A widely used EYFP-based Cre reporter mouse line fails to detect a significant fraction of Cre⁺ T cells

Abstract

The ability to manipulate the mouse genome has contributed heavily to countless discoveries in biology for several decades. The Cre/loxP system enables conditional and inducible deletion of virtually any sequences in the mouse. Because Cre expression does not always predict Cre activity and does not necessarily mirror expression of the endogenous gene(s) whose cis-acting genetic elements were used to construct a given Cre “driver” mouse line, Cre “reporter” mouse lines have been developed in which expression of Cre is signified by expression of one of a variety of fluorescent or otherwise easily detectable proteins, including EYFP, tdTomato, and βgal. It is shown here that one such widely utilized Cre reporter mouse line, which expresses EYFP following Cre-mediated recombination, fails to detect a significant fraction of Cre⁺ T cells, as detected with an otherwise identical tdTomato-expressing Cre reporter. Given the wide use of this and similar Cre reporter mouse lines, these findings have potentially significant implications for a diverse array of studies.

Introduction

Numerous advances in biology have been made possible via the ability to delete specific sequences in the mouse genome. One powerful and highly popular approach to engineering deletions in the mouse genome involves use of the Cre/loxP system,¹ in which mice that have specific sequences flanked by loxP sites in cis (“floxed” genes) are bred to mice in which expression of Cre in cells of interest (Cre “driver” strains) will enable Cre-mediated recombination and deletion of sequences between the loxP sites in the cells of interest. This system is highly versatile and can be used to delete genes or turn on or off various transgenes in an almost limitless variety of ways.

Although quite powerful, interpretation of information derived from Cre/loxP-based approaches is necessarily dependent on the fidelity of Cre expression, and frequently on whether Cre is expressed specifically in all of the cells of interest and only in those cells. It is widely understood that expression of Cre in specific cell types by particular Cre driver strains may not perfectly reflect expression of the endogenous gene whose cis-acting genetic elements were used to construct the Cre driver line. In addition, the efficiency of Cre-mediated recombination can vary widely as a function of the locus, cell type, number of floxed alleles, and other factors. To address this issue, a number of Cre reporter mouse lines have been developed that express fluorescent or otherwise easily detectable proteins in cells in which Cre-mediated recombination has occurred. While clearly not perfect, the use of Cre reporter lines significantly enhances the ability to interpret data developed using Cre/loxP-based approaches. In general, these Cre reporter mouse lines utilize a configuration in which a universally or widely active promoter sequence is separated from the sequence encoding the reporter gene by a loxP-flanked strong transcriptional stop sequence, a so-called “LSL” (lox-stop-lox) arrangement. Cre-mediated recombination removes the transcriptional stop sequence, allowing readthrough and consequent expression of the reporter.

Among the most popular of these Cre reporter mouse lines are those that use EYFP as the reporter. It is reported here that one such line,² available from The Jackson Laboratory (JAX, Bar Harbor, ME, USA) and widely used around the world, fails to detect Cre⁺ cells that are unambiguously detected by an otherwise identical tdTomato-based Cre reporter mouse line.² These findings have important implications for certain types of studies that have relied on EYFP expression as a surrogate for Cre expression.

Materials and methods

Mice

Mice that express Cre recombinase driven by the distal lck (dlck) promoter, which expresses Cre almost exclusively in single positive (SP) thymocytes and mature T cells³ (JAX strain 012837), and Cre reporter mice, which in the presence of Cre recombinase express either EYFP (JAX strain 007903) or tdTomato (JAX strain 007914) from otherwise identical constructs knocked into the Rosa26 locus downstream of a loxP-flanked transcriptional stop sequence,² were obtained from JAX. All mice were bred and maintained in specific pathogen–free facilities in the Northwestern University Feinberg School of Medicine facilities, and all procedures involving mice were approved by the Northwestern Institutional Animal Care and Use Committee and conformed to National Institutes of Health requirements.

Mice with Cre recombinase under the control of the dlck promoter were intercrossed through 2 generations with EYFP Cre reporter mice to produce dlck-Cre/R26E^EYFP/EYFP mice. These dlck-Cre/R26^EYFP/EYFP mice were then intercrossed with (Cre-negative) R26^{tdTomato/tdTomato} mice to produce dlck-Cre/R26^{EYFP/tdTomato} mice.

All mice were used between 8 and 16 weeks old. Both male and female mice were used, with no differences ever observed between male and female mice. All mice were euthanized by CO₂ asphyxiation followed by cervical dislocation.

Flow cytometry

CD4-eFluor 450, CD8-APC, CD19-APC, NK1.1-BV421, and CD3-APC were obtained from eBioscience/Invitrogen and were used at empirically determined optimal concentrations. All staining was carried out on freshly isolated splenocytes or thymocytes on ice for 15 to 30 minutes, followed by one wash and immediate analysis with no fixation. Live/dead gating was performed with a combination of forward and 90-degree light scatter staining. Raw data were generated on a FACSymphony A1 and analyzed with FlowJo software. Gating was performed first on live cells, and then within the live gate on specific subsets. For peripheral T cells, CD4 T cells were CD4⁺CD8⁻ and CD8 T cells were CD8⁺CD4⁻. For thymocytes, gates were set on double negative (DN), double positive (DP), CD4 SP, and CD8 SP in standard CD4 versus CD8 plots. For B cells in the spleen, gates were set on CD19⁺ cells. For NK cells in the spleen, gates were set on NK1.1⁺CD3⁻ cells.

Statistical analysis

Individual experiments included 2 to 4 mice. All results from all mice from any specific set of experiments are included in each figure, with each symbol in each graph representing a single mouse. Cells from different individual mice were never pooled. Differences between groups were analyzed using Mann–Whitney statistical tests and were analyzed within the Prism v10 application. Error bars on all graphs indicate mean ± SD.

Results

A Cre⁺ subset of T cells that express only tdTomato

The Cre reporter mice used in this study harbor constructs inserted into the Rosa26 locus that are identical in all respects, apart from the sequences encoding either EYFP or tdTomato, including the site of integration in the Rosa26 locus, as well as all surrounding cis-acting genetic elements that control gene expression.² Therefore, if these 2 Cre reporters functioned identically, as would be expected, then all peripheral T cells and SP thymocytes would be either EYFP⁺tdTomato⁺ or EYFP⁻tdTomato⁻, with no significant subpopulation that expressed only EYFP or only tdTomato.

Contradicting this prediction, analysis of splenic T cells showed that 100% of mice examined contained a significant subset of both CD4 and CD8 T cells that expressed exclusively tdTomato, in addition to the expected subset that expressed both EYFP and tdTomato (Fig. 1). In contrast, virtually no (<0.2%) CD4 or CD8 T cells expressed only EYFP (Fig. 1). Because these reporter constructs do not “leak,” that is, there is essentially zero expression of either EYFP or tdTomato in the absence of Cre² (Fig. S1), these results show that expression of tdTomato identifies Cre-expressing naive T cells that do not coexpress EYFP.

Figure 1.

A subset of Cre⁺ T cells do not express the EYPF reporter: Spleen cells were analyzed for expression of EYFP and tdTomato on CD4⁺ T cells (A, C) and CD8⁺ T cells (B, D). Representative FACS plots (A, B) and cumulative quantitative data on subsets defined by differential expression of EYFP and tdTomato (C, D) are presented. Results depicted in (C) and (D) represent 13 total dlck-Cre/R26^{EYFP/tdTomato} mice analyzed in 5 separate experiments. ****Different from all other groups by P < 0.0001.

A qualitatively identical pattern of tdTomato and EYFP expression was seen in CD4 SP and CD8 SP thymocytes, with one subset of cells expressing both tdTomato and EYFP and a second subset that was tdTomato⁺EYFP⁻, but no (<1%) corresponding tdTomato⁻EYFP⁺ subset (Fig. 2). However, there was a clear shift in the relative representation of these individual subsets compared to splenic T cells described above. Specifically, the fraction of both CD4 SP and CD8 SP thymocytes that expressed only tdTomato was significantly higher than the respective fraction of peripheral CD4 and CD8 T cells that expressed only tdTomato, with a correspondingly lower fraction of CD4 SP and CD8 SP thymocytes that expressed both tdTomato and EYFP. In addition, CD8 T cells contained a significantly higher fraction of tdTomato⁺ cells that coexpressed EYFP than did CD4 T cells, and CD8 SP thymocytes contained a significantly higher fraction of tdTomato⁺ cells that coexpressed EYFP than did CD4 SP thymocytes (Fig. 2). This difference between CD8 and CD4 peripheral T cells and between CD8 SP thymocytes and CD4 SP thymocytes is consistent with the overall higher penetrance of Cre expression in the CD8 lineage in dlck-Cre mice.³

Figure 2.

A subset of Cre⁺ SP thymocytes do not express the EYFP reporter: Total thymocytes were analyzed for expression of EYFP and tdTomato on CD4 SP thymocytes (A, B) and CD8 SP thymocytes (C, D). Representative FACS plots (A, C) and cumulative quantitative data on subsets defined by differential expression of EYFP and tdTomato (B, D) are presented. Results depicted in (C) and (D) represent 7 total dlck-Cre/R26^{EYFP/tdTomato} mice analyzed in 3 separate experiments. ****Different from all other groups by P < 0.001.

Fidelity of tdTomato expression

To assess whether this discordance between EYFP and tdTomato expression in peripheral T cells and SP thymocytes was due to inappropriate or aberrant expression of tdTomato, and that expression of tdTomato faithfully paralleled the pattern of expression of EYFP under the control of the distal lck promoter, expression of EYFP and tdTomato was also analyzed during earlier stages of T-cell development and in non–T-cell lymphocyte lineages in dlck-Cre/R26^{EYFP/tdTomato} mice.

The original analysis of Cre expression in distal-lck-Cre mice,³ which used a different but structurally similar LSL-containing EYFP-based Cre reporter,⁴ showed that expression of EYFP was confined almost exclusively to SP thymocytes and mature peripheral T cells, with little or no expression in DP and DN thymocytes, <5% expression in NK cells, and no detectable expression in B cells.

This overall pattern was also observed for tdTomato expression in dlck-Cre/R26^{EYFP/tdTomato} mice. In DN thymocytes, 3.19% ± 0.73% expressed tdTomato only, whereas <0.1% expressed either EYFP only or both tdTomato and EYFP (n = 7). In DP thymocytes, 5.01% ± 1.08% expressed tdTomato only, whereas <0.1% expressed either EYFP only or expressed both tdTomato and EYFP (n = 7). For both DN and DP thymocytes, no tdTomato⁺ cells were found in Cre-negative mice (Fig. S2), confirming that tdTomato expression in this small fraction of tdTomato⁺ DN thymocytes and tdTomato⁺ DP thymocytes was due to Cre-mediated recombination.

Expression of tdTomato and EYFP in spleen B cells and NK cells was also analyzed. For B cells (CD19⁺), there were no detectable EYFP⁺ or tdTomato⁺ cells (n = 5), which is completely consistent with the original report.³ For NK cells, defined as NK1.1⁺CD3⁻, which in this study constituted 3.65% ± 0.36% of total spleen cells (n = 5), 3.36% ± 0.57% expressed tdTomato only, 0.047% ± 0.034% expressed EYFP only, and 0.11% ± 0.05% expressed both tdTomato and EYFP (Fig. S3).

Taken together, these data show that the pattern of tdTomato expression closely mirrors that of EYFP expression, albeit with variably higher levels of tdTomato expression. These findings confirm the T-cell specificity of Cre reporter gene expression characteristic of the dlck-Cre driver mouse line, and also show that the existence of a tdTomato⁺EYFP⁻ T-cell subset is not due to aberrant expression of tdTomato in Cre⁻ T cells.

Discussion

Although the Cre/loxP system has been used extensively in mouse biology for decades, a number of possible confounding issues in interpreting data gathered with this system have been noted.^5–9 One such critical issue revolves around the well-understood finding that expression of Cre is not tantamount to recombination mediated by Cre. Although various solutions to this problem have been proposed, one of the more powerful and widely applicable is the use of Cre reporter lines, in which expression of the chosen Cre reporter marks cells with demonstrated Cre recombinase activity. While not perfectly predictive of Cre activity on other floxed loci, as it is well understood that sensitivity to Cre-mediated recombination can vary widely across the genome, the use of Cre reporter lines is nonetheless far superior to having no information at all on Cre expression and activity. Coupled to analysis of reporter expression at the single-cell level, the use of Cre reporter lines clearly makes analysis and interpretation of data gathered in a Cre/loxP system substantially more robust.

The present studies originated from experiments attempting to estimate the rate of “escape mutants,” that is, cells expressing the Cre reporter but not expressing the gene of interest, in a very similar system that also utilized the dlck-Cre driver strain and also sought to express both EYFP and a second gene, both from the Rosa26 locus, both utilizing a standard LSL arrangement. That second gene, while in nearly the identical position in the Rosa26 locus, does not contain the identical cis-acting elements as the EYFP reporter, in contrast to experiments described here. It was therefore unclear whether possible differences in sensitivity and/or efficiency of the 2 Cre targets or the presence of escape mutants was related to differences in surrounding elements and/or other factors. To investigate this further, the experiments described here were carried out using the 2 distinct fluorescent Cre reporters. As mentioned above, these 2 reporter constructs differ only in the sequences that encode the 2 distinct reporters, EYFP and tdTomato; all other features of these reporter alleles, including all cis-acting elements and their precise location in the Rosa26 locus, are completely identical.² This system therefore eliminates any possible effect of differential locus sensitivity or subtle differences in the precise makeup of the reporter constructs.

The key finding is that, contrary to the expectation that there would be coincident expression of both reporters in all cells that expressed either reporter, both CD4 and CD8 SP thymocytes and CD4 and CD8 peripheral naïve T cells, the cell types in which the dlck promoter construct is most active, contained a clear subset of cells that expressed exclusively tdTomato, but essentially no cells that expressed exclusively EYFP. Moreover, this pattern of EYFP versus tdTomato expression was found in 100% of mice analyzed. These findings uncover a previously unknown and unexpected problem with this EYFP-based Cre reporter (and likely others): It fails to detect a significant subset of Cre⁺ cells, here using T cells as the model system.

The molecular basis for the unexpected absence of EYFP in tdTomato⁺ T cells remains obscure. As detailed above, this cannot by definition be due to differences in locus sensitivity, as the 2 reporter constructs are in identical locations within the Rosa26 locus. Similarly, this discrepancy cannot be explained by differences in surrounding sequences, as these are also identical. Furthermore, although direct comparisons are not possible, the levels of expression of EYFP in peripheral CD4 and CD8 cells in the present study are consistent with those reported previously in the original description of the dlck-Cre mouse,³ which used a differently constructed LSL-containing EYFP-based reporter.⁴ This makes it even more unlikely that differences in the surrounding cis-acting genetic elements of similar LSL-containing EYFP reporters can explain the discrepancies in EYFP versus tdTomato expression described here, and suggests that other LSL-containing EYFP reporters are similarly unlikely to detect all Cre⁺ cells. It seems intuitively unlikely that tdTomato-based reporters are inherently more sensitive to Cre-mediated recombination than otherwise identical EYFP-based reporters, as this sensitivity would necessarily map to the tdTomato sequence itself. Finally, if Cre recombinase activity were limiting, this alone would be inadequate to explain why the discrepancy goes only in one direction, that is, allowing for tdTomato⁺EYFP⁻ cells but not the converse.

Regardless of the precise molecular basis for the discrepant expression of EYFP and tdTomato in T cells, the results reported here have clear implications for certain types of experiments. Data derived from this and similar EYFP-based Cre reporters should be interpreted with caution when characterizing novel Cre driver strains. In addition, the results here show that using the EYFP⁻ population of T cells as an internal control for EYFP⁺ T cells is problematical, as the EYFP⁻ subset also contains Cre⁺ T cells. Some types of cell/lineage tracing experiments in which EYFP marking is assumed to be permanent and definitive as to Cre activity may also be more complicated to interpret. In settings of T-cell activation and/or nonsynchronous expression of Cre, as is the case with many inducible systems, this problem may be even more complicated to address, particularly as it remains unclear whether EYFP⁻ cells that are Cre⁺ will remain EYFP⁻ following activation or over long-term persistence in vivo. There are undoubtedly other situations where the lack of EYFP expression in Cre⁺ T cells could affect interpretation of data. Going forward, investigators ideally should use the tdTomato-based reporter whenever compatible with their planned studies, or carefully consider how the use of the EYFP-based reporter line could impact their results.

Author contributions

Geoffrey S. Kansas (Conceptualization [Lead], Data curation [Lead], Formal analysis [Lead], Funding acquisition [Lead], Investigation [Lead], Methodology [Lead], Project administration [Lead], Writing—original draft [Lead], Writing—review & editing [Lead])

Supplementary material

Supplementary material is available at ImmunoHorizons online.

Funding

Support for this project was provided by the National Institutes of Health (grant number 1R21AI151333-01A1).

Conflicts of interest

None declared.

Data availability

All data in this report are available upon request.