Articles
https://doi.org/10.1038/s41587-021-01037-9
1
Molecular Epidemiology Unit, Berlin Institute of Health at the Charité - Universitätsmedizin Berlin, Berlin, Germany. 2
Department of Pediatric
Respiratory Medicine, Immunology and Critical Care Medicine, Charité - Universitätsmedizin Berlin, corporate member of Freie Universität Berlin and
Humboldt-Universität zu, Berlin, Germany. 3
Center for Digital Health, Berlin Institute of Health at the Charité - Universitätsmedizin Berlin, Berlin, Germany.
4
Research group “Dynamics of Early Viral Infection and the Innate Antiviral Response”, division F170, German Cancer Research Center (DKFZ), Heidelberg,
Germany. 5
Max Planck Institute for Molecular Genetics, Berlin, Germany. 6
Institute of Biochemistry, Charité - Universitätsmedizin Berlin, corporate
member of Freie Universität Berlin and Humboldt-Universität zu Berlin, Berlin, Germany. 7
Institute of Medical Immunology, Charité - Universitätsmedizin
Berlin, corporate member of Freie Universität Berlin, Humboldt-Universität zu Berlin, Berlin, Germany. 8
Department of Infectious Diseases and Respiratory
Medicine, Charité - Universitätsmedizin Berlin, corporate member of Freie Universität Berlin, Humboldt-Universität zu Berlin, and Berlin Institute of Health
(BIH), Berlin, Germany. 9
German Center for Lung Research (DZL), associated partner, Berlin, Germany. 10
Institute of Virology, Charité - Universitätsmedizin
Berlin, corporate member of Freie Universität Berlin, Humboldt-Universität zu Berlin, and Berlin Institute of Health (BIH), Berlin, Germany. 11
German Centre
for Infection Research (DZIF), Associated Partner Charité-Universitätsmedizin Berlin, Berlin, Germany. 12
Department of Anesthesiology and Intensive Care,
University Hospital Leipzig, Leipzig, Germany. 13
Charité - Universitätsmedizin Berlin, corporate member of Freie Universität Berlin, Humboldt-Universität zu
Berlin, Berlin, Germany. 14
These authors contributed equally: J. Loske, J. Röhmel, S. Lukassen. 15
These authors jointly supervised this work: M. Binder,
S. Trump, R. Eils, M. A. Mall, I. Lehmann. ✉e-mail: roland.eils@charite.de
I
t has repeatedly been reported that younger individuals have a
substantially lower risk for developing coronavirus disease 2019
(COVID-19), despite a similar risk of infection, as reflected in
dramatically increased mortality with increasing age1–3
. These
observations suggest that children may have a higher capability of
controlling SARS-CoV-2 infection. It has been shown that an early
cell-intrinsic innate immune response, mediated by pattern recognition
receptors (PRRs) and the type I and III interferon (IFN) system,
is crucial for the successful control of SARS-CoV-2 infection4
.
In line with these observations, recent studies compared adults and
children with severe COVID-19 or those presenting to an emergency
department and described an impaired IFN response in
pediatric COVID-19 (refs. 5,6
). However, the molecular mechanisms
protecting against COVID-19 in younger age groups, particularly in
those with no or only mild/moderate symptoms, remain unknown.
To understand the higher capacity of children for controlling
SARS-CoV-2 infection at an early stage, we systematically
characterized the transcriptional landscape of upper airways, an
airway region with high susceptibility for SARS-CoV-2 infec-
tion7
, in SARS-CoV-2-negative and SARS-CoV-2-positive children
and adults.
Results
Different cellular composition in the upper airways of children
and adults. We included study participants of three different
COVID-19 cohorts: the RECAST study focusing on COVID-19
in children and their families, the Pa-COVID-19 study and the
SC2 study8,9
, including SARS-CoV-2-negative and SARS-CoV-
2-positive children (n = 42) and adults (n = 44). The derived
dataset comprised 268,745 cells in total (Fig.  1a). Samples from
the upper airways (nose) were collected from individuals aged 4
weeks to 77 years with a positive SARS-CoV-2 PCR result along
with age-matched SARS-CoV-2-negative controls (Supplementary
Tables  1 and 2). Focusing on early infection, only mild/moderate
COVID-19 cases were considered for this study (Fig. 1a). On
the basis of the single-cell RNA sequencing (scRNA-seq) data, we
Pre-activated antiviral innate immunity in
the upper airways controls early SARS-CoV-2
infection in children
J. Loske   1,14
, J. Röhmel2,14
, S. Lukassen   3,14
, S. Stricker2
, V. G. Magalhães4
, J. Liebig   3
, R. L. Chua   3
,
L. Thürmann   1
, M. Messingschlager   1
, A. Seegebarth1
, B. Timmermann5
, S. Klages5
, M. Ralser6
,
B. Sawitzki   7
, L. E. Sander   8,9
, V. M. Corman   10,11
, C. Conrad3
, S. Laudi   12
, M. Binder   4,15
,
S. Trump   1,15
, R. Eils   3,9,13,15 ✉, M. A. Mall   2,9,15
and I. Lehmann   1,9,13,15
Children have reduced severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection rates and a substantially lower
risk for developing severe coronavirus disease 2019 compared with adults. However, the molecular mechanisms underlying
protection in younger age groups remain unknown. Here we characterize the single-cell transcriptional landscape in the upper
airways of SARS-CoV-2-negative (n = 18) and age-matched SARS-CoV-2-positive (n = 24) children and corresponding samples
from adults (n = 44), covering an age range of 4 weeks to 77 years. Children displayed higher basal expression of relevant pattern
recognition receptors such as MDA5 (IFIH1) and RIG-I (DDX58) in upper airway epithelial cells, macrophages and dendritic
cells, resulting in stronger innate antiviral responses upon SARS-CoV-2 infection than in adults. We further detected distinct
immune cell subpopulations including KLRC1 (NKG2A)+
cytotoxic T cells and a CD8+
T cell population with a memory phenotype
occurring predominantly in children. Our study provides evidence that the airway immune cells of children are primed for virus
sensing, resulting in a stronger early innate antiviral response to SARS-CoV-2 infection than in adults.
Nature Biotechnology | www.nature.com/naturebiotechnology
Articles NatureBiotechnology
Neutrophil
nrMa
moMa
rMa
mDC
CD11c_mDC
pDC
p_T
DNT
B cell
Plasma
cell
IL-17A_CD8
IL-17A_CD4CTL2 CTL1
NK
CD8_Tm
ag_T Treg
p_NKT
Squamous
Basal
Epi_low RNA
Ciliated
Cil-acute-phaseCil-diff
FOXN4
Ionocyte
MC/basophil
Goblet
Club
Low RNA
scRNA-seq
Adults
n = 18
Children
SARS-CoV-2
positive
SARS-CoV-2
negative
SARS-CoV-2
positive
SARS-CoV-2
negative
UMAP 1
UMAP2
ed
c
Cellpopulation(%)
Age (years)
Negative
Positive
SARS-CoV-2
Children versus adults
SARS-CoV-2negative
Mean difference (%)Cell type
Neutrophil
CD8_Tm
CTL2
Basal
Goblet
Ciliated
1.12***
2.94***
21.85***
–6.48***
–18.20***
4.95
Basal
P < 0.0001
Goblet
P = 0.0246
CD8_Tm
P = 0.0017
60 800 20 40
60
40
20
0
40 80600 20 40 80600 20
P = 0.0005
Neutrophil
P = 0.0008
CTL2
P < 0.0001
Ciliated
Children Adults
b
a
n = 23
n = 21n = 24
0
10
20
30
0
10
20
30
0
10
20
30
40
50
8
6
4
2
0
9
6
3
0
40 80600 20
40 80600 20
40 80600 20
Fig. 1 | Age-dependent changes in cell composition of the upper airways. a, Nasal samples of children (n = 42) and adults (n = 44) were collected from
individuals negative and positive (asymptomatic/mild/moderate COVID-19) for SARS-CoV-2 and subjected to scRNA-seq. b, Uniform manifold approximation
and projection (UMAP) showing all identified individual immune and epithelial cell types and states. MC/basophil, mast cells or basophils; moMa,
monocyte-derived macrophages; nrMa, non-resident macrophages; rMa, resident macrophages; mDC, myeloid dendritic cells; CD11c_mDC, CD11c-expressing
mDC; pDCs, plasmacytoid dendritic cells; DNT, double-negative T cells; NK, natural killer cells; NKT, natural killer T cells; p_NKT, proliferating NKT cells; ag_T,
aging T cells; CD8_Tm, memory CD8+
T cells; CTL1 and CTL2, type 1 and 2 cytotoxic T cells; IL-17A_CD8, IL-17A-expressing CD8 T cells; IL-17a_CD4, IL-17A
expressing CD4+
T cells; p_T, proliferating T cells; Cil-diff, differentiating ciliated cells; Epi_low RNA, epithelial cells with low RNA; FOXN4, FOXN4+
cells.
c, Scaled UMAPs displaying 45,000 cells per group reveal pronounced differences in the nasal cell composition of SARS-CoV-2-negative and -positive children
and adults. d, Table showing all cell types/states significantly different between children and adults in non-infected individuals. Given are mean differences in
percent; positive values indicate a higher number of cells of the respective cell population in children compared to adults. Comparisons were performed by
Kruskal–Wallis test followed by Dunn’s two-sided post hoc comparison with Benjamini–Hochberg correction (***P < 0.001; Neutrophil, 5.54 × 10−7
; CD8_Tm,
1.97 × 10−4
; CTL2, 1.83 × 10−5
; Basal, 6.96 × 10−5
; Ciliated, 9.18 × 10−7
; also see Extended Data Fig. 2a). e, Scatter plots representing changes of specific immune
and epithelial cells over time. Depicted are the percentages of the respective cell type/state with respect to all cells for immune cells (left) or epithelial cells
(right) of each individual. Individuals negative for SARS-CoV-2 are indicated in gray, individuals positive for SARS-CoV-2 in red. Lines represent curve fitting
results by local polynomial regression (LOESS). The P values are from linear regression analysis (Benjamini–Hochberg adjusted two-tailed).
Nature Biotechnology | www.nature.com/naturebiotechnology
ArticlesNatureBiotechnology
identified 33 different cell types or states in the upper respiratory
tract of these individuals, including 21 immune and 12 epithelial
cell subtypes (Fig. 1b and Extended Data Fig. 1a,b). We observed
striking differences between the pediatric and adult study participants
regarding the composition of the immune cell and epithelial
cell compartment in the nasal mucosa. While immune cells were
rarely detected in nasal samples from healthy adults, samples from
SARS-CoV-2-negative children contained high amounts of almost
each immune cell subset with an overall dominance of neutrophils
(Fig. 1b,c and Extended Data Fig. 2a). In adults, SARS-CoV-2 infection
was associated with immune cell influx, while the proportion
of immune and epithelial cells remained nearly stable in children
(Fig. 1c and Extended Data Fig. 2a). Upon infection, children’s neutrophils
showed an activated phenotype that was more pronounced
than in infected adults, characterized by the enhanced expression
of, for example, CCL3 and CXCR1/2 (Supplementary Table 4).
Interestingly, many of the epithelial cell populations showed a
clear age dependency with, for example, goblet cells decreasing and
ciliated cells increasing with age (Fig. 1d,e). A recent complementary
study analyzed the cell composition of the nasal mucosa in
healthy and SARS-CoV-2-infected children based on bulk RNA-seq
and cell deconvolution methods. The authors were unable to identify
children-specific goblet cells, but rather described that samples
from healthy children were dominated by a ciliated cell signature,
highlighting the limitations of bulk RNA approaches10
.
SARS-CoV-2 entry receptors are not differentially expressed
in children compared to adults. The expression level of ACE2,
encoding the SARS-CoV-2 entry receptor, and TMPRSS2, FURIN,
CTSB, CTSL and CTSV, encoding entry-associated proteases, was
similar between children and adults and not upregulated by mild/
moderate COVID-19 compared to the uninfected status (Extended
Data Fig. 2b). Hence, these viral entry factors cannot explain the
differences in SARS-CoV-2 pathophysiology between children
and adults.
Children show enhanced viral sensing in airway epithelial cells.
SARS-CoV-2 is a positive-strand RNA virus with a very high rate of
replication11,12
. Hence, the control of SARS-CoV-2 infection requires
an optimal early and coordinated innate antiviral immunity. This
response is activated by various PRRs. Recently, mounting evidence
has been generated in support of MDA5 (IFIH1) as the major PRR
for SARS-CoV-2 in epithelial cells with RIG-I (DDX58) possibly
playing an additional, but minor, role13,14
(own unpublished data).
An important enhancer of viral RNA sensing by MDA5 is LGP2
(DHX58)15
. Importantly, PRRs, in particular MDA5 and LGP2, are
only weakly expressed in many epithelial cell types but are profoundly
upregulated by positive feedback regulation upon viral
infection of the cell or by paracrine exposure to type I or III IFN.
The dynamics of this feedback regulation are crucial for the successful
control of an infecting virus (Fig. 2a). The importance of the
PRR/IFN axis for the successful resolution of SARS-CoV-2 infection
was recently demonstrated by clinical studies finding a strong association
between inborn errors at various loci of the PRR/IFN system
with an increased risk of severe COVID-19 (ref. 16
). Similarly, and
affecting even a much broader fraction of patients, autoantibodies
directed against type I IFNs have been shown to occur at a remarkably
high frequency in patients with severe COVID-19 (ref. 17
).
Notably, we found a significantly higher basal expression level of
the genes coding for RIG-I, MDA5 and LGP2 in epithelial cells in
the upper respiratory tract of healthy children as compared to adults
(Fig. 2b). This result suggests an increased ability of the respiratory
mucosa of children to respond to viral infections, which is further
supported by the highly increased amounts of innate immune cells
in their upper airways (Fig. 1 and Extended Data Fig. 2a). In epithelial
cells of SARS-CoV-2-positive children and adults, we observed
a high expression level of those genes (Fig. 2c) in particular at the
onset of COVID-19 symptoms that tended to decline until day 4
(day 0–4, referred to as early phase) and remained at lower levels in
the later disease phase (days 5 to 12 post symptom onset, referred
to as late phase). It can be assumed that higher basal expression of
these PRRs would permit immediate sensing of SARS-CoV-2 by
MDA5/LGP2 in infected epithelial cells (Fig. 2a). Strikingly, children’s
airway epithelial cells displayed increased expression of these
PRR genes compared to the expression level of these genes in epithelial
cells in SARS-CoV-2-positive adults (Fig. 2b), in particular in
the early disease phase after symptom onset. From day 5 onwards,
virus sensing is largely comparable between children and adults
(Fig. 2b, lower plot).
Following virus sensing, signaling through IRF3/NFκB leads
to the expression of primary antiviral effectors, as well as antiviral
cytokines such as IFNβ and IFNλ (Fig. 2a). IFNs act on epithelial
cells in an autocrine and paracrine manner, further increasing
MDA5/LGP2 responsiveness in the tissue and inducing a broad
range of IFN-stimulated genes (ISGs). While we were not able
to detect the expression of type I and type III IFNs themselves,
ISGs showed an impressive activation pattern in epithelial cells of
SARS-CoV-2-positive children, including many genes previously
shown to exhibit strong antiviral activity against SARS-CoV-2, such
as LY6E (ref. 18
), IFITM2 and BST2 (ref. 19
). In all epithelial cells, and
in particular in ciliated cells, the magnitude of ISG expression considerably
surpassed that of infected adults in both the early and late
infection phases (Fig. 2d and Extended Data Fig. 3) with a generally
decreasing trend in the late phase (Extended Data Fig. 3).
To demonstrate a direct association between MDA5 expression
levels and activation of ISGs upon SARS-CoV-2 infection,
we established an in vitro model using the human lung epithelial
cell line A549, which exhibits very low basal expression levels of
MDA5 similar to expression levels found in nasal epithelial cells
of healthy adults. As expected, based on inefficient MDA5 sensing
in concert with rapid replication and expression of virus-encoded
antagonists20
, only minute amounts of IFNB1 and ISG transcripts
were induced upon SARS-CoV-2 infection in these cells. However,
in cells with moderately increased basal expression of MDA5 by
lentiviral transduction, permitting efficient virus sensing before the
expression of antagonists, we observed a significant induction of the
expression of IFNB1 and key ISGs including MX1, BST2 (tetherin),
RSAD2 (viperin) and IFIT1 (Fig. 2e). These findings corroborate
the central role of MDA5 expression before infection for sensing
SARS-CoV-2 and inducing a swift and robust ISG response.
Enhanced innate immune cell activation in children. Studying
immune–epithelial cell interactions revealed a stronger immune–
epithelial cell cross talk in children versus adults particularly before
infection. Among immune cells, non-resident macrophages (nrMa)
and monocyte-derived macrophages (moMa) as well as CD11c+
dendritic cells (CD11c_mDC) were most interactive (Fig.  3a).
These immune cell subsets showed a higher activation status in
children as demonstrated by an increased expression level of several
cytokine- and chemokine-coding genes such as IL1B, IL8, TNF,
CCL3 and CCL4 (Fig. 3b). We furthermore observed an enhanced
expression level of IFIH1 in moMa, nrMa and CD11c_mDC in children
infected with SARS-CoV-2 and a significant increase of TLR2
in moMa and nrMa in the early phase of infection, suggesting that
these cells might play an additional role in virus sensing and IFN
production. This is further underlined by the fact that moMa, nrMa
and CD11c_mDC of SARS-CoV-2-negative children expressed
IFIH1 and TLR2 at higher levels than those of adults.
Distinct immune cell subpopulations in children. Apart from
the upregulated cell-intrinsic antiviral capacity of airway epithelial
cells, macrophages and dendritic cells, we found specific
Nature Biotechnology | www.nature.com/naturebiotechnology
Articles NatureBiotechnology
b
d
c
a
SARS-CoV-2positiveSARS-CoV-2
negative
DDX58
IFIH1
DHX58
TLR3
IRF3
DDX58
IFIH1
DHX58
TLR3
IRF3
C
lubG
oblet
Sec_diff
C
iliated
C
il_acute_phaseC
il_diff
Squam
ous
FO
XN
4
IonocyteBasalC
lubG
oblet
Sec_diff
C
iliated
C
il_acute_phaseC
il_diff
Squam
ousFO
XN
4
IonocyteBasal
DDX58
IFIH1
DHX58
TLR3
IRF3
EarlyphaseLatephase
AdultsChildren
−1.0
−0.5
0
0.5
1.0
Scaled
exp.
Epithelial cells
AdultsChildren
Neg Pos Neg Pos
e
mRNAexpression
(relativetouninfected)
MDA5low
MDA5high
SARS-CoV-2 infected
24 h
Mock SARS-CoV-2 Mock SARS-CoV-2
A549ACE2
empty vector (MDA5low
) A549ACE2
MDA5 (MDA5high
)
IFNB1 MX1 BST2 RSAD2 IFIT1
0
5
10
15
20 *** ***** *
N
o
sym
ptom
s
Days post-onset of symptoms
IFIH1expression
Days post-onset of symptoms
DDX58expression
−2
−1
0
1
2
Ave. exp.
Pct. exp.
25
50
75
100
C9orf91
IFIT5
LY6E
RSAD2
IFIT3
IFITM2
BST2
MX1
IFIT1
ISG20
IFITM3
MSR1
ZBP1
4
3
2
1
0
4
3
2
1
0
0 1 2 3 4 5 6 7 8 9 10 11 12
N
o
sym
ptom
s
0 1 2 3 4 5 6 7 8 9 10 11 12
TRAF3
TRIM25
IRF3P
STAT1
STAT2
IRF9
P
P
RIPLET
MAVS
IRF7P
IRF3NFκB PP
IFNβ
IFNβ
ISRE
STAT1
STAT2
IRF9
P
P
ISGs
RIG-I
STAT
JAK
LGP2
MDA5
?
TBK1 IKKε
IFNλ
TLR3
TLR7/8
TRIF
MyD88
TRAF6
?
?
IFNλIFNαSARS-CoV-2
Respiratory epithelial cell
Monocytes, DCs
RSAD
BST2
MX1
NEMO
IKKα IKKβ
NFκBP
IFIT1
Fig. 2 | Enhanced viral sensing in children’s epithelial cells. a, Schematic representation of epithelial response to SARS-CoV-2 infection including genes
involved in virus sensing and subsequent IFN response; created with BioRender.com. b, Dot plots depicting the expression of virus-sensing genes in
epithelial cells of children and adults. Each significant increase comparing SARS-CoV-2-negative children (n = 18) with SARS-CoV-2-negative adults
(n = 23), and SARS-CoV-2-positive children during the early (days post-onset of symptoms (dps) ≤ 4, n = 11) or late infection phase (dps 5–12, n = 11) with
SARS-CoV-2-positive adults amid the early (n = 13) or late (n = 8) infection phase, respectively, is marked by a red circle (Benjamini–Hochberg-adjusted
two-tailed, Wilcoxon P < 0.05). Ave. exp., average gene expression; Pct. exp., percentage of cells expressing the gene. c, Violin plots showing expression
of prototypical virus-sensing genes in epithelial cells of SARS-CoV-2-positive patients (n = 45) in relation to days after first symptoms. d, Heat map
showing scaled expression of 171 representative ISGs in all epithelial cells during the early SARS-CoV-2 infection phase (dps ≤ 4). Only selected genes are
annotated; for a completely annotated heat map, see Extended Data Fig. 3. e, Differential expression of selected ISGs in an in vitro model comparing the
response to SARS-CoV-2 infection in MDA5 (encoded by IFIH1)-overexpressing A549ACE2
cells and empty vector controls. n = 3 biologically independent
replicates; error bars show mean ± s.e.m.; *P < 0.05, ***P<0.001 (IFNB1, 5.47 × 10−4
; MX1, 2.63 × 10−2
; BST2, 8.74 × 10−4
; RSAD2, 4.24 × 10−2
; IFIT1,
1.30 × 10−2
) from one-tailed Student t-test.
Nature Biotechnology | www.nature.com/naturebiotechnology
ArticlesNatureBiotechnology
p_T
DNT
IL-17A_CD8
IL-17A_CD4CTL2
CTL1
NK
CD8_Tm
ag_T
Treg
p_NKT
Positive adults
Positive children
Negative children
Negative adults
SARS-CoV-2
−2
−1
0
1
2
Ave. exp.
Pct. exp.
25
50
75
100
a b
UMAP 1
UMAP2
c
d
SARS-CoV-2positiveSARS-CoV-2negative
AdultsChildren
SARS-CoV-2
positive
SARS-CoV-2
negative
AdultsChildren
−3 3 −30
−3
−5
3 60
0
0 3
−5
0
5
−5
0
5
SARS-CoV-2 positiveSARS-CoV-2
negative
AdultsChildrenAdultsChildren AdultsChildren
Late phaseEarly phase
cd11c_m
DC
m
oM
a
nrM
a
cd11c_m
DC
m
oM
a
nrM
a
cd11c_m
DC
m
oM
a
nrM
a
cd11c_m
DC
m
oM
a
nrM
a
cd11c_m
DC
m
oM
a
nrM
a
cd11c_m
DC
m
oM
a
nrM
a
CCL2
CCL3
CCL4
CCL7
CCL8
CXCL9
CXCL10
IL6
IL8
IL18
TNF
IL1B
IFNAR1
IFNAR2
IFNGR1
IFNGR2
TLR1
TLR2
TLR3
TLR7
MYD88
STAT1
IRF7
DHX58
IFIH1
DDX58
0
2
4
6 *** ***
Late phaseEarly phase
CCL5expression
Negative
0
2
4
6
IFNGexpression
*** *** ***
latephaseearlyphase
SARS-CoV-2positive
Interaction
count
0
50
100
150
200
250
Neutrophil
p_T
Epi_lowRNA
mDC
rMa
lowRNA
Plasma_cell
Cil_diff
DNT
Treg
Treg
Treg
IL-17A_CD8
IL-17A_CD4
ag_T
B_cell
MC/basophil
CD8_Tm
CTL2
pDC
CTL1
NK
p_NKT
FOXN4
Squamous
Ciliated
Ionocyte
moMa
CD11c_mDC
nrMa
Cil_acute_phase
Club
Goblet
Basal
Sec_diff
Neutrophil
p_T
Epi_lowRNA
mDC
rMa
lowRNA
Plasma_cell
Cil_diff
DNT
IL-17A_CD8
IL-17A_CD4
ag_T
B_cell
MC/basophil
CD8_Tm
CTL2
pDC
CTL1
NK
p_NKT
FOXN4
Squamous
Ciliated
Ionocyte
moMa
CD11c_mDC
nrMa
Cil_acute_phase
Club
Goblet
Basal
Sec_diff
Neutrophil
p_T
Epi_lowRNA
mDC
rMa
lowRNA
Plasma_cell
Cil_diff
DNT
IL-17A_CD8
IL-17A_CD4
ag_T
B_cell
MC/basophil
CD8_Tm
CTL2
pDC
CTL1
NK
p_NKT
FOXN4
Squamous
Ciliated
Ionocyte
moMa
CD11c_mDC
nrMa
Cil_acute_phase
Club
Goblet
Basal
Sec_diff
Neutrophil
p_T
Epi_lowRNA
mDC
rMa
lowRNA
Plasma_cell
Cil_diff
DNT
IL-17A_CD8
IL-17A_CD4
ag_T
B_cell
MC/basophil
CD8_Tm
CTL2
pDC
CTL1
NK
p_NKT
FOXN4
Squamous
Ciliated
Ionocyte
moMa
CD11c_mDC
nrMa
Cil_acute_phase
Club
Goblet
Basal
Sec_diff
Neutrophil
p_T
Epi_lowRNA
mDC
rMa
lowRNA
Plasma_cell
Cil_diff
DNT
IL-17A_CD8
IL-17A_CD4
ag_T
B_cell
MC/basophil
CD8_Tm
CTL2
pDC
CTL1
NK
p_NKT
FOXN4
Squamous
Ciliated
Ionocyte
moMa
CD11c_mDC
nrMa
Cil_acute_phase
Club
Goblet
Basal
Sec_diff
Treg
Treg
Fig. 3 | Differential immune response in children and adults. a, Heat maps depicting cell–cell communications between all identified cell types derived
from linear-scaled ligand–receptor interaction counts in children (left, n = 18 SARS-CoV-2 negative, n = 11 SARS-CoV-2 positive early infection phase
(dps ≤ 4), n = 11 SARS-CoV-2 positive late infection phase (dps 5–12)) and adults (right, n = 23 SARS-CoV-2 negative, n = 13 SARS-CoV-2 positive
early infection phase, n = 8 SARS-CoV-2 positive late infection phase). b, Dot plots depicting expression of genes involved in antiviral and cytotoxic
response of myeloid dendritic cells and different macrophage populations. Each significant increase comparing SARS-CoV-2-negative children (n = 18)
with SARS-CoV-2-negative adults (n = 23) and SARS-CoV-2-positive children during the early (n = 11) or late (n = 11) infection phase with SARS-CoV-
2-positive adults amid early (n = 13), or late (n = 8) infection phase, respectively, is marked by a red circle (Benjamini–Hochberg-adjusted two-tailed
Wilcoxon, P < 0.05). Ave. exp., average gene expression; Pct. exp., percentage of cells expressing the gene. c, UMAP displaying the different T/NK cell
subpopulations in the nose of children and adults. In comparison, scaled density plots are shown indicating the proportion of the different
T/NK cells in children and adults separated by their SARS-CoV-2 infection status. Yellow represents high density; blue represents low density.
d, Violin plots showing expression of representative immune mediators of the CTL2 subpopulation comparing SARS-CoV-2-negative children (n = 18)
with SARS-CoV-2-negative adults (n = 23) and SARS-CoV-2-positive children during the early (n = 11) or late (n = 11) infection phase with SARS-CoV-
2-positive adults amid the early (n = 13) or late (n = 8) infection phase, respectively. Each dot represents one cell. Plots show median, ***P < 0.001
(IFNG negative: 1.50 × 10−17
, IFNG early phase: 1.14 × 10−47
, IFNG late phase: 1.69 × 10−27
, CCL5 negative: 5.97 × 10−30
, CCL5 late phase: 6.32 × 10−73
),
derived from two-tailed Wilcoxon comparison.
Nature Biotechnology | www.nature.com/naturebiotechnology
Articles NatureBiotechnology
patterns of immune cell subpopulations in children versus
adults. We identified, among others, a subpopulation of KLRC1
(NKG2A)+
cytotoxic T cells (CTL2) occurring predominantly
in children (Fig.  3c). NKG2A is a lectin-like inhibitory receptor
on cytotoxic T cells playing a role in limiting excessive activation,
preventing apoptosis and sustaining the virus-specific
CD8+
T cell response21
. Already without viral infection, this CD8
cytotoxic T cell subset was characterized by a strong expression
level of cytotoxic mediators (Fig. 3d, Extended Data Fig. 4
and Supplementary Table  5). Furthermore, IFNG was highly
expressed in these cells when comparing SARS-CoV-2-negative
children to adults. Upon infection, children were characterized
by a significantly higher expression level of IFNG compared to
adults both in the early phase and in the later phase of infection.
Similarly, the potent chemoattractant gene CCL5 was increased in
children compared to adults with or without infection (Fig. 3d).
The cytotoxic potential and the predominance of this cytotoxic
T cell subset necessary for efficient killing of virus-infected cells
provides further evidence for a better antivirus response in children
compared to adults. In addition, SARS-CoV-2-infected children
showed a distinct CD8+
T cell population (CD8_Tm) with
a memory phenotype that was almost absent in adults (Fig. 3c,
EDF4). It remains unclear whether these cells are beneficial for
protection of the children against future reinfection.
Discussion
Our data provide clear evidence that the epithelial and immune
cells of the upper airways (nose) of children are pre-activated and
primed for virus sensing. This is likely a general feature of the
children’s mucosal immune response, but of particular relevance
for SARS-CoV-2. Very recently, scRNA-seq of fibroblasts infected
with Chikungunya virus showed an extremely narrow window of
opportunity for the cells to express IFNs before viral protein production
shuts off the antiviral system22
. This likely also explains
the differences between SARS-CoV-2 and other respiratory
viruses including respiratory syncytial virus, influenza A virus or
SARS-CoV in terms of the induced host response. SARS-CoV-2 is
characterized by extensive intracellular replication and a remarkable
absence of IFN production and secretion. On the other hand,
SARS-CoV-2 is highly sensitive to treatment with IFNs before
or after infection, as shown in lung epithelial cells, even more
so than SARS-CoV20,23
. Primed virus sensing and a pre-activated
innate immune response in children leads to efficient early production
of IFNs in the infected airways, likely mediating substantial
antiviral effects mirroring those observed in vitro in
IFN-(pre)treated cells. Ultimately, this may lead to reduced virus
replication and faster clearance in children. In fact, several studies
already showed that children are much quicker in eliminating
SARS-CoV-2 compared to adults, consistent with the concept that
they shut down viral replication earlier24–27
. For other respiratory
viruses, such as respiratory syncytial virus and influenza A virus,
that more efficiently induce an IFN response by themselves, a
pre-activated innate immune response may be less relevant. The
enhanced innate antiviral capacity in children together with the
high IFN sensitivity of SARS-CoV-2 may explain why children are
better able to control early-stage infection as compared to adults
and therefore have a lower risk of developing severe COVID-19.
Online content
Any methods, additional references, Nature Research reporting
summaries, source data, extended data, supplementary information,
acknowledgements, peer review information; details of
author contributions and competing interests; and statements of
data and code availability are available at https://doi.org/10.1038/
s41587-021-01037-9.
Received: 8 April 2021; Accepted: 28 July 2021;
Published: xx xx xxxx
References
	1.	 Castagnoli, R. et al. Severe acute respiratory syndrome coronavirus
2 (SARS-CoV-2) infection in children and adolescents: a systematic review.
JAMA Pediatr. 174, 882–889 (2020).
	2.	 Goldstein, E., Lipsitch, M. & Cevik, M. On the effect of age on the
transmission of SARS-CoV-2 in households, schools, and the community.
J. Infect. Dis. 223, 362–369 (2021).
	3.	 O’Driscoll, M. et al. Age-specific mortality and immunity patterns of
SARS-CoV-2. Nature 590, 140–145 (2021).
	4.	 Park, A. & Iwasaki, A. Type I and type III interferons - induction, signaling,
evasion, and application to combat COVID-19. Cell Host Microbe 27,
870–878 (2020).
	5.	 Pierce, C. A. et al. Natural mucosal barriers and COVID-19 in children.
JCI Insight 6, e148694 (2021).
	6.	 Yoshida, M. et al. The local and systemic response to SARS-CoV-2 infection
in children and adults. Preprint at medRxiv https://doi.org/10.1101/
2021.03.09.21253012 (2021).
	7.	 Hou, Y. J. et al. SARS-CoV-2 reverse genetics reveals a variable infection
gradient in the respiratory tract. Cell 182, 429–446 (2020).
	8.	 Chua, R. L. et al. COVID-19 severity correlates with airway epithelium–
immune cell interactions identified by single-cell analysis. Nat. Biotechnol. 38,
970–979 (2020).
	9.	 Trump, S. et al. Hypertension delays viral clearance and exacerbates airway
hyperinflammation in patients with COVID-19. Nat. Biotechnol. (2020).
	10.	Koch, C. M. et al. Immune response to SARS-CoV-2 in the nasal mucosa
in children and adults. Preprint at medRxiv https://doi.org/10.1101/
2021.01.26.21250269 (2021).
	11.	V’Kovski, P., Kratzel, A., Steiner, S., Stalder, H. & Thiel, V. Coronavirus
biology and replication: implications for SARS-CoV-2. Nat. Rev. Microbiol.
19, 155–170 (2021).
	12.	Wolfel, R. et al. Virological assessment of hospitalized patients with
COVID-2019. Nature 581, 465–469 (2020).
	13.	Rebendenne, A. et al. SARS-CoV-2 triggers an MDA-5-dependent interferon
response which is unable to control replication in lung epithelial cells. J. Virol.
95, e02415–e02420 (2021).
	14.	Yin, X. et al. MDA5 governs the innate immune response to SARS-CoV-2
in lung epithelial cells. Cell Rep. 34, 108628 (2021).
	15.	Bruns, A. M., Leser, G. P., Lamb, R. A. & Horvath, C. M. The innate immune
sensor LGP2 activates antiviral signaling by regulating MDA5-RNA
interaction and filament assembly. Mol. Cell 55, 771–781 (2014).
	16.	Zhang, Q. et al. Inborn errors of type I IFN immunity in patients with
life-threatening COVID-19. Science 370, eabd4570 (2020).
	17.	Bastard, P. et al. Auto-antibodies to type I IFNs can underlie adverse
reactions to yellow fever live attenuated vaccine. J. Exp. Med. 218,
e20202486 (2021).
	18.	Pfaender, S. et al. LY6E impairs coronavirus fusion and confers immune
control of viral disease. Nat. Microbiol. 5, 1330–1339 (2020).
	19.	Martin-Sancho, L. et al. Functional landscape of SARS-CoV-2 cellular
restriction. Mol. Cell 81, 2656–2668.e8 (2021).
	20.	Hayn, M. et al. Systematic functional analysis of SARS-CoV-2 proteins
uncovers viral innate immune antagonists and remaining vulnerabilities.
Cell Rep. 35, 109126 (2021).
	21.	Rapaport, A. S. et al. The inhibitory receptor NKG2A sustains virus-specific
CD8+
T cells in response to a lethal poxvirus infection. Immunity 43,
1112–1124 (2015).
	22.	Pott, F. et al. Single-cell transcriptomic analysis of antiviral responses and
viral antagonism in Chikungunya virus-infected synovial fibroblasts. Preprint
at bioRxiv https://doi.org/10.1101/2020.06.07.138610 (2021).
	23.	Lokugamage, K. G. et al. Type I interferon susceptibility distinguishes
SARS-CoV-2 from SARS-CoV. J. Virol. 94, e01410-20 (2020).
	24.	Dufort, E. M. et al. Multisystem inflammatory syndrome in children in New
York state. N. Engl. J. Med. 383, 347–358 (2020).
	25.	Feldstein, L. R. et al. Multisystem inflammatory syndrome in U.S. children
and adolescents. N. Engl. J. Med. 383, 334–346 (2020).
	26.	Weisberg, S. P. et al. Distinct antibody responses to SARS-CoV-2 in
children and adults across the COVID-19 clinical spectrum. Nat. Immunol.
22, 25–31 (2021).
	27.	Whittaker, E. et al. Clinical characteristics of 58 children with a pediatric
inflammatory multisystem syndrome temporally associated with
SARS-CoV-2. JAMA 324, 259–269 (2020).
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in
published maps and institutional affiliations.
© The Author(s), under exclusive licence to Springer Nature America, Inc. 2021
Nature Biotechnology | www.nature.com/naturebiotechnology
ArticlesNatureBiotechnology
Methods
Patient recruitment and ethics approval. Individuals of three different cohorts
were included in this study. Patients of the prospective observational cohort study
Pa-COVID-19 (ref. 28
) and its study arm RECAST (Understanding the increased
resilience of children compared to adults in SARS-CoV-2 infection) were enrolled
between August 2020 and June 2021 at Charité – Universitätsmedizin Berlin.
Further patients were recruited in the prospective SC2-Study8,9
at University
Hospital Leipzig between March 2020 and May 2021. Written informed consent
was given by all patients and/or their parents before inclusion. All three studies
were conducted in accordance with the Declaration of Helsinki and approved by
the respective Institutional Review Boards (Pa-COVID-19/RECAST: EA2/066/20,
SC2: 123/20-ek).
Patient cohort. From the three cohorts, patients classified as asymptomatic, mild
or moderate based on WHO guidelines29
of COVID-19 severity were enrolled.
In total, we analyzed nasal swabs of 45 confirmed SARS-CoV-2-positive patients
comprising 24 children with ages ranging from 4 weeks to 17 years (median
age 9.0 ± 5.6 years, 10 females, 14 males) and from 21 adults between 21 and 76
years (median age 39.0 ± 10.4 years, 12 females, 9 males). None of the children
was hospitalized, but all were in domestic quarantine. Additionally, nasal swabs
of 42 healthy, SARS-CoV-2-negative controls from 18 children between 4 and
16 years (median age 9.0 ± 3.8 years, 8 females, 10 males) and from 23 adults
between 24 and 77 years were included (median age 46.0 ± 16.3 years, 13 females,
10 male; Supplementary Tables 3 and 4). All negative controls were examined for
possible exposure to SARS-CoV-2 by detailed anamnesis. In the cases of known
SARS-CoV-2 exposures, additional serological testing was conducted at the time of
sampling and after 14 days or follow-up interviews were conducted to ensure that
the participant as well as their household members showed no signs of infection for
at least two consecutive weeks and that any routine testing yielded negative results.
Real-time PCR for SARS-CoV-2. RNA was extracted by using the MagNA Pure
96 DNA and Viral NA Small Volume Kit (Roche) on a MagNA Pure 96 System
as recommended by the manufacturer. Real-time PCR with reverse transcription
was performed targeting the envelope (E) gene and nucleocapsid (N) gene on the
Roche Light Cycler 480 system (Tib-Molbiol).
Obtaining a single-cell suspension from human nasal swabs, preparation for
scRNA-seq and subsequent preprocessing of the raw data. Sample processing,
single-cell and library preparation, and data analysis were performed as
documented previously8,9
. Briefly, fresh nasopharyngeal swabs were transferred
into cold DMEM/F12 medium (Gibco, 11039) and within 1 h processed further.
Under biosafety S2, an equal volume of 13 mM dithiothreitol (AppliChem,
A2948) was added to each sample. To achieve higher cell numbers, the solution
was slowly pipetted up and down, and the swab was dipped roughly 20 times
into the medium. Following incubation at 37 °C and 500 r.p.m. for 10 min on
a thermomixer, samples were centrifuged at 350g at 4 °C for 5 min and the
supernatant slowly removed. If the pellet showed any sign of red blood cells, it
was resuspended in 1× PBS (Sigma-Aldrich, D8537), treated with Red Blood Cell
Lysis Buffer (Roche, 11814389001) at 25 °C for 10 min and centrifuged at 350g
at 4 °C for 5 min. If samples were not processed immediately, the cell pellet was
resuspended in DMEM/F12 supplemented with 20% FBS (Gibco, 10500) and 10%
dimethylsulfoxide (Sigma-Aldrich, D8418) and frozen at −80 °C. For the library
preparations, cells were thawed at 37 °C, centrifuged at 350g at 4 °C for 5 min and
further processed according to the protocol. To obtain a single-cell suspension,
Accutase (Thermo Fisher, 00-4555-56) was added to the pellet, and the solution
was incubated at room temperature for 10 min with careful pipetting of the cells
after 5 min. The incubation was stopped by adding DMEM/F12 supplemented with
10% FBS and centrifugation at 350g at 4 °C for 5 min. Subsequently, the supernatant
was removed and the cell pellet was resuspended in 1× PBS (volume was adjusted
to the size of the cell pellet). The suspension was cleared of any cell debris using
a 35-μm cell strainer (Falcon, 352235), and subsequently, cells were counted
with a disposable Neubauer chamber (NanoEnTek, DHC-N01). Cell suspension
was diluted to allow loading of 17,500 cells per sample. A single-cell and unique
barcode emulsion was achieved by mixing the diluted cells with the master mix
and loading them on the chip together with the Gel Beads and Partitioning Oil
using the 10x Genomics Single Cell 3ʹ GEM, Library and Gel Bead Kit v3.1 (10x
Genomics; PN 1000120; PN 1000121; PN 1000213) and loading the chip into
the 10x Chromium Controller. The following reverse transcription, clean-up and
cDNA amplification, as well as the library preparation, were performed according
to the manufacturer’s manual. Notably, to make sure that the virus was inactivated,
we prolonged the incubation at 85 °C during the reverse transcription to 10 min.
Final 3ʹ RNA libraries were pooled for sequencing either on an S2 or S4 flow cell
(S2: up to 13 samples, S4: up to 24 samples) and sequenced on the NovaSeq 6000
Sequencing System (Illumina, paired-end, single-indexing).
Single-cell datasets were aligned and preprocessed using Cellranger 3.0.1.
A custom human hg19 reference genome (10x Genomics, version 3.1.0) with the
SARS-CoV-2 genome (Refseq-ID: NC_045512) added as additional chromosome
was used. For downstream analysis Seurat 3.2.2 was used. Cells with fewer than 3
genes and cells with more than or equal to 15% mitochondrial reads or fewer than
200 genes expressed were discarded. To remove doublets, a cutoff for the number
of unique molecular identifiers and genes was determined manually per sample.
Samples were merged and exported as csv files containing counts and metadata
and imported into scanpy 1.6.030
. Following normalization to 10,000 reads per cell,
expression values were log transformed and highly variable genes were calculated,
which were used as a basis for the following preprocessing steps. The data were
scaled, PCA transformed and aligned using harmony 0.0.5 (ref. 31
) based on 100
principal components. Further alignment was performed using bbknn 1.4.0
(ref. 32
) based on 50 prealigned principal components, 10 trees, 3 neighbors within
batch and a trim setting of 85. Based on the integrated data, UMAP embedding
and Leiden33
clustering were performed. This clustering was used as a basis for
subclustering of immune and epithelial cell populations with the same algorithm.
Clusters of epithelial34–36
and immune37–39
cell populations were assigned to cell
types/stages according to the expression level of different marker genes (Extended
Data Figs. 1 and 4). The T cell and macrophage/dendritic cell clusters were
subclustered and then further refined manually.
The object was stored as h5ad, converted to h5seurat using SeuratDisk version
0.0.0.9014 and imported back into R version 4.1.0.
In total, 268,745 cells were included in the dataset. Cell numbers between
the groups were equally distributed (negative children 51,595; negative adults
62,701; positive adults 51,500) with the exception of the group of positive children
containing higher cell numbers (102,949). Different samples contributed varying
numbers of cells. The percentages of contribution of each sample to its study group
were compared using a Kruskal–Wallis test, which did not indicate significant
differences between groups (P = 0.2).
To enable visual comparisons between UMAPs of different groups, equal
numbers of cells (45,000) per group were randomly sampled using the SubsetData
function in Seurat.
Putative cell–cell interactions were quantified using CellPhoneDB version 2.1.2
using default settings40
. To reduce the influence of individual samples contributing
a larger number of cells and to speed up computation, we capped the number
of cells per sample at 2,000 randomly sampled cells. This was done using the
SubsetData function in Seurat.
Identification of ISG gene set. For the analysis of PRR/IFN responses, a gene
set of the most prominent ISGs expressed by lung epithelial cells was assembled.
As described previously9
, we treated A549 epithelial cells with a mix of IFNβ and
IFNλ for 2, 8 or 24 h, and analyzed transcript levels by microarray analyses using
the Illumina Human HT-12 v3 Expression Beadchip platform at the genomic
and proteomics core facility at DKFZ. We identified ISGs as exhibiting a log2[fold
change] > 0.8 at any time point, yielding 183 genes. We further included ISGs
described to exhibit strong anti-SARS-CoV-2 activity19
(65 top-scoring genes) if
not already included in our list. This eventually yielded a gene set of 217 genes also
expressed in our scRNA-seq.
SARS-CoV-2 infection of MDA5-expressing A549 cells. A549 cells stably
transduced with a lentiviral vector expressing human IFIH1 under the control
of the murine ROSA26 promoter (termed A549 MDA5high
in Fig. 2e) were
provided by Nadine Gillich and Ralf Bartenschlager. We transduced A549 (termed
MDA5low
) and A549 MDA5high
cells using lentiviral vectors encoding human
ACE2 and TMPRSS2 to make them permissive for SARS-CoV-2 infection; to
ensure consistent ACE2/TMPRSS2 expression across experiments, transduction
was freshly performed 24 h before infection of cells. Infection with SARS-CoV-2
(strain BetaCoV/Germany/BavPat1/2020) was performed in our BSL3 facility at a
multiplicity of infection of 0.1, and cells were collected at 24 h post infection. RNA
was extracted using the Monarch Total RNA Miniprep Kit (New England Biolabs)
and reverse transcribed by the High Capacity cDNA Reverse Transcription Kit
(Thermo Fisher Scientific). IFNB1 and ISG transcript levels were then assessed
by real-time PCR using the iTaq Universal SYBR Green Supermix (BioRad)
on a BioRad CFX96 Real-Time PCR Detection System. Data are presented as
mean ± s.e.m. of three biologically independent replicates.
Statistics. Differential gene expression was calculated using rank_genes_groups()
in scanpy version 1.6.0 and corrected for false discovery rates with statsmodels
version 0.9.0 (ref. 41
). Differences in cell type/stage compositions were assessed
using the Kruskal–Wallis rank sum test followed by two-tailed Dunn’s post hoc
test. Age dependencies were calculate fitting a linear regression model corrected
for the COVID-19 status and adjusting the F-test P values using the Benjamini–
Hochberg method42
.
P values for dot plots and violin plots were calculated using the Seurat function
FindMarkers() based on the Wilcoxon test and corrected with the Benjamini–
Hochberg method42
.
To test whether elevated MDA5 and RIG-I levels in A549 cells significantly
increased by IFNB1 and ISG induction upon infection, an unpaired one-tailed
Student t-test of three biologically independent repetitions was performed
(GraphPad Prism v9.1).
Reporting Summary. Further information on research design is available in
the Nature Research Reporting Summary linked to this article.
Nature Biotechnology | www.nature.com/naturebiotechnology
Articles NatureBiotechnology
Data availability
Processed data in the form of a Seurat object are available without access
restrictions on FigShare (https://doi.org/10.6084/m9.figshare.14938755). This
dataset can be used to replicate and extend the analyses presented in this paper.
Due to potential risk of de-identification of pseudonymized RNA-seq data, the
raw sequencing data can be obtained through EGA (EGAS00001005461) for
non-commercial research purposes alone, subject to controlled access as mandated
by EU data protection laws. For access, contact the corresponding author, R.E.
In addition, these data can be further visualized and analyzed in the Magellan
COVID-19 data explorer at https://digital.bihealth.org.
References
	28.	Kurth, F. et al. Studying the pathophysiology of coronavirus disease 2019:
a protocol for the Berlin prospective COVID-19 patient cohort
(Pa-COVID-19). Infection 48, 619–626 (2020).
	29.	Aylward, B., Liang, W. & WHO-China-Joint-Mission. Report of the
WHO-China Joint Mission on Coronavirus Disease 2019 (COVID-19)
(WHO, 2020).
	30.	Wolf, F. A., Angerer, P. & Theis, F. J. SCANPY: large-scale single-cell gene
expression data analysis. Genome Biol. 19, 15 (2018).
	31.	Korsunsky, I. et al. Fast, sensitive and accurate integration of single-cell data
with Harmony. Nat. Methods 16, 1289–1296 (2019).
	32.	Polański, K. et al. BBKNN: fast batch alignment of single cell transcriptomes.
Bioinformatics 36, 964–965 (2020).
	33.	Traag, V. A., Waltman, L. & van Eck, N. J. From Louvain to Leiden:
guaranteeing well-connected communities. Sci. Rep. 9, 5233 (2019).
	34.	Lukassen, S. et al. SARS-CoV-2 receptor ACE2 and TMPRSS2 are primarily
expressed in bronchial transient secretory cells. EMBO J. 39, e105114 (2020).
	35.	Plasschaert, L. W. et al. A single-cell atlas of the airway epithelium reveals the
CFTR-rich pulmonary ionocyte. Nature 560, 377–381 (2018).
	36.	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).
	37.	Mogilenko, D. A. et al. Comprehensive profiling of an aging immune system
reveals clonal GZMK+
CD8+
T cells as conserved hallmark of inflammaging.
Immunity 54, 99–115 (2021).
	38.	Travaglini, K. J. et al. A molecular cell atlas of the human lung from
single-cell RNA sequencing. Nature 587, 619–625 (2020).
	39.	Zhao, S. T. & Wang, C. Z. Regulatory T cells and asthma. J. Zhejiang Univ.
Sci. B 19, 663–673 (2018).
	40.	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. Nat. Protoc. 15,
1484–1506 (2020).
	41.	Seabold, S. & Perktold, J. Statsmodels: econometric and statistical modeling
with Python. Proc. 9th Python in Science Conference 57–61 (2010).
	42.	Benjamini, Y. & Hochberg, Y. Controlling the false discovery rate: a practical
and powerful approach to multiple testing. J. R. Stat. Soc. 57, 289–300 (1995).
Acknowledgements
We thank all patients of the RECAST, Pa-COVID-19 and SC2 studies for donating
nasal samples and clinical data. This study was supported by the BIH COVID-19
research program and the fightCOVID@DKFZ initiative, the European commission
(ESPACE, 874719, Horizon 2020), the German Federal Ministry for Education and
Research (BMBF)-funded de.NBI Cloud within the German Network for Bioinformatics
Infrastructure (de.NBI; 031A537B, 031A533A, 031A538A, 031A533B, 031A535A,
031A537C, 031A534A and 031A532B), the BMBF-funded Medical Informatics Initiative
(HiGHmed, 01ZZ1802A - 01ZZ1802Z), the BMBF-funded projects 01IK20337 and
82DZL0098B1, the German Research Foundation (Deutsche Forschungsgemeinschaft,
DFG)-funded CRC-TR 84 B08, CRC-1449 Z01 and project ID 272983813-TRR179 and
the DFG COVID-19 focus funding project BI1693/2-1. This publication is part of the
Human Cell Atlas—www.humancellatlas.org/publications.
Author contributions
S.T., S. Lukassen, I.L., M.A.M. and R.E. conceived, designed and supervised the project.
B.S., M.R., J.R., L.S., V.M.C. and M.A.M. designed the RECAST cohort. J.R., S.S., L.S.
and S. Laudi recruited the patients and provided the human specimens, clinical data
and annotation of the patients. J. Loske, A.S., J. Liebig, M.M. and R.L.C. performed
scRNA experiments. V.G.M. and M.B. performed and analyzed the in vitro experiments.
J. Loske, L.T., S. Lukassen and R.L.C. analyzed data. B.S., L.S., S. Laudi, V.M.C., M.R.,
C.C., M.B. and M.A.M. contributed with discussion of the results. B.T. and S.K. provided
technical and experimental support. J. Loske, S. Lukassen, S.T., M.B., R.E. and I.L. wrote
the manuscript; all authors read, revised and approved the manuscript.
Competing interests
The authors declare no competing interests.
Additional information
Extended data is available for this paper at https://doi.org/10.1038/s41587-021-01037-9.
Supplementary information The online version contains supplementary material
available at https://doi.org/10.1038/s41587-021-01037-9.
Correspondence and requests for materials should be addressed to R.E.
Peer review information Nature Biotechnology thanks Melanie Neeland and the other,
anonymous, reviewer(s) for their contribution to the peer review of this work.
Reprints and permissions information is available at www.nature.com/reprints.
Nature Biotechnology | www.nature.com/naturebiotechnology
ArticlesNatureBiotechnology
Extended Data Fig. 1 | Cell type marker genes. (a-b) Dot plots depicting average and percent expression of genes used to classify (a) immune and
(b) epithelial cells. Pct. Exp = percentage of cells expressing the gene, Ave. Exp. = average gene expression.
Nature Biotechnology | www.nature.com/naturebiotechnology
Articles NatureBiotechnology
Extended Data Fig. 2 | Cell composition in children and adults and ACE2 expression. (a) Distribution of all cell types/ states in children and adults
separated by SARS-CoV-2 infection status. Given are percentages of the total number of cells. Comparisons by Kruskal-Wallis test followed by Dunn’s
two-sided post hoc comparison (*significance of comparison between children and adults, P < 0.05; Neutrophil: 5.54E-07, mDC: 1.47E-03, rMa: 3.71E-03,
NK: 8.73E-03, p_NKT: 4.96E-02, ag_T: 6.09E-04, Treg: 8.50E-03, DNT: 1.62E-04, CD8_Tm: 1.97E-04, IL17A_CD8: 3.05E-04, IL17A_CD4: 2.68E-03, CTL2:
1.83E-05, CTL1: 3.60E-03, Plasma_cell: 1.83E-05, B_cell: 2.45E-03, immune_total: 1.61E-06, Basal: 6.96E-05, Cil_diff: 2.35E-03, Cil_acute_phase: 8.92E-
06, Ciliated: 9.18E-07, Sec_diff: 5.09E-03). (b) Violin plots show distribution of the SARS-CoV-2 entry receptor ACE2 expression and the gene expression
of associated proteases in epithelial cells of children and adults either with (n=24 children, n=21 adults) or without SARS-CoV-2 infection (n=18 children,
n=23 adults). Plots show median. No significant differences were observed.
Nature Biotechnology | www.nature.com/naturebiotechnology
ArticlesNatureBiotechnology
Extended Data Fig. 3 | Epithelial IFN-response. Heat maps depicting scaled expression (zero mean, unit variance) of known IFN-response genes in all
epithelial cells and separately in ciliated cells across conditions during early and late SARS-CoV-2 infection phase (early: dps ≤ 4: n=11 children, n=13
adults; late dps >4: n=11 children, n=8 adults). Genes were filtered according to a minimum fold change of 1.5 between the lowest and highest expressing
group, with the gene being selected according to their differential expression in all epithelial cells. Genes not meeting the criterion were greyed out to avoid
inflating minor differences as a consequence of the scaling performed. Scaled Exp. = scaled expression.
Nature Biotechnology | www.nature.com/naturebiotechnology
Articles NatureBiotechnology
Extended Data Fig. 4 | Characteristics of NK and T-cells. Expression profile of cytotoxic and aging-related genes in the NK and T-cell subtypes derived
from n=86 nasal swap samples distinguishing the children’s and adults’ immune profile. Ave. Exp. = average gene expression, Pct. Exp. = percentage of
cells expressing the gene.
Nature Biotechnology | www.nature.com/naturebiotechnology