Triple-cell lineage tracing by a dual reporter on a single allele

Genetic lineage tracing is widely used to study organ development and tissue regeneration. Multicolor reporters are a powerful platform for simultaneously tracking discrete cell populations. Here, combining Dre-rox and Cre-loxP systems, we generated a new dual-recombinase reporter system, called Rosa26 traffic light reporter (R26-TLR), to monitor red, green, and yellow fluorescence. Using this new reporter system with the three distinct fluorescent reporters combined on one allele, we found that the readouts of the two recombinases Cre and Dre simultaneously reflect Cre+Dre−, Cre−Dre+, and Cre+Dre+ cell lineages. As proof of principle, we show specific labeling in three distinct progenitor/stem cell populations, including club cells, AT2 cells, and bronchoalveolar stem cells, in Sftpc-DreER;Scgb1a1-CreER;R26-TLR mice. By using this new dual-recombinase reporter system, we simultaneously traced the cell fate of these three distinct cell populations during lung repair and regeneration, providing a more comprehensive picture of stem cell function in distal airway repair and regeneration. We propose that this new reporter system will advance developmental and regenerative research by facilitating a more sophisticated genetic approach to studying in vivo cell fate plasticity.

Mammalian organs are complex, dynamic tissues whose maintenance and repair depend on diverse stem/progenitor cell populations. During multiple types of organogenesis, such as heart, lung, and intestine, one stem/progenitor cell population can contribute to several cell types. For example, the epicardium and endocardium of the heart (1-4), basal cells of the lungs (5-7), and Lgr5 ϩ stem cells of intestinal crypts (8) can give rise to multiple cell types during homeostasis, growth, and regeneration. Some differentiated cell types can also originate from multiple stem/progenitor cell sources. For instance, during heart development, coronary vascular endothelial cells, fibroblasts, pericytes, and adipocytes are derived from both epithelial and endocardial cells, albeit with different proportions of contribution (9 -12). Additionally, cardiomyocytes in the developing heart are derived from both the primary and secondary heart field (13,14). Multiple different epithelial cells in the distal airway are derived from club cells or alveolar type 2 cells and bronchioalveolar stem cells (BASCs) 2 after lung injury (15,16). Fate mapping of multiple cellular origins for their contribution of discrete cell lineages provides critical information for understanding tissue homeostasis renewal and regeneration after injury.
The ability to study multiple cell populations in one tissue is important to understand the complex biological process underlying development, tissue homeostasis, and regeneration. Genetic lineage tracing mediated by DNA site-specific recombination systems has been widely used for cell lineage and fate studies (17). Similar to the Cre-loxP system, multiple site-specific recombination systems, such as Flpe-Frt, Dre-rox, and Nigri-nox, have been developed and used for cell fate tracing studies (17)(18)(19)(20)(21). Two types of reporter systems have been developed and used for cell fates analysis. One type is conventional single-color reporter systems, such as Rosa26-tdTomato, Rosa26-LacZ, or Rosa26-GFP, which are used to track one cell lineage in one color (22)(23)(24), or multicolor systems, such as Brainbow or Confetti, used for clonal analysis of one cell lineage (25,26). Another type is multicolor reporter systems that combine two or more recombination systems for genetic targeting of more than one cell population (usually two). Intersectional dual-reporter systems, such as RC::Fela, R26::FLAP, RC::RLTG, and R26 NZG , can be used to label two distinct cell populations (27)(28)(29)(30). In addition, sequential and exclusive double-reporter systems are also explored for labeling specific cell types more precisely (31)(32)(33). However, because of the exclusive nature of recombination design, these genetic tools only allow simultaneous labeling of two (sub)populations in tissue. Development of a multicolor reporter system capable of labeling three distinct cell populations by noninterfering recombination is useful cro ARTICLE for studying the behavior of more diverse cell types simultaneously in vivo.
In this study, we generated a new dual genetic system that incorporates both the Cre-loxP and Dre-rox recombination systems. Different from previous dual-recombinase reporters that reflect combination readout of Cre/Dre recombinases, the current reporters are parallel for two recombination readouts. Thus, this new dual system has the capacity to label three distinct cell populations simultaneously: Dre ϩ Cre Ϫ , Dre Ϫ Cre ϩ , and Dre ϩ Cre ϩ cell populations. As proof of principle, we traced cardiomyocytes and endothelial cells simultaneously during heart development. Additionally, we used the system to label three stem cell populations (club cells, AT2 cells, and BASCs) in the lungs at homeostasis and to track their cell fate after injury. Our new mouse reporter R26-TLR extents the scope of cell type labeling with one reporter allele, which could be broadly used for diverse cell origin and cell fate studies in development, disease, and regeneration.

Generation and characterization of R26-TLR
The Cre-loxP system is a widely used site-specific recombinase-based system. Like the Cre-loxP system, Dre recombinase specifically targets its recombination site rox (19). For genetic lineage studies, cell-specific promoter-derived Cre (or Dre) removes a loxP-or rox-flanked transcriptional stop cassette (stop) after the Rosa26 promoter, leading to permanent constitutively active expression of the following reporter in Cre-or Dre-expressing cells and their descendants (4). To genetically label multiple cell lineages, we generated a new lineage tracing reporter system that incorporates both the Cre-loxP and Drerox recombinations. This reporter was generated by knocking the CAG-rox-stop-rox-ZsGreen-WPRE-pA-Frt-Neo-Frt sequence into exon1 of the Rosa26 gene locus and knocking the insulator-CAG-loxP-stop-loxP-tdTomato-WPRE-pA sequence into exon2 of the Rosa26 gene locus through two homologous recombinations (Fig. 1A). Whether ZsGreen-WPREpoly(A) after the Dre-rox recombination event can affect tdTomato gene expression by Cre-loxP recombination in the same allele is unknown. We therefore intentionally inserted the insulator between them to block the potential influence between them and ensure that the readouts of Dre-rox and Cre-loxP are specific and efficient. The insulator sequence is in Table S1. Dre and Cre recombinases driven by two different promoters recombine rox or loxP sites, resulting in ZsGreen or tdTomato reporter expression, respectively. Notably, when the two promoters are both active in a cell, the cell expresses ZsGreen and tdTomato at the same time, yielding yellow fluorescence (Fig. 1, B and C). We named this reporter line Rosa26 traffic light reporter (R26-TLR).

Tracing distinct cell populations simultaneously in the heart
To further prove that R26-TLR can be used to trace different/ diverse cell populations simultaneously, we crossed the R26-TLR mouse line with the Tnni3-Dre and Tie2-Cre mouse lines, which specifically target cardiomyocytes and endothelial cell lineages, respectively (32,35). As shown in the design, Dre expression driven by the Tnni3 promoter resulted in ZsGreen gene expression, and Tie2-derived Cre led to tdTomato gene expression (Fig. 2, A and B). For analysis, we collected hearts from E16.5 embryos of three genotypes: Tnni3-Dre;R26-TLR, Tie2-Cre;R26-TLR, and Tnni3-Dre;Tie2-Cre;R26-TLR (Fig.  2C). Whole-mount epifluorescence imaging showed ZsGreen ϩ but not tdTomato ϩ signals detected in Tnni3-Dre;R26-TLR hearts. In contrast, tdTomato ϩ but not ZsGreen ϩ signals in vascular patterns were detected in Tie2-Cre;R26-TLR hearts. As expected, both ZsGreen ϩ and tdTomato ϩ signals were detected in Tnni3-Dre;Tie2-Cre;R26-TLR triple-positive hearts (Fig. 2D). To further verify the labeling specificity, we performed immunostaining on heart sections with the cardiomyocyte marker Tnni3 and the endothelial cell marker Pecam. Immunostaining for ZsGreen, tdTomato, and Tnni3 on Tnni3-Dre;Tie2-Cre;R26-TLR triple-positive heart sections showed that over 99% of all cardiomyocytes were ZsGreen ϩ tdTomato Ϫ (Fig. 2, E and F). We also performed immunostaining for ZsGreen, tdTomato, and Pecam on triple-positive heart sections and detected that more than 99% of endothelial cells were tdTomato ϩ ZsGreen Ϫ (Fig. 2, G and H). The distributions of ZsGreen ϩ cardiomyocytes and tdTomato ϩ endothelial cells were interleaved and connected to each other. Furthermore, we could hardly detect any yellow signals, indicating that the promoters of Tnni3 and Tie2 were less likely to be active in the same cell types (Fig. 2, E and G). Additionally, immunostaining data from the Tnni3-Dre;R26-TLR and Tie2-Cre;R26-TLR double-positive line confirmed no cross-talk between these two Dre-rox and Cre-loxP recombinations (Fig. S1, A and B). Together, these data demonstrate that R26-TLR can be used for simultaneous labeling of distinct cell populations.

Triple-cell lineage tracing Genetic tracing of diverse epithelial cell types simultaneously in the lungs
The lungs are a multifunctional organ consisting of pulmonary vasculature and a respiratory epithelial system that are essential for mammalian survival (36). Multiple stem/progenitor cells participate in the maintenance of lung function during homeostasis and repair (37). In the respiratory system, diverse resident epithelial stem cell populations have been identified from the proximal to the distal airway, which consists of four regions: the trachea, bronchi, bronchioles, and alveoli. The tracheal and bronchial epithelia predominantly consist of basal cells, club cells, ciliated cells, and goblet cells, whereas bronchioles mainly contain ciliated cells, club cells, and neuroendocrine cells (38). The alveolar epithelium includes squamous alveolar type I (AT1) and cuboidal alveolar type II (AT2) cells. These types of epithelial cells are vital for maintenance of the respiratory tract, and they are maintained by self-renewal and differentiation from stem cells (37,39), e.g. club cells give rise to ciliated cells, and AT2 cells generate AT1 cells. Notably, a multipotent stem cell population has been reported to reside at bronchioalveolar duct junctions (BADJs); these BASCs coexpress the club cell maker CC10 and the AT2 cell maker SPC (40). In vivo genetic fate mapping studies have recently demonstrated that CC10 ϩ SPC ϩ BASCs can proliferate and differentiate into bronchiole epithelial club cells and ciliated cells and also alveolar epithelial AT1 and AT2 cells in distinct lung injury models (15,16).

Tracing diverse epithelial cell types simultaneously in lung repair and regeneration
To monitor diverse epithelial cell behaviors labeled by R26-TLR after lung injury, we used two lung injury models: the naphthalene-induced bronchiolar injury model and the bleomycin-induced alveolar injury model. To study the fates of club cells, AT2 cells, and BASCs simultaneously in the bronchiolar injury model, Sftpc-DreER;Scgb1a1-CreER;R26-TLR triple-positive mice were treated with tamoxifen at 7 weeks and then with naphthalene or vehicle (corn oil) 10 days after tamoxifen induction (Fig. 4A). 8 weeks after injury, lung tissues were collected, and sectional immunostaining of ZsGreen, tdTomato, and the club cell maker CC10 or the ciliated cell maker Acetylated tubulin showed that both ZsGreen ϩ tdTomato ϩ BASCs and ZsGreen Ϫ tdTomato ϩ club cells expanded and contributed to bronchiole epithelial cells, including club cells and ciliated cells, for repair of the distal airway (Fig. 4, B, D, and F). The alveolar region was largely undamaged, and we did not detect a significant contribution of ZsGreen ϩ tdTomato Ϫ AT2 cells to ZsGreen Ϫ tdTomato ϩ club cells by naphthalene-induced injury (Fig. 4, B and D). Interestingly, ZsGreen ϩ tdTomato ϩ BASCs were mainly responsible for repair of more terminal bronchioles. Quantification data showed a significant increase in bronchiolar ZsGreen ϩ tdTomato ϩ CC10 ϩ cells (28.92 Ϯ 2.07 naphthalene versus 3.72 Ϯ 0.39 vehicle) and ZsGreen ϩ tdTomato ϩ acetylated tubulin ϩ cells (4.80 Ϯ 0.66 naphthalene versus 0.11 Ϯ 0.07 vehicle) after bronchiolar injury (Fig. 4, C and E).
Next we performed alveolar injury to study the fates of these three types of epithelial cells (Fig. 4G). After bleomycin-induced alveolar injury, ZsGreen ϩ tdTomato ϩ BASCs mainly gave rise to SPC ϩ AT2 cells and T1a ϩ AT1 cells surrounding BADJ regions (Fig. 4, H and J). Quantification data showed a significant increase in alveolar ZsGreen ϩ tdTomato ϩ SPC ϩ cells (49.60 Ϯ 5.87 bleomycin versus 3.96 Ϯ 0.51 vehicle) and ZsGreen ϩ tdTomato ϩ T1a ϩ cells (26.40 Ϯ 1.97 bleomycin versus 0.09 Ϯ 0.03 vehicle) after alveolar injury (Fig. 4, I and K). Conversely, AT2 cells and AT1 cells far from BADJ regions were derived from ZsGreen ϩ tdTomato Ϫ AT2 cells (Fig. 4, H, J, and L). The fates of BASCs after distinct injury models were consistent with previous reports (15,16). Taken together, using lung epithelial cells as an example, the above results demonstrate that R26-TLR can be used to study diverse cell origins and fates during development, disease, and regeneration.

Discussion
In this study, we developed a new genetic tool, R26-TLR, for readout of Cre-loxP and Dre-rox recombination. How different recombination sites are arranged with reporter genes is critical for its use in tracing multiple cell lineages in vivo. The arrangement between recombination sites and reporter genes of R26-TLR is different from all previously reported intersectional dual-reporter systems, such as RC::Fela, R26::FLAP, RC::RLTG, and R26 NZG (27)(28)(29)(30). The model of previous intersectional dual systems is CAG-site1-stop-site1-site2-reporter1-stop-site2-re-porter2. Recombination of recombinase1-site1 results in reporter1 expression. Expression of reporter2 needs recombination of recombinase1-site1 and recombinase2-site2. Because

Triple-cell lineage tracing
of the exclusive nature of two reporters in recombinase-targeted cells, reporter2 can only reflect a subpopulation of report-er1-labeled populations (33). The model of our newly generated R26-TLR is CAG-site1-stop-site1-reporter1-CAG-site2-stop-site2-reporter2, which permits parallel expression of reporters by two different recombination events. We previously reported a dual system, R26-NLR, which was generated based on Nigri-nox and Cre-loxP systems (20). Driven by different cell-specific promoters, e.g. promoter A and promoter B, R26-NLR can be applied to label two distinct A ϩ B Ϫ and A Ϫ B ϩ cell populations simultaneously in the same mouse. However, R26-NLR cannot clearly distinguish the double A ϩ B ϩ cell population from the A ϩ B Ϫ cell population or A Ϫ B ϩ cell population. In this study, because of the different design strategy, R26-TLR integrates the

Triple-cell lineage tracing
advantages of previously reported intersectional reporters and our previous R26-NLR tool. It can be used to label two distinct cell populations and also to label their intersectional subpopu-lations. The new dual R26-TLR system could be used to label all three cell populations: A ϩ B Ϫ , A Ϫ B ϩ , and A ϩ B ϩ (double-positive population). It is a better system than our previous R26-

Triple-cell lineage tracing
NLR system and other previously reported intersectional reporters.
In this study, we used three examples to prove that R26-TLR works in vivo. In the first example, we used three mouse lines, CAG-Dre;R26-TLR, ACTB-Cre;R26-TLR, and CAG-Dre;ACTB-Cre;R26-TLR, to demonstrate that all cells were ZsGreen ϩ tdTomato Ϫ after Dre-rox recombination, ZsGreen Ϫ td-Tomato ϩ after Cre-loxP recombination, or ZsGreen ϩ tdTomato ϩ after both Dre-rox and Cre-loxP recombination, respectively. These data demonstrate that there is no crossreaction between the Dre-rox and Cre-loxP systems. Further, we showed that the Tnni3-Dre;Tie2-Cre;R26-TLR triple-positive line can specifically and efficiently label cardiomyocytes and endothelial cells simultaneously in the heart. In the third example, we used the Sftpc-DreER;Scgb1a1-CreER;R26-TLR triple-positive line to demonstrate that R26-TLR can be used to genetically trace three cell populations, CC10 Ϫ SPC ϩ AT2 cells, CC10 ϩ SPC Ϫ club cells, and the SPC ϩ CC10 ϩ BASCs population, which could not be achieved with previous tools. Their fates could also be clearly tracked after lung injury. In the same mouse, we observed the behavior of three types of cells in the same context after lung injury. BASCs could regenerate distal bronchioles and alveoli near BADJ regions, whereas club cells and AT2 cells could repair other distal airway regions after injury. We use this example to convincingly show that application of R26-TLR yields more comprehensive information than the previously reported intersectional reporters and our previous R26-NLR.
During organogenesis, multiple stem/progenitor cells were heterogeneous and could be identified by expressing more than one gene markers, such as epithelia and mesenchyme. The distribution and cell fate of these diverse cell populations could be well studied using this new tool. In several organs, in the border between two distinct types of epithelia, there existed transitional epithelium that coexpressed bilateral epithelial markers, including the squamous columnar junction of the gastroesophageal junction, cervix junction, and rectum-anus junction (41)(42)(43); BASCs at BADJs of lung epithelium (15,16,40); and limbal stem cells at the narrow zone between the conjunctiva and cornea of the eyes (44,45). Moreover, many double-positive epithelial stem/progenitor cells appeared in addition to transitional epithelium during development. For example, during alveologenesis of the mouse, the epithelial progenitors of the distal tip are Sox9 ϩ Id2 ϩ and responsible for giving rise to bronchiole and alveolar epithelial cells (37,46,47). Axin2 ϩ SPC ϩ AT2 cells were identified as alveolar epithelial progenitors in adult lung (48,49). Basal cells, which can be identified by expression of P63 and Krt5 (double-positive), are multipotent progenitors of lung epithelia (50). On the other hand, mesenchymal cells were also heterogeneous. For instance, embryonic mesenchymal cells can be identified by expressing PDGFRa and PDGFRb, but they were considered to be fibroblasts and pericytes markers in adult tissues, respectively (9,51,52). Whether the expression of PDGFRa and PDGFRb is exclusive or whether the lineage of the PDGFRa ϩ PDGFRb ϩ double-positive mesenchymal population represents a unique cell population that differs from other two single-positive populations remains unknown; their distinct roles in homeostasis and regeneration needed to be investigated further. Because of a lack of related mouse lines, we are still unable to answer these questions by using R26-TLR, but are they are worthy of discussion in the future. We used this new system to simultaneously track three distinct lung epithelial cell populations and examined their discrete contributions to different cell lineages during lung injury and repair. Given the capability of genetic tracing of diverse cell populations without the need of antibody staining, R26-TLR can be used broadly for diverse cell origin and cell fate studies and can also be combined with live imaging for more clear in vivo studies in the future. Moreover, using the strategy of an insulator to permit parallel expression of two recombination results, we can expand them into three or more and, in theory, provide an approach to genetically study multiple cell lineages simultaneously in vivo.
In summary, our newly generated R26-TLR eases the number of mice crossing of different Cre/Dre reporter lines and permits simultaneous tracing of three distinct cell lineages in vivo. Notably, this mouse line extends the capacity of previously reported dual-recombinase-mediated genetic cell labeling and can be used widely to study the origin and fate of diverse cell populations simultaneously in development, disease, and regeneration.

Mice
All animal procedures were reviewed and approved by the ethics committee of the Institutional Animal Care and Use Committee of the Institute of Biochemistry and Cell Biology, Shanghai Institutes for Biological Sciences, Chinese Academy of Sciences. All experiments were performed strictly within the committee's guidelines. The ACTB-Cre, CAG-Dre, Tnni3-Dre, Tie2-Cre, Sftpc-DreER, and Scgb1a1-CreER mice lines were described previously (15,19,32,34,35). The R26-TLR mouse line was created by knocking the CAG-rox-stop-rox-ZsGreen-WPRE-pA-Frt-Neo-Frt sequence into exon1 of the Rosa26 locus and knocking the insulator-CAG-loxP-stop-loxP-tdTomato-WPRE-pA sequence into exon2 of the Rosa26 gene locus using traditional targeted mutation through homologous recombination in embryonic stem cells. The CAG promoter, consisting of the cytomegalovirus enhancer fused to the chicken ␤-actin promoter, is a strong synthetic promoter frequently used to drive high levels of gene expression in mammalian cells, such as mouse cells. Stop is a transcriptional stop cassette. rox and loxP are recombinases recognition sites. WPRE is a woodchuck hepatitis virus posttranscriptional regulatory element for enhancing the stability of RNA transcription. Neo is a screening gene in mouse generation. An insulator is used to block the potentially mutual influence between the two recombination events. ZsGreen and tdTomato are two kinds of fluorescent protein genes. The obtained chimeric mouse lines were crossed to C57BL/6 lines for germline transmission. Generation of R26-TLR was supported by Shanghai Model Organisms Center, Inc. All mice used in this study were kept at C57BL6 backgrounds. The dosage of tamoxifen (Sigma, T5648) treatment was 0.2 mg/g by oral gavage.

Triple-cell lineage tracing Genomic PCR
Genomic DNA was extracted from mouse tails. Briefly, the tail tissues were lysed in the lysis buffer for 8 h at 55°C, and then the mixture was centrifuged at 15,000 rpm for 5 min to obtain a supernatant solution of genomic DNA. Next we used isopropanol to precipitate the DNA and then washed the DNA in 70% ethanol by centrifugation at 15,000 rpm for 3 min. Finally, the DNA was dissolved in double-distilled H 2 O.

Tissue collection and immunofluorescent staining
The immunostaining protocol was performed as described previously (20). Briefly, the tissues were collected after euthanasia of the mice and then fixed in 4% paraformaldehyde (Sigma) for 30ϳ60 min at 4°C according to tissue size. The tissues were washed three times in PBS, and then the tissues were dehydrated at 30% sucrose (dissolved in PBS) at 4°C. After the tissues sank to the bottom of the solution, they were embedded in optimum cutting temperature (Sakura) and stored at Ϫ80°C. 10-m frozen sections were collected on slides. For immunofluorescent staining, the tissue sections were first dried at room temperature and then placed in PBS to remove optimum cutting temperature. Next, the tissue sections were blocked in 5% PBSST (5% donkey serum in PBS and 0.1% Triton X-100 in PBS) for 30 min at room temperature. Then the blocking buffer was discarded, and tissue sections were incubated with primary antibodies at 4°C overnight in the dark. ). An Olympus confocal microscope (FV1200) was used to acquire immunostaining images. Images were analyzed using ImageJ (National Institutes of Health) software.

Bronchiolar injury
Bronchiolar injury was achieved as described previously (15). Naphthalene (Sigma, 84679) was dissolved in sterile corn oil (25 mg ml Ϫ1 ). For bronchiolar injury, Sftpc-DreER;Scgb1a1-CreER; R26-TLR mice were treated with tamoxifen at 7 weeks and then with 250 mg kg Ϫ1 naphthalene or vehicle (corn oil) by naphthalene intraperitoneal injection after 10 days. The lung tissues were collected after 8 weeks of recovery. For analysis, fluores-cence ϩ CC10 ϩ club cell number or fluorescence ϩ acetylatedtubulin ϩ ciliated cell number was quantified at BADJ field.

Alveolar injury
Alveolar injury was induced by bleomycin as described previously (15). Bleomycin (Sigma, B8416) was freshly dissolved in PBS (10 units ml Ϫ1 ) and stored at Ϫ80°C. 10 units ml Ϫ1 bleomycin was diluted to 1 unit ml Ϫ1 with PBS before use. For alveolar injuries, Sftpc-DreER;Scgb1a1-CreER;R26-TLR mice were treated with tamoxifen at 7 weeks and then with 2 units kg Ϫ1 bleomycin or vehicle (PBS) by intratracheal instillation after 10 days. The lungs were collected after 8 weeks. For analysis, fluorescence ϩ SPC ϩ AT2 cells or fluorescence ϩ T1a ϩ AT1 cells were quantified at BADJ fields.

Statistical analysis
The data were acquired from five independent experiments and are presented as mean values Ϯ S.D. Two-sided unpaired Student's t test was used to compare the difference between two groups. p Ͻ 0.001 was considered statistically significant.