SPLINTR – provides insights into the non-genetic transcriptional processes that underpin malignant clonal fitness

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 clear, little is known about the non-genetic mechanisms that contribute to intratumour heterogeneity and malignant clonal fitness. Here, using single-cell profiling and lineage tracing (SPLINTR)—an expressed barcoding strategy—researchers from the Peter MacCallum Cancer Centre trace isogenic clones in three clinically relevant mouse models of acute myeloid leukaemia. They found 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 they genetically validated 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 cells, leukaemia stem cells (LSCs) display heritable clone-intrinsic properties of high, and low clonal output that contribute to the overall tumour mass. The researchers demonstrated 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.

Generation and proof of concept of SPLINTR barcode libraries

Extended Data Fig. 1

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.

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

Fennell KA, Vassiliadis D, Lam EYN et al. (2021) Non-genetic determinants of malignant clonal fitness at single-cell resolution. Nature [Epub ahead of print]. [abstract]

Leave a Reply

Your email address will not be published. Required fields are marked *


Time limit is exhausted. Please reload CAPTCHA.