1
1
A cellular census of healthy lung and asthmatic airway wall identifies novel cell states
2
in health and disease
3
4
5
Vieira Braga, F.A.
1,11
*, Kar, G.
1,11
*, Berg, M.
2,3,*
, Carpaij, O.A.
3,4,§
, Polanski, K.
1,§
, Simon,
6
L.M.
5
, Brouwer, S.
2,3
Gomes, T.
1
, Hesse, L.
2,3
, Jiang, J.
2,3
, Fasouli, E.S.
1,11
, Efremova, M.
1,
7
Vento-Tormo, R.
1
, Affleck, K.
7
, Palit, S.
5
, Strzelecka, P.
1,13,14
, Firth, H.V.
1
, Mahbubani,
8
K.T.A.
6
, Cvejic, A.
1,13,14
, Meyer K.B.
1
, Saeb-Parsy, K.
6
, Luinge, M.
2,3
, Brandsma, C.-A.
2,3
,
9
Timens, W.
2,3
, Angelidis, I.
9
, Strunz, M.
9
, Koppelman, G.H.
3,10
, van Oosterhout, A.J.
7
,
10
Schiller, H.B.
9
, Theis, F.J.
5,8
, van den Berge, M.
3,4
, Nawijn, M.C.
2,3,#,+
& Teichmann,
11
S.A.
1,11,12,#,+
12
13
1.Wellcome Sanger Institute, Wellcome Genome Campus, Hinxton, Cambridge, CB10 1SA,
14
United Kingdom.
15
2.University of Groningen, University Medical Center Groningen, Department of Pathology
16
& Medical Biology, Groningen, The Netherlands. University of Groningen,
17
3.Groningen Research Institute for Asthma and COPD (GRIAC), University of Groningen,
18
Groningen, The Netherlands.
19
4.University Medical Center Groningen, Department of Pulmonology, Groningen, The
20
Netherlands
21
5.Helmholtz Zentrum München, German Research Center for Environmental Health,
22
Institute of Computational Biology, Neuherberg, Germany.
23
6.Department of Surgery, University of Cambridge, and NIHR Cambridge Biomedical
24
Research Centre, Cambridge, United Kingdom.
25
7.Allergic Inflammation Discovery Performance Unit, Respiratory Therapy Area,
26
GlaxoSmithKline, Stevenage, United Kingdom.
27
8.Department of Mathematics, Technische Universität München, Munich, Germany.
28
9.Helmholtz Zentrum München, German Research Center for Environmental Health,
29
Institute of Lung Biology and Disease, Group Systems Medicine of Chronic Lung Disease,
30
and Translational Lung Research and CPC-M bioArchive, Member of the German Center
31
for Lung Research (DZL), Munich, Germany.
32
10.University of Groningen, University Medical Center Groningen, Department of Pediatric
33
Pulmonology and Pediatric Allergology, Beatrix Children’s Hospital, Groningen, The
34
Netherlands.
35
11.Open Targets, Wellcome Genome Campus, Hinxton, Cambridge CB10 1SA, United
36
Kingdom.
37
12.Theory of Condensed Matter Group, Cavendish Laboratory/Dept Physics, University of
38
Cambridge, JJ Thomson Avenue, Cambridge CB3 0EH, UK
39
13.Department of Haematology, University of Cambridge, Cambridge, CB2 0XY, UK
40
14.Cambridge Stem Cell Institute, Cambridge CB2 1QR, UK
41
42
* These authors contributed equally to this work.
43
§ These authors contributed equally to this work.
44
# These authors share senior authorship.
45
+To whom correspondence should be addressed: m.c.nawijn@umcg.nl, st9@sanger.ac.uk
46
47
48
49
50
51
.CC-BY-NC-ND 4.0 International licenseIt is made available under a
(which was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint. http://dx.doi.org/10.1101/527408doi: bioRxiv preprint first posted online Jan. 23, 2019;
2
Summary
52
Human lungs enable efficient gas exchange, and form an interface with the environment
53
which depends on mucosal immunity for protection against infectious agents. Tightly
54
controlled interactions between structural and immune cells are required to maintain lung
55
homeostasis. Here, we use single cell transcriptomics to chart the cellular landscape of
56
upper and lower airways and lung parenchyma in health. We report location-dependent
57
airway epithelial cell states, and a novel subset of tissue-resident memory T cells. In lower
58
airways of asthma patients, mucous cell hyperplasia is shown to stem from a novel mucous
59
ciliated cell state, as well as goblet cell hyperplasia. We report presence of pathogenic
60
effector Th2 cells in asthma, and find evidence for type-2 cytokines in maintaining the altered
61
epithelial cell states. Unbiased analysis of cell-cell interactions identify a shift from airway
62
structural cell communication in health to a Th2-dominated interactome in asthma.
63
64
65
66
.CC-BY-NC-ND 4.0 International licenseIt is made available under a
(which was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint. http://dx.doi.org/10.1101/527408doi: bioRxiv preprint first posted online Jan. 23, 2019;
3
Introduction
67
The lung plays a critical role in both gas exchange and mucosal immunity, and its anatomy
68
serves these functions through (1) the airways that lead air to the respiratory unit, provide
69
mucociliary clearance, and form a barrier against inhaled particles and pathogens; and (2)
70
the alveoli, distal saccular structures where gas exchange occurs. Acute and chronic
71
disorders of the lung are a major cause of morbidity and mortality worldwide
1
. To better
72
understand pathogenesis of lung disease, it is imperative to characterise the cell types of
73
the lung and understand their interactions in health
2,3
and disease. The recent identification
74
of the ionocyte as a novel airway epithelial cell-type
4,5
underscores our incomplete
75
understanding of the cellular landscape of the lung, which limits our insight into the
76
mechanisms of respiratory disease, and hence our ability to design therapies for most lung
77
disorders.
78
79
We set out to profile lung-resident structural and inflammatory cells and their interactions by
80
analysing healthy human respiratory tissue from four sources: nasal brushes, endobronchial
81
biopsies and brushes from living donors, and tissue samples from lung resections and
82
transplant donor lungs. Our single cell analysis identifies differences in the proportions and
83
transcriptional phenotype of structural and inflammatory cells between upper and lower
84
airways and lung parenchyma. Using an unbiased approach to identify tissue-resident CD4
85
T cells in airway wall, we identify a novel tissue migratory CD4 T cell (TMC) that harbours
86
features of both circulating memory cells and of tissue resident memory cells (TRM) CD4 T
87
cells. We demonstrate that many disease-associated genes have highly cell type-specific
88
expression patterns. This holds true for both rare disease-associated genes, such as CFTR
89
mutated in cystic fibrosis, as well as genes associated with a common disease such as
90
asthma.
91
92
In addition, we evaluate the altered cellular landscape of the airway wall in chronic
93
inflammatory disease using bronchial biopsies from asthma patients. We identify a novel
94
epithelial cell state highly enriched in asthma, the mucous ciliated cell. Mucous ciliated cells
95
represent a transitioning state of ciliated cells with molecular features of mucus production,
96
and contribute to mucous cell hyperplasia in this chronic disease. Other changes associated
97
with asthma include increased numbers of goblet cells, intraepithelial mast cells and
98
pathogenic effector Th2 cells in airway wall tissue. We examine intercellular communications
99
occurring in the healthy and asthmatic airway wall, and reveal a remarkable loss of epithelial
100
communication and a concomitant increase in Th2 cell interactions. The newly identified
101
TMC subset interacts with epithelial cells, fibroblasts and airway smooth muscle cells in
102
asthma. Collectively, these data generate novel insights into epithelial cell changes and
103
altered communication patterns between immune and structural cells of the airways, that
104
underlie asthmatic airway inflammation.
105
A human lung cell census identifies macro-anatomical patterns of epithelial cell
106
states across the human the respiratory tree
107
The cellular landscape along the 23 generations of the airways in human lung is expected
108
to differ both in terms of relative frequencies of cell types and their molecular phenotype
6
.
109
We used 10x Genomics Chromium droplet single-cell RNA sequencing (scRNA-Seq) to
110
profile a total of 36,931 single cells from upper and lower airways, and lung parenchyma
111
(Figure 1A, B). We profiled nasal brushes, and (bronchoscopic) brushes and biopsies from
112
airway wall (third to sixth generation) from healthy volunteers. For parenchyma (small
113
respiratory airways and alveoli), we obtained lung tissue from deceased transplant donors,
114
also analysed on the 10x platform, and from non-tumour resection tissue from lung cancer
115
.CC-BY-NC-ND 4.0 International licenseIt is made available under a
(which was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint. http://dx.doi.org/10.1101/527408doi: bioRxiv preprint first posted online Jan. 23, 2019;
4
patients, analysed on a bespoke droplet microfluidics platform based on the Dropseq
116
protocol
7
.
117
118
Integration of the data from nasal epithelium, airway wall and parenchymal tissue reveals a
119
diversity of epithelial, endothelial, stromal and immune cells, with approximately 21 coarse-
120
grained cell types in total (Figures 1 and 2, Extended Figure 1), that can be explored in user-
121
friendly web portal (www.lungcellatlas.org). Analysis of parenchymal lung tissue from
122
resection material using Dropseq led to the identification of 15 coarse-grained cell
123
populations (epithelial and non-epithelial) (Extended Figure 2). Using MatchSCore
8
to
124
quantify the overlap between cell type marker signatures between the two datasets revealed
125
an extensive degree of overlap in cell type identities (Extended Figure 2). In our analysis
126
below, we first concentrate on epithelial cells (Figure 1), and then focus on the stromal and
127
immune compartments (Figure 2).
128
129
In the epithelial lineage, we identified a total of at least 10 cell types across the upper and
130
lower airways and lung parenchyma (Figure 1C, Extended Data Figure 1). We detected
131
multiple basal, club, ciliated and goblet cell states, as well as type-1 (T1) and type-2 (T2)
132
alveolar cells, and the recently described ionocyte
4,5
(Extended Figure 3). Both goblet and
133
ciliated cells were present in the nasal epithelium (Figure 1D). In the lower airways, we
134
detected basal, club and ciliated cells as well as ionocytes, but only very small numbers of
135
goblet cells. T1 and T2 cells were, as expected, only found in the lung parenchyma (Figure
136
1E).
137
138
We did not identify specific clusters of tuft cells or neuroendocrine (NE) cells. Since cell
139
types represented by a small fraction of the data might be missed by unsupervised
140
clustering, we evaluated the expression of known marker genes for NE cells (CHGA,
141
ASCL1, INSM1, HOXB5) and Tuft cells (DCLK1, ASCL2)
4
. NE marker genes identified a
142
small number of cells, present only in lower airways, displaying a transcriptional profile
143
consistent with that of NE cells (extended Figure 4). Tuft cell marker genes did not identify
144
a unique cell population. Ionocytes were found in lower airways, and at very low frequency
145
in upper airways, but were completely absent from the parenchyma. Comparison of the cell
146
populations identified using the two different bronchoscopic sampling methods (brush
147
versus biopsy) in lower airways showed that basal cells were captured most effectively in
148
biopsies, while apical epithelial cells, such as ciliated and club cells were relatively
149
overrepresented in the bronchial brushings (Figure 1D).
150
151
Our dataset allowed us to identify two discrete cell states in basal, goblet and ciliated
152
epithelial cells. Some of these cell phenotypes were restricted to specific anatomical
153
locations along the respiratory tract. Basal cells were present in both upper and lower
154
airways, although at relatively low frequency in upper airways (Figure 1E). The two basal
155
cell states corresponded to differentiation stages, with the less mature Basal 1 cell state
156
expressing higher levels of TP63 and NPPC in comparison to Basal 2 cells (Figure 1F and
157
extended data 1), which were more abundant in bronchial brushes, suggesting a more apical
158
localization for these more differentiated basal cells (Figure 1D). Goblet 1 and 2 cells were
159
both characterized by high expression of CEACAM5, S100A4, MUC5AC and lack of MUC5B
160
(Figure 1F and Extended Figures 1 and 4). Goblet 1 cells specifically express KRT4 and
161
CD36 (Figure 1G and Extended Figure 4). Genes involved with immune function, such as
162
IDO1, NOS2, IL19, CSF3 (Granulocyte-colony stimulating factor) and CXCL10 are
163
expressed at high levels in Goblet 2 cells (Figure 1G and Extended Figure 4). These
164
molecules enriched in Goblet 2 cells are involved in recruitment of neutrophils, monocytes,
165
dendritic cells and T cells
9
. Both goblet cells states are present in upper airway epithelium,
166
.CC-BY-NC-ND 4.0 International licenseIt is made available under a
(which was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint. http://dx.doi.org/10.1101/527408doi: bioRxiv preprint first posted online Jan. 23, 2019;
5
with Goblet 1 cells being more frequent. In contrast, the Goblet 2 cell state was also present
167
in lower airways, albeit at low abundance (Figure 1E).
168
169
Ciliated cell transcriptional phenotypes are also zonated in terms of their presence across
170
macro-anatomical locations, with a discrete ciliated cell state more abundant in upper
171
airways (Ciliated 2) compared to lower airways and parenchyma. Nasal epithelial Ciliated 2
172
cells express pro-inflammatory genes, such as CCL20 (Extended data 3) and higher levels
173
of metabolic genes (ATP12A and COX7A1) and vesicle transport (AP2B1 and SYT5
10
)
174
compared to the Ciliated 1 cell state. In contrast, the Ciliated 1 cells from lower airways
175
specifically expressed genes involved in cytoprotection (PROS1
11
) and fluid reabsorption
176
(FXYD1
12
) (Figure 1H and Extended Figure 4). Interestingly, comparison of the location-
177
specific differences between ciliated and goblet cells identified a transcriptional signature
178
specific for the upper airways present in both epithelial cell types (Extended Figure 4B).
179
180
Next, we assessed the contribution of specific epithelial cell types to Mendelian disease.
181
Cell-type specific expression patterns of genes associated with Mendelian disorders (based
182
on the Online Mendelian Inheritance in Man, OMIM database) confirm ionocytes as
183
particularly high expressers of the CFTR gene, mutated in cystic fibrosis (Figure 1I). These
184
cells also express SCNN1B, mutations of which can cause bronchiectasis, another feature
185
of cystic fibrosis, suggesting a potential key pathological role for ionocytes in both
186
bronchiectasis and cystic fibrosis. In addition, expression of SERPINA1 (Figure 1I) was
187
found to be enriched in type-2 alveolar epithelial cells, underscoring their role in alpha-1-
188
antitrypsin deficiency
13
.
189
Differential anatomical distribution of the stromal and immune components in the
190
human respiratory tree
191
Next, we analysed the single cell transcriptomes of immune and stromal cells from the upper
192
airways, lower airways and the lung parenchyma (Figure 2A). We identified immune clusters
193
of myeloid (macrophages, neutrophils, dendritic cells (DCs) and mast cells) and lymphoid
194
cells (T and NK cells, B cells; Figure 2B, and Extended Figure 5). Immune and stromal cell
195
numbers and composition varied greatly across different anatomical regions (Figure 2A and
196
2C). Nasal brushes contained only a small number of immune cells, with the large majority
197
being dendritic cells. In the lower airways, the fraction of inflammatory cells was significantly
198
larger and relatively enriched for macrophages (Figure 2C and Extended Figure 5), which
199
was directly confirmed by cell composition comparison of upper versus lower airway brushes
200
obtained from the same donor (Extended Figure 5E).
201
202
Macrophages show large donor variation in their phenotype (Extended figure 5), but they all
203
share high expression of MARCO, CCL18 and genes involved in apolipoprotein metabolism
204
(APOC1 and APOE) (Figure 2E). Lung neutrophils express high levels of the granulocyte
205
markers S100A8, S100A12
14
and LILRA5, a receptor poorly characterised in the lungs, that
206
has been shown to have a proinflammatory function in synovial fluid macrophages
15
(Figure
207
2E). DCs were mostly myeloid, with high expression of CD1E, CD1C, CLEC10A (Figure 2E)
208
and of FCER1A (IgE receptor) and CCL17, molecules known to play a key role in
209
inflammatory conditions such as asthma
16
.
210
211
In the droplet RNAseq data sets, we could not distinguish CD4+ and CD8+ T cells and NK
212
cells from each other (Figure 2B). The B cells in our dataset were mostly plasma cells,
213
expressing high levels of JCHAIN (Joining Chain of Multimeric IgA And IgM). IgM+ (IGHM)
214
cells were enriched in the airway lumen and in the lung parenchyma, while IgG3+ (IGHG3)
215
were enriched in airway biopsy samples and were virtually absent from the airway lumen.
216
.CC-BY-NC-ND 4.0 International licenseIt is made available under a
(which was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint. http://dx.doi.org/10.1101/527408doi: bioRxiv preprint first posted online Jan. 23, 2019;