Abstract

A better understanding of the features that define the interaction between cancer cells and immune cells is important for the development of new cancer therapies1. However, focus is often given to interactions that occur within the primary tumour and its microenvironment, whereas the role of immune cells during cancer dissemination in patients remains largely uncharacterized2,3. Circulating tumour cells (CTCs) are precursors of metastasis in several types of cancer4,5,6, and are occasionally found within the bloodstream in association with non-malignant cells such as white blood cells (WBCs)7,8. The identity and function of these CTC-associated WBCs, as well as the molecular features that define the interaction between WBCs and CTCs, are unknown. Here we isolate and characterize individual CTC-associated WBCs, as well as corresponding cancer cells within each CTC–WBC cluster, from patients with breast cancer and from mouse models. We use single-cell RNA sequencing to show that in the majority of these cases, CTCs were associated with neutrophils. When comparing the transcriptome profiles of CTCs associated with neutrophils against those of CTCs alone, we detect a number of differentially expressed genes that outline cell cycle progression, leading to more efficient metastasis formation. Further, we identify cell–cell junction and cytokine–receptor pairs that define CTC–neutrophil clusters, representing key vulnerabilities of the metastatic process. Thus, the association between neutrophils and CTCs drives cell cycle progression within the bloodstream and expands the metastatic potential of CTCs, providing a rationale for targeting this interaction in treatment of 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.

Data availability

Data analysis, statistical testing and visualization were conducted in R (version 3.4.0; R Foundation for Statistical Computing). RNA and exome sequencing data have been deposited in the Gene Expression Omnibus (GEO, NCBI; accession number GSE109761) and the European Nucleotide Archive (ENA, EMBL-EBI; accession number PRJEB24623), respectively. Original R scripts to reproduce data analysis have been deposited to GitHub (accession URL, https://github.com/CMETlab/CTC-WBC). Source data for all mouse experiments are provided. All data are available from the corresponding author upon reasonable request.

Additional information

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

References

  1. 1.

    Khalil, D. N., Smith, E. L., Brentjens, R. J. & Wolchok, J. D. The future of cancer treatment: immunomodulation, CARs and combination immunotherapy. Nat. Rev. Clin. Oncol. 13, 273–290 (2016).

  2. 2.

    Mohme, M., Riethdorf, S. & Pantel, K. Circulating and disseminated tumour cells – mechanisms of immune surveillance and escape. Nat. Rev. Clin. Oncol. 14, 155–167 (2017).

  3. 3.

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

  4. 4.

    Aceto, N. et al. Circulating tumor cell clusters are oligoclonal precursors of breast cancer metastasis. Cell 158, 1110–1122 (2014).

  5. 5.

    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).

  6. 6.

    Pantel, K. & Speicher, M. R. The biology of circulating tumor cells. Oncogene 35, 1216–1224 (2016).

  7. 7.

    Aceto, N., Toner, M., Maheswaran, S. & Haber, D. A. En route to metastasis: circulating tumor cell clusters and epithelial-to-mesenchymal transition. Trends Cancer 1, 44–52 (2015).

  8. 8.

    Stott, S. L. et al. Isolation of circulating tumor cells using a microvortex-generating herringbone-chip. Proc. Natl Acad. Sci. USA 107, 18392–18397 (2010).

  9. 9.

    Xu, L. et al. Optimization and evaluation of a novel size based circulating tumor cell isolation system. PLoS ONE 10, e0138032 (2015).

  10. 10.

    Li, H. et al. Reference component analysis of single-cell transcriptomes elucidates cellular heterogeneity in human colorectal tumors. Nat. Genet. 49, 708–718 (2017).

  11. 11.

    Fridlender, Z. G. et al. Polarization of tumor-associated neutrophil phenotype by TGF-beta: ‘N1’ versus ‘N2’ TAN. Cancer Cell 16, 183–194 (2009).

  12. 12.

    Cristofanilli, M. et al. Circulating tumor cells, disease progression, and survival in metastatic breast cancer. N. Engl. J. Med. 351, 781–791 (2004).

  13. 13.

    Labelle, M., Begum, S. & Hynes, R. O. Direct signaling between platelets and cancer cells induces an epithelial-mesenchymal-like transition and promotes metastasis. Cancer Cell 20, 576–590 (2011).

  14. 14.

    Rothstein, G. et al. Stimulation of neutrophil production in CSF-1-responsive clones. Blood 72, 898–902 (1988).

  15. 15.

    He, J. Q. et al. Association of genetic variations in the CSF2 and CSF3 genes with lung function in smoking-induced COPD. Eur. Respir. J. 32, 25–34 (2008).

  16. 16.

    Verri, W. A., Jr et al. IL-15 mediates antigen-induced neutrophil migration by triggering IL-18 production. Eur. J. Immunol. 37, 3373–3380 (2007).

  17. 17.

    Kacinski, B. M. et al. FMS (CSF-1 receptor) and CSF-1 transcripts and protein are expressed by human breast carcinomas in vivo and in vitro. Oncogene 6, 941–952 (1991).

  18. 18.

    McCracken, J. M. & Allen, L. A. Regulation of human neutrophil apoptosis and lifespan in health and disease. J. Cell Death 7, 15–23 (2014).

  19. 19.

    Canli, O. et al. Myeloid cell-derived reactive oxygen species induce epithelial mutagenesis. Cancer Cell 32, 869–883 (2017).

  20. 20.

    Alexandrov, L. B. et al. Signatures of mutational processes in human cancer. Nature 500, 415–421 (2013).

  21. 21.

    Wellenstein, M. D. & de Visser, K. E. Cancer-cell-intrinsic mechanisms shaping the tumor immune landscape. Immunity 48, 399–416 (2018).

  22. 22.

    Ramasamy, S. et al. Tle1 tumor suppressor negatively regulates inflammation in vivo and modulates NF-κB inflammatory pathway. Proc. Natl Acad. Sci. USA 113, 1871–1876 (2016).

  23. 23.

    Yu, M. et al. Cancer therapy. Ex vivo culture of circulating breast tumor cells for individualized testing of drug susceptibility. Science 345, 216–220 (2014).

  24. 24.

    Macaulay, I. C. et al. Separation and parallel sequencing of the genomes and transcriptomes of single cells using G&T-seq. Nat. Protocols 11, 2081–2103 (2016).

  25. 25.

    Chen, S. et al. Genome-wide CRISPR screen in a mouse model of tumor growth and metastasis. Cell 160, 1246–1260 (2015).

  26. 26.

    Li, H. Aligning sequence reads, clone sequences and assembly contigs with BWA-MEM. Preprint at https://arxiv.org/abs/1303.3997 (2013)

Download references

Acknowledgements

We thank all patients that donated blood for our study, as well as all involved clinicians and study nurses; J. Massagué (Memorial Sloan Kettering Cancer Center), D. Haber and S. Maheswaran (Massachusetts General Hospital and Harvard Medical School) for donating cell lines; G. Christofori for MMTV-PyMT mice and comments on the manuscript, and all members of the Aceto laboratory for feedback and discussions; K. Eschbach and E. Burcklen from the Genomics Facility Basel (D-BSSE of the ETH Zürich) for generating sequencing libraries and performing next-generation sequencing; S. Arnold (D-BSSE of the ETH Zürich) and S. Münst Soysal (University Hospital Basel) for support with sample acquisition and processing; and T. Ryser (Aceto laboratory, University of Basel) for help with CRISPR–Cas9-related experiments. Calculations were performed at sciCORE (http://scicore.unibas.ch/) scientific computing center of the University of Basel. 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 the ETH Zürich and the University of Basel.

Reviewer information

Nature thanks K. Pantel and the other anonymous reviewer(s) for their contribution to the peer review of this work.

Author information

Affiliations

  1. Department of Biomedicine, Cancer Metastasis Lab, University of Basel and University Hospital Basel, Basel, Switzerland

    • Barbara Maria Szczerba
    • , Francesc Castro-Giner
    • , Ilona Krol
    • , Sofia Gkountela
    • , Manuel C. Scheidmann
    • , Cinzia Donato
    • , Ramona Scherrer
    •  & Nicola Aceto
  2. SIB Swiss Institute of Bioinformatics, Lausanne, Switzerland

    • Francesc Castro-Giner
    • , Jochen Singer
    •  & Niko Beerenwinkel
  3. Gynecologic Cancer Center, University Hospital Basel, Basel, Switzerland

    • Marcus Vetter
    • , Christian Kurzeder
    •  & Viola Heinzelmann-Schwarz
  4. Department of Medical Oncology, University Hospital Basel, Basel, Switzerland

    • Marcus Vetter
    • , Julia Landin
    •  & Christoph Rochlitz
  5. Department of Biosystems Science and Engineering, ETH Zurich, Basel, Switzerland

    • Jochen Singer
    • , Christian Beisel
    •  & Niko Beerenwinkel
  6. Breast Center, University of Basel and University Hospital Basel, Basel, Switzerland

    • Christian Kurzeder
    •  & Walter Paul Weber

Authors

  1. Search for Barbara Maria Szczerba in:

  2. Search for Francesc Castro-Giner in:

  3. Search for Marcus Vetter in:

  4. Search for Ilona Krol in:

  5. Search for Sofia Gkountela in:

  6. Search for Julia Landin in:

  7. Search for Manuel C. Scheidmann in:

  8. Search for Cinzia Donato in:

  9. Search for Ramona Scherrer in:

  10. Search for Jochen Singer in:

  11. Search for Christian Beisel in:

  12. Search for Christian Kurzeder in:

  13. Search for Viola Heinzelmann-Schwarz in:

  14. Search for Christoph Rochlitz in:

  15. Search for Walter Paul Weber in:

  16. Search for Niko Beerenwinkel in:

  17. Search for Nicola Aceto in:

Contributions

B.M.S. and N.A. designed the study, performed the experiments and wrote the manuscript. F.C.-G. performed the computational analysis. S.G., I.K., C.D. and R.S. processed blood samples, mouse tissues and performed immunofluorescence staining. M.C.S. performed CRISPR–Cas9-related experiments. M.V., J.L., C.K., V.H.-S., C.R. and W.P.W. provided patient samples and clinical input throughout the project. C.B. generated sequencing data. J.S. and N.B. provided input during computational methods development and data analysis. All authors have read, commented and approved the manuscript in its final form.

Competing interests

N.A. and B.M.S. are listed as inventors in patent applications that are related to CTC clusters and CTC–neutrophil clusters.

Corresponding author

Correspondence to Nicola Aceto.

Extended data figures and tables

  1. Extended Data Fig. 1 CTC capture in patients with breast cancer and in mouse models.

    a, Schematic of the CTC capture strategy with the Parsortix device. b, Schematic of the experimental design. Fifty single CTCs and fifty CTC clusters are spiked into blood to assess capture rate. c, Representative images of CTCs and WBCs captured on the Parsortix device and stained for EpCAM, HER2, EGFR (green) and CD45 (red) (left). Plot showing the mean CTC capture efficiency (right). d, Schematic of the experimental design. Fifty single CTCs are spiked into blood to evaluate artificial CTC aggregation rate during processing. e, Representative image of captured CTCs (left). The plot shows the mean per cent of captured single CTCs, CTC clusters and CTC–WBC clusters (right). n = 3; error bars, s.e.m. (ce). f, Representative images of CTCs from patients and mouse models. n = 34 for patients, n = 8 for NSG-LM2-GFP and NSG-4T1-GFP, n = 7 for BALB/c-4T1-GFP, n = 5 for MMTV-PyMT. g, Plot showing mean CTC counts in patients. n = 34; error bars, s.e.m. h, Plot showing the number of CTCs in each patient-derived CTC–WBC cluster. The red line represents the mean. i, Plot showing mean CTC counts in mouse models. n = 8 for NSG-CDX-BR16-GFP, NSG-LM2-GFP and NSG-4T1-GFP, n = 7 for BALB/c-4T1-GFP, n = 5 for MMTV-PyMT, error bars, s.e.m. j, Plot showing the number of CTCs in each mouse-model-derived CTC–WBC cluster. The red line represents the mean. k, Pie charts displaying the mean percentage of single CTCs (grey), CTC clusters (green) and CTC–WBC clusters (red) in mice upon blood draw via heart puncture (HP) or tumour-draining vessel (TDV) (left). The plots show the mean number of CTCs from the same experiment (right). l, Plot showing fold change of CTC counts, comparing HP versus TDV blood draw. Error bars, s.e.m.; n = 5 for NSG-CDX-BR16-GFP, n = 3 for NSG-LM2-GFP; P values by two-sided Student’s t-test are shown (k, l). m, Plots showing the number of CTCs in each mouse model-derived CTC–WBC cluster, isolated via HP or TDV. The red lines represent the mean. P values by two-sided Student’s t-test are shown. n, Plot showing the mean number of CTCs at day 10 after tumour inoculation, collected from HP, TDV or peripheral circulation (that is, tail vein; PC). Error bars, s.e.m.; n = 3; P values by two-sided Student’s t-test are shown. Source data

  2. Extended Data Fig. 2 Characterization of CTC-associated WBCs.

    a, Bar plot showing the expression levels of WBC marker CD45 in patient samples, including CTC-associated WBCs (red), free-floating peripheral WBCs (blue) and CTCs alone (green). b, Principal component analysis (PCA) of CTC-associated WBCs of patients and five reference WBC populations (n = 50). c, Bar plot showing the expression levels of CD45 in mouse samples, including CTC-associated WBCs (red) and CTCs alone (green). d, PCA of CTC-associated WBCs from all mouse models and five reference WBC populations (n = 47). e, Reference component analysis clustering of CTC-associated WBCs (red) and reference WBCs from mouse models, displaying projection scores of cells (columns, n = 47) on the immune reference panel (rows). f, Representative immunofluorescence images of mouse immune cells stained for CD45, Ly-6G and CD11b. Mouse lymphocytes (CD45+Ly-6GCD11b), mouse monocytes (CD45+Ly-6GCD11b+), and mouse granulocytes (CD45+Ly-6G+CD11blow) are shown (top). Representative immunofluorescence images of mouse lymphocytes and macrophages (peritoneum-derived) stained for F4/80 are shown (bottom). Mouse macrophages display a F4/80+ phenotype, whereas lymphocytes display a F4/80 phenotype (n = 3 for all). g, Representative images of human and mouse T cells, B cells, NK cells, monocytes and granulocytes stained with the Wright–Giemsa protocol to highlight nuclear morphology (left). Wright–Giemsa staining of the human CTC-derived cell line BR16 is also shown (right) (n = 3 for all). h, Representative immunofluorescence images of CTC–neutrophil clusters and CTC–monocyte clusters isolated from mouse models and stained for Ly-6G (gold) and CD11b (blue). CTCs stably express GFP (green). n = 8 for NSG-CDX-BR16-GFP, NSG-LM2-GFP and NSG-4T1-GFP, n = 7 for BALB/c-4T1-GFP. i, Representative images of CTC–neutrophil clusters stained with the Wright–Giemsa protocol to highlight nuclear morphology. n = 8 for NSG-CDX-BR16-GFP, NSG-LM2-GFP and NSG-4T1-GFP. j, Bar graph showing the mean number of CTC–neutrophil clusters and CTC–monocyte clusters in mouse models. Error bars, s.e.m.; n = 8 for NSG-CDX-BR16-GFP, NSG-LM2-GFP, NSG-4T1-GFP and n = 7 for BALB/c-4T1-GFP. k, Heat maps showing the projection scores of mouse-derived (left) and patient-derived (right) CTC-associated neutrophils (columns) on pro-tumoral (N2) neutrophil markers (rows). Source data

  3. Extended Data Fig. 3 Progression-free survival analysis in patients with breast cancer and in mouse models.

    a, Kaplan–Meier progression-free survival analysis comparing patients with one or more CTC–neutrophil cluster per 7.5 ml of peripheral blood (n = 9) versus all patients with no CTC–neutrophil clusters (n = 48). P value by two-sided log-rank test is shown. b, Kaplan–Meier PFS analysis comparing patients with one or more CTC–neutrophil cluster per 7.5 ml of peripheral blood (n = 9) versus patients with one or more CTC per 7.5 ml of peripheral blood but no CTC–neutrophil clusters (n = 21). P value by two-sided log-rank test is shown. c, Kaplan–Meier PFS analysis comparing patients with one or more CTC–neutrophil cluster per 7.5 ml of peripheral blood (n = 9), patients with one or more single CTC per 7.5 ml of peripheral blood but without CTC–neutrophil clusters (n = 14), and patients with one or more CTC cluster per 7.5 ml of peripheral blood but without CTC–neutrophil clusters (n = 7). P value by two-sided log-rank test is shown. Of note, these results are consistent with our previous observations whereby PFS differences in patients with single CTCs versus CTC–clusters were visible only when CTC clusters were present for multiple time points along disease progression. d, Schematic of the experimental design. One hundred CTCs from CTC–neutrophil clusters, CTC clusters or single CTCs are injected in the tail vein of tumour-free recipient mice to measure their metastatic potential. e, Plot showing normalized bioluminescence signal from the lungs of injected mice. n = 5 for all, error bars, s.e.m., P < 0.05 by two-sided Student’s t-test. f, Kaplan–Meier plot showing overall survival of injected mice. n = 5 for all, P values by two-sided log-rank test are shown. g, Representative image of a metastatic lesion in NSG mice injected intravenously with either with CTC–neutrophil clusters, CTC clusters or single CTCs from NSG-BR16-GFP mice. Metastases are stained for pan-cytokeratin (pCK, green) and DAPI (nuclei, blue) (left). The plot shows the mean number of metastatic foci per field of view (right). n = 3; error bars, s.e.m.; P values by two-sided Student’s t-test are shown. h, Representative image of a metastatic lesion in the lungs of BALB/c mice injected intravenously with either with CTC–neutrophil clusters, CTC clusters or single CTCs from BALB/c-4T1-GFP mice. Metastases are stained for pan-cytokeratin (pCK, green) and DAPI (nuclei; blue) (left). The plot shows the mean number of metastatic foci per field of view (right). n = 3; error bars, s.e.m.; P values two-sided Student’s t-test are shown. Source data

  4. Extended Data Fig. 4 Gene expression analysis of scRNA-seq data.

    a, t-SNE analysis of CTCs from CTC–neutrophil clusters and CTCs alone using the 500 most variable genes. t-SNE plots for BALB/c-4T1-GFP samples are coloured by number of detected genes (left) and number of reads per sample (right) (n = 29). b, t-SNE plots for patient samples coloured by number of detected genes (left) and number of reads per sample (right) (n = 68). c, Heat map showing the projection scores of mouse-model-derived CTCs from CTC–neutrophil clusters and CTCs alone in relation to epithelial and mesenchymal genes (n = 59). d, Heat map showing the projection scores of patient-derived CTCs from CTC–neutrophil clusters and CTCs alone in relation to epithelial and mesenchymal genes (n = 68). e, Heat map showing the projection scores of mouse-model-derived CTCs from CTC–neutrophil clusters and CTCs alone in relation to cancer stem-cell genes (n = 59). f, Heat map showing the projection scores of patient-derived CTCs from CTC–neutrophil clusters and CTCs alone in relation to cancer stem-cell genes (n = 68). g, Heat map showing the projection scores of mouse-model-derived CTCs from CTC–neutrophil clusters and CTCs alone in relation to platelet genes (n = 59). h, Heat map showing the projection scores of patient-derived CTCs from CTC–neutrophil clusters and CTCs alone in relation to platelet genes (n = 68).

  5. Extended Data Fig. 5 Proliferation of tumour cells adjacent to neutrophils in primary and metastatic tissues.

    a, Representative immunofluorescence images of NSG-LM2-GFP primary tumour and matched lung metastasis, stained for pan cytokeratin (pCK, green), MPO (gold), Ki67 (purple) and DAPI (nuclei, blue) (n = 3). b, Plots showing the mean per cent of Ki67-positive cancer cells in the primary tumour and metastatic sites (lung or brain) of mouse models, both overall and when considering only those cells that are adjacent to neutrophils. n = 3 for all; error bars, s.e.m.; ns, not significant by two-sided Student’s t-test. c, Representative immunofluorescence images of BR57 primary tumour and matched liver metastasis, stained for pan cytokeratin (pCK, green), MPO (gold), Ki67 (purple) and DAPI (nuclei, blue) (n = 3). d, Plots showing the mean percentage of Ki67-positive cancer cells, both overall and when considering only those cells that are adjacent to neutrophils, in matched primary and metastatic sites of nine patients with breast cancer. Error bars, s.e.m.; ns, not significant by two-sided Student’s t-test. e, Schematic of the experimental design. One hundred CTCs from CTC–neutrophil clusters or CTC alone are injected in the tail vein of recipient mice to measure disseminated tumour cells proliferation. f, Representative images of disseminated tumour cells stained for pan-cytokeratin (pCK, green), Ki67 (purple) and DAPI (nuclei, blue) (left). The plot shows the mean per cent of Ki67-positive DTCs (right). n = 3; error bars, s.e.m.; *P = 0.001 by two-sided Student’s t-test. Source data

  6. Extended Data Fig. 6 Characterization of cytokine-mediated crosstalk within CTC–neutrophil clusters.

    a, Schematic of the experimental design (top). The heat map shows the transcriptional landscape of cytokines and corresponding receptors expressed in at least 20% of CTC–neutrophil clusters (bottom). The cytokine–receptor pairs that are most frequently expressed in human cells are shown in red. b, Schematic of the experimental design (top). The heat map shows the transcriptional landscape of cytokine receptors and corresponding cytokines expressed in at least 20% of CTC–neutrophil clusters (bottom). The cytokine–receptor pairs that are expressed in at least 40% of CTC–neutrophil clusters are shown in red. c, The plot shows the mean 4T1-GFP cell number upon starvation and stimulation with IL-6, IL-1β, TNF-α, OSM or all four cytokines together (cytokine pool). n = 3; error bars, s.e.m.; *P < 0.05 by two-sided Student’s t-test. d, Plots showing the mean percentage of Ki67-positive disseminated tumour cells (DTCs) in the bone marrow of injected mice (n = 3 for all; error bars, s.e.m.; *P < 0.05 by two-sided Student’s t test). e, Plots showing normalized bioluminescence signal from the lungs of injected mice. n = 4 for all; error bars, s.e.m.; *P < 0.05 by two-sided Student’s t-test. f, Tumour growth curves of NSG mice injected with 4T1-Cas9-GFP cells expressing a control vector (Ctrl sgRNA) or sgRNAs targeting Il1r1 or Il6st. n = 3; error bars, s.e.m.; ns, not significant by two-sided Student’s t-test. g, Pie charts displaying the mean percentage of single CTCs (grey), CTC clusters (green) and CTC–neutrophil clusters (gold) of injected mice (left) (n = 3). The plots show the mean fold change of CTC ratios from injected mice (right). n = 3; error bars, s.e.m.; ns, not significant by two-sided Student’s t-test. h, Plots showing the mean per cent of Ki67-positive CTCs from injected mice. n = 3; error bars, s.e.m.; *P = 0.001 by two-sided Student’s t-test. Source data

  7. Extended Data Fig. 7 Mutation analysis of single-cell whole-exome sequencing data.

    a, Somatic mutation rate (mutations per Mb) of CTCs from CTC–neutrophil clusters (n = 14) versus CTCs alone (n = 56), normalized by donor. Lines within the violin plots show the 25th, 50th and 75th percentile, respectively, and dots represent individual CTCs. P value by two-sided Wilcoxon sign-ranked test is shown. b, Somatic mutation rate (mutations per Mb) in all CTCs isolated from donors with CTC–neutrophil clusters (donors (+); n = 6) and donors without CTC–neutrophil clusters (donors (−); n = 5). Lines within the violin plots show the 25th, 50th and 75th percentile, respectively, and dots represent individual CTCs. ns, not significant by two-sided Wilcoxon sign-ranked test. c, Nucleotide substitution pattern among putative somatic mutations in CTCs isolated from donors with CTC–neutrophil clusters (donors (+)) versus donors without CTC–neutrophil clusters (donors (−)). n = 6 for donors (+) and n = 5 for donors (−). Lines within the violin plots show the 25th, 50th and 75th percentile, respectively, and dots represent individual CTCs. P value by two-sided Wilcoxon sign-ranked test is shown. d, Bar plots showing the nucleotide context of given mutations in CTCs alone versus CTCs from CTC–neutrophil clusters. e, Plot showing the age distribution of donors (+) (n = 10) and donors (−) (n = 24). The red lines represent the mean. ns, not significant by two-sided Student’s t-test. f, The tile plot represent genes (columns) containing predicted high-impact mutations in at least two donors (+) and in none of the donors (−). g, Plots showing the mean fold change for MERTK and TLE1 (wild type or mutated) transcripts compared to control (Ctrl) cells. n = 3; error bars, s.e.m.; *P < 0.004 by two-sided Student’s t-test. h, Tumour growth curves representing mean tumour volume measurements of NSG mice injected with 4T1 cells carrying an empty vector (pLOC), wild type or mutated MERTK (left) or TLE1 (right) (n = 3; error bars, s.e.m.; ns, not significant by two-sided Student’s t-test). i, Representative images of the primary tumour of injected mice, stained for pan cytokeratin (pCK, green), myeloperoxidase (MPO, gold) and DAPI (nuclei, blue) (top) (n = 3). The plot shows the mean number of infiltrated neutrophils per field of view within the primary tumour (bottom). Error bars, s.e.m.; n = 3; ns, not significant by two-sided Student’s t-test. j, Representative images of the primary tumour of injected mice, stained for pan cytokeratin (pCK, green), myeloperoxidase (MPO, gold) and DAPI (nuclei, blue) (top) (n = 3). The plot shows the mean number of infiltrated neutrophils per field of view within the primary tumour (bottom). Error bars, s.e.m.; n = 3; *P = 0.002, **P = 0.0007 by two-sided Student’s t-test. k, Pie charts displaying the mean percentage of single CTCs (grey), CTC clusters (green) and CTC–neutrophil clusters (gold) in injected mice. The number of independent biological replicates (n) is shown for each condition. Source data

  8. Extended Data Fig. 8 Co-culture of cancer cells and neutrophils does not lead to the accumulation of key mutational events.

    a, Schematic of the experimental design. Neutrophils were purified from healthy donor blood and cultured with either CTC-derived cell lines (BR16, Brx50) or LM2 cells for 72 h. Tumour cells were collected, and isolated gDNA was processed for WES. b, Tile plot showing the mutation status of all key loci found mutated in patients with CTC–neutrophil clusters. None of the CTC–neutrophil-cluster-associated mutations were detected upon co-culture of cancer cells with neutrophils.

  9. Extended Data Fig. 9 Effects of neutrophil depletion or augmentation in mice.

    a, Plots showing the mean number of neutrophils in the circulation of mice treated with Ly-6G neutralizing antibodies (anti-Ly-6G) (left), or carrying G-CSF overexpressing tumours (right). Error bars, s.e.m.; the number of independent biological replicates (n) is provided in the Source Data; NA, not available; *P < 0.03, **P < 0.0001 by two-sided Student’s t-test. b, Representative images of the primary tumour of NSG-LM2-GFP mice stained for pan cytokeratin (pCK, green), myeloperoxidase (MPO, gold) and DAPI (nuclei, blue) (left). The plots show the mean number of infiltrated neutrophils per field of view within the tumour (right). W, weeks upon tumour development. Error bars, s.e.m.; n = 3; *P < 0.03, **P < 0.0001 by two-sided Student’s t-test. c, Tumour growth curves representing mean tumour volume measurements in the presence or absence of anti-Ly-6G antibodies or G-CSF overexpression. Error bars, s.e.m.; the number of independent biological replicates (n) is provided in the Source Data; ns, not significant by two-sided Student’s t-test. d, Plots showing the mean counts of single CTCs, CTC clusters and CTC-neutrophil clusters in mice. Error bars, s.e.m.; the number of independent biological replicates (n) is provided in the Source Data; ns, not significant; ND, not detected; *P < 0.05 by two-sided Student’s t-test. e, Pie charts displaying the mean percentage of single CTCs (grey), CTC clusters (green) and CTC–neutrophil clusters (gold) in NSG-LM2-GFP and NSG-CDX-BR16-GFP mice treated with anti-Ly-6G antibodies or G-CSF overexpression. W, weeks upon tumour development; the number of independent biological replicates (n) is shown for each condition. f, Plots showing the mean fold change of CTC ratios from NSG-LM2-GFP and NSG-CDX-BR16-GFP mice treated with anti-Ly-6G antibodies or G-CSF overexpression. Error bars, s.e.m.; the number of independent biological replicates (n) is provided in the Source Data; *P = 0.045, **P = 0.01, ***P = 0.004 by two-sided Student’s t-test. g, Representative bioluminescence images of lungs from mice treated with anti-Ly-6G antibodies or G-CSF overexpression (left); the number of independent biological replicates (n) is provided for simplicity directly within the Source Data; W, weeks upon tumour development. The plots show the mean metastatic index of mice treated with anti-Ly-6G antibodies or G-CSF overexpression (right). The number of independent biological replicates (n) is provided for simplicity directly within the Source Data; error bars, s.e.m.; *P < 0.03 **P < 0.01 by two-sided Student’s t-test. h, Kaplan–Meier survival plots showing overall survival rates of mice. The number of independent biological replicates (n) is provided in the Source Data; *P < 0.02 by two-sided log-rank test. i, Schematic of the experiment. NSG, FVB and BALB/c mice were pre-treated with anti-Ly-6G antibodies or control IgG. 4T1-GFP cells or Py2T-GFP cells were then injected into the tail vein to assess metastasis development. j, Plots showing mean normalized bioluminescence signal in the lungs of injected mice. k, Kaplan–Meier survival plot of injected mice. l, Plots showing the mean percentage of Ki67-positive disseminated tumour cells (DTCs) collected from the bone marrow of injected mice. n = 3; error bars, s.e.m.; ns, not significant by two-sided Student’s t-test (jl). m, Bar graph showing the proportion of patients with breast cancer who were treated with G-CSF, related to their CTC status. n = 42 for no CTCs, n = 23 for CTCs, n = 9 for CTC–neutrophil clusters; P value by two-sided Fisher’s exact test is shown. Source data

  10. Extended Data Fig. 10 Expression of cell-adhesion molecules–receptor pairs on CTC–neutrophil clusters.

    a, Schematic of the experimental design (top). The heat map shows the expression landscape of cell-adhesion molecules (CAMs) and corresponding receptors that are expressed in at least 20% of CTC–neutrophil clusters (bottom). The CAM–receptor pairs that are expressed in at least 50% of CTC–neutrophil clusters are shown in red. b, Schematic of the experiment (top). The heatmap shows the expression landscape of CAM receptors and corresponding CAMs that are expressed in at least 20% of CTC–neutrophil clusters (bottom). The CAM–receptor pairs that are expressed in at least 50% of CTC–neutrophil clusters are shown in red. c, Tumour growth curves representing mean tumour volume measurements of mice injected with 4T1-Cas9-GFP cells expressing a control vector (CTRL sgRNA) or sgRNA pools targeting F11r, Icam1, Itgb2 and Vcam1 (CRISPR pool). d, Plot showing the proportion of reads derived from sgRNAs targeting F11r, Icam1, Itgb2 and Vcam1 (4 sgRNAs each) in the 4T1-Cas9-GFP cell line upon library transduction as well as in three primary tumours from NSG-4T1-Cas9-GFP mice. All sgRNAs were represented in the tumour until the end of the experiment. e, Tumour growth curves representing mean tumour volume measurements of mice injected with 4T1-Cas9-GFP cells expressing a control vector (CTRL sgRNA) or individual sgRNAs targeting Vcam1. n = 3; error bars, s.e.m.; ns, not significant by two-sided Student’s t-test (c, e). Source data

Supplementary information

  1. Supplementary Information

    This file contains Supplementary Tables 1-5.

  2. Reporting Summary

  3. Video 1

    Micromanipulation and dissociation of a CTC cluster into single cells. The video shows the isolation of single cells from a CTC cluster through the use of a micromanipulator. The first cell is dissociated and isolated form a multicellular cluster, with this process being then repeated for each individual cell within the cluster. n=98.

Source data

About this article

Publication history

Received

Accepted

Published

DOI

https://doi.org/10.1038/s41586-019-0915-y

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.