ArticlePDF Available

Cell type-dependent differential activation of ERK by oncogenic KRAS in colon cancer and intestinal epithelium

Authors:

Abstract and Figures

Oncogenic mutations in KRAS or BRAF are frequent in colorectal cancer and activate the ERK kinase. Here, we find graded ERK phosphorylation correlating with cell differentiation in patient-derived colorectal cancer organoids with and without KRAS mutations. Using reporters, single cell transcriptomics and mass cytometry, we observe cell type-specific phosphorylation of ERK in response to transgenic KRASG12V in mouse intestinal organoids, while transgenic BRAFV600E activates ERK in all cells. Quantitative network modelling from perturbation data reveals that activation of ERK is shaped by cell type-specific MEK to ERK feed forward and negative feedback signalling. We identify dual-specificity phosphatases as candidate modulators of ERK in the intestine. Furthermore, we find that oncogenic KRAS, together with β-Catenin, favours expansion of crypt cells with high ERK activity. Our experiments highlight key differences between oncogenic BRAF and KRAS in colorectal cancer and find unexpected heterogeneity in a signalling pathway with fundamental relevance for cancer therapy.
Network quantification identifies cell type-specific differences in KRAS to ERK signalling. a Protein phosphorylation and abundance CyTOF data by treatment and cell clusters, as in Fig. 6c. Log2 fold changes to average untreated FLUC-induced control line are given. b Signalling network structure used for modelling. The network was re-parametrised from a starting network, using the experimental data to remove and add connections, denoted by grey and blue arrows, respectively. c Signalling quantification of identifiable network links using Modular Response Analysis. Numbers 1-5 in panels (b) and (c) show network connections with significant differences between clusters in order of detection. Red circles mark MEK-ERK and ERK-MEK connections identified as having different strengths in clusters with high vs. low ERK phosphorylation after KRAS G12V induction. d MRA modelling of differences between KRAS G12V -induced and FLUC control cells within each cluster. K and F mark KRAS G12V and FLUC control cluster pairs, respectively. Cluster pairs exhibiting KRAS G12V -specific differences are shown in red and blue, indicating regulation strengths. e Colour-coded gene expression data from cells sorted by high and low FIRE activity, as indicated. Upper panel shows marker genes (Mki67, encoding Ki67, for proliferative cells, Cd44 and Ephb2 for crypt cells and Kras), lower panel shows 20 significantly regulated genes between the conditions. In total, 269 genes encoding MAPK network components in KEGG were tested. Asterisks indicate dual-specificity phosphatases
… 
Transgenic BRAFV600E, but not KRASG12V disrupts organoids due to high ERK activity. a Simplified representations of transgenes and the RAS-ERK and Wnt/β-catenin pathways, indicating relative positions of the KRAS and BRAF proto-oncogenes. b Organoid survival 4 days after induction of oncogenic KRASG12V or BRAFV600E. Organoids are counted immediately after passaging, and fractions of surviving organoids were calculated at day 4. Control organoids comprise of mixed non-induced cultures of KRASG12V and BRAFV600E lines. c Electron microscopy reveals loss of epithelial integrity after BRAFV600E induction. Images of the intestinal organoid epithelium, 24 h after induction of control FLUC, KRASG12V or BRAFV600E transgenes. Detailed views (right) represent a zoom into areas marked by red boxes in the overviews (left). Detailed views show apical surfaces of adjacent enterocytes with brush border. Red arrows mark desmosomes. Intercellular vacuoles, most visible in the KRASG12V model (marked by *) are likely fixation-induced artefacts, see ref. ²⁷. Scale bars are 10 µm in the overview panels and 500 nm in the detailed view panels. d Quantification of ERK phosphorylation in organoids, 24 h after induction of control, BRAF or KRAS transgenes, using a capillary protein analysis. e Quantification of organoid survival, 4 days after inhibition of EGFR, MEK, ERK and/or induction of BRAFV600E, as in panel (b). Error bars in panels (b), (d) and (e) denote standard deviations. Data shown in panels (b), (d), and (e) are available as a Source Data file
… 
This content is subject to copyright. Terms and conditions apply.
ARTICLE
Cell type-dependent differential activation of ERK
by oncogenic KRAS in colon cancer and intestinal
epithelium
Raphael Brandt 1,9, Thomas Sell 1,2,9, Mareen Lüthen 1,3, Florian Uhlitz1,2,3, Bertram Klinger1,2,
Pamela Riemer 1, Claudia Giesecke-Thiel 4,5, Silvia Schulze1, Ismail Amr El-Shimy 1,2, Desiree Kunkel6,
Beatrix Fauler5, Thorsten Mielke 5, Norbert Mages5, Bernhard G. Herrmann 5,7, Christine Sers 1,3,8,
Nils Blüthgen 1,2,3,8,10 & Markus Morkel 1,3,8,10
Oncogenic mutations in KRAS or BRAF are frequent in colorectal cancer and activate the ERK
kinase. Here, we nd graded ERK phosphorylation correlating with cell differentiation in
patient-derived colorectal cancer organoids with and without KRAS mutations. Using
reporters, single cell transcriptomics and mass cytometry, we observe cell type-specic
phosphorylation of ERK in response to transgenic KRASG12V in mouse intestinal organoids,
while transgenic BRAFV600E activates ERK in all cells. Quantitative network modelling from
perturbation data reveals that activation of ERK is shaped by cell type-specic MEK to ERK
feed forward and negative feedback signalling. We identify dual-specicity phosphatases as
candidate modulators of ERK in the intestine. Furthermore, we nd that oncogenic KRAS,
together with β-Catenin, favours expansion of crypt cells with high ERK activity. Our
experiments highlight key differences between oncogenic BRAF and KRAS in colorectal
cancer and nd unexpected heterogeneity in a signalling pathway with fundamental relevance
for cancer therapy.
https://doi.org/10.1038/s41467-019-10954-y OPEN
1Institute of Pathology, Charité Universitätsmedizin Berlin, Charitéplatz 1, 10117 Berlin, Germany. 2IRI Life Sciences, Humboldt University Berlin, Philippstrasse 13,
10115 Berlin, Germany. 3German Cancer Consortium (DKTK), German Cancer Research Center (DKFZ), 69120 Heidelberg, Germany. 4Department of Cell
Biology, German Rheumatism Research Center, Leibniz Institute, Berlin, Germany. 5Max Planck Institute for Molecular Genetics, Ihnestr. 73, 14195 Berlin,
Germany. 6Berlin-Brandenburg Center for Regenerative Therapies (BCRT), Charité - Universitätsmedizin Berlin, Campus Virchow-Klinikum, Augustenburger Platz
1, 13353 Berlin, Germany. 7Institute for Medical Genetics, Charité Universitätsmedizin Berlin, Hindenburgdamm 30, 12203 Berlin, Germany. 8Berlin Institute of
Health (BIH), Anna-Louise-Karsch-Str. 2, 10178 Berlin, Germany.
9
These authors contributed equally: Raphael Brandt, Thomas Sell.
10
These authors jointly
supervised this work: Nils Blüthgen, Markus Morkel. Correspondence and requests for materials should be addressed to N.B. (email: nils.bluethgen@charite.de)
or to M.M. (email: markus.morkel@charite.de)
NATURE COMMUNICATIONS | (2019) 10:2919 | https://doi.org/10.1038/s41467-019-10954-y | www.nature.com/naturecommunications 1
1234567890():,;
Content courtesy of Springer Nature, terms of use apply. Rights reserved
Multiple signalling pathways, including the mitogen-
activated protein kinase (MAPK) and Wnt/β-catenin
cascades, form a network controlling cellular turnover
in the intestinal epithelium1. Collectively, activities within the
signalling network control stem cell maintenance, cell prolifera-
tion, differentiation into absorptive enterocyte and secretory cells,
and apoptosis. Wnt/β-catenin and MAPK activities are regiona-
lised within the folded single-layered intestinal epithelium. Both
are high in crypts harbouring stem cells and low in differentiated
cells that have migrated away from the crypt base. Oncogenic
mutations activating β-catenin and MAPK perturb intestinal
homeostasis and thereby drive colorectal cancer (CRC) initiation
and progression.
MAPK modules transduce signals downstream of receptor
tyrosine kinases and RAS family GTPases. The consecutive RAF,
MEK and ERK kinases represent a MAPK module frequently
activated in cancer. ERK can phosphorylate and activate a series
of transcription factors orchestrating a complex cellular response
that often is pro-proliferative2. In the normal intestine, EGFR to
ERK signalling is initiated by ligands from the crypt micro-
environment, which are secreted by e.g. epithelial Paneth cells of
the small intestine, Reg4+secretory niche cells of the large
intestine, or adjacent broblasts3,4. In CRC, ERK activity is
supposedly more cell-autonomous due to oncogenic mutations
activating KRAS, NRAS or BRAF (found in 45, 5 and 10% of
CRCs, respectively)5,6, or by de novo expression of EGFR ligands
such as amphiregulin7. Signal transduction to ERK is a main
determinant of cancer development and therapy response5,6,8.
Recent studies suggest that the relationship between ERK-
activating mutations, ERK activity and phenotypic outcome in
CRC is complex. Firstly, mutations in KRAS and BRAF are
associated with distinct CRC development routes: KRAS, but not
BRAF, mutations frequently occur as secondary events after
mutations activating Wnt/β-catenin in the conventional CRC
progression sequence9,10. Conversely, BRAF, and less frequently
KRAS, mutations precede activation of Wnt/β-catenin in the
alternative serrated progression route11,12. The observed dis-
equilibrium between KRAS and BRAF mutations in the con-
ventional vs. serrated pathways of CRC evolution suggest the
existence of functional differences, resulting in distinctive effects
on clinical course and treatment efcacy13. Secondly, ERK activity
appears to be heterogeneous in genetically identical CRC cells.
Cells at the invasive front frequently exhibited higher ERK
phosphorylation levels compared with cells in central areas of the
same cancer, and CRCs with activating KRAS mutations also
showed heterogeneous ERK activity14. Previous studies already
showed heterogeneous Wnt/β-catenin activity in cancer speci-
mens, suggesting a more general paradigm of graded pathway
activities in CRC15. Furthermore, CRC cells have been shown to
exhibit functional differences within a cancer, as only few CRC
cells, so-called cancer stem cells, could initiate new tumours in
xenografts1619. Gradients of surface markers, such as EphB2,
were found to distinguish CRC stem cells with high malignant
potential20. CRC subtypes can share similarities with cell types of
the normal crypt, such as stem cells, enterocytes or secretory cells
in bulk cell analysis21. Finally, because of variable signal trans-
duction and differentiation states, genetically identical CRC
clones exhibit variable proliferative potential and therapeutic
response22.
Experimental techniques with cellular resolution, ranging from
uorescent reporters23 to single-cell transcriptome analyses24,25
and mass cytometry26 hold the promise to disentangle the rela-
tionship between oncogenes, cell differentiation states and cell-
signal transduction while taking into account cellular hetero-
geneity. Here, we ask whether oncogenic forms of KRAS or BRAF
show cell-to-cell heterogeneity in their proclivity to activate ERK.
For this, we use patient-derived and mouse transgenic organoid
cultures that maintain the cell hierarchy of tissue in vitro27.We
assess signalling network states with cellular resolution by mass
cytometry and use BRAFV600E and KRASG12V transgenes to
assess immediate impact of the oncogenes on cell-signal trans-
duction, gene expression programmes and phenotypic outcome.
We discover strong functional differences between the BRAF and
KRAS oncogenes and nd that signal transduction from KRAS to
ERK is shaped by different strengths of feed forward and negative
feedback in a cell type-specic manner.
Results
Graded ERK activity in KRAS-mutant CRC organoids.To
investigate whether oncogenic KRAS enforces constitutive activity
of MEK and ERK kinases, we examined patient-derived three-
dimensional CRC organoid cultures by immunohistochemistry.
We found heterogeneous phosphorylation of both, MEK and
ERK, in organoids with no mutations in the EGFR-RAS-ERK
cascade (line OT326), as well as in KRAS-mutant organoids (line
OT227, carrying a KRASG13D mutation) (Fig. 1a).
We next used mass cytometry to analyse cell differentiation
markers and MEK and ERK phosphorylation side-by side in the
patient-derived organoids (Fig. 1b). We selected the two organoid
lines used above, as well as line OT302, harbouring a KRASG12D
mutation. We found that cells of all three organoid lines formed
gradients with respect to levels of EphB2, a known marker of
CRC hierarchies linked to metastasis and therapy response20.In
two of the three lines (OT326 and OT302), a substantial
proportion of EphB2-low cells was marked by cleaved Caspase
3, suggesting apoptotic removal of cells at the end of their life
span. Intriguingly, all three lines displayed gradients of
phosphorylated MEK and ERK that were largely congruent with
EphB2. These results indicate that patient-derived CRC organoids
contain phosphorylation gradients of MEK and ERK kinases
along an axis dened by cell differentiation. The observed
gradient formed regardless of oncogenic activation of the
upstream KRAS GTPase, and in the absence of tumour stroma
that is not present in the organoids.
BRAFV600E, but not KRASG12V, induces strong ERK activity.
As MEK and ERK activities were graded along a differentiation
axis in patient-derived CRC organoids irrespective of mutational
status of KRAS, we asked whether oncoproteins activating the
MEK-ERK signalling axis exert their activities in a cell type-
specic manner. To study this question, we employed intestinal
organoids of transgenic mice carrying doxycycline-inducible
single copy constructs encoding tdTomato linked to KRASG12V,
BRAFV600E,orrey luciferase (FLUC) as a control in the Gt
(ROSA26)Sor locus (Fig. 2a)28,29.
We initiated organoid cultures by embedding intestinal crypts
from FLUC-, KRASG12V-, and BRAFV600E-inducible mice into
extracellular matrix, as described before27. When we induced
oncoprotein production by adding doxycycline to the culture
media, BRAFV600E led to irreversible disintegration of organoids
within 12 days, whereas transgenic KRASG12V or the FLUC
control protein were well tolerated, even after several passages
(Fig. 2b). To examine whether the BRAF oncogene has
detrimental effects on the epithelium beyond the previously
reported loss of stem cells29,30, we examined histology of the
induced organoids at ultrastructural level using transmission
electron microscopy (Fig. 2c). We found that control and
KRASG12V-induced organoids showed the expected tissue
structure, that is, a single-layered polarised epithelium with
continuous apical and basal surfaces as well as a brush border at
the apical side. Desmosomes, providing lateral cell adhesion, were
ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-10954-y
2NATURE COMMUNICATIONS | (2019) 10:2919 | https://doi.org/10.1038/s41467-019-10954-y | www.nature.com/naturecommunications
Content courtesy of Springer Nature, terms of use apply. Rights reserved
p-MEK p-ERK
OT326 KRAS wt
aHE
b
High
Low
CyTOF signal
2
1
0
–1
–2
–3
–2 0
EphB2
OT326 (KRAS wildtype)OT227 (KRASG13D)
OT227 KRASG13D
OT302 (KRASG12D)
cl-Casp3 p-MEK p-ERK
2
–2 –402–2402–2402–2402
–2–4 0 2 –2–4 0 2 –2–4
PC1
PC2
02 –2–4 0 2
–2 0 2 –2 0 2 –2 0 2
–2
–1
0
1
2
–2
–3
–1
0
1
2
1
0
–1
–2
–3
–2
–1
0
1
2
–2
–3
–1
0
1
2
1
0
–1
–2
–3
–2
–1
0
1
2
–2
–3
–1
0
1
2
1
0
–1
–2
–3
–2
–1
0
1
2
–2
–3
–1
0
1
Fig. 1 Graded MEK and ERK phosphorylation in patient-derived organoids. aHaematoxilin-eosin (HE) staining and phospho-MEK and phospho-ERK
immunohistochemistry of two PD3D lines OT326 and OT227 that are KRAS-wild-type and KRAS-mutant, respectively. Scale bars denote 100 µm for HE
and immunohistochemistry. bCyTOF analysis of PD3Ds. Principal component analyses, colour-coded for EphB2, cleaved Caspase, phospho-MEK and
phospho-ERK are shown. Red, yellow and blue colours of the scale represent high, intermediate and low signals, respectively. CyTOF data is available as a
Source Data le
NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-10954-y ARTICLE
NATURE COMMUNICATIONS | (2019) 10:2919 | https://doi.org/10.1038/s41467-019-10954-y | www.nature.com/naturecommunications 3
Content courtesy of Springer Nature, terms of use apply. Rights reserved
Organoid survival (%)
0
20
40
60
80
100
Organoid survival (%)
+EGFRi
+MEKi
+ERKi
Control
Control
+EGFRi
+MEKi
+ERKi
0
25
50
75
100
c
FLUC
BRAFV600E KRASG12V
Overview Detail
*
*
aloxP lox5171
WNT
Fz
β-Cat.
APC
ERK
MEK
EGFR
EGF
KRAS
PI3K
AKT
BRAF
loxP lox5171
rtTA pA
ROSA26A
TRE tdT 2A KRASG12V
TRE tdT 2A FLUC
TRE tdT 2A BRAFV600E
PGKpA
PGKpA
PGKpA
ins Hygro pA neo pA ins
pA
500 nm
10 μm
d
KRASG12V
BRAFV600E
MEKi
+
+
+
+
+
+
+
+BRAFV600E
b
KRASG12V
BRAFV600E
+
+
e
0
5
10
15
p-ERK (relative units)
Fig. 2 Transgenic BRAFV600E, but not KRASG12V disrupts organoids due to high ERK activity. aSimplied representations of transgenes and the RAS-ERK
and Wnt/β-catenin pathways, indicating relative positions of the KRAS and BRAF proto-oncogenes. bOrganoid survival 4 days after induction of oncogenic
KRASG12V or BRAFV600E. Organoids are counted immediately after passaging, and fractions of surviving organoids were calculated at day 4. Control
organoids comprise of mixed non-induced cultures of KRASG12V and BRAFV600E lines. cElectron microscopy reveals loss of epithelial integrity after
BRAFV600E induction. Images of the intestinal organoid epithelium, 24 h after induction of control FLUC, KRASG12V or BRAFV600E transgenes. Detailed
views (right) represent a zoom into areas marked by red boxes in the overviews (left). Detailed views show apical surfaces of adjacent enterocytes with
brush border. Red arrows mark desmosomes. Intercellular vacuoles, most visible in the KRASG12V model (marked by *) are likely xation-induced artefacts,
see ref. 27. Scale bars are 10 µm in the overview panels and 500 nm in the detailed view panels. dQuantication of ERK phosphorylation in organoids, 24 h
after induction of control, BRAF or KRAS transgenes, using a capillary protein analysis. eQuantication of organoid survival, 4 days after inhibition of EGFR,
MEK, ERK and/or induction of BRAFV600E, as in panel (b). Error bars in panels (b), (d) and (e) denote standard deviations. Data shown in panels (b), (d),
and (e) are available as a Source Data le
ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-10954-y
4NATURE COMMUNICATIONS | (2019) 10:2919 | https://doi.org/10.1038/s41467-019-10954-y | www.nature.com/naturecommunications
Content courtesy of Springer Nature, terms of use apply. Rights reserved
clearly visible. In contrast, BRAFV600E-induced organoids dis-
played a continuous basal surface, whereas the apical side was
grossly distorted, although it contained a brush border as
evidence of polarisation. Nuclei were pleomorphic and no longer
lined up basally but scattered at different positions. Cells were still
attached to each other by desmosome bridges, indicating that the
ongoing epithelial disorganisation was taking place in the
presence of lateral cell adhesion.
To ascertain whether the epithelial disorganisation provoked
by BRAFV600E was correlated with MAPK activity, we measured
phosphorylation of ERK. We found that induction of BRAFV600E,
but not KRASG12V, resulted in high phospho-ERK levels in
intestinal organoids, as determined by capillary protein analysis
(Fig. 2d). BRAFV600E-induced organoid disintegration could be
counteracted by inhibition of the BRAF-downstream MEK and
ERK kinases using AZD6244/Selumetinib31 and BVD-523/
Ulixertinib32, respectively, but not by inhibition of the upstream
EGFR tyrosine kinase receptor family using AZD8931/Sapitinib33
(Fig. 2e), showing that the phenotype is due to excessive MEK-
ERK activity. Indeed, only 24 h after BRAFV600E induction,
almost all direct ERK target genes34 were activated. In contrast,
conditional expression of KRASG12V induced RAS activity as
measured by a RAS-GTP pull-down assay but had no obvious
effect on bulk organoid transcription (Supplementary Fig. 1).
BRAFV600E disrupts intestinal differentiation trajectories.To
uncover potential cellular heterogeneity in response to the
oncogenes, we performed single-cell transcriptome analyses. We
induced FLUC control, BRAFV600E- and KRASG12V-transgenic
organoids for 24 h, prepared single-cell suspensions, and stained
them with a uorescent antibody against the crypt cell marker
CD4435, and with a uorescent dye to eliminate dead cells. Using
single-cell sorting on the transgene-expressing organoids we next
acquired samples of CD44-high crypt and CD44-low villus cells
(see Supplementary Fig. 2 for FACS gating strategy), which were
subjected to single-cell RNA sequencing. In total, we obtained
transcriptomes of 167 cells with >1000 detected genes each, that
were used for further analysis. Single-cell transcriptomes could be
assigned to six interconnected clusters with help of k-means
clustering and were visualised in a t-SNE-based representation
(Fig. 3a). Mapping of signature genes for intestinal stem cells
(ISCs), proliferative TA cells, differentiated enterocytes20 and
secretory Paneth cells4, and the CD44 status as inferred from ow
cytometry (Fig. 3b, c) conrmed the calculated differentiation
trajectories (grey overlay in Fig. 3a): undifferentiated CD44-high
ISC and TA cell signature genes were enriched in clusters 1 and 2
while Paneth cell marker genes were highest in cluster 2, indi-
cating the differentiation route for secretory crypt cells; expres-
sion of enterocyte signature genes increased gradually in clusters
35, marking the CD44-low absorptive lineage.
We next considered the distribution of cells expressing specic
transgenes (Fig. 3b): FLUC control and KRASG12V-expressing
cells intermingled throughout the clusters 15 of the normal cell
differentiation trajectories. BRAFV600E-expressing cells were in
contrast depleted from the central clusters 24, and instead
formed outsider cluster 6, composed entirely of BRAF-induced
cells. Notably, cells in cluster 6 uniformly expressed high levels of
ERK target genes, regardless of whether they were sorted as
CD44-high or CD44-low. Furthermore, cluster 6 cells also highly
expressed Anxa10, which has been identied as a marker for
BRAF-positive serrated adenoma36 (Supplementary Fig. 3). The
single-cell analysis thus showed that BRAFV600E imposed a
specic gene expression programme onto intestinal cells,
independent of their prior differentiation state. Transcriptomes
of KRASG12V-induced cells, in contrast, were undistinguishable
from FLUC control cells; however, we observed that KRASG12V-
induced cells showed a shift towards CD44-high undifferentiated
cell types compared with FLUC controls (see Supplementary
Fig. 2b).
KRASG12V-to-ERK signalling is cell type-specic. To visualise
ERK activity with single-cell resolution in organoids, we
employed the Fra-1-based integrative reporter of ERK (FIRE) that
translates ERK kinase activity into stability of a nuclear yellow-
green venus uorescent protein (Fig. 4a)23. FIRE uorescence in
organoids cultured in normal growth medium containing EGF
was strong in crypts, whereas differentiated villus tissue was
largely FIRE negative (Fig. 4b). In EGF-free medium, ERK activity
in the crypt base persisted, likely due to autocrine and paracrine
signals from EGF-producing Paneth cells4.
We next conditionally expressed FLUC control, KRASG12V-, or
BRAFV600E-encoding transgenes in FIRE-transfected organoids
(Fig. 4c). Transgene induction was often variable, as inferred by
tdTomato uorescence, allowing to compare individual
tdTomato-positive cells with transgene-negative neighbouring
tissue. tdTomato-FLUC control transgene expression had no
inuence on FIRE activity. In contrast, expression of KRASG12V
resulted in increased FIRE signals in crypt cells, which
consistently displayed stronger reporter activity compared with
adjacent KRASG12V-negative cells. Surprisingly, a large majority
of villus cells remained FIRE negative, despite strong tdTomato-
KRASG12V positivity. We conrmed the differential signal
transduction from KRASG12V to ERK using phospho-ERK
immunohistochemistry (Fig. 4d). In line with our FIRE reporter
data, p-ERK-positive cells were largely absent in central
differentiated (Ki67-negative) villus areas of organoids, despite
strong tdTomato-KRASG12V staining. Taken together, our results
show that ERK activity in differentiated villus epithelial cells can
neither be increased by EGF in the medium nor by induction of
oncogenic KRASG12V. However, when we induced BRAFV600E,
we found widespread and strong FIRE signals across the complete
organoid (Fig. 4c). This suggests a strict and cell type-specic
control of signal transduction by oncogenic KRAS, but not BRAF,
in intestinal epithelial cells.
Since FIRE uorescence could distinguish cells responsive to
KRASG12V, we next used the reporter to assist selection of cells
for single-cell RNA sequencing. Our aim was to dene cell types
with high ERK activity, either in response to KRASG12V or as part
of the normal cell hierarchy. For this, we induced organoids with
the integrated ERK reporter for KRASG12V or FLUC, prepared
single-cell suspensions and sorted cells by FACS into 96-well
plates for transcriptome analysis (see Supplementary Fig. 4 for
FACS gating strategy). We focussed on single cells with high
transgene (tdTomato) signal that were either positive or negative
for FIRE (venus) uorescence (Fig. 5a). In total, we obtained 197
single-cell transcriptomes. K-means clustering into eight groups
and t-SNE-based visualisation revealed the cell type distribution
(Fig. 5b, c). Cluster 1 was enriched for undifferentiated crypt (ISC
and TA) marker genes, whereas clusters 24 were dened by
Paneth cell signature genes (Fig. 5d; Supplementary Fig. 5).
Cluster 2 was enriched for Paneth cell markers such as Lyz1,
encoding Lysozyme37, and several genes encoding Defensins,
while other cluster-dening genes such as Mptx1 and Agr2 in
cluster 4 hint at a high degree of Paneth cell heterogeneity.
Clusters 58 formed a differentiation trajectory for absorptive
cells, with Ifabp1 as the top dening gene for clusters 57
(Supplementary Fig. 5).
Using this information, we assessed the distribution of
transcriptomes derived from KRASG12V-induced FIRE-high cells
(Fig. 5c, d). These were conned to distinct aggregates
NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-10954-y ARTICLE
NATURE COMMUNICATIONS | (2019) 10:2919 | https://doi.org/10.1038/s41467-019-10954-y | www.nature.com/naturecommunications 5
Content courtesy of Springer Nature, terms of use apply. Rights reserved
encompassing the undifferentiated cell zone of cluster 1, as well as
transcriptomes inhabiting the outer right rim of the t-SNE
representation that we above assigned to be derived from late-
stage enterocytes and Paneth cells. Immunouorescence micro-
scopy using the Paneth cell marker Lysozyme conrmed high
FIRE activity in this cell type after KRASG12V induction
(Supplementary Fig. 6). In contrast, a central area of the t-SNE
plot encompassing the largest clusters 5 and 6 of bulk enterocytes
was almost devoid of KRASG12-producing FIRE-high cells but
harboured many KRASG12V/FIRE-low cells, conrming that
enterocytes generally cannot activate ERK, even when expressing
oncogenic KRASG12V; however, a specic subset of presumably
late-stage enterocytes displayed high ERK activity.
KRASG12V interacts with GSK3βinhibition. In order to
understand how β-catenin- and MAPK-networks interact in
controlling cell differentiation and ERK phosphorylation in
intestinal epithelium, we performed a network perturbation study
using kinase inhibitors, followed by mass cytometry in
KRASG12V-inducible and FLUC control organoids. For this, we
induced the transgenes in 3-day-old organoids, subsequently
treated them with an GSK3βinhibitor (CHIR99021) for 24 h to
stabilise β-catenin38, and used MEK and p38 inhibitors
(AZD6244 and LY2228820/Ralimetinib39, respectively) for 3 h to
inhibit key kinases as part of the intestinal cell signalling network
(Fig. 6a). We measured a total of 160,000 transgene-positive cells,
representing 12 multiplexed samples.
To discern the immediate effects of KRASG12V and stabilised
β-catenin on intestinal cell hierarchies, we assessed the distribu-
tion of cell type markers (Fig. 6b). As a positive control for the
effect of GSK3βinhibition on β-catenin activity, treatment with
CHIR99021 increased levels of the β-catenin target protein
Axin240. We observed that both induction of KRASG12V and
treatment with the GSK3βinhibitor, increased median levels of
crypt cell markers EphB2, CD44 and CD24, and for all three
proteins, KRASG12V-induced cells that were additionally treated
with the GSK3βinhibitor had the highest levels. These results are
in line with prior evidence that oncogenic KRAS and β-catenin
activities can inhibit or reverse intestinal cell differentiation and
provide clonal benets linked to crypt cell fate41,42.
We used k-means clustering to allocate KRASG12V-induced
cells to six clusters dened by levels of cell type and surface
markers CD24, CD44, EphB2, Krt20 and apoptosis marker
cleaved Caspase 3 (Fig. 6c). p-ERK-positive cells were enriched in
clusters 5 and 6, while cleaved Caspase 3-positive cells were found
in Cluster 5 (Fig. 6d, e). Based on gradual loss of the crypt cell
ab
c
Dimension 2
Dimension 2
Dimension 2
Dimension 1
Dimension 1
Dimension 1
Cluster
CD44
ISC markers TA cell markers Enterocyte markers
Paneth cell markers ERK targets Wnt targets
Signature
genes/cell
2
0
0
–2
–4
2
200
150
50
100
Signature
genes/cell
10
0
5
0
–2
–4
Dimension 2
2
0
–2
–4
2
0
–2
–4
–4 4
0
–4 4
Dimension 2
Dimension 1
Signature
genes/cell
2
60
40
0
20
0
–2
–4
0
–4 4
Dimension 2
Dimension 1
Signature
genes/cell
2
30
0
10
20
0
–2
–4
0
–4 4
Dimension 1
0–4 4
Signature
genes/cell
50
40
10
30
20
Dimension 2
2
0
–2
–4
Dimension 1
0
–4 4
Signature
genes/cell
10
0
5
Dimension 2
2
0
–2
–4
Dimension 1
0–4 4
0–4 4
6
2
135
Transgene
BRAFV600E
KRASG12V
Low
High
FLUC
1
2
3
4
5
6
4
Fig. 3 Differential effects of BRAFV600E or KRASG12V on gene expression and intestinal cell hierarchies. All panels: t-SNE visualisations and clustering of
organoid single-cell transcriptomes clustered with k-means, 24 h after induction of FLUC control, BRAFV600E or KRASG12V transgenes. aColour code for six
k-means clusters, and inferred differentiation trajectories starting at cluster 1 shown as grey overlay. bColour code for transgene and CD44 positivity, as
inferred from ow cytometry. CD44 positivity was used to direct cell selection, and thus relative fractions of CD44-high and -low cells are not
representative. For CD44 status of the cell populations, see Supplementary Fig. 2. cMapping of cell- and pathway-specic differentiation signatures.
Numbers of signature genes detected are given per single-cell transcriptome
ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-10954-y
6NATURE COMMUNICATIONS | (2019) 10:2919 | https://doi.org/10.1038/s41467-019-10954-y | www.nature.com/naturecommunications
Content courtesy of Springer Nature, terms of use apply. Rights reserved
markers EphB2, CD44 and CD24, we concluded that clusters 14
represent a crypt-to-villus gradient, interconnecting with cluster 5
at the end of the differentiation trajectory (Fig. 6f). Clusters 3 and
4 had the lowest phospho-MEK and phospho-ERK levels, and
also contained lowest levels of the Wnt/β-catenin target Axin2, in
agreement with differentiated villus cell status. High levels of
CD24 marked p-ERK-positive cells in cluster 6 as presumptive
Paneth cells4. Interestingly, we observed that fractions of cells
allocated to the clusters were modied by both KRASG12V
induction and GSK3βinhibition (Fig. 6g): both treatments
increased the percentage of cells in cluster 1, representing the
presumptive undifferentiated crypt cells, as inferred from high
levels of markers such as EphB2 and CD441,20. The combination
of KRASG12V and GSK3βinhibition had the greatest effect and
furthermore strongly decreased the fraction of cells in the
apoptotic cell cluster 5. These data suggest that oncogenic KRAS
and β-catenin stabilisation can both favour crypt cell fate over
differentiation, at least on the level of cell type marker expression.
Cell type-specic differences in ERK feedback regulation.To
quantitatively dissect differences in signalling in the cell types, we
assigned the cells measured under different perturbed conditions
to the six clusters dened above according to their shortest
Euclidian distance (Fig. 6c). For each condition and each cluster,
we calculated average phosphorylation levels of MEK, ERK,
BRAF
MEK
KRAS
EGFR
EGF
Fra1-PESTNLS-Venus
ERK
P
tdT-FLUC::FIRE
tdTomato FIRE Overlay
tdT-KRASG12V::FIRE
v
c
c
tdT-KRASG12V
tdTomato
Ki67 p-ERK
d
a
b
c
tdT-BRAFV600E::FIRE
v
c
*
*
*
v
c
c
Non-induced
Red fluorescence
Green fluorescence
+EGF
–EGF
*
v
c
c
c
v
v
c
c
v
c
c
c
v
0100
μm
Fig. 4 Visualisation of ERK activity by FIRE reveals KRASG12V-responsive cells. aSchematic representation of signalling pathway and reporter. bFIRE
activity in wild-type intestinal organoids, in the presence and absence of EGF in the culture medium, as indicated. Asterisk marks isolated FIRE-high villus
cell. cFluorescence microscopy images showing transgene expression (red), FIRE activity (green), and overlays in intestinal organoids, taken 2 days (FLUC,
KRAS) or 1 day (BRAF) after transgene induction. Arrow heads mark KRASG12V/FIRE-high crypt cells, asterisk marks FIRE-high villus cell, respectively.
dImmunohistochemistry of tdTomato, Ki67 and p-ERK in intestinal organoids, as indicated. In panels (c) and (d), c and v demarcate crypt and villus areas,
respectively. Scale bars are 100 µm in all panels
NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-10954-y ARTICLE
NATURE COMMUNICATIONS | (2019) 10:2919 | https://doi.org/10.1038/s41467-019-10954-y | www.nature.com/naturecommunications 7
Content courtesy of Springer Nature, terms of use apply. Rights reserved
4EBP1, p38, ribosomal protein S6 and total protein levels of IκBα
and Axin2, forming a Wnt-, MAPK-, NFκB- and mTOR network
(Fig. 7a, b). Cellular signalling states varied strongly between the
cell clusters 1 and 6. However, comparison between FLUC con-
trol and KRASG12V-expressing organoid cells showed that sig-
nalling within each cell type was very comparable, except for an
increase in phosphorylation levels of MEK and ERK in clusters 5
and 6 in cells expressing KRASG12V. This was in contrast to
results from similar CyTOF experiments performed with the
BRAFV600E-inducible organoid line, which resulted in high MEK
and ERK phosphorylation across all cell type clusters (Supple-
mentary Fig. 7).
We next employed network modelling using Modular
Response Analysis (MRA)43,44. This approach allows to quantify
signal transmission from perturbation-response data and, by
using likelihood ratio tests, to pinpoint which signalling routes
are different between the clusters. The method requires fold
changes in network node activity after perturbation (calculated
from data shown in Fig. 7a) and a literature-derived network (as
shown in Fig. 7b) as input and calculates so-called response
coefcients that reect the strength of signalling interactions.
When we applied this modelling framework to the signalling
perturbation data of the six clusters for the KRAS-mutant
organoid data, we observed that only 5 out of 11 signalling routes
signicantly differed between clusters with two of them
constituting the feed-forward and feedback signalling paths
between MEK and ERK (Fig. 7b, c). In clusters 5 and 6, which
were the clusters that showed elevated phosphorylation levels of
MEK and ERK levels after KRASG12V induction, the model
unveiled that these cells enable RAS-ERK signalling by two
mechanisms: signal transmission from MEK to ERK was
enhanced, while ERK-dependent negative feedback inhibition of
RAF and upstream components was attenuated when compared
with the other clusters. In contrast, clusters 3 and 4, which
exhibited the lowest phospho-ERK levels that were also not
increased by KRASG12V, MEK-to-ERK feed forward signal
transduction was low, and this coincided with strong ERK-
dependent feedback inhibition. Cluster 1, which based on surface
marker expression represents undifferentiated crypt cells, had
strong feedback inhibition according to the model. This cluster
had intermediate phospho-ERK levels as measured by CyTOF
(Fig. 6f) and was FIRE positive (Fig. 4c). However, phospho-ERK
levels were unresponsive to KRASG12V in this cluster and our
model predicts that this could be due to strong ERK feedback.
To model differences between KRASG12V-induced and control
cells within each cluster, we employed comparative MRA
modelling that resulted in cluster-specic signalling models that
consider the inuence of KRASG12V on the signalling network per
cluster (Fig. 7d). We could discern only few differences. Most
importantly, KRASG12V enhanced signalling from MEK to ERK
in clusters 5 and 6. Furthermore, we observed that KRASG12V
modulated the effect of Wnt/Axin2 signalling on mTOR in
cluster 2.
As our functional studies and the modelling showed that RAS-
ERK signal transduction can be differently wired between cell
types, we tested whether this was also true for CRC cell lines.
Indeed, when we compared ERK phosphorylation in response to
transfected KRASG12V in SW48 and Caco2 CRC cells (that have
ab
dCluster
ERK targets
Enterocyte markers
Paneth cell markers
TA cell markers
ISC markers
Wnt targets
Cell features
FIRE (linear scale)
FIRE positive
FIRE negative
TdTomato transgene
(log scale)
c
Dimension 2
Dimension 1
Dimension 2
Dimension 1
Cluster
1
2
3
4
5
6
7
8
Transgene
KRASG12V
FLUC
FIRE
Pos
Neg
Z-value
4
2
0
–2
123 5 6784
FLUC
FIRE neg FLUC
FIRE pos
KRASG12V
FIRE neg KRASG12V
FIRE pos
102
103
104
105
12345678
4
0
–4
–4 0 4 –4 0 4
–8
4
0
–4
–8
1
5
3
8
7
6
2
4
Fig. 5 Single-cell RNA sequencing reveals KRASG12V-responsive and -unresponsive organoid cells. aFluorescence-activated cell sort gates for FIRE-
negative and -positive cells. bt-SNE visualisation colour-coded for eight clusters identied with k-means clustering. Differentiation trajectories starting at
cluster 1 are shown as grey overlay. ct-SNE visualisation displaying colour codes for transgene and FIRE positivity. Filled upward-pointing triangles: FIRE-
high; outlined downward-pointing triangles: FIRE-low. Red: KRASG12V; grey: FLUC. dHeatmap of z-transformed signature scores per cell for cluster cell type
identication. Signature scores correspond to the number of expressed signature genes per cell normalised to gene detection rate and signature length.
Blue: low target gene signature abundance; Red: high target gene signature abundance. Cluster colour codes are given above, and transgene and FIRE
positivity codes are given below the heatmap
ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-10954-y
8NATURE COMMUNICATIONS | (2019) 10:2919 | https://doi.org/10.1038/s41467-019-10954-y | www.nature.com/naturecommunications
Content courtesy of Springer Nature, terms of use apply. Rights reserved
no mutations in KRAS, NRAS or BRAF), we found that Caco2
cells were KRASG12V-responsive, while SW48 cells were
KRASG12V-insensitive, extending a recent study that shows only
subtle effects of RAS mutants in SW4845 (Supplementary Fig. 8).
Our cluster-specic signalling data (Fig. 7a) showed correlated
activities of RAS-ERK, Wnt/β-catenin and other signalling
pathways, as they were generally higher in presumptive crypt
cell clusters 1 and 6, but lower in the presumptive
villus enterocyte clusters 3 and 4. Contrarily, the MRA approach
suggested that Axin2 as read-out of Wnt/β-catenin signalling
was a negative regulator of RAS-ERK. We reasoned that,
as we stimulated Wnt/β-catenin signalling by inhibiting
GSK3β, attenuation of RAS-ERK could be caused by other
targets of GSK3β. To more directly assess the effect of β-catenin
ab
3 d 1 d 1 d 3 h
Time
Treatments
CCM + REM
FLUC
KRASG12V
GSK3βi
cd e
Clusters
cCasp3
Axin2
p-ERK cl.Casp3
+GSK3βi
MEKi
p38i
fg
33 31
15
7
40
16
15
6
9
11
912
102
Axin2
Signal
EphB2 CD44
FLUC
KRASG12V
+FLUC +KRASG12V +FLUC +KRASG12V
+KRASG12V
CD24
101
102
101
102
103
101
102
103
101
123456 123456 123456 123456
102
101
102
101
102
103
101
102
101
102
103
101
102
103
101
Untreated
GSK3βi
Untreated
GSK3βi
Untreated
GSK3βi
Untreated
GSK3βi
6
6
4
4
2
2
1
3
5
0
–2
–5.0 –2.5 0.0
PC1
PC2
2.5
6
102
101
103
102
100
4
2
0
–2
–5.0 –2.5 0.0
PC1
PC2
2.5
6
4
2
0
–2
–5.0 –2.5 0.0
PC1
PC2
2.5
p-ERK
CD24
CD44
EphB2
Fig. 6 CyTOF analysis reveals KRASG12V- and GSK3βinhibitor-responsive p-ERK high cell clusters. aSchematics for generation of network perturbation
data by CyTOF. In short, organoids were established from KRASG12V- and FLUC transgenic mice, induced for transgene expression after 3 days, and treated
with GSK3βinhibitor for 1 day and with MEK and p38 inhibitors for 3 h before harvesting. Finally, 12 samples were subjected to multiplexed CyTOF analysis.
bDistributions of cell type markers in organoid cells induced for FLUC or KRASG12V transgenes plus/minus GSK3βinhibitor treatment. Central lines of
violin plots denote median values. cPCA showing colour code of k-means clustering in KRASG12V-induced cells by EphB2, CD44, CD24, Krt20 and cleaved
Caspase 3 signal strength. d,eMapping of signal strength for p-ERK and cleaved Caspase 3 on PCA, as in (c). fDistribution of EphB2, CD44, CD24, Axin2,
p-ERK and cleaved Caspase 3 signals in clusters 16, as above. Central lines of violin plots denote median values. gFractions of cells in clusters 16, in
organoid cells induced for FLUC or KRASG12V transgenes plus/minus GSK3βinhibitor treatment. Numbers denote percentages of cells in clusters 1, 5, 6.
CyTOF data are available as a Source Data le
NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-10954-y ARTICLE
NATURE COMMUNICATIONS | (2019) 10:2919 | https://doi.org/10.1038/s41467-019-10954-y | www.nature.com/naturecommunications 9
Content courtesy of Springer Nature, terms of use apply. Rights reserved
on RAS-ERK, we therefore performed further experiments
in transgenic organoids in which we induced transgenic
stabilised β-catenin or withdrew Wnt ligands. The data show
that supplementation and abrogation of β-catenin activity
both result in lower ERK phosphorylation in organoids
(Supplementary Fig. 9).
As the results of the modelling pinpointed differences in signal
transduction from KRASG12V to ERK to feed-forward and
feedback signalling between MEK and ERK, we investigated
which molecular mechanism might attenuate ERK activation in
intestinal cells. For this we sorted FLUC control or KRASG12V-
induced organoids with respect to their FIRE reporter levels, as
b
Mki67
Cd44
Ephb2
Kras
FLUC KRAS
FLUC KRAS
FIRE-lowFIRE-high
KEGG
MAPK
Mecom
Cdc25b
Flna
Pla2g4a
Stk3
Stmn1
Chp1
Rras
Ppm1a
Cacna1d
Gadd45a
Fgf14
* Dusp5
Rps6ka4
* Dusp1
* Dusp6
Cacna1h
Pla2g2f
Chp2
Cacna2d1
+FLUC +KRASG12V
1
2
3
4
5
6
p-4EBP1
Axin2
p-ERK
IκBα
p-MEK
p-p38
p-S6
p-4EBP1
Axin2
p-ERK
IκBα
p-MEK
p-p38
p-S6
Perturbations
Clusters
Read-outs for quantitative modelling
(Network nodes)
a
c
+MEKi
+p38i
+GSK3βi -
de
3
RAF
MEK
ERK
RAS
PI3K
AKT
4-EBP1
GSK3β
S6
Axin2
p38
IκBα
mTOR
Link added
1–5 Variable across clusters
2
4
5Link removed
1
ip38
iMEK
iGSK3β
p38−>S6
Axin2−>...RAF−>MEK
MEK−>ERK
GSK3β−>Axin2
ERK−>mTOR−>4EBP1
ERK−>RAF−>MEK
ERK−>S6
ERK−>IkBα
Axin2−>RAS−>p38
Axin2−>...PI3K−>...4EBP1
Axin2−>...PI3K−>...IκBα
–0.26
–0.91
3.41
–9.4
1.8
–6.3
0.24
–0.52
1.04
0.46
–5.7
–5.1
–2.2
–0.26
–0.91
0.44
–9.4
1.85
–6.3
0.24
–0.45
1.04
0.01
–5.7
–5.1
–2.2
–0.26
–0.91
0.34
–9.4
0.43
–6.5
0.24
–1.7
1.04
–0.07
–5.7
–5.1
–2.2
–0.26
–0.91
0.32
–9.4
0.2
–2.0
0.24
–5.1
1.04
0.1
–5.7
–5.1
–2.2
–0.26
–0.91
1.03
–9.4
0.41
–2.8
0.24
–3.5
1.04
0.07
–5.7
–5.1
–2.2
–0.26
–0.91
–9.10–3 –9.10–3 –9.10–3 –9.10–3 –9.10–3 –9.10–3
3.67
–9.4
0.67
–7.1
0.24
–3.0
1.04
0.91
–5.7
–5.1
–2.2
Clusters
p-ERK low
p-ERK high
5
1
3
2
4
Network connections
Clusters
123456
FK FK FK FK FK FK
123456
log2fc
1
0
–1
–2
log2fc
4
0
2
–2
811
Fig. 7 Network quantication identies cell type-specic differences in KRAS to ERK signalling. aProtein phosphorylation and abundance CyTOF data by
treatment and cell clusters, as in Fig. 6c. Log2 fold changes to average untreated FLUC-induced control line are given. bSignalling network structure used
for modelling. The network was re-parametrised from a starting network, using the experimental data to remove and add connections, denoted by grey and
blue arrows, respectively. cSignalling quantication of identiable network links using Modular Response Analysis. Numbers 15 in panels (b) and (c) show
network connections with signicant differences between clusters in order of detection. Red circles mark MEK-ERK and ERK-MEK connections identied as
having different strengths in clusters with high vs. low ERK phosphorylation after KRASG12V induction. dMRA modelling of differences between KRASG12V-
induced and FLUC control cells within each cluster. K and F mark KRASG12V and FLUC control cluster pairs, respectively. Cluster pairs exhibiting KRASG12V-
specic differences are shown in red and blue, indicating regulation strengths. eColour-coded gene expression data from cells sorted by high and low FIRE
activity, as indicated. Upper panel shows marker genes (Mki67, encoding Ki67, for proliferative cells, Cd44 and Ephb2 for crypt cells and Kras), lower panel
shows 20 signicantly regulated genes between the conditions. In total, 269 genes encoding MAPK network components in KEGG were tested. Asterisks
indicate dual-specicity phosphatases
ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-10954-y
10 NATURE COMMUNICATIONS | (2019) 10:2919 | https://doi.org/10.1038/s41467-019-10954-y | www.nature.com/naturecommunications
Content courtesy of Springer Nature, terms of use apply. Rights reserved
above (Fig. 5a) and performed bulk low-input RNA sequencing.
When inspecting 267 genes whose products are implicated in the
MAPK signalling pathway, we noticed three dual-specicity
phosphatases among a total of 20 differentially expressed genes
(Fig. 7e). As these phosphatases are known to dephosphorylate
ERK46, we consider the DUSP1, DUSP5 and DUSP6 gene
products as candidate mediators of attenuated MEK-ERK signal
transmission that we observe in the differentiated cells of
clusters 34.
Discussion
The mass cytometry, reporter assay and single-cell RNA
sequencing data that we present here support a model of
cell type-specic and cell-intrinsic regulation of the terminal
MAP kinase ERK (Fig. 8). We found that oncogenic KRAS can
activate ERK only in specic cell types of the intestinal epithe-
lium, while other cells such as enterocytes were insensitive to ERK
activation by oncogenic KRAS. Our quantitative network model
suggested that ERK activity is attenuated in the latter cells by
increased feedback inhibition and reduced feed-forward signal-
ling from MEK to ERK. In agreement, transcriptome analysis
showed that dual-specicity phosphatases (DUSPs) are selectively
expressed in cells with low ERK activity and may therefore
contribute to cell type-specic suppression of ERK. In addition,
crypt cells expanded in organoids upon induction of KRASG12V,
and this effect was increased when KRASG12V was combined with
GSK3β-mediated β-catenin activity.
The unexpected disparities in the levels of KRASG12V-induced
ERK phosphorylation in different cell types derived from
the intestinal epithelium extend our understanding of how KRAS,
the most prevalent oncogene in CRC, exerts its effects. It is of
note that our study only analysed the G12V mutation of KRAS,
and other mutations such as the more common G12D variant or
mutations at amino acids 13, 61 or 146 may engage different
effectors47,48. Local differences of ERK activity have recently been
found in clinical specimens of CRC, including KRAS-mutant
CRC14. In this previous study, ERK levels were generally higher in
cancer cells adjacent to stromal cells at the invasive front, and
lower in more central areas of cancer specimens, in line with
modulation of ERK activity by cues from the microenvironment.
Our results agree with a model of dynamic ERK activity in cancer
tissues, and we nd in addition that ERK is not only regulated via
external cues from the microenvironment, but also by the cell-
intrinsic differentiation state.
Quantitative modelling of ERK activity from perturbation data
revealed that the markedly distinct abilities of KRASG12V to
activate ERK are due to different strengths of two network con-
nections between cell types, namely MEK to ERK feed forward
and ERK to MEK negative feedback signalling. Negative feedback
within the pathway is well-known, and has been linked to
induction of cellular senescence49 and to therapy resistance in
CRC44,50. Our study has identied dual-specicity phosphatases
as candidate genes modulating MEK-to-ERK signalling during
intestinal differentiation. Indeed, the Dusp1,Dusp5 and Dusp6
genes we have identied to be differentially expressed between
RAS-to-ERK-responsive and -insensitive intestinal cells, encode
ERK-specic phosphatases46. A protein phosphatase network,
including DUSPs, has recently been found to control ERK activity
and differentiation in skin51. Our study suggests the existence of
similar control mechanisms in the intestine. Dual-specicity
phosphatases targeting ERK have been implicated in resistance of
non-small lung cancer to anti-EGFR therapy52. It remains to be
seen whether differences in RAS-ERK signalling beyond the
mutational status of BRAF, NRAS and KRAS have prognostic
value in CRC. Today, CRC patients with wild-type status of the
predictive markers KRAS, NRAS and BRAF are eligible for anti-
EGFR therapy53,54. However, the treatment also generates mixed
outcomes among eligible patients, showing the need to identify
novel markers and mechanisms contributing to differences in
EGFR-RAS-ERK signal transduction and therapy response.
Paneth cells and enterocytes represent the main differentiated
cell types of the intestinal epithelium, and we found that they
have marked differences in their abilities to activate ERK. It is of
note that the former reside in the Wnt/β-catenin-high crypt
compartment, while the latter inhabit a compartment with low β-
catenin activity. Indeed, a functional role for Wnt/β-catenin in
the activation of ERK in intestinal epithelium has been proposed
before55. In our experiments using organoids, p-ERK levels were
lowered when we increased or decreased β-catenin activity
(Supplementary Fig. 9A, B). We hypothesise that this is due to a
loss of all crypt cells or of phospho-ERK-high Paneth cells in the
models with low and high β-catenin activity, respectively. Our
data agree with a recent publication proposing negative interac-
tion between Wnt/β-catenin and ERK signals in intestinal stem
cells56. Indeed, activation of β-catenin by treatment with a GSK3β
inhibitor increased the fraction of cells positive for stem cell
marker CD44, but not levels of MEK or ERK phosphorylation per
cell in the CyTOF experiments.
ERK activity is regulated on multiple levels, including its
subcellular localisation57. We used different approaches to
quantify ERK activity. While mass cytometry and capillary pro-
tein analysis measure total levels of ERK phosphorylation, the
FIRE reporter assesses nuclear ERK activity only23. These activity
measures for crypt cells suggest divergence of total and nuclear
activity. While all crypt cells appeared FIRE positive, the pre-
sumptive undifferentiated crypt cell cluster 1 showed only
intermediate levels of ERK phosphorylation in the CyTOF ana-
lysis. Further analyses will be required to understand how sub-
cellular ERK activity is controlled in a cell type-specic manner.
In our experiments, signal transduction from BRAFV600E to
ERK was relatively independent of cellular context. Extending
previous studies29,30, we found that high levels of ERK activity
induced by oncogenic BRAFV600E are not tolerated in the intes-
tine. This is in contrast to CRC and cell lines, where BRAFV600E
amplications exist and are selected for by MEK inhibition58,59.It
EGF
EGFR
Differentiation
Crypt
Villus
Signalling
β-Catenin
KRASMUT
RAS
BRAF
MEK
ERK DUSPs
TFs
EGF
EGFR
RAS
BRAF
MEK
ERK
TFs
Fig. 8 Model of cell type-specic regulation of ERK activity. ERK is regulated
cell type-specic and cell-intrinsic via different strengths of feedback
inhibition and feed-forward signalling from MEK to ERK. Dual-specicity
phosphatases (DUSPs) are important regulators of ERK activity. β-catenin
and KRASG12V activities modulate cell fate decisions towards the
generation of cells with high ERK activity, likely in part due to low
expression of genes encoding DUSPs
NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-10954-y ARTICLE
NATURE COMMUNICATIONS | (2019) 10:2919 | https://doi.org/10.1038/s41467-019-10954-y | www.nature.com/naturecommunications 11
Content courtesy of Springer Nature, terms of use apply. Rights reserved
thus appears that the corridor for acceptable ERK activity is
tuneable during CRC progression and under selective pressure
exerted, for instance, by targeted therapy. Therefore, our ndings
are reminiscent of the just rightsignalling model that has been
proposed to explain step-wise increases of β-catenin activity in
CRC progression60.
Methods
Generation of transgenic mice. Transgene cassettes were constructed by linking
tdTomato to human BRAF(V600E), KRAS(G12V) and/or murine stabilised
mutant Ctnnb1 (S33A, S37A, T41A, S45A) or rey luciferase via 2A peptides, and
subsequent cloning of these gene combinations into a doxycycline-inducible
expression cassette anked by heterologous loxP sites, and integrated into a pre-
viously modied Gt(ROSA)26Sor locus of F1 hybrid B6/129S6 embryonic stem
cells by Cre recombinase-mediated cassette exchange28. Transgenic animal
experiments were approved by Berlin authorities LAGeSo (G0185/09, G0143/14).
Organoid and cell line culture. Mouse organoid cultures were initiated and
propagated as described before27, using 48-well plates with 15 µl droplets of
Matrigel (Corning) per well overlaid with 300 µl crypt culture medium containing
EGF (50 ng/ml), Noggin (100 ng/ml) and R-Spondin1 (functionally tested from R-
Spondin-conditioned medium; CCM-REN). Transgenes were induced by addition
of 2 µg/ml doxycycline to the medium.
To obtain adenomatous organoids (spheroids) after induction of stabilised β-
catenin, R-Spondin was removed after induction of the transgene encoding β-
cateninstab alone or in combination with KRASG12V. Spheroids were dissociated
with TrypLE (Gibco) for 3 min and Rho kinase inhibitor Y27632 (10 μM) was
added to the culture medium after passaging to prevent anoikis.
For viral transfection, a protocol from reference61 was employed, with
modications: organoids were cultured in the presence of Y27632 and the GSK3β
inhibitor CHIR99021 for 2 days. Next, organoids were disaggregated into single
cells using TrypLE (Gibco) for 5 min at 37 °C. Cell suspensions were spin-occulated
in an ultra-low adhesion round bottom 96-well plate with the virus at 300 g for
45 min. Subsequently, cells were resuspended in Matrigel, and cultured for 2 days
in CCM-REN supplemented with Y27632 and CHIR99021. Medium was then
replaced by CCM-REN containing 2 μg/ml puromycin to select for transfected
cells. As viral transfection initially resulted in organoid pools that were
heterogeneous for FIRE reporter activity, single FIRE positive organoids were
manually selected and propagated before experimental analysis.
Organoid survival was scored as follows: cultures were passaged, inhibitors and
doxycycline were applied with the culture medium directly after passaging.
Individual wells were imaged using the z-stack function of BioZero observation and
analyser software (Keyence) on day 1 and 4 (organoids) or day 1 and 6 (spheroids),
and full focus reconstructed images were used for quantication.
Patient-derived organoids (PD3Ds) were obtained from the biobank of the
Charité - Universitätsmedizin Berlin and experiments were approved by the
ethics commission of Charité - Universitätsmedizin Berlin (EA1/011/18).
Cells were cultivated in 24-well plates (Corning) in medium containing human
bFGF (20 ng/ml, Sigma Aldrich) and EGF (50 ng/ml, Sigma Aldrich), as
published62.
SW48 and Caco2 CRC cells were cultured in L-15 and DMEM, respectively,
supplemented with 10% foetal bovine serum. Cells were transfected using
Lipofectamin 3000 Transfection Reagent (Thermo Fischer) with vectors encoding
BRAFV600E, KRASG12V or FLUC linked to tdTomato and the pTet-on Advanced
Vector (Clontech). Cells were starved in medium containing 0.1% foetal bovine
serum and induced with 2 µg/ml doxycycline 48 h after transfection. Twenty-four
hours later, cells were harvested using TrypLE, washed, rested for 30 min at 37 °C
in starvation medium and xed in 4% paraformaldehyde (PFA) for 15 min at 37 °C.
Fixed cells were washed in PBS/1% bovine serum albumin (BSA), permeabilised in
MeOH at 20 °C overnight, and immunostained with Alexa Fluor 488 mouse anti-
ERK1/2 (pT202/pY204) antibody (1:10; 612592, BD Bioscience) for 30 min. Cell
lines are routinely checked for mycoplasma contamination and panel-sequenced
for oncogenic mutations to conrm identity.
The following inhibitors (SelleckChem ) were employed: AZD6244 (10 μM),
BVD-523 (3 μM), AZD8931 (50 nM), CHIR99021 (6 µM), Getinib (1 μM) and
LY2228820 (200 nM).
Mass cytometry (CyTOF). For CyTOF analysis, we used the following pre-
conjugated antibodies (Fluidigm) as per manufacturers recommendation: CD24
(for mouse: 150-Nd, 3150009B, for human: 169-Tm, 3169004B), CD44 (for mouse:
162-Dy, 3162030B, for human: 166-Er, 3166001B), cleaved Casp3 (142-Nd,
3142004A), p-H2AX [S139] (147-Sm, 3147016A), p-Akt [S473] (152-Sm,
3152005A), p-p38 [T180/Y182] (156-Gd, 3156002A), Ki67 (162-Dy, 3168007B),
IκBα(164-Dy, 3164004A), p-ERK1/2 [T202/Y204] (171-Yb, 3171010A) and p-S6
[S235/236] (175-Lu, 3175009A).
For antibodies not available as metal-conjugates, we used the Maxpar Antibody
Labeling Kit (Fluidigm) according to manufacturers instructions for addition of
the respective metal tags: Axin2 (145-Nd, Abcam, ab32197, 2 µg/ml), p-MEK1/2
[S217/221] (151-Eu, CST, 41G9 , 2 µg/ml), EphB2 (158-Gd, BD, 2H9, 2 µg/ml),
p-4e-BP1 [T37/46] (170-Er, CST, 236B4, 2 µg/ml) and Krt20 (176-Yb, CST,
D9Z1Z, 2 µg/ml). The yield after conjugation was determined using a NanoDrop
spectrometer measuring the absorbance at 280 nm wavelength.
For measurements, organoids were harvested in PBS and digested to a single-
cell solution in 1:1 Accutase (Biolegend) and TrypLE (Gibco) with addition of
100 U/ml Universal Nuclease (Thermo Scientic) at 37 °C. Cells were counted and
a maximum of 500,000 cells were stained with 5 µM Cell-ID Cisplatin (Fluidigm)
in PBS for 5 min at 37°C. After washing, cells were resuspended in their respective
growth medium and allowed to rest for 30 min at 37 °C. Subsequently, cells were
resuspended in BSA/PBS solution, mixed 1:1.4 with Proteomics Stabilizer (Smart
Tube Inc.) and incubated for 10 min at room temperature. Afterwards the cells
were frozen at 80 °C for storage.
One day prior to analysis, cells were thawed in a 37 °C water bath and mixed
with Maxpar Cell Staining Buffer (CSB, Fluidigm). We used the Cell-ID 20-Plex Pd
Barcoding Kit (Fluidigm) to label different samples and performed a downscaled
version of the manufacturers recommended protocol. Cells were washed again in
CSB, then in Barcode Perm Buffer (Fluidigm). After resuspension in 200 µl Barcode
Perm Buffer, 25 µl of the diluted Barcoding Reagents were added to the respective
samples and incubated for 30 min at room temperature. Afterwards cells were
washed twice in CSB, pooled into one tube and counted. In total, 3 × 106cells were
then stained with a surface antibody cocktail for 30 min at room temperature. After
washing in CSB, cells were rexed in 2% methanol-free formaldehyde solution
(Pierce; diluted in Maxpar PBS, Fluidigm) for 10 min at room temperature. Cells
were washed in CSB and put on ice for 10 min. Next, cells were permeabilised by
adding 4 °C methanol for 15 min. Cells were washed twice in CSB and incubated
with a phospho-protein antibody cocktail for 30 min at room temperature. Cells
were washed twice in CSB and incubated with 62.5 nM Cell-ID Intercalator-Ir in
Maxpar PBS for 20 min at room temperature. Cells were washed in Maxpar PBS
and xed in 2% methanol-free formaldehyde overnight at 4 °C. The day after, cells
were washed with CSB and then twice with Milli-Q water. Cell number was
adjusted to 2.55×10
5/ml with Milli-Q water, cells were ltered through a 20 µm
cell strainer (CellTrics, Sysmex) and supplemented 1:10 with EQ Four Element
Calibration Beads (Fluidigm). Data were acquired on a Helios CyTOF system. Mass
cytometry data were normalised using the Helios software and bead-related events
were removed. Doublets were excluded by ltering for DNA content (191Ir and
193Ir) vs. event length, and apoptotic debris removed by a lter in the platin
channel (195Pt). De-convolution of the barcoded sample was performed using the
CATALYST R package version 1.5.363.
Capillary protein quantication and RAS activity assay. Protein sample pre-
paration and quantication of p-ERK was performed using a WES capillary system
(12230 kD Master kit α-RabbitHRP; PS-MK01; Protein Simple) and the antibody
p-ERK/2(T202/Y204) (1:50; #9101, Cell Signaling). Raw p-ERK values were nor-
malised to vinculin (1:30; #4650; Cell Signaling). RAS activity was determined
using the Ras activation Assay Biochem Kit (Cytoskeleton Inc. #BK008), according
to manufacturers instructions and quantied using an LiCor Odyssee CXL scanner
and LiCor Image Studio software.
Immunohistochemistry. Immunohistochemistry was done on PFA-xed and
parafn-embedded tissues. Organoids were xed in 4% PFA for 30 min, while
intestines were xed overnight at room temperature. Subsequently, tissues were
dehydrated in a graded ethanol series, followed by xylene. Tissues were parafne-
embedded, sectioned at 4 µm and mounted on Superfrost Plus slides (Thermo
Fisher Scientic). Sections were deparafnised, rehydrated, bleached for 10 min in
3% H
2
O
2
. Antigens were retrieved using 10 mM Na-citrate, pH 6 for 20 min at
boiling temperature. The following antibodies were used: P-ERK (T202/Y204;
#4370 CellSignal); P-MEK (S217/221; #9121 CellSignal); anti-RFP (1:200; #600-
401-379 Rockland). ImmPRESS secondary antibody and NovaRED substrate kits
(Vector Labs, Burlingame, CA, USA) were used for signal detection, according to
manufacturers protocols.
Immunouorescence and microscopy. For immunouorescence imaging, orga-
noids were washed with PBS and xed in-well with 4% PFA for 30 min at 37 °C.
Fixation was stopped with PBS containing 100 nm Glycine. Cells were blocked and
permeabilised with blocking buffer (PBS containing 1% BSA, 0.2% Triton X-100,
0.05% Tween-20) for at least 2.5 h at room temperature. Samples were incubated
for 36 h at 4 °C with primary antibody against lysozyme (1:250; ab108508, Abcam)
diluted in blocking buffer. After washing with IF-buffer (PBS containing 0.1% BSA,
0.2% Triton X-100, 0.05% Tween-20), samples were incubated for 24 h at 4 °C with
secondary antibody Alexa Fluor 647 anti-rabbit (1:500, 4414, Molecular Probes)
diluted in IF-buffer. Samples were counterstained for 5 min at room temperature
using 0.5 µg/ml DAPI. After washing with IF-buffer, stained cultures were released
from the Matrigel and collected in PBS. Samples were washed, resuspended in
remaining PBS and mounted on slides using Vectashield Antifade Mounting
Medium (H-1000, Vector).
Immunouorescence and FIRE reporter images were taken with a Leica TSC
SPE confocal microscope using an ACS 20× oil-immersion objective, solid-state
lasers (405, 488, 532 and 635 nm) as sources of excitation and LAS X operating
ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-10954-y
12 NATURE COMMUNICATIONS | (2019) 10:2919 | https://doi.org/10.1038/s41467-019-10954-y | www.nature.com/naturecommunications
Content courtesy of Springer Nature, terms of use apply. Rights reserved
software (Leica). Light microscopy images of cultures were taken with a Biozero
microscope using a Plan Apo 4× NA 0.20 objective and Biozero observation and
analyser software (Keyence).
For transmission electron microscopy, organoids were induced for 24 h,
removed from Matrigel and xed overnight at 4 °C in Karnovskysxative
containing 2% PFA and 2.5% glutaraldehyde in PBS, washed three times in PBS
and embedded in 1% LM-agrose. Agorose cubes (~1 mm3) containing several
organoids were subsequently stained, using 0.5% OsO
4
in PBS (2.5 h at RT), 0.1%
Tannic Acid in 100 mM Hepes pH 7.4 (1 h at RT) and 2% Uranyle acetate (1.5 h at
RT), each time followed by three washing steps in PBS. Samples were dehydrated in
a graded ethanol series, embedded in Spurrs resin, thin-sectioned (70 nm, Leica
UC7) and post-contrasted as described64. Regions of ~100150 µm2showing
representative sections through organoids were imaged on a 120 kV Tecnai Spirit
transmission electron microscope (FEI) equipped with a F416 CMOS camera
(TVIPS). Micrographs were recorded automatically at a nal magnication of
4400 × (2.49 nm pixel size at object scale) and 10 µm defocus using Leginon65
and then stitched using TrakEM266.
Flow cytometry and uorescence-activated cell sorting. Flow cytometry of anti-
p-ERK-stained CRC cells resuspended in PBS/1% BSA cells was done using an
Accuri cytometer (BD). Cells were gated for populations displaying different
tdTomato uorescence values (negative, low, medium and high), which correlates
with transgene expression. For each population, the mean anti-p-ERK uorescence
values were determined and normalised to the tdTomato negative fraction of the
corresponding FLUC control experiment.
For uorescence-activated cell sorting of organoid cells, single-cell suspensions
from induced organoids were prepared by digestion with TrypLE (Thermo Fisher
Scientic) in the presence of 2 mM EDTA and 200 U/ml DNAse I. Digestion was
monitored by visual inspection and stopped by crypt culture medium containing
0.2% BSA. Cell suspensions were ltered through 30 µm Celltrix lters and stained
with an anti-CD44-antibody conjugated to Allophycocyanin (APC; clone IM7,
BioLegend) and the Green or Near-IR Live/Dead Fixable Dead Cell Stain Kits (Life
Tech) for subsequent exclusion of dead cells. Single cells were sorted into the 96-
well plates of the Precise WTA Kit with predispensed library chemistry using a BD
FACSAriaII SORP (BD) and a gating strategy as displayed in Supplementary Fig. 3.
Cells were sorted into quadrants of plates to minimise batch effects between plates.
For later analysis of CD44 positivity of the subsets, sorts were performed as
index sorts.
RNA sequencing and bioinformatic analyses. For single-cell RNA sequencing,
the Precise WTA Kit (BD) was used, according to the manufacturersinstructions.
Sequences were produced using NextSeq and/or HiSeq chemistry (Illumina).
Cluster generation on NextSeq 500 followed the instructions of the manufacturer,
at a nal loading concentration of 2 pM on a high-output-owcell. 1% PhiX
was added as quality control, at least 40 million paired reads per pool were
gained during a Paired-End-75 run. Library-pools running on the HiSeq4000-
system were prepared according to Illumina recommendations, loaded with
200 pM concentration and sequenced during a Paired-End-75 run. Again, 1% PhiX
was added as quality control, and at least 40 million read-pairs per pool were
targeted.
Single-cell RNA-sequencing data were pre-processed using the BD Precise
Whole Transcriptome Assay Analysis Pipeline v2.067. Quality control was
performed using scater68. Read counts were normalised using the trimmed mean of
median values (TMM) approach provided with edgeR69. Normalised read counts
were used for k-means clustering and t-SNE visualisation. Differentiation
trajectories in t-SNE plots were determined using slingshot70, with intestinal stem
cell cluster 1 as predened origin. Differentially expressed genes were called on log-
transformed raw counts using a hurdle model provided with R package MAST71.
Top-10 signature genes per cluster were identied by comparing average gene
expressions within cluster to average gene expressions across all other clusters. For
bulk cell RNA sequencing, organoids were induced for 24 h with 2 µg/ml
doxycycline in CCM-REN medium and subsequently dissociated using TrypLE
(Thermo Fisher Scientic). RNA-seq reads were aligned to the mouse genome
GRCm38 using STAR aligner with GENCODE annotation vM11. Differentially
expressed genes were called using DESeq2.
Mathematical modelling. We quantied and locally adjusted the signalling net-
works from the KRAS-mutant perturbation data using an adjusted version of
Modular Response Analysis as implemented in the R package STASNet, version
1.0.272 (https://github.com/molsysbio/STASNet/tree/STASNet1.0.2) as follows: As
input data, we derived representative mean and standard error of the mean values
from the single-cell CyTOF data (trimming the lower and upper 5% of signals) and
a literature-informed prior network. As the apparent response pattern across
clusters was similar, it was decided to use a joint modelling approach, i.e. we rst
quantied the response coefcients for all six clusters by a single set of coefcients
using a combination of latin hypercube sampling with subsequent Levenberg-
Marquardt tting (n=4×10
4) and then iteratively derived and quantied the
signicantly different signalling coefcients between clusters using a likelihood
ratio test. Afterwards we searched for biologically justiable link extensions lacking
in the network to better describe the data. The whole procedure was repeated until
no further justiable link additions could be found, followed by a removal round of
statistically insignicant links.
Reporting summary. Further information on research design is available in
the Nature Research Reporting Summary linked to this article.
Data availability
scRNA-seq and bulk RNA-seq data are available from GEO repository under accession
numbers GSE115242 and GSE115234, respectively. Data underlying Figs. 1, 6,
Supplementary Fig. 7 (CyTOF data) and Fig. 2b, d, e, and Supplementary Figs. 1c, 6, 8, 9
are provided as source data les. Modelling information is available as a source data le.
Received: 6 June 2018 Accepted: 12 June 2019
References
1. Beumer, J. & Clevers, H. Regulation and plasticity of intestinal stem cells
during homeostasis and regeneration. Development 143, 36393649 (2016).
2. Stelniec-Klotz, I. et al. Reverse engineering a hierarchical regulatory network
downstream of oncogenic KRAS. Mol. Syst. Biol. 8, 601 (2012).
3. Sasaki, N. et al. Reg4+deep crypt secretory cells function as epithelial niche for
Lgr5+stem cells in colon. Proc. Natl Acad. Sci. USA 113, E5399E5407 (2016).
4. Sato, T. et al. Paneth cells constitute the niche for Lgr5 stem cells in intestinal
crypts. Nature 469, 415418 (2010).
5. Pylayeva-Gupta, Y., Grabocka, E. & Bar-Sagi, D. RAS oncogenes: weaving a
tumorigenic web. Nat. Rev. Cancer 11, 761774 (2011).
6. Holdereld, M., Deuker, M. M., McCormick, F. & McMahon, M. Targeting
RAF kinases for cancer therapy: BRAF-mutated melanoma and beyond. Nat.
Rev. Cancer 14, 455467 (2014).
7. Ciardiello, F. et al. Differential expression of epidermal growth factor-related
proteins in human colorectal tumors. Proc. Natl Acad. Sci. USA 88, 77927796
(1991).
8. Papke, B. & Der, C. J. Drugging RAS: know the enemy. Science 355,
11581163 (2017).
9. Fearon, E. R. & Vogelstein, B. A genetic model for colorectal tumorigenesis.
Cell 61, 759767 (1990).
10. Fearon, E. R. Molecular genetics of colorectal cancer. Annu. Rev. Pathol. 6,
479507 (2011).
11. Rad, R. et al. A genetic progression model of Braf(V600E)-induced intestinal
tumorigenesis reveals targets for therapeutic intervention. Cancer Cell 24,
1529 (2013).
12. Yachida, S., Mudali, S., Martin, S. A., Montgomery, E. A. & Iacobuzio-
Donahue, C. A. Beta-catenin nuclear labeling is a common feature of sessile
serrated adenomas and correlates with early neoplastic progression after
BRAF activation. Am. J. Surg. Pathol. 33, 18231832 (2009).
13. Morkel, M., Riemer, P., Sers, C., Bläker, H. & Sers, C. Similar but different:
distinct roles for KRAS and BRAF oncogenes in colorectal cancer
development and therapy resistance. Oncotarget 6, 2078520800 (2015).
14. Blaj, C. et al. Oncogenic effects of high MAPK activity in colorectal cancer
mark progenitor cells and persist irrespective of RAS mutations. Cancer Res.
77, 17631774 (2017).
15. Hlubek, F. et al. Heterogeneous expression of Wnt/beta-catenin target genes
within colorectal cancer. Int. J. Cancer 121, 19411948 (2007).
16. Snippert, H. J. et al. Prominin-1/CD133 marks stem cells and early
progenitors in mouse small intestine. Gastroenterology 136, 21872194.e1
(2009).
17. Zhu, L. et al. Prominin 1 marks intestinal stem cells that are susceptible to
neoplastic transformation. Nature 457, 603607 (2009).
18. OBrien, C. A., Pollett, A., Gallinger, S. & Dick, J. E. A human colon cancer cell
capable of initiating tumour growth in immunodecient mice. Nature 445,
106110 (2007).
19. Todaro, M. et al. Colon cancer stem cells dictate tumor growth and resist cell
death by production of interleukin-4. Cell Stem Cell 1, 389402 (2007).
20. Merlos-Suárez, A. et al. The intestinal stem cell signature identies colorectal
cancer stem cells and predicts disease relapse. Cell Stem Cell 8, 511524
(2011).
21. Sadanandam, A. et al. A colorectal cancer classication system that associates
cellular phenotype and responses to therapy. Nat. Med. 19, 619625 (2013).
22. Kreso, A. et al. Variable clonal repopulation dynamics inuence
chemotherapy response in colorectal cancer. Science 339, 543548 (2013).
23. Albeck, J. G., Mills, G. B. & Brugge, J. S. Frequency-modulated pulses of ERK
activity transmit quantitative proliferation signals. Mol. Cell 49, 249261
(2013).
NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-10954-y ARTICLE
NATURE COMMUNICATIONS | (2019) 10:2919 | https://doi.org/10.1038/s41467-019-10954-y | www.nature.com/naturecommunications 13
Content courtesy of Springer Nature, terms of use apply. Rights reserved
24. Grün, D. et al. Single-cell messenger RNA sequencing reveals rare intestinal
cell types. Nature 525, 251255 (2015).
25. Li, H. et al. Reference component analysis of single-cell transcriptomes
elucidates cellular heterogeneity in human colorectal tumors. Nat. Genet. 49,
708718 (2017).
26. Bodenmiller, B. et al. Multiplexed mass cytometry proling of cellular
states perturbed by small-molecule regulators. Nat. Biotechnol. 30, 858867
(2012).
27. Sato, T. et al. Single Lgr5 stem cells build crypt-villus structures in vitro
without a mesenchymal niche. Nature 459, 262265 (2009).
28. Vidigal, J. A. et al. An inducible RNA interference system for the functional
dissection of mouse embryogenesis. Nucleic Acids Res. 38, e122 (2010).
29. Riemer, P. et al. Transgenic expression of oncogenic BRAF induces loss of
stem cells in the mouse intestine, which is antagonized by β-catenin activity.
Oncogene 34, 31643175 (2015).
30. Tong, K. et al. Degree of tissue differentiation dictates susceptibility to BRAF-
driven colorectal cancer. Cell Rep. 21, 38333845 (2017).
31. Yeh, T. C. et al. Biological characterization of ARRY-142886 (AZD6244), a
potent, highly selective mitogen-activated protein kinase kinase 1/2 inhibitor.
Clin. Cancer Res. 13, 15761583 (2007).
32. Sullivan, R. J. et al. First-in-class ERK1/2 inhibitor ulixertinib (BVD-523) in
patients with MAPK mutant advanced solid tumors: Results of a phase I dose-
escalation and expansion study. Cancer Discov. 8, 184195 (2018).
33. Barlaam, B. et al. Discovery of AZD8931, an equipotent, reversible inhibitor of
signaling by EGFR, HER2, and HER3 receptors. ACS Med. Chem. Lett. 4,
742746 (2013).
34. Uhlitz, F. et al. An immediatelate gene expression module decodes ERK
signal duration. Mol. Syst. Biol. 13, 928 (2017).
35. Wang, F. et al. Isolation and characterization of intestinal stem cells based on
surface marker combinations and colony-formation assay. Gastroenterology
145, 383-95.e121 (2013).
36. Gonzalo, D. H. et al. Gene expression proling of serrated polyps identies
annexin A10 as a marker of a sessile serrated adenoma/polyp. J. Pathol. 230,
420429 (2013).
37. Peeters, T. & Vantrappen, G. The Paneth cell: a source of intestinal lysozyme.
Gut 16, 553558 (1975).
38. Yin, X. et al. Niche-independent high-purity cultures of Lgr5+intestinal stem
cells and their progeny. Nat. Methods 11, 106112 (2014).
39. Ishitsuka, K. et al. p38 mitogen-activated protein kinase inhibitor LY2228820
enhances bortezomib-induced cytotoxicity and inhibits osteoclastogenesis in
multiple myeloma; therapeutic implications. Br. J. Haematol. https://doi.org/
10.1111/j.1365-2141.2008.07044.x (2008).
40. Lustig, B. et al. Negative feedback loop of Wnt signaling through upregulation
of conductin/axin2 in colorectal and liver tumors. Mol. Cell. Biol. 22,
11841193 (2002).
41. Schwitalla, S. et al. Intestinal tumorigenesis initiated by dedifferentiation and
acquisition of stem-cell-like properties. Cell 152,2538 (2013).
42. Vermeulen, L. et al. Dening stem cell dynamics in models of intestinal tumor
initiation. Science 342, 995998 (2013).
43. Kholodenko, B. N. et al. Untangling the wires: a strategy to trace functional
interactions in signaling and gene networks. Proc. Natl Acad. Sci. USA https://
doi.org/10.1073/pnas.192442699 (2002).
44. Klinger, B. et al. Network quantication of EGFR signaling unveils potential
for targeted combination therapy. Mol. Syst. Biol. 9, 673 (2013).
45. Hood, F. E. et al. Isoform-specic Ras signaling is growth factor dependent.
Mol. Biol. Cell mbc-E18 (2019).
46. Kidger, A. M. & Keyse, S. M. The regulation of oncogenic Ras/ERK
signalling by dual-specicity mitogen activated protein kinase phosphatases
(MKPs). Semin. Cell Dev. Biol. https://doi.org/10.1016/j.semcdb.2016.01.009
(2016).
47. Hunter, J. C. et al. Biochemical and Structural Analysis of Common Cancer-
Associated KRAS Mutations. Mol. Cancer Res. https://doi.org/10.1016/j.
semcdb.2016.01.009 (2015).
48. Hammond, D. E. et al. Differential reprogramming of isogenic colorectal
cancer cells by distinct activating KRAS mutations. J. Proteome Res.https://
doi.org/10.1021/pr501191a (2015).
49. Courtois-Cox, S. et al. A negative feedback signaling network underlies
oncogene-induced senescence. Cancer Cell https://doi.org/10.1016/j.
ccr.2006.10.003 (2006).
50. Prahallad, A. et al. Unresponsiveness of colon cancer to BRAF(V600E)
inhibition through feedback activation of EGFR. Nature 483, 100103 (2012).
51. Mishra, A. et al. A protein phosphatase network controls the temporal and
spatial dynamics of differentiation commitment in human epidermis. Elife
https://doi.org/10.7554/eLife.27356 (2017).
52. Phuchareon, J., McCormick, F., Eisele, D. W. & Tetsu, O. EGFR inhibition
evokes innate drug resistance in lung cancer cells by preventing Akt activity
and thus inactivating Ets-1 function. Proc. Natl Acad. Sci. USA https://doi.org/
10.1073/pnas.1510733112 (2015).
53. Karapetis, C. S. et al. K-ras mutations and benet from cetuximab in advanced
colorectal cancer. N. Engl. J. Med. 359, 17571765 (2008).
54. Amado, R. G. et al. Wild-type KRAS is required for panitumumab efcacy in
patients with metastatic colorectal cancer. J. Clin. Oncol. 26, 16261634
(2008).
55. Phelps, R. A. et al. A two-step model for colon adenoma initiation and
progression caused by APC loss. Cell 137, 623634 (2009).
56. Kabiri, Z. et al. Wnt signaling suppresses MAPK-driven proliferation of
intestinal stem cells. J. Clin. Invest. https://doi.org/10.1172/JCI99325 (2018).
57. Lenormand, P., Brondello, J. M., Brunet, A. & Pouysségur, J. Growth factor-
induced p42/p44 MAPK nuclear translocation and retention requires both
MAPK activation and neosynthesis of nuclear anchoring proteins. J. Cell Biol.
https://doi.org/10.1083/jcb.142.3.625 (1998).
58. Corcoran, R. B. et al. BRAF gene amplication can promote acquired
resistance to MEK inhibitors in cancer cells harboring the BRAF V600E
mutation. Sci. Signal. 3, ra84 (2010).
59. Little, A. S. et al. Amplication of the driving oncogene, KRAS or BRAF,
underpins acquired resistance to MEK1/2 inhibitors in colorectal cancer cells.
Sci. Signal. 4, ra17 (2011).
60. Albuquerque, C. et al. The just-rightsignaling model: APC somatic
mutations are selected based on a specic level of activation of the beta-
catenin signaling cascade. Hum. Mol. Genet. 11, 15491560 (2002).
61. Fujii, M., Matano, M., Nanki, K. & Sato, T. Efcient genetic engineering of
human intestinal organoids using electroporation. Nat. Protoc. 10, 14741485
(2015).
62. Schütte, M. et al. Molecular dissection of colorectal cancer in pre-clinical
models identies biomarkers predicting sensitivity to EGFR inhibitors. Nat.
Commun. 8, 14262 (2017).
63. Chevrier, S. et al. Compensation of signal spillover in suspension and imaging
mass cytometry. Cell Syst. https://doi.org/10.1016/j.cels.2018.02.010 (2018).
64. Matz, P. et al. Footprint-free human fetal foreskin derived iPSCs: A tool for
modeling hepatogenesis associated gene regulatory networks. Sci. Rep. https://
doi.org/10.1038/s41598-017-06546-9 (2017).
65. Suloway, C. et al. Automated molecular microscopy: the new Leginon system.
J. Struct. Biol. 151,4160 (2005).
66. Cardona, A. et al. TrakEM2 software for neural circuit reconstruction. PLoS
ONE 7, e38011 (2012).
67. Fu, G. K., Wilhelmy, J., Stern, D., Fan, H. C. & Fodor, S. P. A. Digital encoding
of cellular mRNAs enabling precise and absolute gene expression
measurement by single-molecule counting. Anal. Chem. 86, 28672870
(2014).
68. McCarthy, D. J., Campbell, K. R., Lun, A. T. L. & Wills, Q. F. Scater: pre-
processing, quality control, normalization and visualization of single-cell
RNA-seq data in R. Bioinformatics 33, 11791186 (2017).
69. Robinson, M. D., McCarthy, D. J. & Smyth, G. K. edgeR: a Bioconductor
package for differential expression analysis of digital gene expression data.
Bioinformatics 26, 139140 (2010).
70. Street, K. et al. Slingshot: Cell lineage and pseudotime inference for single-cell
transcriptomics. bioRxiv 128843, https://doi.org/10.1101/128843 (2017).
71. Finak, G. et al. MAST: a exible statistical framework for assessing
transcriptional changes and characterizing heterogeneity in single-cell RNA
sequencing data. Genome Biol. 16, 278 (2015).
72. Dorel, M. et al. Modelling signalling networks from perturbation data.
Bioinformatics https://doi.org/10.1093/bioinformatics/bty473 (2018).
Acknowledgements
The authors gratefully acknowledge excellent technical assistance by Anja Sieber (IRI Life
Sciences and Charité Universitätsmedizin Berlin), Gaby Bläss and Sonja Banko (MPIMG,
Berlin) for RAS activity assays, mouse genotyping and mouse care, respectively. The
authors also gratefully acknowledge the help of lab students Ekaterina Eroshok and
Maximilian Anders (Molecular Medicine Masters programme, Charité Uni-
versitätsmedizin Berlin) with immunohistochemistry in the early phase of this project.
We received the FIRE plasmid as a kind gift from John Albeck, UC Davis. The work was
in part funded by Deutsche Forschungsgemeinschaft (MO2783/2-1 to M.M.), Berlin
School of Integrative Oncology (to N.B. and M.M.), the German Cancer Consortium
DKTK (to N.B. and M.M.), the Federal Ministry of Education and Research BMBF
(StemNet 01EK1604B to N.B.; ColoSys 031L0081 to C.S. and N.B.) and the Berlin
Institute of Health (to N.B., C.S. and M.M.).
Author contributions
R.B., T.S., M.L., P.R., C.G.-T., S.S., D.K., N.M., B.F. and I.A.E.-S. conducted, analysed and
interpreted experiments; F.U., T.S. and B.K. performed bioinformatic analyses; M.M.,
N.B., C.S., P.R., T.M. and B.G.H. conceived, designed, interpreted experiments and/or
supervised parts of the study; M.M., N.B. and B.K. wrote the paper.
ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-10954-y
14 NATURE COMMUNICATIONS | (2019) 10:2919 | https://doi.org/10.1038/s41467-019-10954-y | www.nature.com/naturecommunications
Content courtesy of Springer Nature, terms of use apply. Rights reserved
Additional information
Supplementary Information accompanies this paper at https://doi.org/10.1038/s41467-
019-10954-y.
Competing interests: The authors declare no competing interests.
Reprints and permission information is available online at http://npg.nature.com/
reprintsandpermissions/
Peer review information: Nature Communications thanks Rony Seger and other
anonymous reviewer(s) for their contribution to the peer review of this work. Peer
reviewer reports are available.
Publishers note: Springer Nature remains neutral with regard to jurisdictional claims in
published maps and institutional afliations.
Open Access This article is licensed under a Creative Commons
Attribution 4.0 International License, which permits use, sharing,
adaptation, distribution and reproduction in any medium or format, as long as you give
appropriate credit to the original author(s) and the source, provide a link to the Creative
Commons license, and indicate if changes were made. The images or other third party
material in this article are included in the articles Creative Commons license, unless
indicated otherwise in a credit line to the material. If material is not included in the
articles Creative Commons license and your intended use is not permitted by statutory
regulation or exceeds the permitted use, you will need to obtain permission directly from
the copyright holder. To view a copy of this license, visit http://creativecommons.org/
licenses/by/4.0/.
© The Author(s) 2019
NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-10954-y ARTICLE
NATURE COMMUNICATIONS | (2019) 10:2919 | https://doi.org/10.1038/s41467-019-10954-y | www.nature.com/naturecommunications 15
Content courtesy of Springer Nature, terms of use apply. Rights reserved
1.
2.
3.
4.
5.
6.
Terms and Conditions
Springer Nature journal content, brought to you courtesy of Springer Nature Customer Service Center GmbH (“Springer Nature”).
Springer Nature supports a reasonable amount of sharing of research papers by authors, subscribers and authorised users (“Users”), for small-
scale personal, non-commercial use provided that all copyright, trade and service marks and other proprietary notices are maintained. By
accessing, sharing, receiving or otherwise using the Springer Nature journal content you agree to these terms of use (“Terms”). For these
purposes, Springer Nature considers academic use (by researchers and students) to be non-commercial.
These Terms are supplementary and will apply in addition to any applicable website terms and conditions, a relevant site licence or a personal
subscription. These Terms will prevail over any conflict or ambiguity with regards to the relevant terms, a site licence or a personal subscription
(to the extent of the conflict or ambiguity only). For Creative Commons-licensed articles, the terms of the Creative Commons license used will
apply.
We collect and use personal data to provide access to the Springer Nature journal content. We may also use these personal data internally within
ResearchGate and Springer Nature and as agreed share it, in an anonymised way, for purposes of tracking, analysis and reporting. We will not
otherwise disclose your personal data outside the ResearchGate or the Springer Nature group of companies unless we have your permission as
detailed in the Privacy Policy.
While Users may use the Springer Nature journal content for small scale, personal non-commercial use, it is important to note that Users may
not:
use such content for the purpose of providing other users with access on a regular or large scale basis or as a means to circumvent access
control;
use such content where to do so would be considered a criminal or statutory offence in any jurisdiction, or gives rise to civil liability, or is
otherwise unlawful;
falsely or misleadingly imply or suggest endorsement, approval , sponsorship, or association unless explicitly agreed to by Springer Nature in
writing;
use bots or other automated methods to access the content or redirect messages
override any security feature or exclusionary protocol; or
share the content in order to create substitute for Springer Nature products or services or a systematic database of Springer Nature journal
content.
In line with the restriction against commercial use, Springer Nature does not permit the creation of a product or service that creates revenue,
royalties, rent or income from our content or its inclusion as part of a paid for service or for other commercial gain. Springer Nature journal
content cannot be used for inter-library loans and librarians may not upload Springer Nature journal content on a large scale into their, or any
other, institutional repository.
These terms of use are reviewed regularly and may be amended at any time. Springer Nature is not obligated to publish any information or
content on this website and may remove it or features or functionality at our sole discretion, at any time with or without notice. Springer Nature
may revoke this licence to you at any time and remove access to any copies of the Springer Nature journal content which have been saved.
To the fullest extent permitted by law, Springer Nature makes no warranties, representations or guarantees to Users, either express or implied
with respect to the Springer nature journal content and all parties disclaim and waive any implied warranties or warranties imposed by law,
including merchantability or fitness for any particular purpose.
Please note that these rights do not automatically extend to content, data or other material published by Springer Nature that may be licensed
from third parties.
If you would like to use or distribute our Springer Nature journal content to a wider audience or on a regular basis or in any other manner not
expressly permitted by these Terms, please contact Springer Nature at
onlineservice@springernature.com
... In intestinal stem cells (ISCs), loss of APC function mutations drives intestinal adenomas by enhancing intracellular Wnt signaling [16]. Brandt et al. found that oncogenic KRAS, together with β-Catenin, favoured the expansion of crypt cells with high ERK activity [17]. At present, molecular targets of colon cancer comprise EGFR, VEGF, ERBB2, BRAF, KRAS, PD-1, CTLA-4, NTRK etc. ...
Article
Full-text available
Studies of both, microbiota and target therapy associated with gene mutations in colorectal cancer, (CRC) have attracted increasing attention. However, only a few of them analyzed the combined effects on CRC. we analyzed differences in intestinal microbiota of 44 colorectal cancer patients and 20 healthy controls (HC) using 16S rRNA gene sequencing of fecal samples. For 39 of the CRC patients, targeted Next Generation Sequencing (NGS) was carried out at formalin fixed paraffin embedded (FFPE) samples to identify somatic mutation profiles. Compared to the HC group, the microbial diversity of CRC patients was significantly lower. In the CRC group, we found a microbiome that was significantly enriched for strains of Bifidobacterium, Bacteroides, and Megasphaera whereas in the HC group the abundance of Collinsella, Faecalibacterium, and Agathobacter strains was higher. Among the mutations detected in the CRC group, the APC gene had the highest mutation rate (77%, 30/39). We found that the KRAS mutant type was closely associated with Faecalibacterium, Roseburia, Megamonas, Lachnoclostridium, and Harryflintia. Notably, Spearman correlation analysis showed that KRAS mutations were negatively correlated with the existence of Bifidobacterium and positively correlated with Faecalibacterium. By employing 16S rRNA gene sequencing, we identified more unique features of microbiota profiles in CRC patients. For the first time, our study showed that gene mutations could directly be linked to the microbiota composition of CRC patients. We hypothesize that the effect of a targeted colorectal cancer therapy is also closely related to the colorectal flora, however, this requires further investigation.
... Improved surface rendering tools such as Imaris 9.5 can be applied to serial confocal fluorescence images to enhance visualization and quantification capacity of cells expressing markers of proliferation and self-renewal in primary and secondary CRC tumor organoids ( Figure 6E). In addition, extensive analysis of cellular heterogeneity can be conducted on tumor organoids by assessing the abundance and distribution various markers, using either cytometry approaches (fluorescence, CyTOF) or single-cell genomics upon completing steps -32-34 (Brandt et al., 2019;Chen et al., 2018;Gonzalez-Exposito et al., 2019). ...
Article
Full-text available
Organoids can enable the study of solid tumors initiated from a single cancer stem cell (CSC) ex vivo. We describe a serial tumor organoid plating protocol using primary colorectal cancer (CRC) tissues as a rapid and cost-efficient approach to evaluate the impact of therapeutic interventions on CSC functions. We detail the isolation of primary colorectal CSCs, organoid embedding, serial passaging, and CSC-related analytical techniques. For complete details on the use and execution of this protocol, please refer to Masibag et al. (2021) and Bergin et al. (2021).
... Our description of the wiring of the RAS/MAPK pathway in neuroblastoma will support the design of clinical trials using combinatorial treatments to prevent or overcome therapy resistance. In addition, the framework described here could be used to analyse signalling in tumours of individual patients While it will be technically challenging to assess signalling network responses in tumour patients, ex vivo cultures-so-called avatars-could be an option [45,46]. We envision that learning features of robustness and vulnerability of tumours from signalling models on cell line panels might greatly reduce the required set of perturbations in those avatars that are sufficient to inform a model, and allow reliable stratification and prediction of treatment options. ...
Article
Full-text available
Very high risk neuroblastoma is characterised by increased MAPK signalling, and targeting MAPK signalling is a promising therapeutic strategy. We used a deeply characterised panel of neuroblastoma cell lines and found that the sensitivity to MEK inhibitors varied drastically between these cell lines. By generating quantitative perturbation data and mathematical modelling, we determined potential resistance mechanisms. We found that negative feedbacks within MAPK signalling and via the IGF receptor mediate re-activation of MAPK signalling upon treatment in resistant cell lines. By using cell-line specific models, we predict that combinations of MEK inhibitors with RAF or IGFR inhibitors can overcome resistance, and tested these predictions experimentally. In addition, phospho-proteomic profiling confirmed the cell-specific feedback effects and synergy of MEK and IGFR targeted treatment. Our study shows that a quantitative understanding of signalling and feedback mechanisms facilitated by models can help to develop and optimise therapeutic strategies. Our findings should be considered for the planning of future clinical trials introducing MEKi in the treatment of neuroblastoma.
Article
Wnt pathways are important for the modulation of tissue homeostasis, and their deregulation is linked to cancer development. Canonical Wnt signaling is hyperactivated in many human colorectal cancers due to genetic alterations of the negative Wnt regulator APC. However, the expression levels of Wnt-dependent targets vary between tumors, and the mechanisms of carcinogenesis concomitant with this Wnt signaling dosage have not been understood. In this study, we integrate whole-genome CRISPR/Cas9 screens with large-scale multi-omic data to delineate functional subtypes of cancer. We engineer APC loss-of-function mutations and thereby hyperactivate Wnt signaling in cells with low endogenous Wnt activity and find that the resulting engineered cells have an unfavorable metabolic equilibrium compared with cells which naturally acquired Wnt hyperactivation. We show that the dosage level of oncogenic Wnt hyperactivation impacts the metabolic equilibrium and the mitochondrial phenotype of a given cell type in a context-dependent manner. These findings illustrate the impact of context-dependent genetic interactions on cellular phenotypes of a central cancer driver mutation and expand our understanding of quantitative modulation of oncogenic signaling in tumorigenesis.
Article
Autophagy, an evolutionarily highly conserved cellular degradation process, plays the Janus role (either cytoprotective or death-promoting) in colorectal cancer, so the targeting of several key autophagic pathways with small-molecule compounds may be a new therapeutic strategy. In this review, we discuss autophagy-associated cell death pathways and key cytoprotective autophagy pathways in colorectal cancer. Moreover, we summarize a series of small-molecule compounds that have the potential to modulate autophagy-associated cell death or cytoprotective autophagy for therapeutic purposes. Taken together, these findings demonstrate the Janus role of autophagy in colorectal cancer, and shed new light on the exploitation of a growing number of small-molecule compounds to target autophagy in future cancer drug discovery. Teaser: Autophagy plays both cytoprotective and death-promoting roles in colorectal cancer, so the targeting of some autophagic pathways with pharmacological small-molecule compounds is a potential therapeutic strategy.
Article
Full-text available
Current treatment options for patients with advanced colorectal cancers (CRC) include anti-EGFR/HER1 therapy with the blocking antibody cetuximab. Although a subset of patients with KRAS wild-type disease initially respond to the treatment, resistance develops in almost all cases. Relapse has been associated with the production of the ligand heregulin (HRG) and/or compensatory signaling involving the receptor tyrosine kinases HER2 and HER3. Here we provide evidence that triple HER receptor blockade based on a newly developed bispecific EGFRxHER3-targeting antibody (scDb-Fc) together with the HER2 blocking antibody trastuzumab effectively inhibited HRG-induced HER receptor phosphorylation, downstream signaling, proliferation and stem cell expansion of DiFi and LIM1215 CRC cells. Comparative analyses revealed that the biological activity of scDb-Fc plus trastuzumab was sometimes even superior to that of the combination of the parental antibodies, with PI3K/Akt pathway inhibition correlating with improved therapeutic response and apoptosis induction as seen by single cell analysis. Importantly, growth suppression by triple HER targeting was recapitulated in primary KRAS wild-type patient-derived organoid (PDO) cultures exposed to HRG. Collectively, our results provide strong support for a pan-HER receptor blocking approach to combat anti-EGFR therapy resistance of KRAS wild-type CRC tumors mediated by the upregulation of HRG and/or HER2/HER3 signaling.
Article
Dysregulation of DNA methylation patterns and non‐coding RNA, including miRNAs, has been implicated in colon cancer, and these changes may occur early in the development of carcinoma. In this study, the role of epigenetics as early changes in colon tumorigenesis was examined through paired sample analysis of patient‐matched normal, adenoma and carcinoma samples. Global methylation was assessed by genomic 5‐methyl cytosine (5‐mC) and long interspersed nuclear element‐1 (LINE‐1) promoter methylation by pyrosequencing. KRAS mutations were also assessed by pyrosequencing. Expression of miRNA, specifically, two microRNA genes—miR‐200a and let‐7c—was analysed using RT‐qPCR. Differences in global methylation in adenomas were not observed, compared with normal tissue. However, LINE‐1 methylation was decreased in adenomas (p = .056) and carcinomas (p = .011) compared with normal tissue. Expressions of miRNA, miR‐200a and let‐7c were significantly higher in adenomas than normal tissues (p = .008 and p = .045 respectively). Thus the significant changes in LINE‐1 methylation and microRNA expression in precancerous lesions support an early role for epigenetic changes in the carcinogenic process. Epigenetic characteristics in adenomas may provide potential diagnostic and prognostic therapeutic targets early in cancer development at the adenoma stage.
Article
Full-text available
Cancer metastasis causes most cancer-related deaths, and modeling cancer invasion holds potential in drug discovery and companion diagnostics. Although 2D cocultures have been developed to study cancer invasion, it is challenging to recreate the 3D cancer invasion of an individual cancer patient. Here, we report an acoustic bioprinting technology that can precisely construct tumor microtissues for modeling cancer invasion in 3D. By using acoustic droplet technology, we can precisely encapsulate cancer associated fibroblasts (CAFs) derived from a colorectal cancer patient into gel droplets and print them into a 3D CAF microtissue. After depositing a tumor organoid derived from the same patient, our 3D bio-printed microtissue can be used to model cancer cell migration and invasion from the tumor organoid to the 3D CAF microtissue. We further used 3D bio-printed microtissues to investigate cancer invasion dynamics as well as their treatment response using time-lapse imaging. Thus, our acoustic 3D bioprinting technology can be widely used for establishing various microtissues for modeling cancer invasion and other diseases, highlighting its potential in personalized treatment.
Article
Cancer therapy often results in heterogeneous responses in different metastatic lesions in the same patient. Inter- and intratumor heterogeneity in signaling within various tumor compartments and its impact on therapy are not well characterized due to the limited sensitivity of single-cell proteomic approaches. To overcome this barrier, we applied single-cell mass cytometry with a customized 26-antibody panel to PTEN-deleted orthotopic prostate cancer xenograft models to measure the evolution of kinase activities in different tumor compartments during metastasis or drug treatment. Compared with primary tumors and circulating tumor cells (CTC), bone metastases, but not lung and liver metastases, exhibited elevated PI3K/mTOR signaling and overexpressed receptor tyrosine kinases (RTK) including c-MET protein. Suppression of c-MET impaired tumor growth in the bone. Intratumoral heterogeneity within tumor compartments also arose from highly proliferative EpCAM-high epithelial cells with increased PI3K and mTOR kinase activities coexisting with poorly proliferating EpCAM-low mesenchymal populations with reduced kinase activities; these findings were recapitulated in epithelial and mesenchymal CTC populations in patients with metastatic prostate and breast cancer. Increased kinase activity in EpCAM-high cells rendered them more sensitive to PI3K/mTOR inhibition, and drug-resistant EpCAM-low populations with reduced kinase activity emerged over time. Taken together, single-cell proteomics indicate that microenvironment- and cell state-dependent activation of kinase networks create heterogeneity and differential drug sensitivity among and within tumor populations across different sites, defining a new paradigm of drug responses to kinase inhibitors. Significance: Single-cell mass cytometry analyses provide insights into the differences in kinase activities across tumor compartments and cell states, which contribute to heterogeneous responses to targeted therapies.
Thesis
Introduction : Les dyschromies cutanées en mosaïque ont fait suspecter de longue date l’implication d’un mosaïcisme génétique sous-jacent. Ces évènements post-zygotiques sont cependant difficilement détectables par les techniques conventionnelles. Ainsi, les bases génétiques des dyschromies en mosaïque étaient restées mal connues. Matériel et méthodes : la cohorte M.U.S.T.A.R.D rassemble des échantillons d’ADN de biopsies cutanées de patients porteurs d’un mosaïcisme pigmentaire. Après une analyse phénotypique spécialisée, ces échantillons sont étudiés en séquençage à forte profondeur, d’exome (ES) en trio, ou ciblé. Les données sont analysées à l’aide d’un pipeline dédié, permettant la détection de variations ponctuelles en mosaïque (mSNV) mais également de diverses anomalies chromosomiques en mosaïque. Résultats : De 2013 à 2019, 101 patients ont été inclus. Un ES a été réalisé pour 56 patients, identifiant un mSNV chez 12 patients, dans 7 gènes dont 4 nouveaux (RHOA, DOCK1, GNA13, TFE3), et une anomalie chromosomique chez 17 patients. Une étude ciblée de ces gènes chez 40 autres patients était positive pour 17, soit un rendement diagnostique global à 55% (46/84). Conclusion : Ce travail illustre l’importance d’une approche bioinformatique polyvalente, combinée à une expertise clinique, pour la détermination des causes génétiques des dyschromies cutanées en mosaïque. Il a également mis en évidence le rôle de la voie de signalisation dépendant des Rho GTPases, faisant progresser notre compréhension de la physiopathologie des dyschromies en mosaïque, une étape nécessaire à l’amélioration de la prise en charge de ces patients porteurs de maladies rares et complexes.
Preprint
Full-text available
Single-cell transcriptomics allows researchers to investigate complex communities of heterogeneous cells. These methods can be applied to stem cells and their descendants in order to chart the progression from multipotent progenitors to fully differentiated cells. While a number of statistical and computational methods have been proposed for analyzing cell lineages, the problem of accurately characterizing multiple branching lineages remains difficult to solve. Here, we introduce a novel method, Slingshot, for inferring multiple developmental lineages from single-cell gene expression data. Slingshot is a uniquely robust and flexible tool for inferring developmental lineages and ordering cells to reflect continuous, branching processes.
Article
Full-text available
HRAS, NRAS, and KRAS isoforms are almost identical proteins that are ubiquitously expressed and activate a common set of effectors. In vivo studies have revealed that they are not biologically redundant; however, the isoform specificity of Ras signaling remains poorly understood. Using a novel panel of isogenic SW48 cell lines endogenously expressing wild-type or G12V-mutated activated Ras isoforms, we have performed a detailed characterization of endogenous isoform-specific mutant Ras signaling. We find that despite displaying significant Ras activation, the downstream outputs of oncogenic Ras mutants are minimal in the absence of growth factor inputs. The lack of mutant KRAS-induced effector activation observed in SW48 cells appears to be representative of a broad panel of colon cancer cell lines harboring mutant KRAS. For MAP kinase pathway activation in KRAS-mutant cells, the requirement for coincident growth factor stimulation occurs at an early point in the Raf activation cycle. Finally, we find that Ras isoform-specific signaling was highly context dependent and did not conform to the dogma derived from ectopic expression studies.
Article
Full-text available
Motivation: Intracellular signalling is realised by complex signalling networks which are almost impossible to understand without network models, especially if feedbacks are involved. Modular Response Analysis (MRA) is a convenient modelling method to study signalling networks in various contexts. Results: We developed the software package STASNet that provides an augmented and extended version of MRA suited to model signalling networks from incomplete perturbation schemes and multi-perturbation data. Using data from the DREAM challenge, we show that predictions from STASNet models are among the top-performing methods. We applied the method to study the effect of SHP2, a protein that has been implicated in resistance to targeted therapy in colon cancer, using a novel data set from the colon cancer cell line Widr and a SHP2-depleted derivative. We find that SHP2 is required for MAPK signalling, whereas AKT signalling only partially depends on SHP2. Availability: An R-package is available at https://github.com/molsysbio/STASNet. Supplementary information: Supplementary data are available at Bioinformatics online.
Article
Full-text available
Background: Single-cell transcriptomics allows researchers to investigate complex communities of heterogeneous cells. It can be applied to stem cells and their descendants in order to chart the progression from multipotent progenitors to fully differentiated cells. While a variety of statistical and computational methods have been proposed for inferring cell lineages, the problem of accurately characterizing multiple branching lineages remains difficult to solve. Results: We introduce Slingshot, a novel method for inferring cell lineages and pseudotimes from single-cell gene expression data. In previously published datasets, Slingshot correctly identifies the biological signal for one to three branching trajectories. Additionally, our simulation study shows that Slingshot infers more accurate pseudotimes than other leading methods. Conclusions: Slingshot is a uniquely robust and flexible tool which combines the highly stable techniques necessary for noisy single-cell data with the ability to identify multiple trajectories. Accurate lineage inference is a critical step in the identification of dynamic temporal gene expression.
Article
Full-text available
The advent of mass cytometry increased the number of parameters measured at the single-cell level while decreasing the extent of crosstalk between channels relative to dye-based flow cytometry. Although reduced, spillover still exists in mass cytometry data, and minimizing its effect requires considerable expert knowledge and substantial experimental effort. Here, we describe a novel bead-based compensation workflow and R-based software that estimates and corrects for interference between channels. We performed an in-depth characterization of the spillover properties in mass cytometry, including limitations defined by the linear range of the mass cytometer and the reproducibility of the spillover over time and across machines. We demonstrated the utility of our method in suspension and imaging mass cytometry. To conclude, our approach greatly simplifies the development of new antibody panels, increases flexibility for antibody-metal pairing, opens the way to using less pure isotopes, and improves overall data quality, thereby reducing the risk of reporting cell phenotype artifacts. Signal spillover exists in mass cytometry and complicates the development of antibody panels and the interpretation of the data. Here, we characterize this spillover and its effects on the data. Further, we present an approach to estimate and correct for it, which was implemented in a newly developed R package called CATALYST. We show how this method can be used to correct for channel crosstalk in suspension and imaging mass cytometry data throughout the linear range of the instrument.
Article
Full-text available
Oncogenic mutations in BRAF are believed to initiate serrated colorectal cancers; however, the mechanisms of BRAF-driven colon cancer are unclear. We find that oncogenic BRAF paradoxically suppresses stem cell renewal and instead promotes differentiation. Correspondingly, tumor formation is inefficient in BRAF-driven mouse models of colon cancer. By reducing levels of differentiation via genetic manipulation of either of two distinct differentiation-promoting factors (Smad4 or Cdx2), stem cell activity is restored in BRAFV600E intestines, and the oncogenic capacity of BRAFV600E is amplified. In human patients, we observe that reduced levels of differentiation in normal tissue is associated with increased susceptibility to serrated colon tumors. Together, these findings help resolve the conditions necessary for BRAF-driven colon cancer initiation. Additionally, our results predict that genetic and/or environmental factors that reduce tissue differentiation will increase susceptibility to serrated colon cancer. These findings offer an opportunity to identify susceptible individuals by assessing their tissue-differentiation status.
Article
Full-text available
Ulixertinib (BVD-523) is an ERK1/2 kinase inhibitor with potent preclinical activity in BRAF- and RAS-mutant cell lines. In this multicenter phase I trial (NCT01781429), 135 patients were enrolled to an accelerated 3 + 3 dose-escalation cohort and six distinct dose-expansion cohorts. Dose escalation included 27 patients, dosed from 10 to 900 mg twice daily and established the recommended phase II dose (RP2D) of 600 mg twice daily. Ulixertinib exposure was dose proportional to the RP2D, which provided near-complete inhibition of ERK activity in whole blood. In the 108-patient expansion cohort, 32% of patients required dose reduction. The most common treatment-related adverse events were diarrhea (48%), fatigue (42%), nausea (41%), and dermatitis acneiform (31%). Partial responses were seen in 3 of 18 (17%) patients dosed at or above maximum tolerated dose and in 11 of 81 (14%) evaluable patients in dose expansion. Responses occurred in patients with NRAS-, BRAF V600-, and non-V600 BRAF-mutant solid tumors.SIGNIFICANCE: Here, we describe the first-in-human dose-escalation study of an ERK1/2 inhibitor for the treatment of patients with advanced solid tumors. Ulixertinib has an acceptable safety profile with favorable pharmacokinetics and has shown early evidence of clinical activity in NRAS- and BRAF V600- and non-V600-mutant solid-tumor malignancies. Cancer Discov; 8(2); 1-12. ©2017 AACR.
Article
Full-text available
Epidermal homeostasis depends on a balance between stem cell renewal and terminal differentiation. The transition between the two cell states, termed commitment, is poorly understood. Here we characterise commitment by integrating transcriptomic and proteomic data from disaggregated primary human keratinocytes held in suspension to induce differentiation. Cell detachment induces several protein phosphatases, five of which - DUSP6, PPTC7, PTPN1, PTPN13 and PPP3CA - promote differentiation by negatively regulating ERK MAPK and positively regulating AP1 transcription factors. Conversely, DUSP10 expression antagonises commitment. The phosphatases form a dynamic network of transient positive and negative interactions that change over time, with DUSP6 predominating at commitment. Boolean network modelling identifies a mandatory switch between two stable states (stem and differentiated) via an unstable (committed) state. Phosphatase expression is also spatially regulated in vivo and in vitro. We conclude that an auto-regulatory phosphatase network maintains epidermal homeostasis by controlling the onset and duration of commitment.
Article
Full-text available
Induced pluripotent stem cells (iPSCs) are similar to embryonic stem cells and can be generated from somatic cells. We have generated episomal plasmid-based and integration-free iPSCs (E-iPSCs) from human fetal foreskin fibroblast cells (HFF1). We used an E-iPSC-line to model hepatogenesis in vitro. The HLCs were characterized biochemically, i.e. glycogen storage, ICG uptake and release, UREA and bile acid production, as well as CYP3A4 activity. Ultra-structure analysis by electron microscopy revealed the presence of lipid and glycogen storage, tight junctions and bile canaliculi- all typical features of hepatocytes. Furthermore, the transcriptome of undifferentiated E-iPSC, DE, HE and HLCs were compared to that of fetal liver and primary human hepatocytes (PHH). K-means clustering identified 100 clusters which include developmental stage-specific groups of genes, e.g. OCT4 expression at the undifferentiated stage, SOX17 marking the DE stage, DLK and HNF6 the HE stage, HNF4α and Albumin is specific to HLCs, fetal liver and adult liver (PHH) stage. We use E-iPSCs for modeling gene regulatory networks associated with human hepatogenesis and gastrulation in general.
Article
Intestinal homeostasis depends on a slowly proliferating stem cell compartment in crypt cells, followed by rapid proliferation of committed progenitor cells in the transit amplifying (TA) compartment. The balance between proliferation and differentiation in intestinal stem cells (ISCs) is regulated by Wnt/β-catenin signaling, although the mechanism remains unclear. We previously targeted PORCN, an enzyme essential for all Wnt secretion, and demonstrated that stromal production of Wnts was required for intestinal homeostasis. Here, a PORCN inhibitor was used to acutely suppress Wnt signaling. Unexpectedly, the treatment induced an initial burst of proliferation in the stem cell compartment of the small intestine, due to conversion of ISCs into TA cells with a loss of intrinsic ISC self-renewal. This process involved MAPK pathway activation, as the proliferating cells in the base of the intestinal crypt contained phosphorylated ERK1/2, and a MEK inhibitor attenuated the proliferation of ISCs and their differentiation into TA cells. These findings suggest a role for Wnt signaling in suppressing the MAPK pathway at the crypt base to maintain a pool of ISCs. The interaction between Wnt and MAPK pathways in vivo has potential therapeutic applications in cancer and regenerative medicine.