iFlpMosaics enable the multispectral barcoding and high-throughput comparative analysis of mutant and wild-type cells

Abstract

To understand gene function, it is necessary to compare cells carrying the mutated target gene with normal cells. In most biomedical studies, the cells being compared are in different mutant and control animals and, therefore, do not experience the same epigenetic changes and tissue microenvironment. The experimental induction of genetic mosaics is essential to determine a gene cell-autonomous function and to model the etiology of diseases caused by somatic mutations. Current technologies used to induce genetic mosaics in mice lack either accuracy, throughput or barcoding diversity. Here we present the iFlpMosaics toolkit comprising a large set of new genetic tools and mouse lines that enable recombinase-dependent ratiometric induction and single-cell clonal tracking of multiple fluorescently labeled wild-type and Cre-mutant cells within the same time window and tissue microenvironment. The labeled cells can be profiled by multispectral imaging or by fluorescence-activated flow cytometry and single-cell RNA sequencing. iFlpMosaics facilitate the induction and analysis of genetic mosaics in any quiescent or progenitor cell, and for any given single or combination of floxed genes, thus enabling a more accurate understanding of how induced genetic mutations affect the biology of single cells during tissue development, homeostasis and disease.

Fig. 1. iFlpMosaics allow efficient and ratiometric labeling of mutant and wild-type cells.

a, Flp/FlpO-ERT2-induced recombination of the iFlpMosaic allele leads to one of two possible outcomes: cells expressing MTomato and Cre (mutant) or cells expressing MYFP (wildtype) (Extended Data Fig. 2). A confocal micrograph shows some labeled retina astrocytes (AC), endothelial cells (EC, ERG⁺ nuclei) and blood cells (BC). b, The Tg-iFlp^{MTomato-H2B-GFP-Cre/MYFP-H2B-Cherry-FlpO} allele and confocal micrographs showing the full spectral separation of the four fluorescent labels expressed in mutant and wild-type retina cells (ECs and few ACs). c, The proportion of MTomato⁺ and MYFP⁺ cells detected by FACS. n.s., not significant. d, A comparison of MTomato intensity and Cre expression. e, Recombination of the Cre-reporter Rosa26^-LSL-YFP in MTomato⁺ cells (white arrowheads) after induction of the R26-iFlp^{MTomato-Cre/MYFP} allele by Apln-FlpO. There are few false positives (red arrows, MTomato⁺). f, qRT-PCR assessing gene deletion efficiency. P values are indicated on top of the horizontal lines. g, Induction of R26-iFlp^{MTomato-Cre/MYFP} in mice carrying a floxed allele for Rbpj, an essential gene for arterial development, results in only 3.5% MTomato-2A-Cre⁺ cells in arteries (A), versus 96% in veins (V) and capillaries. h, Immunostaining for VEGFR2 and quantification of gene deletion in P7 heart sections ECs (Apln⁺). i, Immunostaining for RBPJ and quantification of gene deletion in adult lung epithelium (induced with FlpO-ERT2). j, Immunostaining for MYC and MYCN and quantification of genetic deletion in P7 retina ECs (Apln⁺). k, Immunostaining for DLL4 and quantification of gene deletion in adult lung epithelium (induced with FlpO-ERT2). The red arrows indicate the cells with MTomato-2A-Cre expression, the green arrows indicate the cells with MYFP expression and the gray arrows indicate the nonfluorescent/wild-type cells. The data are presented as mean values ± standard deviation. Scale bars, 50 μm for b, e, g, h and j and 10 μm for i and k. Source data

Fig. 2. Generation of ubiquitous and tissue-specific FlpO-ERT2 mouse lines.

a, Recombination of the R26-iFlp^{MTomato-Cre/MYFP} allele induced by the published R26^{CAG-FlpO-ERT2} allele after a single dose of tamoxifen. b, A schematic of novel Tg(Ins-CAG-FlpO-ERT2) alleles containing DNA elements to enhance FlpO-ERT2 expression (Extended Data Fig. 2). c, A comparison of FlpO copy number by genomic DNA qRT-PCR of founders (F) 1–5. d–g, A comparison by FACS and histology of the recombination efficiency of the different FlpO-ERT2-expressing mouse lines/founders at embryonic, postnatal and adult stages. The fold change is indicated for the most substantial changes. CardioM, cardiomyocytes; LSK⁺, Lin⁻Sca1⁺c-Kit⁺ cells; MP, myeloid progenitors. h, CRISPR–Cas9 targeting with a guide RNA (gRNA) and a donor DNA for the generation of novel endothelial-specific Cdh5-FlpO-ERT2-expressing mouse lines from the published Cdh5-CreERT2 mouse line. i, A comparison of FlpO and Cre copy number by genomic DNA qRT-PCR. j, Recombination efficiency of founder number 1 of the newly generated Tg(Cdh5-FlpO-ERT2) mouse line in embryonic, postnatal and adult tissues and some representative confocal micrographs. White box delimits the magnified area shown to the right. The data are presented as mean values ± standard deviation. Scale bars, 100 μm. Source data

Fig. 3. Ratiometric iFlpMosaics allow high-throughput assessment of a cell-autonomous gene function.

a,b, R26-iFlp^{MTomato-Cre/MYFP} mosaics were induced with Apln-FlpO (recombines endothelial and derived hematopoietic cells at E9.5) on control (wildtype) and Myc-floxed backgrounds, and the tissues were analyzed at P7. The chart and FACS data correspond to CD31⁺ (ECs, in a) and CD45⁺ (blood, in b) cells. The absolute fold change of the log₂ ratio is indicated in red and P values in black. c, Similar FACS analysis of the indicated organs from R26-iFlp^{MTomato-Cre/MYFP} mice carrying the ACTB-FlpE allele. d, iFlpMosaic deletion of Foxo1/3/4 from P1 to P15 increases the FACS frequency of MTomato⁺/mutant cells in the heart but not in the liver or lungs. e, Confocal micrographs of control and Foxo1/3/4 mosaic P6 and P15 retinas from animals induced at P1. Dense clusters of ERG⁺ ECs are indicated with yellow circles. The chart shows ratiometric analysis of retinal immunostainings for MTomato, MYFP and ERG (labels EC nuclei). f, Confocal micrographs of P6 retinas showing FOXO1 immunostaining and deletion exclusively in MTomato⁺ cells. The data are presented as mean values ± standard deviation. Scale bars, 100 μm. Source data

Fig. 4. High-resolution iFlpMosaics reveal the impact of genetic mutations on single-cell clonal dynamics.

a, Quantification of Ki67⁺ cells (in cycle) and EdU⁺ cells (S phase) in livers. b, Left: a schematic of the alleles and how they can be used to fluorescently barcode and track the clonal expansion of single mutant/MTomato⁺ and wild-type (WT)/MYFP⁺ cells from P1 to P20. Right: representative confocal micrographs with magnified insets (white boxes) of P20 livers (from animals induced with 4-OHT at P1) with different recombination frequencies (Extended Data Fig. 6). c, The clone size distributions of hepatocytes at both low and high recombination rates. d, The whole-liver weight at P1 and P20. e, A quantification of MYFP⁺ (Myc^WT) and MTomato⁺ (Myc^KO) clone size frequency. f, A chart depicting the normalized expansion of Myc^WT and Myc^KO hepatocytes. g, The representative confocal micrographs of P20 pancreas from control and Myc-floxed mice carrying the indicated iFlpMosaic alleles. Right: inverted LUT images depicting the clone mapping (Fiji image analysis scripts) according to their dual color code. h, The quantification of clone size frequency. i, A chart depicting the normalized expansion of Myc^WT and Myc^KO pancreatic cells. j, log₂ ratio of MTomato⁺ cells to MYFP⁺ cells in control and Myc-floxed pancreas, showing a 3.7-fold decrease of MTomato⁺ (Myc^KO) cells over 20 days. k, Representative confocal micrographs of adult livers from control and Myc-floxed mice in which the indicated alleles were induced 14 days before tissue collection. Right: a semiautomatic mapping of clones according to their fluorescent barcode. Note the loss of MTomato⁺ cells in the Myc-floxed background. l, The clone size, showing that WT hepatocytes rarely divide or expand over a 2 week period. m, The clone frequency, showing the lack of impact of Cre expression in WT MTomato⁺ cells. n, The percentages of proliferative (composed of more than one cell) and nonproliferative clones. o, log₂ ratio of the MTomato⁺ and MYFP⁺ cell frequency in control and adult Myc-floxed livers. p, The clone frequency Myc^WT and Myc^KO hepatocytes. q, The percentages of proliferative and nonproliferative Myc^WT and Myc^KO clones. The data are presented as mean values ± standard deviation. Scale bars, 100 μm. Source data

Fig. 5. iDre/Flp^Progenitor enables the induction of genetic mosaics in the progeny of single cells.

a, A schematic of the iDre/Flp^Progenitor DNA construct used to generate transgenic mice. b, A strategy for the generation of the Tg(Ins-CAG-DreERT2) allele. c,d, Retinal confocal micrographs showing that the iDre/Flp^Progenitor allele increases the recombination frequency of the iFlpMosaic allele, when induced either by the CAG-DreERT2 (c) or Cdh5-FlpOERT2 (d) alleles. H2B-V5 labels the nuclei of cells expressing the iDre/Flp^Progenitor allele. e, A schematic of the genetic recombination cascade and cellular markers in the progenitor and daughter cells. f, The kinetics of the induced recombination cascade determined by FACS of in vitro cultured fibroblasts. Top: plots and chart show the hCD2 reporter signals when contrasted with the autofluorescence control PerCP dye channel. n.s., not significant. Bottom: FACS plots show the recombination frequency and expression of the iFlpMosaic allele reporters over time, within hCD2⁺ cells. g, A confocal analysis of the P6 pancreas, 3 days after a very low dose of 4-OHT induction, showing twin-spot clones. Only H2B-V5⁺ (iDre/Flp^Progenitor+) cells give rise to twin-spot clones having both Tomato⁺ and YFP⁺ cells (orange bar). h, Induction with a single dose of 4-OHT (40 mg kg⁻¹) 5 days before collecting the tissues shows that the new Ki67-2A-DreERT2 allele (Extended Data Fig. 7f) drives recombination in Ki67⁺ progenitor cells of the indicated organs, generating twin-spot clones. Background immuno, background or noise signal from immunostaining with anti-GFP/YFP; Exp., expected frequencies of recombination. The data are presented as mean values ± standard deviation. Scale bars, 24 μm in g and h, and 500 μm in c and d. Source data

Fig. 6. Combining scRNA-seq with ratiometric iFlpMosaics to reveal cell-autonomous gene function.

a, A schematic of the induction and analysis stages (four embryos were used). b, Uniform manifold approximation and projections (UMAPs) showing Rbpj^KO (MTomato⁺) and Rbpj^Wt (MYFP⁺) cells and dot plots with the top markers used to identify the major cell types. CNS, central nervous system; PNS, peripheral nervous system. c, A dot plot showing the expression for Rbpj whole mRNA (mostly the 3′ undeleted mRNA) and the deleted floxed exons 6–7. Bottom: the expression of the canonical NOTCH/RBPJ signaling target Hes1 (regulated also by other pathways) and MKi67. d,Histobars with the proportions of the different clusters identified in c. e, Comparative UMAP showing that mosaic deletion of Rbpj changes the relative proportion of different cell types. f, Analysis within the endothelial cell cluster, including general embryo, brain and liver ECs, with some cells undergoing endothelial-to-mesenchymal transition. g,h, Rbpj^KO mutant cells present a deregulation of arterial (Gja4 and Gja5) (g) and tip cell (Esm1, Kcne3 and Apln) marker genes (h). i, Gene set enrichment analysis (GSEA) comparing Rbpj^Wt with Rbpj^KO cells (see Extended Data Fig. 8 for the analysis of all other cell clusters). NES, normalized enrichment score. j, UMAPs showing the differential distribution of the Notch1/2/3^KO (MTomato⁺) and Notch1/2/3^Wt (MYFP⁺) cells (collected from four embryos) and the different clusters they form. k, A dot plot showing the frequency and amplitude of expression for the three Notch receptors whole mRNA (mostly the 3′ undeleted mRNA in the case of Notch2 and Notch3) and their deleted floxed exons. Bottom: the expression of the Notch receptors (whole mRNA) and their canonical signaling targets Hes1/Hey1/Hey2 (regulated also by other pathways) and the G2/M phase marker MKi67. l, The histobars and UMAP showing that mosaic deletion of Notch1/2/3 (MTomato⁺ cells) changes the relative proportion of different cell types, particularly neural cells (Extended Data Figs. 9 and 10).

Extended Data Fig. 1. False positives and false negatives with Cre-dependent mosaic genetics.

a, Schematic showing how a standard Cre-reporter can recombine and label cells without the deletion of any floxed gene (G) (false positives) and how the floxed gene can be deleted in non-Cre-recombined cells (false negatives). b, Different recombination efficiencies of different Rosa26 Cre-reporters in postnatal day 7 mice having the Cdh5-CreERT2 allele, despite all localizing to the Rosa26 locus and having a similar genetic distance between LoxP sites. c,d, Analysis of FACS or immunohistochemistry data reveals that Cre-reporters only accurately report recombination of themselves, not that of other floxed alleles (reporters), particularly at low tamoxifen doses. Data are presented as mean values +/− SD. Source data

Extended Data Fig. 2. iFlpMosaics are neither toxic nor leaky.

a, b, Schematic diagrams of the novel Rosa26-iFlp^{MTomato-Cre/MYFP} and Tg-iFlp^{MTomato-H2B-GFP-Cre/MYFP-H2B-Cherry-FlpO} alleles, showing the genetic distances between the mutually exclusive FRT site pairs and the expected outcomes after FlpO/FlpO-ERT2 recombination. c, d Confocal microscopy and FACS analysis of mouse ES cells used to generate Rosa26-iFlp^{MTomato-Cre/MYFP} mice 3 days after transfection with FlpO-expressing plasmids. e-g, Frequency of recombination and expression detected by microscopy in ES cells and ECs derived from embryoid bodies (EBs). h, Relative frequency of MTomato-2A-Cre+ and MYFP+ cells in ES cells (in vitro) and in blood (in vivo); the absence of change over time shows that permanent expression of Cre is non-toxic to cells. i, j FACS analysis of FlpO versus Cre-dependent leakiness in adult organs of mice carrying the standard Rosa26-LSL-YFP allele in combination with iFlpMosaics alleles. FlpO leakiness was not detected (0% MTomato+ cells), and Cre-non-self-leakiness (only detectable with the additional Rosa26-LSL-YFP allele) was observed only in a very small fraction of cells from animals carrying the Tg-iFlp^{MTomato-H2B-GFP-Cre/MYFP-H2B-Cherry-FlpO} allele and not at all in animals carrying the Rosa26-iFlp^{MTomato-Cre/MYFP} allele. k, CreERT2 alleles (that is Cdh5-CreERT2) can be leakier than iFlpMosaics. Scale bars 50μm (c and g left) and 200μm in f and g center and right. Data are presented as mean values +/− SD. For statistics see Source Data file 1. Source data

Extended Data Fig. 3. Comparative analysis of recombination efficiency in FlpO-ERT2 mouse lines.

a-c, Representative FACs plots of different adult organs (see quantitative data in Fig. 2g), showing MYFP+ and MTomato+ cell frequencies in mice carrying the iFlp^{MTomato-Cre/MYFP} allele in combination with the published R26^{CAG-FlpO-ERT2} or the newly generated Tg(Ins-CAG-FlpO-ERT2)^F2 allele. Induction was 3 daily consecutive injections of tamoxifen (60mg/kg) and tissues collected 7 days after. d, Confocal micrograph of a flat-mounted adult mouse aorta 2 weeks after induction. e, FACS analysis showing the lack of leakiness of the Tg(Ins-CAG-FlpO-ERT2)^F2 and iFlpMosaic alleles in the absence of induction/tamoxifen. f, g, Representative FACS plots showing that in both the heart and the lung CD45+ blood cells are much more affected by Myc deletion than CD31+ ECs. Scale bar 100μm. Source data

Extended Data Fig. 4. Validation of iFlpMosaics for ratiometric quantitative gene function analysis in cell migration and proliferation.

a, Representative confocal micrographs of a P6 retina from a Rbpj-floxed mouse carrying the iFlp^{MTomato-Cre/MYFP} and Apln-FlpO alleles. Apln is an X-chromosomal gene (mosaically expressed in females) expressed at the angiogenic front of growing vessels, recombining in angiogenic front ECs that later form the entire vasculature. Apln also recombines in non-endothelial (ERG-negative) cells in the retina. The right micrograph shows ERG-signal segmentation (EC nuclei), with nuclei of MTomato+ (Rbpj^KO) cells in red and nuclei of MYFP+ (Rbpj^WT) cells in green. Grey nuclei are in non-recombined ERG+ cells. Dashed white lines demark the leading edge (LE), angiogenic front (AF), and mature area (MA). A, arteries; V, veins. The chart shows the log2 ratios of MTomato+ to MYFP+ ECs in each retinal region, showing the enrichment of Rbpj^KO cells in the LE and AF, which suggests that loss of this gene induces cell migration to the front of the plexus. b, c Representative confocal micrographs of a P6 retina from Rbpj-floxed mouse carrying the iFlp^{MTomato-H2BGFP-Cre/MYFP-H2BCherry} and Apln-FlpO alleles. This allows the nuclear labeling of Rbpj^KO (H2B-GFP+) and Rbpj^WT (H2B-Cherry+) ECs and blood cells, which is more convenient for cell object segmentation and quantification of nuclear-specific proteins, such as the proliferation and cell-cycle arrest markers Ki67 and p21. Nuclear segmentation outlines: red, Cherry+/Ki67- or p21-; pink, Cherry+/Ki67+ or p21+; green, GFP+/Ki67- or p21-; cyan, GFP+/Ki67+ or p21+. Data are presented as mean values +/− SD. For statistics see Source Data file 1. Scale bars, 150μm. Source data

Extended Data Fig. 5. Combining iFlpMosaic with iFlp^Chromatin yields higher single-cell clonal resolution.

a, Representative confocal micrograph of adult liver sections from mice carrying the indicated alleles. Combination of these membrane and nuclear reporter alleles allows up to 3 mutant and 1 wildtype cell barcodes (labeling). Note that binucleated (2x2n) or polyploid (4n) hepatocytes can have more multispectral barcodes. b, Representative confocal micrograph of a P7 retina from an animal containing the indicated alleles and induced at P1 with 4-OHT. The bar shows the ratiometric frequency of each nuclear/H2B+ fluorescent marker from the iFlp^Chromatin allele. c, Representative confocal micrograph of the indicated adult organ sections. Mononucleated diploid cells (kidney and lung) express only 1 of 3 possible nuclear/chromatin markers, whereas multinucleated or polyploid cells (hepatocytes and cardiomyocytes) can express a combination of 2 or 3 possible markers (6 possible chromatin barcodes in total). d, Recombination and cell barcoding possibilities in animals carrying the indicated alleles. The representative confocal micrographs of a P7 retina show that a fraction of wildtype and mutant cells can both be labeled in 4 different ways, substantially increasing single-cell clonal resolution. The panel to the right illustrates how a Fiji image analysis script automatically detects, segments, and pseudocolors the nuclei of dual-labeled cells (cells with both nuclei and membrane labelled). e, Recombination and barcoding possibilities in a mononucleated diploid cell in animals carrying the indicated alleles. f, Representative confocal micrographs of an adult liver, collected 2 weeks after tamoxifen induction, show that mutant hepatocytes can be labeled with up to 16 different fluorescent barcodes, depending on whether the cells are mononucleated (diploid) or multinucleated (polyploid). Wildtype hepatocytes can have up to 8 different fluorescent barcodes (only 6 shown). Scale bars, 100μm. Source data

Extended Data Fig. 6. Mapping the clonal expansion of single mutant and wildtype cells with iFlpMosaics.

a,b, Representative confocal images (low magnification large tile scans) of entire P20 liver sections from mice carrying the indicated alleles. Recombination of the reporter alleles was induced at P1 and resulted in different recombination frequencies. Right panels show the results of automatic image processing with Fiji scripts, which generate spatial maps of dual-labeled clones of cells (with membrane and nuclear labeling). The bigger the circle, the bigger the clone. c, Quantification and representative confocal micrographs of proliferative pancreatic cells (Ki67+ in cycle or EdU+ in S-phase) at different postnatal stages. d, Quantification and representative confocal micrograph of Ki67+/DAPI+ cells in the adult liver. e, Representative confocal micrographs of adult liver sections from control and Myc-floxed mice carrying the indicated alleles. Left panels show endogenous fluorescent signals from the reporter alleles. Center panels show sections from the same mice immunostained with antibodies to DsRed (detects MTomato and H2B-Cherry), GFP (detects MYFP and H2B-GFP), or the HA epitope (detects MTFP1 and H2B-Cerulean) together with Fiji-script-generated spatial maps of dual-labeled clones (with membrane and nuclear labeling) reveal hepatocyte ploidy and clone size. Graphs show the percentages of mononucleated cells, binucleated cells, and 2-cell mononuclear clones. In control mice, MYFP+ (WT) and MTomato+ (WT) diploid cells occur in similar proportions, whereas in Myc-floxed mutants MTomato+ (Myc^KO) diploid cells are more frequently binucleated and give rise to fewer 2-cell mononucleated clones than MYFP+ (Myc^WT) cells. Scale bars, 50μm (c, d), the rest 300μm. Data are presented as mean values +/− SD. For numerical data see Source Data file 1. Source data

Extended Data Fig. 7. iDre/Flp^Progenitor enables the effective induction of genetic mosaics from single progenitor cells.

a, FACS plots showing that the iDre/Flp^Progenitor allele significantly increases the recombination frequency of the iFlp^{MTomato-Cre/MYFP} allele when combined with the Tg(Ins-CAG-FlpO-ERT2) allele, because it is much more sensitive to FlpO-ERT2 activity. b,c, In animals lacking the iDre/Flp^Progenitor allele, most single-cell derived clones are formed by just one type of cell (MTomato+ or MYFP+). With the iDre/Flp^Progenitor allele, it is possible to induce clones formed by mutant (MTomato+) and wildtype (MYFP+) cells derived from the same single progenitor cells. d, Representative confocal micrograph of a P7 retina from an animal carrying the indicated alleles and induced with a very low dose of 4-OHT (0,04 mg/kg) at P3. After induction of the iDre/Flp^Progenitor allele, cells express the nuclear marker H2B-V5 and FlpO. After cell division, FlpO will recombine the iFlp^{MTomato-Cre/MYFP} allele, generating MYFP+ and MTomato+ progeny cells derived from a single initial recombination event. Cells at the angiogenic front disperse/migrate more after division than cells in the mature plexus. e, Time-lapse confocal microscopy of mouse ES cells derived from a mouse with the indicated genotype. 24h after administration of 4-OHT (1μM), ES cells first co-express the marker hCD2 at their surface (labelled with anti-hCD2-APC) and FlpO. During the subsequent hours, the iFlpMosaic-recombined progeny of these single progenitor ES cells express either MTomato-2A-Cre+ or MYFP+ (twin clones). f, Through the use of CRISPR-Cas9 mediated gene targeting, we generated several founder mice carrying the Mki67-2A-DreERT2 knock-in allele detected with primers P1 to P3. On the right a representative confocal micrograph from a growing hair follicle (see also Fig. 5h) containing many labelled cells 5 days after induction. Scale bars 500 μm in b, c, 100μm in d, 20μm in e, and 25μm in f. Data are presented as mean values +/− SD. For statistics see Source Data file 1. Source data

Extended Data Fig. 8. Single cell RNA-seq analysis combined with ratiometric iFlpMosaics uncovers the cell-autonomous Rbpj gene function in diverse cell types.

a, Umaps and barplots showing the identified clusters and their frequencies among MYFP+ (Rbpj^WT) and MTomato+ (Rbpj^KO) cells. Dot plots show the frequency (size) and expression level (color intensity) for the top cluster marker genes. b, Top differentially expressed genes per cluster between MYFP+ (Rbpj^WT) and MTomato+ (Rbpj^KO) cells. Most of the charts also show the expression of Rbpj, its canonical target Hes1, and the proliferation markers Ki67 (cells in G2/M of the cell-cycle) and Cdkn1a (likely arrested cells). c, Gene set enrichment analysis (GSEA) pathways with their normalized enrichment score. Source data

Extended Data Fig. 9. Single cell RNA-seq analysis combined with ratiometric iFlpMosaics uncovers the combined NOTCH1/2/3 receptors cell-autonomous function in diverse embryo cell types.

a, Umaps and barplots showing the identified clusters and their frequencies among MYFP+ (Notch1/2/3^WT) and MTomato+ (Notch1/2/3^KO) cells. Dot plots showing the frequency (size) and expression level (color intensity) for the top cluster marker genes. b, Top differentially expressed genes per cluster between MYFP+ (Notch1/2/3^WT) and MTomato+ (Notch1/2/3^KO) cells. Dot plots also show the expression of Notch1 (note that Notch2 and Notch3 3′mRNA is still expressed after deletion and detected by scRNAseq, as shown in Fig. 6k), its canonical target genes (Hes and Hey, these are also regulated by other pathways), and the proliferation markers Ki67 (labels cells in G2/M of the cell-cycle) and Cdkn1a (likely arrested cells). c, Gene set enrichment analysis (GSEA) pathways with their normalized enrichment score. Source data

Extended Data Fig. 10. Single cell RNA-seq analysis combined with ratiometric iFlpMosaics uncovers the combined NOTCH1/2/3 receptors cell-autonomous function in diverse embryo cell types.

a, Umaps and barplots showing the identified clusters and their frequencies among MYFP+ (Notch1/2/3^WT) and MTomato+ (Notch1/2/3^KO) cells. Dot plots showing the frequency (size) and expression level (color intensity) for the top cluster marker genes. b, Top differentially expressed genes per cluster between MYFP+ (Notch1/2/3^WT) and MTomato+ (Notch1/2/3^KO) cells. Dot plots also show the expression of Notch1 (note that Notch2 and Notch3 3′mRNA is still expressed after deletion and detected by scRNAseq, as shown in Fig. 6k), its canonical target genes (Hes and Hey, these are also regulated by other pathways), and the proliferation markers Ki67 (labels cells in G2/M of the cell-cycle) and Cdkn1a (likely arrested cells). c, Gene set enrichment analysis (GSEA) pathways with their normalized enrichment score. Source data