Introduction

Genetic modification such as gene knockout in a precise spatiotemporal manner forms the basis for understanding cellular and molecular mechanisms in multiple biological processes^{1, 2}. The powerful Cre-loxP recombination system is the most widely used approach employed to knockout or over-express specific genes in vivo³. Cre recombinase, derived from bacteriophage P1 gene, targets short palindrome DNA sequence loxP (34bp) and recombines two loxP-flanked sequences for deletion^{4, 5}. Application of the Cre-loxP system in mouse genetics has revolutionized gene functional analysis and significantly advanced our understanding of many developmental and pathophysiological processes at the molecular level^{6, 7}. To date, most of the mouse genes have been successfully targeted by two loxP sites (floxed allele), which could be recombined by cell-type specific Cre lines. For many in vivo genetic studies, a tissue-specific gene knockout approach has been widely used as a gold standard to dissect gene function in a particular cell type during a biological process.

While constitutively active Cre is robust and efficient in recombination for gene deletion, Cre lacks temporal control when the promoter driving Cre is active from the embryonic to the adult stage. Therefore, a temporally active Cre is needed to avoid early mortality as a result of the ablation of an indispensable gene at a key developmental stage. To overcome the temporal limitation of Cre, CreER, the fusion of Cre with a mutated estrogen receptor, is generated to enable its activation only after tamoxifen (Tam) treatment, and interaction of ER with Tam induces nuclear translocation of CreER from the cytoplasm⁸. As a result, the activity of inducible CreER could be controlled by Tam treatment, thus restricting CreER activity at a specific time window and permitting gene functional analysis at a higher spatiotemporal resolution than Cre⁹. As the loxP-flanked sequences and their locations in genome or chromatin influence its accessibility for recombination, the recombination efficiency of CreER varies significantly among different floxed alleles, ranging from easy-to inert-to-recombine alleles¹⁰. For example, CreER has high efficiency in targeting the easy-to-recombine alleles such as generic reporter R26-tdT¹¹, but exhibits low efficiency on some other reporters such as R26-Confetti¹², or may not excise some floxed gene alleles that are inert or resistant for targeting, leading to failure of gene deletion in some tdT⁺ cells and causing false-positive tracing fate (Figure 1A). In other words, tdT expression may not always necessarily denote gene deletion in the labeled cells. The notion that the pattern of Cre-expression based on a reporter (e.g. R26-tdT) represents the pattern of Cre-mediated recombination is less rigorous¹³. It is therefore not correct to assume the successful deletion of the gene of interest based on the deletion of another gene (or reporter) by the same CreER mouse line.

Figure 1.

One way to increase the recombination efficiency of CreER is to treat mice with many times of Tam, as the high dosage of Tam in theory could increase the chances of CreER-mediated recombination. However, the increased dosage of Tam treatment is toxic and could lead to many side effects on the phenotype, creating confounding effects on the study^{14, 15}. To increase the inducible recombination efficiency of floxed alleles, Lao et al. reported a mosaic mutant analysis with spatial and temporal control of recombination (MASTR), in which initial FlpoER-mediated recombination induces GFPCre expression, that subsequently recombines floxed alleles effectively in GFP⁺ cells¹⁶. Similarly, a Flp-induced mosaic analysis system with Cre or Tomato (MASCOT) has been reported to express Cre and reporter simultaneously in a cell¹⁷. While MASTR and MASCOT enable spatiotemporal labeling of mosaic mutant cells utilizing the current resources of floxed mouse lines, it may not be suitable for effective gene knockout at a tissue/population level due to low recombination efficiency initiated by Flp recombinase. In order to improve the inducible recombination efficiency, Tian et al. have also reported a self-cleaved inducible CreER (sCreER) that switches inducible CreER into a constitutively active Cre, effectively recombining floxed allele for gene manipulation¹⁰. However, self-cleaved inducible Cre mice have to be newly generated each time for targeting different promoters, which is time-consuming and the system is not compatible with current resources largely based on CreER mouse lines. A recent study has reported an iSuRe-Cre strategy to induce and report Cre-dependent genetic modifications utilizing conventional CreER tools¹⁸. The reporter expression reflects gene knockout in cells by iSuRe-Cre. However, two limitations hinder its widespread applications. First, without any induction, the iSuRe-Cre transgene is leaky in some tissues such as heart and skeletal muscle. Second, the initial recombination for releasing the iSuRe-Cre allele induced by CreER is far from efficient in some tissues compared with easy-to-recombine alleles such as R26-tdT¹⁸, limiting its usage for efficient gene knockout at a tissue level. Thus, a new method is needed to ensure gene knockout in cells as effectively and efficiently as recombination on easy-to-recombine alleles, thus allowing efficient gene deletion at a population level.

Conventional Cre-loxP system uses tissue-specific promoter to drive Cre, thus the precision of genetic targeting solely depends on promoter activity. Now it is known that many promoters are not as specific as previously reported^19–21. The ectopic or unwanted Cre expression in other cell types may result in unspecific cell labeling or gene deletion, leading to many confounding issues or controversies in multiple fields of research^22–24. In addition, some cell types cannot be clearly distinguished by one marker from another, and thus could not be specifically and genetically targeted by only one gene promoter-driven recombinase²⁵. To circumvent this limit, dual recombinases-mediated genetic targeting utilizes two promoters to independently drive Cre and Dre²⁶. Dre is another bacteriophage recombinase that targets rox sites and is orthogonal to the Cre-loxP system²⁷. While cells of interest could be labeled by specific reporters that are responsive to dual recombinases, gene deletion requires a final readout on singular Cre recombinase for targeting floxed gene alleles.

In this study, we developed two genetic strategies: roxCre and loxCre, in which the rox-Stop-rox (RSR) and loxP-Stop-loxP cassettes are respectively inserted into the Cre coding region, such that Cre transcription and translation are not properly carried out before the removal of the Stop cassette. We did not detect any spontaneous leakiness by roxCre or loxCre. We found that CreER-induced loxCre effectively deletes genes in reporter-labeled cells, ensuring efficient gene knockout for the evaluation of lineage-specific gene function. In addition, DreER-induced roxCre efficiently deletes genes in cells for both specific and efficient intersectional genetic studies. We expect that these tools would overcome the issues of inefficient and non-specific genetic modifications, enhancing our ability to more precisely manipulate genes in a particular cell lineage for a better understanding of gene functions in multiple biomedical fields.

Results

Design of roxCre for DreER-induced mCre expression

A sequential genetic approach has been recently reported to delete genes²⁸, which uses Dre-induced CreER expression by removing the loxP-Stop-loxP sequence ahead. As aforementioned, the recombination carried out by the released CreER for excising floxed gene alleles may not be as efficient as reporters (e.g. R26-tdT). A possible strategy to control Cre activity is to engineer a rox-flanked Stop sequence ahead of its DNA (R26-RSR-Cre) that will be removed after Dre-rox recombination (Figure 1B). However, such a strategy is not ideal, as the upstream transcriptional Stop sequence might not completely prevent Cre transcription, leading to the leakiness issue associated with the iSuRe-Cre strategy¹⁸, for instance (Figure 1C). We then inserted a longer Stop sequence or reversed the direction of the Cre DNA by generating R26-RSR-Cre2 and R26-R-reverseCre-R lines, respectively, that still displayed heavy leakiness as Cre was independently activated for recombination without being crossed with a Dre mouse line (Figure 1D-G). To enable efficient dual recombinases-mediated gene knockout, we need to design a strategy for DreER-induced expression of a constitutively active Cre, and the system should not exhibit leakiness.

We first generated roxCre, in which an RSR cassette was inserted into the coding region of Cre, splitting Cre into N- and C-terminal segments (Figure 1H). After removal of RSR by Dre-rox recombination, Cre is recombined containing one remaining rox site within its coding sequence, hereafter termed as modified Cre (mCre, Figure 1H). To ensure that mCre has the same recombination efficiency as a conventional Cre, we inserted a rox sequence at 12 different sites of Cre individually, most of which were located between helix domains (Figure 1I). As insertions of additional amino acids of rox would change the original sequence of Cre, potentially leading to reduced activity, we screened for their recombination efficiency by transfecting cells with 12 versions of mCre (mCre1 to mCre12) and the responding GFP reporter (Figure 1J). We found that mCre1, 2, 3, 4, 6, 7, 10, and 11 displayed a comparable efficiency as the conventional Cre; whereas mCre5, 8, 9, and 12 exhibited impaired recombination efficiency (Figure 1K, L). These data demonstrate that at least some of the mCre generated in our study were as efficient as the conventional Cre in vitro.

Next, we sought to screen for the most robust version of mCre in vivo. We generated four roxCre knock-in mice based on mCre1, 4, 7, and 10 as determined from the above in vitro screening (Figure S1-S2). In these roxCre lines, RSR inserted into Cre was removed by Dre-rox recombination resulting in the generation of mCre (Figure 1J). We then used hepatocyte- and endothelial cell-specific promoters albumin (Alb) and VE-cadherin (Cdh5) to drive roxCre, and generate Alb-roxCre1-tdT, Cdh5-roxCre4-tdT, Alb-roxCre7-GFP, and Cdh5-roxCre10-GFP knock-in mice, respectively (Figure S1-S2). No fluorescence signal was detected in Alb-roxCre1-tdT, Cdh5-roxCre4-tdT, Alb-roxCre7-GFP, and Cdh5-roxCre10-GFP knock-in mice (Figure S1-S2), demonstrating no reporter leakiness without Dre-rox recombination. These mice were crossed with R26-GFP²⁹ or R26-tdT¹¹ reporters. Nor did we detect leakiness as a result of any Cre activity when they were crossed with R26-tdT or R26-GFP reporter (Figure S1-S2). These data demonstrate that roxCre was functionally efficient yet non-leaky.

DreER-induced mCre robustly recombines inert alleles

The mCre system was originally designed for inducible gene knockout to study biological functions. To more rigorously evaluate the recombination effectiveness among different versions of mCre, we used an inducible DreER to temporally and specifically release mCre in hepatocytes or endothelial cells. Although mCre efficiently labeled virtually all targeted cells (Figure S3A-E), a strong Cre is not needed to drive easy-to-recombine alleles such as R26-tdT reporter in this case. To identify a strong mCre, we attempted to target one of the most inert alleles for recombination, R26-Confetti¹², which is often used as reporters for clonal analysis due to its sparse labeling with rare recombination events^{10, 12}. In the Cre-loxp recombination principle, Cre can remove the sequence between two loxp sites that have the same transcription direction and turn over the sequence in the middle of two loxp sites that have the opposite transcription direction. As the recombination efficiency of CreER is limited, R26-Confetti only can be recombined into YFP, or nGFP (nuclear GFP), or mCFP (membrane CFP), or RFP (Figure S3F).

We first crossed R26-DreER;R26-Confetti mice with Alb-roxCre1-tdT (group 1) and Alb-roxCre7-GFP (group 2) mice, and Tam-induced Dre-rox recombination yielded Alb-mCre1-tdT and Alb-mCre7-GFP mice, respectively (Figure 2A). In this system, the constitutively active mCre would completely remove the sequence between two loxp sites that have the same transcription direction, and recombine R26-Confetti into two sets of reporter pairs: YFP-nGFP pair and mCFP-RFP pair, which could be detected in a single hepatocyte due to the poly-nuclear or polyploidy feature of hepatocytes (Figure 2A). In each pair, reporters can equally express, as mCre is strong enough to turn over the sequence in the middle of two loxp sites that have the opposite transcription direction (Figure S3G). Therefore, we used YFP/mCFP to detect mCre recombination efficiency and tdT/RFP or GFP to trace its endogenous reporter activity that was driven by Alb promoter in Alb-mCre1-tdT and Alb-mCre7-GFP mice, respectively (Figure 2A). Besides, Alb-CreER²⁶;R26-Confetti was used as a control under the same Tam treatment (group 3, Figure 2A).

Figure 2.

In fact, DreER-rox recombination in hepatocytes is not 100%. Therefore, we quantified YFP and mCFP expression in hepatocytes with DreER-rox recombination to evaluate the strength of mCre activity. We found sparse YFP⁺ or mCFP⁺ hepatocytes (0.39 ± 0.04%) in Alb-CreER;R26-Confetti mice. On the other hand, 99.94 ± 0.02% of tdT⁺ hepatocytes (DreER recombined) were YFP⁺ or mCFP⁺ in R26-DreER;Alb-roxCre1-tdT;R26-Confetti mice; and 76.83 ± 3.20% of GFP⁺ hepatocytes (DreER recombined) were YFP⁺ or mCFP⁺ in R26-DreER;Alb-roxCre7-GFP;R26-Confetti mice (Figure 2B, C). Similarly, we found that 39.81 ± 2.61% of tdT⁺ endothelial cells and 74.15 ± 3.64% of GFP⁺ endothelial cells in small intestine were YFP⁺ or mCFP⁺ in R26-DreER;Cdh5-roxCre4-tdT;R26-Confetti and R26-DreER;Cdh5-roxCre10-GFP;R26-Confetti mice, respectively, compared with sparse labeling in Cdh5-CreER;R26-Confetti mice (Figure 2D-F). Similar results were also observed in other tissues and organs (Figure S4). These data demonstrate that inducible DreER controlled roxCre activation, and mCre1 was the most efficient recombinase assisting the recombination of inert alleles in vivo.

Dre-induced mCre efficiently deletes floxed genes

Next, we examined the efficiency of gene knockout by Dre-induced mCre. We generated the Alb-roxCre1-tdT;Ctnnb1^flox/floxmouse line to examine if Dre-induced mCre could efficiently delete Ctnnb1 gene that encodes β-catenin (Ctnnb1^flox)³⁰. To enhance the precision and specificity of targeted cells, we employed intersectional genetic targeting using dual recombinases, and generated the Cyp2e1-DreER mouse line that labeled peri-central hepatocytes and also renal epithelial cells (Figure S5). Crossing Cyp2e1-DreER with hepatocyte-specific Alb-roxCre1-tdT mice achieved genetic targeting of peri-central hepatocytes exclusively, circumventing issues associated with ectopic targeting of renal cells by Cyp2e1-DreER line (Figure S5) or unwanted targeting of peri-portal hepatocytes by Alb-CreER line²⁶.

We generated Cyp2e1-DreER;Alb-roxCre1-tdT;Ctnnb1^flox/floxtriple knock-in mice (mutant), in which Tam-induced DreER-rox recombination yielded mCre that subsequently targeted floxed Ctnnb1 allele (Figure 3A). We treated mutant mice and their littermate control Cyp2e1-DreER;Alb-roxCre1-tdT;Ctnnb1^flox/+ mice with Tam at 8 weeks of age, and purified tdT⁺ hepatocytes for analysis at 3 days after Tam treatment (Figure 3B). qRT-PCR analysis showed significantly reduced Ctnnb1 and Glul in tdT⁺ hepatocytes of the mutant group, compared to that of the control group (Figure 3C, D). We then collected liver samples at 3 days (D) and 4 weeks (W) post-Tam for further analysis (Figure 3E). Immunostaining for tdT, β-catenin, GS, and E-CAD on liver sections revealed similar levels of β-catenin and GS expression in the 3D mutant liver than the control liver (Figure 3F). However, expression of β-catenin and GS almost disappeared in the 4W mutant sample compared with the control (Figure 3G), suggesting efficient deletion of Ctnnb1 and down-regulation of Wnt signaling downstream gene target GS. Western blotting against β-catenin and qRT-PCR against Ctnnb1 of purified tdT⁺ hepatocytes at 4W post-Tam revealed a near complete loss of Ctnnb1 at both the protein and mRNA levels (Figure 3H-K). Additionally, Wnt downstream target genes including Glul, Axin2, Cyp1a2, Cyp2e1, Oat, Tcf7, Lect2, Tbx3, Slc1a2, and Rhbg were significantly reduced in the mutant than the control group (Figure 3L). Taken together, these data demonstrate that DreER-induced mCre enabled efficient intersectional genetic manipulation in the cells of interest.

Figure 3.

R26-loxCre-tdT is efficiently recombined by CreER

Having successfully established roxCre1 for Dre-induced mCre expression, we iterated the system to enable mCre induction by CreER, a more widely available tool for biomedical research. We replaced the rox sites of roxCre1 with loxP sites by inserting a loxP-Stop-loxP cassette into Cre cDNA at the same locus as roxCre1(Figure 4A). We then generated a new R26-loxCre-tdT mouse line, in which mCre and tdT would be expressed simultaneously after Stop removal, allowing tdT as a surrogate marker for mCre detection in the same cell (Figure 4C). We then designed two experiments to verify the baseline leakiness of Cre expression (Figure 4B) and to examine if mCre/tdT could be expressed after Stop removal (Figure 4C), respectively, by R26-loxCre-tdT;R26-tdT (Strategy 1) and by injecting AAV8-Cre virus into R26-loxCre-tdT mice (Strategy 2). Immunostaining for tdT on tissue sections revealed no tdT expression in Strategy 1 demonstrating minimal to negligible leakiness at the baseline. The robust tdT expression in the liver in Strategy 2 showed mCre/tdT was allowed to express after Stop removal by AAV2/8-hTBG-Cre(Figure 4D).

Figure 4.

We then confirm if the new design can improve the labeling efficiency compared to the conventional approach by crossing endothelial cell-specific Cdh5-CreER with the conventional reporter R26-tdT (Strategy 3) and the new reporter R26-roxCre-tdT mice (strategy 4) under the same protocol of Tam treatment (Figure 4E). Immunostaining for tdT and VE-Cad on tissue sections revealed no difference in the percentage of VE-Cad⁺ endothelial cells expressing tdT between strategies 3 and 4 (Figure 4F-G), suggesting that recombination of the R26-roxCre-tdT allele by CreER was as efficient and easy as R26-tdT which is an easy-to-recombine reporter mouse line. Therefore, R26-roxCre-tdT can be served as an easy-to-recombine reporter, enabling efficient recombination to generate mCre/tdT in CreER-expressing cells, without leakiness.

R26-loxCre-tdT adaptor ensures efficient recombination

We next examined whether mCre released from loxCre could also display high recombination efficiency for the inert-to-recombine allele R26-Confetti. We generated the triple knock-in Cdh5-CreER;R26-loxCre-tdT;R26-Confetti mouse line in which mCre was released for confetti recombination upon Tam treatment, and Cdh5-CreER;R26-Confetti was used as a control (Figure 5A). Both mouse strains were treated with Tam at 8 weeks of age and their tissue samples were collected for analysis at one week post-Tam (Figure 5B). As tdT expression is also detected in R26-loxCre-tdT after the first recombination by CreER, we only compared the YFP and mCFP fluorescence signals from R26-Confetti (Figure 5A). Wholemount-fluorescence imaging showed significantly more YFP and mCFP signals in the retina of the Cdh5-CreER;R26-loxCre-tdT;R26-Confetti mice, compared with that of Cdh5-CreER;R26-Confetti mice (Figure 5C). Immunostaining on frozen sections revealed significantly higher percentage of endothelial cells expressing YFP/mCFP in various vascularized organs of the Cdh5-2a-CreER;R26-loxCre-tdT;R26-Confetti mice, compared with that of the Cdh5-2a-CreER;R26-Confetti mice (Figure 5D). Of note, virtually all tdT⁺ endothelial cells simultaneously expressed YFP/mCFP (Figure 5D). These data demonstrate that the R26-loxCre-tdT line could be used as an adaptor to enhance CreER-mediated recombination with a simultaneous expression of tdT as an accurate indicator.

Figure 5.

R26-loxCre-tdT ensures gene deletion in tdT⁺ cells

To examine if R26-roxCre-tdT enables Alb-CreER to more efficiently delete genes, we crossed Alb-CreER;R26-roxCre-tdT with Ctnnb1^flox/floxto generate Alb-CreER;R26-roxCre-tdT;Ctnnb1^flox/flox mice for evaluating the efficiency of releasing mCre and also mCre-mediated Ctnnb1 deletion in tdT⁺ hepatocytes (Figure 6A). We also crossed Alb-CreER;R26-tdT2 with Ctnnb1^flox/flox to generate Alb-CreER;R26-tdT2;Ctnnb1^flox/flox mice to evaluate recombination efficiency of both R26-tdT2 and Ctnnb1^flox/flox alleles (Figure 6A). Additionally, Alb-CreER;R26-roxCre-tdT;Ctnnb1^flox/+ mice were used as a control for heterozygous gene deletion (Figure 6B). We treated Alb-CreER;R26-tdT2;Ctnnb1^flox/flox and Alb-CreER;R26-roxCre-tdT;Ctnnb1^flox/+ mice with five dosages of Tam, and treated Alb-CreER;R26-roxCre-tdT;Ctnnb1^flox/flox mice only with one dosage of Tam, and collected tissues for analysis to determine Ctnnb1 gene deletion at 4 weeks after Tam treatment (Figure 6B). Immunostaining data showed that β-catenin, GS, E-CAD, and Cyp2e1 were significantly reduced in Alb-CreER;R26-roxCre-tdT;Ctnnb1^flox/floxmice, but were readily detectable in both Alb-CreER;R26-tdT2;Ctnnb1^flox/floxand Alb-CreER;R26-roxCre-tdT;Ctnnb1^flox/+ mice (Figure 6C, D). Of note, a small number of hepatocytes continued to express Cyp2e1 in Alb-CreER;R26-roxCre-tdT;Ctnnb1^flox/flox mice (Figure 6D). These Cyp2e1-expressing hepatocytes were unanimously tdT⁻, and not a single tdT⁺ hepatocyte expressed Cyp2e1 (Figure 6D), suggesting that these Cype2e1⁺tdT⁻ hepatocytes were likely not targeted by Alb-CreER for the recombination to release mCre. Western blotting of β-catenin and qRT-PCR of Ctnnb1 of isolated hepatocytes revealed almost complete deletion of Ctnnb1 in the Alb-CreER;R26-tdT2;Ctnnb1^flox/floxmice compared with those in control groups (Figure 6E-G). Additionally, those Wnt downstream or regulated genes including Glul, Axin2, Cyp1a2, Cyp2e1, Oat, Tcf7, Lect2, Tbx3, Slc1a2, and Rhbg were significantly reduced in hepatocytes derived from the Alb-CreER;R26-tdT2;Ctnnb1^flox/floxmice, compared with those from the Alb-CreER;R26-tdT2;Ctnnb1^flox/floxand Alb-CreER;R26-roxCre-tdT;Ctnnb1^flox/+ mice, respectively (Figure 6H).

Figure 6.

Discussion

In this study, we respectively generated two genetic tools based on roxCre and loxCre to enhance the effectiveness of Cre-mediated gene deletion even under the control of a less efficient DreER or CreER driver. The R26-loxCre-tdT line permitted efficient recombination to release mCre, which greatly enhanced the effectiveness of subsequent gene deletion in tdT⁺ cells. The inclusion of a fluorescent reporter with roxCre or loxCre allele allowed us to readily detect cells with gene deletion, thus facilitating further detailed characterization of these cells both in situ and ex vivo. Furthermore, the strategy of roxCre further enhances the precision for cell fate mapping with genetic deletion simultaneously in Dre⁺Cre⁺ cells through intersectional genetic targeting. For example, we have showcased the power of roxCre for genetic manipulation in peri-central hepatocytes, resulting in efficient gene knockout in tdT⁺ hepatocytes specifically. Additionally, we have demonstrated the efficient deletion of Wls and Ctnnb1 genes in tdT⁺ endothelial cells and hepatocytes, respectively, by combining R26-loxCre-tdT adaptors with CreER lines. Altogether, we have demonstrated that the R26-loxCre-tdT mouse line could be widely applicable to any CreER-bearing mouse for enhancing its effectiveness in genetic modification with simultaneous cell lineage tracing in multiple biomedical research fields.

Coupling gene deletion with reporter activity can greatly facilitate the accuracy of studying gene function in targeted cells by in situ investigation with or without specific cell isolation for ex vivo experiments. Assuming successful gene deletion in the cell type of interest because of efficient cell labeling by a lineage-specific reporter mediated by the same Cre is incorrect, as recombination efficiency varies significantly among different mouse lines. In fact, many factors can potentially impact Cre-loxP recombination efficiency. Concerning the recombinase enzyme, the type of recombinase, the strength of the promoter that drives the recombinase, the expression level of the recombinase, and the translocation efficiency of an inducible recombinase such as CreER from the cytoplasm to the nucleus after Tam induction could all influence the recombination efficiency. For the targeted DNA sites, the genomic loci where targeted sites are placed, loxP-flanked sequence length, and even the inserted DNA sequence per se could also influence the recombination efficiency. We clearly observed the differences in recombination efficiency of different alleles (e.g. R26-tdT, R26-tdT2, and R26-Confetti), even located in the same genomic locus (e.g. Rosa26) mediated by the same CreER under the same Tam treatment protocol, as evidenced by high efficiency in R26-tdT, medium efficiency in R26-tdT2, and low efficiency in R26-Confetti. Importantly, R26-loxCre-tdT was found to be efficient for recombination mediated by CreER, which was comparable to that of the R26-tdT allele, thus enabling efficient labeling of cells of interest. Furthermore, R26-loxCre-tdT also enabled CreER to efficiently ablate genes in any cells that had been genetically traced with tdT, synchronizing genetic modification and lineage tracing in the labeled cells.

Despite the high demand in the quest for efficient genetic modification in genetically labeled cells is significant, no efficient tool has been available to date. A previous study attempted to generate an adaptor for CreER to efficiently delete genes in reporter-labeled cells by a transgenic iSuRe-Cre¹⁸. However, the leakiness of iSuRe-Cre on the Rosa26 locus was too high in the male germline. Therefore, a transgenic mouse line was reconstructed subjected to iSuRe-Cre mediated recombination, which still displays consistently leaky recombination in some tissues such as cardiac and skeletal muscle¹⁸. The leakiness was likely due to the inability of the Stop cassette to completely block the transcription of Cre in iSuRe-Cre. Consistently, we also found leakiness as a result of Cre transcription even in the presence of Stop ahead in the R26-Stop-Cre mouse line (Figure 1C). These leaky events, despite being at different levels, demonstrate that transcription of recombinases such as Cre could not be easily inhibited. Instead of blocking its transcription by Stop, we inserted Stop in between the N- and C-terminus of Cre, thus preventing Cre expression completely at the translational level. We found that the removal of Stop rendered mCre, which efficiently recombined different types of alleles examined in our study. Additionally, we also observed inefficient recombination in hepatocyte-specific Alb-CreER;iSuRe -Cre mice, rendering this line less effective in conditional gene deletion in hepatocytes at a population level. In contrast, R26-loxCre-tdT was an easy-to-recombine allele for Alb-CreER mediated recombination, allowing tdT labeling and gene deletion simultaneously in the majority of hepatocytes. Therefore, our R26-loxCre-tdT line has circumvented the leakiness issue associated with iSuRe-Cre, exhibited as an easy-to-recombine allele for CreER targeting, and synchronized fate mapping with efficient genetic modification robustly, efficiently and temporally in the same cells.

The efficiency of CreER-loxP recombination depends not only on the location of the loxP-containing alleles but also on the strength of CreER driven by an active promoter. For instance, R26-loxCre-tdT was an easy-to-recombine allele compared with other tested R26 alleles, but the recombination efficiency to release mCre for subsequent recombination depends on the strength of the selected CreER lines for genetic targeting. It is known that different CreER mouse lines vary significantly for recombination, largely due to the differences in expression levels of CreER and Tam induction efficacy. Therefore, we should select a relatively stronger CreER line for efficient recombination on R26-loxCre-tdT to release mCre for gene deletion at a population level. In addition to bulk gene deletion, the R26-loxCre-tdT also offers an alternative way for mosaic analysis of gene function by sparse labeling. Previous studies have used Cre-loxP to over-express gene and fluorescence reporter simultaneously in a cell for functional genetic mosaic (ifgMosaic) analysis³¹. While ifgMosaic could over-express some dominantly negative genes for functional study, this system could not take advantage of rich resources from many conventionally available floxed mouse lines for loss-of-function study. Mosaic mutant analysis with spatial and temporal control of recombination (MASTR) utilized FlpoER for tissue-specific expression of GFPCre to delete floxed genes¹⁶. Compared with FlpoER lines, CreER is a more broadly used recombinase that most laboratories use in gene deletion experiments. A recent study using dual recombinase-mediated cassette exchange (MADR) also permits stable labeling of mutant cells expressing transgenic elements from defined chromosomal loci³². However, the use of viruses and electroporation, albeit convenient and rapid, are inefficient to specifically target any type of cells for in vivo studies. The powerful mosaic analysis with double markers (MADM) enables simultaneous lineage tracing of a pair of mutant and control sibling cells with distinct fluorescence reporters, allowing precise mosaic analysis of gene function in any cell^{33, 34}. Nevertheless, the MADM cassette has to be combined with the mutant null allele, which is not readily available for synchronizing reporter expression and genetic modification in cells of interest. The ensured gene deletion with tdT reporter by R26-roxCre-tdT could be coupled with any present CreER and loxP alleles for potential mosaic analysis. This could be achieved by adjusting Tam at a lower dosage so that the recombination efficiency to release mCre could be low in cells of interest for sparse tdT labeling. We believe that R26-loxCre-tdT could be useful for robust, efficient, and temporal gene deletion at a population level and mosaic analysis at a single-cell level with improvement in the future.

In this study, we deleted the Wnt signaling gene Ctnnb1 in hepatocytes to present how roxCre and loxCre respectively enable efficient recombination. The impact of deleting Ctnnb1 in a specific hepatocyte lineage on homeostasis, development, and injury recovery requires further investigation. Besides, roxCre and loxCre should be involved in more gene functional studies, in which knocking out genes continuously leads to embryonic death due to their crucial role in development.Both loxCre and roxCre incorporate a fluorescent reporter tdT for genetic lineage tracing, revealing efficient gene knockout in cells marked by the reporter simultaneously. In other words, one reporter tdT can only show cells with gene knockout. One direction to improve the current R26-loxCre-tdT for mosaic analysis in the future is to generate a second distinct surrogate reporter for wild-type cells as an internal control during mosaic analysis, similar to that in the MADM system.

Materials and Methods

Mice

Experiments using mice (Mus musculus) were carried out with the study protocols (SIBCB-S374-1702-001-C1) approved by the Institutional Animal Care and Use Committee of Center for Excellence in Molecular Cell Science (CEMCS), Shanghai Institute of Biochemistry and Cell Biology, Chinese Academy of Sciences. The R26-DreER³⁵, Alb-CreER²⁶, R26-GFP²⁹, R26-tdT¹¹, R26-tdT2³⁶, R26-RSR-tdT³⁷, R26-Confetti¹²,and Ctnnb1-flox³⁰ mouse lines were used as previously described. New knock-in mouse lines R26-RSR-Cre, R26-RSR-Cre2, R26-R-reverseCre-R, Cdh5-CreER, Alb-roxCre1-tdT, Alb-roxCre7-GFP, Cdh5-roxCre4-tdT, Cdh5-roxCre10-GFP, Cyp2e1-DreER, and R26-loxCre-tdT were generated by homologous recombination using CRISPR/Cas9 technology. These new mouse lines were generated by the Shanghai Model Organisms Center, Inc. (SMOC). These mice were bred in a C57BL6/ICR mixed background. All mice were housed at the laboratory Animal center of the Center for Excellence in Molecular Cell Science in a Specific Pathogen Free facility with individually ventilated cages. The room has controlled temperature (20–25 °C), humidity (30–70%), and light (12 hours light-dark cycle). For the determination of the embryonic period of sampling, the day on which the vaginal plug was examined in female mice was recorded as E0.5. As the sex is not relevant to the topic of this study, male and female mice ranging in age from E13.5 to 12W (week) were allocated and mixed into the experimental groups in this study. No data in mouse experiments were excluded.

Genomic PCR

Genomic DNA was prepared from the mouse toes or embryonic tails. Tissues were precipitated by centrifugation at maximum speed for 1 min at room temperature. After that, tissues were lysed by lysis buffer (100 mM Tris-HCl, 5 mM EDTA, 0.2% SDS, 200 mM NaCl, and 100 μg/mL Proteinase K) at 55 °C overnight. About 750μL pure ethanol was added to the lysis mixture and mixed thoroughly, followed by centrifugation at maximum speed for 5 min at room temperature to collect the DNA precipitation. Then the supernatant was discarded and the mixture was dried at 55 °C for 1 hour. About 100–200μL double-distilled H₂O was added to dissolve the DNA. The genomic PCR primer pairs were designed for the mutant alleles spanning both endogenous genomic fragments and insert fragments. All the new genomic PCR primer sequences would be provided upon request.

In vitro screening of mCre for efficient recombination

To test if the remaining rox sequence after Dre-rox recombination would impact Cre activity, the pCDNA3.1³⁸ (Invitrogen, V79020) was used to express 12 types of modified Cre (mCre). For testing the Cre activity, 500ng pCDNA3.1-mCre plasmids, or pCDNA3.1 (negative control), or pHR-CMV-nlsCRE (positive control, Invitrogen, 12265) were mixed with 500ng pCAG-loxp-stop-loxp-ZsGreen (Addgene, 51269) plasmid. In-vitro transfection experiment protocol is performed according to that described previously³⁹. For cell culture medium, DMEM (ThermoFisher, 11965092) was supplemented with 10% fetal bovine serum (fetal bovine serum, 10099141c) for preparing fresh complete culture medium. Poly-D-Lysine (ThermoFisher, A3890401) precoated coverslips (biosharp, BS-14-RC) or 10 cm dish (ThermoFisher, 150466) overnight to dry, and sterile water was used to flush it for three times. A cryogenic vial of QBI293 was put in a 37°C water bath, then thawed liquid contents into a 50 mL conical tube (Corning,430829) prefilled with 5 mL prewarmed fresh complete culture medium. After centrifuging at 125 × g for 5 min, the supernatant was discarded and cells were resuspended in 10 mL complete medium. One-third of the cells were plated in a 10 cm coated dish and incubated in cultures at 37°C. On the second day, the culture medium was discarded, and the cell layer was briefly rinsed with prewarmed PBS (Gibco, C10010500BT). After that, PBS was removed and 1 mL 0.25% Trypsin solution (with EDTA, Gibco, 25200072) was added. The cells were incubated at 37°C for 3 min., and a 6mL complete growth medium was added to aspirate cells by gently pipetting. The medium-containing cells were transferred to a 50 mL conical tube and centrifuged at 125 × g for 5 min. The supernatant was discarded, and cells were resuspended in 5.2 mL complete culture medium. 24 well plates were placed into coated coverslips and added 1mL complete culture medium. 200µL cells were added to every experimental well and incubated cultures at 37°C for 9 hours to grow up to ∼80%. The new prewarmed fresh complete culture medium replaced the old for 1mL per well. Lipofectamine™ 3000 Transfection Reagent (Thermo Fisher, L3000015) was used for plasmid transfection. Samples were incubated in cultures at 37°C for 24 hours. Samples were washed by PBS once and mounted on slides by VECTASHIELD^®. Antifade Mounting Medium with DAPI (Vector, H-1200). Images were acquired using an Olympus microscope (Olympus, BX53).

Tamoxifen injection and sample collection

For induction of CreER or DreER, tamoxifen (Sigma, T5648) was dissolved in corn oil and was administered to mice by oral gavage. Cyp2e1-DreER;Alb-roxCre1-tdT;Ctnnb1^flox/+and Cyp2e1-DreER;Alb-roxCre1-tdT;Ctnnb1^flox/flox mice were treated with 0.05 mg Tam per gram mouse body weight for one dose. For comparing tdT⁺ cells, Cyp2e1-DreER;R26-RSR-tdT and Cyp2e1-DreER;Alb-roxCre1-tdT mice were treated with 0.05 mg Tam per gram mouse body weight. R26-DreER;Alb-roxCre1-tdT;R26-Confetti, R26-DreER;Cdh5-roxCre4-tdT;R26-Confetti, R26-DreER;Alb-roxCre7-GFP;R26-Confetti, R26-DreER;Cdh5-roxCre10-GFP;R26-Confetti, R26-DreER; R26-DreER;Alb-roxCre7-GFP;R26-GFP, Alb-CreER;R26-Confetti, Cdh5-CreER;R26-Confetti, Alb-CreER;R26-tdT2;R26-Confetti, Alb-CreER;iSuRe-Cre;R26-Confetti, Alb-CreER;R26-loxCre-tdT;R26-Confetti, Alb-CreER;R26-loxCre-tdT;Ctnnb1^flox/flox, and Cyp2e1-DreER;R26-RSR-tdT mice were treated with 0.2 mg Tam per gram mouse body weight for one dose. Cdh5-CreER;R26-tdT, Cdh5-CreER;R26-loxCre-tdT, Cdh5-CreER;R26-Confetti, Cdh5-CreER;R26-loxCre-tdT;R26-Confetti, Alb-CreER;R26-tdT2;Ctnnb1^flox/flox, and Alb-CreER;R26-loxCre-tdT;Ctnnb1^flox/+ mice were treated with 0.2 mg Tam per gram mouse body weight one dose per day for 5 days. R26-RSR-Cre2;R26-GFP and R26-R-reverseCre-R;R26-tdT mice were sacrificed at 8 weeks old for analysis. Mice, both males and females, at the age of 8–12 weeks were used for experiments with similar-aged mice for both control and experimental groups.

Whole-mount imaging and sectioning

The tissue samples were fixed with 4% paraformaldehyde (PFA) (Sigma, P6148-500g, wt/vol in PBS) for 1 or 2 hours (stomach, intestine, colon) at 4°C, followed by washing with PBS three times. The fixed tissues were placed in an agarose-filled petri dish for bright-field and fluorescence imaging by a Zeiss stereoscopic microscope (AxioZoom V16). For cryo-sections, tissues were sectioned to slides of 10-μm thickness after dehydration by 30% sucrose (Sinopharm, H-10021463, wt/vol in PBS) overnight and pre-embedding with OCT (Sakura) at 4 °C for 1 hour.

Immunostaining

Immunostaining was performed as previously described⁴⁰. 0.2% PBST was prepared with 0.2% (vol/vol) Triton X-100(Sigma, T9284) dissolved in PBS. Tissue sections were blocked with 2.5% normal donkey serum and 0.1% 4’6-diamidino-2-phenylindole (DAPI, Vector lab, D21490) dissolved in 0.2% PBST for 30 mins after washing with PBS three times. The tissue sections were incubated with primary antibody diluted in 0.2% PBST at 4°C overnight. On the next day, sections were incubated with secondary antibodies diluted in 0.2% PBST at room temperature for 30 min, followed by PBS washing for three times. The slides were washed with PBS for three times. The slides were mounted with a mounting medium (Vector Lab). For weak signals, the endogenous peroxidase activity was quenched before blocking. Horseradish peroxidase or biotin-conjugated secondary antibodies and a Tyramide signal amplification kit (PerkinElmer) were used after incubating the primary antibodies. For primary antibodies of murine origin, mouse immunoglobulins were blocked with an anti-mouse Fab antibody (Jackson, 715-007-003, 1:100). The included primary antibodies are listed as follows: tdT (Rockland, 600-401-379, 1:500; or Rockland, 200-101-379, 1:500), GFP (Invitrogen, A11122, 1:500), GFP (Rockland, 600-101-215M, 1:500), GFP (Nacalai, 04404-84; 1:500), GS (Abcam, ab49873, 1:1000), β-catenin (BD Pharmingen, 610153, 1:200), E-CAD (R&D, AF748, 1:500), Ep-CAM (Abcam, ab92382, 1:500), VE-Cad (R&D, AF1002, 1:100), and CYP2E1(Abcam, ab28146, 1:100). The corresponding secondary antibodies (JIR or Abcam) were diluted according to the instructions. Images were captured by using a Nikon confocal (Nikon A1 FLIM) or an Olympus confocal (FV3000), and captured images were analyzed by Image J2 (version 2.9.0/1.54f) and Photoline (version 23.02) software. We collected five random fields from each liver section for quantification. Mutant and control sections were processed at the same time to avoid potential batch differences during staining. Imaging of all immunostained slides was performed under the same exposure and contrast conditions using the same confocal microscope.

Imaging of Confetti fluorescence reporters

All the fluorescence images of R26-confetti were taken by a Leica confocal (Leica SP8 will). All the sections were collected freshly and blocked with 0.2% PBST containing DAPI for 30 mins after washing with PBS three times. The parameters of excitement light and emission light for all channels are set as follows: DAPI (ex.405nm, em.415-450nm), CFP (ex.458nm, em.463-481nm), GFP (ex.488nm, em.495-511nm), YFP (ex.520nm, em.527-545nm), tdT (ex.546nm, em.555-590nm), markers for antibody staining (ex.647nm, em.657-700nm). Captured images were analyzed by Image J2 (version 2.9.0/1.54f) and Photoline (version 23.02) software.

Hepatocyte dissociation

Mouse primary hepatocytes were isolated by the two-step collagenase perfusion method, which was modified from a previous protocol⁴¹. Briefly, mice were anesthetized and the liver was exposed through an incision in the lower abdomen. The inferior vena cava and portal vein were also exposed. A needle was inserted into the inferior vena cava and secured with a hemostatic clamp around the needle. The portal vein was cut immediately when the mouse liver was perfused with a perfusion medium buffer (containing 0.5 mM EGTA) for 5 minutes using a peristaltic pump. Then, the liver was perfused with medium containing collagenase type I (150 U/ml; Gibco, 17100-017) for 2-5 min to adequately digest the liver. After the gallbladder was removed, the liver was dissected with cold DMEM to free the hepatic cells. Then the cell suspension was passed through a 70-μm cell strainer (BD Biosciences, 352350) and centrifuged at 50 x g for 3 min at 4 °C. The supernatant was removed, and cells were resuspended in Percoll (GE Healthcare, 17-0891-01)/DMEM/10 x PBS (Sangon, B548117, diluted in 1:1) (1:1:0.1) mixture and centrifuged at 300 x g for 5 min at 4 °C. After the supernatant was removed, cells were dissected with cold DMEM and centrifuged at 50 x g for 3 min at 4 °C. Purified hepatocytes were collected for FACS analysis, qRT-PCR or Western Blot analysis.

Flow cytometric analysis and isolation of tdT⁺ hepatocytes

Cells were stained with the following antibodies: anti-mouse CD45 APC (eBioscience, 17-0451-82, 1:400), anti-mouse CD140a APC (eBioscience, 17-1401-81, 1:100), anti-mouse CD31 APC (eBioscience, 17-0311-82, 1:40). All antibodies were diluted in 1∼2mL DMEM (according to cells number) and cells were incubated with antibodies for 30 min in the dark at 4 °C. After staining, 10mL DMEM was added to stop staining, and the cell suspension was passed through a 70-μm cell strainer. Cells were centrifuged at 50 x g for 3 min at 4 °C and dissected with relevant suspended solution. For no DAPI control group, their suspended solution is 0.1% Dnase I (Worthington, LS002139, diluted by DMEM), otherwise, the suspended solution is 0.1% Dnase I mixed with 0.1% DAPI. Hepatocytes were sorted by FACS Aria SORP machine. Cell viability was assessed with DAPI staining. APC^-tdT⁺ hepatocytes were collected for subsequent experiments.

Total RNA extraction and qRT-PCR

Total RNA was extracted from the liver of indicated mice or hepatocytes isolated from indicated mice as previously described⁴². Cells or homogenized tissues were lysed with Trizol (Invitrogen, 15596018), and total RNA was extracted according to the manufacturer’s instructions. For each sample, 1 μg of total RNA was reversely transcribed into cDNA using the Prime Script RT kit (Takara, RR047A). The SYBR Green qPCR master mix (Thermo Fisher Scientific, 4367659) was used and quantitative RT-PCR was performed on QuantStudio 6 Real-Time PCR System (Thermo Fisher Scientific). Gapdh was used as the internal control. Sequences of all primers would be provided upon request.

Western blot

Liver tissues were collected at the indicated stages. All samples were lysed in RIPA lysis buffer (Beyotime, P0013B) containing protease inhibitors (Roche, 11836153001) for 30 min on ice, and then centrifuged at 15,000 x g for 15 min to collect the supernatant. All samples were separated with 30μL for testing the protein concentration by Pierce^TM BCA Protein Assay kits (Thermo Scientific^TM,23227). The remaining samples were mixed with 5 x loading buffer (Beyotime, p0015L) and boiled at 100 °C for 5 min. 40μg samples were added to every well in precast gradient gels (Beyotime, P0469M) with 1x running buffer (Epizyme, PS105S, should be diluted into 1x). After running, samples were transferred onto Immobilon® PVDF membranes (Millipore, IPVH00010). After blocking in blocking buffer (Epizyme, PS108P), the membranes were incubated with primary antibodies (diluted by primary antibody dilution buffer (Epzyme, PS114)) overnight at 4 °C, then washed for three times and incubated with HRP-conjugated secondary antibodies (diluted by 1xTBST (Epzyme, PS103S, should be diluted into 1x)). Samples were incubated with chemiluminescent HRP substrate (Thermo Fisher Scientific, WBKLS0500), and related signals were detected by MiniChemi 610 Plus (Biogp). The following antibodies were used: β-catenin (BD Pharmingen, 610153, 1: 5000), GAPDH (Proteintech, 60004-1-IG, 1:2000), β-actin (Epzyme, LF202,1:1000), HRP-donkey-a-mouse (JIR, 715-035-150, 1:5000), and HRP-goat a rabbit IgG (JIR, 111-035-047, 1:5000).

AAV injection

The AAVs used in this study were purchased from Taitool Biotechnologies company (Shanghai, China). Briefly, AAVs were produced with cis-plasmids containing the full TBG promoter, which is specifically active in hepatocytes, and Cre expression is under the control of TBG promoter, replication incompetent AAV2/8-TBG-Cre virus (AAV8-TBG-Cre, S0657-8-H5) was packaged and purified before application to mice. Mice were injected intraperitoneally at 1x 10¹¹ genome copies of the virus per mouse. The above newly generated virus as well as targeting plasmids will be provided upon request.

Statistical analysis

Statistical analyses were performed using GraphPad Prism (Version 9.5.1). All the continuous variables were expressed as means ± standard error of the mean (SEM). One-way ANOVA was used to detect statistical significance between three experimental groups. The statistical difference between the two experimental groups was determined using the unpaired student’s t-test. The probability (P) values of less than 0.05 were considered statistically significant.

Acknowledgements

This study was supported by the National Key Research & Development Program of China (2023YFA1800700, 2022YFA1104200, 2023YFA1801300, 2020YFA0803202), National Natural Science Foundation of China (82088101, 32370897, 32100648, 32370783, 32100592), CAS Project for Young Scientists in Basic Research (YSBR-012), Shanghai Pilot Program for Basic Research-CAS, Shanghai Branch (JCYJ-SHFY-2021-0), Research Funds of Hangzhou Institute for Advanced Study (2022ZZ01015, B04006C01600515), Shanghai Municipal Science and Technology Major Project, and the New Cornerstone Science Foundation through the New Cornerstone Investigator Program and the XPLORER PRIZE.

Author contributions

M.S., J.L., and B.Z. designed the study, analyzed the data, and wrote the manuscript; X.L., K.L., W.P., W.W., S.Z., and H.Z. bred the mice and performed experiments; K.O.L. edited the manuscript and provided intellectual input; B.Z. conceived, supervised, and organized the study.

Declaration of interests

The authors declare no competing interests.

Data availability

All data included in this study are included in the manuscript. The raw data are available upon request.