Skip to main content

Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

In vivo profiling of metastatic double knockouts through CRISPR–Cpf1 screens

Abstract

Systematic investigation of the genetic interactions that influence metastatic potential has been challenging. Here we developed massively parallel CRISPR–Cpf1/Cas12a crRNA array profiling (MCAP), an approach for combinatorial interrogation of double knockouts in vivo. We designed an MCAP library of 11,934 arrays targeting 325 pairwise combinations of genes implicated in metastasis. By assessing the metastatic potential of the double knockouts in mice, we unveiled a quantitative landscape of genetic interactions that drive metastasis.

This is a preview of subscription content

Access options

Rent or Buy article

Get time limited or full article access on ReadCube.

from$8.99

All prices are NET prices.

Fig. 1: In vivo profiling of metastatic double knockouts by MCAP.
Fig. 2: Identification of synergistic mutation combinations.
Fig. 3: Nf2 and Trim72 mutations jointly promote lung metastasis in vivo.

Data availability

MCAP data, sequences of oligos and library design are described in the Methods section and Supplementary Tables. All vectors and libraries have been deposited to Addgene and are available to the academic community. Cell lines and all data supporting this work will be made available to the academic community upon reasonable request to the corresponding author. Genomic sequencing data have been deposited with NCBI SRA (PRJNA515306).

Code availability

Key scripts used to process and analyze the data will be available to the academic community upon reasonable request.

References

  1. 1.

    Lambert, A. W., Pattabiraman, D. R. & Weinberg, R. A. Cell 168, 670–691 (2017).

    CAS  Article  PubMed  PubMed Central  Google Scholar 

  2. 2.

    Schunder, E., Rydzewski, K., Grunow, R. & Heuner, K. Int. J. Med. Microbiol. 303, 51–60 (2013).

    CAS  Article  PubMed  PubMed Central  Google Scholar 

  3. 3.

    Zetsche, B. et al. Cell 163, 759–771 (2015).

    CAS  Article  PubMed  PubMed Central  Google Scholar 

  4. 4.

    Zetsche, B. et al. Nat. Biotechnol. 35, 31–34 (2017).

    CAS  Article  PubMed  Google Scholar 

  5. 5.

    Chen, S. et al. Cell 160, 1246–1260 (2015).

    CAS  Article  PubMed  PubMed Central  Google Scholar 

  6. 6.

    Kumar, M. S. et al. Genes Dev. 23, 2700–2704 (2009).

    CAS  Article  PubMed  PubMed Central  Google Scholar 

  7. 7.

    Robinson, D. R. et al. Nature 548, 297–303 (2017).

    CAS  Article  PubMed  PubMed Central  Google Scholar 

  8. 8.

    Wong, A. S. L., Choi, G. C. G., Cheng, A. A., Purcell, O. & Lu, T. K. Nat. Biotechnol. 33, 952–961 (2015).

    CAS  Article  PubMed  PubMed Central  Google Scholar 

  9. 9.

    Wong, A. S. L. et al. Proc. Natl Acad. Sci. USA 113, 2544–2549 (2016).

    CAS  Article  PubMed  Google Scholar 

  10. 10.

    Roguev, A. et al. Nat. Methods 10, 432–437 (2013).

    CAS  Article  PubMed  PubMed Central  Google Scholar 

  11. 11.

    Han, K. et al. Nat. Biotechnol. 35, 463–474 (2017).

    CAS  Article  PubMed  PubMed Central  Google Scholar 

  12. 12.

    Shen, J. P. et al. Nat. Methods 14, 573–576 (2017).

    CAS  Article  PubMed  PubMed Central  Google Scholar 

  13. 13.

    Horlbeck, M. A. et al. Cell 174, 953–967 (2018).

    CAS  Article  PubMed  Google Scholar 

  14. 14.

    Najm, F. J. et al. Nat. Biotechnol. 36, 179–189 (2018).

    CAS  Article  PubMed  Google Scholar 

  15. 15.

    Erard, N., Knott, S. R. V. & Hannon, G. J. Mol. Cell 67, 348–354 (2017).

    CAS  Article  PubMed  PubMed Central  Google Scholar 

  16. 16.

    Boettcher, M. et al. Nat. Biotechnol. 36, 170–178 (2018).

    CAS  Article  PubMed  PubMed Central  Google Scholar 

  17. 17.

    Chow, R. D., Kim, H. R. & Chen, S. Nat. Commun. 9, 1903 (2018).

    Article  PubMed  PubMed Central  Google Scholar 

  18. 18.

    Kim, H. K. et al. Nat. Methods 14, 153–159 (2017).

    CAS  Article  PubMed  Google Scholar 

  19. 19.

    Kim, H. K. et al. Nat. Biotechnol. 36, 239–241 (2018).

    CAS  Article  PubMed  Google Scholar 

  20. 20.

    Kleinstiver, B. P. et al. Nat. Biotechnol. 37, 276–282 (2019).

    Article  PubMed  Google Scholar 

  21. 21.

    Chow, R. D., Wang, G., Ye, L. & Chen, S. Protocol Exchange https://doi.org/10.1038/protex.2019.018 (2019).

  22. 22.

    Langmead, B., Trapnell, C., Pop, M. & Salzberg, S. L. Genome Biol. 10, R25 (2009).

    Article  PubMed  PubMed Central  Google Scholar 

  23. 23.

    Li, H. & Durbin, R. Bioinformatics 25, 1754–1760 (2009).

    CAS  Article  PubMed  PubMed Central  Google Scholar 

  24. 24.

    Li, H. et al. Bioinformatics 25, 2078–2079 (2009).

    Article  PubMed  PubMed Central  Google Scholar 

  25. 25.

    Koboldt, D. C. et al. Genome Res. 22, 568–576 (2012).

    CAS  Article  PubMed  PubMed Central  Google Scholar 

  26. 26.

    Zhang, J., Kobert, K., Flouri, T. & Stamatakis, A. Bioinformatics 30, 614–620 (2014).

    CAS  Article  PubMed  Google Scholar 

  27. 27.

    Martin, M. EMBnet J. 17, 10–12 (2011).

    Article  Google Scholar 

Download references

Acknowledgements

We thank all members of the Chen laboratory, as well as various colleagues in the Department of Genetics, Systems Biology Institute, Immunobiology Program, BBS Program, MSTP Program, Comprehensive Cancer Center and Stem Cell Center at Yale, for their assistance and scientific discussion. We thank the Center for Genome Analysis, Center for Molecular Discovery, Pathology Tissue Services, Histology Services, High Performance Computing Center, West Campus Analytical Chemistry Core and West Campus Imaging Core and Keck Biotechnology Resource Laboratory at Yale, for technical support. S.C. is supported by Yale SBI/Genetics Startup Fund, Damon Runyon Dale Frey Award (No. DFS-13-15), Melanoma Research Alliance (No. 412806 and 16-003524), St-Baldrick’s Foundation (No. 426685), Breast Cancer Alliance, Cancer Research Institute, AACR (No. 499395 and 17-20-01-CHEN), The Mary Kay Foundation (No. 017-81), The V Foundation (No. V2017-022), Ludwig Family Foundation, DoD (No. W81XWH-17-1-0235), Sontag Foundation (DSA Award), Chenevert Family Foundation and NIH/NCI (No. 1DP2CA238295-01, 1R01CA231112-01, 1U54CA209992-8697, 5P50CA196530-A10805 and 4P50CA121974-A08306). R.D.C. and M.B.D. are supported by the Yale MSTP training grant from NIH (No. T32GM007205). G.W. is supported by CRI Irvington and RJ Anderson Postdoctoral Fellowships. A.C. is supported by a Yale PhD training grant from NIH (No. T32GM007223).

Author information

Affiliations

Authors

Contributions

R.D.C. designed the study, performed experiments, developed statistical algorithms and computational pipelines and analyzed the data. G.W. and L.Y. performed molecular, cellular and animal experiments and MCAP readout. A.C. optimized in vivo screens and assisted with validation experiments. H.R.K. and L.S. assisted with cloning and validation experiments. M.B.D. assisted with flow cytometry experiments. Y.E. assisted with in vitro assays. S.C. conceived the study, provided conceptual advice, secured funding and supervised the work. R.D.C. and S.C. wrote the manuscript.

Corresponding author

Correspondence to Sidi Chen.

Ethics declarations

Competing interests

The authors have filed a provisional patent related to this work.

Additional information

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

Integrated supplementary information

Supplementary Figure 1 Double knockout of Nf1 and Pten by a single Cpf1 crRNA array.

a, Schematic maps of the constructs for double knockout (DKO) experiments using a single Cpf1 crRNA array. b, One-step cloning strategy for DKO experiments using CRISPR–Cpf1. c, Schematic of the experimental approach for evaluating whether both permutations of an Nf1 and Pten dual-crRNA array can induce mutagenesis at both target loci. d, Seven days after lentiviral infection, genomic DNA was collected for mutation analysis by Nextera sequencing. For each treatment condition, mutations were identified at the genomic loci targeted by crPten (left column, blue) and by crNf1 (right column, orange). Variant frequencies associated with each mutation are shown in the boxes to the right; for each condition, the top five most frequent variants are shown. The total mutation frequencies are noted below.

Supplementary Figure 2 Evaluation of in vivo library diversity in the absence of mutagenesis.

a, Experimental design to evaluate the suitability of the in vivo transplant model for high-throughput genetic interrogation. To model neutral selection in the absence of mutagenesis, KPD cells were infected with a lentiviral library containing random 8-mer barcodes, for a theoretical total of 48 = 65,536 unique barcodes. Twelve days after injection into mice, genomic DNA was extracted from the nodules for barcode sequencing and assessment of library diversity. b, Bar plot detailing the percentage of all possible 8-mers (n = 65,536) that were recovered in each sample (cell pool, n = 1 infection replicate; nu/nu mice, n = 2 mice; Rag–/– mice, n = 4 mice). c, Tukey box plots (IQR boxes with 1.5 × IQR whiskers and all outliers shown individually) of the abundances of all possible 8-mers (n = 65,536) in each sample (cell pool, n = 1 infection replicate; nu/nu mice, n = 2 mice; Rag–/– mice, n = 4 mice).

Supplementary Figure 3 Analysis of the barcoded MCAP-MET plasmid library.

a, Density plot showing the distribution of MCAP-MET library abundance. All 11,934 crRNA arrays were detected in the plasmid library. b, Density plot of the number of unique barcodes associated with each crRNA array. A total of 774,296 unique barcoded crRNA arrays (BC arrays) were detected in the MCAP-MET plasmid library. c, Histogram (left) and CDF (right) of the data in b, fitted to a negative binomial distribution (n = 774,296 BC arrays). d, Empirical CDF of the abundance of all detected BC arrays (n = 774,296) in the MCAP-MET library (left), and a violin density plot of the abundances (right). e, Scatter plot of the normalized MCAP-MET library abundance in plasmid (n = 1) and averaged cell pools (n = 6).

Supplementary Figure 4 Clonal analysis in primary tumors and lung metastases at ≥0.001% frequency.

a, Tukey box plots (IQR boxes with 1.5 × IQR whiskers) detailing the distribution of clone frequencies in pre-injection cell pools (n = 6), indicated by the abundances of each BC array in each sample (n = 164,597, 184,002, 170,638, 172,803, 171,011 and 171,509 BC arrays, from top to bottom). b, Bar plot of the number of clones present at ≥0.001% frequency in cell pools, primary, tumors and lung metastases. c, Violin density plots with overlaid data points of the log2 number of clones present at ≥0.001% frequency in cell pools (gray, n = 6 cell replicates), primary tumors (orange, n = 10 mice) and lung metastases (blue, n = 37 from ten mice). Cells versus primary tumors (two-sided Wilcoxon rank sum test, P = 0.0002), cells versus lung metastases (P = 0.0001) and primary tumors versus lung metastases (P = 0.0162). d, Dot plot of the relative frequencies of clones at ≥0.001% frequency. Relative frequencies are expressed as percentages of total reads in each sample. Points are colored and labeled below by cell sample/mouse ID as in b. e, Empirical CDF of all clones at ≥0.001% frequency, expressed as percentages of total reads in each sample, aggregated by sample type: cell pools (gray, n = 6 cell replicates), primary tumors (orange, n = 10 mice) and lung metastases (blue, n = 37 from ten mice). The clone size distributions in primary tumors and lung metastases were significantly different (two-sided Kolmogorov–Smirnov test, P < 2.2 × 10–16). f, Violin density plots of Shannon diversity indices in cell pools (gray, n = 6 cell replicates), primary tumors (orange, n = 10 mice) and lung metastases (blue, n = 37 from ten mice) for clones at ≥0.001% frequency. Cells versus primary tumors (two-sided Wilcoxon rank sum test, P = 0.0002), cells versus lung metastases (P = 3.28 × 10–7) and primary tumors versus lung metastases (P = 0.0212).

Supplementary Figure 5 Clonal analysis in primary tumors and lung metastases at ≥0.01% frequency.

a, Bar plot of the number of clones present at ≥0.01% frequency in primary tumors and lung metastases. Note that cell samples do not have clones passing this frequency cutoff due to the high diversity in the population, and therefore were not included in this Figure. b, Violin density plots of the number of clones present at ≥0.01% frequency in primary tumors (orange, n = 10 mice) and lung metastases (blue, n = 37 from ten mice). Collectively, primary tumors had significantly more clones at ≥0.01% frequency than lung metastases (two-sided Wilcoxon rank sum test, P = 0.0023). c, Dot plot of the relative frequencies of clones at ≥0.01% frequency across primary tumors and lung metastases. Relative frequencies are expressed as percentages of total reads in each sample. Points are colored by mouse ID, as annotated below and in a. d, Empirical CDF of all clones at ≥0.01% frequency in primary tumors (orange, n = 10 mice) and lung metastases (blue, n = 37 from ten mice), expressed as percentages of total reads in each sample and aggregated by sample type. The clone size distributions in primary tumors and lung metastases were significantly different (two-sided Kolmogorov–Smirnov test, P = 0.0412). e, Violin density plots of Shannon diversity indices in primary tumors (orange, n = 10 mice) and lung metastases (blue, n = 37 from ten mice) for clones at ≥0.01% frequency. Primary tumors were significantly more diverse with regard to clone frequency distribution (two-sided Wilcoxon rank sum test, P = 0.0183).

Supplementary Figure 6 Representation of MCAP-MET crRNA array library in plasmid, cells, primary tumors and lung metastases.

a, Heat map of pairwise Spearman correlation coefficients of crRNA array log2 RPM abundance from MCAP-MET plasmid library (n = 1), MCAP transduced cells before transplantation (day 7 or 14 post infection, n = 6 cell replicates), primary tumors (n = 10 mice), and lung metastases (n = 37 from ten mice). b, Tukey box plots (IQR boxes, with 1.5 × IQR whiskers and all points overlaid) of crRNA array abundances (n = 11,934 arrays) from MCAP-MET profiling experiments. MCAP-MET plasmid library (n = 1), MCAP transduced cells before transplantation (day 7 or 14 post infection, n = 6 cell replicates), primary tumors (n = 10 mice) and lung metastases (n = 37 from ten mice).

Supplementary Figure 7 Binary FDR-based crRNA array enrichment analyses.

a, Venn diagram of gene pairs that were enriched in ≥50% of primary tumors or lung metastases. b,c, Histogram detailing the percentage of independent crRNA arrays that were enriched in primary tumors (b) or lung metastases (c) for each single gene (left) or gene pair (middle). Right, table of the top genes/gene pairs in terms of the percentage of independent crRNA arrays that were enriched in primary tumors (b) or lung metastases (c). Text colors correspond to the histograms. df, Enrichment bar plots of multiple independent crRNA arrays targeting Nf2_Rb1 (d), Nf2_Pten (e) and Nf2_Trim72 (f) in lung metastases (n = 37 from ten mice).

Supplementary Figure 8 Additional comparisons of crRNA array abundances.

ac, Scatter plot of MCAP-MET crRNA array abundance in cell pools versus primary tumors (a), cell pools versus lung metastases (b) and primary tumors versus lung metastases (c). Data are shown in terms of average log2 RPM across the indicated sample type. To illustrate the null distribution, the linear regression line of NTC-NTC control arrays is overlain. Shading on the regression line denotes the 95% confidence interval. d, Scatter plot of MCAP-MET single knockout (SKO; n = 26 genes) and double knockout (DKO; n = 325 gene pairs) abundances in cell pools (n = 6 cell replicates) versus lung metastases (n = 37 from ten mice). Data shown in terms of average log2 RPM for the indicated sample type, after first averaging the constituent crRNA arrays for each gene/gene pair. The linear regression was calculated over the entire library, with the 95% confidence interval shaded in. Significant outliers (two-sided outlier test, adjusted P < 0.05) are outlined and enlarged, with s.e.m. error bars.

Supplementary Figure 9 Additional analyses of synergistic metastasis drivers.

ae, Tukey box plots (IQR boxes with 1.5 × IQR whiskers and notched 95% confidence interval of median) detailing the log2 RPM abundances of the indicated genotypes in lung metastases (n = 37 from ten mice), with associated two-sided Wilcoxon rank sum P values and SynCo scores noted. a, Chd1_Kmt2d. b, Chd1_Nf2. c, Jak1_Kmt2c. d, Nf1_Pten. e, Kmt2d_Pten. Statistics are in reference to the DKO genotype (purple) and colored according to the corresponding SKO genotype (green and orange).

Supplementary Figure 10 Relative selective advantages of gene pair versus single gene knockouts.

Heat map of the change in log2 RPM abundance in lung metastases (n = 37 from ten mice) for each single gene knockout, relative to each modulatory second knockout. A positive value (red) means that the second knockout (rows) granted a relative selective advantage over the reference knockout (columns), whereas a negative value (blue) means the second knockout was relatively disadvantageous compared with the single knockout.

Supplementary Figure 11 Mutation signatures of human primary tumors compared with metastases.

a, Bar plot of the mutation frequencies associated with the 26 genes represented in the MCAP-MET library, in the TCGA PanCan primary tumor dataset (left) and the MET500 metastasis dataset (right). b, Scatter plot comparing the mutation frequencies in a (n = 26 genes). Spearman R = 0.799, P = 9.887 × 10–7. c,d, Venn diagrams of JAK1 and KMT2C (c) or CHD1 and KMT2D (d) mutations in the MET500 (n = 486 tumors) and TCGA PanCan cohorts (n = 9,870 tumors). Statistical significance of the overlap was assessed by hypergeometric test.

Supplementary Figure 12 Characterization of the Nf2_Trim72 gene pair in vitro.

a, Representative T7E1 assays (n = 5 independent infection replicates) to quantify mutation efficiency of the top-performing Nf2 crRNA (left) and Trim72 crRNA (right), in cells expressing five different crRNA arrays (annotated below). b, Representative flow cytometry analysis of KPD cells after 2 h in culture with 10 μM EdU, stained by EdU-APC (n = 3 infection replicates for each condition). c, Quantification of EdU incorporation (mean ± s.e.m.) of KPD cells transduced with Rosa26 + Rosa26, Nf2 + Rosa26, Trim72 + Rosa26 or Nf2 + Trim72 (n = 3 infection replicates for each condition). Nf2 + Trim72 versus Nf2 + Rosa26, P = 0.3068; versus Trim72 + Rosa26, P = 0.5450; versus Rosa26 + Rosa26, P = 0.6484. Statistical significance was assessed by two-sided unpaired Welch’s t-test. d, Representative images of invasive cells in a Matrigel invasion assay at 24 h (n = 6 assay replicates per condition, aggregated from three independent experiments). Scale bar, 100 μm. e, Quantification of the number of invasive cells in Matrigel invasion assays at 24 h (n = 6 assay replicates per condition, aggregated from three independent experiments; mean ± s.e.m.). Statistical significance was assessed by two-sided Wilcoxon rank sum test. Nf2 + Trim72 versus Nf2 + Rosa26, P = 0.0281; versus Trim72 + Rosa26, P = 0.0281; versus Rosa26 + Rosa26, P = 0.0043. Nf2 + Rosa26 versus Rosa26 + Rosa26, P = 0.0195; versus Trim72 + Rosa26, P = 0.6147. Trim72 + Rosa26 versus Rosa26 + Rosa26, P = 0.0455. *P < 0.05, **P < 0.01, ***P < 0.001.

Supplementary Figure 13 Additional characterization of the Nf2_Trim72 gene pair in vivo.

a, Examples of luciferase imaging at 23 days post infection in mice bearing Rosa26 + Rosa26, Nf2 + Rosa26, Trim72 + Rosa26 or Nf2 + Trim72 tumors (n = 8 mice per condition). b, Representative bright-field images of primary tumors derived from KPD cells transduced with Rosa26 + Rosa26, Nf2 + Rosa26, Trim72 + Rosa26 or Nf2 + Trim72 crRNA arrays, at 28 days post infection (n = 8 mice per condition). Scale bar, 2 mm. c, Representative bright-field images of lungs from mice bearing Rosa26 + Rosa26, Nf2 + Rosa26, Trim72 + Rosa26 or Nf2 + Trim72 primary tumors at 28 days post infection (n = 8 mice for each condition). Arrowheads indicate metastatic lung nodules. Scale bar, 2 mm.

Supplementary information

Supplementary Information

Supplementary Figures 1–13

Reporting Summary

Supplementary Protocol

In vivo combinatorial knockout screens using CRISPR–Cpf1.

Supplementary Table 1

Genes targeted in the MCAP-MET library.

Supplementary Table 2

Oligos and spacer sequences in the MCAP-MET library.

Supplementary Table 3

Normalized crRNA array-level abundances in plasmid, cell pools, primary tumors, and lung metastases.

Supplementary Table 4

Synergy analysis in lung metastases.

Supplementary Table 5

Genetic interaction matrix.

Supplementary Table 6

Mutation frequencies of MCAP-MET genes in TCGA Pancan and MET500 cohorts.

Source Data, Figure 3

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Chow, R.D., Wang, G., Ye, L. et al. In vivo profiling of metastatic double knockouts through CRISPR–Cpf1 screens. Nat Methods 16, 405–408 (2019). https://doi.org/10.1038/s41592-019-0371-5

Download citation

Further reading

Search

Quick links

Nature Briefing

Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.

Get the most important science stories of the day, free in your inbox. Sign up for Nature Briefing