iSuRe-HadCre is an essential tool for effective conditional genetics

Abstract Methods for modifying gene function at high spatiotemporal resolution in mice have revolutionized biomedical research, with Cre-loxP being the most widely used technology. However, the Cre-loxP technology has several drawbacks, including weak activity, leakiness, toxicity, and low reliability of existing Cre-reporters. This is mainly because different genes flanked by loxP sites (floxed) vary widely in their sensitivity to Cre-mediated recombination. Here, we report the generation, validation, and utility of iSuRe-HadCre, a new dual Cre-reporter and deleter mouse line that avoids these drawbacks. iSuRe-HadCre achieves this through a novel inducible dual-recombinase genetic cascade that ensures that cells expressing a fluorescent reporter had only transient Cre activity, that is nonetheless sufficient to effectively delete floxed genes. iSuRe-HadCre worked reliably in all cell types and for the 13 floxed genes tested. This new tool will enable the precise, efficient, and trustworthy analysis of gene function in entire mouse tissues or in single cells.


Introduction
Understanding how genes function requires methods to deplete or enhance their expression, thus allowing assessment of the ensuing phenotypic consequences.The current goldstandard method uses the Cre-Lox technology, which enables precise loss-or gain-of-gene-function ( 1 ).However, this technology often produces variable results and requires careful validation of the intended conditional genetic deletions ( 2 ,3 ).This is largely because different floxed genes vary in their sensitivity to Cre-mediated recombination ( 4 ,5 ).Validation of gene deletion involves either whole-organ or cell-type-specific FACS isolations in bulk, qRT-PCR, or western blot analyses with specific probes.However, these approaches lack single cell resolution and the tissues extracted for genetic deletion validation cannot be the same as those used for phenotype or imaging analysis, frequently resulting in 'genetically blind' results.The alternative in situ validation of protein or mRNA loss after deletion of a floxed gene is often difficult due to lack of good probes or antibodies giving a good signal-noise ratio at single cell resolution, in every cell expressing the gene.
As an alternative, many scientists infer the intended deletion of a floxed gene by using standard Cre-reporters to label cells that recombined the reporter allele.However, recombination of generic Cre-reporters does not correlate with recombination of other floxed genes (6)(7)(8), particularly when using tamoxifen-inducible Cre lines, in which recombination activity is weaker and often incomplete, targeting only a fraction of the tissue cells.
We previously generated the iSure-Cre allele, that significantly facilitated conditional genetics because it linked the expression of a single and easy to detect fluorescent reporter, to the permanent expression of Cre, which unlike CreERT2, efficiently recombines and deletes floxed genes ( 7 ).This guarantees that, irrespective of the level of CreERT2 expression and the intensity of tamoxifen induction, all reporter-expressing cells will have permanent expression of Cre, drastically reducing the occurrence of false positives.False positives occur when cells expressing a given Cre-reporter, do not have recombination of the intended floxed gene, and are a major cause of variability and inaccuracy in biomedical research using mouse genetics.Despite its advantages, the first-generation iSuRe-Cre technology also has its caveats, such as relatively low sensitivity to CreERT2 activity, leakiness, and the possibility of toxicity in cells permanently expressing Cre ( 7 ).
Here, we present the characterization of several new mouse lines that were designed to overcome these limitations.Only one of these showed a significant improvement compared with the original iSuRe-Cre line.This new line, which we call iSuRe-HadCre , is several fold brighter, more sensitive to induction by CreERT2 / tamoxifen, has no leakiness, and ensures that Cre is only transiently expressed, thus preventing Crerelated toxicity.iSuRe-HadCre is based on a new tight and inducible recombination cascade that first converts inducible CreERT2 activity into constitutive Cre expression, followed by an inbuilt FlpO-recombinase-dependent step that switches off Cre and FlpO expression and simultaneously activates expression of a fluorescent reporter.We have confirmed this new mouse line works reliably with all CreERT2 lines and the 13 floxed genes tested.Given its characteristics, the iSuRe-HadCre allele will be an essential tool for laboratories performing conditional gene function analyses, and particularly those interested in performing single cell genetic studies or epistasis analysis.
All mouse husbandry and experimentation was conducted using protocols approved by local animal ethics committees and authorities (Comunidad Autónoma de Madrid CAM-PROEX 167 / 17, 130 / 19 and 164.8 / 20).The CNIC mouse colony (Mus musculus) is maintained in racks with individual ventilation cages according to current Spanish and European legislation (RD 53 / 2013 and EU Directive 63 / 2010, respectively).Mice have a dust-and pathogen-free bedding, sufficient nesting, and environmental enrichment material for the development of species-specific behavior.All mice have access to food and water 'ad libitum' in environmental conditions of 45-65% relative humidity, temperatures of 21-24 • C, and a 12 h / 12 h light / dark cycle.In addition, animal welfare is preserved through an animal health surveillance program that follows FELASA recommendations for specific pathogenfree facilities.We used Mus musculus with the C57BL6, C57BL6x129SV or C57BL6xDBA2 genetic backgrounds.

DNA constructs, engineering and genome targeting
The basic elements of the different DNA constructs depicted in the figures were obtained from Addgene or by Gene Synthesis and assembled by standard DNA cloning methods.Most sequences were obtained from DNA constructs previously generated in our lab ( 7 , 31 , 32 ).To obtain the gene targeting vector to produce the iSuRe-HadCre allele, we modified an existing plasmid from Addgene (#22799).New sequences were obtained by direct Miniprep DNA or PCR-based cloning with restriction enzymes.The unique restriction sites and the sequential modular cloning strategy were designed and selected using DNAstar SeqBuilder (Lasergene) in concert with our previous experience with commercially available restriction enzymes and PCR kits.The basic elements of the DNA constructs are depicted in several figures with the following abbreviations: ROSA26 (DNA sequences of the PAGE 3 OF 21 mouse ROSA26 locus located in chromosome 6), CAG prom (strong CMV enhancer + B-actin promoter), INS (Insulator), FRT (short DNA sequences recognized by the recombinase Flp or codon-optimized FlpO), Lox (short DNA sequences recognized by the recombinase Cre, LoxP, or LoxN variants, which are mutually incompatible), Int-Cre (Cre-containing intron), ERT2 (estrogen receptor domain that when fused to Cre is inducible / activable by tamoxifen), 2A (2A peptide sequence of the Thosea asigna virus (TaV) used for equimolar expression / translation of upstream and downstream proteins), Int-FlpO-Stop (intron containing FlpO followed by a stop codon), PEST (the Odc1 protein domain, which signals proteins for degradation), H2B (histone 2B, used in fusions to provide for chromatin localization), V5 (unique epitope for immunodetection), WPRE (woodchuck hepatitis virus element that enhances RNA stability and transport, enhancing transgene expression), pA (polyA transcription stop signal), N-PhiM (gene encoding a non-fluorescent protein that is often used as a reporter / marker of promoter expression in the absence of recombination), MbTomato (gene encoding a membrane-tagged tdTomato fluorescent protein that emits in the red spectrum when excited by a yellow laser), and PGK-Neo (selection marker previously used in embryonic stem cell gene targeting approaches).
iSuRe-CrePEST V1 donor DNA was injected together with guide RNA (sequence AATCCA GA GGTTGATTA GCG) into mouse eggs collected from an intercross between male iSuRe-Cre and female DBA / Bl6 mice.The iSuRe-CrePEST V2 and iSuRe-HadCre donor DNAs were injected into mouse eggs together with the guide RNA GC AAC ACGATCCCGCC ACC A. All donor-DNA-plus-guide-RNA injections included standard CRISPR Alt RNA and Cas9 protein (IdTDNA).Several founders were obtained from these injections.After breeding to the C57Bl6 background, the progeny were PCR screened and then crossed with CreERT2 or other mouse lines to confirm proper gene targeting and the expression or function of distinct construct elements.

Immunostaining
For immunostaining of mouse retinas, eyes were dissected from mouse pups and fixed by incubation with agitation for 20 min in 4% PFA in PBS (diluted from a stock of 16% PFA; EMS 15710).After two washes in PBS, retinas were microdissected from the eyes and refixed in 4% PFA for 45 min with agitation.
For rabbit anti-Myc immunostaining, eyes were fixed only 1 h on ice in PFA4% and retinas microdissected.
Dissected retinas were blocked and permeabilized by incubation for 1 h in PBST (0.3% Triton X-100, 3% fetal bovine serum (FBS) and 3% donkey serum in PBS).Samples were then washed twice in PBST and incubated with primary antibodies (see Supplementary Table S3 ) diluted in PBST overnight at 4 • C with agitation.Retinas were then washed 5 × 20 min in PBST diluted 1:2.Samples were then incubated in this same solution for 2 h at room temperature with Alexa-conjugated secondary antibodies.After 3 × 15 min washes in PBS containing 0.15% Triton X-100, retinas were washed 2 × 15 min in PBS and mounted in Fluoromount-G (SouthernBiotech).For combining rabbit anti-ERG-647 with rabbit anti-Myc immunostaining, retinas were incubated with 5% rabbit serum for 30 min at r / t after the Alexa 594 AffiniPure Fab Fragment Donkey Anti-Rabbit secondary antibody, and then incubated with rab-bit anti-ERG-Alexa-647 diluted in 0.15% Triton, 1.5% BSA, 1.5% Rabbit serum overnight at 4ºC, followed by washes and mounting as described above.
For immunostaining of organs cryosections, tissues were fixed for 2 h in 4% PFA in PBS at 4 • C.After three washes in PBS for 10 min each, organs were stored overnight in 30% sucrose (Sigma) in PBS.Organs were then embedded in OCT (Sakura) and frozen at − 80 • C. Cryosections (15 μm) were cut on a cryostat (Leica), washed three times for 10 min each in PBS, and blocked and permeabilized in PBS containing 10% donkey serum (Millipore), 10% fetal bovine serum (FBS) and 1% Triton X-100.Primary antibodies were diluted in blocking / permeabilization buffer and incubated overnight at 4 • C.This step was followed by three 10-min washes in PBS and incubation for 2 h with conjugated secondary antibodies and 4,6-diamidino-2-phenylindole (DAPI) in PBS at room temperature.After three washes in PBS, sections were mounted with Fluoromount-G (SouthernBiotech).All antibodies used are listed in Supplementary Table S3 .
For adult aortas paraffin sections immunostaining with antibodies detecting RBPJ, aortas were fixed in 4% PFA in PBS at 4 • C, overnight.After three washes in PBS for 10 min each, aortas were stored overnight in 70% ethanol and then processed and embedded in paraffin.Paraffin sections (5 μm) were cut on a microtome (Leica), In brief, sections were dewaxed and rehydrated, followed by antigen retrieval: for anti-RBPJ immunostaining, antigen retrieval was carried out in subboiling sodium citrate buffer (10 mM, pH 6.0) for 20 min, and slides were then allowed to cool down to room temperature for 30 min.Sections were then washed twice with PBS and then incubated for 10 min in BloxAll (Vector laboratories SP-6000) to quench endogenous peroxidase activity.Next, slides were washed twice for 5 min each in wash buffer (0.3M NaCl, 0.05M Tris-HCl pH7.5, 0.1% Tween), followed by blocking for 1 h in PBS containing 10% FBS, 0.3% Triton and 10% donkey serum.Sections were then incubated with primary antibody in antibody incubation buffer (5% donkey serum, 1% FBS in wash buffer) overnight at 4 • C.After three washes, slides were incubated at room temperature for 1h with biotinconjugated secondary antibodies anti-rabbit-HRP secondary antibody, and, after washing, sections were incubated another hour with ABC reagent (Vector Laboratories, PK-6100).Sections were then washed and the signal was amplified using the TSA fluorescein kit (NEL701A001KT).Sections were counterstained with DAPI and mounted with Fluoromount-G (South-ernBiotech).

Image acquisition and analysis
Whole mount tissue samples and sections were imaged at high resolution with a Leica SP8 (+ Navigator) or a Stellaris confocal microscope, using 10 ×-63 × objectives for confocal scanning.We acquired individual fields or tiles of large areas.All images shown are representative of results obtained for each group and experiment.For each comparison, all animals were dissected and processed under exactly the same conditions.For comparisons of phenotypes or signal intensities, images were obtained using the same laser excitation and confocal scanner detection settings.Fiji / ImageJ was used to threshold, select, and quantify objects in confocal micrographs.For quantifications, we used manual and automatic ImageJ public plugins and customized Fiji macros.In general, endothelial cell nuclei (ERG+) were segmented via the Threshold / Find maxima functions.For each segmented nucleus (each ROI), we determined reporter (MbTomato or YFP) and marker (p21 or Ki67 or Myc or Rbpj) positivity and defined new ROIs with different assigned objects.Positivity was assigned when the overall ROI pixel intensity was above a defined threshold.Counts for individual cellular reporter-marker combinations were automatically obtained and then confirmed by visual inspection of each segmented ROI.Data were copied and further analyzed in R, Excel or Graphpad.

Isolation of lung fibroblasts from adult mice
To establish primary cell cultures of lung fibroblasts we followed a published protocol ( 33 ), with some modifications.Briefly, lungs were dissected from mice under sterile conditions and placed in sterile PBS.In a cell culture hood, lungs were removed from PBS and then chopped into small fragments with scissors and placed in 10 ml of digestion buffer containing DPBS (Thermofisher 14040141) with Liberase TL (0.14 Wunsch units / ml; Merck 5401020001) and 1 × antibiotic / antimycotic (Thermofisher 15240096).Tissue in digestion buffer was incubated in a water bath at 37ºC for 30min, with mixing every 2-3 min.After this period, the solution was pipetted up and down to break clumps and after 10 ml of culture medium containing DMEM / F12 (Thermofisher 11320033), 15% FBS and 2 × antibiotic / antimycotic solution was added.This cell suspension was centrifuged at 500g for 5 min.The cell pellet was washed two times with culture medium to remove liberase completely.At the end cells were resuspended in 12mL of culture medium and seeded in a p100 dish and incubated at 37ºC, 5% CO 2 , 3% O 2 .Medium was changed after 3-5 days, when fibroblasts had crawled out of tissue fragments.Fibroblasts were then expanded and maintained in full medium at 37ºC, 5% CO2, 3% O 2 for expansion and only transferred to normoxic conditions to carry out experiments.
To induce recombination, cells were trypsinized, resuspended and plated in culture medium containing 0.2 micromolar of 4-OH tamoxifen, for 4 h, and washed after in culture medium.Cells were then trypsinized at different timepoints after induction (24,48,72, 96 and 120 h), stained with DAPI and analysed or sorted in a FACS machine.A minimum of 20000 cells per experimental group were sorted into 300 μl of a solution containing DPBS and 10% FBS.At the end, cells were centrifuged at 500g, for 5min at 4ºC, and supernatants removed.Cell pellets were stored at -80ºC until all samples were ready to be further processed.Genomic DNA was extracted from cells pellets by incubating each pellet in 40 μl of DirectPCR Lysis Reagent (Viagen 301-C) supplemented with proteinase K (0.33 mg / ml) and incubated at 55ºC overnight.Proteinase K was inactivated by incubating samples at 85ºC for 45 min. 1 μl of each sample was directly used for quantitative real time PCR using Taqman universal master mix and an Applied Biosystems QuantStudio5 machine.

Derivation of mouse embryonic stem cells
To derive genetically modified mouse ES cells, we intercrossed mice containing the desired alleles ( iSure-HadCre , with or without UBC-CreERT2 and Cdh5-CreERT2 ), and the CNIC pluripotent cells unit expanded in vitro their blastocysts according to established protocols ( 34 ).Briefly, blastocysts were transferred individually to a 24 well plate containing feeder layers of freshly inactivated MEFs and in ES-2i medium (DMEM / Glutamax, GIBCO 31966-021; NEAA, ßmercaptoethanol, LIF, 20% Serum Replacement, 3 μM CHIR and 1 μM PD).Blastocysts were cultured without disturbance for 3 days.From day 4 the medium was changed every other day, and when each ICM had grown sufficiently, was disaggregated by gently trypsinization and individually passaged to a new 24-well plate.Following several passages independent cell lines were genotyped, expanded and frozen.Some of the ES cell clones containing the desired alleles were treated with 4-OHT (concentration ranging from 0.01 to 1 μM) before or after differentiating them to endothelial cells for 6 days on a OP9 feeder monolayer in basal alpha MEM Media.

Flow cytometry and fluorescence-activated cell sorting
With the exception of the FACS data presented in Figure 1 D (see below), mouse organs were minced and digested to a single-cell suspension by incubation at 37 • C for 30 min with 2.5 mg / ml type I collagenase (Thermofisher), 2.5 mg / ml dispase II (Thermofisher) and 50 ng / ml DNAseI (Roche).Cells were resuspended in 5-10 volumes of FACS buffer (PBS lacking Ca 2+ or Mg 2+ and containing 10% dialyzed FBS (Termofisher)) and filtered through a 70 μm filter to remove nondissociated tissue.Erythroid cells were removed by incubating cell suspensions in blood lysis buffer (0.15 M NH 4 Cl, 0.01 M KHCO 3 and 0.01 M EDTA in distilled water) for 10 min on ice.Cells were washed in FACS buffer, spun at 400g and resuspended in the required volume of buffer for direct FACS analysis (300 μl) or immunostaining (100-200 μl).
For the analysis of endothelial cells (CD31+, CD45-), cells were incubated at 4 • C for 30 min with APC-conjugated rat anti-mouse CD31 (1:200, BD Pharmingen, 551262) and rat anti-mouse CD45-APC-Cy7 (1:200, BD Pharmingen 557659).DAPI (1:1000) was added prior to cell analysis.For the analysis or isolation of cells from dissociated tissues, viable cells were selected by the absence of DAPI fluorescence.All viable cells were interrogated by examining FSC and SSC to select by size and complexity and by comparing FSC-H and FSC-W repeated with SSC-H and SSC-W in order to discern single cells.An additional channel lacking any endogenous or fluorescent label was acquired to detect and exclude autofluorescence.Cells were selected according to their absence of DAPI and the intensity of APC, APC-Cy7, endogenous EYFP and endogenous MbTomato signals.Flow cytometry analysis and FACS were performed with BD Fortessa or BD Aria Cell Sorter.Experimental data were analyzed using Flow JO 10.5.0 or BD FACS DIVA v8.0.1 software.
For the analysis presented in Figure 1 D, skin, lung, and brain samples were dissociated as follows: For ear skin, sacrificed adult mice were perfused with PBS for 5 min to wash out erythrocytes from tissues.Ear skin was isolated, minced with dissection scissors, and digested in PBS containing 0.2 mg / ml DNase I (Roche), 0.2% FBS (Gib-co™), and 5mg / ml collagenase II (Sigma) for 10 min at 37 • C with shaking at 950 rpm.Digestion was stopped by adding 10 μl 0.5M EDTA, and samples were filtered through a 50 μm CellTrics filter (Sysmex).The filter was washed twice with FACS buffer (PBS, 0.5% FBS, 2 mM EDTA), after which cells were pelleted and resuspended in FACS buffer.Dead cells were labeled with sytox blue dead cell stain (Life Technologies), and GFP + cells were passed on a BD LSRFortessa TM Cell Analyzer For the isolation of lung tissue, adult mice were perfused via the right ventricle with PBS for 5 min to wash out erythrocytes from the lungs.The thorax was opened, and the lungs were inflated via the trachea with a 1 ml solution of PBS containing 10 mg / ml dispase II (Sigma) and left to incubate for 5-10 min.The lungs were then removed, placed in a clean petri dish, and injected with 1.0 mg / ml collagenase II (Sigma) and 0.2 mg / ml DNase I (Roche) in PBS and incubated for 5 min.Lungs were transferred to tubes containing 1.5 ml of 1 mg / ml collagenase II and 0.2 mg / ml DNase I in PBS and minced into small pieces using scissors.Tissue samples were further incubated at 37 • C with shaking at 750 rpm for 10 min.Samples were then sheared by repeated pipetting with a 1000 μl pipette and incubated for an additional 5 min.Digestion was quenched by addition of EDTA to a final concentration of 2 mM.The resulting cell suspensions were then filtered through 50 μm CellTrics filters (Sysmex).Filters were washed twice with FACS buffer (PBS, 0.5% FBS, 2 mM EDTA), after which cells were pelleted and resuspended in FACS buffer.Dead cells were labeled with sytox blue dead cell stain, and GFP + cells were passed on a BD LSRFortessa™ Cell Analyzer.Raw data were analyzed with FlowJo software version 10.5.0.
For the dissociation of adult brain tissue, sacrificed mice were perfused with PBS for 5 min to wash out erythrocytes.Brains were digested to a single cell suspension using the Neural Tissue Dissociation Kit (P) (Miltenyi Biotec).Briefly, the brain was removed from the skull and minced with scissors until no large pieces remained.Minced tissue was transferred to a tube containing 5 ml of papain-based enzyme mix and incubated for 17 min at 37 • C with slow rotation.Enzyme mix 2 was added, and the tissue was sheared by repeated pipetting with a 1000 μl pipette and incubated for an additional 12 min.The tissue was then sheared further by passing through a 20 G syringe needle and incubated for another 10 min.The resulting cell suspension was further diluted by addition of 10 ml cold PBS and passed through a nylon 70 μm cell mesh (Corning).Cells were pelleted and resuspended in 500 μl FACS buffer, and GFP + cells were immediately passed on a BD LSRFortessa™ Cell Analyzer.

DNA profiling of FACS-sorted cells
Cells were suspended in PBS with no Ca 2 + or Mg 2 + and containing 10% dialyzed FBS (Termofisher) and were sorted according to YFP, MbTomato, anti-CD31-APC or anti-CD45-APC-Cy7 fluorescent signals.
For DNA isolation, between 5 and 40 thousand cells were obtained per sample.Cells were sorted at a slow flow rate (high purity scale) and with a 100 μm nozzle.Cells were pelleted by spinning at 500g for 5 min and resuspended in 25 μl DirectPCR (Cell) Lysis Reagent for PCR (VIAGEN Cat #301-C) containing a final concentration of 0.4 mg / ml proteinase K.After incubation at 55 • C overnight, the proteinase was inactivated at 85 • C for 45 min.Semi-quantitative and competitive PCR ( Supplementary Figure S6 ) was performed with 1 μl of DNA per sample using the primers listed in Supplementary Table S1 .Groups of 2-4 primers were used per PCR reaction.Some of the PCRs served as DNA input controls (for semi-quantitative PCR), and others targeted the different floxed gene sequences.
To quantify the amount of the different iSuRe-HadCre allele DNA elements before and after recombination ( Supplementary Figure S4 A), we used the online software from Integrated DNA technologies (IdT) to design the probe and primer pairs (listed in Supplementary Table S2 ).Quantitative real-time PCR was performed with 1 μl of DNA using Taqman universal master mix (Applied Biosystems 4304437) and an Applied Biosystems QuantStudio5 machine.

Single cell RNAseq analysis
RNA-seq data can be viewed at the Gene Expression Omnibus (GEO) under accession number GSE245726.Instructions and code to reproduce all scRNA-seq results can be found at https:// doi.org/ 10.5281/ zenodo.11220492 .For scRNA-seq data processing the following pipeline was followed.For alignment and quantification of gene expression the reference transcriptome was built using mouse genome GRCm38 and ensembl gene build version 98 (sep2019.archive.ensembl.org).The Cre, sv40pA, Tomato or WPRE transgenes sequences expressed in the samples were added to the reference.Gene metadata were obtained from the corresponding Ensembl BioMart archive.Reads from hashtags and transcripts were processed, aligned, and quantified using the CellRanger v7.1.0pipeline.Single cell analysis was based on the Seurat R 50 package.Low quality cells were filtered out using the following criteria: total counts > 2000 and below 75 000, genes detected > 500 and below, Mitochondrial transcripts content < 5%, Haemoglobin transcripts < 5%, Hashtag counts > 4 applying a 0.99 positive quantile threshold with the function HTODemux, where doublets were filtered out.Cells were de-multiplexed using the sample hashtag antibody signals (Biolegend).Counts were log-normalized and cells expressing normalized counts of Cdh5 ≤0.9 and WPRE ≤0.5 were removed to select Tomato + expressing ECs for the analysis and remove other contaminant cell types.

Statistical analysis
All bar graphs show mean ± s.d.Experiments were repeated with independent animals.Comparisons between two groups of samples with a Gaussian distribution were by unpaired two-tailed Student t -test.Comparisons among more than two groups were made by one or two-way analysis of variance (ANOVA) followed by multiple comparison tests as indicated in Source Data File 1 .All calculations were done in Excel, and final datapoints were analyzed and represented with Graph-Pad Prism.No randomization or blinding was used, and animals or tissues were selected for analysis based on their genotype, the detected Cre-dependent recombination frequency, and the quality of multiplex immunostaining.Sample sizes were chosen according to the observed statistical variation and published protocols.

Caveats with the first-generation iSuRe-Cre allele
The first generation iSuRe-Cre allele is a transgene located in chromosome 17.Once combined with a CreERT2 expressing allele, recombination can be induced with the ligand tamoxifen, which leads to the permanent equimolar expression of the reporter MbTomato and Cre (Figure 1 A).This allele ensures high Cre activity and deletion of floxed genes in cells expressing the reporter, increasing the efficiency of conditional  genetics in these cells.However, and as reported previously, and detailed further below, there are a number of caveats with the first generation iSuRe-Cre technology.
The first is the inadvertent recombination of the allele in a large fraction of adult myocytes in heart and muscle, a phenomenon we call self-leakiness (Figure 1 B).This self-leakiness arises in the absence of any exogenous Cre or CreERT2 activity and is likely caused by skipping of the transcription stop signals and residual expression of the downstream MbTomato-2A-Cre cassette.This residual expression is likely to occur in cells in which the upstream CAG promoter is strongly transcribed (CMV enhancer element + beta-actin promoter).Myocytes highly express this promoter because of their continuously high production of actin, and this appears to facilitate iSuRe-Cre self-recombination in these cells as animals age ( 7 ).More recently, we observed the phenomenon of non-self-leakiness in other cell types.In this case, the very low level of leaky MbTomato-2A-Cre expression is insufficient to self-recombine the iSuRe-Cre allele, but is enough to recombine other floxed alleles that are more sensitive to Cre / CreERT2 activity, particularly in adult mice (Figure 1 C-E).In addition, when this non-self-leakiness confers a competitive proliferative advantage to the sporadically recombined cells, for instance when using iSuRe-Cre in conjunction with the Rosa26-LSL-Pik3ca H1047R allele, it leads to the development of inadvertent Pik3ca H1047R -driven vascular malformations (Figure 1 F-H and Supplementary Figure S1 A).As expected, self-leakiness and non-self-leakiness are both increased in animals harboring 2 copies of the iSuRe-Cre allele (Figure 1 I).
The second caveat is the relatively low sensitivity of the iSuRe-Cre allele to Cre / CreERT2 activity compared with other Rosa26 -based reporter lines (Figure 1 J), which increases the frequency of cells having recombination of other floxed alleles but missing recombination of the iSuRe-Cre allele, named here as false negatives (Figure 1 K and Supplementary Figure S1 A).This relatively low sensitivity to recombination may be because of the DNA structure or genomic location of the iSuRe-Cre allele.
A third potential caveat with the first generation iSuRe-Cre line is the permanent expression of Cre in reporter-expressing cells.Although this feature is what ensures full genetic deletions and significantly decreases the frequency of false positives (Figure 1 K, L), it also has the potential to generate cell toxicity.Particularly in vitro , Cre is toxic when expressed at high levels or for long periods, presumably due to recombination of native DNA sequences resembling LoxP sites (35)(36)(37)(38).We detected no overt Cre toxicity in cells of iSuRe-Cre mice when it was initially generated on the 129Sv / C57Bl6 background ( 7 ).However, after two generations of breeding to the reference C57Bl6 mouse strain, we began to notice a decrease in the percentage of germline recombined iSuRe-Cre mutants among newborns (Figure 1 M).We also obtained evidence that when the allele was induced in C57Bl6 retinal ECs between postnatal days 1 (P1) and P3, a large fraction of the iSuRe-Cre + cells seen at P6 could not be detected at P28, a phenomenon not seen with other standard Cre-reporters (Figure 1 N).Most of the MbTomato-2A-Cre + cells were outcompeted during vascular development and did not contribute to the capillary network, instead contributing only to more mature vessels (Figure 1 O).This could reflect a decrease in cell fitness or toxicity resulting from the continuously high Cre expression.FACS analysis revealed that iSuRe-Cre expression was 4-8 times higher in postnatal angiogenic retinal and brain ECs than in other ECs ( Supplementary Figure S1 B), which may aggravate this issue specifically in these cells.However, in most of our experiments with mice on other mixed genetic backgrounds (129SvxC57Bl6 or CD1) or at adult stages, we did not detect Cre-toxicity, even when iSuRe-Cre was combined with additional Cre expressing alleles and a very high percentage of cells expressed the iSuRe-Cre allele (Figure 1 P and Supplementary Figure S1 C).
Since the expression of transgenes, including the synthetic CAG promoter used in the iSuRe-Cre allele, varies according to the cell type, organ, stage of tissue development ( Supplementary Figure S1 B), and the mouse genetic background ( 39 ), with this allele it is not possible to ensure the absence of Cre toxicity in all cell types, developmental stages or mouse strains.

Design and characterization of iSuR e-CrePES T alleles
We set out to improve the iSuRe-Cre technology by generating more refined genetic constructs and mouse lines.Our first attempt involved the design of a construct that would be non-leaky in all cell types and that would allow us to induce the expression of a modified Cre with weaker activity (CrePEST).CrePEST is a fusion of Cre with the PEST domain of ornithine decarboxylase ( Odc1 ) to destabilize Cre, a strategy that has been used in Drosophila to bypass Cre toxicity ( 40 ,41 ).We first targeted the iSuRe-CrePEST construct to the same iSuRe-Cre mouse allele on chromosome 17 using CRISPR / Cas9-mediated homologous-dependent repair (Figure 2 A).
To overcome leakiness, we fused Cre to ERT2, so that any skipping of the upstream transcriptional stop signal would result in the expression of CreERT2, and not Cre (compare Figure 2 A iSuRe-Cre to Figure 2 B iSuRe-CrePEST v1 ).In this way, any residual expression of CreERT2 would be trapped in the cytoplasm, which would prevent inadvertent recombination of other floxed alleles (leakiness).This strategy proved to be effective, with animals containing the iSuRe-CrePEST v1 allele showing minimal leakiness in cardiomyocytes and skeletal myocytes (Figure 2 C and Supplementary Figure S2 A).
Recombination of the iSuRe-CrePEST v1 allele by CreERT2 results in a cascade of recombination events (Figure 2 B).Surprisingly, a significant fraction of MbTomato + cells did not recombine other floxed reporter alleles (and were hence false positives) even 6 days after the induction of recombination and expression of the iSuRe-CrePEST v1 allele (Figure 2 D and Supplementary Figure S2 B).This low recombination efficiency could be due either to weak expression or activity of FlpO in the first step of the cascade (after the first Cre-ERT2 dependent LoxP recombination) or to weak CrePEST activity after FlpO recombination.Analysis of the expression of H2B-V5, a marker of cells that have undergone FlpO recombination (see Figure 2 B genetic cascade), showed that only a small fraction of MbTomato + cells expressed H2B-V5 (Figure 2 E, F).This demonstrates that the intermediate cascade step of FlpO-dependent recombination, necessary for full construct activation, does not occur in a significant fraction of MbTomato+ cells, even 4 days after induction with 4-OH tamoxifen (4-OHT), possibly due to relatively weak expression and activity of FlpO driven by the recombined iSuRe-CrePEST v1 allele.We also noticed that whereas cells with a strong MbTomato+ signal co-expressed H2B-V5, cells weakly expressing MbTomato were H2B-V5 negative (Figure 2 E, G).Importantly, the frequency of Rosa26-Lox-Stop-Lox-EYFP allele recombination was much higher in the few H2B-V5 + cells than in H2B-V5 -cells (Figure 2 E, H, I), suggesting that CrePEST expression is sufficient to fully recombine other floxed alleles.Overall, these results show that the iSuRe-CrePEST v1 allele is unreliable because it generates many false positives.The sensitivity to CreERT2 recombination of the iSuRe-CrePEST v1 allele is also significantly lower than that of other Rosa 26 -based reporter alleles (Figure 2 E, H).
We previously observed that Rosa26 reporter alleles, particularly the Rosa26-Lox-Stop-Lox-TdTomato (Ai14) allele ( 17 ), are more sensitive to CreERT2 / tamoxifen-induced recombination than the iSuRe-Cre allele (Figure 1 J).Since the genetic distance between LoxP sites is similar in the iSuRe-Cre allele (1.1kb) and the Rosa26-Lox-Stop-Lox-TdTomato (Ai14) allele (0.9kb), we hypothesized that the higher sensitivity of the latter to CreERT2 recombination was due either to its different genomic location or to the DNA sequences included in its floxed transcription stop cassette (LoxP-3xpA-LoxP).Therefore, we assembled a Rosa26 -targeting construct containing exactly the same C A G-LoxP-3xpA-LoxP cassette followed by a conditional Cre-LoxN-ERT2-stopCodon-LoxN-PEST-2A-MbTomato cassette (Figure 3 A).This strategy relies on the incompatibility between LoxN sites and LoxP sites, such that the native Cre sequence will still be retained after the two initiating Cre / CreERT2-dependent deletion events (between the 2 LoxP sites and the 2 LoxN sites).Analysis of the relative recombination frequency between this new Rosa26-iSuRe-CrePEST v2 allele and the internal reference Rosa26-LSL-YFP allele revealed that it is around 8 times more sensitive to CreERT2 activity at 4 mg / kg tamoxifen than the previous iSuRe-Cre allele (Figure 3 B), despite the presence of two independent LoxP and LoxN flanked cassettes.This higher sensitivity significantly reduces the occurrence of false negatives.We also confirmed the Rosa26-iSuRe-CrePEST v2 allele is not leaky (Figure 3 C).
We next examined the potential for toxicity arising from permanent CrePEST expression driven by the strong Rosa26-C A G promoter (Figure 3 A), particularly in highly proliferative or migratory cells that we previously found to be sensitive to the high Cre levels driven by the iSuRe-Cre allele (Figure 1 M-O).Unfortunately, expression of MbTomato-2A-CrePEST in retinal ECs for 4 days significantly decreased the frequency of these cells relative to EYFP + cells (Figure 3 E), indicating cumulative toxicity linked to high CrePEST expression.Consistent with this finding, induction of the Rosa26-iSuRe-CrePEST v2 allele at high frequency significantly compromised retinal angiogenesis (Figure 3 F).This outcome was associated with decreased proliferation (Ki67) and increased expression of the cell-cycle inhibitor and senescence marker p21 (Figure 3 G, H).We also crossed these mice with C57Bl6 Sox2-Cre and Tie2-Cre mice.Permanent expression of the induced Rosa26-iSuRe-CrePEST v2 allele in all cells of the mouse embryo resulted in 100% embryonic lethality, being the Rosa26-iSuRe-CrePEST v2 allele more toxic than the iSuRe-Cre allele (compare Figure 3 I and J with Figure 1 M and 1 P).This significantly higher toxicity of the Rosa26-iSuRe-CrePEST v2 allele is likely due to its much stronger promoter activity ( Supplementary Figure S3 ).
Taken together, these results show that the Rosa26-iSuRe-CrePEST v2 allele is more sensitive to CreERT2 and is not leaky, however, the permanent and high expression of CrePEST is also toxic to mammalian cells and detrimental to tissue development.

Design and validation of the iSuRe-HadCre allele
Given the results obtained, we surmised that we had to develop a new construct enabling high and transient Cre expression.To achieve this, we assembled the iSuRe-HadCre construct and targeted it to the ROSA26 locus (Figure 4 A).In mice with this allele, tamoxifen-induced CreERT2 expression triggers a sequential genetic cascade in which Cre and FlpO are first transiently expressed at equimolar levels, given the viral 2A peptide used ( 42 ), followed by an inbuilt FlpOdependent switch-off mechanism that simultaneously selfdeletes both the Cre and FlpO genes and activates expression of the reporter MbTomato (Figure 4 A).We selected the FlpO recombinase and FRT sites due to their incompatibility with the Cre / CreERT2 system and their significantly lower recombination frequency in mammalian cells at 37 • C ( 43 ).With this approach, we reasoned that Cre would have sufficient time to execute its action on other floxed genes before FlpO accumulated to the level needed to recombine and excise the large (2.8 kb) FRT-flanked cassette (Figure 4 Aii and Aiii) from the Rosa26-iSuRe-HadCre allele.
To test the effectiveness of the iSuRe-HadCre design, we first performed a temporal FACS and histology analysis of the sequential genetic cascade, examining tissues collected at 24-120 h post-induction.The combination of the 4-OHT-CreERT2 activation and high Cre expression resulted in immediate recombination and expression of the Rosa26-Lox-Stop-Lox-EYFP allele, being the protein fluorescence detectable in most cells at 24-48 h postinduction (Figure 4 B-D).In contrast, due to the lag in the induction of the weaker FlpO recombinase activity ( 43 ), and the subsequently slower FlpO-dependent genetic recombination, most cells expressed the iSuRe-HadCre MbTomato reporter only at 48-72 h post induction.This could be confirmed experimentally, since Cre and FlpO recombine the iSuRe-HadCre allele immediately in the first 24-48 h ( Supplementary Figure S4 A), while the Tomato reporter shows only at 48-72 h (Figure 4 B-D).We predicted that this lag in the FlpO expression and recombination-dependent switch off activity on the iSuRe-HadCre allele, would be sufficient to express enough Cre and ensure efficient Cre-mediated deletion of other floxed genes during this time period.Indeed, > 95% of MbTomato + cells were also YFP + at all stages and tissues analysed (Figure 4 C, D), demonstrating the effectiveness of recombination in cells with transient Cre expression driven by the iSuRe-HadCre allele.Further analysis of this is presented in the next section.
We next compared the sensitivities to CreERT2 of the iSuRe-HadCre and iSuRe-Cre lines.To have an internal control, we used animals also having the Rosa26-Lox-Stop-Lox-EYFP allele and obtained the relative recombination data (MbTomato:YFP recombination log ratio).This analysis showed that sensitivity to CreERT2 and the relative recombination of reporter alleles vary within the same cell type (ECs) and across developmental stages and organs, being the Rosa26-iSuRe-HadCre allele superior to the iSuRe-Cre allele (Figure 4 E and Supplementary Figure S4 B), and in some cases also superior to that of the Rosa26-Lox-Stop-Lox-EYFP allele (MbTomato:YFP recombination log ratio > 0 in Figure 4 E), particularly when two tamoxifen doses were used.The Rosa26-iSuRe-HadCre allele has two floxed cassettes (flanked by the wildtype LoxP sites and the mutant LoxN sites).Af-ter recombination of the LoxP-flanked cassette, high CreERT2 expression is induced (Figure 4 A, i), and this significantly enhances the recombination of the more difficult to recombine LoxN cassette ( Supplementary Figure S4 A) after a second tamoxifen dose (Figure 4 E and F).Besides being more sensitive to Cdh5-CreERT2 allele induction, the Rosa26-iSuRe-HadCre allele was also more sensitive to Myh11-CreERT2 induction in smooth muscle cells (SMCs, Figure 4 G).By using three high-dose (60 mg / kg each) tamoxifen injections, we also confirmed that the intermediate FlpO-dependent recombination is not evaded in quiescent adult tissues, which are usually more difficult to recombine (Figure 4 H and Supplementary Figure S4 C).
In addition, we found the expression and signal intensity of the reporter expressed by the Rosa26-iSuRe-HadCre allele to be significantly higher which enhances its detection by direct fluorescence imaging, FACS or scRNAseq (Figure 5 A and Supplementary Figure S5 A, S5 B).Moreover, the Rosa26-iSuRe-HadCre was not leaky in any cell type or organ analyzed (Figure 5 B, C and Supplementary Figure S5 C).

iSuRe-HadCre significantly increases the efficiency of Cre-genetics in all cell types
A key feature of the iSuRe-Cre technology is the very high percentage of true positives (cells expressing a fluorescent reporter that also have recombination of other floxed alleles), which results from the permanent co-expression of Cre and the reporter.In contrast, reporter-expressing cells in iSuRe-HadCre mice express Cre only transiently.To check for the efficiency of recombination in cells expressing the iSuRe-HadCre allele reporter, we intercrossed several reporters and the iSuRe-Cre allele with the Cdh5-CreERT2 allele and analysed embryos, postnatal and adult mice.Remarkably, the frequency of Rosa26-Lox-Stop-Lox-EYFP reporter allele recombination was very high in MbTomato+ cells of the iSuRe-HadCre line, much higher than in cells expressing other classical reporters and similar to cells permanently expressing Cre driven by iSuRe-Cre (Figure 5 D-F).We also intercrossed the iSuRe-HadCre allele with the TgUBC-CreERT2 allele ( 11 ), that is ubiquitously expressed and recombines most cell types in different organs.A single injection of 4-OHT showed that the iSuRe-HadCre allele also enables very efficient recombination of the Rosa26-Lox-Stop-Lox-EYFP reporter within MbTomato+ cells, for all cell types and organs analysed (Figure 5 G and Supplementary Figure S5 D).The average rate of true positives among all the organs analysed was 95%, similar to the first generation iSuRe-Cre ( 7 ), and significantly higher than the 20-30% obtained with standard Cre-reporters (

iSuRe-HadCre enables complete gene deletion in single cells or whole tissues
To confirm the utility and broad applicability of the iSuRe-HadCre allele for reliable single cell or whole tissue loss-offunction genetics, we analysed deletion efficiency for a total of 13 different floxed genes, with a genetic distance between LoxP sites ranging from 1.5 to 4.8 kb ( Supplementary Figure S6 A).Recombination efficiency correlates negatively with the genetic distance between LoxP sites, but is also strongly influenced by the gene locus and its DNA sequence ( 6-8 ), being highly variable.In contrast to reporter alleles, a full gene deletion requires the deletion of two floxed alleles.We initially used the standard Rosa26-Lox-Stop-Lox-EYFP reporter allele, containing 2.2kb between LoxP sites, as an internal comparative reference.The Rosa26-iSuRe-HadCre and Rosa26-Lox-Stop-Lox-EYFP alleles were independently combined with the Notch1 floxed (3.5 kb between LoxP sites for each allele) and Cdh5-CreERT2 alleles, a single high 4-OHT dose was injected at P1, and retinas were collected at P6.
Notch1 is an essential gene for the formation of arteries during vascular development ( 32 ), and decreased Dll4-Notch signaling induces higher retinal vessel density ( 44 ).Nevertheless, wildtype and Notch1 floxed mutants carrying the Rosa26-Lox-Stop-Lox-EYFP allele had similar percentages of YFP + arterial cells, artery length, or vessel density (Figure 6 A-C), which could suggest a lack of NOTCH1 function in these processes, or a lack of Notch1 deletion in YFP + cells.Importantly, in animals carrying the Rosa26-iSuRe-HadCre allele, only ∼1.5% of arterial cells expressed the MbTomato reporter, despite a very high frequency of MbTomato + cells in neighbouring capillaries and veins (Figure 6 D-F).These data indicate that the efficiency of Notch1 floxed deletion is very high only in animals and cells expressing the Rosa26-iSuRe-HadCre allele.This allowed us for the first time to uncover the full extent of the retinal Notch1 loss-of-function phenotype.The observed increase in vascular density after deleting Notch1 (Figure 6 E) is significantly more pronounced than published before ( 45 ,46 ), suggesting incomplete Notch1 deletion in all previous studies.
We also quantified the recombination efficiency in single cells of retinas exposed to a low dose of tamoxifen.Of the many ECs expressing the iSuRe-HadCre allele, all recombined the Rosa26-Lox-Stop-Lox-EYFP allele and none formed arteries, whereas many cells with recombination of only the Rosa26-Lox-Stop-Lox-EYFP reporter allele formed arteries, which were thus all false positives since they did not delete Notch1 (Figure 6 G).Single-cell RNAseq analysis of cells isolated by FACS confirmed the very efficient deletion of Notch1 , and loss of its mRNA, in ECs collected 3 days after 4-OHT injection (Figure 6 H).Overall, this data demonstrates that gene deletion with iSuRe-HadCre occurs quickly and is highly efficient, similar to iSuRe-Cre .
Next, we performed a similar analysis in mice carrying Rbpj floxed alleles.Rbpj is another gene essential for arterial development, and the very low percentage of MbTomato + cells found in arteries confirmed that this gene was also very efficiently deleted in the large majority of cells recombining the iSuRe-HadCre allele (Figure 6 I).Semi-quantitative PCR analysis in bulk FACS-sorted cells from postnatal and adult animals revealed that the iSuRe-HadCre allele enabled also highly efficient simultaneous deletion of Notc h1, Notc h2, and Notch3 (6 floxed alleles) or of Rbpj and Myc in cells of different organs ( Supplementary Figure S6 B-H).Note that a small percentage of single cell events isolated by FACS will have contaminant wildtype cell RNA and DNA, as shown before ( 47 ), and these retain the non-recombined floxed genes.
One of the most important genes for angiogenesis is Kdr / Vegfr2, that encodes for the most important VEGF receptor (VEGFR2).We have previously observed that is very difficult to delete this gene with the Cdh5-CreERT2 allele during angiogenesis, presumably due to the floxed allele structure, but also due to the loss of the Vegfr2 mutant ECs during angiogenesis, that are outcompeted by non-mutant cells over time ( 7 ,45 ).The iSuRe-HadCre allele allowed us to delete Vegfr2 very effectively during retina angiogenesis, and in this way assess the real impact of fully deleting Vegfr2 on vascular development (Figure 7 A, B).Even when the induced recombination was very low, single iSuRe-HadCre / MbTomato + cells had deletion of VEGFR2, since they were excluded from the angiogenic front (Figure 7 C, D).Immunostaining   analysis confirmed the loss of VEGFR2 expression in iSuRe-HadCre / MbTomato+ ECs, but not in adjacent reporter negative cells (Figure 7 E).Efficiency of deletion was also very high in iSuRe-HadCre / MbTomato+ adult liver ECs (Figure 7 F).These are known to be very sensitive to the loss of VEGF / VEGFR2 signalling ( 48 ,49 ).When the iSuRe-HadCre allele was induced in mosaic fashion, most Vegfr2 mutant (MbTomato+) ECs were depleted in the adult liver capillaries, and survived only in larger portal and central vessels (Figure 7 F lower panel).In mosaic animals, the Cdh5-CreERT2+ ECs that did not undergo iSuRe-HadCre recombination, failed to delete Vegfr2 , and formed most adult liver capillaries.Deletion of Vegfr2 in lymphatic ECs with Prox1-CreERT2 was also very effective in iSuRe-HadCre / Tomato + cells, at both high and low (mosaic) recombination rates (Figure 7 G and Supplementary Figure S7 ).
To further demonstrate the efficiency of the iSuRe-HadCre allele, we determined the rate of genetic deletion and performed phenotypic analysis of postnatal and adult mice carrying multiple other floxed genes.Immunostaining and confocal analysis revealed a very effective deletion of all these genes in iSuRe-HadCre expressing cells, but not in cells lacking the expression of this allele (Figure 8 ).Simultaneously deletion of three Foxo genes, shown before to partially compensate for each other ( 23 ), is very efficient with iSuRe-HadCre after a single injection of 4-OHT (Figure 8 A, left).The analysis of genetic mosaics showed that many of the Cdh5-CreERT2+ ECs failed to delete these 3 Foxo genes.On the other hand, iSuRe-HadCre / MbTomato+ ECs effectively deleted Foxos and did not form tip cells when surrounded by wildtype cells (Figure 8 A, right).Jagged1 and Dll4 were also effectively deleted in animals containing the iSuRe-HadCre allele after a single dose of 4-OHT (Figure 8 B, C), unlike the three doses of tamoxifen (P1, P2, P3) used before to delete these genes ( 50 ).Dll4 loss induced vascular hyperplasia, associated to cell-cycle arrest (Ki67-), increased filopodia (Figure 8 D), and the upregulation of the tip cell marker Esm1, whereas Jagged1 loss strongly inhibited angiogenesis, as shown previously ( 44 ,50 ).Flt1 / Vegfr1 was also only effectively deleted in iSuRe-HadCre / MbTomato+ cells (Figure 8 E, F).We found that for some genes, like Myc and Mycn , their protein is detectable in only around 29% of wildtype ECs during angiogenesis, being impossible to determine where are the mutant cells without the use of iSuRe-HadCre (Figure 8 G).Simultaneous deletion of these two genes, that correspond to deletion of 4 alleles containing 4.6kb and 4.8kb of DNA between LoxP sites ( Supplementary Figure S6 A), was very effective only in iSuRe-HadCre / MbTomato + cells (Figure 8 G).This data shows how the iSuRe-HadCre allele can be very relevant for multiple gene deletions in whole tissues or single cells, a method essential for epistasis analysis in mice.
We also tested the efficiency of iSuRe-HadCre with the Myh11-CreERT2 line that is expressed in SMCs.iSuRe-HadCre + cells effectively deleted Rbpj in SMCs (Figure 8 H).
Overall, these data confirm that when cells recombine and express the iSuRe-HadCre allele, even after a single and low tamoxifen pulse, they express enough Cre to efficiently recombine other floxed genes.In contrast, conditional genetics without using this allele is blind, even when using standard reporter alleles, because cells often do not recombine the intended floxed genes, thus generating many false positives that prevent the accurate determination of a gene function in single cells or entire tissues.

Transient Cre expression driven by iSuRe-HadCre is non-toxic
We assessed toxicity induced by expression of the Rosa26-iSuRe-HadCre, Rosa26-iSuRe-CrePEST v2 and iSuRe-Cre alleles in mice carrying these alleles on the C57Bl6 background.Only animals containing the Rosa26-iSuRe-HadCre allele survived upon recombination with both the Tie2-Cre allele (blood cells and ECs) and the Sox2-Cre allele (all embryo cells).These expressed the allele in all target cells (Figure 9 A, B), confirming the lack of toxicity in cells with iSuRe-HadCre recombination and expression.
Given its higher sensitivity and recombination efficiency, the Rosa26-iSuRe-HadCre allele can also be used to significantly reduce the cellular toxicity associated with the consecutive tamoxifen injections usually required for conditional genetics analysis.In line with previous studies describing cell toxicity and impaired angiogenesis when high doses of tamoxifen are delivered to activate CreERT2 ( 51 ,52 ), we found that 4-OHT induces p21+ arrested / senescent ECs when animals express CreERT2 and display high rates of recombination of the Rosa26-Lox-Stop-Lox-EYFP reporter (Figure 9 C, D).To more precisely determine the potential short-term cell toxicity arising just after iSuRe-HadCre allele activation, we compared the frequency of p21 expression in cells recombining the Rosa26-Lox-Stop-Lox-EYFP or Rosa26-iSuRe-HadCre alleles 24 and 48 h after tamoxifen injection.During this period, cells recombining the iSuRe-HadCre allele will have strong but transient Cre expression, whereas cells exclusively recombining the Rosa26-Lox-Stop-Lox-EYFP allele will only have tamoxifen-induced CreERT2.The additional transient Cre expression driven by the Rosa26-iSuRe-HadCre allele had no significant impact on p21 expression at 24 or 48 h (Figure 9 E, F), despite the very efficient deletion of floxed alleles after a single injection of tamoxifen (Figures 6 -8 ).This data suggest that the toxicity induced by tamoxifen in cells expressing CreERT2 is independent of canonical or productive Cre recombination.This is in line with previous reports showing that tamoxifen administration to animals expressing CreERT2 is more toxic than Cre expression itself ( 38 ,51 ).Given that the Rosa26-iSuRe-HadCre allele increases the efficiency of gene deletion without increasing cellular toxicity, it can be used to efficiently delete genes even at lower doses of tamoxifen, and in this way prevent toxicity associated to the use of high doses of tamoxifen to induce CreERT2-dependent floxed gene deletions.

Discussion
Conditional Cre genetics has been used for decades and remain the gold standard for the analysis of gene function ( 1 , 3 , 39 ).Recent CRISPR / Cas9-based gene targeting approaches do not match it for genetic deletion efficiency and precision, especially in inducible genetic studies.However, despite the clear advantages of the Cre / Lox system, its temporal control and efficiency varies greatly with different Cre transgenes and floxed alleles ( 2 , 5 , 6 , 53 ), and this significantly decreases its reliability, particularly when applied to mosaic or single-cell genetics analyses that require lower or incomplete CreERT2 activity.
iSuRe-Cre was the first CreERT2-inducible, dual reporter-Cre-expressing mouse allele that overcame the problem of false positives in conditional Cre genetics ( 7 ).However, this first-generation transgene also had several caveats, including leakiness, relatively low sensitivity to Cre / CreERT2 activity, and the risk of toxicity in cells with high and permanent Cre expression ( 7 ).We found now that this toxicity varies across cell types, developmental stages, and mouse genetic backgrounds, as reviewed recently ( 38 ).All of these variables are difficult to control and predict when using different mouse models containing the iSuRe-Cre allele, and can impact the final phenotype of the analyzed cells.
To overcome these caveats, we set out to develop and characterize several new mouse lines, at each step encountering problems with the improved designs.The iSuRe-CrePEST v1 line overcame the leakiness of iSuRe-Cre , particularly in myocytes, but suffered from the same low sensitivity to CreERT2 recombination and did not provide certainty that reporterexpressing cells had also recombined other floxed alleles.The Rosa26-iSuRe-CrePEST v2 allele was non-leaky and significantly more sensitive to CreERT2 because of its location in the Rosa26 locus, but the CAG-promoter expression from this locus was too high, and the very high and permanent coexpression of the reporter and CrePEST still induced significant toxicity in most tested cell types.
These unexpected results prompted us to design a similar Rosa26-targeting construct that would maintain nonleakiness and the high sensitivity to CreERT2 induction, but that at the same time would only support the induction of transient, but sufficient, expression of Cre.The solution we devised was to engineer a construct enabling the co-expression of Cre at equimolar levels with the much weaker recombinase FlpO (43), which would, after a delay, switch off the system and turn on the expression of a fluorescent reporter.In this way, the reporter would label cells that definitely had, but no longer have, high Cre (and FlpO) expression, as indicated by the name, iSuRe-HadCre .This new design and strategy proved to be effective and overcame all the problems we encountered with the previous models.In contrast with its predecessors, the iSuRe-HadCre allele is located in the ubiquitously expressed Rosa26 locus, shows no leakiness, is highly sensitive to CreERT2 activity, supports higher reporter expression, has no Cre-related toxicity, and still ensures a very high rate of floxed gene deletion (true positives).Overall, we tested the efficiency of iSure-HadCre with UBC-CreERT2, Cdh5-CreER T2, Prox1-CreER T2 and Myh11-CreERT that recombine most cell types, and with 13 different floxed genes.Deletion of all these genes was very effective with iSuRe-HadCre , even when multiple genes containing very large floxed cassettes (up to 4.8 kb) had to be deleted simultaneously.This is essential to perform reliable conditional functional genetics, particularly genetic epistasis analysis.
Is known that different floxed genes have very different sensitivities to Cre / CreERT2 activity, and their recombination rate may also vary with the cellular stage or differentiation state.Even the same gene may be differentially recombined depending on the cell type and the chromatin conformation of its locus.Genes not expressed in a given cell type and present in heterochromatin regions, may be also more difficult to delete by Cre, but they are in principle also less relevant to conditionally delete.In this work, we analysed multiple genes deletion rates across several cellular stages (embryonic, neonatal and adult) and in different cell types, and have not found significant differences in the efficiency of iSuRe-HadCre .Future work may address if higher or permanent Cre expression is necessary to fully delete some other floxed genes, at the cost of potentially causing cellular toxicity, as observed with the first generation iSuRe-Cre .
New tools that enhance the reliability of conditional genetics are gaining importance as technology moves towards single-cell mosaic genetics, which, when applied in situ , requires low tamoxifen doses or mosaic CreERT2 induction in relatively few cells.We have shown that the iSuRe-HadCre allele guarantees a very high rate of true positives in single induced cells.This allele allowed us to confirm how single cells losing Notch1, Rbpj, Kdr and Foxo1 / 3 / 4 change their mobilization or differentiation behaviour across a growing vascular bed.Standard reporters and tools did not enable the induction and identification of cells with deletion of these genes, preventing determination of their important role in these processes.
Single-cell technologies are now standard for profiling control and mutant cells ex situ .Given that current single-cell sequencing technologies only detect the mRNA of some genes in single cells, and that most floxed genes conserve their nonfloxed DNA and mRNA sequences after deletion, this often generates profiles of pseudomutant cells whose true genetic status cannot be determined.Of the 13 floxed genes analysed, 7 conserve their 3´mRNA after Cre-deletion and therefore their deletion cannot be determined by scRNAseq.Some antibodies we used also failed to detect the partial protein deletion by immunostaining given that they detect the protein upstream or downstream of the deletion.We have provided multiple evidence here that many of the CreERT2 expressing cells do not have deletion of the intended floxed genes after tamoxifen, indicating that the iSuRe-HadCre allele is necessary for the reliable isolation of bona fide mutant cells for reliable single-cell functional omics.
In addition to providing a new mouse model, we also derived embryonic stem cells from iSuRe-HadCre mice with or without the UBC-CreERT2 and Cdh5-CreERT2 alleles ( Supplementary Figure S8 ).These can be easily modified by CRISPR / Cas9 to include other floxed alleles ( 54 ), and in this way perform reliable single cell conditional genetics in all differentiated cell lineages in vitro .The novel iSuRe-HadCre recombinase genetic cascade ensuring high transient Cre activity without toxicity can also be engineered in other cell lines or animal models.Given its ease of use, high efficiency, and lack of toxicity, we believe the iSuRe-HadCre allele will be essential for laboratories conducting conditional gene-function analysis with the Cre / lox system.

PAGE 6 OF 21 NucleicFigure 1 .
Figure 1.Ca v eats of the iSuR e-Cre allele.( A ) Simplified schematic of the published iSuR e-Cre allele.After combining it with an y other CreER T2 allele, is possible to induce its recombination with tamoxifen and analyse cells with permanent expression of the reporter MbTomato and Cre. ( B ) FACS analysis of iSuRe-Cre self-leakiness in organs from adult (8 weeks old) and postnatal mice (postnatal day 7).Note that major leakiness occurs only in adult mice.( C ) FACS analysis of iSuRe-Cre self-leakiness (MbTomato + cells) versus non-self-leakiness with the Rosa26-Lox -Stop-Lox -EYFP allele (YFP + cells).In general, non-self-leakiness is higher than self-leakiness.( D ) FACS analysis of iSuRe-Cre self-leakiness (MbTomato + cells in red bar) versus non-self-leakiness with the R osa26-L o x-MbTomato-L o x-MbEGFP allele (GFP + cells in animals with or without iSuRe-Cre).( E ) Confocal micrographs of adult mouse tissues, showing inadvertent recombination in a smooth muscle cell (SMC) and EC. ( F-H) Images of vascular lesions on snout ( F ), brain ( G ) and ear skin ( H ) of uninduced adult iSuRe-Cre Rosa26-Lox -Stop-Lox -Pik3ca H1047R mice.( I ) Stereomicroscopy images of leg muscle and heart tissue showing endogenous MbTomato expression due to self-leakiness in iSuRe-Cre mice, higher in animals with 2 copies (ki / ki) of the allele.( J ) FACS analysis of the sensitivity of the indicated alleles to CreERT2-dependent recombination.( K ) FACS analysis of the frequency of false positives (MbTomato or iSuRe+,YFP -), true positives (iSuRe + ,YFP + ), and false negatives (YFP + , iSuRe -) in liver ECs after the administration of different tamoxifen doses to mice carrying the indicated alleles.( L ) FACS analysis of the frequency of false and true positives among iSuRe-Cre / Tomato + cells in postnatal day 7 liver ECs. ( M ) Expected and obtained survival ratios of animals containing the germline recombined iSuRe-Cre allele (originally on the G4 or C57Bl6x129 background) across generations of breeding with the C57Bl6 strain.( N ) Confocal micrographs used to compare the frequency of cells positive for MbTomato-2A-Cre+ (from the iSuRe-Cre allele) and for GFP+ (from the Rosa26-Lox -mT-Lox -mG allele) in retinas of animals in the C57Bl6 background.The drop in the frequency of MbTomato-2A-Cre + cells from P6 to P28 indicates Cre toxicity in ECs during the development of retinal vessels.( O ) R epresentativ e conf ocal micrograph sho wing that when animals containing the three indicated alleles are in the C57Bl6 back ground, MbTomato / iSure-Cre + cells are outcompeted by EYFP + cells during retinal vascular development.( P ) Expected and obtained survival ratios for mice containing the iSuRe-Cre and Tie2-Cre alleles in the mixed C57Bl6 x CD1 genetic background.These mice have similar frequencies of MbTomato + blood and ECs (iSuRe-Cre + or R26-LSL-TdTomato + ), suggesting no toxicity in this case.Confocal micrographs to the right show representative variable recombination and expression in two retinas of the same animal, with no major differences in angiogenesis (ECs are isolectinB4 + ).Data are presented

Figure 2 .
Figure 2. Characterization of the iSuR e-CrePES T v1 allele.( A ) Crispr-Cas9 genetic targeting of the preexisting iSuRe-Cre allele, present in Chromosome 17, with the indicated donor DNA to generate the new iSuRe-CrePEST v1 allele shown below.( B ) Genetic cascade and expression outcomes after induction of the iSuR e-CrePES T v1 allele with CreERT2 + tamoxifen.After the first recombination event induced by the activation of CreERT2 with tamoxifen, cells will express MbTomato, CreERT2 and FlpO.In a second step, MbTomato + cells with high FlpO activity will undergo genetic fusion of the PEST domain to Cre and co-express equimolar amounts of MbTomato, CrePEST, and H2B-V5.CrePEST activity is higher than CreERT2 but, due to its higher degradation, presumably lo w er than Cre activity.Other abbreviations and construct elements are detailed in the DNA engineering section in Methods.( C ) FACS analysis of MbTomato + cells in iSuRe-Cre and iSuRe-CrePEST v1 animals reveals significantly less self-leakiness in the latter.( D ) FACS analysis and charts showing the frequency of false positives (MbTomato + , YFP -) and true positives (MbTomato + , YFP + ) among MbTomato + cells from se v eral adult organs of animals containing the Cdh5-CreERT2, iSuRe-CrePEST v1 and Rosa26-LSL-YFP alleles (yellow bars) or the Cdh5-CreERT2, iSuRe-Cre , and Rosa26-LSL-YFP alleles (magenta bars).Recombination efficiency (true positives frequency) is lower in mice containing the iSuRe-CrePEST v1 allele.( E ) P6 retina confocal micrographs showing the efficiency of the recombination cascade (MbTomato activation followed by H2B-V5) of the iSuRe-CrePEST v1 allele (induced with Cdh5-CreERT2 and tamoxifen at P1). Red arrows indicate cells with only MbTomato expression (false positives); yellow arrows, cells with MbTomato and EYFP expression (partial recombination of the iSuRe-CrePEST v1 allele); and white arrows, cells with expression of MbTomato, EYFP, and H2B-V5 (true positives).( F ) Chart showing the low efficiency (% of V5+ cells) of the second, FlpO-mediated, iSuRe-CrePEST v1 recombination event in MbTomato+ cells (these were initially induced by CreERT2 at P1). ( G ) Quantification of MbTomato reporter signal intensity (from confocal images) in cells that ha v e undergone only the first recombination e v ent (H2B-V5 -) or all recombination e v ents (H2B-V5 + ).Each dot is one cell intensity.( H ) Frequency of cells expressing the different reporters combinations (EYFP, MbTomato, and H2B-V5) among all recombined cells in P6 retinas of mice containing the alleles iSuR e-CrePES T v1 , R osa26-LSL-YFP and Cdh5-CreER T2.Note the v ery lo w frequency of MbTomato + , H2B-V5 + cells, but that most of these are YFP + (true positives).( I ) Frequency of false positives (YFP -) and true positives (YFP + ) among MbTomato -cells (green bar) or MbTomato + , H2B-V5 -cells (red bars) or MbTomato + , H2B-V5 + cells (orange bars), re v ealing that only H2B-V5 + cells accurately indicate cells with high Cre activity, thanks to the permanent CrePEST expression.Data are presented as mean values ± s.d.For statistics, see Source Data File 1 .Scale bar, 200 μm.

Figure 3 .
Figure 3. Characterization of the R osa26-iSuR e-CrePES T v2 allele.( A ) Schematic of the R osa26-iSuR e-CrePES Tv2 allele.Tamoxifen-induced CreERT2 activity leads to deletion of the stop cassette containing three Sv40 polyA sequence repeats, triggering co-expression of CrePEST and the reporter MbTomato.( B ) Confocal microscopy analysis of the sensitivity to CreERT2-induced recombination of the Rosa26-iSuRe-CrePEST v2 and iSuRe-Cre alleles (MbTomato + ) relative to the internal control Rosa26-LSL-EYFP allele (each dot represents one retina sample).( C ) FACS analysis of the self-leakiness frequency (MbTomato + cells) in adult organs of iSuRe-Cre versus Rosa26-iSuRe-CrePEST v2 animals (each dot represents a measure in one animal).( D ) Quantification of confocal micrographs reflecting the relative frequency of MbTomato+ cells ( Rosa26-iSuRe-CrePEST v2 ) and YFP + cells ( Rosa26-LSL-EYFP ) 24 and 96 h after tamoxifen induction at P4.Most MbTomato + cells are already YFP + (true positives, yellow arrows) at 24 h postinduction.Very few are MbTomato + and YFP -(false positives, red arrow) at 24 h, and almost none later.( E ) MbTomato-2A-CrePEST + cells become 6 times less frequent 4 da y s after induction, indicating toxicity of CrePEST expression.( F-H ) Retina confocal micrographs and corresponding charts showing a significant decrease in angiogenesis and cell proliferation in animals containing the induced Rosa26-iSuRe-CrePEST v2 allele.Each dot in the charts represents one large retina field (in total 15403 cells were quantified in G and 8795 in H). ( I, J ) Animals containing the Sox2-Cre (recombines all cells) or Tie2-Cre (recombines only ECs and blood) allele in combination with the R osa26-iSuR e-CrePES T v2 allele are not born.Data are presented as mean values ± s.d.For statistics, see Source Data File 1 .Scale bars, 70 μm in D and 120 μm in E-H.

PAGE 11 OF 21 Figure 4 .
Figure 4. Design and validation of the Rosa26-iSuRe-HadCre mouse allele.( A ) Construct and sequential genetic cascade after induction of the R osa26-iSuR e-HadCre allele.After induction of CreERT2 with tamoxifen, the first recombination event results in deletion of either the LoxP or the LoxN cassette, or of both L o x-flank ed genetic cassettes.Since the L o xP-flank ed DNA cassette is easier to recombine than the L o xN-flank ed cassette, in the presence of relatively low levels of CreERT2 induction, LoxP recombination will predominate after a single 4-OHT injection (as illustrated in the figure), and this will trigger very strong expression of CreERT2 after.This strong CreERT2 expression will facilitate / enhance LoxN cassette recombination, especially if a second injection of tamoxifen (4-OHT) is delivered after 24 h.After recombination of both L o xP and L o xN-flank ed cassettes (after a single or multiple injections of tamoxifen), there will be strong, equimolar co-expression of Cre and the weaker FlpO recombinase.FlpO will recombine FRT sites, self-deleting the construct and activating expression of the reporter MbTomato in cells that had, but no longer have, high levels of Cre expression.( B ) FACS analysis of the temporal dynamics of the genetic cascade upon 4-OHT induction.Littermate pups with the indicated alleles were injected with a single 4-OHT dose at P4, and lung cells were collected at P5, P6 or P7. ( C ) FACS analysis of the kinetics of recombination upon induction of the R osa26-iSuR e-HadCre allele in fibroblasts derived from the indicated mice.( D ) Confocal analysis of the temporal dynamics of the genetic cascade upon induction of the R osa26-iSuR e-HadCre allele in retinal vessels after 15 mg / kg dose of 4-OHT.Note that the dose of 4-OHT used is low and will not recombine all the reporters in all cells.( E ) R elativ e inducibility of the iSure-Cre and R osa26-iSuR e-HadCre alleles in relation to the reference Rosa26-LSL-EYFP allele at different de v elopment al st ages and in different organs.( F ) FACS analysis of the correlation bet ween the number of t amoxifen doses and the frequency of recombination.Two doses induce higher recombination rates of the Rosa26-iSuRe-HadCre allele in relation to the Rosa26-LSL-EYFP allele.( G ) Comparative analysis on the recombination rate in adult aorta SMCs (Myh11-CreERT2 + ).( H ) Comparative analysis of recombination rates at very high doses of tamoxifen in adult animals, some with 2 copies (2 ×) of the iSuRe-HadCre allele.Each dot in the charts represents the mean value obtained in one animal.Data are presented as mean values ± s.d.For statistics, see Source Data File 1 .Scale bars, 200 μm.

PAGE 13 OF 21 Figure 5 .
Figure 5. R osa26-iSuR e-HadCre is bright, non-leaky and generates a high rate of true positiv es. ( A ) Comparativ e FACS analy sis of reporter (MbTomato) intensity in recombined cells at different stages and in different organs.( B ) Comparative FACS analysis of self-leakiness in iSure-Cre and iSuRe-HadCre mice organs.( C ) R epresentativ e FACS plots showing no leakiness in iSuRe-HadCre mice organs.( D ) Comparative FACs analysis of the rate of true positives (YFP + ) and false positives (YFP -) at embryonic stages in different mouse lines when combined with the compatible reference Rosa26-LSL-YFP allele.( E ) Comparative confocal analysis of the rate of true and false positives in postnatal retinas from different mouse lines when combined with the compatible reference Rosa26-LSL-YFP allele.( F ) Comparative FACS analysis of the rate of true and false positives in adult liver and lung cells; recombination in these quiescent cells is generally more difficult to achie v e than in embryonic or postnatal cells.( G , H ) Analysis by FACS of the frequencies of YFP+ (blue laser) and MbTomato+ (y ello w laser) CD31 − CD45 − cells (all organ cells e x cept CD45+ blood and CD31+ ECs) extracted from the indicated mice six da y s after receiving a single injection of 4-OHT.Note the difference in the frequency of false positives between the iSuRe-HadCre and the Rosa26-LSL-tdTomato line.Data are presented as mean values ± s.d.For statistics, see Source Data File 1 .Scale bars, 200 μm.

Figure 6 .PAGE 15 OF 21 Figure 7 .
Figure 6.R osa26-iSuR e-HadCre increases the efficiency and reliability of conditional genetics.( A-C ) R epresentativ e conf ocal micrographs of P6 retinal vessels from control mice ( A ) and Notch1 loss-of-function (LOF) mice ( B ) with high recombination rates of the R26-LSL-EYFP allele; vessels were stained with anti-ERG (labels EC nuclei, for object segmentation), isolectinB4 (labels EC surface), and anti-GFP (detects YFP).Quantification of the contribution of EYFP + ECs (ERG + , isolectinB4 + ) to arteries (indicated with a red A), arterial length, and vascular density reveals no significant differences, indicating poor Notch1 deletion ( C ). ( D-F ) R epresentativ e conf ocal micrographs of P6 retinal v essels from control mice ( D ) and Notch1 LOF mice ( E ) with high recombination rates of the iSuRe-HadCre allele.Quantification of the contribution of MbTomato + ECs to arteries, arterial length, and vascular density shows a very significant difference, confirming complete deletion of the Notch1 gene in MbTomato + cells in animals containing the iSuRe-HadCre allele.( G ) Comparative analysis of the contribution of R26-LSL-EYFP (YFP+) and iSuRe-HadCre (MbTomato + ) ECs to retinal arteries (indicated with a red A), versus all capillaries, confirming that only cells expressing the recombined iSuRe-HadCre allele have full deletion of Notch1 and cannot form arteries. ( H ) scRNAseq analysis of ECs collected from livers and hearts by FACS indicate strong Notch1 deletion in cells expressing the iSuRe-Cre or iSuRe-HadCre alleles.( I ) Representative confocal micrograph showing retinal vessels (IsolectinB4 + , ERG + ) and quantification of the contribution of MbTomato + ECs to the entire vascular network versus arteries, confirming very efficient deletion of Rbpj in the large majority of MbTomato + cells.Data are presented as mean values ± s.d.For statistics, see Source Data File 1 .Scale bars, 200 μm.

PAGE 17 OF 21 Figure 8 .Figure 9 .
Figure 8. iSuRe-HadCre enables the deletion of multiple floxed genes in the entire target tissue or in single cells.( A ) Retina confocal micrographs sho wing effectiv e deletion of Fo x o genes in MbTomato+ ECs (FO X O1 imunost aining and phenot ypes characteristic of full Fo x os depletion, such as defective EC sprouting and dense vasculature).( B ) Confocal micrographs showing that one single injection of 4-OHT at P1 efficiently recombines iSuRe-HadCre in all retina ECs and deletes Dll4 , leading to the strong upregulation of the tip cell marker gene Esm1 in most cells, and cell-cycle exit.ERG + ECs become Ki67-at the angiogenic front.( C ) One single injection of 4-OHT at P1 efficiently recombines iSuRe-HadCre in all retina ECs and deletes the gene Jagged1 , leading to the strong inhibition of angiogenesis.( D ) The membrane-tagged Tomato protein from iSuRe-HadCre allows visualization of filopodia by confocal microscopy.Deletion of the indicated genes change the number of filopodia per vessel length.( E , F ) Effective deletion of Flt1 / Vegfr1 is indicated by the loss of FLT1 protein in MbTomato+ ECs from the indicated mice.Note that Flt1 antibody signal is noisy.Specific signal is perinuclear / cytoplasmic.( G ) Retina confocal micrographs of animals with mosaic induction of the iSuRe-HadCre allele showing that Myc and Mycn (both detected with Anti-Myc) are only effectively deleted in cells expressing the iSuRe-HadCre allele (MbTomato + cells, white and orange arrows).( H ) Confocal micrographs of mouse aortas showing the effective deletion of Rbpj in MbTomato + of SMCs (Myh11+).White arrows indicate some nuclei / cells with MbTomato expression and lacking RBPJ, and green arrows without MbTomato expression and retaining RBPJ expression.Histogram showing the % of cells expressing RBPJ.Data are presented as mean values ± s.d.For statistics, see Source Data File 1 .Scale Bars 200 μm, except D, 50 μm and H, 25 μm.