ARTICLE DOI: 10.1038/s41467-018-04527-8 OPEN Composite regulation of ERK activity dynamics underlying tumour-speciﬁc traits in the intestine 1,2 3 4 5 6 2,7 Yu Muta , Yoshihisa Fujita , Kenta Sumiyama , Atsuro Sakurai , M. Mark Taketo , Tsutomu Chiba , 2 8,9 1,5 5 Hiroshi Seno , Kazuhiro Aoki , Michiyuki Matsuda & Masamichi Imajo Acting downstream of many growth factors, extracellular signal-regulated kinase (ERK) plays a pivotal role in regulating cell proliferation and tumorigenesis, where its spatiotemporal dynamics, as well as its strength, determine cellular responses. Here, we uncover the ERK activity dynamics in intestinal epithelial cells (IECs) and their association with tumour characteristics. Intravital imaging identiﬁes two distinct modes of ERK activity, sustained and pulse-like activity, in IECs. The sustained and pulse-like activities depend on ErbB2 and EGFR, respectively. Notably, activation of Wnt signalling, the earliest event in intestinal tumor- igenesis, augments EGFR signalling and increases the frequency of ERK activity pulses through controlling the expression of EGFR and its regulators, rendering IECs sensitive to EGFR inhibition. Furthermore, the increased pulse frequency is correlated with increased cell proliferation. Thus, ERK activity dynamics are deﬁned by composite inputs from EGFR and ErbB2 signalling in IECs and their alterations might underlie tumour-speciﬁc sensitivity to pharmacological EGFR inhibition. 1 2 Department of Pathology and Biology of Diseases, Graduate School of Medicine, Kyoto University, Kyoto 606-8051, Japan. Department of Gastroenterology and Hepatology, Graduate School of Medicine, Kyoto University, Kyoto 606-8507, Japan. Department of Systems Science, Graduate School of Informatics, Kyoto University, Kyoto 606-8501, Japan. Laboratory for Mouse Genetic Engineering, Quantitative Biology Center, RIKEN, Osaka 5 6 565-0874, Japan. Laboratory of Bioimaging and Cell Signaling, Graduate School of Biostudies, Kyoto University, Kyoto 606-8501, Japan. Division of Experimental Therapeutics, Graduate School of Medicine, Kyoto University, Kyoto 606-8501, Japan. Kansai Electric Power Hospital, Osaka 553-0003, Japan. Division of Quantitative Biology, Okazaki Institute for Integrative Bioscience, National Institute for Basic Biology, National Institutes of Natural Sciences, Okazaki, Aichi 444-8787, Japan. Department of Basic Biology, Faculty of Life Science, SOKENDAI (Graduate University for Advanced Studies), Myodaiji, Okazaki, Aichi 444-8787, Japan. Correspondence and requests for materials should be addressed to M.I. (email: firstname.lastname@example.org) NATURE COMMUNICATIONS (2018) 9:2174 DOI: 10.1038/s41467-018-04527-8 www.nature.com/naturecommunications 1 | | | 1234567890():,; ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-04527-8 he extracellular signal-regulated kinase (ERK) signalling pathway and its interaction with other pathways can be readily pathway regulates a variety of biological processes analyzed. Here, by taking the full advantage of the organoid Tincluding cell proliferation, survival, differentiation, and culture method and a highly sensitive biosensor for ERK activity, 1, 2 tumorigenesis . Since ERK activation promotes proliferation of we uncover the ERK activity dynamics in IECs. We demonstrate many types of cells, its deregulated/constitutive activation is often the presence of two distinct modes of ERK activity, sustained, observed in various cancers. Among many growth factor recep- constant activity and pulse-like activity, both in vivo and in vitro. tors, epidermal growth factor receptor (EGFR) plays a pivotal role Our analyses show that the two modes of ERK activity are gen- in activating ERK in normal and cancerous epithelia , therefore, erated by different EGFR family receptors. Moreover, we reveal EGFR–ERK signalling has been of particular interest in cancer that Wnt signalling activation alters the ERK signalling dynamics, 4, 5 biology . In the classical view, EGF stimulation simply triggers which underlies the enhanced responsiveness of tumour cells to 1, 6 transient and short-lived ERK activation . However, recent EGFR inhibition. studies using a highly sensitive biosensor for ERK activity have revealed that EGF signalling can generate complex spatio- Results 8–10 temporal ERK activity at the single cell level . For instance, In vivo imaging of ERK activity in the mouse small intestine. certain types of cultured cells show considerable heterogeneity in To reveal the ERK activity dynamics in the intestinal epithelium, ERK activity due to spontaneous ERK activation pulses and its we used transgenic mice ubiquitously expressing a highly sensi- lateral propagation to adjacent cells, both of which were asso- tive Förster resonance energy transfer (FRET) biosensor for ERK 8, 10 ciated with cell proliferation . Similarly, propagation of ERK activity (EKAREV-NLS) (Fig. 1a) . The small intestine of activity and its correlation with cell proliferation were also EKAREV-NLS mice was observed under an inverted two-photon observed in the mouse skin . Notably ERK activity dynamics as excitation microscope (Fig. 1b). By this approach, ERK activity well as its overall strength can be a critical determinant of cell represented by the FRET/CFP ratio could be live-imaged at a 8, 9 proliferation . Moreover, difference in ERK activity dynamics single-cell resolution in areas ranging from the crypt bottom to leads to different outputs in some biological processes. For the villus (Supplementary Fig. 1a). To validate the speciﬁcity of example, in PC12 cells, treatment with NGF or FGF induces the biosensor, we intravenously administered a known activator 12, 13 prolonged ERK activation and neuronal differentiation , of the ERK pathway, 12-O-tetradecanoylphorbol-13-acetate whereas EGF treatment generates only transient, pulse-like ERK (TPA), or a MEK inhibitor, PD0325901. As expected, TPA activation without inducing the differentiation . Despite its increased the FRET/CFP ratio, whereas the MEK inhibitor obvious importance, however, how ERK activity dynamics decreased it (Fig. 1c–f), indicating that EKAREV-NLS faithfully are regulated and how they affect the physiological processes monitors ERK activity. Thus, hereafter, we use the FRET/CFP remains unknown. ratio as an index of ERK activity. In vivo time-lapse imaging then The intestinal epithelium is one of the representative tissues in revealed that IECs at the crypt exhibited sporadic ERK activity which EGFR–ERK signalling regulates both normal homoeostasis pulses (Fig. 1g–j and Supplementary Movie 1). The ERK activity and tumorigenesis . In this tissue, actively dividing stem cells pulses ﬁred spontaneously in each cell (Fig. 1g, h and Supple- expressing a marker gene, LGR5, support the rapid and constant mentary Fig. 1b) or in some cases were propagated from adjacent renewal of the entire epithelium . In mice, depletion of either cells within single crypt units (Fig. 1i, j and Supplementary EGFR or three of its ligands, EGF, amphiregulin, and TGF-α, Fig. 1c, d). The duration of single ERK activity pulses and the impairs proliferation of intestinal stem cells (ISCs) and progenitor velocity of propagation in IECs were comparable to those in 16, 17 cells . Conversely, excessive activation of EGFR signalling by cultured cells . These results demonstrate two modes of ERK depleting its negative regulator, Lrig1, induces ISC expansion and activity in the intestinal epithelium: the sustained, basal activity 18, 19 tumour development . In line with these ﬁndings, EGFR sig- that was evident by the decrease in ERK activity after MEK nalling has also been implicated in human colorectal cancer inhibitor treatment and the pulse-like activity that may arise (CRC). Typically, sporadic CRCs develop through the spontaneously in each cell or be propagated from adjacent cells. adenoma–carcinoma sequence, an archetypal model of multi-step carcinogenesis . In this model, mutations in the adenomatous polyposis coli (APC) gene have been regarded as the earliest and ERK activity dynamics in intestinal organoids. To facilitate the the rate-limiting events of tumour initiation. Following APC analysis of ERK activity dynamics in IECs, we generated intestinal mutations, sequential accumulation of other genetic mutations organoids from EKAREV-NLS mice and cultured them in med- including KRAS, BRAF, PIK3CA, SMAD4, and TP53 mutations ium containing EGF, Noggin, and R-spondin 1 (ENR medium) 20–22 transforms the tissue to malignant tumours . In addition, (Fig. 2a, b). The growth rate and morphology of the EKAREV- EGFR overexpression is also observed in human CRCs, and is NLS organoids were indistinguishable from those of wild-type 23–26 associated with poor prognosis . Pharmacological inhibition of organoids. As observed in vivo, TPA increased the FRET/CFP EGFR signalling has been shown to be effective against these ratio (Fig. 2c–e), whereas PD0325901, a potent MEK inhibitor, cancers . However, mutations in KRAS or BRAF desensitize decreased it (Fig. 2f–h). The FRET/CFP ratios correlated well CRCs to EGFR inhibition , suggesting that RAS-RAF-ERK sig- with the fraction of phosphorylated ERK (Supplementary Fig. 2a). nalling mediates the tumour-promoting activity of EGFR signal- In addition, we also observed the expected effects of a BRAF ling. Collectively, these reports suggest that EGFR–ERK signalling inhibitor (BRAFi) on basal ERK activity (Fig. 2i): high con- is a key driver of stem/progenitor cell proliferation and tumour centration of BRAFi decreased the basal ERK activity, whereas progression in the intestinal epithelium in both mice and humans. low concentration of BRAFi contrastingly increased it, a phe- However, EGFR–ERK signalling dynamics and their regulatory nomenon known as paradoxical activation . Collectively, these mechanisms remain unknown due to technical difﬁculties. results demonstrate that the EKAREV-NLS biosensor can Recent advances in detecting ERK activity using ﬂuorescent monitor ERK activity in intestinal organoids as sensitively as biosensors and culturing primary intestinal epithelial cells (IECs) in vivo. as organoids have paved the way to visualize EGFR–ERK sig- We next examined whether the spatiotemporal dynamics of nalling dynamics in this tissue. Since intestinal organoids com- ERK activity in vivo are recapitulated in intestinal organoids. prise IECs without any genetic mutations and can be cultured in Time-lapse imaging of intestinal organoids revealed that most serum-free media, dynamic regulation of the EGFR–ERK cells exhibited ERK activity pulses as observed in vivo (Fig. 3a–c, 2 NATURE COMMUNICATIONS (2018) 9:2174 DOI: 10.1038/s41467-018-04527-8 www.nature.com/naturecommunications | | | NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-04527-8 ARTICLE a b c TPA 475 nm CFP 433 nm YFP 1.6 ERK substrate Binding domain Pre Post Linker 433 nm FRET 0.9 530 nm d e f MEK inhibitor *** 1.2 *** 1.2 1.8 1.0 Pre Post 1.0 0.8 0.9 0.8 Pre Post Pre Post TPA MEK inhibitor g h 1.6 Cell 1 Cell 1 Cell 1 Cell 1 Cell 1 Cell 2 Cell 2 Cell 2 Cell 2 Cell 2 Cell 3 1.4 Cell 3 Cell 3 Cell 3 Cell 3 0 min 33 min 46 min 57 min 1.2 0.8 1.4 FRET/CFP 1.0 015 30 45 60 Elapsed time (min) i j Cell 1 Cell 1 Cell 2 Cell 1 Cell 1 Cell 2 Cell 2 Cell 1 1.4 Cell 2 Cell 2 Cell 3 Cell 3 Cell 3 Cell 3 Cell 3 1.2 0 min 24 min 39 min 57 min 0.8 1.4 FRET/CFP 1.0 015 30 45 60 Elapsed time (min) Fig. 1 In vivo imaging of ERK activity dynamics in the mouse small intestine. a Schematic representation of the structure of the FRET biosensor for ERK activity, EKAREV-NLS, and its mechanism of action. Upon activation, ERK phosphorylates the substrate sequence in the biosensor. The WW domain speciﬁcally binds to the phosphorylated substrate, which brings CFP and YFP into close proximity. In this “closed” conformation state, excitation energy absorbed by CFP is transferred to YFP without radiation, thereby enabling YFP to emit ﬂuorescence. The biosensor returns from the “closed” to the original “open” conformation by phosphatase-dependent dephosphorylation of the substrate sequence. b Experimental setting of in vivo imaging of the small intestine. The mouse small intestine was exteriorized, ﬁxed on the microscope stage, and observed with an inverted two-photon excitation microscope under inhalation anaesthesia. c The representative FRET/CFP images of the small intestine of EKAREV-NLS mice before and after administration of 0.1 mg −1 kg body weight of TPA. d Bee swarm plots showing the ERK activity (FRET/CFP ratio) in each cell before and after the TPA (n = 50 cells pooled from −1 three crypts). e The representative FRET/CFP images of the small intestine of EKAREV-NLS mice before and after administration of 5 mg kg body weight of a MEK inhibitor (PD0325902). f Bee swarm plots showing the FRET/CFP ratios in each cell before and after MEK inhibitor treatment (n = 50 cells pooled from three crypts). g–j In vivo time-lapse imaging of intestinal crypts of EKAREV-NLS mice. The representative images (g, i) and quantiﬁed data from three selected cells (cells 1–3) (h, j) are shown. Note that the ERK activity pulses are spontaneously generated (g, h), or propagated from adjacent cells (i, j). The original time course data of ERK activity used for h and j before smoothing by the moving average are shown in Supplementary Fig. 1b, c. Scale bars, 50 µm. Red lines represent mean. Error bars represent s.e.m. Mann–Whitney U-tests were used for comparison (d, f). *P < 0.05, **P < 0.001, ***P < 0.0001 NATURE COMMUNICATIONS (2018) 9:2174 DOI: 10.1038/s41467-018-04527-8 www.nature.com/naturecommunications 3 | | | Normalized ERK activity (FRET/CFP) FRET/CFP FRET/CFP ERK activity ERK activity Normalized ERK activity (FRET/CFP) (FRET/CFP) (FRET/CFP) ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-04527-8 a b c TPA ENR medium (EGF, Noggin, R-spondin 1) Treatment 1.4 Establish 0 min 10 min organoids ~4 days 0.3 1.8 0.7 FRET/CFP Passage Observation with 2PM 20 min 30 min d e f MEK inhibitor *** 1.3 1.6 1.2 1.4 1.5 TPA 1.2 0 min 7 min 1.1 1.0 1.0 0.8 0.8 0.9 Pre Post –5 0 10 20 30 TPA 20 min 30 min Elapsed time (min) g h i MEK inhibitor *** 1.1 1.4 1.2 BRAF inhibitor 0.1 μM 1.0 1.2 1.1 1 μM 0.9 1.0 1.0 10 μM 0.8 0.8 0.9 0.8 0.6 –5 0 10 20 30 Pre Post –5 010 20 30 Elapsed time (min) Elapsed time (min) MEK inhibitor Fig. 2 Characterization of mouse intestinal organoids expressing the FRET biosensor for ERK activity. a Experimental setting for live imaging of intestinal organoids. Organoids were established from the small intestine of EKAREV-NLS mice, and cultured in media containing EGF, Noggin, and R-spondin1. Imaging was performed within 4 days of the last passage. b A representative FRET/CFP ratio image of an EKAREV-NLS organoid cultured under normal conditions. c–h Time-lapse imaging of ERK activity in the EKAREV-NLS organoids. Organoids were treated with 1 μM TPA (c–e) or 200 nM MEK inhibitor (PD0325901) (f–h) at time point 0. c, f Representative time-lapse images of organoids treated with TPA (c) or a MEK inhibitor (f) are shown. d, g Time courses of the average FRET/CFP values in the organoids (n = 90 (d) and 55 (g) cells). The FRET/CFP values in individual cells were normalized to the mean values before the treatment. e, h Bee swarm plots of the FRET/CFP values in each cell before and after the treatment (n = 90 (e) and 66 (h) cells). The FRET/CFP values in individual cells were normalized to the mean values before the treatment. i Time courses of the average ERK activity (FRET/CFP ratio) in EKAREV-NLS organoids treated with 10 μM, 1 μM, or 100 nM of a BRAF inhibitor (SB590885) (n = 30, 35, and 31 cells, respectively). Scale bars, 50 µm. Red lines represent mean. Error bars represent s.e.m. Mann–Whitney U-tests were used for comparison (e, h). *P < 0.05, **P < 0.001, ***P < 0.0001 Supplementary Fig. 2b, c, and Supplementary Movie 2). Temporal propagation of ERK activity pulses (Fig. 3g, h, Supplementary autocorrelation analysis of ERK activity did not show any Fig. 2d, and Supplementary Movie 3). Propagation of ERK apparent periodicity (Fig. 3d), suggesting the stochastic nature of activity pulses was diminished either by an EGFR inhibitor, the pulses. Although ERK activity pulses occurred spontaneously PD153035, or by a broad-spectrum matrix metalloproteinase in individual cells, propagation of the pulses was not observed in inhibitor, marimastat, which should suppress shedding of EGFR 37, 38 intestinal organoids cultured in the ENR medium, which contains ligands (Fig. 3i, j). These results suggest that, in intestinal high concentrations of EGF. Since propagation of ERK activity organoids, propagation of ERK activity pulses requires shedding 8, 32, 33 was mediated by shedding of EGFR ligands in cultured cells , of EGFR ligands and the resulting EGFR activation. Thus, the we reasoned that EGF supplemented in the ENR medium ERK activity dynamics observed in the intestinal epithelium were decreased the sensitivity of cells to the lower amount of EGFR successfully recapitulated in intestinal organoids. ligands secreted by cells through negative-feedback mechan- 34–36 isms . To test this idea, the organoids were cultured in the absence of EGF (in the NR medium). No difference was observed ErbB2 and EGFR generate distinct modes of ERK activity.We in the basal ERK activity (Fig. 3e), probably due to adaptation. As next investigated the upstream signalling pathways that con- expected, organoids cultured in NR, but not in ENR, exhibited tribute to ERK activity dynamics. Because EGFR and ErbB2 (a.k.a ERK activation upon EGF addition (Fig. 3f). Furthermore, human EGFR2, HER2) are often overexpressed in gastrointestinal organoids cultured in NR occasionally exhibited cell-to-cell cancers and agents targeting these receptors have been used in 4 NATURE COMMUNICATIONS (2018) 9:2174 DOI: 10.1038/s41467-018-04527-8 www.nature.com/naturecommunications | | | Normalized ERK activity Normalized ERK activity (FRET/CFP) (FRET/CFP) Normalized ERK activity Normalized ERK activity (FRET/CFP) (FRET/CFP) Normalized ERK activity FRET/CFP FRET/CFP (FRET/CFP) NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-04527-8 ARTICLE FRET/CFP a b 1.0 1.5 0 min 4 min 15 min 27 min 0.3 1.8 FRET/CFP 0 45 90 Elapsed time (min) c d e n.s. Cell 1 1.8 1.2 1.4 Cell 1 Cell 2 Cell 2 Cell 3 1.6 Cell 3 0.6 1.2 1.4 1.0 1.2 1 –0.6 0.8 030 60 90 015 30 45 60 ENR NR Elapsed time (min) Δ time (min) FRET/CFP 0.3 1.7 NR 1.3 Cell 3 ENR NR Cell 2 1.2 EGF Cell 1 1.1 15 min 21 min 27 min 33 min FRET/CFP 0.4 1.7 NR + EGFRi 0.9 –5 0 10 20 30 Elapsed time (min) NR condition 1.8 Cell 1 0 min 30 min 60 min 90 min Cell 2 1.6 Cell 3 FRET/CFP 0.4 1.7 NR + MMPi 1.4 1.2 015 30 45 Elapsed time (min) 0 min 30 min 60 min 90 min Fig. 3 Spontaneous pulse-like ERK activation and propagation of ERK activity in intestinal organoids. a–d Time-lapse imaging of the EKAREV-NLS organoids was performed for 90 min at 1.5 min intervals (n = 71 cells). a Representative time-lapse images of ERK activity (FRET/CFP ratio) in an organoid showing pulse-like ERK activation. b Heat map showing the time course of ERK activity (FRET/CFP ratio) in each cell. c Time courses of ERK activity in three representative cells. The original data before smoothing by the moving average are shown in Supplementary Fig. 2c. d Temporal autocorrelation coefﬁcients of ERK activity in the three cells shown in c. e, f Time-lapse imaging of the EKAREV-NLS organoids cultured under EGF-starved conditions for 24 h. e Bee swarm plots of ERK activity in organoids cultured under the normal condition (ENR) or EGF-starved condition (NR) (n = 193 and 148 cells pooled from three organoids). f Time course of the average ERK activity in the control or EGF-starved organoids after stimulation with EGF. The organoids −1 were cultured under the normal (ENR) or EGF-starved condition (NR), and then stimulated with 50 ng ml of EGF at time point 0 (ENR: n = 129, NR: n = 59 cells). g Representative time-lapse images of an organoid cultured under the NR condition showing propagation of ERK activity. ERK activity (FRET/CFP ratio) is shown in golden pseudo colour mode. h Time courses of ERK activity in the three representative cells marked in g. i, j Representative time-lapse images of organoids treated with an EGFR inhibitor (PD153035, 1 μM) (i) or an MMP inhibitor (marimastat, 100 μM) (j). Organoids were cultured in NR media for 24 h, and subsequently treated with either inhibitor. ERK activity is shown in the golden pseudo colour mode. Scale bars, 50 µm. Red lines represent mean. Error bars represent s.e.m. Mann–Whitney U-test was used for comparison (e). *P < 0.05, **P < 0.001, ***P < 0.0001, n.s., not signiﬁcant NATURE COMMUNICATIONS (2018) 9:2174 DOI: 10.1038/s41467-018-04527-8 www.nature.com/naturecommunications 5 | | | ERK activity Normalized ERK activity ERK activity (FRET/CFP) (FRET/CFP) (FRET/CFP) Autocorrelation coefficient Cell no. Normalized ERK activity (FRET/CFP) ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-04527-8 a b ENR condition ENR condition n.s. 1.6 *** 1.1 Inhibitor EGFRi 1.4 ErbB2i EGFRi + ErbB2i 1.0 1.2 1.0 0.9 0.8 0.6 0.8 Pre Post Pre Post –5 0 10 20 30 EGFRi ErbB2i Elapsed time (min) c d ENR condition NR condition n.s. *** 1.3 1.4 *** EGFRi ErbB2i 1.2 1.2 EGF 1.1 1.0 1.0 0.8 0.9 –5 0 10 20 30 Control dnEGFR dnErbB2 Elapsed time (min) e f ENR condition ENR condition *** ERK-pulse (+) ERK-pulse (−) n.s. 100% *** 75% 50% 25% 0% 0 EGFRi −− + + EGFRi −− + + –− ++ ErbB2i −− ++ ErbB2i g h ENR condition ENR condition n.s. ** * ** *** 0 0 01 0.1 10 01 0.1 10 BRAF inhibitor (μM) BRAF inhibitor (μM) 3, 27, 39 cancer therapy , we examined the effect of an EGFR inhi- conﬁrm the effects of EGFR and ErbB2 inhibition on basal ERK bitor, PD153035, and an ErbB2 inhibitor, CP-724714, on ERK activity, dominant negative forms of EGFR (dnEGFR) and ErbB2 activity dynamics. Basal ERK activity was markedly decreased by (dnErbB2), both of which are truncated mutants lacking their 40, 41 CP-724714 (ErbB2i), but not by PD153035 (EGFRi) (Fig. 4 a, b). respective intracellular domains , were introduced into the As expected, EGFRi but not ErbB2i abrogated EGF-induced ERK intestinal organoids by using lentiviruses. Expression of dnErbB2, activation (Fig. 4c), validating the speciﬁcity of the inhibitors. To but not that of dnEGFR, signiﬁcantly decreased the basal ERK 6 NATURE COMMUNICATIONS (2018) 9:2174 DOI: 10.1038/s41467-018-04527-8 www.nature.com/naturecommunications | | | Normalized ERK activity Normalized ERK activity (FRET/CFP) (FRET/CFP) Mean pulse duration (min) Normalized ERK activity Normalized ERK activity (FRET/CFP) (FRET/CFP) Frequency (/cell/day) Frequency (/cell/day) NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-04527-8 ARTICLE Fig. 4 EGFR and ErbB2 generate two distinct modes of ERK activity in intestinal organoids. a, b EKAREV-NLS organoids were treated with EGFRi (PD153035) (1 μM), and/or ErbB2i (CP-724714) (10 μM), at time point 0. a Time courses of average ERK activity under each condition (EGFRi: n = 70, ErbB2i: n = 62, EGFRi + ErbB2i: n = 57 cells). b Quantiﬁcation of ERK activity before and after treatment with EGFRi or ErbB2i (EGFRi: n = 113, ErbB2i: n = 147 cells, pooled from two organoids). c Time course of average ERK activity in organoids that were cultured under EGF-starved condition (NR) for 24 h, −1 treated with either EGFRi or ErbB2i for 60 min, and then stimulated with 50 ng ml of EGF (EGFRi: n= 48, ErbB2i: n = 32 cells). d Quantiﬁcation of ERK activity in organoids infected with a control lentivirus (control) or lentiviruses expressing either a dominant negative form of EGFR (dnEGFR) or that of ErbB2 (dnErbB2) (Control: n= 134, dnEGFR: n = 129, dnErbB2: n = 151 cells, pooled from ﬁve organoids cultured in the ENR medium). e, f Quantiﬁcation of ERK activity pulses in EKAREV-NLS organoids cultured in the ENR medium and treated with EGFRi and/or ErbB2i (−/−: n = 71, EGFRi/−: n = 49, −/ErbB2i: n = 36, EGFRi/ ErbB2i: n = 53 cells). Organoids were treated with indicated inhibitors and imaged for 90 min. ERK activity data were smoothened by 6-min moving average, and ﬁtted to ﬂat lines or multi-peak functions. e The proportion of cells exhibiting the pulse-like ERK activation (ERK-pulse ) under each condition. f Frequencies of ERK activity pulses under each condition. Duration (g) and frequencies (h) of ERK activity pulses in organoids cultured in the ENR medium and treated with 0, 10, 1, or 0.1 μM of a BRAF inhibitor (SB590885) for 90 min (n = 55, 52, 29, and 65 cells, respectively). Scale bars, 50 µm. Red lines represent mean. Error bars represent s.e.m. Mann–Whitney U-test (b) and Steel–Dwass test (d, f–h) were used for comparison. *P < 0.05, ** P < 0.001, ***P < 0.0001, n.s., not signiﬁcant activity (P < 0.0001) (Fig. 4d). Moreover, knockdown of ErbB2 or (Fig. 5c, d). Moreover, adenoma-derived organoids were less EGFR by short hairpin RNA exerted similar effects on the basal sensitive to ErbB2i compared to the normal organoids (Figs. 5d ERK activity (Supplementary Fig. 3a, b). Taken together, these and 4b). Meanwhile, EGFRi, but not ErbB2i, decreased the fre- results show that ErbB2, but not EGFR, mainly drives the basal quency of the ERK activity pulses (Fig. 5e) and the proportion of ERK activity in intestinal organoids. ERK-pulse cells in adenoma-derived organoids (Fig. 5f). We then examined the role of EGFR and ErbB2 in generating Treatment with both inhibitors almost completely suppressed the the ERK activity pulses. To extract the quantitative parameters of ﬁring of ERK activity pulses (Fig. 5e, f). Altogether, these results the pulses, time-course data of ERK activity in each cell were indicate that EGFR signalling is a predominant driver for basal ﬁtted to horizontal lines (corresponding to the sustained, constant ERK activity and spontaneous ERK activity pulses in adenoma- activity) or multi-peak functions (for the pulse-like activity). We derived organoids. categorized cells exhibiting one or more ERK activity pulses, as + − ERK-pulse , or otherwise as ERK-pulse (Supplementary Wnt signalling activation alters ERK activity dynamics. Fig. 3c–e). Treatment with EGFRi, but not that with ErbB2i, Although deregulated/constitutive activation of Wnt signalling decreased the proportion of ERK-pulse cells (Fig. 4e). Moreover, caused by APC mutations is known as the initial genetic event in the frequency of ERK activity pulses was substantially reduced by the adenoma-carcinoma sequence , accumulating evidences indi- EGFRi, but not by ErbB2i (Fig. 4f). Upon co-inhibition of EGFR cate that tumour cells have already acquired epigenetic alterations 43 44 and ErbB2, the ERK activity pulses disappeared almost before the APC mutations , . Therefore, we inquired whether completely (Fig. 4e, f). In line with this, knockdown of EGFR, activation of Wnt signalling alone could recapitulate the altered but not that of ErbB2, reduced the frequency of ERK activity ERK activity dynamics in adenoma-derived organoids. We utilized pulses (Supplementary Fig. 3f). These results show that, in a GSK3 inhibitor, CHIR99021, which causes β-catenin accumula- intestinal organoids, ERK activity pulses are generated by EGFR tion thereby activating canonical Wnt/β-catenin signalling. kinase activity whereas basal ERK activity is maintained by ErbB2 CHIR99021 did not affect the basal ERK activity at least for 24 h kinase activity. Removal of R-spondin 1 or Noggin did not affect (Fig. 6a, b); however, CHIR99021 surprisingly changed the cellular basal ERK activity, its dependency on ErbB2, and the dependency responses to EGFRi and ErbB2i. In organoids treated with of ERK activity pulses on EGFR (Supplementary Fig. 4a–f), CHIR99021, EGFRi decreased the basal ERK activity more potently negating involvement of these proteins in controlling ERK than did ErbB2i (Fig. 6c), as observed in the adenoma-derived activity dynamics in intestinal organoids. Interestingly, treatment organoids. Simultaneous addition of EGFRi and ErbB2i most with a BRAF inhibitor, SB590885, not only decreased the strongly suppressed the basal ERK activity (Fig. 6c). The effects of frequency but also prolonged the duration of ERK activity pulses CHIR99021 disappeared after 24 h incubation in ENR medium (Fig. 4g, h). As the shape of pulses generally depends on the (without CHIR99021), suggesting that CHIR99021-induced sensi- activity of the pulse generator, BRAF might be one of the pulse tivity to EGFRi was reversible (Fig. 6d). CHIR99021 also increased generators in this system. the proportion of cells with ERK activity pulses (ERK-pulse ) (Fig. 6e) and the frequency of the pulses (Fig. 6f). The ﬁring of the EGFR governs distinct ERK activity dynamics in adenoma ERK activity pulse was suppressed almost completely by EGFRi and cells. The above results demonstrated distinct roles for EGFR and slightly by ErbB2i (Fig. 6e, f).Notably,Wnt ligand stimulationor ErbB2 in regulating the ERK activity dynamics in normal cells. overexpression of constitutively active β-catenin exerted effects We next investigated how these receptors regulate ERK activity similar to those of CHIR99021 on ERK activity dynamics (Sup- dynamics during intestinal tumorigenesis. Organoids were plementary Fig. 5). These results suggest that activation of Wnt Δ716 generated from adenomas developed in Apc mice, signalling increases the contribution of EGFR signalling to ERK which have a truncation mutation in Apc , followed by intro- activity thereby altering the ERK activity dynamics similar to those duction of the biosensor for ERK. Under our experimental con- in adenoma-derived organoids. ditions, approximately half of the cells in the adenoma-derived To examine the biological signiﬁcance of Wnt signalling- organoids expressed the biosensor (Fig. 5a). There was no sig- dependent alterations in ERK activity dynamics, we conducted niﬁcant difference in basal ERK activity between the normal and cell cycle analyses in intestinal organoids. We generated intestinal adenoma-derived organoids (P = 0.7599) (Fig. 5a, b). However, a organoids from transgenic mice expressing a cell cycle reporter striking difference was observed in the presence of EGFRi and (Fucci2a), which marks S/G2/M phase cells and G1 phase cells by ErbB2i: treatment with EGFRi, which did not affect basal ERK mVenus and mCherry, respectively (Fig. 6g). Consistent with activity in the normal organoids (Fig. 4a, b), signiﬁcantly an essential role of ERK in cell proliferation, MEK inhibitor decreased it in the adenoma-derived organoids (P < 0.0001) treatment strongly reduced the proportion of S/G2/M phase cells NATURE COMMUNICATIONS (2018) 9:2174 DOI: 10.1038/s41467-018-04527-8 www.nature.com/naturecommunications 7 | | | ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-04527-8 Adenoma-derived organoids n.s. a b 2.0 Normal Adenoma-derived 2.0 1.5 Pre Post 1.0 1.0 0.2 2.0 0.5 FRET/CFP Pre Post Normal Adenoma d ef Adenoma-derived Adenoma-derived Adenoma-derived organoids organoids organoids *** * *** *** ERK-pulse (+) ERK-pulse (–) 1.4 40 n.s. 100% *** 1.2 75% 1.0 50% 0.8 25% 0.6 0 0% Pre Post Pre Post Pre Post EGFRi−− + + EGFRi−− + + EGFRi ErbB2i EGFRi + ErbB2i ErbB2i −− ++ ErbB2i −− ++ Fig. 5 Adenoma-derived organoids exhibit increased dependence on EGFR signalling. a, b Organoids were generated from the normal epithelium Δ716 or adenomas of the Apc mouse small intestine, and infected with lentiviruses expressing a FRET biosensor for ERK activity (EKAREV-NLS). a Representative images of ERK activity (FRET/CFP ratio) in organoids derived from the normal epithelium and adenomas. b Bee swarm plots of ERK activity in the normal epithelium-derived and adenoma-derived organoids (Normal: n = 61, Adenoma: n = 99 cells, from more than three organoids). c–f Adenoma-derived organoids expressing EKAREV-NLS were treated with 1 μM of an EGFR inhibitor (EGFRi), PD153035, and/or 10 μM of an ErbB2 inhibitor (ErbB2i), CP-724714. c Representative images of adenoma-derived organoids before and after treatment with EGFRi or ErbB2i. d Bee swarm plots of ERK activity before and after inhibitor treatment (EGFRi: n = 120, ErbB2i: n = 119, EGFRi + ErbB2i: n = 109 cells, pooled from three organoids). e, f Quantiﬁcation of ERK activity pulses in adenoma-derived organoids. Organoids were treated with EGFR and/or ErbB2 inhibitors, and then imaged for 90 min. ERK activity data from each cell were processed as described for Fig. 4e(−/−: n = 40, EGFRi/−: n = 45, −/ErbB2i: n = 32, EGFRi/ErbB2i: n = 29 cells). Frequencies of the pulse-like ERK activity (e) and the proportion of cells with pulse-like ERK activity (ERK-pulse )(f) under each condition. Scale bars, 50 µm. Red lines represent mean. Error bars represent s.e.m. Mann–Whitney U-test (b, d) and Steel–Dwass test (e) were used for comparison. *P < 0.05, **P < 0.001, ***P < 0.0001, n.s., not signiﬁcant (Fig. 6h). ErbB2i, but not EGF depletion or EGFRi, signiﬁcantly inquired whether administration of a GSK3 inhibitor affects decreased the proportion of S/G2/M phase cells (P < 0.0001) EGFR signalling in the mouse intestinal epithelium, as does it in (Fig. 6h). The effect of ErbB2i was further enhanced by intestinal organoids. To this end, CHIR99021 was administered simultaneous addition of EGFRi (Fig. 6h). As expected, removal to EKAREV-NLS mice, and ERK activity before and after of R-spondin 1 or Noggin also decreased S/G2/M phase cells, administration of a clinically used EGFR inhibitor, erlotinib was consistent with their roles in maintaining undifferentiated IECs observed. Administration of CHIR99021 increased the expression (Supplementary Fig. 6a, b). Even under the R-spondin 1- or of several genes, which were also increased by CHIR99021 in Noggin-depleted condition, ErbB2i reduced S/G2/M phase cells intestinal organoids and have been reported as the Wnt-target more strongly than EGFRi (Supplementary Fig. 6a, b). These genes, in the intestinal epithelium (Supplementary Fig. 7a). results suggest that ErbB2 primarily drives cell cycle progression Intravital imaging demonstrated that erlotinib decreased the basal in normal intestinal organoids and that EGFR serves as the ERK activity in the intestinal epithelium of CHIR99021-treated auxiliary driver. Thus, basal ERK activity driven by mice but not in that of control mice (Fig. 7a). Notably, ErbB2 signalling might play a major role in cell proliferation in CHIR99021 increased the expression of EGFR protein in the the normal organoids. Notably, treatment with CHIR99021 or crypt region that mainly comprised stem cells and progenitor Wnt ligands, which increased the frequency of ERK activity cells (Fig. 7b, c). The EGFR protein was expressed more abun- pulses without affecting the basal ERK activity (Fig. 6f, Supple- dantly in the crypt than in the villus (Fig. 7b), where Wnt sig- mentary Fig. 5c), increased proliferating cells in an EGFR- nalling activity should be low . Moreover, EGFR expression was Δ716 dependent manner (Fig. 6i, Supplementary Fig. 6c–e). Thus, the markedly increased in adenomas of Apc mice (Fig. 7d, e), as increased frequency of ERK activity pulses correlated with well as adenoma-derived organoids (Fig. 7f, g) and CHIR99021- increased cell proliferation after those treatments. treated organoids (Supplementary Fig. 7b, c). Thus, the increased expression of EGFR might be involved in the Wnt signalling- dependent augmentation of EGFR signalling. The expression GSK3 inhibitor treatment enhances EGFR signalling in vivo. levels of ErbB2 were comparable among adenomas and the normal epithelium (Supplementary Fig. 7d). Finally, we examined The results from intestinal organoid studies suggest the Wnt signalling-dependent control of EGFR signalling in IECs. We thus whether alteration in EGFR signalling activity affects cell 8 NATURE COMMUNICATIONS (2018) 9:2174 DOI: 10.1038/s41467-018-04527-8 www.nature.com/naturecommunications | | | Normalized ERK activity (FRET/CFP) Normalized ERK activity (FRET/CFP) Frequency (/cell/day) FRET/CFP ErbB2i EGFRi NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-04527-8 ARTICLE a b n.s. c EGFRi and/or ErbB2i 1.4 1.1 EGFRi ENR ENR+CHIR 1.8 1.2 ErbB2i 1.0 EGFRi + ErbB2i 1.0 0.8 0.9 0.3 0.6 0.8 ENR ENR+CHIR –5 0 10 20 30 Elapsed time (min) Imaging ENR ENR ENR+CHIR ENR ENR+CHIR e f ERK-pulse (+) ERK-pulse (–) n.s. Re-ENR ENR ENR+CHIR ENR 100% *** –48 h –24 h 0 1.4 n.s. n.s. 75% ** 1.2 50% 1.0 25% 0% 0.8 CHIR −+ + + CHIR −+ + + EGFRi −+ − − EGFRi −+ − − EGFRi Pre Post Pre Post Pre Post ErbB2i −− −+ ErbB2i −− −+ ENR ENR+CHIR Re-ENR S/G2/M phase G1 phase gh i S/G2/M phase G1 phase *** *** CAG mCherry-hCdt1 T2A mVenus-hGem ** n.s. n.s. *** *** *** *** 100% 100% hGem / hCdt1 75% 75% G2 G1 50% 50% 25% 25% 0% 0% CHIR −+ + + + EGFRi −+ − − + ErbB2i −− − + + Fig. 6 Pharmacological activation of Wnt signalling promotes cell proliferation by augmenting EGFR signalling. a Representative images and b quantiﬁcation of ERK activity in organoids treated with or without CHIR99021 (5 μM) for 24 h (n = 210 and 213 cells (from the left), from three organoids). c Time courses of ERK activity in organoids treated with CHIR99021 for 24 h and then treated with EGFRi (PD153035), and/or ErbB2i (CP-724714) (n = 45, 29, and 29 cells from the left). d Quantiﬁcation of ERK activity before (Pre) and after (Post) EGFRi treatment (bottom) (n = 30, 44, and 53 cells from the left). Organoids cultured in ENR, ENR+CHIR, or ENR+CHIR99021 for 24 h and then in ENR for another 24 h (Re-ENR) were treated with EGFRi (top). e, f Quantiﬁcation of ERK activity pulses in organoids treated with CHIR99021 for 24 h and then with EGFRi and/or ErbB2i for 90 min during imaging. The proportion of cells with ERK activity pulses (ERK-pulse )(e), and the frequency of the pulses (f) under each condition (n = 36, 47, 47, and 25 cells from the left). g Schematic structure of Fucci2a (top), cell cycle diagram marked with corresponding ﬂuorescence (bottom left), and the representative image of organoid expressing Fucci2a (bottom right). h, i Proportion of cells in the S/G2/M phases in the Fucci2a organoids. h Organoids were cultured for 24 h in ENR, EGF-deﬁcient medium (NR), or ENR supplemented with EGFRi ErbB2i, and/or a MEK inhibitor (MEKi) (n = 158, 264, 400, 359, 423, and 243 cells (from the left), from more than four organoids). i Organoids treated with CHIR99021 for 24 h were subsequently treated with EGFRi, and/or ErbB2i for another 24 h (n = 156, 1062, 461, 220, and 234 cells (from the left), from more than four organoids). Scale bars, 50 µm. Red lines represent mean. Error bars represent s.e.m. Mann–Whitney U-test (b, d), Steel–Dwass test (f), and χ test with BH procedure (h, i) were used for comparison.*P < 0.05, **P < 0.001, ***P < 0.0001, n.s., not signiﬁcant NATURE COMMUNICATIONS (2018) 9:2174 DOI: 10.1038/s41467-018-04527-8 www.nature.com/naturecommunications 9 | | | ENR NR EGFRi ErbB2i EGFRi + ErbB2i MEKi Normalized ERK activity (FRET/CFP) FRET/CFP S/G2/M cells (%) Normalized ERK activity (FRET/CFP) Normalized ERK activity S/G2/M cells (%) (FRET/CFP) Frequency (/cell/day) 2.0 1.5 1.0 0.5 ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-04527-8 a b c Hoechst EGFR/E-Cadherin n.s. *** 1.2 ** 1.0 1.0 0.8 0.2 Pre Post Pre Post Control CHIR Erlotinib Control CHIR d e Hoechst EGFR E-Cadherin EGFR/E-Cadherin 4.0 *** n.s. *** 3.0 1.0 2.0 1.0 Adenoma Crypt Villi Normal epithelium f g Hoechst EGFR ErbB2 EGFR/ErbB2 ENR condition 1.5 *** 1.0 1.0 0.5 Normal Adenoma *** Control Erlotinib CHIR99021 CHIR + Erlotinib n.s. *** CHIR−+ − + Erlotinib−+ + − proliferation and survival in the intestinal epithelium. We found signalling and promotes proliferation and survival of IECs, and that administration of CHIR99021 promoted proliferation of crypt that enhancement of EGFR signalling renders cells highly sensitive cells, whereas erlotinib did not exhibit a signiﬁcant effect (Fig. 7h, to EGFR inhibition in vivo. It is also noteworthy that autopho- i, Supplementary Fig. 7e, f). Remarkably, simultaneous adminis- sphorylation of ErbB2 was suppressed by ErbB2i, but not by tration of CHIR99021 and erlotinib strongly suppressed the pro- EGFRi, in normal and adenoma organoids (Supplementary liferation of crypt cells (Fig. 7h, i, Supplementary Fig. 7e, f), Fig. 8a–d). Thus, EGFR might be dispensable for ErbB2 activation. promoted apoptosis (Supplementary Fig. 7g), and severely dis- turbed the crypt–villus structures (Fig. 7h). These results suggest Wnt signalling controls ERK activity via EGFR regulators. that in vivo administration of CHIR99021 enhances EGFR Finally, we examined whether Wnt signalling-induced gene 10 NATURE COMMUNICATIONS (2018) 9:2174 DOI: 10.1038/s41467-018-04527-8 www.nature.com/naturecommunications | | | Adenoma Normal Normalized ERK activity Hoechst/EdU organoid organoid Adenoma Normal epithelium (FRET/CFP) CHIR99021 Control EGFR/E-Cadherin intensity EGFR/ErbB2 intensity EdU cells/crypt section EGFR/E-Cadherin intensity NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-04527-8 ARTICLE Fig. 7 EGFR signalling is augmented in adenomas and CHIR99021-treated mouse intestine in vivo. a Bee swarm plots of ERK activity in the vehicle- (Control) or CHIR99021-treated (CHIR) mouse intestinal epithelium before (Pre) and after (Post) treatment with an EGFR inhibitor, erlotinib. Mice were −1 −1 injected with vehicle or 20 mg kg body weight of CHIR99021 for 3 days and then treated with 100 mg kg body weight erlotinib for 30 min (Control: n = 48, CHIR: n = 61 cells from three crypts). b, c Immunoﬂuorescence staining of the small intestine in the mice injected with vehicle (Control) or CHIR99021 for 3 days using anti-EGFR and anti-E-cadherin antibodies. b The ratio of EGFR and E-cadherin staining intensities is shown in the IMD mode according to the colour scale. Counterstaining was performed with Hoechst. Note that strong staining of stromal cells results from non-speciﬁc binding of the secondary antibodies used here (anti-mouse IgG). c Quantiﬁcation of the EGFR/E-cadherin staining intensity ratio in each cell under vehicle or CHIR99021-treated Δ716 condition (n = 50 cells). d Immunoﬂuorescence staining of an adenoma and the normal small intestine of Apc mice with anti-EGFR and anti-E-cadherin antibodies. EGFR/E-cadherin staining intensity ratio is shown in the IMD mode. e Quantiﬁcation of the EGFR/E-cadherin ratio in each cell located in the indicated regions (n = 50 cells). f Immunoﬂuorescence staining of normal and adenoma-derived organoids with anti-EGFR and anti-ErbB2 antibodies. EGFR/ErbB2 staining intensity ratio is shown in the IMD mode. g Quantiﬁcation of the EGFR/ErbB2 ratio in each cell in the organoids (n = 40 cells pooled from at least two organoids). h, i EdU staining of the small intestine of CHIR99021- and/or erlotinib-treated mice. Mice were injected with vehicle, 20 mg −1 −1 + kg body weight of CHIR99021, and/or 100 mg kg body weight of erlotinib for 3 days. i Quantiﬁcation of EdU cells per crypt section (−/−: n = 46, CHIR/–: n = 42, −/Erlotinib: n = 54, and CHIR/Erlotinib: n= 45 crypts from three mice). Scale bars, 50 µm. Red lines represent mean. Error bars represent s.e.m. Mann–Whitney U-test (a, c, g), and Steel–Dwass test (e, i) were used for comparison.*P < 0.05, **P < 0.001, ***P < 0.0001 expression changes regulate EGFR–ERK signalling dynamics. For constant component and EGFR-dependent pulse-like component this purpose, we analyzed gene expression proﬁles of normal (Fig. 8j). Remarkably, Wnt signalling activation, which often (ENR) and CHIR99021-treated organoids (ENR + CHIR) by occurs during intestinal tumorigenesis, enhanced EGFR–ERK using microarrays. We identiﬁed 81 and 274 transcripts whose signalling, at least in part, by controlling expression of EGFR and expression levels were upregulated and downregulated, respec- its regulators, such as Egﬂ6, Lrig3, and Troy (Figs. 7 and 8). This tively, in CHIR99021-treated organoids by more than 3 folds alteration increased the contribution of EGFR to the basal ERK (Fig. 8a, Supplementary Data 1). Gene set enrichment analysis activity, increased the frequency of ERK activity pulses, and revealed that CHIR99021 induced gene expression changes promoted proliferation of IECs (Figs. 6 and 7). Thus, Wnt sig- similar to those in the Apc-knockout epithelium and colorectal nalling activation sensitized IECs to pharmacological inhibition of adenomas (Fig. 8b–e). Indeed, gene sets comprising genes EGFR signalling (Fig. 7). These results show that distinct downregulated and upregulated after Apc knockout were most upstream tyrosine kinase receptors can generate different modes signiﬁcantly enriched in the CHIR99021-dependent down- of ERK activity dynamics to control cellular responses, and that regulated and upregulated gene list, respectively (Supplementary deregulation of such mechanisms might underlie the character- Table 1). Notably, CHIR99021 altered the expression of four istic features of cancer cells. genes whose protein products promote (Egﬂ6, Flna, and Troy) or Upon Wnt signalling activation, augmented EGFR signalling 47–50 suppress (Lrig3) EGFR signalling (Supplementary Data 1). increased the frequency of ERK activity pulses, which was cor- The expression levels of Dusps, Sprys, Spreds, and MIG6, which related with the increased proliferation of IEC (Figs. 6 and 7). also regulate EGFR–ERK signalling , were not signiﬁcantly Importantly, however, these results do not indicate that the altered by CHIR99021 (Supplementary Fig. 9a). RT-PCR analyses pulsatile nature of EGFR–ERK signalling per se is needed for the conﬁrmed that Egﬂ6, Flna, and Troy were upregulated, and that promotion of cell proliferation. Notably, ErbB2 inhibition Lrig3 was downregulated in CHIR99021-treated organoids decreased the basal ERK activity less efﬁciently in adenomas or in (Fig. 8f) and adenoma-derived organoids (Fig. 8g). To examine GSK3 inhibitor-treated organoids than in normal organoids whether these genes mediate the altered ERK activity dynamics in (Figs. 5 and 6). This shows that, after Wnt signalling activation, adenomas, we performed exogenous expression and knockdown basal ERK activity is maintained at a constant level through of these genes in adenoma-derived organoids (Supplementary adaptation to enhanced EGFR signalling, which increases the Fig. 9b). The knockdown of Egﬂ6 or expression of Lrig3 abolished relative contribution of EGFR to the ERK activity. the sensitivity of adenoma cells to EGFR inhibition (Fig. 8h). We demonstrated that EGFR, but not ErbB2, is a major gen- Meanwhile, the frequency of ERK activity pulses was signiﬁcantly erator of the ERK activity pulses (Fig. 4e, f). The distinct con- decreased by the Lrig3 expression (P = 0.0001) and the Troy tribution of EGFR and ErbB2 to ERK activity dynamics might be knockdown (P = 0.0009) in adenoma organoids (Fig. 8i). attributed to the different regulatory mechanisms involved. Since Knockdown of Flna did not signiﬁcantly affect ERK activity EGFR–ERK signalling activity can be affected by the ﬂuctuation dynamics in either experiment (Fig. 8h, i). These results suggest of the topical concentration of EGFR ligands and also be regu- 35, 36, 53 that decreased expression of Lrig3 and increased expression of lated by multiple negative feedback mechanisms , these Egﬂ6 and Troy, probably together with enhanced EGFR effects might introduce a pulsatile nature into EGFR signalling. In expression, coordinately promotes EGFR–ERK signalling in contrast to EGFR, ErbB2 does not have any known ligands and its adenoma cells. activity is mainly regulated by dimerization . In addition, ErbB2- containing heterodimers, compared to dimers without ErbB2, exhibit slower rates of endocytosis and dissociation from Discussion 55–57 ligands . Considering all these studies, the ErbB2 homodimer The spatiotemporal regulation of ERK signalling has been con- and ErbB2-containing heterodimers might be more stable and sidered a key determinant of cellular responses in many biological 51, 52 transmit sustained, constant inputs to ERK. Importantly, contexts , and its disturbance has been implicated in many knockdown of EGFR did not affect basal ERK activity (Supple- diseases including cancers . Here, we successfully visualized ERK mentary Fig. 3b), which is dependent on ErbB2 kinase activity. activity in the intestinal epithelium and unveiled the roles of Moreover, an ErbB2 inhibitor, but not an EGFR inhibitor, EGFR and ErbB2 in regulating ERK activity dynamics. We found reduced autophosphorylation of ErbB2 (Supplementary that EGFR kinase activity is required for ERK activity pulses, and Fig. 8a–d), suggesting that the majority of ErbB2 autopho- that ErbB2 is the primary tyrosine kinase governing basal ERK sphorylation is mediated by ErbB2, but not by EGFR. Thus, activity and cell proliferation in intestinal organoids (Figs. 4 ErbB2 seems to be activated mainly via homodimerization, and 6). ERK activity in IECs comprises the ErbB2-dependent NATURE COMMUNICATIONS (2018) 9:2174 DOI: 10.1038/s41467-018-04527-8 www.nature.com/naturecommunications 11 | | | ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-04527-8 a bc Genes downregulated after Apc KO Genes upregulated after Apc KO via Myc 0.0 0.8 Upregulated: 0.6 81 genes –0.2 (0.28%) 0.4 –0.4 0.2 5 –0.6 0.0 –0.8 ENR+CHIR ENR ENR+CHIR ENR de Genes downregulated Genes upregulated Downregulated: in colorectal adenoma in colorectal adenoma 274 genes (0.95%) 0.4 0.0 0.2 –0.2 0 0.0 –8 –6 –4 –2 04 2 6 –0.4 -0.2 log (ENR+CHIR/ENR) –0.6 ENR+CHIR ENR ENR+CHIR ENR f g ENR 4 200 * Normal organoid ENR+CHIR Adenoma organoid 100 2 ** n.s. * * Egfl6 Flna Troy Lrig3 Dusp4 Dusp10 Egfl6 Flna Troy Lrig3 h i Adenoma organoid n.s. n.s. Adenoma organoid ** *** *** 1.4 n.s. n.s. 1.2 20 * ** 1.0 0.8 10 0.6 EGFRi Pre Post Pre Post Pre Post Pre Post Pre Post Control shEgfl6 shFlna shTroy Lrig3 expression Normal epithelium Enhanced Wnt activity Promoted cell proliferation EGFR EGFR ErbB2 ErbB2 Time EGFR EGFR inhibition EGFR inhibition Egfl6 Adenoma formation Troy Lrig3 ErbB2 ErbB2 Time Limited effect on Suppressed cell proliferation cell proliferation although our data could not rule out the involvement of EGFR receptor level are involved in different ERK activity dynamics in and ErbB3 in the ErbB2 activation. In any case, our results cultured mammary epithelial cells . Since our results also suggest indicate that the upstream receptors under the control of different the involvement of Raf in generating ERK activity pulses (Fig. 4g, regulatory mechanisms can generate distinct modes of ERK h), multiple feedback loops might govern the pulses at different activity dynamics. Consistently, distinct mechanisms acting at the levels of the signalling cascade. It should be clariﬁed that EGFR 12 NATURE COMMUNICATIONS (2018) 9:2174 DOI: 10.1038/s41467-018-04527-8 www.nature.com/naturecommunications | | | –log (p -value) Normalized ERK activity (FRET/CFP) Relative mRNA level ERK activity ERK activity Enrichment score Enrichment score Relative mRNA level Frequency (/cell/day) Control Enrichment score Enrichment score Relative mRNA level shEgfl6 shFlna shTroy Lrig3 NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-04527-8 ARTICLE Fig. 8 Wnt signalling activation affects EGFR–ERK signalling dynamics through regulating expression of multiple molecules. a A volcano plot depicting the fold changes in gene expression levels between normal (ENR) and CHIR99021-treated organoids (ENR + CHIR), and statistical signiﬁcance of the changes. b–e Enrichment plots from gene set enrichment analysis (GSEA). GESA plots for genes downregulated after Apc knockout (b), genes upregulated after Apc knockout through Myc (c), genes downregulated in colorectal adenoma (d), and genes upregulated in colorectal adenoma (e) are shown. f, g RT-PCR analysis revealed that Egﬂ6, Flna, and Troy were upregulated, and that Lrig3 was downregulated in both CHIR99021-treated and adenoma-derived organoids. The relative mRNA levels of indicated genes in normal (ENR) versus CHIR99021-treated organoids (ENR + CHIR) (f), and those in normal versus adenoma-derived organoids (g) are shown (n = 3 samples containing more than ten organoids). h, i Adenoma-derived organoids expressing the ERK biosensor were infected with lentiviruses expressing control vector (control), Lrig3 (Lrig3), or shRNAs for Egﬂ6 (shEgﬂ6), Flna (shFlna), or Troy (shTroy). h Bee swarm plots of ERK activity in organoids before (Pre) and after (Post) EGFR inhibitor treatment under each condition (n = 80 cells pooled from two organoids). i The frequency of ERK activity pulses under each condition. Time-lapse imaging was performed for 90 min (n = 50 cells). j Schematic representation of ERK activity dynamics generated by kinase activity of EGFR and ErbB2 in the normal and Wnt signalling-activated intestinal epithelia. Red lines represent mean. Error bars represent s.e.m. Welch’s t test (f, g), Mann–Whitney U-test (h), and Steel–Dwass test (i) were used for comparison.*P < 0.05, **P < 0.001, ***P < 0.0001 has a kinase activity-independent function as a heterodimeriza- ErbB2compared to inhibition of either one. Consistent with this tion partner of ErbB2. However, our results did not support the ﬁnding, simultaneous inhibition of EGFR and ErbB2 has been contribution of EGFR to ErbB2 activation, as discussed above. shown to be effective in refractory CRCs , though inhibition of Thus, the observed functions of EGFR are likely to be mediated ErbB2 alone was ineffective. Thus, our ﬁndings and the proposed by intracellular signalling triggered by EGFR kinase activity, model are in good agreement with the clinical efﬁcacy of EGFR which is referred to as EGFR signalling in this study. and ErbB2 inhibition in CRC treatment. Nevertheless, it should Different dynamics of receptor tyrosine kinases could lead to be noted that there are signiﬁcant differences between mouse distinct dynamics of the downstream ERK activity . For exam- adenomas and human CRCs. In particular, mouse adenomas are ple, in PC12 cells, activated EGFR rapidly undergoes the clathrin- usually assumed to have only Apc mutations, whereas human based endocytosis and the following degradation, whereas acti- CRCs often contain many genetic mutations, which might affect vation of TrkA, a receptor for NGF, induces its translocation to cellular responses to EGFR inhibition. Indeed, CRCs harbouring long-lived signalling endosomes, where TrkA avoids degradation RAS or RAF mutations are resistant to EGFR inhibitors, and and mediates sustained ERK activation . Thus, mechanisms initially sensitive CRCs can also acquire the resistance via several regulating dynamics of speciﬁc receptors can regulate ERK mechanisms . activity dynamics. In this study, we identiﬁed Egﬂ6, Lrig3, and Recent studies have shown that ErbB signalling regulates the 18, 19 Troy as important regulators of EGFR–ERK signalling dynamics transition between quiescence and proliferation of ISCs . For in IECs (Fig. 8). Previously, these proteins have been reported to instance, a pan-ErbB negative regulator, Lrig1, is expressed in a 47–49 regulate EGFR signalling in cultured cells . Notably, altered quiescent, long-lived subpopulation of ISCs, which is distinct expression of these regulators was associated with changes in ERK from the Lgr5 population and serves as an origin of intestinal 18, 19, 61 activity dynamics in adenoma cells (Fig. 8). Thus, elucidation of tumours . Lrig1 depletion increases the expression of ErbB mechanisms by which these regulators function in IECs would family receptors and enhances ERK activity, resulting in ISC 18, 19 provide a clue about how ERK activity dynamics can be altered expansion and adenoma formation . Thus, Lrig1 plays a during tumorigenesis and how it affects tumour cell phenotypes. crucial role in controlling quiescence and proliferation of ISCs. In EGFR signalling has been implicated in many cancers and its this study, we showed that Wnt signalling activation, increases pharmacological inhibition has been considered promising for the expression of EGFR, but not ErbB2 (Fig. 7). Since Lrig1 4, 5 18, 19 cancer treatment . Our results might provide a therapeutic broadly regulates the expression of ErbB receptors , selective rationale for why EGFR inhibitors can speciﬁcally suppress the induction of EGFR suggests Lrig1-independent regulation of growth of certain types of CRCs without severe adverse effects. EGFR expression. More recently, Basak and colleagues have Since strong activation of Wnt signalling augments EGFR sig- reported that EGFR–ERK signalling mediates ISC proliferation nalling and renders cells highly dependent on EGFR signalling and that inhibition of EGFR and ErbB2, MEK, or ERK induces (Figs. 6 and 7), it can be readily speculated that pharmacological reversible quiescence of ISCs . Notably, they also showed that inhibition of EGFR preferentially targets CRCs rather than nor- inhibition of EGFR alone could suppress proliferation of ISCs . mal IECs harbouring weaker Wnt signalling activity. Con- In contrast, we did not observe any signiﬁcant alterations in cell sistently, administration of a GSK3 inhibitor (CHIR99021), which proliferation after EGFR inhibition for 24 h. This discrepancy promoted expression of, at least a fraction of, Wnt-target genes, could be due to the difference in experimental settings, since they increased the sensitivity of normal IECs to EGFR inhibition mainly focused on the effects of long-term perturbation on ISC in vivo (Fig. 7). Importantly, however, the effects of CHIR99021 functions whereas we focused on the immediate effects of the administration and Wnt signalling activation are not completely treatment, which more likely reﬂect alterations in rapidly cycling the same, as many events other than GSK3β inhibition occur in progenitors rather than in ISCs. Since stem cells harbour strong Wnt signalling and GSK3 plays various roles independently of Wnt signalling activity , inhibition of EGFR alone might sup- Wnt signalling. Also, the effects of CHIR99021 were not as strong press their proliferation in the long term. as those of Wnt signalling activation; CHIR99021 promoted cell In conclusion, we have revealed ERK activity dynamics regu- proliferation only weakly whereas genetic activation of Wnt sig- lated by two distinct upstream receptors, EGFR and ErbB2, in the nalling has been shown to cause hyperplasia in the intestinal intestinal epithelium and shown that alterations in the dynamics epithelium. Nevertheless, CHIR99021 exerted similar effects on caused by Wnt signalling activation underlie the high sensitivity ERK activity dynamics and expression of several genes both of tumour cells to EGFR inhibition. These ﬁndings highlight the in vitro and in vivo. In addition to EGFR signalling, our results importance of understanding ERK activity dynamics and their also demonstrate the signiﬁcant contribution of ErbB2 signalling role in controlling cellular functions at the single cell level, issues to ERK activity in IECs. Indeed, proliferation of IECs was more that could not be addressed by conventional biochemical effectively suppressed by concomitant inhibition of EGFR and approaches. Similar phenomena could have occurred for many NATURE COMMUNICATIONS (2018) 9:2174 DOI: 10.1038/s41467-018-04527-8 www.nature.com/naturecommunications 13 | | | ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-04527-8 other signalling pathways and molecules, elucidation of which Intestinal organoid culture. Small intestinal organoids were generated according to the published protocol . Brieﬂy, intestinal crypts were isolated from the mouse should be an important future research area to improve the small intestine by incubation for 30 min at 4 °C in PBS containing 2 mM EDTA, understanding of cellular drug responses at the single cell level. and cultured in Advanced DMEM/F12 Medium (Thermo Fisher Scientiﬁc) sup- Live imaging with highly sensitive FRET biosensors could be a −1 −1 plemented with EGF (50 ng ml , PeproTech), Noggin (100 ng ml , PeproTech), −1 powerful tool to address these issues and shed new light on the and R-spondin1 (500 ng ml , R&D Systems). Adenoma cells were isolated from Δ716 the Apc mouse small intestinal adenomas by incubation for 60 min at 4 °C in mechanisms involved in many physiological and pathological PBS containing 2 mM EDTA. Organoids were embedded in Growth Factor processes. Reduced Matrigel (Corning), and maintained by serial passaging. Organoids were imaged using an incubator-type two-photon excitation microscope, LCV110-MPE, within 4 days of the last passage in order to minimize the autoﬂuorescence emitted Methods from the cell debris accumulating in their central cavities. In some experiments, Lentivirus production and infection of intestinal organoids. The FRET bio- organoids were treated with TPA (1 μM, LC laboratories), PD0325901 (200 nM, sensor for ERK has been described previously . The dominant negative form of Calbiochem), PD153035 (1 μM, MedChemexpress), CP-724714 (10 μM, AdooQ), EGFR (dnEGFR) is a C-terminally truncated mutant encoding amino acid 1 to Marimastat (100 μM, R&D Systems), or SB590885 (100 nM–10 μM, LKT Labora- 679 , and the dominant negative form of ErbB2 (dnErbB2) encodes 1 to 693, tories). For EGF starvation, organoids were cultured in media without EGF for 24 respectively . Both were generated by PCR using mouse cDNA encoding the h. For pharmacological activation of Wnt signalling, organoids were cultured in corresponding full-length protein as the template. Lentivirus expressing a con- media supplemented with CHIR99021 (5 μM, Cayman Chemical) for 24 h. For stitutively active form of β-catenin was described previously . Constructs Wnt ligand stimulation, organoids were cultured in media supplemented with expressing shRNAs were made by inserting corresponding oligonucleotides into a −1 mouse recombinant Wnt3a (100 ng ml , R&D Systems). We generated organoids CSII-U6-MCS-puro vector . The target sequences of shRNAs are shown in Sup- expressing the dominant negative form of either EGFR or ErbB2 and adenoma- plementary Table 2. The FRET biosensor was cloned into pCSIIbsr, and dominant derived organoids expressing EKAREV using the aforementioned lentiviruses as negative constructs were cloned into pCSIIpuro, both of which are derived from described previously . All organoids were generated from mice housed in a spe- pCSII-EF (a gift from Hiroyuki Miyoshi) , to produce lentiviruses. For virus ciﬁc pathogen-free facility, and assumed to be free from mycoplasma production, Lenti-X 293T cells (Clontech) were transfected with the pCSII plas- contamination. mids, together with psPAX2 and pCMV-VSV-G-RSV-Rev plasmids. The culture supernatants containing the lentiviruses were collected 48 h after transfection, ﬁl- Image processing. Acquired images were analyzed with MetaMorph software tered, concentrated with the Lenti-X concentrator (Clontech), and then used to 30,68 (Universal Imaging) as described previously . Brieﬂy, FRET efﬁciency was infect intestinal organoids. visualized as FRET/CFP ratio images shown in the intensity-modulated display mode, in which eight colours from red to blue represent the FRET/CFP ratio and Mice. Transgenic mice expressing the FRET biosensor for ERK (EKAREV mice) the 32 grades of colour intensity represent the ﬂuorescence intensity of CFP have been described previously . FRET mice were backcrossed to C57BL/6N Jcl according to the colour scale shown in each ﬁgure. In order to analyze the Z-stack (CLEA Japan) for more than ten generations. To date, no disease or anomaly has images, we performed maximum intensity projection, in which voxels with the been observed in these mice. Fucci2a mice were provided by the RIKEN BRC maximum intensity at each x–y coordinate were projected in the visualization 45 Δ716 through the National Bio-Resource Project of the MEXT, Japan . Apc mice plane. To reduce the noise of in vivo imaging data, a 5 × 5 median ﬁlter was applied have been reported previously . Mice were housed in a speciﬁc pathogen-free to the FRET/CFP ratio images. In addition, time course data of the FRET/CFP facility and provided with a standard diet and water ad libitum. In some experi- values in each cell were smoothened by 6-min moving averages. Quantitative −1 parameters of pulsatile ERK activation were obtained as follows. First, FRET/CFP ments, CHIR99021 (20 mg kg body weight, Cayman Chemical), erlotinib (100 −1 mg kg body weight, Wako Pure Chemical Industries), and/or vehicle (dimethyl values in each cell were calculated for all time frames. Second, the FRET/CFP values were smoothened by 6-min moving averages. Subsequently, the smoothed sulfoxide, DMSO) were intraperitoneally injected daily to 7-week-old C57BL/6 mice for 3 days. No statistical method was used to predetermine sample size. The time series for each cell was ﬁtted to either a ﬂat line or a multi-peak function: experiments were not randomized. The investigators were not blinded to allocation ω ðN ¼ 0Þ during experiments and outcome assessment. The animal protocols were reviewed ytðÞ ¼ ; max y ðÞ t ; ¼ ; y ðÞ t þ ω ðN 1Þ and approved by the Animal Care and Use Committee of Kyoto University fg 1 N 0 Graduate School of Medicine (No. 10584). where N and w indicate the number of ERK activity pulses and basal ERK activity, respectively, max() is a function returning the maximum value of input variables Microscopy. For two-photon excitation microscopy (2PM), we used an and y (t) indicates time-course of the ith ERK activity pulse represented by FV1200MPE-IX83 inverted microscope (Olympus) equipped with a 30×/1.05 NA silicon oil-immersion objective lens (UPLSAPO 30XS; Olympus), an LCV110-MPE y ðÞ t ¼ A φðÞ t t ; i i i i incubator microscope (Olympus) equipped with a 25×/1.05 water-immersion objective lens (XLPLN 25XWMP2; Olympus), and an InSight DeepSee Laser where A and t are the amplitude and timing of the ith ERK pulse, respectively, and i i (Spectra Physics). The laser power was set to 3–18%. The scan speed was set φ ðÞ s is a radial basis function. We here adopted a cosine function within a single between 4–12.5 μs per pixel. Z-stack images were acquired at 1–10 μm intervals. In period, i.e., φ (s)=cos(ω s) +1if π=ω s π=ω or φ (s) = 0 otherwise. Para- i i i i i time-lapse analyses, images were recorded every 1–3 min. The excitation wave- meters (A , t , and w ) were optimized by minimizing the square error between a i i i length for CFP was 840 nm. We used an IR-cut ﬁlter (BA685RIF-3), two dichroic ﬁtting function and the observed ERK activity. If one of the amplitudes A was mirrors (DM505 and DM570), and two emission ﬁlters (BA460-500 for CFP and greater than threshold, cells whose data were ﬁtted to the multi-peak function were BA520-560 for YFP) (Olympus). Confocal images were acquired with an FV1000/ + − deﬁned as “ERK-pulse ”, otherwise they were deﬁned as “ERK-pulse ”. The data IX83 confocal microscope (Olympus) equipped with a 30×/1.05 NA silicon oil- analysis was performed with MATLAB software (MathWorks). immersion objective lens (UPLSAPO 30XS; Olympus). Immunoﬂuorescence staining. Small intestine specimens were washed with cold In vivo observation of the small intestine. Intravital imaging of the small PBS, and ﬁxed overnight in 10% formalin in PBS at 4 °C. The tissues were intestine was performed as described previously . Mice were anaesthetized with sequentially treated with PBS containing 12, 15, and 18% sucrose (for more than 2 1.5–2% isoﬂurane (Abbott) inhalation and placed in the supine position on an h for each treatment), embedded in O.C.T. compound (Tissue-Teck), frozen, and electric heating pad maintained at 37 °C. Before surgery, the abdominal area of the sectioned at 8-μm thickness. Sections were subjected to immunoﬂuorescence mouse was disinfected using 70% ethanol. A small vertical incision was made on staining with the following primary antibodies: anti-EGFR (Medical & Biological the right side of the abdominal wall. The small intestine was pulled out of the Laboratories, clone 6F1), anti-ErbB2 (Cell Signaling Technology, clone 29D8), anti- abdominal cavity, and both proximal and distal sides of the small intestine of phospho-ErbB2 (Cell Signaling Technology, clone 6B12), anit-Ki67 (Abcam, interest were ligated using 5-0 surgical silk sutures (Nesco Suture). After PBS was ab15580), and anti-E-cadherin (Cell Signaling Technology, clone 24E10). The administered into the intestinal cavity with a 29-gauge needle, the small intestine sections were treated with 0.5% Triton X-100 in TBS for 10 min. Antigen retrieval was put on a cover glass placed on a heat-stage maintained at 37 °C, and ﬁxed with was performed by treating samples for 20 min in Tris-EDTA buffer, pH 8.0. Alexa surgical sutures to minimize peristalsis. No drugs were used to stop peristalsis of Fluor 488 conjugated goat anti-rabbit IgG (Molecular Probes, no. A-11008), Alexa the intestinal tract. Instead, we dilated the small intestine by injecting PBS into the Fluor 546 conjugated goat anti-rabbit IgG (Molecular Probes, no. A-11035), or cavity in order to minimize the peristalsis. Time-lapse imaging was then performed Alexa Fluor 488 conjugated goat anti-mouse IgG (Molecular Probes, no. A-11029; with an FV1200MPE-IX83 two-photon excitation microscope under control of the dilution: 1:500) were used as secondary antibodies. Counterstaining was performed −1 FluoView software (Olympus). In some experiments, 0.1 mg kg body weight of with Hoechst 33342 (Molecular Probes, no. H3570). For EdU staining, mice were −1 −1 TPA (LC laboratories) or 5 mg kg body weight of PD0325901 (EMD Millipore) injected intraperitoneally with EdU (80 μgg body weight) 2 h before euthani- −1 was administered via the caudal vein, and 100 mg kg body weight of erlotinib zation. Organoids were treated with 10 μM of EdU for 2 h before staining. EdU (Wako Pure Chemical Industries) was injected intraperitoneally during imaging. staining was performed using the Click-iT EdU Alexa Fluor 488 Imaging Kit (Life Mice were euthanized after the experiments. Technologies, no. C10337), according to the manufacturer’s instructions. 14 NATURE COMMUNICATIONS (2018) 9:2174 DOI: 10.1038/s41467-018-04527-8 www.nature.com/naturecommunications | | | NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-04527-8 ARTICLE Extraction of RNA and qRT-PCR. IECs were isolated by incubating and shaking 13. Qiu, M.-S. & Green, S. H. PC12 cell neuronal differentiation is associated with intestinal specimens in cold PBS containing 5 mM EDTA. Total RNA was prolonged p21ras activity and consequent prolonged ERK activity. Neuron 9, extracted from the isolated IECs or intestinal organoids by using an RNeasy Mini 705–717 (1992). Kit (QIAGEN). The extracted total RNA was reverse-transcribed into cDNA by 14. Fiske, W. H., Threadgill, D. & Coffey, R. J. ERBBs in the gastrointestinal tract: using a High-Capacity cDNA Reverse Transcription Kit (Thermo Fisher Scientiﬁc). recent progress and new perspectives. Exp. Cell Res. 315, 583–601 (2009). Quantitative PCR analyses were performed by using an Applied Biosystems Ste- 15. Barker, N. et al. Identiﬁcation of stem cells in small intestine and colon by pOne Real-Time PCR System and Power SYBR Green PCR Master Mix (Thermo marker gene Lgr5. Nature 449, 1003–1007 (2007). Fisher Scientiﬁc). The expression data were normalized to those of GAPDH or 16. Miettinen, P. J., Berger, J. E., Meneses, J. & Phung, Y. Epithelial immaturity β-actin. and multiorgan failure in mice lacking epidermal growth factor receptor. Nature 376, 337 (1995). 17. Troyer, K. L. et al. Growth retardation, duodenal lesions, and aberrant ileum Microarray analysis. For microarray analysis, we performed two independent architecture in triple null mice lacking EGF, amphiregulin, and TGF-α. series of experiments. Total RNA was extracted from intestinal organoids cultured Gastroenterology 121,68–78 (2001). in the ENR media supplemented with or without CHIR99021 (5 μM) for 4 days as described above. Synthesis of cDNA, in vitro transcription and labelling of cRNA, 18. Wong, V. W. et al. Lrig1 controls intestinal stem-cell homeostasis by negative regulation of ErbB signalling. Nat. Cell Biol. 14, 401–408 (2012). and hybridization to the Mouse Gene 1.0 ST arrays (Affymetrix, Santa Clara, CA, USA) were performed according to the manufacturer's protocols. The CEL ﬁles 19. Powell, A. E. et al. The pan-ErbB negative regulator Lrig1 is an intestinal stem were analyzed by Transcriptome Analysis Console (TAC) 4.0 software (Thermo cell marker that functions as a tumor suppressor. Cell 149, 146–158 (2012). Fisher Scientiﬁc). Expression signals of all genes (probe sets) were calculated using 20. Fearon, E. R. & Vogelstein, B. A genetic model for colorectal tumorigenesis. the RMA algorithm. Gene set enrichment analysis was performed by using GSEA Cell 61, 759–767 (1990). 3.0 software (Broad Institute) . 21. Vogelstein, B. et al. Genetic alterations during colorectal-tumor development. N. Engl. J. Med. 319, 525–532 (1988). 22. Rajagopalan, H. et al. Tumorigenesis: RAF/RAS oncogenes and mismatch- Statistical analysis. Mann–Whitney U-test was performed to analyze the statis- repair status. Nature 418, 934–934 (2002). tical difference between two sets of data. Welch’s t test was performed when the 23. Radinsky, R. et al. Level and function of epidermal growth factor receptor number of samples was less than 6. As a multiple comparison test, the Steel–Dwass predict the metastatic potential of human colon carcinoma cells. Clin. Cancer test was used after validation of homoscedasticity by the Bartlett’s test. BH pro- Res. 1,19–31 (1995). cedure was used for adjustment for multiple comparisons in χ test. P values less 24. Spano, J. P. et al. Impact of EGFR expression on colorectal cancer patient than 0.05 were considered statistically signiﬁcant. No statistical method was used to prognosis and survival. Ann. Oncol. 16, 102–108 (2005). predetermine sample size. All statistical analyses were performed using the JMP Pro 13 software and R software (ver. 3.4.1). 25. Nicholson, R., Gee, J. & Harper, M. EGFR and cancer prognosis. Eur. J. Cancer 37,9–15 (2001). 26. O’Dwyer P., J. & Benson, A. B. 3rd Epidermal growth factor receptor-targeted Data availability. The data that support the ﬁndings of this study are available therapy in colorectal cancer. Semin. Oncol. 29,10–17 (2002). within the article and its Supplementary Information or from the corresponding 27. Fakih, M. G. Metastatic colorectal cancer: current state and future directions. author upon reasonable request. The microarray data have been deposited in the J. Clin. Oncol. 33, 1809–1824 (2015). Gene Expression Omnibus (GEO) database under the accession code GSE110257 28. Zhao, B. et al. Mechanisms of resistance to anti-EGFR therapy in colorectal (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc = GSE110257). cancer. Oncotarget 8, 3980–4000 (2017). 29. Sato, T. et al. Single Lgr5 stem cells build crypt-villus structures in vitro Received: 11 August 2017 Accepted: 2 May 2018 without a mesenchymal niche. Nature 459, 262–265 (2009). 30. Kamioka, Y. et al. Live imaging of protein kinase activities in transgenic mice expressing FRET biosensors. Cell Struct. Funct. 37,65–73 (2012). 31. Holderﬁeld, M., Nagel, T. & Stuart, D. Mechanism and consequences of RAF kinase activation by small-molecule inhibitors. Br. J. Cancer 111, 640–645 (2014). 32. Joslin, E. J. et al. Structure of the EGF receptor transactivation circuit References integrates multiple signals with cell context. Mol. Biosyst. 6, 1293–1306 (2010). 1. Shaul, Y. D. & Seger, R. The MEK/ERK cascade: from signaling speciﬁcity to 33. Fan, H. & Derynck, R. Ectodomain shedding of TGF‐α and other diverse functions. Biochim. Biophys. Acta 1773, 1213–1226 (2007). transmembrane proteins is induced by receptor tyrosine kinase activation and 2. Dhillon, A. S., Hagan, S., Rath, O. & Kolch, W. MAP kinase signalling MAP kinase signaling cascades. EMBO J. 18, 6962–6972 (1999). pathways in cancer. Oncogene 26, 3279–3290 (2007). 34. Lito, P., Rosen, N. & Solit, D. B. Tumor adaptation and resistance to RAF 3. Schneider, M. R. & Yarden, Y. The EGFR-HER2 module: a stem cell approach inhibitors. Nat. Med. 19, 1401–1409 (2013). to understanding a prime target and driver of solid tumors. Oncogene 35, 35. Avraham, R. & Yarden, Y. Feedback regulation of EGFR signalling: decision 2949–2960 (2016). making by early and delayed loops. Nat. Rev. Mol. Cell Biol. 12, 104–117 4. Dokala, A. & Thakur, S. Extracellular region of epidermal growth factor (2011). receptor: a potential target for anti-EGFR drug discovery. Oncogene 36, 36. Cirit, M., Wang, C. C. & Haugh, J. M. Systematic quantiﬁcation of negative 2337-2344 (2017). feedback mechanisms in the extracellular signal-regulated kinase (ERK) 5. Wieduwilt, M. J. & Moasser, M. M. The epidermal growth factor receptor signaling network. J. Biol. Chem. 285, 36736–36744 (2010). family: biology driving targeted therapeutics. Cell. Mol. Life Sci. 65, 1566–1584 37. Diaz-Rodriguez, E., Montero, J. C., Esparis-Ogando, A., Yuste, L. & Pandiella, (2008). A. Extracellular signal-regulated kinase phosphorylates tumor necrosis factor 6. Marshall, C. J. Speciﬁcity of receptor tyrosine kinase signaling: transient versus alpha-converting enzyme at threonine 735: a potential role in regulated sustained extracellular signal-regulated kinase activation. Cell 80, 179–185 shedding. Mol. Biol. Cell 13, 2031–2044 (2002). (1995). 38. Umata, T. et al. A dual signaling cascade that regulates the ectodomain 7. Komatsu, N. et al. Development of an optimized backbone of FRET biosensors shedding of heparin-binding epidermal growth factor-like growth factor. J. for kinases and GTPases. Mol. Biol. Cell 22, 4647–4656 (2011). Biol. Chem. 276, 30475–30482 (2001). 8. Aoki, K. et al. Stochastic ERK activation induced by noise and cell-to-cell 39. Lordick, F. & Janjigian, Y. Y. Clinical impact of tumour biology in the propagation regulates cell density-dependent proliferation. Mol. Cell 52, management of gastroesophageal cancer. Nat. Rev. Clin. Oncol. 13, 348–360 529–540 (2013). (2016). 9. Albeck, J. G., Mills, G. B. & Brugge, J. S. Frequency-modulated pulses of ERK 40. Livneh, E. et al. Reconstitution of human epidermal growth factor receptors activity transmit quantitative proliferation signals. Mol. Cell 49, 249–261 and its deletion mutants in cultured hamster cells. J. Biol. Chem. 261, (2013). 12490–12497 (1986). 10. Matsubayashi, Y., Ebisuya, M., Honjoh, S. & Nishida, E. ERK activation 41. Jones, F. E. & Stern, D. F. Expression of dominant-negative ErbB2 in the propagates in epithelial cell sheets and regulates their migration during wound mammary gland of transgenic mice reveals a role in lobuloalveolar healing. Curr. Biol. 14, 731–735 (2004). development and lactation. Oncogene 18, 3481–3490 (1999). 11. Hiratsuka, T. et al. Intercellular propagation of extracellular signal-regulated 42. Oshima, M. et al. Loss of Apc heterozygosity and abnormal tissue building in kinase activation revealed by in vivo imaging of mouse skin. eLife 4, e05178 nascent intestinal polyps in mice carrying a truncated Apc gene. Proc. Natl (2015). Acad. Sci. USA 92, 4482–4486 (1995). 12. Greene, L. A. & Tischler, A. S. Establishment of a noradrenergic clonal line of 43. Okuchi, Y. et al. Identiﬁcation of aging-associated gene expression signatures rat adrenal pheochromocytoma cells which respond to nerve growth factor. that precede intestinal tumorigenesis. PLoS ONE 11, e0162300 (2016). Proc. Natl Acad. Sci. USA 73, 2424–2428 (1976). NATURE COMMUNICATIONS (2018) 9:2174 DOI: 10.1038/s41467-018-04527-8 www.nature.com/naturecommunications 15 | | | ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-04527-8 44. Shih, I.-M. et al. Evidence that genetic instability occurs at an early stage of 65. Miyoshi, H., Blömer, U., Takahashi, M., Gage, F. H. & Verma, I. M. colorectal tumorigenesis. Cancer Res. 61, 818–822 (2001). Development of a self-inactivating lentivirus vector. J. Virol. 72, 8150–8157 45. Mort, R. L. et al. Fucci2a: a bicistronic cell cycle reporter that allows Cre (1998). mediated tissue speciﬁc expression in mice. Cell Cycle 13, 2681–2696 (2014). 66. Mizuno, R. et al. In vivo imaging reveals PKA regulation of ERK activity 46. Gregorieff, A. & Clevers, H. Wnt signaling in the intestinal epithelium: from during neutrophil recruitment to inﬂamed intestines. J. Exp. Med. 211, endoderm to cancer. Genes Dev. 19, 877–890 (2005). 1123–1136 (2014). 47. Ding, Z. et al. A novel signaling complex between TROY and EGFR mediates 67. Koo, B.-K. et al. Controlled gene expression in primary Lgr5 organoid glioblastoma cell invasion. Mol. Cancer Res. 16, 322-332 (2018). cultures. Nat. Methods 9,81–83 (2012). 48. Chen, J. et al. EGFL6, a potential novel ligand of EGFR, play roles in 68. Aoki, K. & Matsuda, M. Visualization of small GTPase activity with Nasopharyngeal cacinoma metastasis through establishing invasive and long- ﬂuorescence resonance energy transfer-based biosensors. Nat. Protoc. 4, distant metastatic niche by paracrine and autocrine. Cancer Res. 75, 118–118 1623–1631 (2009). (2015). 69. Subramanian, A. et al. Gene set enrichment analysis: a knowledge-based 49. Cai, M. et al. Inhibition of LRIG3 gene expression via RNA interference approach for interpreting genome-wide expression proﬁles. Proc. Natl Acad. modulates the proliferation, cell cycle, cell apoptosis, adhesion and invasion of Sci. USA 102, 15545–15550 (2005). glioblastoma cell (GL15). Cancer Lett. 278, 104–112 (2009). 50. Fiori, J. L. et al. Filamin A modulates kinase activation and intracellular Acknowledgements trafﬁcking of epidermal growth factor receptors in human melanoma cells. We are grateful to H. Miyoshi (Division of Experimental Therapeutics, Graduate School Endocrinology 150, 2551–2560 (2009). of Medicine, Kyoto University) for technical advice. We thank K. Hirano, K. Takakura, 51. Ebisuya, M., Kondoh, K. & Nishida, E. The duration, magnitude and S. Kobayashi, S. Fujiwara, N. Nishimoto, Y. Takeshita, A. Kawagishi, and the Medical compartmentalization of ERK MAP kinase activity: mechanisms for providing Research Support Center of Kyoto University for technical assistance. We are also signaling speciﬁcity. J. Cell Sci. 118, 2997–3002 (2005). grateful to the members of the Matsuda Laboratory for their helpful input. M.M. was 52. Wainstein, E. & Seger, R. The dynamic subcellular localization of ERK: supported by the Nakatani Foundation, CREST JPMJCR1654 and JSPS KAKENHI Grant mechanisms of translocation and role in various organelles. Curr. Opin. Cell Numbers 15H02397, 15H05949 “Resonance Bio”, and 16H06280 “ABiS”. M.I. was Biol. 39,15–20 (2016). supported by the Takeda Science Foundation and JSPS KAKENHI Grant Num- 53. Wells, A. et al. Ligand-induced transformation by a noninternalizing bers 16K21106, 18H05100, and 18K06929. epidermal growth factor receptor. Science 247, 962–964 (1990). 54. Klapper, L. N. et al. The ErbB-2/HER2 oncoprotein of human carcinomas may function solely as a shared coreceptor for multiple stroma-derived growth Author contributions factors. Proc. Natl Acad. Sci. USA 96, 4995–5000 (1999). Y.M., M.I., K.A., T.C., H.S. and M.M. designed the project. Y.M. and M.I. performed all 55. Baulida, J., Kraus, M. H., Alimandi, M., Di Fiore, P. P. & Carpenter, G. All the experiments. Y.F. developed an algorithm for image processing. K.S. and M.M.T. ErbB receptors other than the epidermal growth factor receptor are developed transgenic mice. A.S. performed microarray analyses. Y.M. analyzed the data. endocytosis impaired. J. Biol. Chem. 271, 5251–5257 (1996). Y.M., M.I., and M.M. prepared the manuscript. M.I. and M.M. supervised the study. 56. Worthylake, R., Opresko, L. K. & Wiley, H. S. ErbB-2 ampliﬁcation inhibits down-regulation and induces constitutive activation of both ErbB-2 and epidermal growth factor receptors. J. Biol. Chem. 274, 8865–8874 Additional information Supplementary Information accompanies this paper at https://doi.org/10.1038/s41467- (1999). 57. Lenferink, A. E. et al. Differential endocytic routing of homo‐and hetero‐ 018-04527-8. dimeric ErbB tyrosine kinases confers signaling superiority to receptor heterodimers. EMBO J. 17, 3385–3397 (1998). Competing interests: The authors declare no competing interests. 58. Sparta, B. et al. Receptor level mechanisms are required for epidermal growth factor (EGF)-stimulated extracellular signal-regulated kinase (ERK) activity Reprints and permission information is available online at http://npg.nature.com/ pulses. J. Biol. Chem. 290, 24784–24792 (2015). reprintsandpermissions/ 59. Valdez, G. et al. Trk-signaling endosomes are generated by Rac-dependent macroendocytosis. Proc. Natl Acad. Sci. USA 104, 12270–12275 (2007). Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims in 60. Sartore-Bianchi, A. et al. Dual-targeted therapy with trastuzumab and published maps and institutional afﬁliations. lapatinib in treatment-refractory, KRAS codon 12/13 wild-type, HER2- positive metastatic colorectal cancer (HERACLES): a proof-of-concept, multicentre, open-label, phase 2 trial. Lancet Oncol. 17, 738–746 (2016). 61. Powell, A. E. et al. Inducible loss of one Apc allele in Lrig1-expressing Open Access This article is licensed under a Creative Commons progenitor cells results in multiple distal colonic tumors with features of Attribution 4.0 International License, which permits use, sharing, familial adenomatous polyposis. Am. J. Physiol. Gastrointest. Liver Physiol. adaptation, distribution and reproduction in any medium or format, as long as you give 307, G16–G23 (2014). appropriate credit to the original author(s) and the source, provide a link to the Creative 62. Basak, O. et al. Induced quiescence of Lgr5+stem cells in intestinal organoids Commons license, and indicate if changes were made. The images or other third party enables differentiation of hormone-producing enteroendocrine cells. Cell Stem material in this article are included in the article’s Creative Commons license, unless Cell 20, 177–190 (2017). e174. indicated otherwise in a credit line to the material. If material is not included in the 63. Imajo, M., Miyatake, K., Iimura, A., Miyamoto, A. & Nishida, E. A molecular article’s Creative Commons license and your intended use is not permitted by statutory mechanism that links Hippo signalling to the inhibition of Wnt/β‐catenin regulation or exceeds the permitted use, you will need to obtain permission directly from signalling. EMBO J. 31, 1109–1122 (2012). the copyright holder. To view a copy of this license, visit http://creativecommons.org/ 64. Okamoto, K., Fujisawa, J., Reth, M. & Yonehara, S. Human T-cell leukemia licenses/by/4.0/. virus type-I oncoprotein Tax inhibits Fas-mediated apoptosis by inducing cellular FLIP through activation of NF-kappaB. Genes Cells 11, 177–191 (2006). © The Author(s) 2018 16 NATURE COMMUNICATIONS (2018) 9:2174 DOI: 10.1038/s41467-018-04527-8 www.nature.com/naturecommunications | | |
Nature Communications – Springer Journals
Published: Jun 5, 2018
It’s your single place to instantly
discover and read the research
that matters to you.
Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.
All for just $49/month
Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly
Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.
All the latest content is available, no embargo periods.
“Whoa! It’s like Spotify but for academic articles.”@Phil_Robichaud