A cancer rainbow mouse for visualizing the functional genomics of oncogenic clonal expansion

Abstract

Field cancerization is a premalignant process marked by clones of oncogenic mutations spreading through the epithelium. The timescales of intestinal field cancerization can be variable and the mechanisms driving the rapid spread of oncogenic clones are unknown. Here we use a Cancer rainbow (Crainbow) modelling system for fluorescently barcoding somatic mutations and directly visualizing the clonal expansion and spread of oncogenes. Crainbow shows that mutations of ß-catenin (Ctnnb1) within the intestinal stem cell results in widespread expansion of oncogenes during perinatal development but not in adults. In contrast, mutations that extrinsically disrupt the stem cell microenvironment can spread in adult intestine without delay. We observe the rapid spread of premalignant clones in Crainbow mice expressing oncogenic Rspondin-3 (RSPO3), which occurs by increasing crypt fission and inhibiting crypt fixation. Crainbow modelling provides insight into how somatic mutations rapidly spread and a plausible mechanism for predetermining the intratumor heterogeneity found in colon cancers.

Fig. 1

Engineering cancer rainbow mice. a The current two-step model of intestinal field cancerization. Somatic mutations increase stem cell fitness and lead to the fixation of premalignant clones within crypts (orange). Crypt fission results in duplication of premalignant crypts and the lateral propagation of somatic fields. b The Crainbow expression vector incorporates 4-tandem cassettes downstream of the ubiquitous chicken-β-actin promoter (CAG). The construct confers ubiquitous expression of a membrane-targeted and chemically inducible near-infrared fluorogen-activating peptide (FAP-Mars1) as a control. Cell type specific activation of Cre recombinase mediates recombination at one of three positions through inclusion of three pairs of orthogonal lox sites (LoxN: white triangle, Lox2272: hatched triangle, and LoxP: filled triangle). Single-copy transgene insertion provides a single outcome per cell and barcoding of each genetic fate by fluorescent imaging. c Candidate tumor driver genes are synthesized/PCR amplified and then In-Fusion^® cloned into a bicistronic expression cassette. Multi-site Gateway™ cloning is used to directionally clone each pENTR plasmid into a Gateway™-compatible ROSA-targeting vector in a single step.

Fig. 2

Validating Crainbow functionality. a Diagram of NCAT-Crainbow mice. See also Table 1 and Supplementary Fig. 1. b NCAT^VilCre small intestine (N = 10 mice, PND9–PND20) was prepared as a whole-mount and confocal imaged for XFP expression in the intestinal crypts. c Inset in “b” at higher magnification. d Crypts were color segmented and counted. The experimentally observed value was normalized to the predicted stochastic outcome (0.33/XFP). The asterisk denotes statistical significance by one-way ANOVA (TagBFP vs. mTFP1: p < 1e–6, TagBFP vs. mKO: p < 1e–6, mTFP1 vs. mKO: not significant). NCAT^CreER/T2 mice (N = 8, 19–22 weeks of age) were injected with 200 mg/kg tamoxifen. Mice were sacrificed 12 days post tamoxifen injection, and the small intestine was vibratome sectioned and antibody stained to recover the quenched TagBFP signal. e Sections were imaged by confocal microscopy for XFP expression (TagBFP: cyan, EYFP: mTFP1, mKO: magenta; segmented nuclear masks shown and overlaid with surface rendered tissue outline). f Rotated inset in “e” at higher power. g Nuclei were segmented, counted, and normalized to the predicted stochastic outcome (0.33/XFP). The asterisk denotes statistical significance by one-way ANOVA (TagBFP vs. mTFP1: p < 1e–6, TagBFP vs. mKO: p < 1e–6, mTFP1 vs. mKO: 8.4e–4). (SEM included for each graph). Scale Bars = 100 µm in b, c, f and 2 mm in e. Source data are provided as a “Source Data file”.

Fig. 3

Widespread expansion of oncogenic clones during perinatal development. a Diagram of MCAT-Crainbow mice. See also Table 1 and Supplementary Fig. 5. b MCAT^VilCre small intestine (N = 10 mice, 3–6 weeks of age) prepared as a wholemount and confocal imaged. c Inset in “b” at higher magnification. d MCAT^VilCre Crypts were color segmented, counted and normalized to the positional bias calculated in NCAT^VilCre mice. Asterisk denotes statistical significance by one-way ANOVA (mTFP1 vs. EYFP: p = 0.003, mTFP1 vs. mKO: p = 0.016, EYFP vs. mKO = 3e–6). e Immunostaining for FLAG, V5, or HA epitopes (magenta) specific to each ßcat isoform in MCAT^VilCre small intestine vibratome slices and merged with fluorescent lineage markers (mTFP1: cyan, EYFP: yellow, and mKO: orange). Arrows denote isoform expression with cognate lineage reporter (FLAG and mTFP1, V5 and EYFP, and HA and mKO). Corresponding insets depict higher magnification images. Arrowheads denote membrane-localized ßcat, whereas asterisk denotes nuclear-localized ßcat. Epitope stains (magenta) are also presented as merged and as a single-channel image with its cognate fluorescent lineage reporter (green). f HEK cells were transiently transfected with MCAT isoforms, fixed, stained, and imaged for the indicated epitope (magenta) and fluorescent reporter (green). Cells were also cotransfected with epithelial cadherin (CDH1) as indicated. Arrows denote sequestration of ßcat at the plasma membrane, and the asterisk denotes nuclear ßcat. g Wnt signalling activity for each oncogene in the absence of CDH1 (solid bar) or in the presence of overexpressed CDH1 (hatched bar) (N = 6 wells per condition and independently repeated in four experiments). TOP FLASH activity was normalized to WNT/RSPO-stimulated control cells (dashed line). Asterisk denotes statistical significance by two-way ANOVA and Bonferroni’s multiple comparisons test (cyan < 1e–6, yellow = 0.01, magenta = 0.02). (SEM included for each graph). Scale Bars = 1 mm in b, 100 µm in c/e, 15 µm in e: insets 1–3, and 10 µm in f. Source data are provided as a “Source Data file”.

Fig. 4

The crypt microenvironment impairs the spread of oncogenic clones in adults. a, b MCAT^CreER/T2 mice (>6 weeks of age) were I.P. injected with 200 mg/kg of tamoxifen and sacrificed 3 days (N = 8) or 8 weeks later (N = 5). The small intestine was vibratome sectioned and imaged. c Nuclei were segmented, counted, and normalized to the total volume of tissue imaged. Clone spread was determined by comparing the change in total recombined nuclei for each fate at day 3 and 8 weeks. Statistical significance by mixed effects modelling and Sidak multicomparison correction (∆Nßcat: p = 0.99, Ccat/Lef: p = 0.14 ∆Nßcat∆C: p = 0.99). d, e MCAT^CreER/T2 (>6 weeks of age) were injected three times (Day 1, 3, 5) with 200 mg/kg tamoxifen and sacrificed at Day 7 (N = 4) or at 8 weeks (N = 5). The small intestine was whole mount imaged by confocal microscopy. Representative crypts are outlined and f the relative fraction of crypt fixation events for each ßcat isoform was measured. Statistical significance by one-way ANOVA and Holm-Sidak’s multiple comparison test (∆Nßcat vs. Ccat/Lef1: p = 0.011 (∆Nßcat vs. ∆Nßcat∆C: p = 0.0062, Ccat/Lef1 vs. ∆Nßcat∆C p = 0.259). g MCAT^VilCre crypts were isolated and cultured as in the indicated media (+/−c59 inhibitor at 10 nM) and then imaged by confocal microscopy. h High-magnification view of wells (N = 6) from each condition in “g” and rotated corresponding insets. i MCAT^VilCre organoids were subcultured (P0–P5, and N = 5–10 wells per subculture) and the entire well imaged by confocal microscopy and organoids of each lineage were counted. Relative clone spreading for each organoid fate was calculated by normalization to the crypt number (C) at isolation (see also Supplementary Fig. 6c). The asterisk denotes statistical significance by mixed effects modelling and Dunnett post hoc relative to Crypt control (C) (∆Nßcat: p = 0.03, p = 0.002, p < 0.0001 and Ccat/Lef1: p < 0.0001 for each asterisk and ∆Nßcat∆C p < 0.0001 for each asterisk). j Organoid formation efficiency for MCAT crypts cultured in “g–h.” The asterisk denotes statistical significance by two-way ANOVA and Tukey’s multiple comparisons test. ENRY conditions: not significant between ßcat isoforms, whereas Ccat/Lef1 was significantly different from ∆Nßcat and ∆Nßcat∆C for each treatment (ENRY + c59: p = 0.0003, ENY: p = 0.03, and ENY + c59: p = 0.01). (SEM included for each graph). Scale bars = 200 µm in a, b, 100 µm in d, e, 2 mm in g, and 200 µm in h. Source data are provided as a “Source Data file”.

Fig. 5

Ectopic expansion of the intestinal stem cell compartment by expression of oncogenic RSPO3. a Multiprobe RNA FISH and confocal microscopy for Top2a (cyan), Lgr5 (yellow), and Rspo3 (magenta) in paraffin-embedded sections of the small intestine (SI) and colon (Col). b ROBO^VilCre mice expressing oncogenic human RSPO3. Position 0: control and membrane-targeted FAP-Mars1, Position 1: mTFP1 (cyan) coexpressed with 3×FLAG-RSPO3, EYFP (yellow) coexpressed with V5-PTPRK^e1:RSPO3^e2–5, and Position 3: mKO (magenta) coexpressed with 3×HA-PTPRK^e1–7:RSPO3^e2–5 (see also Supplementary Fig. 7). c Gross examination of the ROBO^VilCre mice and gastrointestinal tracts at PND16 compared with WT littermate controls. WT and ROBO d small intestines (SI) and e colons (Col) were paraffin embedded and sectioned for H&E pathology and coregistry of indicated cell type markers by multiprobe RNA FISH. Images are insets from whole-slide imaged Swiss rolls (see Supplementary Figs. 8–13). f MCAT^VilCre and ROBO^VilCre mice were injected with EdU (magenta) and coimaged for each lineage reporter by confocal microscopy (mTFP1: cyan, EYFP: yellow, mKO: orange). g ROBO^VilCre organoids can grow without exogenous RSPO3 (ENY) and were imaged by confocal microscopy for each lineage reporter. Scale bars = 10 µm in a, 2 cm in c, 100 µm in d, e, 50 µm in f, and 1 mm in g. Source data are provided as a “Source Data file”.

Fig. 6

Microenvironmental oncogenesis attenuates crypt fixation. a Vibratome cross-sections of MCAT^VilCre and ROBO^VilCre small intestines were confocal imaged and tiled for each lineage reporter (PND17). b Higher magnification of areas indicated in “a”. c Whole-mount confocal imaging of small intestines for each fluorescent lineage reporter in ROBO^VilCre mice. d Whole-mount confocal imaging of MCAT^VilCre small intestines. e Indicated callouts (1–8) from “c” at higher magnification. f The percentage of each crypts that are monoclonal (crypt fixation), biclonal, or triclonal was calculated for NCAT^VilCre mice (N = 3 and 938 crypts), MCAT^VilCre mice (N = 9, and 6006 crypts analyzed at PND17) and ROBO^VilCre mice (N = 6 and 3557 crypts analyzed). (SEM included for each graph). g Wholemount confocal imaging of ROBO^VilCre colon at PND18. h Region of interest in “g” at higher magnification. Arrows show examples of crypts where fixation has not occurred. Scale bars = 1 mm in a, 100 µm in b/d, g, h, 200 µm in c, and 40 µm in e. Source data are provided as a “Source Data file”.

Fig. 7

Microenvironmental oncogenesis drives heterogenous premalignant clones in adults. a Single cells were isolated from PND16 WT and ROBO^VilCre by fractionation to enrich for crypt epithelium. scRNAseq and tSNE visualization was performed for WT and ROBO^VilCre cells. (SC: stem cell, TA: transiently amplifying cell, ENT: enterocytes that included bottom (b1: S/G2M phase and b2: G1), middle (m), and top (t), and were recently described, SECR: secretory Paneth and Goblet cells, IMMU: immune, ENTEND: enteroendocrine, FIBR: fibroblast). For cell type markers, see also Supplementary Fig. 14. b Heatmap visualization of cell types present in WT and ROBO^VilCre isolates as a percentage of total cells. c Heat map visualization of cell cycle phase as a percentage of each cell type. d–g ROBO^CreER/T2 (N = 9 mice, 6–15 weeks of age) were I.P. injected with 200 mg/kg of tamoxifen and sacrificed 3 days (N = 4) or 8 weeks later (N = 5), and the small intestine was vibratome sectioned and imaged by tiling confocal microscopy. Confocal imaged vibratome sections of ROBO^CreER/T2 small intestines at d 3 days or e 8 weeks post tamoxifen injection. f Nuclei were segmented and counted and normalized to the total volume imaged. Clone spread was determined by comparing the total recombined nuclei for each fate at 8 weeks relative to day 3. The asterisk denotes statistical significance by mixed effects modelling and Sidak multicomparison correction (RSPO3: <0.000001, PTPRK^e1:RSPO3^e2–5 (P1:R2-5): p = 0.31 and PTPRK^e1–7:RSPO3^e2–5 (P1-7:R2-5): p = 0.94). g Evidence for attenuated crypt fixation 8 weeks post tamoxifen injection as revealed by chimeric crypts in whole-mount confocal imaging. (SEM included for each graph). Scale bars = 100 µm in d, e, g. Source data are provided as a “Source Data file”.

Fig. 8

Models for the initiation and spread of oncogenic clones. a Early acquisition of somatic mutations during a critical period drives efficient spread of oncogenic clones before adulthood. Intestinal development selects for ISCs with increased fitness. This could bias crypt fixation (i.e., neutral drift dynamics) or crypt formation during development. Premalignant clones can then spread throughout the intestinal epithelium due to the high rate of crypt fission that is observed in mice (postnatal days 5–22, P5–22) and humans (0–2 years of age). b Microenvironmental oncogenesis induces crypt fission and inhibits crypt fixation (i.e., neutral drift dynamics) in the adult intestine. This results in an expanded stem cell compartment and the rapid spread of heterogeneous premalignant clones in the adult intestine.