The shaping and functional consequences of the microRNA landscape in breast cancer

Journal name:
Date published:
Published online

MicroRNAs (miRNAs) show differential expression across breast cancer subtypes, and have both oncogenic and tumour-suppressive roles1, 2, 3, 4, 5, 6. Here we report the miRNA expression profiles of 1,302 breast tumours with matching detailed clinical annotation, long-term follow-up and genomic and messenger RNA expression data7. This provides a comprehensive overview of the quantity, distribution and variation of the miRNA population and provides information on the extent to which genomic, transcriptional and post-transcriptional events contribute to miRNA expression architecture, suggesting an important role for post-transcriptional regulation. The key clinical parameters and cellular pathways related to the miRNA landscape are characterized, revealing context-dependent interactions, for example with regards to cell adhesion and Wnt signalling. Notably, only prognostic miRNA signatures derived from breast tumours devoid of somatic copy-number aberrations (CNA-devoid) are consistently prognostic across several other subtypes and can be validated in external cohorts. We then use a data-driven approach8 to seek the effects of miRNAs associated with differential co-expression of mRNAs, and find that miRNAs act as modulators of mRNA–mRNA interactions rather than as on–off molecular switches. We demonstrate such an important modulatory role for miRNAs in the biology of CNA-devoid breast cancers, a common subtype in which the immune response is prominent. These findings represent a new framework for studying the biology of miRNAs in human breast cancer.

At a glance


  1. Factors shaping the miRNA landscape across the breast cancer genome.
    Figure 1: Factors shaping the miRNA landscape across the breast cancer genome.

    a, miRNA position, detectability and association with CNAs and host gene transcription. Panels top to bottom: chromosome schematic; CNA frequency across cohort (red: amplification, >3 copies; pink: gain, 3 copies; light blue: heterozygous deletion, 1 copy; dark blue: homozygous deletion, 0 copies) and key breast cancer genes5 and miRNAs (red, oncogene; blue, tumour suppressor); location of miRNAs (black, detected; grey, non-detected); miRNAs significantly altered by CNAs (red, amplifications/gains; blue, heterozygous/homozygous deletions); miRNAs resident inside coding genes (green, correlated miRNA–host expression; grey, uncorrelated). miRNA counts are summarized across 100-kb bins. b, miRNA expression is uncommonly copy-number driven. The proportion of mRNA or miRNA expression variation explained by underlying copy-number changes, estimated by a GAM. c, Contributors to miRNA expression. Pairwise correlations between: all miRNA–mRNA pairs; each miRNA and its nearest mRNA neighbour, including 91 miRNAs on opposite strand of mRNA, 227 miRNAs encompassed by the coding strand of mRNAs, 59 miRNA pairs derived from a single pre-miRNA and 42 miRNAs arising from putative polycistrons. d, miRNA promoters span fewer CpG islands than mRNA promoters. The proportion of mRNA, miRNAs or a random selection of >6,000 genomic loci whose nearest upstream promoter-proximal regulatory module is within 2 kb of predicted CpG or methylated CpG (mCpG) islands.

  2. Cellular processes connected to the miRNA landscape.
    Figure 2: Cellular processes connected to the miRNA landscape.

    a, miRNA variation versus sample meta-data and mRNA expression. GAM evaluation of the contribution of all variable mRNAs or established clinical, histopathological or molecular parameters to the observed variation in expression of 302 variable miRNAs (intensity 9.36±2.24). miRNAs related to immune response (miR-150, miR-155, miR-142-5p, miR-142-3p and miR-146a) are boxed. LN, lymph-node positive; PR, progesterone receptor status. b, Groups of miRNAs are concerted by key cellular pathways. Unsupervised clustering of GAM values from all variable mRNA–miRNA combinations (Euclidean distance; Ward’s minimum variance agglomeration). Gene Ontology terms significantly enriched in particular branches are listed (Supplementary Table 8). ECM, extracellular matrix; TS, tumour suppressor. c, The miRNA–mRNA landscape is dominated by positive associations. GAM values versus the directional Pearson correlation values for all mRNA–miRNA pairs or only pairs comprising miRNAs and their predicted or experimentally validated targets. d, miRNA–mRNA relations differ systematically in ER+ and ER samples. The GAM values calculated for ER+ and ER samples separately are contrasted, with mRNA–miRNA pairs pertaining to named Gene Ontology (GO) terms highlighted. Insets show expression of individual example pairs (generalized trends discussed in the text are circled approximately).

  3. miRNAs have an increased prognostic value in the genomically stable iClust4.
    Figure 3: miRNAs have an increased prognostic value in the genomically stable iClust4.

    a, Individual miRNAs are not robust prognostic factors in breast cancer. Survival prognosis by individual miRNAs, using a multivariate Cox proportional hazard model adjusted for age at diagnosis, tumour size, lymph-node status, grade and ER status and corrected for multiple testing. Numbers below boxes denote number of samples in each subtype. iClust4m, iClust4mimic. b, Example hazard ratios across multiple studies. Log2 proportional hazard ratios and 95th confidence intervals across external data sets (Supplementary Fig. 1a) for miRNAs with adjusted Pvalues <0.1 across our entire cohort. Size of squares are proportional to the weight assigned to each result. c, Compound miRNA signatures are more prognostic in iClust4. PC1 was calculated for subsets of miRNAs with an individual prognostic value of <0.2 in select subtypes, and used as a stratifier for Kaplan–Meier survival analysis. Pvalues: multivariate Cox analyses. d, The iClust4 miRNA signature is prognostic in other subtypes and external cohorts. Kaplan–Meier survival analysis stratified by the PC1 based on 17 canonical miRNAs from iClust4; validated in external data sets with a balanced sample composition. Proportion of samples divided by PC1 is colour coded as in the key. Numbers of samples are in brackets. ‘miRNAs’ denotes the number of the 17 iClust4 miRNAs represented in the study. Pvalues: univariate Cox model (not all stratifying variables were available for the external data sets). Detailed PC1-stratified log2 proportional hazard ratios are shown in insets, cropped at (−3,2).

  4. miRNAs have a significant modulatory role in iClust4.
    Figure 4: miRNAs have a significant modulatory role in iClust4.

    a, Example of effector–modulator–target relationship. Left, effector (E)–modulator (M)–target (T) schematic. Right, the interaction between effector mRNA ALCAM and its two target mRNAs, ZEB3 and CLDN3, differs in the subsets of samples with high or low expression of miR-301a, classifying the miRNA as a conceptual modulator. b, miRNAs have an increased modulator activity in iClust4. The number of effector–target pairs for each potential miRNA modulator across tumour subtypes. c, iClust4 has an increased number of highly connected modulators and effectors. Distribution of the number of highly connected modulators (>50 targets; arbitrary cutoff) per effector. For ER+ the boxes show the distribution across ten random samplings of 299 ER+ tumours, used as a contrast to the 299 ER tumours. d, e, Network of immune-response-related Gene Ontology terms enriched among iClust4 targets. miRNAs (blue) modulate Gene Ontology terms descendants of ‘immune response’ (grey) in iClust4 (d). Highest connected miRNA and Gene Ontology term hubs are magnified (e). put-miR, putative miRNA.


  1. Le Quesne, J. & Caldas, C. Micro-RNAs and breast cancer. Mol. Oncol. 4, 230241 (2010)
  2. Buffa, F. M. et al. microRNA-associated progression pathways and potential therapeutic targets identified by integrated mRNA and microRNA expression profiling in breast cancer. Cancer Res. 71, 56355645 (2011)
  3. Enerly, E. et al. miRNA-mRNA integrated analysis reveals roles for miRNAs in primary breast tumors. PLoS ONE 6, e16915 (2011)
  4. Farazi, T. A. et al. MicroRNA sequence and expression analysis in breast tumors by deep sequencing. Cancer Res. 71, 44434453 (2011)
  5. Lyng, M. B. et al. Global microRNA expression profiling of high-risk ER+ breast cancers from patients receiving adjuvant tamoxifen mono-therapy: a DBCG study. PLoS ONE 7, e36170 (2012)
  6. The Cancer Genome Atlas Network. Comprehensive molecular portraits of human breast tumours. Nature 490, 6170 (2012)
  7. Curtis, C. et al. The genomic and transcriptomic architecture of 2,000 breast tumours reveals novel subgroups. Nature 486, 346352 (2012)
  8. Wang, K. et al. Genome-wide identification of post-translational modulators of transcription factor activity in human B cells. Nature Biotechnol. 27, 829837 (2009)
  9. Perou, C. M. et al. Molecular portraits of human breast tumours. Nature 406, 747752 (2000)
  10. Olive, V., Jiang, I. & He, L. mir-17–92, a cluster of miRNAs in the midst of the cancer network. Int. J. Biochem. Cell Biol. 42, 13481354 (2010)
  11. Valastyan, S. & Weinberg, R. A. miR-31: a crucial overseer of tumor metastasis and other emerging roles. Cell Cycle 9, 21242129 (2010)
  12. Blenkiron, C. et al. MicroRNA expression profiling of human breast cancer identifies new markers of tumor subtype. Genome Biol. 8, R214 (2007)
  13. Luqmani, Y. A., Al Azmi, A., Al Bader, M., Abraham, G. & El Zawahri, M. Modification of gene expression induced by siRNA targeting of estrogen receptor α in MCF7 human breast cancer cells. Int. J. Oncol. 34, 231242 (2009)
  14. Seitz, H. et al. A large imprinted microRNA gene cluster at the mouse Dlk1-Gtl2 domain. Genome Res. 14, 17411748 (2004)
  15. Krol, J., Loedige, I. & Filipowicz, W. The widespread regulation of microRNA biogenesis, function and decay. Nature Rev. Genet. 11, 597610 (2010)
  16. Bezman, N. A., Chakraborty, T., Bender, T. & Lanier, L. L. miR-150 regulates the development of NK and iNKT cells. J. Exp. Med. 208, 27172731 (2011)
  17. Faraoni, I., Antonetti, F. R., Cardone, J. & Bonmassar, E. miR-155 gene: a typical multifunctional microRNA. Biochim. Biophys. Acta 1792, 497505 (2009)
  18. Xu, W. D., Lu, M. M., Pan, H. F. & Ye, D. Q. Association of microRNA-146a with Autoimmune Diseases. Inflammation 35, 15251529 (2012)
  19. Andreopoulos, B. & Anastassiou, D. Integrated analysis reveals hsa-miR-142 as a representative of a lymphocyte-specific gene expression and methylation signature. Cancer Inform. 11, 6175 (2012)
  20. Teschendorff, A. E., Miremadi, A., Pinder, S. E., Ellis, I. O. & Caldas, C. An immune response gene expression module identifies a good prognosis subtype in estrogen receptor negative breast cancer. Genome Biol. 8, R157 (2007)
  21. Peng, X. et al. Identification of miRs-143 and -145 that is associated with bone metastasis of prostate cancer and involved in the regulation of EMT. PLoS ONE 6, e20341 (2011)
  22. Li, B. et al. Down-regulation of miR-214 contributes to intrahepatic cholangiocarcinoma metastasis by targeting Twist. FEBS J. 279, 23932398 (2012)
  23. Git, A. et al. PMC42, a breast progenitor cancer cell line, has normal-like mRNA and microRNA transcriptomes. Breast Cancer Res. 10, R54 (2008)
  24. Guo, H., Hu, X., Ge, S., Qian, G. & Zhang, J. Regulation of RAP1B by miR-139 suppresses human colorectal carcinoma cell proliferation. Int. J. Biochem. Cell Biol. 44, 14651472 (2012)
  25. Castellano, L. et al. The estrogen receptor-α-induced microRNA signature regulates itself and its transcriptional response. Proc. Natl Acad. Sci. USA 106, 1573215737 (2009)
  26. Boquest, A. C. et al. Isolation and transcription profiling of purified uncultured human stromal stem cells: alteration of gene expression after in vitro cell culture. Mol. Biol. Cell 16, 11311141 (2005)
  27. Scheel, C. et al. Paracrine and autocrine signals induce and maintain mesenchymal and stem cell states in the breast. Cell 145, 926940 (2011)
  28. Schmidt, W. M., Spiel, A. O., Jilma, B., Wolzt, M. & Muller, M. In vivo profile of the human leukocyte microRNA response to endotoxemia. Biochem. Biophys. Res. Commun. 380, 437441 (2009)
  29. Bair, E. & Tibshirani, R. Semi-supervised methods to predict patient survival from gene expression data. PLoS Biol. 2, e108 (2004)
  30. Fabbri, M. et al. MicroRNAs bind to Toll-like receptors to induce prometastatic inflammatory response. Proc. Natl Acad. Sci. USA 109, E2110E2116 (2012)

Download references

Author information

  1. These authors contributed equally to this work.

    • Heidi Dvinge &
    • Anna Git


  1. Cancer Research UK Cambridge Institute and Department of Oncology, University of Cambridge, Li Ka Shing Centre, Robinson Way, Cambridge CB2 0RE, UK

    • Heidi Dvinge,
    • Anna Git,
    • Stefan Gräf,
    • Suet-Feung Chin &
    • Carlos Caldas
  2. European Molecular Biology Laboratory–European Bioinformatics Institute, Wellcome Trust Genome Campus, Cambridge CB10 1SD, UK

    • Mali Salmon-Divon
  3. Department of Molecular Biology, Ariel University Center of Samaria, Ariel 40700, Israel

    • Mali Salmon-Divon
  4. Department of Preventive Medicine, University of Southern California, Los Angeles, California 90033, USA

    • Christina Curtis &
    • Andrea Sottoriva
  5. Genome Sciences Centre, BC Cancer Agency, 675 West 10th Avenue, Vancouver V5Z 1L3, Canada

    • Yongjun Zhao &
    • Martin Hirst
  6. Department of Pathology and Laboratory Medicine, University of British Columbia, Vancouver V6T 2B5, Canada

    • Yongjun Zhao,
    • Martin Hirst,
    • Gulisa Turashvili &
    • Sam Aparicio
  7. Wellcome Trust Cancer Research UK Gurdon Institute and Department of Biochemistry, University of Cambridge, The Henry Wellcome Building of Cancer and Developmental Biology, Cambridge CB2 1QN, UK

    • Javier Armisen &
    • Eric A. Miska
  8. Cambridge Breast Unit, Addenbrooke’s Hospital, Cambridge University Hospital NHS Foundation Trust and NIHR Cambridge Biomedical Research Centre, Cambridge CB2 2QQ, UK

    • Elena Provenzano &
    • Carlos Caldas
  9. Molecular Oncology, British Columbia Cancer Research Centre, Vancouver V5Z 1L3, Canada

    • Gulisa Turashvili &
    • Sam Aparicio
  10. Department of Histopathology, School of Molecular Medical Sciences, University of Nottingham, Nottingham NG5 1PB, UK

    • Andrew Green &
    • Ian Ellis
  11. Cambridge Experimental Cancer Medicine Centre, Cambridge CB2 0RE, UK

    • Carlos Caldas
  12. Present addresses: Basic Sciences Division, Fred Hutchinson Cancer Research Center, Seattle, Washington 98109, USA (H.D.); Department of Medicine, University of Cambridge, Addenbrooke’s Hospital, Hills Road, Cambridge CB2 0QQ, UK (S.G.).

    • Heidi Dvinge &
    • Stefan Gräf


H.D. and A.Git led the analysis and drafted the manuscript with C.Caldas; A.Git, S.G., S.A. and C.Caldas designed and coordinated the study; A.Git carried out all microarray and quantitative reverse transcriptase PCR laboratory work. Sequencing data were provided by Y.Z., M.H., J.A., E.A.M. and S.A. and analysed by M.S.-D., who also analysed external epigenetic data; S.G. designed custom microarray probes and contributed to array pre-processing. C.Curtis and A.S. processed external CNA data. S.-F.C., E.P., A.Green, I.E., G.T., S.A. and C.Caldas coordinated collection and processing of clinical material and associated clinical and histopathological information. S.A. and C.Caldas are joint senior authors and project co-leaders.

Competing financial interests

The authors declare no competing financial interests.

Corresponding authors

Correspondence to:

The raw non-coding RNA microarray data is available through the European Genome–Phenome Archive (, which is hosted by the EBI, under accession number GAS00000000122.

Author details

Supplementary information

PDF files

  1. Supplementary Information (8 MB)

    This file contains Supplementary Methods, Supplementary Appendix containing R source code for the modulatory effect of miRNAs, Supplementary References and Supplementary Figures 1-8.

Zip files

  1. Supplementary Data (2.9 MB)

    This file contains Supplementary Tables 1-9. Supplementary Table 1 contains detailed clinical and histopathological information for the 1,302 tumors; Supplementary Table 2 shows annotation, potential cross-hybridization and distribution of 853 detectable miRNAs; Supplementary Table 3 shows miRNAs regulated by CNAs across the entire cohort or in ER+/ER- sub-cohorts; Supplementary Table 4 contains a list of minimal common regions of CNAs (across the entire cohort or in ER+/ER- sub-cohorts) which contain miRNAs and mRNAs; Supplementary Table 4 contains a list of minimal common regions of CNAs (across the entire cohort or in ER+/ER- sub-cohorts) which contain only miRNAs; Supplementary Table 5 shows intensity and correlation of detectable sibling miRNAs; Supplementary Table 6 contains a list of differentially expressed miRNAs between ER+ and ER- sub-cohorts or between Pam50 subtypes; Supplementary Table 7 shows calculated pairwise generalized additive model values for variable mRNAs and miRNAs; Supplementary Table 8 contains a list of significantly enriched Gene Ontology Biological Process terms in mRNAs clustered by miRNA-mRNA GAM values and Supplementary Table 9 contains a summary of ER and Her2 statuses of the breast cancer cell lines used in this study.

Additional data