parenchyma, nerve bundle, perichondrium, airway smooth muscle, arterial vessel, venous vessel, undefined morphology, mesothelium, pulmonary vessel, immune infiltrate and small airway (on parenchyma sections only) (Supplementary Table 5). The addition of four Visium FFPE sections with better conservation of morphology allowed to further annotate peribronchium and distinguish exclusively mucous glands from the rest of the seromucous glands. The annotation of mucous only glands was very conservative and exclusive to distinctly separate mucous-only glands, as the transcripts from high counts serous glands regions often contaminated the neighbouring spots. The annotation was used to compute average estimated cell abundance of each cell type across annotated microenvironments, which was normalised per cell type and presented as a dotplot.

Single cell and nuclei RNA-seq analysis

The CellRanger unfiltered matrices were used as an input for the SoupX algorithm¹²² to remove ambient RNA contamination. The single cell RNA-seq libraries were corrected according to the tutorial (https://github.com/constantAmateur/SoupX). For each single nuclei RNA-seq library, CellRanger filtered nuclei were first subjected to a set of QC filters. Pass QC nuclei were processed using standard scanpy pipeline and were clustered at a relatively low resolution to form 5-10 clusters. Nuclei that didn't pass QC were assigned to those clusters by logistic regression. This clustering was then passed to SoupX, which used it to derive a set of cluster-specific genes for automatic estimation of contamination rate, which together with estimated soup composition were used by SoupX to clean the data. Default values were used when calling SoupX's functions, except for "autoEstCont()" where "soupQuantile" was set to 0.8 to improve stability of estimation. The single-cell and nuclei libraries with SoupX correction were analysed using the standard scanpy 1.7.1 workflow¹²³. The cells with >4000 counts in nuclei and 20,000 counts in the cells were removed. In the cells, droplets with >10,000 features were removed. Lower threshold of 1000 features was applied to donor A37 due to ambient RNA contamination that was difficult to remove. The rough level clustering was used to annotate master cell types. Master cell types were further subtracted and re-analysed with scanpy workflow including new highly variable genes detection. Typically, between 1000 and 3000 highly variable genes (HVG-s) were used for calculating 40 principal components that were used for calculating the UMAP coordinates. Data integration via Harmony¹²⁴ or BBKNN¹²⁵ or scVI¹²⁶ was used with either “Material” as input to correct between cells and nuclei, or “Material and Donor” in order to take into account the variability between donors. Clustering at different resolutions was done in order to determine biologically meaningful cell types. Doublets and contaminated cells were detected via overclustering and observing markers from more than one cell type, and in case of nuclei-seq also observing higher counts. Doublet clusters were removed. Iterative clustering approach was used to further derive clusters for less abundant cell types. In addition to known marker genes, new ones were derived using the scanpy’ rank_genes_groups function. Cell types and master clusters were annotated according to known and newly derived markers as in Supplementary File 1. The UMAP of the airway epithelial cell types was achieved by integrating the data with published human airway epithelial cells²¹, including previously un-annotated serous and mucous cells identified from the unprocessed data of their study. Altogether we detected all cell types from at least two different locations and three different donors. The distribution of cell types with respect to Location, Material and Donor variables are shown in Supplementary Table 7.

Post-processing analysis

Gene set enrichment analysis was performed in GSEA online tool (https://www.gsea-msigdb.org/gsea/index.jsp)¹²⁷ for specific gene sets and in gProfileR (https://biit.cs.ut.ee/gprofiler/gost). The analysis of gene expression in GTEx tissues was performed in the GTEx portal (https://gtexportal.org/home/). Cell-cell interaction analysis was performed with CellPhoneDB (https://www.cellphonedb.org/)¹²⁸ and with CellChat (http://www.cellchat.org/)⁹⁵. In order to cope with donor to donor differences, the dataset was downsampled to a set number of cells per every donor-celltype combination.

Pseudotime analysis for selected cell populations was performed with Monocle3¹²⁹, its functionality to infer a pseudotime based on UMAP coordinates. The root was identified as the cell with the highest combined expression of canonical progenitor markers (VCAN for chondrocytes; TGM2, HMCN2 and SULF1 for smooth muscle). Cell trajectory analysis was performed using the scVelo package (v0.2.1)¹³⁰ and specifying stochastic model.

Label transfer was performed via Azimuth tool (https://azimuth.hubmapconsortium.org/) with the lung reference data⁹. CellTypist⁷⁵ was used to train a logistic regression model from our dataset for label transfer using the celltypist.train function (tag-value pairs: use_SGD=True, feature_selection=True). Human protein atlas (https://www.proteinatlas.org/humanproteome/tissue) was used for extracting images of protein stainings with antibodies on human tissues.

BCR and TCR analysis from VDJ-data

VDJ analysis was done with Scirpy 0.6.0 (https://icbi-lab.github.io/scirpy/)¹³¹. For TCR data, clonotypes were defined based on CDR3 nucleotide sequence identity. For BCR data, clonotypes were defined based on the Hamming distance between CDR3 amino acid sequences with a cutoff of 2 and orphan VJ chains removed. In both cases, V gene identity was required and the CDR3 sequence similarity was evaluated across all of a cell pair’s V(D)J chains.

Poisson linear mixed model for cell type composition analysis

Poisson regression with various metadata as covariates was applied to adjust confounding effects on the cell type count data as previously described^132,133. We used Location as biological factor and Protocol, Material (scRNA-seq vs snRNA-seq) and Donor as technical factors in the model as random effects to overcome the collinearity (see Supplementary Notes in Elmentaite et al.¹³³ for more details).

Variance in gene expression

In order to determine the effects of the metadata features on the expression data, a linear mixed model was used⁵⁸. Genes expressed in less than 5% of the samples were filtered out. The count matrix was then normalised and log-transformed. The percentages of variance in gene expression data explained by each metadata feature were obtained through fitting the linear mixed model. The Bayes factor was then computed to determine the gene-specific effects of some metadata features in the expression data, assigning an effect size and a local true sign rate for all genes analysed. Genes presenting a local true sign rate (LTSR) value greater than 0.9 were considered significantly affected by the metadata feature analysed. See Supplementary Notes of ⁵⁸ for more details.

Functional GWAS analysis (fGWAS)

The fGWAS approach to determine disease relevant cell types is described elsewhere³⁴. Summary statistics for the selected GWAS study of COPD¹³⁴ were obtained via Open Targets Genetics (https://genetics.opentargets.io/study/GCST007431; https://www.ebi.ac.uk/gwas/studies/GCST007431).

Cell-cell interaction analysis

Cell-cell interaction from scRNA-seq data was predicted using CellChat⁹⁵. B plasma subsets were combined (B-plasma) and cell types of interest (B-naive, B-memory, SMG-Duct, SMG-Mucous, SMG-Serous, CD4-naive/CM, CD4-EM/Effector and CD4-TRM) were downsampled to 200 cells per cell type per donor. Analyses were performed both with individual CD4-T cell subsets and all CD4 subsets combined. Normalised count matrix along with cell annotation metadata was processed through the standard CellChat pipeline, except that the communication probability was calculated with a truncated mean of 10%.

1. Angelidis, I. et al. An atlas of the aging lung mapped by single cell transcriptomics and deep tissue proteomics. Nat. Commun. 10 , 963 (2019).

2. Kato, A., Hulse, K. E., Tan, B. K. & Schleimer, R. P. B-lymphocyte lineage cells and the respiratory system. J. Allergy Clin. Immunol. 131, 933–57; quiz 958 (2013).

3. Schiller, H. B. et al. The Human Lung Cell Atlas: A High-Resolution Reference Map of the Human Lung in Health and Disease. Am. J. Respir. Cell Mol. Biol. 61 , 31–41 (2019).

4. Ardini-Poleske, M. E. et al. LungMAP: The Molecular Atlas of Lung Development Program. Am. J. Physiol. Lung Cell. Mol. Physiol. 313, L733–L740 (2017).

5. Wilbrey-Clark, A., Roberts, K. & Teichmann, S. A. Cell Atlas technologies and insights into tissue architecture. Biochemical Journal vol. 477 1427–1442 (2020).

6. Han, X. et al. Mapping the Mouse Cell Atlas by Microwell-Seq. Cell 173, 1307 (2018).

7. Tabula Muris Consortium et al. Single-cell transcriptomics of 20 mouse organs creates a Tabula Muris. Nature 562, 367–372 (2018).

8. Montoro, D. T. et al. A revised airway epithelial hierarchy includes CFTR-expressing ionocytes. Nature 560, 319–324 (2018).

9. Travaglini, K. J. et al. A molecular cell atlas of the human lung from single-cell RNA sequencing. Nature 587, 619–625 (2020).

10. Plasschaert, L. W. et al. A single-cell atlas of the airway epithelium reveals the CFTR-rich pulmonary ionocyte. Nature 560, 377–381 (2018).

11. Vieira Braga, F. A. et al. A cellular census of human lungs identifies novel cell states in health and in asthma. Nat. Med. 25 , 1153–1163 (2019).

12. Madissoon, E. et al. scRNA-seq assessment of the human lung, spleen, and esophagus tissue stability after cold preservation. Genome Biol. 21 , 1 (2019).

13. Ruiz García, S. et al. Novel dynamics of human mucociliary differentiation revealed by single-cell RNA sequencing of nasal epithelial cultures. Development 146, (2019).

14. Sikkema, L. et al. An integrated cell atlas of the human lung in health and disease. bioRxiv (2022) doi:10.1101/2022.03.10.483747 .

15. Kadur Lakshminarasimha Murthy, P. et al. Human distal lung maps and lineage hierarchies reveal a bipotent progenitor. Nature 604, 111–119 (2022).

16. Angelidis, I. et al. An atlas of the aging lung mapped by single cell transcriptomics and deep tissue proteomics. Nat. Commun. 10 , 963 (2019).

17. Carraro, G. et al. Transcriptional analysis of cystic fibrosis airways at single-cell resolution reveals altered epithelial cell states and composition. Nat. Med. 27 , (2021).

18. Habermann, A. C. et al. Single-cell RNA sequencing reveals profibrotic roles of distinct epithelial and mesenchymal lineages in pulmonary fibrosis. Sci Adv 6 , eaba1972 (2020).

19. Reyfman, P. A. et al. Single-Cell Transcriptomic Analysis of Human Lung Provides Insights into the Pathobiology of Pulmonary Fibrosis. Am. J. Respir. Crit. Care Med. 199, 1517–1536 (2019).

20. Adams, T. S. et al. Single-cell RNA-seq reveals ectopic and aberrant lung-resident cell populations in idiopathic pulmonary fibrosis. Sci Adv 6 , eaba1983 (2020).

21. Goldfarbmuren, K. C. et al. Dissecting the cellular specificity of smoking effects and reconstructing lineages in the human airway epithelium. Nat. Commun. 11 , 2485 (2020).

22. Ordovas-Montanes, J. et al. Allergic inflammatory memory in human respiratory epithelial progenitor cells. Nature 560, 649–654 (2018).

23. Sterlin, D. et al. IgA dominates the early neutralizing antibody response to SARS-CoV-2. Sci. Transl. Med. 13 , (2021).

24. Sun, X. et al. A census of the lung: CellCards from LungMAP. Dev. Cell 57 , 112–145.e2 (2022).

25. Kleshchevnikov, V. et al. Comprehensive mapping of tissue cell architecture via integrated single cell and spatial transcriptomics. doi:10.1101/2020.11.15.378125 .

26. Xia, H. et al. Identification of a Cell-of-Origin for Fibroblasts Comprising the Fibrotic Reticulum in Idiopathic Pulmonary Fibrosis. Am. J. Pathol. 184, 1369 (2014).

27. Gomez, I. G. et al. TWEAK-Fn14 Signaling Activates Myofibroblasts to Drive Progression of Fibrotic Kidney Disease. J. Am. Soc. Nephrol. 27 , 3639–3652 (2016).

28. Bauer, Y. et al. A novel genomic signature with translational significance for human idiopathic pulmonary fibrosis. Am. J. Respir. Cell Mol. Biol. 52 , 217–231 (2015).

29. Kanamori-Katayama, M. et al. LRRN4 and UPK3B are markers of primary mesothelial cells. PLoS One 6 , (2011).

30. Kapoor, V. N. et al. Gremlin 1 + fibroblastic niche maintains dendritic cell homeostasis in lymphoid tissues. Nat. Immunol. 22 , 571–585 (2021).

31. Wang, X. et al. Follicular dendritic cells help establish follicle identity and promote B cell retention in germinal centers. J. Exp. Med. 208, 2497–2510 (2011).

32. Marshall, A. J. et al. FDC-SP, a novel secreted protein expressed by follicular dendritic cells. J. Immunol. 169, (2002).

33. Elliot, J. G. et al. Aggregations of lymphoid cells in the airways of nonsmokers, smokers, and subjects with asthma. Am. J. Respir. Crit. Care Med. 169, 712–718 (2004).

34. Elmentaite, R. et al. Cells of the human intestinal tract mapped across space and time. bioRxiv 2021.04.07.438755 (2021) doi:10.1101/2021.04.07.438755 .

35. Tsukui, T. et al. Collagen-producing lung cell atlas identifies multiple subsets with distinct localization and relevance to fibrosis. Nat. Commun. 11 , 1–16 (2020).

36. Lee, J.-H. et al. Anatomically and Functionally Distinct Lung Mesenchymal Populations Marked by Lgr5 and Lgr6. Cell 170, 1149–1163.e12 (2017).

37. Baarsma, H. A. et al. Noncanonical WNT-5A signaling impairs endogenous lung repair in COPD. J. Exp. Med. 214, 143–163 (2017).

38. Doherty, L. & Sanjay, A. LGRs in Skeletal Tissues: An Emerging Role for Wnt-Associated Adult Stem Cell Markers in Bone. JBMR Plus 4 , e10380 (2020).

39. Bochukova, E. G. et al. Rare mutations ofFGFR2causing apert syndrome: identification of the first partial gene deletion, and anAluelement insertion from a new subfamily. Human Mutation vol. 30 204–211 (2009).

40. Robertson, S. Disorders. in GeneReviews (eds. Adam, M. P. et al.) (University of Washington, Seattle, 2008).

41. Chen, B., Banton, M. C., Singh, L., Parkinson, D. B. & Dun, X.-P. Single Cell Transcriptome Data Analysis Defines the Heterogeneity of Peripheral Nerve Cells in Homeostasis and Regeneration. Front. Cell. Neurosci. 15 , 624826 (2021).

42. Gerber, D. et al. Transcriptional profiling of mouse peripheral nerves to the single-cell level to build a sciatic nerve ATlas (SNAT). (2021) doi:10.7554/eLife.58591 .

43. Renthal, W. et al. Transcriptional Reprogramming of Distinct Peripheral Sensory Neuron Subtypes after Axonal Injury. Neuron 108, 128–144.e9 (2020).

44. Wolbert, J. et al. Redefining the heterogeneity of peripheral nerve cells in health and autoimmunity. Proc. Natl. Acad. Sci. U. S. A. 117, 9466–9476 (2020).

45. Adameyko, I. & Ernfors, P. Nerves Do It Again: Donation of Mesenchymal Cells for Tissue Regeneration. Cell Stem Cell vol. 24 195–197 (2019).

46. Vila Ellis, L. et al. Epithelial Vegfa Specifies a Distinct Endothelial Population in the Mouse Lung. Dev. Cell 52 , 617–630.e6 (2020).

47. Niethamer, T. K. et al. Defining the role of pulmonary endothelial cell heterogeneity in the response to acute lung injury. Elife 9 , (2020).

48. Gillich, A. et al. Capillary cell-type specialization in the alveolus. Nature 586, 785–789 (2020).

49. Schupp, J. C. et al. Integrated Single-Cell Atlas of Endothelial Cells of the Human Lung. Circulation 144, 286–302 (2021).

50. Murfee, W. L., Skalak, T. C. & Peirce, S. M. Differential arterial/venous expression of NG2 proteoglycan in perivascular cells along microvessels: identifying a venule-specific phenotype. Microcirculation 12 , 151–160 (2005).

51. Proebstl, D. et al. Pericytes support neutrophil subendothelial cell crawling and breaching of venular walls in vivo. J. Exp. Med. 209, 1219–1234 (2012).

52. Nowicki-Osuch, K. et al. Molecular phenotyping reveals the identity of Barrett’s esophagus and its malignant transition. Science vol. 373 760–767 (2021).

53. Widdicombe, J. H. & Wine, J. J. Airway Gland Structure and Function. Physiol. Rev. 95 , 1241–1319 (2015).

54. Meyrick, B., Sturgess, J. M. & Reid, L. A reconstruction of the duct system and secretory tubules of the human bronchial submucosal gland. Thorax 24 , 729–736 (1969).

55. Hegab, A. E. et al. Isolation and in vitro characterization of basal and submucosal gland duct stem/progenitor cells from human proximal airways. Stem Cells Transl. Med. 1 , (2012).

56. Tata, A. et al. Myoepithelial Cells of Submucosal Glands Can Function as Reserve Stem Cells to Regenerate Airways after Injury. Cell Stem Cell 22 , 668–683.e6 (2018).

57. Hegab, A. E. et al. Novel stem/progenitor cell population from murine tracheal submucosal gland ducts with multipotent regenerative potential. Stem Cells 29 , 1283–1293 (2011).

58. Young, A. M. H. et al. A map of transcriptional heterogeneity and regulatory variation in human microglia. Nat. Genet. 53 , 861–868 (2021).

59. Borchers, M. T. et al. The role of T cells in the regulation of acrolein-induced pulmonary inflammation and epithelial-cell pathology. Res. Rep. Health Eff. Inst. 5–29 (2009).

60. Motz, G. T. et al. Chronic cigarette smoke exposure primes NK cell activation in a mouse model of chronic obstructive pulmonary disease. J. Immunol. 184, 4460–4469 (2010).

61. Wortham, B. W., Eppert, B. L., Flury, J. L., Morgado Garcia, S. & Borchers, M. T. TLR and NKG2D signaling pathways mediate CS-induced pulmonary pathologies. PLoS One 8 , e78735 (2013).

62. Deprez, M. et al. A Single-Cell Atlas of the Human Healthy Airways. Am. J. Respir. Crit. Care Med. 202, 1636–1645 (2020).

63. Chakarov, S. et al. Two distinct interstitial macrophage populations coexist across tissues in specific subtissular niches. Science 363, (2019).

64. Evren, E. et al. Distinct developmental pathways from blood monocytes generate human lung macrophage diversity. Immunity 54 , 259–275.e7 (2021).

65. Pirzgalska, R. M. et al. Sympathetic neuron-associated macrophages contribute to obesity by importing and metabolizing norepinephrine. Nat. Med. 23 , 1309–1318 (2017).

66. Wolf, Y. et al. Brown-adipose-tissue macrophages control tissue innervation and homeostatic energy expenditure. Nat. Immunol. 18 , 665–674 (2017).

67. Hulsmans, M. et al. Macrophages Facilitate Electrical Conduction in the Heart. Cell 169, 510–522.e20 (2017).

68. Chang, D., Sharma, L. & Dela Cruz, C. S. Chitotriosidase: a marker and modulator of lung disease. Eur. Respir. Rev. 29 , (2020).

69. Artur Krężel, W. M. The Functions of Metamorphic Metallothioneins in Zinc and Copper Metabolism. Int. J. Mol. Sci. 18 , (2017).

70. Subramanian Vignesh, K. & Deepe, G. S., Jr. Metallothioneins: Emerging Modulators in Immunity and Infection. Int. J. Mol. Sci. 18 , (2017).

71. Takano, H. et al. Protective role of metallothionein in acute lung injury induced by bacterial endotoxin. Thorax 59 , 1057–1062 (2004).

72. Mukaida, N. Pathophysiological roles of interleukin-8/CXCL8 in pulmonary diseases. Am. J. Physiol. Lung Cell. Mol. Physiol. 284, L566–77 (2003).

73. Reynolds, G. et al. Developmental cell programs are co-opted in inflammatory skin disease. Science 371, (2021).

74. Hadley, G. A., Bartlett, S. T., Via, C. S., Rostapshova, E. A. & Moainie, S. The epithelial cell-specific integrin, CD103 (alpha E integrin), defines a novel subset of alloreactive CD8+ CTL. The Journal of Immunology 159, 3748–3756 (1997).

75. Conde, C. D. et al. Cross-tissue immune cell analysis reveals tissue-specific adaptations and clonal architecture in humans. doi:10.1101/2021.04.28.441762 .

76. Piet, B. et al. CD8 T cells with an intraepithelial phenotype upregulate cytotoxic function upon influenza infection in human lung. Journal of Clinical Investigation vol. 121 2254–2263 (2011).

77. Wu, T. et al. Lung-resident memory CD8 T cells (TRM) are indispensable for optimal cross-protection against pulmonary virus infection. J. Leukoc. Biol. 95 , 215–224 (2014).

78. Ghilas, S. et al. NK cells orchestrate splenic cDC1 migration to potentiate antiviral protective CD8 T cell responses. doi:10.1101/2020.04.23.057463 .

79. Böttcher, J. P. et al. NK Cells Stimulate Recruitment of cDC1 into the Tumor Microenvironment Promoting Cancer Immune Control. Cell 172, 1022–1037.e14 (2018).

80. Ma, M. et al. NKG2CNKG2A Natural Killer Cells are Associated with a Lower Viral Set Point and may Predict Disease Progression in Individuals with Primary HIV Infection. Front. Immunol. 8 , 1176 (2017).

81. Fang, M. et al. CD94 is essential for NK cell-mediated resistance to a lethal viral disease. Immunity 34 , 579–589 (2011).

82. Triebel, F. et al. LAG-3, a novel lymphocyte activation gene closely related to CD4. Journal of Experimental Medicine vol. 171 1393–1405 (1990).

83. Siegers, G. M., Barreira, C. R., Postovit, L.-M. & Dekaban, G. A. CD11d β2 integrin expression on human NK, B, and γδ T cells. J. Leukoc. Biol. 101, 1029–1035 (2017).

84. Treiner, E. et al. Selection of evolutionarily conserved mucosal-associated invariant T cells by MR1. Nature 422, 164–169 (2003).

85. Corthésy, B. Multi-faceted functions of secretory IgA at mucosal surfaces. Front. Immunol. 4 , 185 (2013).

86. Kunkel, E. J. & Butcher, E. C. Plasma-cell homing. Nat. Rev. Immunol. 3 , (2003).

87. Yoshida, M. et al. The local and systemic response to SARS-CoV-2 infection in children and adults. medRxiv 2021.03.09.21253012 (2021).

88. Morteau, O. et al. An indispensable role for the chemokine receptor CCR10 in IgA antibody-secreting cell accumulation. J. Immunol. 181, 6309–6315 (2008).

89. O’Connor, B. P. et al. BCMA is essential for the survival of long-lived bone marrow plasma cells. J. Exp. Med. 199, 91–98 (2004).

90. Soutar, C. A. Distribution of plasma cells and other cells containing immunoglobulin in the respiratory tract of normal man and class of immunoglobulin contained therein. Thorax 31 , 158–166 (1976).

91. Collin, A. M. et al. Lung immunoglobulin A immunity dysregulation in cystic fibrosis. EBioMedicine 60 , 102974 (2020).

92. Zhu, J. et al. Plasma cells and IL-4 in chronic bronchitis and chronic obstructive pulmonary disease. Am. J. Respir. Crit. Care Med. 175, 1125–1133 (2007).

93. Rowley, A. H. et al. IgA plasma cell infiltration of proximal respiratory tract, pancreas, kidney, and coronary artery in acute Kawasaki disease. J. Infect. Dis. 182, 1183–1191 (2000).

94. Wilson, E. & Butcher, E. C. CCL28 Controls Immunoglobulin (Ig)A Plasma Cell Accumulation in the Lactating Mammary Gland and IgA Antibody Transfer to the Neonate. J. Exp. Med. 200, 805–809 (2004).

95. Jin, S. et al. Inference and analysis of cell-cell communication using CellChat. Nat. Commun. 12 , 1–20 (2021).

96. Lee, A. Y. S. et al. Expression of Membrane-Bound CC Chemokine Ligand 20 on Follicular T Helper Cells in T–B-Cell Conjugates. Front. Immunol. 0 , (2017).

97. Elgueta, R. et al. CCR6-dependent positioning of memory B cells is essential for their ability to mount a recall response to antigen. J. Immunol. 194, (2015).

98. Bowman, E. P. et al. Developmental switches in chemokine response profiles during B cell differentiation and maturation. J. Exp. Med. 191, (2000).

99. He, B. et al. Intestinal bacteria trigger T cell-independent immunoglobulin A(2) class switching by inducing epithelial-cell secretion of the cytokine APRIL. Immunity 26 , 812–826 (2007).

100. Beagley, K. W. et al. Interleukins and IgA synthesis. Human and murine interleukin 6 induce high rate IgA secretion in IgA-committed B cells. J. Exp. Med. 169, 2133–2148 (1989).

101. Hirano, T. et al. Complementary DNA for a novel human interleukin (BSF-2) that induces B lymphocytes to produce immunoglobulin. Nature 324, 73–76 (1986).

102. Ladjemi, M. Z. et al. Increased IgA production by B-cells in COPD via lung epithelial interleukin-6 and TACI pathways. Eur. Respir. J. 45 , 980–993 (2015).

103. Nish, S. A. et al. T cell-intrinsic role of IL-6 signaling in primary and memory responses. Elife 3 , (2014).

104. Gong, Y.-Z. et al. Differentiation of follicular helper T cells by salivary gland epithelial cells in primary Sjögren’s syndrome. Journal of Autoimmunity vol. 51 57–66 (2014).

105. Mercedes Rincon, C. G. I. Role of IL-6 in Asthma and Other Inflammatory Pulmonary Diseases. Int. J. Biol. Sci. 8 , 1281 (2012).

106. Savelikhina, I., Ostrovskyy, M., Ostrovska, K., Kulynych-Miskiv, M. & Varunkiv, O. Proinflammatory cytokine IL-6 detetion in severe COPD patients: Focus on roflumilast. Eur. Respir. J. 52 , (2018).

107. Tillie-Leblond, I. et al. Balance between proinflammatory cytokines and their inhibitors in bronchial lavage from patients with status asthmaticus. Am. J. Respir. Crit. Care Med. 159, (1999).

108. Rossi, G. A. et al. Human ciliated bronchial epithelial cells: expression of the HLA-DR antigens and of the HLA-DR alpha gene, modulation of the HLA-DR antigens by gamma-interferon and antigen-presenting function in the mixed leukocyte reaction. Am. J. Respir. Cell Mol. Biol. 3 , (1990).

109. Kalb, T. H., Chuang, M. T., Marom, Z. & Mayer, L. Evidence for accessory cell function by class II MHC antigen-expressing airway epithelial cells. Am. J. Respir. Cell Mol. Biol. 4 , (1991).

110. Cagnoni, F. et al. CD40 on Adult Human Airway Epithelial Cells: Expression and Proinflammatory Effects. The Journal of Immunology 172, 3205–3214 (2004).

111. Gormand, F. et al. CD40 expression by human bronchial epithelial cells. Scand. J. Immunol. 49 , (1999).

112. Tanaka, H. et al. CD40 and IFN-gamma dependent T cell activation by human bronchial epithelial cells. J. Med. Invest. 48 , (2001).

113. Dimitriou, I. D., Kapsogeorgou, E. K., Moutsopoulos, H. M. & Manoussakis, M. N. CD40 on salivary gland epithelial cells: high constitutive expression by cultured cells from Sjögren’s syndrome patients indicating their intrinsic activation. Clin. Exp. Immunol. 127, (2002).

114. Shenoy, A. T. et al. Antigen presentation by lung epithelial cells directs CD4+ TRM cell function and regulates barrier immunity. Nat. Commun. 12 , 1–16 (2021).

115. Ladjemi, M. Z. et al. Bronchial Epithelial IgA Secretion Is Impaired in Asthma. Role of IL-4/IL-13. Am. J. Respir. Crit. Care Med. 197, 1396–1409 (2018).

116. Planté-Bordeneuve, T. et al. The pIgR-IgA system as a new player in lung fibrosis. Airway cell biology and immunopathology (2021) doi:10.1183/13993003.congress-2021.pa867 .

117. Bleier, B. S., Ramanathan, M. & Lane, A. P. COVID-19 Vaccines May Not Prevent Nasal SARS-CoV-2 Infection and Asymptomatic Transmission. Otolaryngology–Head and Neck Surgery vol. 164 305–307 (2021).

118. Tiboni, M., Casettari, L. & Illum, L. Nasal vaccination against SARS-CoV-2: Synergistic or alternative to intramuscular vaccines? International Journal of Pharmaceutics vol. 603 120686 (2021).

119. Lund, F. E. & Randall, T. D. Scent of a vaccine. Science vol. 373 397–399 (2021).

120. Krishnaswami, S. R. et al. Using single nuclei for RNA-seq to capture the transcriptome of postmortem neurons. Nat. Protoc. 11 , 499–524 (2016).

121. Radtke, A. J. et al. IBEX: A versatile multiplex optical imaging approach for deep phenotyping and spatial analysis of cells in complex tissues. Proc. Natl. Acad. Sci. U. S. A. 117, 33455–33465 (2020).

122. Young, M. D. & Behjati, S. SoupX removes ambient RNA contamination from droplet-based single-cell RNA sequencing data. Gigascience 9 , (2020).

123. Wolf, F. A., Angerer, P. & Theis, F. J. SCANPY: large-scale single-cell gene expression data analysis. Genome Biol. 19 , 15 (2018).

124. Korsunsky, I. et al. Fast, sensitive and accurate integration of single-cell data with Harmony. Nat. Methods 16 , 1289–1296 (2019).

125. Polański, K. et al. BBKNN: fast batch alignment of single cell transcriptomes. Bioinformatics 36 , 964–965 (2020).

126. Lopez, R., Regier, J., Cole, M. B., Jordan, M. I. & Yosef, N. Deep generative modeling for single-cell transcriptomics. Nat. Methods 15 , 1053–1058 (2018).

127. Subramanian, A. et al. Gene set enrichment analysis: a knowledge-based approach for interpreting genome-wide expression profiles. Proc. Natl. Acad. Sci. U. S. A. 102, 15545–15550 (2005).

128. Efremova, M., Vento-Tormo, M., Teichmann, S. A. & Vento-Tormo, R. CellPhoneDB: inferring cell–cell communication from combined expression of multi-subunit ligand–receptor complexes. Nature Protocols vol. 15 1484–1506 (2020).

129. Cao, J. et al. The single-cell transcriptional landscape of mammalian organogenesis. Nature 566, 496–502 (2019).

130. Bergen, V., Lange, M., Peidli, S., Wolf, F. A. & Theis, F. J. Generalizing RNA velocity to transient cell states through dynamical modeling. Nat. Biotechnol. 38 , 1408–1414 (2020).

131. Sturm, G. et al. Scirpy: a Scanpy extension for analyzing single-cell T-cell receptor-sequencing data. Bioinformatics 36 , 4817–4818 (2020).

132. Stephenson, E. et al. Single-cell multi-omics analysis of the immune response in COVID-19. Nat. Med. 27 , 904–916 (2021).

133. Elmentaite, R. et al. Single-Cell Sequencing of Developing Human Gut Reveals Transcriptional Links to Childhood Crohn’s Disease. Dev. Cell 55 , 771–783.e5 (2020).

134. Shrine, N. et al. Author Correction: New genetic signals for lung function highlight pathways and chronic obstructive pulmonary disease associations across multiple ancestries. Nat. Genet. 51 , 1067 (2019).

Yes there is potential Competing Interest. In the past three years, SAT has received remuneration for consulting and Scientific Advisory Board Membership from Genentech, Roche, Biogen, GlaxoSmithKline, Foresite Labs and Qiagen. SAT is a co-founder, board member and holds equity in Transition Bio.

FiguresSupplementary.pdf
Supplementarylegends.docx
SupplementaryFile1.docx
Supplementary File 1. Cell type marker genes.Marker genes for all the described cell types.
SupplementaryTable1.xlsx
Supplementary Table 1. Sample and donor info for single-cell RNA-seq.Sample ID-s and corresponding location, spatial code, material of cells/nuclei, protocol, enrichment and dissociation notes, Donor ID-s with age, BMI, gender and smoking history if available, 10x version, Gene expression sequencing run ID and corresponding BCR and TCR sample ID-s for single cell RNA-seq samples.
SupplementaryTable2.xlsx
Supplementary Table 2. Sample info for single-nuclei RNA-seq.Sample ID-s and corresponding information about pooling, donors, location, protocol and 10x version for single nuclei RNA-seq samples.
SupplementaryTable3.xlsx
Supplementary Table 3. Sample info for Visium Spatial Transcriptomics.Sample and image ID-s, location and donor information, permeabilisation time and Visium slide ID for Visium ST samples.
SupplementaryTable4.xlsx
Supplementary Table 4. Cell type annotation comparisons.Comparison of the finest level of annotations in the current study with the finest level of annotations from the Integrated Human Lung Cell Atlas href="https://paperpile.com/c/uW1uOL/qfdOz" rel="noopener noreferrer" target="_blank">¹⁴, the previous most recent healthy lung cell atlas href="https://paperpile.com/c/uW1uOL/o7q3" rel="noopener noreferrer" target="_blank">⁹ , the LungMAP integrated atlas href="https://paperpile.com/c/uW1uOL/wMX6C" rel="noopener noreferrer" target="_blank">²⁴ and the corresponding ‘CellCards’ from the LungMAP literature review href="https://paperpile.com/c/uW1uOL/wMX6C" rel="noopener noreferrer" target="_blank">²⁴.
SupplementaryTable5.xlsx
Supplementary Table 5. Manual annotation of micro-anatomical tissue environments on Visium ST. Manual annotation indicates presence (y) of various tissue regions on every Visium ST sample. The annotations on the tissue can be seen as categories on the accompanying loupe files stored in DCP and in the online portal.
SupplementaryTable6.xlsx
Supplementary Table 6. Antibody information for IBEX staining. Staining cycle, antibody protein names, clone, conjugate, Vendor, catalogue number and dilution used.
Supplementarytable7.xlsx
Supplementary Table 7. Proportions and numbers of cells from scRNA-seq and snRNA-seq data by Location, Material and Donor.

Download PDF

Journal Publication

published 21 Dec, 2022

Read the published version in Nature Genetics →

Version 1

posted

You are reading this latest preprint version

A spatial multi-omics atlas of the human lung reveals a novel gland-associated immune niche

Status:

Journal Publication

Version 1

Abstract

Figures

Introduction

Results And Discussion

Summary and Conclusions

Declarations

Acknowledgements

Author Contributions

Declaration of Interest

Data availability

Methods

Aim and study design

Experimental Methods

Access to human tissue and ethics oversight

Tissue dissociation and single cell sequencing

Single nucleus sequencing

Visium Spatial Transcriptomics

RNAScope

Tissue preservation and antibody staining

Computational analysis

Mapping of Gene expression libraries

Spatial mapping of cell types using Visium Spatial Transcriptomics and cell2location

Single cell and nuclei RNA-seq analysis

Post-processing analysis

BCR and TCR analysis from VDJ-data

Poisson linear mixed model for cell type composition analysis

Variance in gene expression

Functional GWAS analysis (fGWAS)

Cell-cell interaction analysis

References

Additional Declarations

Supplementary Files

Status:

Journal Publication

Version 1