PreprintPDF Available
Preprints and early-stage research may not have been peer reviewed yet.

Abstract

Lung carcinoid tumors, also referred to as pulmonary neuroendocrine tumors or lung carcinoids, are rare neoplasms of the lung with a more favorable prognosis than other subtypes of lung cancer. Still, some patients suffer from relapsed disease and metastatic spread while no consensus treatment exists for metastasized carcinoids. Several recent single-cell studies have provided detailed insights into the cellular heterogeneity of more common lung cancers, such as adeno- and squamous cell carcinoma. However, the characteristics of lung carcinoids on the single-cell level are yet completely unknown.To study the cellular composition and single-cell gene expression profiles in lung carcinoids, we applied single-cell RNA sequencing to three lung carcinoid tumor samples and normal lung tissue. The single-cell transcriptomes of carcinoid tumor cells reflected intertumoral heterogeneity associated with clinicopathological features, such as tumor necrosis and proliferation index. The immune microenvironment was specifically enriched in noninflammatory monocyte-derived myeloid cells. Tumor-associated endothelial cells were characterized by distinct gene expression profiles. A spectrum of vascular smooth muscle cells and pericytes predominated the stromal microenvironment. We found a small proportion of myofibroblasts exhibiting features reminiscent of cancer-associated fibroblasts. Stromal and immune cells exhibited potential paracrine interactions which may shape the microenvironment via NOTCH, VEGF, TGFβ and JAK/STAT signaling. Moreover, single-cell gene signatures of pericytes and myofibroblasts demonstrated prognostic value in bulk gene expression data.Here, we provide first comprehensive insights into the cellular composition and single-cell gene expression profiles in lung carcinoids, demonstrating the non-inflammatory and vessel-rich nature of their tumor microenvironment, and outlining relevant intercellular interactions which could serve as future therapeutic targets.
1
The single-cell transcriptional landscape of lung carcinoid
tumors
Philip Bischoff1,2, Alexandra Trinks1,2, Jennifer Wiederspahn1,3, Benedikt Obermayer4, Jan
Patrick Pett4, Philipp Jurmeister1,2,5, Aron Elsner6, Tomasz Dziodzio2,6, Jens-Carsten
Rückert6, Jens Neudecker6, Christine Falk7,8, Dieter Beule4, Christine Sers1,2,9, Markus
Morkel1,2,9, David Horst1,2,9, Frederick Klauschen1,5,9,10, Nils Blüthgen1,2,3,9,10
Affiliations:
1) Charité - Universitätsmedizin Berlin, corporate member of Freie Universität Berlin and
Humboldt-Universität zu Berlin, Institute of Pathology, Charitéplatz 1, 10117 Berlin,
Germany
2) Berlin Institute of Health at Charité - Universitätsmedizin Berlin, Charitéplatz 1, 10117
Berlin, Germany
3) IRI Life Sciences, Humboldt University of Berlin, Philippstrasse 13, 10115 Berlin, Germany
4) Berlin Institute of Health at Charité - Universitätsmedizin Berlin, Core Unit Bioinformatics,
Charitéplatz 1, 10117 Berlin, Germany
5) Institute of Pathology, LMU Munich, Thalkirchner Straße 36, 80337 München, Germany
6) Charité - Universitätsmedizin Berlin, corporate member of Freie Universität Berlin and
Humboldt-Universität zu Berlin, Department of Surgery, Campus Charité Mitte and
Campus Virchow-Klinikum, Charitéplatz 1, 10117 Berlin, Germany
7) Institute of Transplant Immunology, Hannover Medical School, Carl-Neuberg-Str. 1, 30625
Hannover, Germany
8) DZIF, German Center for Infectious Diseases, TTU-IICH Hannover-Braunschweig site,
38124 Braunschweig, Germany
9) German Cancer Consortium (DKTK), Partner Site Berlin, and German Cancer Research
Center (DKFZ), 69120 Heidelberg, Germany
10) joint last authors
Corresponding author: Philip Bischoff, philip.bischoff@charite.de, +49 30 450 536 127
Competing interests: The authors declare no potential conflicts of interest.
.CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted December 8, 2021. ; https://doi.org/10.1101/2021.12.07.471416doi: bioRxiv preprint
2
Abstract
Lung carcinoid tumors, also referred to as pulmonary neuroendocrine tumors or lung
carcinoids, are rare neoplasms of the lung with a more favorable prognosis than other subtypes
of lung cancer. Still, some patients suffer from relapsed disease and metastatic spread while
no consensus treatment exists for metastasized carcinoids. Several recent single-cell studies
have provided detailed insights into the cellular heterogeneity of more common lung cancers,
such as adeno- and squamous cell carcinoma. However, the characteristics of lung carcinoids
on the single-cell level are yet completely unknown.
To study the cellular composition and single-cell gene expression profiles in lung carcinoids,
we applied single-cell RNA sequencing to three lung carcinoid tumor samples and normal lung
tissue. The single-cell transcriptomes of carcinoid tumor cells reflected intertumoral
heterogeneity associated with clinicopathological features, such as tumor necrosis and
proliferation index. The immune microenvironment was specifically enriched in non-
inflammatory monocyte-derived myeloid cells. Tumor-associated endothelial cells were
characterized by distinct gene expression profiles. A spectrum of vascular smooth muscle cells
and pericytes predominated the stromal microenvironment. We found a small proportion of
myofibroblasts exhibiting features reminiscent of cancer-associated fibroblasts. Stromal and
immune cells exhibited potential paracrine interactions which may shape the microenvironment
via NOTCH, VEGF, TGFβ and JAK/STAT signaling. Moreover, single-cell gene signatures of
pericytes and myofibroblasts demonstrated prognostic value in bulk gene expression data.
Here, we provide first comprehensive insights into the cellular composition and single-cell gene
expression profiles in lung carcinoids, demonstrating the non-inflammatory and vessel-rich
nature of their tumor microenvironment, and outlining relevant intercellular interactions which
could serve as future therapeutic targets.
.CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted December 8, 2021. ; https://doi.org/10.1101/2021.12.07.471416doi: bioRxiv preprint
3
Introduction
Lung cancer is a heterogeneous disease comprising different histopathological subtypes.
Besides adenocarcinomas and squamous cell carcinomas, the 2015 WHO classification
established the category of pulmonary neuroendocrine tumors (NETs) [1]. This category
comprises the high-grade subtypes small cell lung cancer (SCLC) and large cell
neuroendocrine carcinoma (LCNEC), and the low- and intermediate-grade NETs of the lung,
also referred to as typical and atypical carcinoids, respectively. Lung carcinoids contribute to
1 % of lung cancer cases [2] with an increasing incidence over the last decades [3]. On
average, lung carcinoids have a better outcome than conventional lung cancers. Typical
carcinoids and atypical carcinoids, of which the latter are specified by higher mitotic rate or
presence of tumor necrosis, have a 5-year survival rate of approximately 90% and 70%,
respectively [3, 4]. About 10% of carcinoid patients present with regional lymph node
metastasis [3, 5]. Atypical carcinoids have a higher risk of lymphonodal and systemic
metastatic spread, and recurrent disease [5, 6]. However, no consensus exists for a
standardized systemic therapeutic regimen of metastasized lung carcinoids [7].
The more common subtypes of lung cancer, i.e., adenocarcinomas and squamous cell
carcinomas, are related to smoking and characterized by high tumor mutational burden. In
contrast, lung carcinoids affect younger patients and non-smokers, harbor a significantly lower
mutational load and a different spectrum of oncogenic mutations [8]. Consequently, novel
targeted and immune therapies, which have already improved the outcome in lung adeno- and
squamous cell carcinomas [9], cannot easily be translated to lung carcinoids. Moreover,
predicting the efficacy of modern targeted and immune therapies is limited by intratumoral
heterogeneity, where tumors may harbor primary resistant tumor cell subclones, as well as the
complex tumor microenvironment, modulating immune responses against the tumor. Single-
cell gene expression profiling allows to overcome this limitation and has already provided
valuable insights into the cellular heterogeneity of lung adenocarcinomas [10-14].
In this study, we comprehensively analyzed the cellular composition of lung carcinoids
by applying single-cell RNA sequencing to three carcinoid tumor and normal lung tissue
samples. We show that single-cell gene expression profiles of carcinoid tumor cells reflect
clinicopathological features and allow assignment to recently defined molecular clusters [15].
Further, we found that the tumor microenvironment was characterized by differentiating
monocyte-derived myeloid cells with non-inflammatory features, tumor-associated endothelial
cells, a spectrum of vascular smooth muscle cells and pericytes, and myofibroblasts with
cancer-associated fibroblast-like features. Our analysis provides the basis for further studies
of the lung carcinoid tumor microenvironment, potential prognostic and predictive biomarkers
as well as novel therapeutic targets.
.CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted December 8, 2021. ; https://doi.org/10.1101/2021.12.07.471416doi: bioRxiv preprint
4
Results
Single-cell RNA sequencing uncovers the cellular diversity of lung carcinoids
To explore the cellular composition of lung carcinoids and their tumor microenvironment on the
single-cell level, we collected fresh tissue samples of tumor tissue and normal lung
parenchyma from 3 previously untreated lung carcinoid patients undergoing primary surgery
(patients 1-3, Fig. 1A). All three patients showed tumor cells growing in solid nests with
expression of the neuroendocrine marker proteins synaptophysin and chromogranin A (Fig.
1B). Tumors comprised one typical carcinoid (patient 1), one atypical carcinoid with high
proliferative activity (patient 2, see Supp. Fig. 1A for Ki67 immunostaining), and one atypical
carcinoid with focal tumor necrosis (patient 3, see Supp. Fig. 1B for HE staining of necrotic
area). Both atypical carcinoid cases (patients 2 and 3) had regional lymph node metastases at
the time of diagnosis. Tissue samples were enzymatically dissociated and subjected to single-
cell RNA sequencing using a commercial droplet-based system. Single-cell gene expression
data of 7 normal lung tissue samples from a previously published cohort (patients 4-10) [14]
were included in the subsequent analyses. Altogether, we analyzed 73,105 single-cell
transcriptomes of which 64,697 high-quality transcriptomes remained after quality control and
filtering (Fig. 1C, see Supp. Fig. 1C-D for quality control parameters).
Visualization of single-cell transcriptomes by uniform manifold approximation and
projection (UMAP) revealed distinct shifts between normal and tumor tissue samples (Fig. 1D).
Note that single-cell transcriptomes of different patients overlapped in many clusters, excluding
systematic batch effects across samples (Fig. 1E). In the epithelial, immune and stromal cell
compartment, which were defined by gene expression of canonical marker genes (Supp. Fig.
1E), we observed tumor-specific changes (Fig. 1F). In the tumor tissue samples, we mostly
found epithelial and stromal single-cell transcriptomes, whereas immune single-cell
transcriptomes were more abundant in the normal tissue samples (Fig. 1G).
.CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted December 8, 2021. ; https://doi.org/10.1101/2021.12.07.471416doi: bioRxiv preprint
5
Figure 1: Single-cell RNA sequencing of lung carcinoids
(A) Clinical characteristics of 3 lung carcinoid patients analyzed in this study. (B) H&E staining
and immunohistochemical staining for neuroendocrine marker proteins of 3 lung carcinoid
patients. (C) Simplified schematic representation of the single-cell RNA sequencing workflow.
(D-F) UMAPs based on the top 10 principal components of all single-cell transcriptomes after
filtering, color-coded by (D) tissue type, (E) patients, and (F) main cell type. (G) Proportions of
main cell types per sample.
Intertumoral heterogeneity of lung carcinoids reflects clinicopathological features and
molecular subtypes
To further analyze the epithelial cell compartment, epithelial single-cell transcriptomes were
subset and re-clustered. Epithelial cell clusters overrepresented in normal or tumor tissue
samples were assigned as normal or tumor cell clusters, respectively (Supp. Fig. 2A, 2B). We
observed that normal cell clusters were shared by different patients whereas tumor cell clusters
were highly patient-specific (Fig. 2A). In the normal cell clusters, using canonical marker genes
and predefined gene signatures [16, 17], we identified alveolar epithelial type 1 and 2, ciliated,
club, and basal cells (Fig. 2A-B, Supp. Fig. 3A-B). As indicated by the highly patient-specific
tumor cell clusters, we found many differentially expressed genes in the tumor cells between
.CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted December 8, 2021. ; https://doi.org/10.1101/2021.12.07.471416doi: bioRxiv preprint
6
patients (Fig. 2C), such as CD44, TFF3 and EGFR, which correlated with differential protein
expression, as shown by immunohistochemistry (Fig. 2D). Highly expressed genes in the
typical carcinoid of patient 1 comprised many that have been associated with good prognosis,
such as MT1G, MT1M, MT1X, PCK1, LPL, CD44 [18]. While we found distinct transcriptional
differences between different tumor cases, transcriptional profiles within individual tumors were
quite homogeneous and varied mainly depending on the number of reads and genes per cell
(Supp. Fig. 2C-F).
To further explore interpatient heterogeneity, we inferred different functional traits from
the single-cell gene expression profiles, namely cell cycle phase, proportion of mitochondrial
reads and signaling pathway activity. The tumor cells of patient 2 had the highest proportion
of cells in S phase while at the same time showing the highest Ki67 proliferation index in
immunohistochemistry (Fig. 2E, see Supp. Fig. 1A for Ki67 immunostaining). The highest
proportion of mitochondrial reads was observed in tumor cells of patient 3 which was
characterized by focal tumor necrosis (Fig. 2F, see Supp. Fig. 1B for H&E of necrotic area).
Tumor cell transcriptomes of patient 3 had high scores for EGFR pathway activity and strong
EGFR expression on the protein level (Fig. 2D, 2G). The pathway activity scores for estrogen
and androgen receptor signaling correlated with the patient’s sex (Fig. 2G, see Fig. 1A for
clinical characteristics). Recently, it has been shown that lung carcinoids can be subtyped into
three distinct molecular clusters based on transcriptional and epigenetic features [15]. In our
single-cell gene expression profiles, we could assign the tumors of patients 2 and 3 to cluster
A1, and the tumor of patient 1 both to cluster A2 and cluster B (Fig. 2H). Notably, the immune
and stromal cell compartment exhibited only minor expression scores of molecular cluster
gene signatures.
Taken together, lung carcinoid tumor single-cell transcriptomes revealed intertumoral
heterogeneity which reflected different clinical and histomorphological features, such as
patient’s sex, tumor proliferative activity and tumor necrosis, as well as recently proposed
molecular clusters of lung carcinoids.
.CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted December 8, 2021. ; https://doi.org/10.1101/2021.12.07.471416doi: bioRxiv preprint
7
Figure 2: Intertumoral heterogeneity of tumor cells in lung carcinoids
(A) UMAPs based on the top 10 principal components of all epithelial single-cell transcriptomes
color-coded by tissue type, patient, and cell type. (B) Average gene expression of selected
marker genes of normal epithelial cell types, for cell type color code see (A). (C) Differentially
expressed genes in tumor epithelial cells grouped by patients, top 20 genes showed per
patient, for patient color code see (A). (D) Immunohistochemical staining of proteins encoded
by selected differentially expressed genes indicated by black arrowheads in (C). (E) Proportion
of tumor epithelial cells assigned to different cell cycles, grouped by patient. (F) Proportion of
mitochondrial reads in tumor epithelial transcriptomes, grouped by patient. (G) Mean pathway
activity scores of tumor epithelial cells, grouped by patient, and normal epithelial cells, grouped
.CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted December 8, 2021. ; https://doi.org/10.1101/2021.12.07.471416doi: bioRxiv preprint
8
by patient groups. (H) Module scores of marker genes of molecular clusters according to Alcala
et al. [15] in epithelial, immune, and stromal cells, grouped by patient.
The immune microenvironment of lung carcinoids is characterized by non-inflammatory
monocyte-derived myeloid cells
To discover the cellular composition of the immune microenvironment, immune single-cell
transcriptomes were subset and re-clustered. We identified a variety of different cell types
within the immune cell compartment using canonical marker genes and predefined gene
signatures [16, 17] (Fig. 3A-B, Supp. Fig. 3A-B).
We identified different lymphoid cells, such as conventional T cells, CD8+ T cells, NK
cells, B cells and plasma cells (Fig. 3A, Supp. Fig. 4A-B, Supp. Table 1-2). On average, the
proportion of lymphoid cell types in tumor tissues closely resembled normal tissues while we
also noted some interpatient heterogeneity in both tumor and normal tissues. Gene signatures
reflecting naiveness, cytotoxicity or exhaustion [13] of lymphoid cells showed similar
expression scores in normal and tumor tissues (Fig. 3C). Note that signature scores for
exhaustion were low in all lymphoid cell types, contrasting the microenvironment of lung
adenocarcinoma in a reference dataset, which was enriched in exhausted T cells (Supp. Fig.
5A-B) [14]. Together, the lymphoid microenvironment of lung carcinoids resembles the
lymphoid cell compartment of normal lung parenchyma.
Within the myeloid cell compartment, we identified monocytes, dendritic cells,
macrophages and mast cells with different abundancies in tumor and normal tissues (Fig. 3A,
Supp. Fig. 4A, Supp. Table 3-4). Among the monocytes, the classical monocyte cluster 2 was
enriched in tumor tissues (Fig. 3D). Conventional dendritic cells comprised two clusters of
which the CD141+ cluster 1 was mostly found in tumor tissues (Fig. 3E). Beyond, we identified
a tumor-enriched cell cluster with high expression levels of both monocyte markers, such as
CD14, and LGMN, a gene upregulated in differentiating monocytes [19] (Fig. 3B). We conclude
that this cell cluster represents the spectrum of monocyte-derived myeloid cells differentiating
either into macrophages, as shown by high APOC1 and APOE expression in cluster 3, or into
dendritic cells, as shown by high S100A8 and S100A9 expression in cluster 1 [20, 21] (Supp.
Fig. 4D). While the proportions of monocytes and conventional dendritic cells were
heterogeneously increased across patients, monocyte-derived myeloid cells were consistently
increased across all three carcinoid tumors analyzed (p = 0.0070). Monocyte-derived myeloid
cells were further characterized by high expression of SELENOP, C1QA, C1QB, C1QC and
the chemokines CCL3 and CCL4 (Fig. 3F). Compared to normal tissues, monocyte-derived
myeloid cells in tumor tissues showed equal to slightly lower expression scores of various gene
signatures related to inflammation and immune response (Fig. 3G). In a reference dataset of
lung adenocarcinoma, the microenvironment was composed of both pro- and non-
.CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted December 8, 2021. ; https://doi.org/10.1101/2021.12.07.471416doi: bioRxiv preprint
9
inflammatory clusters (Supp. Fig. 5A-B). In contrast, our results indicate that the lung carcinoid
immune microenvironment is predominated by non-inflammatory monocyte-derived myeloid
cells.
The composition of the myeloid cell compartment was to some degree heterogeneous
across the three lung carcinoids analyzed by single-cell RNA sequencing. In order to study
interpatient heterogeneity in a larger cohort, we quantified the expression of marker genes of
characteristic immune cell types in a published bulk gene expression dataset of lung carcinoids
[15, 22]. Here, we found that marker genes of the tumor-enriched monocyte-derived myeloid
cell cluster 2 and 3 were associated with atypical carcinoids, albeit not correlated with overall
survival (Supp. Fig. 4E), indicating microenvironmental differences between lung carcinoid
subtypes.
.CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted December 8, 2021. ; https://doi.org/10.1101/2021.12.07.471416doi: bioRxiv preprint
10
Figure 3: Composition of the immune tumor microenvironment in lung carcinoids
(A) UMAPs based on the top 10 principal components of all immune single-cell transcriptomes,
split by tissue type, color-coded by cell type, and proportions of lymphoid and myeloid cell
types per tissue type and, for tumor samples, per patient, Mann-Whitney U test, ** = p<0.01.
(B) Average gene expression of selected marker genes of immune cell types, for cell type color
.CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted December 8, 2021. ; https://doi.org/10.1101/2021.12.07.471416doi: bioRxiv preprint
11
code see (A). (C) Module scores of gene signatures related to naiveness, cytotoxicity and
exhaustion in different lymphoid cell types, split by tissue type, for cell type color code see (A).
(D) Average gene expression of selected marker genes of monocyte subsets and relative size
of monocyte clusters, for tissue type color code see (C). (E) Average gene expression of
selected marker genes of conventional dendritic cell subsets and relative size of conventional
dendritic cell clusters, for tissue type color code see (C). (F) Differentially expressed genes in
myeloid cells in tumor samples, grouped by cell type, top 10 genes shown per cell type, for cell
type color code see (A), black arrowheads indicate genes mentioned in the main text. (G)
Module scores of gene signatures related to immune response in different myeloid cell types,
split by tissue type, for cell type color code see (A), for tissue type color code see (C).
Vascular cells and CAF-like myofibroblasts constitute the stromal microenvironment of
lung carcinoids
To gain insight into the composition of the stromal microenvironment, stromal single-cell
transcriptomes were subset and re-clustered. Here, we identified different clusters of
endothelial, fibroblastic and smooth muscle cells using canonical marker genes and predefined
gene signatures [16, 17] (Fig. 4A-4B, Supp. Fig. 3A-B).
Among the endothelial cells we could distinguish bronchial, capillary, arterial and
venous endothelial cells based on predefined marker gene signatures [16] (Fig. 4A, Supp. Fig.
3A, Supp. Table 5-6). While different subtypes of endothelial cells were present in normal lung
parenchyma, tumor tissues were significantly enriched in bronchial-type endothelial cells (p =
0.0070) (Fig. 4A, Supp. Fig. 6A). Here, the majority of endothelial transcriptomes was obtained
from the tumor of patient 1, correlating with dense vascularization as shown by immunostaining
(Supp. Fig. 6B). Endothelial cells in tumor tissues showed high mRNA expression of INSR, a
marker gene of tumor-associated endothelial cells [23], and high INSR protein expression,
contrasting normal lung tissue (Supp. Fig. 6C-D). Moreover, we found high expression of
genes that have been related to angiogenesis, such as VWA1, COL15A1, IGFBP7 and GSN
[10] (Supp. Fig. 6D), suggesting a phenotype of tumor-associated endothelial cells in the
microenvironment of lung carcinoids comparable to those found in the microenvironment of
lung adenocarcinoma (Supp. Fig. 5A, C) [14].
Within the fibroblastic and smooth muscle cell compartment in tumor tissues, we found
myofibroblasts, vascular smooth muscle cells and pericytes, whereas fibroblasts were
significantly decreased compared to normal tissues (p = 0.0091) (Fig. 4A, Supp. Table 7-8).
Myofibroblasts were strongly enriched in tumor tissues (p = 0.0074) and showed high
expression of extracellular matrix components, such as COL1A1, COL3A1 and COL6A3, as
well as matrix-degrading enzymes, such as SULF1 (Fig. 4C), suggesting that these cells might
be involved in extracellular matrix remodeling. Moreover, myofibroblasts were characterized
.CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted December 8, 2021. ; https://doi.org/10.1101/2021.12.07.471416doi: bioRxiv preprint
12
by high activity of TGFβ signaling (Fig. 4D). Within the myofibroblasts, cluster 1 showed a
specific overexpression of FAP and MMP11 as well as a higher expression of various
collagens, compared to myofibroblast cluster 2 and other fibroblast and smooth muscle cell
clusters (Fig. 4E). In an independent cohort of lung carcinoids characterized by bulk RNA
sequencing [15, 22], the marker gene signature of myofibroblast cluster 1 was significantly
associated with atypical carcinoids and correlated with worse overall survival (Supp. Fig. 6E).
Together, these data indicate that myofibroblasts in lung carcinoid tumor tissues exhibit
biological traits characteristic of cancer-associated fibroblasts.
The stromal microenvironment of carcinoids was predominated by vascular smooth
muscle cells (p = 0.0160) and pericytes, contrasting the microenvironment of lung
adenocarcinoma which mainly contained myofibroblasts (Supp. Fig. 5A, C) [14]. The highest
proportion of pericytes was found in case 1 which was diagnosed as a typical carcinoid
whereas fewer pericytes were found in patients 2 and 3 diagnosed with atypical carcinoids
(Fig. 4A). Correspondingly, in an independent lung carcinoid cohort characterized by bulk RNA
sequencing [15, 22], the pericyte marker gene signature was associated with typical carcinoids
and correlated with better overall survival (Supp. Fig. 6F). Pericytes in tumor tissues showed
a high expression of RGS5, a gene involved in pericyte development, and ACTA2, a smooth
muscle marker gene (Fig. 4F). Smooth muscle cells expressed low levels of pericyte marker
genes, such as COX4I2 and PDGRB (Fig. 4C). We found these genes expressed in a graded
fashion suggesting that pericytes and vascular smooth muscle cells rather form a continuum
than discrete cell types in tumor tissues (Fig. 4G). These results show that the stromal
microenvironment of lung carcinoids is composed of myofibroblasts reminiscent of cancer-
associated fibroblasts, and a spectrum of vascular smooth muscle cells and pericytes.
Myofibroblasts and pericytes may be linked to worse and better overall survival, respectively.
.CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted December 8, 2021. ; https://doi.org/10.1101/2021.12.07.471416doi: bioRxiv preprint
13
Figure 4: Composition of the stromal tumor microenvironment in lung carcinoids
(A) UMAPs based on the top 10 principal components of all stromal single-cell transcriptomes
split by tissue type, color-coded by cell type, and proportions of endothelial and
fibroblastic/smooth muscle cell types per tissue type and, for tumor samples, per patient,
Mann-Whitney U test, * = p<0.05, ** = p<0.01. (B) Average gene expression of selected marker
genes of stromal cell types, for cell type color code see (A). (C) Differentially expressed genes
in fibroblastic/smooth muscle cells, grouped by cell type, top 10 genes shown per cell type, for
cell type color code see (A), black arrowheads indicate genes mentioned in the main text. (D)
.CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted December 8, 2021. ; https://doi.org/10.1101/2021.12.07.471416doi: bioRxiv preprint
14
Mean pathway activity scores of different fibroblastic/smooth muscle cell clusters, mesothelial
cells excluded, black arrowheads indicate pathways and cell types mentioned in the main text.
(E) Average expression of selected marker genes of myofibroblast cluster 1 as indicated by
black arrowhead. (F) Average gene expression of pericytes in tumor versus normal tissues,
top 10 genes overexpressed in tumor tissues indicated, genes mentioned in the main text in
bold. (G) UMAPs of fibroblastic/smooth muscle cells, colored by gene expression of canonical
pericyte marker genes.
Interactions between tumor microenvironmental cells potentially activate NOTCH,
VEGF, TGFβ, and JAK/STAT signaling
We observed that different cell types are specifically enriched in the lung carcinoid
microenvironment (Fig. 5A). In order to delineate functional relationships between
microenvironmental and tumor cells, we quantified potential paracrine receptor-ligand
interactions [24]. Interestingly, most potential interactions were found between cell types of the
stromal microenvironment, involving tumor-associated endothelial cells, myofibroblasts,
vascular smooth muscle cells and pericytes, whereas tumor cells are less involved in potential
paracrine interactions (Fig. 5B, Supp. Table 9). Note that the number of potential interactions
was independent from the number of cells or mean number of mRNA counts per cell type
(Supp. Fig. 7A-B). Focusing on the most relevant signaling pathways, we found many
interactions potentially activating NOTCH, TGFβ, VEGF, and JAK/STAT signaling (Fig. 5C).
Tumor-associated endothelial cells receive potentially VEGF, TGFβ, and NOTCH pathway-
activating signals, while myofibroblast mainly receive potentially TGFβ pathway-activating
signals, both via various paracrine and autocrine interactions. Dendritic cells, monocytes and
monocyte-derived myeloid cells receive potentially JAK/STAT pathway-activating signals
mainly via autocrine and paracrine interactions with other immune cells. These results indicate
that autocrine and paracrine interactions between various stromal and immune cells may
shape the lung carcinoid tumor microenvironment.
.CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted December 8, 2021. ; https://doi.org/10.1101/2021.12.07.471416doi: bioRxiv preprint
15
Figure 5: Potential paracrine interactions within the lung carcinoid tumor
microenvironment
(A) Characteristic cell types of the lung carcinoid tumor microenvironment and selected cell
type marker genes. (B) Number of potential auto-/paracrine interactions between characteristic
cell types of the lung carcinoid tumor microenvironment, calculated using the CellPhoneDB
algorithm. (C) Number of potential cell-cell interactions filtered for high-confidence receptors
and ligands of relevant signaling pathways, grouped by interaction families. Each heatmap
shows potential interactions where the respective receptor is expressed in the cell type
indicated above.
.CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted December 8, 2021. ; https://doi.org/10.1101/2021.12.07.471416doi: bioRxiv preprint
16
Discussion
In this study, we analyzed the tumor heterogeneity and cellular composition of the tumor
microenvironment in lung carcinoids. By applying single-cell RNA sequencing to tumors from
three patients, we outline the single-cell landscape of lung carcinoids in unprecedented depth
and comprehensiveness. We could show that tumor cell transcriptomes reflect high
intertumoral but low intratumoral heterogeneity. The immune microenvironment was
characterized by non-inflammatory monocyte-derived myeloid cells, classical monocytes and
conventional dendritic cells, while the lymphoid cell compartment was comparable to normal
lung parenchyma. The stromal microenvironment was composed of tumor-associated
endothelial cells, myofibroblasts with features of cancer-associated fibroblasts, and a spectrum
of pericytes and vascular smooth muscle cells.
Since gene expression profiles are linked to the biological behavior of tumors,
transcriptional subtypes have been defined for many tumor entities. Recently, Alcala et al.
described three molecular clusters of lung neuroendocrine neoplasms based on transcriptome
and methylome profiling [15]. However, it is not known to what extent information from bulk
multi-omic profiling originate from tumor cells or the associated non-neoplastic immune and
stromal cells. Indeed, molecular subtypes of some entities have been shown to be mainly
driven by features of the tumor microenvironment, such as in colorectal cancer [25]. In our
dataset, we could show that assignment of lung carcinoids to recently defined molecular
clusters is not substantially driven by immune or stromal cells, but rather represent tumor-
intrinsic features. Nonetheless, molecular clusters of lung neuroendocrine neoplasms have
been suggested to be associated with distinct cell types of the tumor microenvironment [15].
Although the size of our dataset does not allow to define patient subgroups based on tumor
microenvironment composition, we observed that the tumor of patient 1 was assigned to
molecular cluster B and harbored the highest proportion of monocytes while patient 3 was
assigned to cluster A1 and harbored the highest proportion of conventional dendritic cells.
Exemplarily, this underlines the proposed association of molecular clusters of lung
neuroendocrine neoplasms with tumor microenvironment composition [15].
Many studies have dissected the composition of the immune microenvironment of lung
cancer and its potential effects on response to immune checkpoint blockade, being an
important pillar in treatment of advanced disease [26]. However, the immune cellular diversity
in lung carcinoids has much less been studied. It has been described that only a small
proportion of carcinoids are substantially infiltrated by CD8+ T cells, which does not correlate
with survival [27]. While most studies report no expression of PD-L1 in lung carcinoids at all
[27, 28], some studies report a small proportion of PD-L1-postive cases and a correlation of
PD-L1 expression with metastatic spread [29]. We observed that the composition of the
lymphoid cell compartment closely resembled normal lung parenchyma which is in line with a
.CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted December 8, 2021. ; https://doi.org/10.1101/2021.12.07.471416doi: bioRxiv preprint
17
recent study analyzing the lung carcinoid immune microenvironment by flow cytometry [30].
While it has been discussed that carcinoid tumors are not eligible for immune checkpoint
inhibitor therapy due to their low mutational and neoantigen load [31], still, combined anti-PD1
and anti-CTL4 blockade has shown efficacy in individual advanced atypical carcinoid cases
[32]. Beyond CD8+ T cells, myeloid cells in the microenvironment can modulate the response
to immune checkpoint inhibitors [33]. Across all three lung carcinoids analyzed, we observed
a consistent decrease in tissue-resident alveolar macrophages and an increase in monocyte-
derived myeloid cells, compared to normal lung tissue, which has likewise been observed in
lung adenocarcinomas [10, 14]. Monocyte-derived myeloid cells can give rise to monocyte-
derived dendritic cells or monocyte-derived macrophages exhibiting different functionalities in
the tumor microenvironment [20]. In our dataset, gene expression patterns favored a
differentiation towards monocyte-derived macrophages. However, complementary information
on protein expression is necessary to determine the lineage commitment of these cells since
cell types are yet mainly defined by surface marker profiles obtained in FACS studies [20].
Furthermore, in our study, monocyte-derived myeloid cells showed low expression scores of
various inflammation-related pathways, high expression of SELENOP1, which has been
associated with M2 polarization of tumor-associated macrophages, high expression of the
cytokines CCL3 and CCL4, which both can exert pro- or anti-tumorigenic functions [34, 35],
and high expression of all components of the C1q protein complex, which has been found to
have tumor-promoting features [36, 37]. Together, we conclude that monocyte-derived myeloid
cells exhibit rather non-inflammatory and pro-tumorigenic features in the lung carcinoid tumor
microenvironment.
The microenvironment of neuroendocrine neoplasms often harbors a dense vascular
network. We found that tumor-associated endothelial cells in lung carcinoids exhibit a distinct
gene expression profile and share many highly expressed genes, such as INSR, VWA,
COL15A1, IGFBP7 and GSN, with tumor-associated endothelial cells of more aggressive
cancers, such as lung adenocarcinoma [10, 14]. In addition, we observed a high proportion of
vascular smooth muscle cells and pericytes. Antiangiogenic drugs have been in clinical trials
and the VEGFR inhibitor sunitinib has been approved for therapy of pancreatic neuroendocrine
tumors [31]. Interestingly, compared to normal tissues, tumor-associated pericytes expressed
high levels of RGS5, which has been found to be overexpressed in developing pericytes during
embryogenesis [38] and is associated with reduced response to VEGF inhibition in mouse
models [39]. Moreover, we observed high expression of smooth muscle actin ACTA2 in tumor-
associated pericytes, which has been proposed as a marker for tumors refractory to VEGFR2
inhibition in a pancreatic neuroendocrine tumor mouse model [40]. Furthermore, the
microenvironment of lung carcinoids contained a small proportion of myofibroblasts which were
characterized by high TGFβ and hypoxia signaling, high expression of matrix components,
.CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted December 8, 2021. ; https://doi.org/10.1101/2021.12.07.471416doi: bioRxiv preprint
18
matrix degrading enzymes, and marker genes such as FAP, all being features of cancer-
associated fibroblasts [41, 42].
The tumor microenvironment of lung carcinoids and lung adenocarcinomas shared
some features, such as tumor-associated endothelial cells, cancer-associated myofibroblasts,
and monocyte-derived myeloid cells. Other microenvironmental features were specific for lung
carcinoids, such as the predominance of vascular smooth muscle cells and pericytes, and the
normal-like composition of the lymphoid cell compartment. Beyond these patient-overarching
features of the lung carcinoid microenvironment, we observed inter-patient heterogeneity in its
cellular composition. In a larger cohort of lung carcinoids profiled by bulk RNA sequencing, we
found that certain cell types such as monocyte-derived myeloid cells, pericytes, and
myofibroblasts might be associated with different histological subtypes of lung carcinoids
(typical versus atypical) and patient prognosis. However, since the microenvironment of lung
carcinoids is rather sparse yet complex, its cellular composition can only to a limited extent be
inferred from bulk gene expression data. Therefore, our study forms a basis for subsequent
single-cell transcriptome profiling or multiplex immunofluorescence studies of larger cohorts.
In the future, a more detailed and comprehensive understanding of the tumor
microenvironment could reveal specific cell types that are eligible for novel targeted therapies,
and provide valuable prognostic and predictive information to improve the clinical management
of lung carcinoid patients.
.CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted December 8, 2021. ; https://doi.org/10.1101/2021.12.07.471416doi: bioRxiv preprint
19
Methods
Collection of tissue specimens
Fresh tissue samples of approximately 0.1-0.5 cm³ of tumor tissue and normal lung
parenchyma were obtained during intraoperative pathologist consultation. Informed consent
was obtained from all patients were. Research was approved by vote EA4/164/19 of the ethics
committee of Charité - Universitätsmedizin Berlin. Normal tissue samples of patients 4-10 have
already been part of a previous study (patient 4 = P018, patient 5 = P019, patient 6 = P027,
patient 7 = P029, patient 8 = P030, patient 9 = P031 and patient 10 = P033) [14].
Tissue dissociation and single cell isolation
For transport, tissue samples were stored for max. 3 hours on ice in Tissue Storage Solution
(Miltenyi). First, tissue samples were minced into pieces of max. 1 mm³ using two scalpels.
Minced tissue samples were disaggregated using the Tumor Dissociation Kit, human (Miltenyi)
according to the manufacturer’s protocol in a gentleMACS Octo Dissociator with heaters
(Miltenyi) using the preinstalled program 37C_h_TDK_1 for 30-45 min. Subsequently, cell
suspensions were filtered through 100 µm filters and kept at 4°C or on ice for all subsequent
steps. Next, cells were pelleted by centrifugation at 300 g for 5 min in BSA-coated low-binding
tubes, and resuspended in 1 ml ACK buffer for 60 seconds for erythrocyte lysis. Cells were
washed with DMEM, again pelleted, and resuspended in PBS. After filtering the cell
suspensions through 20 µm filters, debris was removed using the Debris Removal Solution
(Miltenyi) according to the manufacturer’s protocol. Finally, cell concentration was determined
using a Neubauer chamber.
Single-cell RNA sequencing
Immediately after single cell isolation, 10,000 single cells per tissue sample were subjected to
barcoding and library preparation, using the Chromium Single Cell 3´Reagent Kit v3 (10x
Genomics) and the Chromium Controller (10x Genomics) according to the manufacturer’s
protocol. Libraries were sequenced on a HiSeq 4000 Sequencer (Illumina) at average 240 mio.
reads per library, resulting in average approx. 50,000 reads per cell.
H&E and immunostaining
For hematoxylin and eosin (H&E), and immunohistochemical staining, 3-5 µm tissue sections
were prepared from formalin-fixed and paraffin-embedded (FFPE) tissue.
For H&E staining, tissue sections were incubated in acidic haemalum staining solution
(Waldeck) for 8 min, washed, and incubated in eosin staining solution (Sigma-Aldrich) for 2.5
min at room temperature using a Tissue-Tek Prisma Plus slide stainer (Sakura).
For antigen retrieval, tissue sections were incubated in CC2 buffer (for mouse anti-INSR) or
CC1 mild buffer (for all other antibodies, Ventana Medical Systems) for 30 min at 100°C.
Sections were incubated with the primary antibody for 60 min at room temperature, washed,
.CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted December 8, 2021. ; https://doi.org/10.1101/2021.12.07.471416doi: bioRxiv preprint
20
and incubated with the secondary antibody for 30 minutes at room temperature. Antibodies
were diluted in Dako Real Antibody Diluent (Dako, S2022). Staining was performed on the
BenchMark XT immunostainer (Ventana Medical Systems).
The following primary antibodies were used: mouse anti-Synaptophysin (1:50, clone 27G12,
Leica, NCL-L-SYNAP-299), rabbit anti-Chromogranin A (1:100, clone EP38, Epitomics, AC-
0037), mouse anti-CD44 (1:50, clone DF1485, Dako, M7082), rabbit anti-EGFR (prediluted,
Roche, 790-4347), rabbit anti-TFF3 (1:250, Abcam, ab108599), Rabbit anti-ERG (prediluted,
clone EPR3864, Roche, 790-4576), mouse anti-Ki67 (1:50, clone MIB-1, Dako, M7240),
mouse anti-INSR (1:50, clone CT-3, Invitrogen, AHR0271).
Slides were imaged using a Pannoramic SCAN 150 slide scanner (3DHISTECH).
Single-cell gene expression analysis
Preprocessing
After sequencing, reads were aligned and UMIs quantified using Cellranger 3.0.2 (10x
Genomics) with reference transcriptome GRCh38. All subsequent analyses were performed in
R using the toolkit Seurat v4 [43], if not stated otherwise. Single-cell gene expression data of
all patients were merged and filtered for the following quality parameters: 500-10,000 genes
detected, 1,000-100,000 UMIs counted, fraction of mitochondrial reads <40%, and fraction of
hemoglobin reads <5%. Single-cell gene expression data was normalized using the
scTransform function with default parameters, and the number of UMIs per cell and the fraction
of mitochondrial reads was regressed out.
Cell type annotation
After principal component analysis (PCA), the top 10 principal components were used for
clustering and UMAP embedding of single-cell transcriptomes. Main cell types (epithelial,
immune, stromal) were assigned based on cluster-wise expression of canonical cell type
marker genes (resolution = 0.3, otherwise default parameters). The dataset was split into three
main cell type subsets, and PCA, clustering and UMAP embedding was rerun on each subsets
using the top 10 principal components and a clustering resolution of 2 with otherwise default
parameters. In order to assign epithelial, immune, and stromal cell types, selected cell type
marker genes according from Habermann et al. [44] and Tata et al. [45], and cell type
signatures according to Vieira Braga et al. [17] and Travaglini et al. [16] were used. Clusters
contaminated with epithelial or immune transcriptomes were identified by expression of
EPCAM or PTPRC, respectively, and removed from the dataset prior to subsequent analyses.
In the epithelial subset, cell clusters which were overrepresented in tumor tissue samples were
annotated as tumor cells.
Differential gene expression analysis
Prior to differential gene expression analysis of epithelial cells, tumor cells from tumor samples
were subset and gene expression rescaled. Immune and stromal subsets were split into
.CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted December 8, 2021. ; https://doi.org/10.1101/2021.12.07.471416doi: bioRxiv preprint
21
lymphoid, myeloid, endothelial and fibroblastic/smooth muscle subsets, and gene expression
rescaled. Next, marker genes of each cell cluster were calculated against all other clusters of
the subset using the FindAllMarkers function with Wilcoxon rank-sum test and the following
parameters: include only positive markers, proportion of expressing cells inside the cluster ≥
0.25, difference between proportions of expressing cells inside and outside the cluster ≥ 0.25,
log2 fold change between cells inside and outside the cluster ≥ 0.25.
Functional analysis
Cell cycle phases were assigned using the CellCycleScoring function. The AddModuleScore
function was used to score the expression of functional relevant gene signatures: the hallmark
signatures of the collection of the Broad Institute [46], naiveness, cytotoxicity and exhaustion
signatures according to Guo et al. [13], and molecular cluster marker genes according to Alcala
et al. [15]. For the latter, the top 50 upregulated genes in each individual molecular cluster
versus the two other clusters were combined and used as gene sets. Oncogenic signaling
pathway activity scores were computed using the R toolkit Progeny [47, 48] based on the top
500 genes with otherwise default parameters. The CellPhoneDB toolkit was used with default
parameters to calculate potential cell-cell interactions [24]. The curated list of high-confidence
ligands and receptors of oncogenic pathways can be found in ref. [14].
Bulk gene expression and survival analysis
Bulk gene expression data was downloaded from the GitHub repository
https://github.com/IARCbioinfo/DRMetrics [22] and clinical data from [15] was added. After
filtering out genes located on sex and mitochondrial chromosomes, bulk gene expression data
was normalized using the VarianceStabilizingTransformation function of the DESeq2 toolkit.
Data on histological subtype (typical vs. atypical) was available for 75 carcinoid cases. Overall
survival data was available for 76 carcinoid cases. Single-cell gene expression data was split
into myeloid, lymphoid, endothelial and fibroblastic/smooth muscle subsets and rescaled.
Marker genes were calculated as described above. Next, marker gene lists were used as gene
sets for single-sample gene set enrichment analysis (ssGSEA) [49] of the bulk gene expression
data using the gsva function of the R toolkit GSVA assuming Gaussian distribution with
otherwise default parameters. For survival analyses, ssGSEA enrichment scores were
dichotomized (ES > median or ≤ median). Survival curves, log-rank statistics and Cox
regression were calculated using the R packages survival and survminer.
Code and data availability
The code used for analyses is available from https://github.com/bischofp/lung_carcinoid. Gene
expression count data is available from [link provided upon acceptance of the manuscript].
.CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted December 8, 2021. ; https://doi.org/10.1101/2021.12.07.471416doi: bioRxiv preprint
22
Funding
The work was in part funded by the Berlin Institute of Health (to PB, PJ, DH, MM, CS and NB),
and the German Cancer Consortium DKTK (to MM and NB).
PB is participant in the BIH-Charité Junior Clinician Scientist Program funded by the Charité -
Universitätsmedizin Berlin and the Berlin Institute of Health.
PJ is participant in the BIH-Charité Digital Clinician Scientist Program funded by the Charité -
Universitätsmedizin Berlin and the Berlin Institute of Health and the German Research
Foundation (DFG).
Author contributions
NB, FK, PB conceived and designed the study;
PB, AT, FK, DH, AE, TD, JN, JR contributed to clinical sample acquisition and preparation;
AT, PB conducted experiments;
PB, BO, JP, DB, NB performed bioinformatic analyses;
PB, FK, NB, PJ, CF, DH, MM, CS analyzed and interpreted data and/or supervised parts of
the study;
PB wrote the manuscript;
FK, NB, MM, DH revised the manuscript;
all authors provided critical feedback and helped shaping the research, analysis, and
manuscript.
Competing Interests
The authors declare no competing interest.
Acknowledgement
We thank Manuela Pacyna-Gengelbach and Barbara Meyer-Bartell for excellent technical
assistance.
.CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted December 8, 2021. ; https://doi.org/10.1101/2021.12.07.471416doi: bioRxiv preprint
23
References
1. Travis, W.D., et al., The 2015 World Health Organization Classification of Lung Tumors:
Impact of Genetic, Clinical and Radiologic Advances Since the 2004 Classification. J
Thorac Oncol, 2015. 10(9): p. 1243-1260.
2. Naalsund, A., et al., Carcinoid lung tumors--incidence, treatment and outcomes: a
population-based study. Eur J Cardiothorac Surg, 2011. 39(4): p. 565-9.
3. Petursdottir, A., et al., Pulmonary carcinoid tumours: incidence, histology, and surgical
outcome. A population-based study. Gen Thorac Cardiovasc Surg, 2020. 68(5): p. 523-
529.
4. Aydin, E., et al., Long-term outcomes and prognostic factors of patients with surgically
treated pulmonary carcinoid: our institutional experience with 104 patients. Eur J
Cardiothorac Surg, 2011. 39(4): p. 549-54.
5. Rea, F., et al., Outcome and surgical strategy in bronchial carcinoid tumors: single
institution experience with 252 patients. Eur J Cardiothorac Surg, 2007. 31(2): p. 186-91.
6. Thomas, C.F., Jr., H.D. Tazelaar, and J.R. Jett, Typical and atypical pulmonary carcinoids
: outcome in patients presenting with regional lymph node involvement. Chest, 2001.
119(4): p. 1143-50.
7. Caplin, M.E., et al., Pulmonary neuroendocrine (carcinoid) tumors: European
Neuroendocrine Tumor Society expert consensus and recommendations for best practice
for typical and atypical pulmonary carcinoids. Ann Oncol, 2015. 26(8): p. 1604-20.
8. Derks, J.L., et al., New Insights into the Molecular Characteristics of Pulmonary Carcinoids
and Large Cell Neuroendocrine Carcinomas, and the Impact on Their Clinical
Management. J Thorac Oncol, 2018. 13(6): p. 752-766.
9. Howlader, N., et al., The Effect of Advances in Lung-Cancer Treatment on Population
Mortality. N Engl J Med, 2020. 383(7): p. 640-649.
10. Kim, N., et al., Single-cell RNA sequencing demonstrates the molecular and cellular
reprogramming of metastatic lung adenocarcinoma. Nat Commun, 2020. 11(1): p. 2285.
11. Lambrechts, D., et al., Phenotype molding of stromal cells in the lung tumor
microenvironment. Nat Med, 2018. 24(8): p. 1277-1289.
12. Lavin, Y., et al., Innate Immune Landscape in Early Lung Adenocarcinoma by Paired
Single-Cell Analyses. Cell, 2017. 169(4): p. 750-765 e17.
13. Guo, X., et al., Global characterization of T cells in non-small-cell lung cancer by single-
cell sequencing. Nat Med, 2018. 24(7): p. 978-985.
14. Bischoff, P., et al., Single-cell RNA sequencing reveals distinct tumor microenvironmental
patterns in lung adenocarcinoma. Oncogene, 2021.
.CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted December 8, 2021. ; https://doi.org/10.1101/2021.12.07.471416doi: bioRxiv preprint
24
15. Alcala, N., et al., Integrative and comparative genomic analyses identify clinically relevant
pulmonary carcinoid groups and unveil the supra-carcinoids. Nat Commun, 2019. 10(1):
p. 3407.
16. Travaglini, K.J., et al., A molecular cell atlas of the human lung from single-cell RNA
sequencing. Nature, 2020.
17. Vieira Braga, F.A., et al., A cellular census of human lungs identifies novel cell states in
health and in asthma. Nat Med, 2019. 25(7): p. 1153-1163.
18. Swarts, D.R., et al., An exploration of pathways involved in lung carcinoid progression
using gene expression profiling. Carcinogenesis, 2013. 34(12): p. 2726-37.
19. Solberg, R., et al., Legumain expression, activity and secretion are increased during
monocyte-to-macrophage differentiation and inhibited by atorvastatin. Biol Chem, 2015.
396(1): p. 71-80.
20. Haas, L. and A.C. Obenauf, Allies or Enemies-The Multifaceted Role of Myeloid Cells in
the Tumor Microenvironment. Front Immunol, 2019. 10: p. 2746.
21. Collin, M. and V. Bigley, Human dendritic cell subsets: an update. Immunology, 2018.
154(1): p. 3-20.
22. Gabriel, A.A.G., et al., A molecular map of lung neuroendocrine neoplasms. Gigascience,
2020. 9(11).
23. Kiflemariam, S., et al., Tumor vessel up-regulation of INSR revealed by single-cell
expression analysis of the tyrosine kinome and phosphatome in human cancers. Am J
Pathol, 2015. 185(6): p. 1600-9.
24. Efremova, M., et al., CellPhoneDB: inferring cell-cell communication from combined
expression of multi-subunit ligand-receptor complexes. Nat Protoc, 2020. 15(4): p. 1484-
1506.
25. Lee, H.O., et al., Lineage-dependent gene expression programs influence the immune
landscape of colorectal cancer. Nat Genet, 2020.
26. House, I.G., et al., Macrophage-Derived CXCL9 and CXCL10 Are Required for Antitumor
Immune Responses Following Immune Checkpoint Blockade. Clin Cancer Res, 2020.
26(2): p. 487-504.
27. Kasajima, A., et al., Inflammation and PD-L1 expression in pulmonary neuroendocrine
tumors. Endocr Relat Cancer, 2018. 25(3): p. 339-350.
28. Tsuruoka, K., et al., PD-L1 expression in neuroendocrine tumors of the lung. Lung Cancer,
2017. 108: p. 115-120.
29. Vesterinen, T., et al., PD-1 and PD-L1 expression in pulmonary carcinoid tumors and their
association to tumor spread. Endocr Connect, 2019. 8(8): p. 1168-1175.
30. Stankovic, B., et al., The immune microenvironment in typical carcinoid lung tumour, a
brief report of four cases. Scand J Immunol, 2020. 92(2): p. e12893.
.CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted December 8, 2021. ; https://doi.org/10.1101/2021.12.07.471416doi: bioRxiv preprint
25
31. Cives, M., et al., The Tumor Microenvironment in Neuroendocrine Tumors: Biology and
Therapeutic Implications. Neuroendocrinology, 2019. 109(2): p. 83-99.
32. Klein, O., et al., Immunotherapy of Ipilimumab and Nivolumab in Patients with Advanced
Neuroendocrine Tumors: A Subgroup Analysis of the CA209-538 Clinical Trial for Rare
Cancers. Clin Cancer Res, 2020. 26(17): p. 4454-4459.
33. Peranzoni, E., et al., Myeloid Cells as Clinical Biomarkers for Immune Checkpoint
Blockade. Front Immunol, 2020. 11: p. 1590.
34. Ntanasis-Stathopoulos, I., D. Fotiou, and E. Terpos, CCL3 Signaling in the Tumor
Microenvironment. Adv Exp Med Biol, 2020. 1231: p. 13-21.
35. Mukaida, N., S.I. Sasaki, and T. Baba, CCL4 Signaling in the Tumor Microenvironment.
Adv Exp Med Biol, 2020. 1231: p. 23-32.
36. Bulla, R., et al., C1q acts in the tumour microenvironment as a cancer-promoting factor
independently of complement activation. Nat Commun, 2016. 7: p. 10346.
37. Mangogna, A., et al., Is the Complement Protein C1q a Pro- or Anti-tumorigenic Factor?
Bioinformatics Analysis Involving Human Carcinomas. Front Immunol, 2019. 10: p. 865.
38. Cho, H., et al., Pericyte-specific expression of Rgs5: implications for PDGF and EDG
receptor signaling during vascular maturation. FASEB J, 2003. 17(3): p. 440-2.
39. Mitchell, T.S., et al., RGS5 expression is a quantitative measure of pericyte coverage of
blood vessels. Angiogenesis, 2008. 11(2): p. 141-51.
40. Franco, M., et al., Use of a mouse model of pancreatic neuroendocrine tumors to find
pericyte biomarkers of resistance to anti-angiogenic therapy. Horm Metab Res, 2011.
43(12): p. 884-9.
41. Yoshida, G.J., Regulation of heterogeneous cancer-associated fibroblasts: the molecular
pathology of activated signaling pathways. J Exp Clin Cancer Res, 2020. 39(1): p. 112.
42. Petrova, V., et al., The hypoxic tumour microenvironment. Oncogenesis, 2018. 7(1): p. 10.
43. Hao, Y., et al., Integrated analysis of multimodal single-cell data. Cell, 2021. 184(13): p.
3573-3587 e29.
44. Habermann, A.C., et al., Single-cell RNA sequencing reveals profibrotic roles of distinct
epithelial and mesenchymal lineages in pulmonary fibrosis. Sci Adv, 2020. 6(28): p.
eaba1972.
45. Tata, P.R. and J. Rajagopal, Plasticity in the lung: making and breaking cell identity.
Development, 2017. 144(5): p. 755-766.
46. Liberzon, A., et al., The Molecular Signatures Database (MSigDB) hallmark gene set
collection. Cell Syst, 2015. 1(6): p. 417-425.
47. Schubert, M., et al., Perturbation-response genes reveal signaling footprints in cancer
gene expression. Nat Commun, 2018. 9(1): p. 20.
.CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted December 8, 2021. ; https://doi.org/10.1101/2021.12.07.471416doi: bioRxiv preprint
26
48. Holland, C.H., et al., Robustness and applicability of transcription factor and pathway
analysis tools on single-cell RNA-seq data. Genome Biol, 2020. 21(1): p. 36.
49. Barbie, D.A., et al., Systematic RNA interference reveals that oncogenic KRAS-driven
cancers require TBK1. Nature, 2009. 462(7269): p. 108-12.
.CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted December 8, 2021. ; https://doi.org/10.1101/2021.12.07.471416doi: bioRxiv preprint
ResearchGate has not been able to resolve any citations for this publication.
Article
Full-text available
Recent developments in immuno-oncology demonstrate that not only cancer cells, but also the tumor microenvironment can guide precision medicine. A comprehensive and in-depth characterization of the tumor microenvironment is challenging since its cell populations are diverse and can be important even if scarce. To identify clinically relevant microenvironmental and cancer features, we applied single-cell RNA sequencing to ten human lung adenocarcinomas and ten normal control tissues. Our analyses revealed heterogeneous carcinoma cell transcriptomes reflecting histological grade and oncogenic pathway activities, and two distinct microenvironmental patterns. The immune-activated CP²E microenvironment was composed of cancer-associated myofibroblasts, proinflammatory monocyte-derived macrophages, plasmacytoid dendritic cells and exhausted CD8+ T cells, and was prognostically unfavorable. In contrast, the inert N³MC microenvironment was characterized by normal-like myofibroblasts, non-inflammatory monocyte-derived macrophages, NK cells, myeloid dendritic cells and conventional T cells, and was associated with a favorable prognosis. Microenvironmental marker genes and signatures identified in single-cell profiles had progonostic value in bulk tumor profiles. In summary, single-cell RNA profiling of lung adenocarcinoma provides additional prognostic information based on the microenvironment, and may help to predict therapy response and to reveal possible target cell populations for future therapeutic approaches.
Article
Full-text available
The simultaneous measurement of multiple modalities represents an exciting frontier for single-cell genomics and necessitates computational methods that can define cellular states based on multimodal data. Here, we introduce “weighted-nearest neighbor” analysis, an unsupervised framework to learn the relative utility of each data type in each cell, enabling an integrative analysis of multiple modalities. We apply our procedure to a CITE-seq dataset of 211,000 human peripheral blood mononuclear cells (PBMCs) with panels extending to 228 antibodies to construct a multimodal reference atlas of the circulating immune system. Multimodal analysis substantially improves our ability to resolve cell states, allowing us to identify and validate previously unreported lymphoid subpopulations. Moreover, we demonstrate how to leverage this reference to rapidly map new datasets and to interpret immune responses to vaccination and coronavirus disease 2019 (COVID-19). Our approach represents a broadly applicable strategy to analyze single-cell multimodal datasets and to look beyond the transcriptome toward a unified and multimodal definition of cellular identity.
Article
Full-text available
Background Lung neuroendocrine neoplasms (LNENs) are rare solid cancers, with most genomic studies including a limited number of samples. Recently, generating the first multi-omic dataset for atypical pulmonary carcinoids and the first methylation dataset for large-cell neuroendocrine carcinomas led us to the discovery of clinically relevant molecular groups, as well as a new entity of pulmonary carcinoids (supra-carcinoids). Results To promote the integration of LNENs molecular data, we provide here detailed information on data generation and quality control for whole-genome/exome sequencing, RNA sequencing, and EPIC 850K methylation arrays for a total of 84 patients with LNENs. We integrate the transcriptomic data with other previously published data and generate the first comprehensive molecular map of LNENs using the Uniform Manifold Approximation and Projection (UMAP) dimension reduction technique. We show that this map captures the main biological findings of previous studies and can be used as reference to integrate datasets for which RNA sequencing is available. The generated map can be interactively explored and interrogated on the UCSC TumorMap portal (https://tumormap.ucsc.edu/?p=RCG_lungNENomics/LNEN). The data, source code, and compute environments used to generate and evaluate the map as well as the raw data are available, respectively, in a Nextjournal interactive notebook (https://nextjournal.com/rarecancersgenomics/a-molecular-map-of-lung-neuroendocrine-neoplasms/) and at the EMBL-EBI European Genome-phenome Archive and Gene Expression Omnibus data repositories. Conclusions We provide data and all resources needed to integrate them with future LNENs transcriptomic studies, allowing meaningful conclusions to be drawn that will eventually lead to a better understanding of this rare understudied disease.
Article
Full-text available
Although single-cell RNA sequencing studies have begun to provide compendia of cell expression profiles1–9, it has been difficult to systematically identify and localize all molecular cell types in individual organs to create a full molecular cell atlas. Here, using droplet- and plate-based single-cell RNA sequencing of approximately 75,000 human cells across all lung tissue compartments and circulating blood, combined with a multi-pronged cell annotation approach, we create an extensive cell atlas of the human lung. We define the gene expression profiles and anatomical locations of 58 cell populations in the human lung, including 41 out of 45 previously known cell types and 14 previously unknown ones. This comprehensive molecular atlas identifies the biochemical functions of lung cells and the transcription factors and markers for making and monitoring them; defines the cell targets of circulating hormones and predicts local signalling interactions and immune cell homing; and identifies cell types that are directly affected by lung disease genes and respiratory viruses. By comparing human and mouse data, we identified 17 molecular cell types that have been gained or lost during lung evolution and others with substantially altered expression profiles, revealing extensive plasticity of cell types and cell-type-specific gene expression during organ evolution including expression switches between cell types. This atlas provides the molecular foundation for investigating how lung cell identities, functions and interactions are achieved in development and tissue engineering and altered in disease and evolution.
Article
Full-text available
Immune checkpoint inhibitors are becoming standard treatments in several cancer types, profoundly changing the prognosis of a fraction of patients. Currently, many efforts are being made to predict responders and to understand how to overcome resistance in non-responders. Given the crucial role of myeloid cells as modulators of T effector cell function in tumors, it is essential to understand their impact on the clinical outcome of immune checkpoint blockade and on the mechanisms of immune evasion. In this review we focus on the existing clinical evidence of the relation between the presence of myeloid cell subsets and the response to anti-PD(L)1 and anti-CTLA-4 treatment. We highlight how circulating and tumor-infiltrating myeloid populations can be used as predictive biomarkers for immune checkpoint inhibitors in different human cancers, both at baseline and on treatment. Moreover, we propose to follow the dynamics of myeloid cells during immunotherapy as pharmacodynamic biomarkers. Finally, we provide an overview of the current strategies tested in the clinic that use myeloid cell targeting together with immune checkpoint blockade with the aim of uncovering the most promising approaches for effective combinations.
Article
Full-text available
Pulmonary fibrosis (PF) is a form of chronic lung disease characterized by pathologic epithelial remodeling and accumulation of extracellular matrix (ECM). To comprehensively define the cell types, mechanisms, and mediators driving fibrotic remodeling in lungs with PF, we performed single-cell RNA sequencing of single-cell suspensions from 10 nonfibrotic control and 20 PF lungs. Analysis of 114,396 cells identified 31 distinct cell subsets/states. We report that a remarkable shift in epithelial cell phenotypes occurs in the peripheral lung in PF and identify several previously unrecognized epithelial cell phenotypes, including a KRT5 ⁻ / KRT17 ⁺ pathologic, ECM-producing epithelial cell population that was highly enriched in PF lungs. Multiple fibroblast subtypes were observed to contribute to ECM expansion in a spatially discrete manner. Together, these data provide high-resolution insights into the complexity and plasticity of the distal lung epithelium in human disease and indicate a diversity of epithelial and mesenchymal cells contribute to pathologic lung fibrosis.
Article
Full-text available
Accumulating evidence indicates that intratumoral heterogeneity contributes to the development of resistance to anticancer therapeutics. Fibroblasts, which are components of the paraneoplastic stroma, play a crucial role in the wound-healing process. Activated fibroblasts accumulate in the wound and are involved in many aspects of the tissue remodeling cascade that initiates the repair process and prevents further tissue damage. The pathophysiological roles of cancer-associated fibroblasts (CAFs) in the heterogeneous tumor microenvironment have attracted increasing interest. CAFs play crucial roles in tumor progression and the response to chemotherapy. Several cytokines and chemokines are involved in the conversion of normal fibroblasts into CAFs, and some of these form a feedback loop between cancer cells and CAFs. In addition, the physical force between tumor cells and CAFs promotes cooperative invasion or co-migration of both types of cells. Pro-inflammatory cytokines, such as leukemia inhibitory factor (LIF) and interleukin-6 (IL-6), are secreted by both cancer cells and CAFs, and mediate the epigenetic modification of CAFs. This enhances the pro-tumorigenic function of CAFs mediated by promoting actomyosin contractility and extracellular matrix remodeling to form the tracks used for collective cancer cell migration. The concept of intra-tumoral CAF heterogeneity refers to the presence of inflammatory CAFs with low levels of α-smooth muscle actin (α-SMA) and high levels of IL-6 expression, which are in striking contrast to transforming growth factor-β (TGF-β)-dependent myofibroblastic CAFs with high α-SMA expression levels. CAF populations that suppress tumor growth and progression through stroma-specific Hedgehog (Hh) activation have been detected in different murine tumor models including those of the bladder, colon, and pancreas. A new therapeutic strategy targeting CAFs is the "stromal switch," in which tumor-promoting CAFs are changed into tumor-retarding CAFs with attenuated stromal stiffness. Several molecular mechanisms that can be exploited to design personalized anticancer therapies targeting CAFs remain to be elucidated. Strategies aimed at targeting the tumor stroma as well as tumor cells themselves have attracted academic attention for their application in precision medicine. This novel review discusses the role of the activation of EGFR, Wnt/β-catenin, Hippo, TGF-β, and JAK/STAT cascades in CAFs in relation to the chemoresistance and invasive/metastatic behavior of cancer cells. For instance, although activated EGFR signaling contributes to collective cell migration in cooperation with CAFs, an activated Hippo pathway is responsible for stromal stiffness resulting in the collapse of neoplastic blood vessels. Therefore, identifying the signaling pathways that are activated under specific conditions is crucial for precision medicine.
Article
Full-text available
Immunotherapy for metastatic colorectal cancer is effective only for mismatch repair-deficient tumors with high microsatellite instability that demonstrate immune infiltration, suggesting that tumor cells can determine their immune microenvironment. To understand this cross-talk, we analyzed the transcriptome of 91,103 unsorted single cells from 23 Korean and 6 Belgian patients. Cancer cells displayed transcriptional features reminiscent of normal differentiation programs, and genetic alterations that apparently fostered immunosuppressive microenvironments directed by regulatory T cells, myofibroblasts and myeloid cells. Intercellular network reconstruction supported the association between cancer cell signatures and specific stromal or immune cell populations. Our collective view of the cellular landscape and intercellular interactions in colorectal cancer provide mechanistic information for the design of efficient immuno-oncology treatment strategies. Single-cell RNA sequencing of colorectal tumors highlights associations between lineage-related gene expression signatures and specific stromal and immune cell populations.
Article
Background Lung cancer is made up of distinct subtypes, including non–small-cell lung cancer (NSCLC) and small-cell lung cancer (SCLC). Although overall mortality from lung cancer has been declining in the United States, little is known about mortality trends according to cancer subtype at the population level because death certificates do not record subtype information. Methods Using data from Surveillance, Epidemiology, and End Results (SEER) areas, we assessed lung-cancer mortality and linked deaths from lung cancer to incident cases in SEER cancer registries. This allowed us to evaluate population-level mortality trends attributed to specific subtypes (incidence-based mortality). We also evaluated lung-cancer incidence and survival according to cancer subtype, sex, and calendar year. Joinpoint software was used to assess changes in incidence and trends in incidence-based mortality. Results Mortality from NSCLC decreased even faster than the incidence of this subtype, and this decrease was associated with a substantial improvement in survival over time that corresponded to the timing of approval of targeted therapy. Among men, incidence-based mortality from NSCLC decreased 6.3% annually from 2013 through 2016, whereas the incidence decreased 3.1% annually from 2008 through 2016. Corresponding lung cancer–specific survival improved from 26% among men with NSCLC that was diagnosed in 2001 to 35% among those in whom it was diagnosed in 2014. This improvement in survival was found across all races and ethnic groups. Similar patterns were found among women with NSCLC. In contrast, mortality from SCLC declined almost entirely as a result of declining incidence, with no improvement in survival. This result correlates with limited treatment advances for SCLC in the time frame we examined. Conclusions Population-level mortality from NSCLC in the United States fell sharply from 2013 to 2016, and survival after diagnosis improved substantially. Our analysis suggests that a reduction in incidence along with treatment advances — particularly approvals for and use of targeted therapies — is likely to explain the reduction in mortality observed during this period.
Article
Purpose: Combination immunotherapy with anti-CTLA-4 and anti-PD-1 blockade has demonstrated significant clinical activity across several tumour types. Neuroendocrine tumors (NETs) are a heterogeneous group of rare tumors with limited treatment options. CA209-538 is a clinical trial of combination immunotherapy with ipilimumab and nivolumab in rare cancers, including advanced NETs. Experimental design: CA209-538 is a prospective multicentre clinical trial in patients with advanced rare cancers. Patients received treatment with nivolumab at a dose of 3mg/kg and ipilimumab at 1mg/kg every three weeks for four doses, followed by nivolumab 3mg/kg every two weeks and continued for up to 96 weeks, until disease progression or the development of unacceptable toxicity. Response was assessed every 12 weeks by RECIST 1.1. The primary endpoint was clinical benefit rate. Results: Twenty-nine patients with advanced NETs received treatment. Three patients (10%) had low, 13 (45%) intermediate and 13 (45%) high grade tumors; lung was the most common primary site (39%). The objective response rate was 24% with a CBR of 72%; 43% of patients with pancreatic neuroendocrine neoplasms (NENs) and 33% of patients with atypical bronchial carcinoid achieved an objective response. The median progression free survival was 4.8 months (95% CI: 2.7, 10.5) and overall survival 14.8 months (95% CI: 4.1,21.3). Immune-related toxicity was reported in 66% of patients with 34% experiencing grade 3/4 events. Conclusions: Combination immunotherapy with ipilimumab and nivolumab demonstrated significant clinical activity in subgroups of patients with advanced NETs including patients with atypical bronchial carcinoid and high grade pancreatic NENs.