Tamoxifen-independent recombination of reporter genes limits lineage tracing and mosaic analysis using CreERT2 lines
The CreERT2/loxP system is widely used to induce conditional gene deletion in mice. One of the main advantages of the system is that Cre-mediated recombination can be controlled in time through Tamoxifen administration. This has allowed researchers to study the function of embryonic lethal genes at later developmental timepoints. In addition, CreERT2 mouse lines are commonly used in combination with reporter genes for lineage tracing and mosaic analysis. In order for these experiments to be reliable, it is crucial that the cell labeling approach only marks the desired cell population and their progeny, as unfaithful expression of reporter genes in other cell types or even unintended labeling of the correct cell population at an undesired time point could lead to wrong conclusions. Here we report that all CreERT2 mouse lines that we have studied exhibit a certain degree of Tamoxifen-independent, basal, Cre activity. Using Ai14 and Ai3, two commonly used fluorescent reporter genes, we show that those basal Cre activity levels are sufficient to label a significant amount of cells in a variety of tissues during embryogenesis, postnatal development and adulthood. This unintended labelling of cells imposes a serious problem for lineage tracing and mosaic analysis experiments. Importantly, however, we find that reporter constructs differ greatly in their susceptibility to basal CreERT2 activity. While Ai14 and Ai3 easily recombine under basal CreERT2 activity levels, mTmG and R26R-EYFP rarely become activated under these conditions and are therefore better suited for cell tracking experiments.
KeywordsCreERT2 Cre/loxP system Lineage tracing Mosaic analysis Tamoxifen-independent recombination Reporter-gene
Genetic studies in mice have revolutionized biological research and have been fundamental to develop animal models for the study of human diseases. Although there are obvious differences between mice and humans (Perlman 2016), mouse-based animal models are often the first choice and allow us to investigate the mechanisms of disease initiation and progression in great detail (Nguyen and Xu 2008). A combination of factors places the mouse at the forefront as a genetic model organism. The mouse is closer related to humans than other frequently used non-mammalian model organisms such as worms and flies and thus far better suited to investigate the complex physiological systems that mammals share. Moreover, the mouse genome is completely sequenced (Mouse Genome Sequencing Consortium et al. 2002) and an ever-expanding repertoire of molecular and genetic techniques allow sophisticated experimental design.
A breakthrough for the mouse as a model system came with the introduction of techniques that allowed the creation of knock-in and knock-out mice. The ability to activate or inactivate a gene of interest has allowed scientists to experimentally test hypotheses regarding the function of specific genes and to determine their role during development, physiology and in pathological settings. While we will continue to learn from conventional knock out techniques, the approach has its shortcomings. The global deletion of genes that are essential for embryonic development is often only partially informative. While the function of such genes can be studied in the embryo, it is often difficult to distinguish primary from secondary effects and it remains unknown if such genes have roles later in development or in homeostasis (Turgeon and Meloche 2009).
Reporter gene expression in the absence of Tamoxifen induction reveals inherent background recombination of the CreERT2/loxP system
Fluorescent reporter genes can become activated by background Cre activity in several CreERT2 lines
Reporter constructs differ in their susceptibility to basal CreERT2 activity
To corroborate that the divergence in fluorescent cell numbers between reporter lines was the result of recombination events, we designed PCRs to detect the presence of recombined DNA in the Ai14 and R26R-EYFP reporters, respectively (see methods for detail). We generated mice that simultaneously carried the Ai14 and R26R-EYFP reporters, and mice that carried both reporters and Cdh5-CreERT2. This approach ensured that both reporters were exposed to the same amounts of basal CreERT2 activity. On these mice we subsequently run PCRs to detect recombination events prior to and after tamoxifen induction (Fig. 5c). After 38 PCR cycles, abundant product was obtained for both the Tamoxifen-treated Ai14 and R26R-EYFP samples, while in the absence of Tamoxifen only PCR product for the Ai14 reporter could be detected. To dismiss the possibility that the absence of PCR product in the R26R-EYFP sample was the result of low PCR efficiency we repeated the PCRs with increased numbers of cycles in order to improve the yield. After 41 cycles, the amount of PCR product increased in both the Tamoxifen-treated and un-treated Ai14 sample. In contrast, we only obtained a faint band in the non-Tamoxifen-treated R26R-EYFP sample, despite the fact that the product was strongly amplified in the Tamoxifen-treated sample. In the absence of Cdh5-CreERT2, no product was detected in any PCR. The fact that the difference in amount of PCR product between Tamoxifen-treated versus non-treated samples is substantially higher for R26R-EYFP than for Ai14, strongly suggests that Ai14 recombines more easily than R26R-EYFP when exposed to the same degree of basal CreERT2 activity.
Basal CreERT2 leakage levels may hamper lineage tracing experiments
Taken together, our results show that reporter genes can differ widely in their susceptibility to basal CreERT2 leakage and that Ai14 and Ai3 should be used with caution in lineage tracing, pulse-chase or mosaic analysis experiments.
Mouse lines and Tamoxifen administration
Three CreERT2 lines were utilized: Cdh5(PAC)-CreERT2 (Pitulescu et al. 2010), Pdgfb-CreERT2 (Claxton et al. 2008) and Prox1Cre ERT2 (Bazigou et al. 2011); as well as four fluorescent reporter lines: Ai14 and Ai3 (Madisen et al. 2009), mTmG (Muzumdar et al. 2007), and R26R-EYFP (Srinivas et al. 2001).
Induction of postnatal mice was performed by oral gavage to the lactating females, with a daily dose of 100 μl 20 mg/ml Tamoxifen (T5648, Sigma-Aldrich) from P0 to P2. Tamoxifen was first diluted in 10% ethanol and further diluted in 90% corn oil (C8267, Sigma-Aldrich).
Induction of adult mice was performed by oral gavage of 100 μl 20 mg/ml Tamoxifen with two doses on alternating days. Both male and female mice were used for our analyses.
Microvascular fragment isolation, EC culture and immunohistochemistry
Primary endothelial cells from microvascular fragments were isolated and cultured as described before (Niaudet et al. 2015). Briefly, brains from P7 mice were dissected and minced with 1% collagenase (Sigma C6885), 1% penicillin/streptomycin (P/S, Life Technologies 15140-122) DMEM (low glucose, pyruvate; Gibco 31885023) for 20 min while stirring at 300 rpm and 37 °C and neutralized with 20% fetal bovine serum (FBS), 1% PS, DMEM. The suspension was filtered through a 70 µm cell strainer (BD Falcon) and centrifuged at 520 g for 5 min. The cells were resuspended in 1% PS, 0.5 mg/mL heparin (Sigma H3149), DMEM and incubated in rotation at room temperature for 30 min with Dynabeads (Invitrogen) previously coated with rat-anti mouse PECAM1 (BD Pharmingen#553370) antibody. Bead were dissociated with Tryp-LE 10x (Life Technologies). EC were cultured in ECGM2 medium (Promo Cell #C22011) at 37 °C, 5% CO2.
EC in culture were fixed 5 min in 3% PFA (paraformaldehyde) at room temperature and then blocked during 2 h at room temperature in 5% normal donkey serum, 0.2% BSA (bovine serum albumin), 0.1% Triton X-100, PBS (CaCl2, MgCl2). Cells were incubated overnight at 4 °C with goat anti-VE-Cadherin antibody (Santa Cruz, sc-6458, 1:500) in 0.1% BSA, 0.05 Triton X-100, PBS. After washes in PBS, cells were incubated with donkey anti-goat 488 during 3 h at room temperature, followed by incubations with 0.1 μM 647P-conjugated phalloidin (Promocell, PK-PF647P-7-01) for 30 min, and 1 μg/ml Hoechst (Life Sciences, H3570) for 15 min.
Eyeballs of P7 or P8 mice were fixed in 4% formaldehyde at 4 °C during 2 h and then washed in PBS and dissected. Retinas were permeabilized and blocked at 4 °C overnight in PBS, 1% BSA, 0.5% Triton X-100, 5% Normal donkey serum (Jackson ImmunoResearch). Ai3 and Ai14 retinas were incubated with isolectin B4 conjugated with Alexa Fluor-647 (IB4, ThermoFisher Scientific I32450, 1:200) in PBlec: PBS, 0,1 mM CaCl2, 0,1 mM MgCl2, 0,1 mM MnCl2, 1% Triton X-100 pH 6.8 for 4 h at room temperature. R26R-EYFP and mTmG retinas were also incubated with chicken anti-GFP (Abcam ab13970, 1:500) during that time. This was followed by three 30′ washes in PBS, 0.5% BSA, 0.25% Triton X-100. R26R-EYFP and mTmG retinas were then incubated overnight at 4 °C with the Donkey anti-chicken 488 (Jackson ImmunoResearch, 703-545-155) in PBS, 0.5% BSA, 0,25% Triton X-100 and washed again three times 30′ in PBS. Retinas were mounted with ProLong Gold Antifade Mountant (Life Technologies, P36930) and imaged on a confocal laser-scanning microscope (Leica TCS SP8) with a 10X magnification objective.
Brains were fixed in 4% formaldehyde for 4 h at 4 °C upon dissection, and cut in 50 μm thickness sections on a vibratome. Brain sections were blocked and permeabilized at 4 °C overnight in PBS, 1% bovine serum albumin (BSA), 0,5% Triton X-100, 5% Normal donkey serum. Ai3 and Ai14 brain sections were incubated with Alexa Fluor647-conjugated IB4 (1:200) in PBlec during two days at 4 °C. In addition, R26R-EYFP and mTmG were also incubated with chicken anti-GFP. Three washes of one hour were performed in PBS, 0,5% BSA, 0,25% Triton X-100. Afterwards, R26R-EYFP and mTmG brain sections were incubated over two days at 4 °C with Donkey anti-chicken 488 and washed again, while no additional incubation was performed on Ai3 and Ai14 brains. Sections were mounted with ProLong Gold Antifade Mountant and imaged on a confocal laser-scanning microscope (Leica TCS SP8) with a 10X magnification objective.
PCR analysis of recombination efficiency
The following PCR primers were used to detect the recombined “floxed” R26R-EYFP reporter: P1: CCAGGGTTTCCTTGATGATGTC and P2: GTGGCGGATCTTGAAGTTCAC. These primers anneal upstream and downstream from the floxed site and can amplify two different DNA fragments: a “full length” region including the floxed sequence (3444 base pairs), or a shorter fragment when recombination has occurred (775 bp). The PCR programs for R26R-EYFP consisted of 38 or 41 cycles with 30′’ denaturalization at 94 °C, 45’’ annealing at 53 °C and 45’’ extension at 72 °C. This program favors the generation of the 775 bp band and does not produce a 3444 bp fragment.
For the Ai14 reporter we used: P1: GGTTCGGCTTCTGGCGTGTGACC and P2: AAGGCCGGCCGAATTCGATCTAGC. These primers also bind upstream and downstream from the floxed Ai14 region, and can amplify the full-length region of 1162 bp or a recombined fragment of 291 bp. The PCR programs for Ai14 were as follows: 38 or 41 cycles of 30’’ denaturalization at 94 °C, 45’’ annealing at 60 °C and 30’’ extension at 72 °C. These programs favor the appearance of the recombined band; however, the full-length fragment can also be detected (not shown). To determine the size of the PCR products we used the GeneRuler 50 bp DNA ladder (ThermoFisher Scientific).
Retina and brain reporter-recombination analysis
The number of cells expressing the fluorescent reporter was measured with the plugin “Analyze particles” on Fiji software, using images of the whole retina or brain to count the number of cells that were reporter-positive. For retinas and brains with a very high recombination, this method was not valid due to the overlap of reporter-positive cells forming a continuum. In these cases, we measured the average size of an individual cell and divided all the reporter-positive area by the size of a cell to estimate the number of recombinant cells.
Whole-mount immunofluorescence of skin samples
Whole mount tissue (ear skin/embryonic skin) was fixed in 4% paraformaldehyde for 2 h at room temperature, permeabilized in 0.3% Triton X-100 in PBS (PBST) for 10 min and blocked for 2 h in PBST + 3% milk. Samples were then incubated with primary antibodies at 4 °C overnight in blocking buffer, followed by several washes with PBST and incubation with secondary antibodies for 2 h at room temperature. Stained samples were washed and mounted with Mowiol.
The following primary antibodies were used: rat anti-mouse PECAM-1 (553370, BectonDickinson), rat anti-mouse LYVE-1 (103-PA50AG, Reliatech), goat anti-mouse Nrp2 (AF567, R&D Systems) and rabbit anti-GFP (A11122, Thermo Fisher). Secondary antibodies conjugated to Cy3, Alexa Fluor 488 or 647 were obtained from Jackson ImmunoResearch.
The possibility to control the timing of recombination events via Tamoxifen administration has made the CreERT2/loxP system enormously popular. Apart from its wide usage in conditional expression or deletion of target genes, it is progressively used in combination with floxed fluorescent reporters for cell tracking experiments such as lineage tracing, or mosaic analysis. The usefulness of the CreERT2/loxP system depends on its faithful expression. Any expression in cell populations other than the desired ones or an expression in even the correct cell population at the wrong time can lead to erroneous conclusions. Along those lines, Tamoxifen-independent Cre activity and subsequent activation of reporter genes has previously been reported for individual CreERT2 lines (Papoutsi et al. 2015; Kristianto et al. 2017), but it has remained unclear if the reported examples are rare exceptions or represented a general problem. In this study, we analyzed three CreERT2 lines: Cdh5(PAC)Cre-ERT2, Pdgfb-CreERT2, and Prox1-CreERT2 and find that all exhibit a certain degree of basal CreERT2 activity. Our observations indicate that sequestration of CreERT2 to the cytoplasm is not 100% efficient and that leakage of CreERT2 into the nucleus occurs stochastically. The frequency of these events seems to correlate with the abundance of CreERT2 molecules in the cell. In addition, the amount of reporter activation also seems to depend on the reporter line. Here we compared the four commonly used reporter lines: Ai14, Ai3, mTmG and R26R-EYFP, and find that Ai14 and Ai3 become substantially easier recombined than mTmG and R26R-EYFP. One of the possible reasons for this differential reporter activation could lie in the distance between the loxP sites, which has been shown to have an effect on recombination efficiency (Zheng et al. 2000). The loxP sites flanking the STOP codon in the mTmG and R26R-EYFP reporter constructs are further apart than in the other two lines (Fig. 1b), conferring them a higher recombination threshold. Another factor that may affect recombination is chromatin state, which can alter the accessibility of Cre to the loxP sites (Vooijs et al. 2001).
It could further be hypothesized that differences in promoter strength between the investigated reporters might influence the amount of cells being labeled. Looking at Ai14 and R26R-EYFP this appears, at first sight, a plausible explanation, since Ai14 contains a synthetic CAG promoter, which is substantially stronger than the native Rosa26 promoter of R26R-EYFP (Chen et al. 2011). However, the argument seems less valid in light of the fact that the same CAG promoter is also present in the mTmG reporter, which labels substantially less cells than Ai14.
Differences in the brightness of the fluorescent proteins can neither explain the discrepancies in the amount of labeled cells. While tdTomato is a brighter fluorescent protein than EYFP, equally many cells are labeled by the Ai14 and the Ai3 reporters, yet Ai3 and R26R-EYFP, both of which express EYFP, differ substantially in the amount of cells that become labeled as a consequence of basal CreERT2 activity. In addition, throughout this study, EYFP expression was always amplified with an antibody staining which resulted in a uniformly strong EYFP signal in cells that had undergone recombination. Thus, while one might expect that promoter strength or the brightness of the reporter fluorescent protein could have an effect on the amounts of cells being labeled; neither of those factors seems to have a major impact.
In order for lineage tracing to be reliable, it is key that the labeling approach only marks the desired cell population and their progeny. For this reason, CreERT2 leakage and subsequent reporter activation events during earlier developmental stages should be taken into consideration. The progeny of cells in which the promoter controlling CreERT2 expression was temporarily expressed will inherit the activated reporter. One such example is the labelling of microglia cells in Cdh5(PAC)Cre-ERT2, Ai14 mice. Without appropriate controls, this finding could have been erroneously interpreted as a transdifferentiation of microglia from endothelial cells.
For mosaic analysis, the reporter expression must, in addition, be as closely correlated to the gene deletion as possible. In such experiments, reporter expression should be crosschecked with immunohistochemistry to confirm the loss of the gene of interest in the labeled cell population. In cases where reliable immunostaining cannot be achieved, such correlation could be based on a cell-autonomous phenotype (Laviña et al. 2018). However, with the introduction of the ifgMosaic mice (Pontes-Quero et al. 2017), technically superior options now exist. In this approach, gene recombination and reporter activation are fully linked through bicistronic expression.
Open access funding provided by Uppsala University. We would like to thank Ralf Adams for the Cdh5(PAC)Cre-ERT2 line and Marcus Fruttiger for the Pdgfb-CreERT2 line. We thank Pia Peterson, Jana Chmielniakova, Cecilia Olsson, Hélène Leksell, Sofie Sjöberg and Henrik Ortsäter for technical assistance.
Conceptualization: T.M., K.G. Methodology and analysis: A.A., I.M., N.D., T.M., K.G. Writing: A.A., K.G. Funding acquisition: C.B., T.M., K.G.
This work was supported by the Swedish Cancer Foundation [CAN2015/771 to C. Betsholtz, CAN 2016/535 to TM], The Swedish Research Council [VR2015-00550 to C. Betsholtz, 542-2014-3535 to TM], the European Research Council [2011-294556 to C. Betsholtz, ERC-2014-CoG-646849 to TM], Knut och Alice Wallenbergs Stiftelse [2012.0272 to C. Betsholtz, 2015.0030 to CB and TM], Fondation Leducq [14-CVD-02 to C. Betsholtz and K. Gaengel.], the Wenner-Gren Foundation [to C. Betsholtz and K. Gaengel.].
Compliance with ethical standards
Conflict of interest
The authors declare that they have no conflict of interest.
Animal housing, as well as the experiments performed, were in accordance with Swedish legislation and were approved by the local animal ethics committees prior to experimentation. The protocols included in this study were approved by the Uppsala Committee on the Ethics of Animal Experiments (permit numbers C115/15, C111515/16, C130/15). All efforts were made to minimize animal suffering. Both female and male mice were used. This article does not contain any studies with human participants performed by any of the authors.
- Chen C-m, Krohn J, Bhattacharya S, Davies B (2011) A comparison of exogenous promoter activity at the ROSA26 locus using a PhiC31 integrase mediated cassette exchange approach in mouse ES cells. PLoS ONE 6(8):e23376. https://doi.org/10.1371/journal.pone.0023376 CrossRefPubMedPubMedCentralGoogle Scholar
- Fonseca MI, Chu S-H, Hernandez MX, Fang MJ, Modarresi L, Selvan P, MacGregor GR, Tenner AJ (2017) Cell-specific deletion of C1qa identifies microglia as the dominant source of C1q in mouse brain. J Neuroinflammation 14(1):48. https://doi.org/10.1186/s12974-017-0814-9 CrossRefPubMedPubMedCentralGoogle Scholar
- Hall B, A Limaye, and AB Kulkarni (2009) Overview: generation of gene knockout mice. Curr Protoc Cell Biol Chapter 19 (September): Unit 19.12 19.12.1–17. https://doi.org/10.1002/0471143030.cb1912s44
- Laviña B, Castro M, Niaudet C, Cruys B, Álvarez-Aznar A, Carmeliet P, Bentley K, Brakebusch C, Betsholtz C, Gaengel K (2018) Defective endothelial cell migration in the absence of Cdc42 leads to capillary-venous malformations. Development 145(13):dev161182. https://doi.org/10.1242/dev.161182 CrossRefPubMedGoogle Scholar
- Zhang Y, Ulvmar MH, Stanczuk L, Martinez-Corral I, Frye M, Alitalo K, Mäkinen T (2018) Heterogeneity in VEGFR3 levels drives lymphatic vessel hyperplasia through cell-autonomous and non-cell-autonomous mechanisms. Nat Commun 9(1):1296. https://doi.org/10.1038/s41467-018-03692-0 CrossRefPubMedPubMedCentralGoogle Scholar
Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.