E-cadherin is required for metastasis in multiple models of breast cancer

Article metrics

Abstract

Metastasis is the major driver of death in patients with cancer. Invasion of surrounding tissues and metastasis have been proposed to initiate following loss of the intercellular adhesion protein, E-cadherin1,2, on the basis of inverse correlations between in vitro migration and E-cadherin levels3. However, this hypothesis is inconsistent with the observation that most breast cancers are invasive ductal carcinomas and express E-cadherin in primary tumours and metastases4. To resolve this discrepancy, we tested the genetic requirement for E-cadherin in metastasis using mouse and human models of both luminal and basal invasive ductal carcinomas. Here we show that E-cadherin promotes metastasis in diverse models of invasive ductal carcinomas. While loss of E-cadherin increased invasion, it also reduced cancer cell proliferation and survival, circulating tumour cell number, seeding of cancer cells in distant organs and metastasis outgrowth. Transcriptionally, loss of E-cadherin was associated with upregulation of genes involved in transforming growth factor-β (TGFβ), reactive oxygen species and apoptosis signalling pathways. At the cellular level, disseminating E-cadherin-negative cells exhibited nuclear enrichment of SMAD2/3, oxidative stress and increased apoptosis. Colony formation of E-cadherin-negative cells was rescued by inhibition of TGFβ-receptor signalling, reactive oxygen accumulation or apoptosis. Our results reveal that E-cadherin acts as a survival factor in invasive ductal carcinomas during the detachment, systemic dissemination and seeding phases of metastasis by limiting reactive oxygen-mediated apoptosis. Identifying molecular strategies to inhibit E-cadherin-mediated survival in metastatic breast cancer cells may have potential as a therapeutic approach for breast cancer.

Access optionsAccess options

Rent or Buy article

Get time limited or full article access on ReadCube.

from$8.99

All prices are NET prices.

Fig. 1: E-cad loss increases invasion and dissemination into 3D collagen I.
Fig. 2: E-cad loss inhibits metastasis.
Fig. 3: E-cad loss decreases colony formation.
Fig. 4: Disseminated E-cad cells frequently undergo TGFβ-dependent ROS-mediated apoptosis.

Data availability

The next-generation RNA-seq data is available from NCBI Gene Expression Omnibus with accession number GSE114011. The numerical data underlying all figure panels are available in the Supplementary Information. Additional requests for information or data will be fulfilled by the corresponding author upon request.

Code availability

Semi-custom code was developed for RNA-seq analysis and has been made available on GitHub at https://github.com/baderzone/ecad_2019.

References

  1. 1.

    Bogenrieder, T. & Herlyn, M. Axis of evil: molecular mechanisms of cancer metastasis. Oncogene 22, 6524–6536 (2003).

  2. 2.

    Berx, G. et al. E-cadherin is a tumour/invasion suppressor gene mutated in human lobular breast cancers. EMBO J. 14, 6107–6115 (1995).

  3. 3.

    Frixen, U. H. et al. E-cadherin-mediated cell-cell adhesion prevents invasiveness of human carcinoma cells. J. Cell Biol. 113, 173–185 (1991).

  4. 4.

    Li, C. I., Anderson, B. O., Daling, J. R. & Moe, R. E. Trends in incidence rates of invasive lobular and ductal breast carcinoma. J. Am. Med. Assoc. 289, 1421–1424 (2003).

  5. 5.

    Nguyen-Ngoc, K. V. et al. ECM microenvironment regulates collective migration and local dissemination in normal and malignant mammary epithelium. Proc. Natl Acad. Sci. USA 109, E2595–E2604 (2012).

  6. 6.

    Cheung, K. J., Gabrielson, E., Werb, Z. & Ewald, A. J. Collective invasion in breast cancer requires a conserved basal epithelial program. Cell 155, 1639–1651 (2013).

  7. 7.

    Christofori, G. & Semb, H. The role of the cell-adhesion molecule E-cadherin as a tumour-suppressor gene. Trends Biochem. Sci. 24, 73–76 (1999).

  8. 8.

    Chambers, A. F., Groom, A. C. & MacDonald, I. C. Dissemination and growth of cancer cells in metastatic sites. Nat. Rev. Cancer 2, 563–572 (2002).

  9. 9.

    Sosa, M. S. et al. NR2F1 controls tumour cell dormancy via SOX9- and RARβ-driven quiescence programmes. Nat. Commun. 6, 6170 (2015).

  10. 10.

    Wheelock, M. J., Shintani, Y., Maeda, M., Fukumoto, Y. & Johnson, K. R. Cadherin switching. J. Cell Sci. 121, 727–735 (2008).

  11. 11.

    Davies, S. R., Watkins, G., Douglas-Jones, A., Mansel, R. E. & Jiang, W. G. Bone morphogenetic proteins 1 to 7 in human breast cancer, expression pattern and clinical/prognostic relevance. J. Exp. Ther. Oncol. 7, 327–338 (2008).

  12. 12.

    Demircan, B. et al. Comparative epigenomics of human and mouse mammary tumors. Genes Chromosom. Cancer 48, 83–97 (2009).

  13. 13.

    Xue, W. et al. A cluster of cooperating tumor-suppressor gene candidates in chromosomal deletions. Proc. Natl Acad. Sci. USA 109, 8212–8217 (2012).

  14. 14.

    Jovanovic, I. P. et al. Interleukin-33/ST2 axis promotes breast cancer growth and metastases by facilitating intratumoral accumulation of immunosuppressive and innate lymphoid cells. Int. J. Cancer 134, 1669–1682 (2014).

  15. 15.

    Gao, D. et al. Myeloid progenitor cells in the premetastatic lung promote metastases by inducing mesenchymal to epithelial transition. Cancer Res. 72, 1384–1394 (2012).

  16. 16.

    Day, M. L. et al. E-cadherin mediates aggregation-dependent survival of prostate and mammary epithelial cells through the retinoblastoma cell cycle control pathway. J. Biol. Chem. 274, 9656–9664 (1999).

  17. 17.

    Adamson, G. M. & Billings, R. E. Tumor necrosis factor induced oxidative stress in isolated mouse hepatocytes. Arch. Biochem. Biophys. 294, 223–229 (1992).

  18. 18.

    Liu, R. M. & Desai, L. P. Reciprocal regulation of TGF-β and reactive oxygen species: A perverse cycle for fibrosis. Redox Biol. 6, 565–577 (2015).

  19. 19.

    Johnson, T. M., Yu, Z. X., Ferrans, V. J., Lowenstein, R. A. & Finkel, T. Reactive oxygen species are downstream mediators of p53-dependent apoptosis. Proc. Natl Acad. Sci. USA 93, 11848–11852 (1996).

  20. 20.

    LeBleu, V. S. et al. PGC-1α mediates mitochondrial biogenesis and oxidative phosphorylation in cancer cells to promote metastasis. Nat. Cell Biol. 16, 992–1003 (2014).

  21. 21.

    Kerksick, C. & Willoughby, D. The antioxidant role of glutathione and N-acetyl-cysteine supplements and exercise-induced oxidative stress. J. Int. Soc. Sports Nutr. 2, 38–44 (2005).

  22. 22.

    Herrera, B. et al. Reactive oxygen species (ROS) mediates the mitochondrial-dependent apoptosis induced by transforming growth factor β in fetal hepatocytes. FASEB J. 15, 741–751 (2001).

  23. 23.

    Massagué, J. TGFβ in Cancer. Cell 134, 215–230 (2008).

  24. 24.

    Kleer, C. G., van Golen, K. L., Braun, T. & Merajver, S. D. Persistent E-cadherin expression in inflammatory breast cancer. Mod. Pathol. 14, 458–464 (2001).

  25. 25.

    Rodriguez, F. J., Lewis-Tuffin, L. J. & Anastasiadis, P. Z. E-cadherin’s dark side: possible role in tumor progression. Biochim. Biophys. Acta 1826, 23–31 (2012).

  26. 26.

    Sundfeldt, K. et al. E-cadherin expression in human epithelial ovarian cancer and normal ovary. Int. J. Cancer 74, 275–280 (1997).

  27. 27.

    Kim, S. A. et al. Loss of CDH1 (E-cadherin) expression is associated with infiltrative tumour growth and lymph node metastasis. Br. J. Cancer 114, 199–206 (2016).

  28. 28.

    McCart Reed, A. E. et al. An epithelial to mesenchymal transition programme does not usually drive the phenotype of invasive lobular carcinomas. J. Pathol. 238, 489–494 (2016).

  29. 29.

    Xu, Y. et al. Breast tumor cell-specific knockout of Twist1 inhibits cancer cell plasticity, dissemination, and lung metastasis in mice. Proc. Natl Acad. Sci. USA 114, 11494–11499 (2017).

  30. 30.

    Beerling, E. et al. Plasticity between epithelial and mesenchymal states unlinks EMT from metastasis-enhancing stem cell capacity. Cell Rep. 14, 2281–2288 (2016).

  31. 31.

    Lambert, A. W., Pattabiraman, D. R. & Weinberg, R. A. Emerging biological principles of metastasis. Cell 168, 670–691 (2017).

  32. 32.

    Yates, C. C., Shepard, C. R., Stolz, D. B. & Wells, A. Co-culturing human prostate carcinoma cells with hepatocytes leads to increased expression of E-cadherin. Br. J. Cancer 96, 1246–1252 (2007).

  33. 33.

    Brabletz, T. To differentiate or not—routes towards metastasis. Nat. Rev. Cancer 12, 425–436 (2012).

  34. 34.

    Guy, C. T., Cardiff, R. D. & Muller, W. J. Induction of mammary tumors by expression of polyomavirus middle T oncogene: a transgenic mouse model for metastatic disease. Mol. Cell. Biol. 12, 954–961 (1992).

  35. 35.

    Maroulakou, I. G., Anver, M., Garrett, L. & Green, J. E. Prostate and mammary adenocarcinoma in transgenic mice carrying a rat C3(1) simian virus 40 large tumor antigen fusion gene. Proc. Natl Acad. Sci. USA 91, 11236–11240 (1994).

  36. 36.

    Muzumdar, M. D., Tasic, B., Miyamichi, K., Li, L. & Luo, L. A global double-fluorescent Cre reporter mouse. Genesis 45, 593–605 (2007).

  37. 37.

    Boussadia, O., Kutsch, S., Hierholzer, A., Delmas, V. & Kemler, R. E-cadherin is a survival factor for the lactating mouse mammary gland. Mech. Dev. 115, 53–62 (2002).

  38. 38.

    Badea, T. C., Wang, Y. & Nathans, J. A noninvasive genetic/pharmacologic strategy for visualizing cell morphology and clonal relationships in the mouse. J. Neurosci. 23, 2314–2322 (2003).

  39. 39.

    Nguyen-Ngoc, K. V. et al. 3D culture assays of murine mammary branching morphogenesis and epithelial invasion. Methods Mol. Biol. 1189, 135–162 (2015).

  40. 40.

    Langmead, B., Trapnell, C., Pop, M. & Salzberg, S. L. Ultrafast and memory-efficient alignment of short DNA sequences to the human genome. Genome Biol. 10, R25 (2009).

  41. 41.

    Kim, D. et al. TopHat2: accurate alignment of transcriptomes in the presence of insertions, deletions and gene fusions. Genome Biol. 14, R36 (2013).

  42. 42.

    Anders, S., Pyl, P. T. & Huber, W. HTSeq—a Python framework to work with high-throughput sequencing data. Bioinformatics 31, 166–169 (2015).

  43. 43.

    Love, M. I., Huber, W. & Anders, S. Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biol. 15, 550 (2014).

  44. 44.

    Cheung, K. J. et al. Polyclonal breast cancer metastases arise from collective dissemination of keratin 14-expressing tumor cell clusters. Proc. Natl Acad. Sci. USA 113, E854–E863 (2016).

  45. 45.

    Liberzon, A. et al. Molecular signatures database (MSigDB) 3.0. Bioinformatics 27, 1739–1740 (2011).

Download references

Acknowledgements

We thank all members of the Ewald Laboratory for critical discussions and J. C. Ramirez for assistance in quantifying the area of metastases and cryo-sectioning. We thank H. Zhang from the Johns Hopkins School of Public Health Flow Cytometry Core Facility and H. Hao from the Microarray and Deep Sequencing Core Facility for technical assistance. A.J.E. received support for this project through grants from: The Breast Cancer Research Foundation/Pink Agenda (BCRF-16-048, BCRF-17-048, BCRF-18-048), the Metastatic Breast Cancer Network, Twisted Pink, Hope Scarves and the National Institutes of Health/National Cancer Institute (U01CA217846, U54CA2101732, 3P30CA006973). V.P. was supported in part by an Isaac and Lucille Hay Fellowship. J.S.B. received support for this project through a grant from the National Institutes of Health/National Cancer Institute (U01CA217846). Both A.J.E. and J.S.B. received support from the Jayne Koskinas Ted Giovanis Foundation for Health and Policy and the Breast Cancer Research Foundation, private foundations committed to critical funding of cancer research. The opinions, findings, conclusions and recommendations expressed in this Letter are those of the authors and not necessarily those of the Jayne Koskinas Ted Giovanis Foundation for Health and Policy or the Breast Cancer Research Foundation, or their respective directors, officers or staff. Research in the Aceto laboratory is supported by the European Research Council, the Swiss National Science Foundation, the Swiss Cancer League, the Basel Cancer League, the two Cantons of Basel through ETH Zürich and the University of Basel.

Author information

V.P. and A.J.E. conceptualized the project and designed most experiments. V.P. performed most experiments. I.K., B.M.S. and N.A. performed CTC enumeration. Y.S. and J.S.B. performed RNA-seq analysis. V.P., Y.S., J.S.B. and A.J.E. contributed to the interpretation of the sequencing data. V.P. and A.J.E. wrote the manuscript with useful input from all authors.

Correspondence to Andrew J. Ewald.

Ethics declarations

Competing interests

A.J.E. and V.P. are listed as inventors on a patent application related to the use of antibodies as cancer therapeutics. A.J.E. is listed as an inventor on a patent application related to the use of keratin-14 as a prognostic indicator for breast cancer outcomes. A.J.E.'s spouse is an employee of ImmunoCore. J.S.B. is a founder and director of Neochromosome and a member of the scientific advisory board of AI Therapeutics. I.K., B.M.S. and N.A. are listed as inventors in patent applications related to CTCs and cancer treatment. N.A. is a paid consultant for pharmaceutical and insurance companies with an interest in liquid biopsy.

Additional information

Publisher’s note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Peer review information Nature thanks Johanna Ivaska, Erik Thompson and the other, anonymous, reviewer(s) for their contribution to the peer review of this work.

Extended data figures and tables

Extended Data Fig. 1 E-cad expression is retained during several intermediate stages of metastasis.

a, E-cad expression is determined by immunofluorescence at various stages of metastasis following orthotopic transplantation of MMTV-PyMT; mT/mG tumour organoids into NSG host mice. Observations described in this figure were made across at least three independent MMTV-PyMT tumours. b, Tumour organoids embedded in collagen I express membrane-localized E-cad. Scale bar, 50 μm. c, Representative tile scan showing E-cad expression within a primary tumour section. Scale bar, 500 μm. Three selected regions within the tumour (yellow boxes; scale bar, 200 μm) express high levels of membrane-localized E-cad. dg, E-cad is also expressed in invasion strands (d), locally disseminated units (e), intravasated units (f) and distant metastases (g) in vivo. Scale bar, 50 μm. Arrowhead, disseminated unit (e) or intravasated unit (f). Yellow box in f marks a magnified inset (scale bar, 20 μm).

Extended Data Fig. 2 E-cad loss decreases migratory persistence ex vivo and increases invasion in vivo.

a, E-cad+ or E-cad cells from adeno-Cre-treated MMTV-PyMT; Cdh1+/+ or Cdh1fl/fl organoids, respectively, were manually tracked as they migrated within collagen I. Blue lines, cell tracks; circles, last tracked position of the cell at that time point. Scale bar, 50 μm. b, c, Disseminating E-cad cancer cells exhibit decreased migratory persistence (b) and displacement (c) relative to E-cad+ cells. Bar, median. ****P < 0.0001 and *P = 0.027 (Mann–Whitney test, two-sided). d, Adeno-Cre-transduced MMTV-PyMT; Cdh1+/+ or Cdh1fl/fl organoids were transplanted into cleared mammary fat pads of immunocompromised NSG mice. Tumour sizes were monitored twice weekly and collected after ~6–8 weeks. e, Tumours arising from Cdh1fl/fl organoids were smaller than those arising from control organoids at the corresponding timepoint. Data are mean ± s.e.m. ****P < 0.0001 (regression analysis). f, Representative micrographs of primary tumours arising from Cdh1+/+ or Cdh1fl/fl donor tissue. Scale bar, 500 μm. Control tumours have relatively similar amounts of mT+ and mG+ cancer cells. By contrast, mG+ (Cre+E-cad) cancer cells constitute <10% of Cdh1fl/fl tumours. Data are mean ± s.e.m. ****P < 0.0001 (Kruskal–Wallis test). g, Left, representative tile scan of a primary tumour arising from control organoids (scale bar, 500 μm); an enlarged inset of the tumour–stroma border is shown (scale bar, 50 μm). Right, ~94% of the tumour border exhibits a pushing boundary. h, Left, representative tile scan of a mG+ (E-cad) region of a primary tumour arising from Cdh1fl/fl organoids (scale bar, 500 μm); enlarged insets of the tumour–stroma border are shown (scale bars, 50 μm and 20 μm). Right, more than 80% of the E-cad tumour edge has an invasive morphology. Source data

Extended Data Fig. 3 E-cad- cancer cells retain epithelial gene expression.

a, Control adeno-Cre-transduced MMTV-PyMT; Cdh1+/+ tumours are E-cad+, keratin+, vimentin. mG+ (Cre+) cells in adeno-Cre-transduced MMTV-PyMT; Cdh1fl/fl tumours are E-cad, keratin+ and vimentin. Dotted lines define mT+, E-cad+ regions within Cdh1fl/fl tumours. Adjacent host-derived stroma serves as a positive control for vimentin expression. Scale bar, 50 μm. Repeated across at least three independent Cdh1+/+ and Cdh1fl/fl tumours. b, Metastases in control transplant mice are keratin+. mG+ (Cre+) metastases arising in adeno-Cre-transduced MMTV-PyMT; Cdh1fl/fl transplant mice are keratin+. Scale bar, 50 μm. Repeated across sections from at least three independent Cdh1+/+ and Cdh1fl/fl mice. c, Lung metastases in control tail-vein mice are E-cad+ and have membrane localized β-catenin. By contrast, mG+ metastases in mice injected with adeno-Cre-transduced MMTV-PyMT; Cdh1fl/fl clusters are E-cad and β-catenin. Scale bar, 50 μm. Repeated across sections from at least three independent Cdh1+/+ and Cdh1fl/fl mice. d, Left, representative micrographs of metastases arising in adeno-Cre-transduced MMTV-PyMT; Cdh1+/+ or Cdh1fl/fl transplant mice. Scale bar, 100 μm. Right, projected surface area of metastases arising in adeno-Cre-transduced MMTV-PyMT Cdh1+/+ and Cdh1fl/fl host mice is represented. Box plots show the median, box edges represent the first and third quartiles, and the whiskers extend to 5th and 95th percentiles. ****P < 0.0001 (Mann–Whitney test, two-sided). e, Heat map of canonical EMT transcripts (left) and cadherin family members (right). RNA-seq was performed by comparing transcriptomes of four adeno-Cre-treated MMTV-PyMT; Cdh1+/+ organoids to five adeno-Cre-treated MMTV-PyMT; Cdh1fl/fl organoids. P values were calculated for the Wald test. Genome-wide significance = 1.7 × 10−6 (FWER = 0.05). Source data

Extended Data Fig. 4 E-cad loss increases invasion and dissemination but prevents metastasis across several assays.

a, All metastases arising from adeno-Cre-transduced MMTV-PyMT; Cdh1+/+ organoids are E-cad+. Only mG+ metastases arising from adeno-Cre-transduced MMTV-PyMT; Cdh1fl/fl organoids are E-cad. b, Number of macrometastases in mice transplanted with adeno-Cre-transduced MMTV-PyMT; Cdh1+/+ or Cdh1fl/fl tumour organoids, as sorted by the colour of the metastasis (mT, mG or mT/mG mixed). Horizontal line shows the median. ****P < 0.0001 (Kruskal–Wallis test). c, Number of macrometastases per mouse after tail-vein injections of adeno-Cre-transduced MMTV-PyMT; Cdh1+/+ or Cdh1fl/fl clusters, as sorted by the colour of the metastasis (mT, mG or mT/mG mixed) is represented. Horizontal line shows the median. ****P < 0.0001 (two-way ANOVA). d, Schematic of 3D-invasion assays using organoids from the same MMTV-PyMT; Cdh1fl/fl tumour with or without adeno-Cre transduction. e, Representative DIC images of MMTV-PyMT; Cdh1fl/fl organoids with or without adeno-Cre. Arrowheads, dissemination events. Scale bar, 50 μm. f, g, There is an increase in invasion (f) and dissemination (g) in Cdh1fl/fl organoids treated with adeno-Cre, relative to uninfected controls. Box plots show the median, box edges represent the first and third quartiles, and the whiskers extend to 5th and 95th percentiles. ****P < 0.0001 (Mann–Whitney test, two-sided). h, i, Schematic of the transplant assay using MMTV-PyMT; Cdh1fl/fl organoids with or without adeno-Cre (h). There are no observable E-cad macrometastases (i). Horizontal line shows the median. **P = 0.0022 (Mann–Whitney test, two-sided). j, k, Schematic of the tail-vein assay using MMTV-PyMT; Cdh1fl/fl organoids with or without adeno-Cre (j). E-cad- cancer cells are defective at tumour-cell seeding (k). Horizontal line shows the median. **P = 0.0087 (Mann–Whitney test, two-sided). l, Schematic of orthotopic transplantation of adeno-Cre-transduced MMTV-PyMT; Cdh1+/+ or MMTV-PyMT; Cdh1fl/fl tumour organoids into immunocompetent FVB host mice. Lungs from these mice were collected after ~8 weeks and the number of metastases was counted. m, Number of observed macrometastases in FVB mice transplanted with adeno-Cre-transduced MMTV-PyMT; Cdh1+/+ or Cdh1fl/fl tumour organoids, sorted by the colour of the metastasis. Horizontal line shows the median. **P = 0.0033, *P = 0.024 (two-way ANOVA). n, Left, all metastases arising in control mice are E-cad+. Only mG+ cells containing metastases in adeno-Cre-transduced MMTV-PyMT; Cdh1fl/fl host mice are E-cad. Right, E-cad cancer cells do not contribute to observable macrometastases in adeno-Cre-transduced MMTV-PyMT; Cdh1fl/fl immunocompetent host mice. Horizontal line shows the median. ****P < 0.0001 (Mann–Whitney test, two-sided). Source data

Extended Data Fig. 5 Tamoxifen-inducible Cdh1 deletion increases invasion and decreases metastasis ex vivo and in vivo.

a, Schema for tail-vein injections using FACS-sorted populations of purely E-cad+ or E-cad cancer clusters. b, E-cad cancer cell clusters form significantly fewer metastatic colonies compared to E-cad+ clusters. ***P = 0.0002 (Mann–Whitney test, two-sided). c, Schema for invasion assay using tamoxifen-treated MMTV-PyMT; Cdh1fl/fl; CreER organoids. Control organoids were also tamoxifen-treated and were isolated from MMTV-PyMT; Cdh1fl/fl tumours without CreER expression. d, Representative western blot depicting E-cad protein levels (loading control on same gel; three replicates of E-cad were quantified for summary graph). Data are mean ± s.d. *P = 0.029 (Mann–Whitney test, two-sided). e, Representative DIC micrographs of collagen-embedded, tamoxifen-treated control and CreER-expressing organoids. Arrowheads, dissemination events. Scale bar, 50 μm. f, g, Tamoxifen-treated organoids isolated from CreER-expressing tumours are more invasive (f) and disseminative (g) than control organoids. Box plots show the median, box edges represent the first and third quartiles, and the whiskers extend to 5th and 95th percentiles. ****P < 0.0001 (Mann–Whitney test, two-sided). h, Schematic of in situ Cdh1 deletion. Organoids isolated from MMTV-PyMT; Cdh1fl/fl and Cdh1fl/fl; CreER tumours were transplanted into immunocompromised NSG mice. Once these mice developed tumours of up to 8–10 mm, tamoxifen was injected intraperitoneally, and tumour growth was monitored twice weekly. Tumours were collected ~six weeks after transplantation. i, There is a decrease in tumour growth rates after Cdh1 deletion relative to control tumours. Data are mean ± s.e.m. ****P < 0.0001 by regression analysis. j, Representative tile scan of a MMTV-PyMT; Cdh1fl/fl; CreER primary tumour. Tamoxifen was injected to delete Cdh1. mT+ cells did not express Cre and are E-cad+; mG+ cancer cells are E-cad. Scale bar, 500 μm. Magnified insets for E-cad+ (yellow box; scale bar, 50 μm) and E-cad (grey box; scale bar, 50 μm) regions of the tumour boundary. Source data

Extended Data Fig. 6 E-cad loss promotes invasion and suppresses metastasis in a basal model of IDC.

a, Schematic of 3D invasion assay using C3(1)-tag tumour organoids. Cdh1 deletion is induced by adeno-Cre. Control organoids are treated with adeno-GFP. b, Representative western blot depicting protein levels of E-cad in adeno-GFP and adeno-Cre-transduced C3(1)-tag; Cdh1fl/fl organoids (loading control on same gel; four replicates of E-cad were quantified for summary graph). Data are mean ± s.d., *P = 0.029 (Mann–Whitney test, two-sided). c, Representative timelapse DIC images of adeno-GFP and adeno-Cre-transduced C3(1)-tag; Cdh1fl/fl tumour organoids. Scale bar, 50 μm. d, There is a significant increase in dissemination in adeno-Cre-transduced C3(1)-tag; Cdh1fl/fl organoids relative to control organoids. Box plots show the median, box edges represent the first and third quartiles, and the whiskers extend to 5th and 95th percentiles; ****P < 0.0001 (Mann–Whitney test, two-sided). e, Adeno-GFP or adeno-Cre-transduced C31(1)-Tag; Cdh1fl/fl organoids were transplanted into the cleared mammary fat pads of NSG mice. The tumour boundary was examined for invasive morphology. f, Left, control C3(1)-tag tumours typically organize and invade collectively. Right, loss of E-cad causes an increase in invasion and dissemination along the tumour–stroma interface (scale bar, 500 μm). Magnified insets show invasive borders (scale bar, 100 μm). g, CTCs were isolated by cardiac puncture performed on NSG mice transplanted with adeno-GFP or adeno-Cre-transduced C3(1)-tag; Cdh1fl/fl organoids. Left, there is a significant decrease in the number CTCs arising from E-cad cancer cells. Horizontal line shows the median. *P = 0.026 (Mann–Whitney test, two-sided). Right, representative images of E-cad+ and E-cad CTCs. Scale bar, 20 μm. h, Schema for transplant and tail-vein assays using adeno-GFP or adeno-Cre-transduced C3(1)-tag; Cdh1fl/fl tumour organoids. All metastases in control mice are E-cad+, whereas only metastases containing mG+ cancer cells in mice injected with adeno-Cre transduced organoids are E-cad. i, Left, representative micrographs of E-cad+ and E-cad metastases arising in adeno-GFP and adeno-Cre-transduced C3(1)-tag; Cdh1fl/fl transplant mice, respectively. Scale bar, 50 μm. Right, E-cad cancer cells rarely contribute to macrometastases. Horizontal line shows the median. **P = 0.002 (Mann–Whitney test, two-sided). j, Left, whole-lung images of metastases arising after tail-vein injections of adeno-GFP or adeno-Cre-transduced C3(1)-tag; Cdh1fl/flcancer cell clusters (scale bar, 1 cm), with magnified insets for smaller lung areas. Arrowheads, metastases. Right, E-cad cancer cells rarely contribute to metastases in tail-vein assay. Horizontal line shows the median. ****P < 0.0001 (Mann–Whitney test, two-sided). Source data

Extended Data Fig. 7 Loss of E-cad in a triple-negative IDC PDX increases invasion and dissemination but decreases colony formation.

a, Gating strategy to isolate tumour cell clusters (2–4 cells each). b, c, Flow sorting strategy to isolate mT+ (E-cad+) tumour cell clusters from tamoxifen treated MMTV-PyMT; Cdh1fl/fl organoids (b) or mG+ (E-cad) tumour-cell clusters from tamoxifen treated MMTV-PyMT; Cdh1fl/fl; CreER organoids (c). Scale bar, 10 μm. d, Organoids were isolated from a triple-negative PDX tumour and were divided into three groups for treatment with lentiviral shRNA against luciferase or E-cad (two shRNA clones). Puromycin was used to positively select transduced cells. Organoids were embedded in 3D collagen I to assay their invasive phenotype. e, Representative western blot depicting decreased levels of E-cad protein when treated with Cdh1 shRNA (loading control on same gel; three replicates of E-cad were quantified for summary graph). Data are mean ± s.d. *P = 0.02 (Kruskal–Wallis test). f, Representative micrographs of PDX tumour organoids embedded in collagen I. Scale bar, 50 μm. g, h, Knockdown of E-cad in PDX organoids significantly increases invasion (g) and dissemination (h). Horizontal line shows the median. ****P < 0.0001, ***P = 0.0002 (Kruskal–Wallis test). i, Schema for colony formation assay after E-cad knockdown in a triple-negative PDX model. j, Representative micrographs of colonies arising from PDX-derived cancer cell clusters treated with luciferase shRNA or Cdh1 shRNA (two shRNA clones). Scale bar, 50 μm. k, Knockdown of E-cad in PDX-derived cancer cells decreases colony formation. *P = 0.034 (Kruskal–Wallis test). Source data

Extended Data Fig. 8 Loss of E-cad causes an increase in apoptosis, ROS accumulation and TGFβ signalling.

a, Representative western blot depicting no significant changes in total protein levels of the dormancy maker, NR2F1, in adeno-Cre-transduced MMTV-PyMT; Cdh1+/+ or Cdh1fl/fl organoids (loading control on same gel; three replicates of NR2F1 were quantified for summary graph). Data are mean ± s.d. ****P < 0.0001 (Mann–Whitney test, two-sided). b, RNA-seq was used to compare gene expression changes in adeno-Cre-transduced MMTV-PyMT; Cdh1+/+ and Cdh1fl/fl organoids. c, Heat map of differentially expressed transcripts in adeno-Cre-transduced MMTV-PyMT; Cdh1+/+ and Cdh1fl/fl tumour organoids. RNA-seq was performed by comparing transcriptomes of four adeno-Cre-treated MMTV-PyMT; Cdh1+/+ organoids to five adeno-Cre-treated MMTV-PyMT; Cdh1fl/fl organoids. P values were calculated for the Wald test. Genome-wide significance = 1.7 × 10−6 (FWER = 0.05). d, Schema of transcripts known to be involved in metastasis that are upregulated (red boxes) or downregulated (blue boxes) as a consequence of E-cad loss in MMTV-PyMT tumour organoids. e, Differentially expressed transcripts that have previously been shown to regulate apoptosis are highlighted (red) among all differential transcripts (black). All transcripts are represented in grey. RNA-seq was performed by comparing transcriptomes of four adeno-Cre-treated MMTV-PyMT; Cdh1+/+ organoids to five adeno-Cre-treated MMTV-PyMT; Cdh1fl/fl organoids. Raw P values are reported without multiple-testing correction. Fold change is reported as experimental/control for experimental > control and as −control/experimental for control > experimental. Genome-wide significance = 1.7 × 10−6 (FWER = 0.05). f, Proportion of disseminating E-cad+ or E-cad cancer cells displaying apoptotic morphologies. g, CC3 is localized to E-cad disseminating cells in adeno-Cre = transduced MMTV-PyMT; Cdh1fl/fl tumour organoids (scale bar, 50 μm), with zoomed insets for the organoid bulk and disseminated cell (scale bar, 20 μm). Arrows, disseminated cells. h, Mean intensity of ROS in E-cad+ or E-cad disseminated cells relative to the corresponding organoid bulk. ****P < 0.0001, **P = 0.002 (Mann–Whitney test, two-sided). i, SMAD2/3 is diffusely expressed within E-cad+, MMTV-PyMT cancer cells, whereas it is nuclear localized after Cdh1 deletion. Scale bar, 20 μm. Observations were made across at least three independent tumours. Source data

Extended Data Fig. 9 E-cad loss is associated with increased apoptosis at several stages of metastasis in vivo.

a, CC3 immunoreactivity was used to detect the relative abundance of apoptosis in transplanted adeno-Cre-transduced MMTV-PyMT; Cdh1+/+ or Cdh1fl/fl tumours. b, Representative micrographs of E-cad+ primary tumours depicting the rarity of CC3+ cancer cells within the tumour bulk, invasion or dissemination events (magnified insets; scale bar, 20 μm). Scale bar, 50 μm. c, Representative micrographs of E-cad primary tumours depicting a significant proportion of CC3+ cancer cells within the tumour bulk, dissemination events and invasion (magnified insets; scale bar, 20 μm). Scale bar, 50 μm. d, The number of CC3+ cancer cells within E-cad primary tumours increases significantly relative to E-cad+ primary tumours. Horizontal line shows the median. ***P = 0.0008 (Mann–Whitney test, two-sided). e, Left, representative micrographs marking CC3+ cancer cells within E-cad+ or E-cad metastases. Scale bar, 20 μm. Arrowhead, CC3+, E-cad metastasis. Right, there is a significant increase in the amount of apoptosis in E-cad metastases relative to E-cad+ metastases. Horizontal line shows the median. *P = 0.015 (Mann–Whitney test, two-sided). f, Schema of colony-formation assay using MMTV-PyMT; E-cad+ or E-cad cancer cells in the presence of 100 nM or 1 μM of doxorubicin, paclitaxel and cisplatin. g, E-cad cancer cells are no more or less sensitive to chemotherapies compared to E-cad+ cancer cells. Data are mean ± s.d. **P = 0.0065 (E-cad+), 0.0044 (E-cad) and ***P = 0.0005 (Ecad+), 0.0003 (E-cad) (Kruskal–Wallis test). Source data

Extended Data Fig. 10 Effects of apoptosis, oxidative stress and TGFβ inhibition on dissemination of E-cad+ or E-cad MMTV-PyMT organoids.

a, Schema of invasion assay in the presence of a pan-caspase inhibitor (z-VAD-FMK). b, Representative micrographs of adeno-Cre-transduced MMTV-PyMT; Cdh1+/+ or Cdh1fl/fl organoids in the presence of z-VAD-FMK. Scale bar, 50 μm. c, d, There is a dose-dependent increase in invasion (c) and dissemination (d) of Cdh1fl/fl organoids relative to control organoids in the presence of z-VAD-FMK. Box plots show the median, box edges represent the first and third quartiles, and the whiskers extend to 5th and 95th percentiles. ****P < 0.0001 (Kruskal–Wallis test). e, Schema of invasion assay in the presence of NAC, soluble TGFβ or TGFβR1 inhibitor (SB525334). f, Representative images of tamoxifen-treated MMTV-PyMT; Cdh1fl/fl or Cdh1fl/fl; CreER organoids in the presence of these inhibitors. Treatment with soluble TGFβ increases dissemination of E-cad+ and E-cad organoids while inhibition of TGFβR1 decreases dissemination. Treatment with NAC does not change dissemination of E-cad+ or E-cad organoids. Scale bar, 50 μm. Box plots show the median, box edges represent the first and third quartiles, and the whiskers extend to 5th and 95th percentiles. ****P < 0.0001 (Kruskal–Wallis test). g, Schema of tail-vein assays using NAC pre-treated E-cad+ or E-cad cancer cell clusters. MMTV-PyMT; Cdh1fl/fl organoids were treated with adeno-GFP or adeno-Cre, trypsinized into small clusters, incubated with NAC for 24 h, and injected via the tail vein of NSG host mice. Lungs from these mice were collected 48 h or 1 week after injection and number of micro- and macrometastases, respectively, were counted. h, There is no significant difference in the number of micrometastases arising from E-cad+ or E-cad cancer cells after NAC pre-treatment. Horizontal line shows the median. (Mann–Whitney test, two-sided). i, NAC pretreatment partially rescued the ability of E-cad cancer cells to form macrometastases relative to E-cad+ cancer cells. Horizontal line shows the median. **P = 0.0051 (Mann–Whitney test, two-sided). Source data

Supplementary information

Supplementary Information

This file contains Supplementary Data Sections 1–3

Reporting Summary

Supplementary Information

This file contains the uncropped blots

Video 1

Control MMTV-PyMT organoids invade collectively into 3D collagen I Representative DIC timelapse of an adeno-Cre transduced MMTV-PyMT; E-cad+/+ organoid (E-cad+) embedded in 3D collagen I. Representative across at least 3 independent tumors.

Video 2

E-cad deletion increases invasion and local dissemination of MMTV-PyMT organoids in 3D collagen I Representative DIC timelapse of an adeno-Cre transduced MMTV-PyMT; E-cadfl/fl organoid (mosaic E-cad-) embedded in 3D collagen I. Representative across at least 3 independent tumors.

Video 3

Control C3(1)-Tag organoids invade collectively into 3D collagen I Representative DIC timelapse of an adeno-GFP transduced C3(1)-Tag; E-cadfl/fl organoid (E-cad+) embedded in 3D collagen I. Representative across 3 independent tumors.

Video 4

E-cad deletion increases invasion and local dissemination of C3(1)-Tag organoids in 3D collagen I Representative DIC timelapse of an adeno-Cre transduced C3(1)-Tag; E-cadfl/fl organoid (mosaic E-cad-) embedded in 3D collagen I. Representative across 3 independent tumors.

Video 5

Disseminated E-cad- cells frequently undergo apoptosis Representative confocal timelapse of an E-cad- cancer cell undergoing apoptosis as it disseminates from an adeno-Cre transduced MMTV-PyMT; E-cadfl/fl organoid embedded in 3D collagen I. Representative across at least 3 independent tumors.

Source data

Source Data Fig. 1

Source Data Fig. 2

Source Data Fig. 3

Source Data Fig. 4

Source Data Extended Data Fig. 2

Source Data Extended Data Fig. 3

Source Data Extended Data Fig. 4

Source Data Extended Data Fig. 5

Source Data Extended Data Fig. 6

Source Data Extended Data Fig. 7

Source Data Extended Data Fig. 8

Source Data Extended Data Fig. 9

Source Data Extended Data Fig. 10

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Comments

By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.