The Crick Institute
The Crick Institute
News Article | November 22, 2016
Scientists from the Crick Institute, London and the Hebrew University, Jerusalem, discover a protein that plays a key role in turning cancer tumor cells into cancer stem cells that are able to renew outbreaks of the disease An international study led by scientists from the Crick Institute in London and the Hebrew University of Jerusalem revealed a survival mechanism in cancer cells that allows the disease to erupt again even after aggressive treatment. In a paper published in Science (LINK) the researchers describe the mechanism by which cancer tumor cells become cancer stem cells that can sustain long-term growth. When cancer develops, the generated cells are not uniform in their biological properties and contribute differently to tumor development. Only a small portion of cancer cells can form new tumors or metastases, and these are called "cancer stem cells". This disparity between tumor cells poses major challenges in understanding the nature of the tumor, its sensitivity to drugs, and planning an effective treatment that will eliminate all tumor cells. "Many chemotherapy drugs leave a small amount of cancer stem cells that cause a renewed outbreak of the disease after a few years. It is therefore important to identify cancer stem cells in tumors and characterize the differences between the different tumor cells as the basis for detecting weak spots in the course of the development of the disease," explained Prof. Eran Meshorer, head of the Laboratory for stem cells and epigenetics in the Institute of Life Sciences and a member of the Edmond and Lily Safra Center for Brain Sciences (ELSC) of The Hebrew University of Jerusalem. Cancer stem cells are not limited to the tumor itself and they are able to engage again in healthy environment and stimulate the disease. To study the characteristics of those unique cells, Prof. Meshorer and doctoral student Alva Biran from the Hebrew University teamed up with Dr. Paula Scaffidi and Christina Morales Torres from The Crick Institute in London. The international research team also included Dr. Ayelet Hashahar Cohen of the Hebrew University, Dr. Rotem Ben-Hamo and Professor Sol Efroni from Bar-Ilan University, and Dr. Tom Misteli from the National Cancer Institute, NIH. The research team found that in a number of cancer types, those cancer stem cells lose one of their DNA packaging proteins - H1.0. By binding to DNA, H1.0 silences the expression of the genes it binds to. "We found that the disappearance of H1.0 is crucial for the cancer cells to remain immortal. To understand the mechanism of action, we mapped its interaction with DNA and found that it binds to the genes' regulatory regions. When H1.0 levels go down, the genes to which it binds can be activated. These genes, it turns out, are the ones which provide the cancer cell with its immortal potential," explained Prof. Meshorer. The study is based on epigenetics - a scientific field that investigates gene expression in DNA by switching genes on and off. In order to identify the cancer stem cells from other cells in the tumor, the research team studied epigenetic mechanisms that distinguish between the least-sorted cells, with endless division properties and a potential to create growth, and the more sorted cells which lack this ability. The results showed an inverse relation between H1.0 and the division of cancer cells: "As the H1.0 levels fall, the greater the potential of uncontrolled division of cells. In contrast, high levels of the protein prevent this process. We found that the disappearance of protein H1.0 is characteristic of cancer stem cells and it is necessary to maintain the ability of partition and the potential for growth creation." The discovery could open the door for medical intervention in cancer stem cells aimed at the restoration of high levels of H1.0 in all cancer cells and by that blocking the differentiation of cancer cells. While further research is needed to understand the effectiveness of H1.0 protein in preventing the spread of cancer growth, this research advances significantly the study of the mechanisms of cancer stem cells and the relatively new epigenetic approach to cancer research
News Article | December 16, 2015
The MMTV-PyMT mice were a gift from E. Sahai, MMTV-PyMT actin-GFP (mice expressing green fluorescent protein under the control of the actin promoter), Gcsf-null and Rag1-null mice were a gift from J. Huelsken, MMTV-PyMT actin-luciferase (mice expressing firefly luciferase under the control of the actin promoter) transgenic line was a gift from D. Bonnet, Rosa26R-eGFP-DTA mice were a gift from C. Reis e Sousa. Ela2-Cre knock-in mice and Alox5-null mice were purchased from European Mouse Mutant Archive (EMMA) and Jackson Laboratory, respectively. All mouse strains have been described previously30, 31, 32, 33, 34, 35, 36, 37. All strains of mice were in >10 generations FVB/N and/or C57BL/6 background except Gcsf-null, Ela2-Cre and Rosa26R-eGFP-DTA mice that were used in mixed background with littermate controls. Female mice were used between 6–9 weeks of age, except spontaneous cancer models. Breeding and all animal procedures were performed in accordance with UK Home Office regulations under project license PPL/80/2531. Where applicable, mice were anaesthetized with IsoFlo (isoflurane, Abbott Animal Health) and temporally treated with the analgesics Vetergesic (Alstoe Animal Health) and/or Rimadyl (Pfizer Animal Health). For tumour studies under the project licence PPL/80/2531, the overruling determinant was animal welfare. The National Cancer Research Institute (NCRI) Guidelines for the Welfare and Use of Animals in Cancer Research were followed. When assessing primary tumour growth, a mean diameter of 1.5 cm for single tumours was not exceeded. However, for multifocal disease such as MMTV-PyMT cancer, provided that there were no additional adverse welfare consequences for the animal, the total superficial tumour burden was allowed to exceed these dimensions when essential for the achievement of the scientific objective, namely spontaneous metastasis. Mice were monitored daily for signs of adverse effects. The source data for primary tumour growth are in Supplementary Fig. 3. FVB/N wild-type mice were used for MMTV-PyMT tumour cell transplantations to isolate lung neutrophils. Rag1-null mice were used when using human or mouse GFP or luciferase-expressing tumour cells. Primary MMTV-PyMT, MMTV-PyMT actin-GFP or MMTV-PyMT actin-luciferase cells (105–106 cells per injection), the unmarked or stably mouse phosphoglycerate kinase 1 (PGK) promoter-GFP-expressing mouse mammary cancer cell line 4T1 (105 cells per injection) and the unmarked or stably actin-GFP-expressing human breast cancer cell line MDA-MB-231 (1–2 × 106 cells per injection) were used. For experimental metastasis, tumour cells were re-suspended in 100 μl PBS and tail vein injected. For orthotopic transplantations, tumour cells were re-suspended in 50 μl growth-factor-reduced Matrigel (Costar) and transplanted within the fourth mammary fat pad on both flanks (MMTV-PyMT and MDA-MB-231 cells) or one flank only (4T1 cells). MMTV-PyMT+ mice that spontaneously developed a primary tumour and had visible lung metastasis were used to determine immune cell presence in the lung and neutrophil presence in other organs together with tumour-free littermate controls. For determination of timing and dynamics of lung infiltration by neutrophils and cancer cells, MMTV-PyMT+ mice harbouring 1.5–2 g spontaneously developed tumours were used. Neutrophil infiltration was quantified by flow cytometry and histological staining of lung sections for S100A9 and cancer cell presence by examination of six histological lung sections (100 μm apart) for PyMT staining to confirm the pre-metastatic state. The timing of neutrophil infiltration into the pre-metastatic lung before cancer cells was confirmed in FVB/N wild-type mice carrying two primary tumours originating from orthotopic injection of primary MMTV-PyMT cancer cells and used for analysis (daily treated with anti-Ly6G or control IgG antibody starting 24 h before tumour cell implantation). Mice were culled and analysed about 6 weeks after spontaneous primary tumour onset; no differences were observed in tumour onset among the different genotypes. Rat anti-Ly6G antibody38, 39 (12.5 μg per mouse; clone 1A8 from BioXcell) or rat IgG isotype control (provided by the Cell Services Unit of The Crick Institute) in 100 μl saline were administered daily via intraperitoneal injection. Zil (LKT Laboratories) dissolved in DMSO (Sigma) or DMSO alone was fed to mice by pipetting on the back of the tongue once a day at a dosage of 100 μg Zil per g mouse weight. Rag1-null mice were orthotopically transplanted with unlabelled mammary tumour cells 4 weeks before labelled tumour cell injection via the tail vein (MMTV-PyMT and 4T1 105 cells, MDA-MB-231 106 cells). Anti-Ly6G or Zil treatment for 2 weeks (except 4T1, 10 days) started 1 day before intravenous injection of cancer cells. Then, total primary tumour burden, neutrophil presence in the lung, spontaneous lung metastasis incidence from the transplanted primary tumour and/or experimentally induced lung metastasis originating from the intravenously injected cancer cells was analysed. Of note, exclusively experimental metastasis are present in lung harbouring MDA-MB-231 cells, while predominantly spontaneous metastases are visible in lung harbouring 4T1 cells due to the high spontaneous metastasis rate of primary 4T1 tumours. Only GFP+ experimental metastasis induced by cancer cell injection was quantified in these experiments. Primary MMTV-PyMT cells were either cell sorted for BLT2 and/or CysLT2 presence or absence, or treated for 3 days on collagen-coated dishes with either neutrophil-conditioned medium or LTB4 and LTC/D/E4. Subsequently, 103–104 cells were orthotopically transplanted into the mammary gland or 106 cells injected via the tail vein into Rag1-null mice and mammary tumour growth or lung metastasis incidence analysed about 3 weeks thereafter. To analyse total cancer cells at early stages, Rag1-null mice were injected with 0.5–1 × 106 MMTV-PyMT actin-GFP cells via the tail vein followed 12 h later by intravenous injection of 25 × 106 neutrophils (freshly isolated from MMTV-PyMT tumour-transplanted mice) or 12, 24 and 36 h later by intravenous injection of 200 μl lung neutrophil-conditioned or control sphere medium (described later). Cancer cells in the lung were analysed 3 days after the initial tumour cell injection for frequencies of CD90+ MICs among GFP+CD24+ (non-MIC) cancer cells. For determination of effects of neutrophils or neutrophil-conditioned medium on metastatic burden, Rag1-null mice were intravenously injected with 1–10 × 105 MMTV-PyMT actin-GFP or actin-luciferase cells followed immediately, 2 and 4 days later, by injection of 25 × 106 neutrophils or 3–5 times every 12 h by injection of 200 μl lung neutrophil-conditioned medium. Metastatic burden was determined by flow cytometric analysis of GFP+ cancer cells 1 week or bioluminescence imaging of luciferase+ cancer cells 2–4 weeks thereafter, respectively. Rag1-null mice were transplanted with 106 Gcsf-null primary MMTV-PyMT cancer cells into two mammary glands and tumour growth, spontaneous metastatic incidence and neutrophil presence in the lung were analysed 4 weeks thereafter. C57BL/6 wild-type mice were lethally irradiated (dosage: 2× 600 rad, 4 h apart) and 24 h later injected via the tail vein with 2 × 106 bone marrow cells freshly isolated from C57BL/6 or Alox5-null donor mice. Bone marrow chimaeric mice were orthotopically transplanted with 106 MMTV-PyMT cells into the fourth mammary fat pad on both sides 8 weeks after bone marrow reconstitution and primary tumour size, neutrophil infiltration into the lung and lung metastasis were analysed 6 weeks later. Chimaeric mice were generated in a pure C57BL/6 background, therefore MMTV-PyMT cells from the same background were used to generate primary tumours. In this lower tumorigenic background, metastasis only occurs in about 50% of the mice. No alteration in this penetrance was observed between wild-type and Alox5-null bone marrow chimaeric mice. Figure 4b quantifies animals harbouring metastatic disease. Percentage of bone marrow reconstitution was calculated by isolating total DNA from bone marrow of chimaeras and semi-quantitative PCR with a calibration curve from 100% wild-type DNA mixed at defined ratios with 100% Alox5-null DNA. PCR was performed using Redtag reagents (Sigma) (primers are listed in Supplementary Information) and 25 amplification cycles before loading an agarose gel. Ratio between wild-type and Alox5-null band was calculated for every mouse and percentage chimaerism was determined by comparison with calibration curve. Chimaerism was consistently between 80 and 96%. Mice inoculated with actin-luciferase-expressing MMTV-PyMT cells were shaved around the chest area and injected with 3 mg XenoLight D-luciferin potassium salt (PerkinElmer) in PBS into the peritoneum 5 min before imaging for at least 45 min using the IVIS Spectrum Pre-clinical In vivo Imaging System (PerkinElmer). The maximum bioluminescence intensity signal for the lung of every mouse was determined using Living Image 4.3.1 software. Mouse lung tissue was fixed in 4% paraformaldehyde in PBS for 24 h and embedded in paraffin blocks. Four-micrometre sections were stained. The breast cancer tissue array paired with metastatic tumours, 96 samples (1.5 mm), was purchased from Abcam (ab178118). H&E staining was performed according to standard procedures. For immunohistochemistry, either secondary horseradish peroxidase (HRP)-conjugated antibodies were used in combination with DAP Peroxidase substrate or the VECTASTAIN ABC kit (all Vector Laboratories) according to the manufacturer’s instructions. Specific primary antibodies were used (see Supplementary Information), visualization of cell nuclei was performed with haematoxylin and analysis employed the Nikon Eclipse 90i light microscope and NIS-elements software. MMTV-PyMT cell isolation was described in detail previously8. In brief, primary MMTV-PyMT tumours, liver, spleen and lung were dissected, minced, and digested with Liberase (Roche) and DNaseI (Sigma) in HBSS and passed through a 100 μm cell strainer. Some tumour cells were used for cell culture at this point. Bone marrow cells were isolated by crushing the femur and tibia and blood collected via bleeding from the tail vein with heparin (Sigma) as a coagulant. For flow cytometric analysis or further purification, single-cell suspensions of tumour, liver, spleen, lung, bone marrow and blood were subjected to hypotonic lysis (Red Blood Cell Lysis Solution, Miltenyi) to remove erythrocytes and washed with 1×PBS/2 mM EDTA/0.5% BSA. Prepared single-cell suspensions of mouse tissues and in vitro treated cancer cells were incubated with mouse FcR Blocking Reagent (Miltenyi) followed by incubation with (a combination) of specific pre-labelled antibodies or in combination with fluorescently labelled secondary antibodies (Invitrogen) (see Supplementary Information). Dead cells were stained with 4′,6-diamidino-2-phenylindole (DAPI) or propidium iodide (PI; both Sigma). The LSRFortessa cell analyser running FACSDiva software (BD Biosciences) and FlowJo software was used. Tumour cells were flow-sorted using the Influx cell sorter running FACS Sortware sorter software (BD Biosciences). MMTV-PyMT cells were used in experiments immediately after sorting and sorted 4T1 cells cultured in adherent conditions for 3 days before western blot analysis. Freshly isolated lung cells from wild-type mice orthotopically transplanted with MMTV-PyMT tumours were incubated with mouse FcR Blocking Reagent (Miltenyi), APC-coupled anti-Ly6G (clone 1A8) antibody (BD Bioscience) followed by incubation with magnetic anti-APC microbeads (Miltenyi). Magnetically labelled neutrophils were isolated using LS columns (Miltenyi) according to the manufacturer’s instructions. Neutrophil purity and viability was measured by flow cytometry. Some isolated Ly6G+ cells were smeared onto a glass slide and air-dried overnight followed by H&E staining to evaluate cell morphology. Remaining neutrophils were kept in sphere medium at a concentration of 106 neutrophils per 150 μl medium for 14 h to allow conditioning. Neutrophils and cell debris were removed by centrifugation and conditioned medium occasionally snap-frozen before use. All used cell lines were provided by the Cell Services Unit of The Crick Institute, which routinely tests for Mycoplasma contamination and were not further authenticated in our laboratory. Cell lines were cultured in DMEM medium supplemented with 10% fetal bovine serum (DMEM/FCS, both Invitrogen). Freshly isolated MMTV-PyMT cells were cultured overnight on PureCol collagen (Advanced Biomatrix)-coated dishes in growth medium DMEM/F12 with 2% FBS, 20 ng ml−1 EGF (Invitrogen) and 10 μg ml−1 insulin (Sigma) before use in experiments. All in vitro and in vivo experiments involving primary MMTV-PyMT cells were performed with at least two primary tumour cell preparations from different spontaneous MMTV-PyMT+ mice. Unless otherwise specified, each in vitro and in vivo experiment was performed with a different tumour cell preparation. Primary MMTV-PyMT cells were cultured in sphere medium on collagen-coated dishes, 4T1 and MDA-MB-231 cells in DMEM/FCS on uncoated dishes or in non-attachment conditions for the indicated periods of time under presence of (as indicated for every experiment): control sphere medium, neutrophil-conditioned medium, 100% ethanol control (EtOH, Sigma), DMSO control, 1 μM LTB4, 100 nM LTC/D/E4 (Cysteinyl Leukotriene HPLC Mixture I), 3 μM BLT2 inhibitor LY255283, 0.3 μM CysLT2 inhibitor BAY-u9773 (all Cayman Chemical), 1 μM Zil and/or 1 nM pan-MEK inhibitor PD0325901 (provided by J. Downwards) followed by further tests or analysis. The sphere formation assay was described previously8. In brief, 104 total MMTV-PyMT or flow-sorted cells per well were plated in ultra-low-attachment 96-well plates (Costar) in 100 μl sphere medium DMEM/F12 supplemented with B-27, 20 ng ml−1 EGF, 20 ng ml−1 FGF (all Invitrogen) and 4 μg ml−1 heparin (Sigma) or neutrophil-conditioned medium. After 7–10 days, if not otherwise indicated, all formed spheres were quantified from images taken with the inverted Leica DM IRBE light and fluorescence microscope. The area of the plane passing through the sphere centre was measured for every sphere (sphere size) using ImageJ software and the areas of all formed spheres were summed up. The obtained number was divided by total number of plated cells. This value represents the sphere formation index (SFI) per cell for every experimental group. Freshly isolated MMTV-PyMT cells were either only treated for 3 days in adherent conditions before sphere assay or directly treated during the sphere assay with neutrophil-conditioned medium or LTB4 and/or LTC/D/E4 or Zil, as indicated. When cells were passaged, cells were quantified by cell counting and re-plated in equal numbers per well for the next passage approximately every 7 to 10 days. Rag1-null mice carrying MMTV-PyMT tumours were treated daily for 3 days with Zil and intravenously injected with 105 GFP-expressing MMTV-PyMT cancer cells. BrdU (1 mg per mouse) was intraperitoneally injected 18 h after GFP+ cancer cells and lungs were harvested and digested 6 h later. In vitro 3-day MMTV-PyMT or 4T1 cells treated as indicated in adherent conditions were pulsed with 30 μM BrdU (Sigma) for 3 h and harvested. Cells were incubated with fluorescently labelled anti-CD24 and/or anti-CD90.1 antibody if indicated. BrdU Flow Kit (BD Bioscience) was used for staining followed by analysis by flow cytometry. Primary MMTV-PyMT cells were cultured on collagen-coated dishes for 3 days supplemented with either LTB4 and LTC/D/E4 or Zil followed by incubation with fluorescently labelled anti-CD90.1 and anti-CD24 antibodies and analysis by flow cytometry. 4T1 and MDA-MB-231 cell lines were cultured in DMEM/FCS supplemented with LTB4 and LTC/D/E4 for 3 days in adherent conditions followed by either staining with fluorescently labelled anti-CD49f, anti-BLT2, anti-CysLT2 and/or anti-CD44 antibodies or using the ALDEFLUOR kit (StemCell Technologies) according to the manufacturer’s instructions and analysed by flow cytometry. Neutrophils were freshly isolated from the lungs of wild-type or MMTV-PyMT tumour-bearing mice. RNA isolation was performed using MagMAX-96 Total RNA Isolation Kit and cDNA synthesis using SuperScript III Reverse Transcriptase. Quantitative PCR reactions were performed using EXPRESS SYBR GreenER reagents with the Applied Biosystems 7500 Fast Real-Time PCR System (all Invitrogen) and specific primers (see Supplementary Information). Ethanol was used to precipitate protein from cell culture medium before analysis using either the enzyme immunoassays (EIAs) LTC/D/E4 Biotrak EIA System (Amersham) or the LTB4 EIA Kit (Cayman Chemical) according to the manufacturer’s instructions. Primary MMTV-PyMT cells grown on collagen-coated dishes were cultured overnight in DMEM/F12 with B-27, and 4 μg ml−1 heparin (Sigma) before treatment with 1 μM LTB4 or 100 nM LTC/D/E4. Unsorted or sorted LTR-reduced 4T1 cells were stimulated with LTB4, LTC/D/E4, BLT2 inhibitor LY255283 and/or CysLT2 inhibitor BAY-u9773 as indicated. Cells were washed and protein isolated using RIPA buffer (25 mM Tris-hydrogen chloride pH 7.6, 50 mM sodium chloride, 1% NP-40, 1% sodium deoxycholate, 0.1% soduim dodecyl sulfate) freshly supplemented with 1 μM sodium pyrophosphate, 1 μM B-glycerophosphate, 1 μM sodium vanadiumoxide, 1 μM sodium fluoride, 1 μM sodium molybdate (all Sigma) and cOmplete ULTRA Tablets (Roche), and processed by standard western blot techniques. Membranes were blocked with 5%BSA in PBS with 0.5% Tween-20 (Sigma) and incubated with specific primary antibodies (see Supplementary Information). ECL Western Blotting System including secondary antibodies and Hyperfilm ECL (both Amersham) were used. Protein lysates of 3 h LTB4-stimulated MDA-MB-231 cells were analysed using the Proteome Profiler Human Phospho-Kinase Array Kit (R&D systems) according to the manufacturer’s instructions. Western blot quantification was performed on scanned films using ImageJ software. Data analyses used GraphPad Prism version 7. The data are presented as mean ± standard error of the mean, individual values, ‘scatter plot with Tukey box and whiskers’ and/or ‘scatter plot with column bar’ graphs and were analysed using Student’s t-tests (paired or unpaired according to the experimental setting), Mann–Whitney tests, one-sample t-tests and two-way ANOVA as indicated in the legends. Data were pooled from at least two experiments, except Fig. 4c, i, k and Extended Data Figs 2d, 4d–m, 5a, b, f, h, 6k, 10e, in which data are at least biological triplicates generated in parallel. Two-way ANOVA was performed when the control groups between experiments were significantly different. Western blot in Extended Data Fig. 8i, k, the proteome profiler dot blot in Extended Data Fig. 8d and BrdU incorporation of 4T1 cells in Extended Data Fig. 10k were performed once. Extended Data Fig. 3b (mRNA expression) compares biological triplicates of the pre-metastatic to a representative control (wild-type) value. The experiments were not randomized and there was no blinding as animals or samples were marked. No statistical methods were used to predetermine sample sizes. Sample sizes were based on previous experience with the models8, 14. n values represent biological replicates, with the exception of the sphere assays, for which both technical and biological replicates are shown. Differences were considered significant when P < 0.05 and are indicated as NS, not significant, *P < 0.05, **P < 0.01, ***P < 0.001.
Peng T.,Rockefeller University |
Thinon E.,Rockefeller University |
Thinon E.,The Crick Institute |
Hang H.C.,Rockefeller University
Current Opinion in Chemical Biology | Year: 2016
Protein fatty-acylation in eukaryotes has been associated with many fundamental biological processes. However, the diversity, abundance and regulatory mechanisms of protein fatty-acylation in vivo remain to be explored. Herein, we review the proteomic analysis of fatty-acylated proteins, with a focus on N-myristoylation and S-palmitoylation. We then highlight major challenges and emerging methods for direct site identification, quantitation, and lipid structure characterization to understand the functions and regulatory mechanisms of fatty-acylated proteins in physiology and disease. © 2015 Elsevier Ltd.
Thinon E.,Rockefeller University |
Thinon E.,The Crick Institute |
Percher A.,Rockefeller University |
Hang H.C.,Rockefeller University
ChemBioChem | Year: 2016
Dietary unsaturated fatty acids, such as oleic acid, have been shown to be covalently incorporated into a small subset of proteins, but the generality and diversity of this protein modification has not been studied. We synthesized unsaturated fatty-acid chemical reporters and determined their protein targets in mammalian cells. The reporters can induce the formation of lipid droplets and be incorporated site-specifically onto known fatty-acylated proteins and label many proteins in mammalian cells. Quantitative proteomics analysis revealed that unsaturated fatty acids modify similar protein targets to saturated fatty acids, including several immunity-associated proteins. This demonstrates that unsaturated fatty acids can directly modify many proteins to exert their unique and often beneficial physiological effects in vivo. © 2016 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
del Pozo Martin Y.,The Crick Institute |
Park D.,The Crick Institute |
Ramachandran A.,The Crick Institute |
Ombrato L.,The Crick Institute |
And 7 more authors.
Cell Reports | Year: 2015
During metastatic colonization, tumor cells must establish a favorable microenvironment or niche that will sustain their growth. However, both the temporal and molecular details of this process remain poorly understood. Here, we found that metastatic initiating cells (MICs) exhibit a high capacity for lung fibroblast activation as a result of Thrombospondin 2 (THBS2) expression. Importantly, inhibiting the mesenchymal phenotype of MICs by blocking the epithelial-to-mesenchymal transition (EMT)-associated kinase AXL reduces THBS2 secretion, niche-activating ability, and, consequently, metastatic competence. Subsequently, disseminated metastatic cells revert to an AXL-negative, more epithelial phenotype to proliferate and decrease the phosphorylation levels of TGF-β-dependent SMAD2-3 in favor of BMP/SMAD1-5 signaling. Remarkably, newly activated fibroblasts promote this transition. In summary, our data reveal a crosstalk between cancer cells and their microenvironment whereby the EMT status initially triggers and then is regulated by niche activation during metastatic colonization. © 2015 The Authors.