Abstract
All cancers emerge after a period of clonal selection and subsequent clonal expansion. Although the evolutionary principles imparted by genetic intratumour heterogeneity are becoming increasingly clear1, little is known about the non-genetic mechanisms that contribute to intratumour heterogeneity and malignant clonal fitness2. Here, using single-cell profiling and lineage tracing (SPLINTR)—an expressed barcoding strategy—we trace isogenic clones in three clinically relevant mouse models of acute myeloid leukaemia. We find that malignant clonal dominance is a cell-intrinsic and heritable property that is facilitated by the repression of antigen presentation and increased expression of the secretory leukocyte peptidase inhibitor gene (Slpi), which we genetically validate as a regulator of acute myeloid leukaemia. Increased transcriptional heterogeneity is a feature that enables clonal fitness in diverse tissues and immune microenvironments and in the context of clonal competition between genetically distinct clones. Similar to haematopoietic stem cells3, leukaemia stem cells (LSCs) display heritable clone-intrinsic properties of high, and low clonal output that contribute to the overall tumour mass. We demonstrate that LSC clonal output dictates sensitivity to chemotherapy and, although high- and low-output clones adapt differently to therapeutic pressure, they coordinately emerge from minimal residual disease with increased expression of the LSC program. Together, these data provide fundamental insights into the non-genetic transcriptional processes that underpin malignant clonal fitness and may inform future therapeutic strategies.
This is a preview of subscription content, access via your institution
Access options
Access Nature and 54 other Nature Portfolio journals
Get Nature+, our best-value online-access subscription
$29.99 / 30 days
cancel any time
Subscribe to this journal
Receive 51 print issues and online access
$199.00 per year
only $3.90 per issue
Buy this article
- Purchase on Springer Link
- Instant access to full article PDF
Prices may be subject to local taxes which are calculated during checkout
Similar content being viewed by others
Data availability
All high-throughput sequencing data supporting the findings of this study have been deposited in the NCBI Gene Expression Omnibus under primary accession code GSE161676.
Code availability
The code that support the findings of this study are available online (https://atlassian.petermac.org.au/bitbucket/scm/daw/splintr_paper_code and https://atlassian.petermac.org.au/bitbucket/scm/daw/splintr_barcode_seq_pipeline).
References
Turajlic, S., Sottoriva, A., Graham, T. & Swanton, C. Resolving genetic heterogeneity in cancer. Nat. Rev. Genet. 20, 404–416 (2019).
Marine, J. C., Dawson, S. J. & Dawson, M. A. Non-genetic mechanisms of therapeutic resistance in cancer. Nat. Rev. Cancer 20, 743–756 (2020).
Rodriguez-Fraticelli, A. E. et al. Single-cell lineage tracing unveils a role for TCF15 in haematopoiesis. Nature 583, 585–589 (2020).
Lin, D. S. et al. DiSNE movie visualization and assessment of clonal kinetics reveal multiple trajectories of dendritic cell development. Cell Rep. 22, 2557–2566 (2018).
Greaves, M. When one mutation is all it takes. Cancer Cell 27, 433–434 (2015).
Visvader, J. E. Cells of origin in cancer. Nature 469, 314–322 (2011).
Krivtsov, A. V. et al. Cell of origin determines clinically relevant subtypes of MLL-rearranged AML. Leukemia 27, 852–860 (2013).
Pendleton, M., Lindsey, R. H. Jr, Felix, C. A., Grimwade, D. & Osheroff, N. Topoisomerase II and leukemia. Ann. N. Y. Acad. Sci. 1310, 98–110 (2014).
Somervaille, T. C. & Cleary, M. L. Identification and characterization of leukemia stem cells in murine MLL-AF9 acute myeloid leukemia. Cancer Cell 10, 257–268 (2006).
Krivtsov, A. V. et al. Transformation from committed progenitor to leukaemia stem cell initiated by MLL-AF9. Nature 442, 818–822 (2006).
Dahlin, J. S. et al. A single-cell hematopoietic landscape resolves 8 lineage trajectories and defects in Kit mutant mice. Blood 131, e1–e11 (2018).
Matsuo, H. et al. Fusion partner-specific mutation profiles and KRAS mutations as adverse prognostic factors in MLL-rearranged AML. Blood Adv. 4, 4623–4631 (2020).
The Cancer Genome Atlas Research Network. Genomic and epigenomic landscapes of adult de novo acute myeloid leukemia. N. Engl. J. Med. 368, 2059–2074 (2013).
Shaffer, S. M. et al. Rare cell variability and drug-induced reprogramming as a mode of cancer drug resistance. Nature 546, 431–435 (2017).
Raj, A. & van Oudenaarden, A. Nature, nurture, or chance: stochastic gene expression and its consequences. Cell 135, 216–226 (2008).
Larsson, A. J. M. et al. Genomic encoding of transcriptional burst kinetics. Nature 565, 251–254 (2019).
MacPherson, L. et al. HBO1 is required for the maintenance of leukaemia stem cells. Nature 577, 266–270 (2020).
Coombs, C. C., Tavakkoli, M. & Tallman, M. S. Acute promyelocytic leukemia: where did we start, where are we now, and the future. Blood Cancer J. 5, e304 (2015).
Jones, C. L., Inguva, A. & Jordan, C. T. Targeting energy metabolism in cancer stem cells: progress and challenges in leukemia and solid tumors. Cell Stem Cell 28, 378–393 (2021).
Zuber, J. et al. Mouse models of human AML accurately predict chemotherapy response. Genes Dev. 23, 877–889 (2009).
Duy, C. et al. Chemotherapy induces senescence-like resilient cells capable of initiating AML recurrence. Cancer Discov. 11, 1542–1561 (2021).
Bell, C. C. et al. Targeting enhancer switching overcomes non-genetic drug resistance in acute myeloid leukaemia. Nat. Commun. 10, 2723 (2019).
Rambow, F. et al. Toward minimal residual disease-directed therapy in melanoma. Cell 174, 843–855 (2018).
Fong, C. Y. et al. BET inhibitor resistance emerges from leukaemia stem cells. Nature 525, 538–542 (2015).
Shlush, L. I. et al. Tracing the origins of relapse in acute myeloid leukaemia to stem cells. Nature 547, 104–108 (2017).
Zorita, E., Cusco, P. & Filion, G. J. Starcode: sequence clustering based on all-pairs search. Bioinformatics 31, 1913–1919 (2015).
Silver, D. P. & Livingston, D. M. Self-excising retroviral vectors encoding the Cre recombinase overcome Cre-mediated cellular toxicity. Mol. Cell 8, 233–243 (2001).
Kebschull, J. M. & Zador, A. M. Cellular barcoding: lineage tracing, screening and beyond. Nat. Methods 15, 871–879 (2018).
Naik, S. H., Schumacher, T. N. & Perie, L. Cellular barcoding: a technical appraisal. Exp. Hematol. 42, 598–608 (2014).
Ye, M. et al. Hematopoietic differentiation is required for initiation of acute myeloid leukemia. Cell Stem Cell 17, 611–623 (2015).
Naik, S. H. et al. Diverse and heritable lineage imprinting of early haematopoietic progenitors. Nature 496, 229–232 (2013).
Langmead, B. Aligning short sequencing reads with Bowtie. Curr. Protoc. Bioinform. 32, 11.7.1–11.7.14 (2010).
Li, H. et al. The Sequence Alignment/Map format and SAMtools. Bioinformatics 25, 2078–2079 (2009).
Scheinin, I. et al. DNA copy number analysis of fresh and formalin-fixed specimens by shallow whole-genome sequencing with identification and exclusion of problematic regions in the genome assembly. Genome Res. 24, 2022–2032 (2014).
McLaren, W. et al. The Ensembl variant effect predictor. Genome Biol. 17, 122 (2016).
Stuart, T. et al. Comprehensive integration of single-cell data. Cell 177, 1888–1902 (2019).
Butler, A., Hoffman, P., Smibert, P., Papalexi, E. & Satija, R. Integrating single-cell transcriptomic data across different conditions, technologies, and species. Nat. Biotechnol. 36, 411–420 (2018).
Stoeckius, M. et al. Cell hashing with barcoded antibodies enables multiplexing and doublet detection for single cell genomics. Genome Biol. 19, 224 (2018).
Hafemeister, C. & Satija, R. Normalization and variance stabilization of single-cell RNA-seq data using regularized negative binomial regression. Genome Biol. 20, 296 (2019).
Becht, E. et al. Dimensionality reduction for visualizing single-cell data using UMAP. Nat. Biotechnol. 37, 38–44 (2018).
Zappia, L. & Oshlack, A. Clustering trees: a visualization for evaluating clusterings at multiple resolutions. Gigascience 7, giy083 (2018).
Finak, G. et al. MAST: a flexible statistical framework for assessing transcriptional changes and characterizing heterogeneity in single-cell RNA sequencing data. Genome Biol. 16, 278, (2015).
Michailidou, K. et al. Association analysis identifies 65 new breast cancer risk loci. Nature 551, 92–94, (2017).
Marjanovic, N. D. et al. emergence of a high-plasticity cell state during lung cancer evolution. Cancer Cell 38, 229–246 (2020).
Azizi, E. et al. Single-cell map of diverse immune phenotypes in the breast tumor microenvironment. Cell 174, 1293–1308 (2018).
Acknowledgements
We thank members of the Dawson laboratory, particularly S. Jackson, O. Gilan and K. Knezevic for technical assistance with experiments; F. Rossello for advice and discussions related to data analysis and D. Zalcenstein and J. Gao for advice regarding lentiviral barcode design and cloning. We thank the following funders for fellowship and grant support: Peter MacCallum Foundation Grant (to D.V.), VCA mid-career fellowship (to E.Y.N.L.), CSIRO Future Science Fellowship in Synthetic Biology (to T.S.W.), NHMRC Early Career Fellowship grant (no. 1052195, to D.C.M.) NHMRC Investigator Grant (no. 1196749, to M.A.D.), Cancer Council Victoria Dunlop Fellowship (to M.A.D), Howard Hughes Medical Institute international research scholarship (no. 55008729, to M.A.D), NHMRC Investigator Grant (no. 1196755, to S.J.D.), CSL Centenary fellowship (to S.J.D.) and NHMRC Project Grant (no. 1128984, to S.J.D. and M.A.D.).
Author information
Authors and Affiliations
Contributions
K.A.F., D.V., S.H.N. and M.A.D. designed/analysed experiments. K.A.F., D.V. and M.A.D. wrote the manuscript with contributions from all of the authors. K.A.F. developed SPLINTR with assistance from D.C.M., E.Y.N.L. and D.V. with joint supervision by S.H.N. and M.A.D. All experimental work was performed by K.A.F. and D.V. with assistance from J.J.B., T.S., L.M., A.G. and Y.-C.C. under the supervision of M.A.D. All computational analyses were performed by D.V. with assistance from E.Y.N.L., S.H., Q.W. and T.S.W. under the supervision of E.S.W., S.H.N. and M.A.D. Critical reagents and resources were provided by L.M.K., S.H.N., S.-J.D. and M.A.D.
Corresponding author
Ethics declarations
Competing interests
M.A.D. has been a member of advisory boards for CTX CRC, Storm Therapeutics, Celgene and Cambridge Epigenetix. S.-J.D. has been a member of advisory boards for AstraZeneca. S.-J.D. has received research funding from Genentech. M.A.D. and S.-J.D. receive research funding from CTx CRC. The other authors declare no competing interests.
Additional information
Peer review information Nature thanks Ross Levine and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. Peer reviewer reports are available.
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Extended data figures and tables
Extended Data Fig. 1 Generation and proof of concept of SPLINTR barcode libraries.
a, Three independent SPLINTR barcode libraries were generated, each containing a distinct fluorescent reporter and barcode structure (Methods). SPLINTR barcodes are constitutively transcribed upon genome integration, enabling tracking of clonally related cells together with their individual transcriptomes. b, Base-position analysis of SPLINTR barcode structures. The designated combinations of strong (G or C) and weak (A or T) bases results in barcode structures that can be computationally deconvoluted from other SPLINTR libraries. c, Venn diagrams showing number of unique barcodes in each library identified via overlap of two deep-sequenced PCR technical replicates per library. Boxplots showing distributions of Hamming (d) and Levenshtein edit (e) distances between individual barcodes calculated for each individual library. 1000 barcodes present in each reference library were randomly sampled without replacement (n = 100 resampling events per barcode library, GFP library, n = 170,885 barcodes, BFP library, n = 672,982 barcodes and mCherry library, n = 1,324,188 barcodes) and the average hamming/edit distance between all pairwise combinations were computed. Boxplots span the upper quartile (upper limit), median (centre) and lower quartile (lower limit). Whiskers extend a maximum of 1.5x IQR. f, Schematic of a pilot in vitro SPLINTR barcoding experiment. g, Proportional bubbleplot showing barcode distributions derived from barcode-seq from Pool #1 and #2 and scRNA-seq from Pool #1. Bubble size scales with clone size. h, Pearson correlation matrix between normalized barcode repertoires from Pool #1 and #2 barcode-seq and Pool #1 scRNA-seq. Correlation R values are shown. i, UMAP projection of Pool #1 scRNA-seq dataset. Louvain clusters are indicated. Cells containing a detected SPLINTR barcode are highlighted red. Cells without a barcode are shown in grey. j, The number of distinct SPLINTR barcodes detected per cell in the Pool #1 scRNA-seq dataset. Doublets were classified as cells containing a unique combination of two or more SPLINTR barcodes and are shown in red. Singlet cells are in blue. k, Total UMI counts per cell for cells grouped according to their predicted doublet status as described in (j) (doublet = 469 cells, singlet = 7494 cells, unknown = 2388 cells). Unknown indicates cells in which no SPLINTR barcode was detected. Boxplots span the upper quartile (upper limit), median (centre) and lower quartile (lower limit). Whiskers extend a maximum of 1.5x IQR.
Extended Data Fig. 2 Clonal dominance is an intrinsic feature of leukaemia initiating clones, regardless of cell-of-origin.
Results are representative of n = 4 mice per experimental group. a, Schematic of an in vivo barcoding experiment investigating the clonal competition and dynamics in MLL-AF9 AML derived from an HSC (Lin-, c-Kit +, Sca-1+) or GMP (Lin-, c-Kit +, CD34+, FcγRHI) cell-of-origin. For the FACS gating strategy, see Supplementary Information Figure 1. b, Disease burden based on CD45.2 donor cell expression in the bone marrow and spleen of Ptprca (top) or NSG (bottom) mice transplanted with either HSC or GMP derived leukaemia. All mice were collected at the same timepoint. Bars represent the median. c, Histogram detailing the total number of GMP (red) or HSC- derived leukaemic clones (green) comprising the 95th percentile from bulk barcode-seq at Baseline and Disease in Ptprca (top) or NSG (bottom) mice transplanted with a 50:50 mix of either HSC or GMP-derived leukaemias. d, Violin plots showing the clone size expressed as log10 transformed counts per million (CPM) of HSC or GMP-derived clones collected from NSG (top, HSC: n = 15 clones, GMP: n = 14 clones) or Ptprca (bottom, HSC: n = 13 clones, GMP: n = 10 clones) mice that received 50:50 HSC/GMP competition transplants. Boxplots span the upper quartile (upper limit), median (centre) and lower quartile (lower limit). Whiskers extend a maximum of 1.5x IQR. Two-sided T-test. NSG p = 0.55, Ptprca p = 0.13. e, Heat maps of pairwise Pearson correlation values for HSC derived clones (top) and GMP derived clones (bottom) from Barcode-seq samples from mice transplanted with single MLL-AF9 HSC or GMP-derived leukaemias or a 50:50 HSC/GMP cell mix. f, Bubble plot showing distribution of HSC and GMP derived leukaemic clone sizes in barcode-seq data from Baseline (T0) and disease timepoints from mice that received transplants of 50:50 HSC/GMP cell mix. Clones ordered based on their abundance at the baseline timepoint (Ranked abundance at baseline). Dominant clones (those above 2% frequency) from GMP or HSC groups are coloured, all other clones are shown in grey. HSC-derived leukaemias were barcoded with the GFP SPLINTR library. GMP-derived leukaemias were barcoded with the mCherry SPLINTR library.
Extended Data Fig. 3 SPLINTR identifies both LSC and non-LSC like transcriptional signatures in disease causing clones.
a, Schematic of an in vivo experiment to identify the transcriptional features of dominant murine MLL-AF9 + KrasG12D leukaemic clones that drive disease. b, Correlation plot between the frequencies of SPLINTR barcoded bone marrow (left) and spleen (right) clones at the Baseline and Disease timepoints (n = 4 mice). Inset values indicate adjusted R2 and p-values using the F-test for linear regression on frequencies of barcodes detected at both timepoints. Clone barcode frequencies were quantified by barcode-seq. Dashed line indicates a threshold frequency (2%) above which a clone is considered dominant. (n = 4 mice) c, UMAP projection of scRNA-seq data from SPLINTR barcoded MLL-AF9 + KrasG12D cells taken at the baseline timepoint prior to transplantation. Top row, left to right: Louvain clusters, cells comprising dominant clones, cells coloured according to module enrichment scores for genes upregulated in leukaemic stem cells (LSC Signature - UP). Bottom row, left to right: Cells coloured according to module enrichment scores for genes upregulated in normal long-term haematopoetic stem cells (LT-HSCs), common myeloid progenitors (CMPs) and granulocyte/macrophage progenitors (preGM/GMPs). q10 = 10th quantile, q90 = 90th quantile. Module enrichment scores for genes upregulated in LSCs (d), normal LT-HSC (e) or GMPs (f) or CMPs (g) for each dominant clone vs. all other cells (cells per group left to right: n = 24, 34, 29, 147, 115, 32, 13115). Asterisks indicate the result of a two-sided T-test using Holm multiple testing correction for each clone vs. all other cells. Number of cells shown above each category. Adjusted p-values (left to right) for (d): 6e−07, 3.2e−07, 0.33, 0.25, 3e−34, 0.92. (e): 1, 1, 1, 0.00012, 2.2e−14, 1. (f): 2.6e−07, 9.1e−07,1, 1, 1.2e−45, 1. (g): 1.5e−10, 9.1e−11, 0.76, 6.9e−06, 3.7e−55, 0.76. **** = p-adj < 0.001. Boxplots for (d) to (g) span the upper quartile (upper limit), median (centre) and lower quartile (lower limit). Whiskers extend a maximum of 1.5x IQR. h, Integration of scRNA-seq data sets from MLL-AF9 primary leukaemia cells from (c) and healthy progenitor HSPC populations from Dahlin et al. 2018. Top-left panel: UMAP projection showing Louvain clusters. Top-middle panel: Cells labelled according to their sample of origin. Primary MLL-AF9 baseline timepoint cells (MLL-AF9), Dahlin et al. lineage negative, c-Kit positive cells (LK), and Dahlin et al. lineage negative, Sca-1 positive and c-Kit positive cells (LSK). Top-right: UMAP projection with dominant primary MLL-AF9 clones highlighted. Bottom row: UMAP projections showing gene set module enrichment scores for genes upregulated in Long Term Haematopoietic Stem Cells (LT-HSC) (left), Common Myeloid Progenitors (CMPs) (middle) and Granulocyte/Macrophage progenitors (preGM/GMPs) (right). q10 = 10th quantile, q90 = 90th quantile. i, Hypergeometric test for Louvain clusters enriched for dominant MLL-AF9 clones. Negative log10 transformed P values per cluster are shown. Red clusters are significantly enriched for dominant clones. Dashed line indicates a p = 0.05 significance threshold. j, The percentage of the 3 different cell types comprising each of the Louvain clusters in the integrated dataset. Clusters enriched for dominant MLL-AF9 clones are shown in red.
Extended Data Fig. 4 Generation and genetic characterization of secondary and tertiary leukaemic cells.
a, Schematic of a SPLINTR experiment using 2° MLL-AF9 cells. Timepoints where whole exome sequencing (WES) and low-coverage whole genome sequencing (LC-WGS) analysis was conducted are indicated. b, Results of whole exome sequencing (WES) of SPLINTR barcoded tertiary MLL-AF9 leukaemic cells at Endpoint (TN), expressed as mutations per Megabase (1 million base pairs). Points indicate raw result from individual mice (n = 2). Lines indicate the mean. Full lists of variants identified can be found in Extended Data Table 3. c, Copy number variant profiles from LC-WGS data of SPLINTR barcoded tertiary MLL-AF9 leukaemic cells from bone marrow at Endpoint (top) and a sex and age matched healthy wild type C57BL/6 littermate control (bottom). d, UMAP projection of reanalysed MLL-AF9 + KrasG12D Baseline scRNA-seq data from Fig. 2. Cells comprising Shared dominant clones are highlighted (top) and normalized expression level of transcripts derived from the MLL-AF9 retroviral construct (bottom).
Extended Data Fig. 5 Pre-existing transcriptional determinants of clonal dominance in MLL-AF9 leukaemia.
a, Heat maps of pairwise Pearson correlation values for Barcode-seq samples from MLL-AF9, MLL-AF9 + KrasG12D and MLL-AF9 + Flt3ITD groups (n = 4-5 mice). b, UMAP projections of Baseline (T0) scRNA-seq datasets from MLL-AF9 (left column), MLL-AF9 + KrasG12D (middle column) and MLL-AF9 + Flt3ITD (right column) groups as in Fig. 1. Rows show results of Louvain clustering (top), module enrichment scores for genes upregulated in LSCs (middle) or the MHC Class I signature (bottom). Heat maps showing the top differentially expressed genes in dominant clones vs all other cells at the Baseline timepoint for the MLL-AF9 (c) and MLL-AF9 + Flt3ITD (d) groups. A random sample of all other cells within the same genotype is shown on each heat map. Biological pathways upregulated (e) and downregulated (f) in dominant clones for each individual genotype and the Competition 80:15:5 group at Endpoint (TN). Number of genes in each group are indicated at the top. g, UMAP projections of scRNA-seq data from Baseline and Disease stage MLL-AF9 + KrasG12D cells showing expression patterns of B2m, Myb and Slpi. q10 = 10th quantile, q90 = 90th quantile. Disease scRNA-seq data integrated from two independent mice. h, Boxplots of normalized Slpi expression (log1p transformed) in cells comprising dominant clones vs. all other cells for the MLL-AF9 (cells per group left to right: n = 17, 21, 41, 3, 7, 2, 2, 9088), MLL-AF9 + KrasG12D (cells per group left to right: n = 14, 18, 29, 443, 63, 386, 9, 27, 36, 120, 136, 1, 45, 17, 80, 23, 162, 50, 1, 48, 13, 41, 10304) and MLL-AF9 + Flt3ITD (cells per group left to right: n = 3257, 664, 1066, 501, 110, 26, 306, 34, 24, 11, 32) groups. One-sided Wilcoxon rank sum test with Benjamini-Hochberg multiple test correction for each clone vs. all other cells (Other) within each group. Adjusted p-values for MLL-AF9 left to right: 0.022, 2e−10, 1, 0.022, 0.022, 0.022, 1. Adjusted p-values for MLL-AF9 + KrasG12D left to right: 0.85, 0.00023, 6.6e−09, 0.13, 7.5e−14, 1, 0.12, 1.7e−10, 2.9e−10, 1.1e−36, 8.8e−11, 0.66, 4.3e−05, 0.51, 8.8e−06, 1.5e−09, 2.2e−31, 6.2e−13, 0.13, 5e−13, 0.03, 2.8e−13. Adjusted p-values for MLL-AF9 + Flt3ITD left to right: 0.24, 1, 1, 1, 1, 0.47, 7.1e−17, 1, 0.066, 0.66, 1, 1, 0.0089, 1, 1, 1, 1, 0.46, 0.46, 1.6e−07, 0.17. * = p-adj < 0.05, ** = p-adj < 0.01, *** = p-adj < 0.001, **** = p-adj < 0.0001. i, UMAP projections of Disease stage MLL-AF9 (left column), MLL-AF9 + KrasG12D (middle column) and MLL-AF9 + Flt3ITD (right column) scRNA-seq datasets. Rows show B2m expression (top) and module enrichment scores for genes upregulated (middle) or downregulated (bottom) in LSCs in each dataset. Boxplots span the upper quartile (upper limit), median (centre) and lower quartile (lower limit). Whiskers extend a maximum of 1.5x IQR.
Extended Data Fig. 6 Results of in vivo modelling of clonal competition.
Proportional bubble plot of Barcode-seq data showing dominant clones in the Competition 1:1:1 (a) and Competition 80:15:5 (b) groups at Baseline and Disease timepoints. Red, green and blue regions show MLL-AF9 + Flt3ITD, MLL-AF9 and MLL-AF9 + KrasG12D derived clones respectively (n = 4 mice per group) c, Correlation plots of clonal repertoires of MLL-AF9 + KrasG12D clones from baseline and disease samples in the Competition 1:1:1 (left) or Competition 80:15:5 (right) groups (n = 4 mice per group) d, Venn diagram indicating the number of shared and distinct MLL-AF9 + KrasG12D clones at Disease between Kras only, Competition 1:1:1 and Competition 80:15:5 groups. e, Clone size at Baseline (T0) expressed as log10 transformed counts per million (CPM) of Competition 80:15:5 specific clones (n = 22 clones) vs Kras specific clones (n = 22 clones). Boxplots span the upper quartile (upper limit), median (centre) and lower quartile (lower limit). Whiskers extend a maximum of 1.5x IQR. One-sided T-test (*, p = 0.018). f, Representative flow cytometry analysis of the disease burden in bone marrow and spleen 2 weeks after equal numbers of cells from each genotype were injected into separate mice. Plots are representative of 1 mouse per group. g, Proportional bubble plot of normalized Barcode-seq data showing clonal repertoire of Kras clones from Baseline samples in the Competition 1:1:1, Competition 80:15:5 or MLL-AF9 + KrasG12D only groups. h, Correlation plots of the barcode repertoires present in Baseline cells sampled from the MLL-AF9 + KrasG12D only transplants and the Competition 1:1:1 and Competition 80:15:5 transplants. Inset values indicate adjusted R2 and p-values using a one-tailed F-test for linear regression on CPM of barcodes detected at both timepoints. Results in this figure are representative of n = 4 mice per experimental group.
Extended Data Fig. 7 Competition-specific clones display altered transcriptional burst kinetics compared to KRAS-specific clones.
a, Normalized Slpi (top) and B2m (bottom) expression in MLL-AF9 + KrasG12D Baseline scRNA-seq cells comprising Kras-specific clones (red, n = 617 cells) Competition-specific clones (blue, n = 926 cells) Baseline-specific clones (green, n = 965 cells) or all other clones (grey, n = 12,049 cells). Boxplots indicate median and inter-quartile range (IQR). Whiskers extend a maximum of 1.5x IQR. Two-sided T-test using Holm multiple testing correction for each group vs. all other clones (Other clones). Kras-specific (Slpi: ****, p-adj = 2e−134; B2m: ****, p-adj = 6.2e−36), Competition-specific (Slpi: ****, p-adj = 6.3e−32; B2m: ****, p-adj = 7.9e−36), Baseline-specific (Slpi: ****, p-adj = 5.8e−13; B2m: p-adj = 0.91). b, Histogram showing the percentage of cells belonging to Kras-specific, Competition-specific or Baseline-specific clones present in each Louvain cluster for the MLL-AF9 + KrasG12D Baseline scRNA-seq dataset. c, Permutation analysis of Shannon diversity scores derived from the Louvain cluster occupancy of Kras-specific (n = 617 cells) and Competition-specific (n = 926 cells) groups. Centre point indicates raw Shannon diversity values with error bars reflecting the standard error of 100 resampling events with replacement. Two-sided Mann Whitney U test. *, p = 0.006847. d, Scatter plots of log10 transformed burst frequency and burst size values for 5188 genes with inferred transcriptional kinetics based on MLL-AF9 + KrasG12D Baseline scRNA-seq data from Kras-specific (n = 617 cells) and Competition-specific cells (n = 926 cells). Red points indicate genes with a statistically significant difference (Benjamini-Hochberg adjusted p-value < 0.01) in Competition-specific cells vs Kras-specific cells. e, Venn diagram showing overlap of genes with significantly increased or decreased burst size or burst frequency in the Competition-specific group versus the Kras-specific group. Biological process gene ontologies for genes with a significantly increased transcriptional burst frequency (f) or burst size (g) in Competition-specific cells vs Kras-specific cells. Selected gene sets with Benjamini-Hochberg adjusted p-value < 0.05 are shown.
Extended Data Fig. 8 Dominant clones possess a distinct chromatin accessibility landscape.
a, Number of clones comprising the 95th percentile at Baseline (T0) and Endpoint (TN) bone marrow and spleen samples in NSG and Ptprca mice that received transplants from the same pool of barcoded MLL-AF9 + KrasG12D cells. b, Proportional bubble plots of clones identified via Barcode-seq at Baseline and Disease timepoints for each mouse in each immune background. For each sample, clones are sorted in descending rank order according to their percentage abundance in the Baseline sample (n = 3 mice per group) c, Permutation analysis of Shannon diversity scores derived from the Louvain cluster occupancy of Ptprca only (n = 398 cells) or Ptprca and NSG specific (Shared, n = 150 cells) dominant clones from Fig. 2. Center point indicates raw Shannon diversity values with error bars reflecting the standard error of 1000 resampling events with replacement. Two-sided Mann Whitney U test. *, p = 1e−11. d, Gene expression heat map showing the top differentially expressed genes in dominant clones from the NSG-only, Ptprca-only and shared groups vs all other cells at the Baseline timepoint. A random sample of all other cells is shown on the heat map. Genes of interest are highlighted. e, Biological pathways enriched in the set of genes upregulated in NSG-specific clones (NSG_up) or downregulated in NSG (NSG_down) and Ptprca specific clones (BL6_down). Number of genes in each group are indicated above. f, UMAP projections of scRNA-seq dataset from Baseline MLL-AF9 + KrasG12D cells from Fig. 3 showing expression of Cebpα, Cebpβ and Cebpδ. g, UMAP projection of scATAC-seq dataset from Baseline MLL-AF9 + KrasG12D cells from Fig. 3 showing results of Louvain clustering and label transfer to the corresponding scRNA-seq dataset as in Fig. 3. h, Results of differential accessibility analysis of cells in scATAC-seq clusters 1 and 7 (Dominant) vs cells in all other clusters (Other) at three exemplar loci B2m (top), Kit (middle) Cebpε (bottom). Boxes indicate regions of significantly altered accessibility in the Dominant group. i, The top six enriched transcription factor motifs in regions significantly differentially accessibile in cells comprising scATAC-seq clusters 1 and 7 (clusters enriched for dominant clones) vs cells in all other clusters. j, Results of Cebpε, Cebpα and isotype control (IgG/Neg) Chromatin Immuno-precipitation coupled qPCR using primers against the promoter of Slpi and a negative-control region (Gene desert) in MLL-AF9 + KrasG12D cells. Mean ± s.d. of n = 3 independent experiments are shown. k, Western blot analysis of Baseline Slpi protein levels in MLL-AF9 + KrasG12D cells transduced with guides targeting Slpi or non-targeting guide control immediately prior to the in vitro and in vivo competition assays in Fig. 2. (representative of n = 2 independent experiments). For uncropped blot, see Supplementary Information Figure 2. l, Flow cytometry analysis of bone marrow cells collected from 50:50 group mice. Representative results from n = 4 mice.
Extended Data Fig. 9 Serial transplantation reveals a heritable transcriptional program of clonal dominance.
a, Bubble plots of clones identified via Barcode-seq for Donor mouse #1 and corresponding recipient NSG or Ptprca mice at secondary transplant Baseline and Disease timepoints. Bubbleplots indicate primary transplant clones (left) or secondary transplant clones (middle and right). For each sample, clones are sorted in descending rank order according to their percentage abundance in the Baseline sample. n = 2 mice per group. b, UMAP projection of scRNA-seq datasets from bone marrow (top) and spleen (bottom) derived clones at Baseline prior to secondary transplantation. Cells comprising the top three most dominant clones from the primary transplantation are highlighted. c, Network plots of enriched Gene Ontology Biological Process terms for genes upregulated (left) and downregulated (right) in secondary transplant dominant clones. d, Gene expression heat map showing the top differentially expressed genes in cells comprising dominant clones from the secondary transplantation (red) vs all other cells (grey). A random sample of all other cells is shown on the heat map. Genes positively or negatively correlated with clonal dominance are highlighted red or blue respectively. e, UMAP projection of individual scRNA-seq datasets from bone marrow (top) and spleen (bottom) derived clones at Baseline prior to secondary transplantation. Louvain clusters, cells comprising individual dominant clones in the secondary transplantation and expression patterns of Slpi, Cebpε, B2m and the LSC signature are shown from left to right. q10 = 10th quantile, q90 = 90th quantile.
Extended Data Fig. 10 Distinct responses of high and low output AML clones to chemotherapy treatment.
a, Abundance (proportion) of each dominant clone from the primary transplantation shown across the Baseline #2 and Disease #2 timepoints of the secondary transplantation. Clone abundance in the bone marrow (top) and spleen (bottom) are shown. Difference to mean indicates the difference between mean frequencies of each clone in the primary versus secondary transplants. b, Proportions of secondary transplant clones for Donor mouse #1 and recipient Ptprca and NSG mice. High output clones contribute >5% of the total disease burden whereas Mid output clones contribute 1-5%. Low output clones contribute below 1% of the total disease burden. Points represent individual clones and are coloured according to their primary transplant clone identity. Low output clones are not coloured. c, UMAP projections of the integrated scRNA-seq dataset shown in Fig. 4 showing c-Myc expression, and module enrichment scores for the Boroviak Diapause, LSC and Ramalho stemness gene signatures. q10 = 10th quantile, q90 = 90th quantile. d, UMAP projections of Baseline timepoint scRNA-seq data from MLL-AF9 + KrasG12D cells prior to transplantation. Top row left to right: Louvain clusters, embedded locations of cells comprising low to High output clones (red), Mid/High to low output clones (blue) clones that are extinguished with chemotherapy treatment (green) and clones that survive chemotherapy (purple). Bottom row left to right: Slpi expression, Cebpε expression B2m expression, module enrichment scores for the Boroviak Diapause and Fridman Senescence signatures. q10 = 10th quantile, q90 = 90th quantile. e, Gene expression heat map showing the top differentially expressed genes in cells comprising Mid/High to Low output and Low to High output clones in the Vehicle sample and at each treatment timepoint. Normalized expression shown on scale. f, Selected enriched Biological Process gene ontologies for the set of genes up and downregulated in Low to High output clones vs Mid/High to Low output clones in the Vehicle sample and at each treatment timepoint. g, Selected enriched Biological Process gene ontologies for the set of genes up and downregulated in Mid/High to Low output clones vs Low to High output clones in the Vehicle sample and at each treatment timepoint.
Supplementary information
Supplementary Information
Supplementary Figs. 1 and 2.
Supplementary Table 1
A list of oligos.
Supplementary Table 2
Hashtag information.
Supplementary Table 3
Exome data.
Supplementary Table 4
Gene sets.
Supplementary Table 5
Bursting analysis.
Rights and permissions
About this article
Cite this article
Fennell, K.A., Vassiliadis, D., Lam, E.Y.N. et al. Non-genetic determinants of malignant clonal fitness at single-cell resolution. Nature 601, 125–131 (2022). https://doi.org/10.1038/s41586-021-04206-7
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1038/s41586-021-04206-7
This article is cited by
-
Gene-expression memory-based prediction of cell lineages from scRNA-seq datasets
Nature Communications (2024)
-
Tissue mosaicism following stem cell aging: blood as an exemplar
Nature Aging (2024)
-
The potential of epigenetic therapy to target the 3D epigenome in endocrine-resistant breast cancer
Nature Structural & Molecular Biology (2024)
-
Single-cell lineage capture across genomic modalities with CellTag-multi reveals fate-specific gene regulatory changes
Nature Biotechnology (2023)
-
Unraveling non-genetic heterogeneity in cancer with dynamical models and computational tools
Nature Computational Science (2023)
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.
