Genetic Cre-loxP assessment of epicardial cell fate using Wt1-driven Cre Alleles

Introduction

Epicardium is a specialized mesothelium that is crucial for heart development. Cells from the proepicardium, an outpouching of the septum transversum, seed the surface of the developing heart at approximately embryonic day 9.5 of mouse gestation (E9.5). These cells spread as an epithelial sheet over the surface of the heart to form the epicardium.^1-3 Epicardium regulates heart development both by exchanging paracrine signals with myocardium, and by undergoing epithelial to mesenchymal transition (EMT) to form mesenchymal cells known as epicardium-derived cells (EPDCs), which differentiate into several different lineages of the developing heart. Epicardium is quiescent in the normal adult heart, but is reactivated in the setting of myocardial injury.^{4, 5} The developmental plasticity and paracrine activity of epicardium has raised interest in its potential applications to modulate remodeling and regeneration of the injured heart.^{4, 6, 7}

Studies of epicardium have been facilitated by development of Cre alleles that permit selective labeling and isolation of epicardial cells and their derivatives. A number of such alleles has been reported.^8-13 Each has its own strengths and weaknesses, and none described to date are completely specific to epicardium. Nevertheless, through proper consideration of the properties of each Cre allele, it has been possible to make advances in understanding the function and fate of epicardium in heart development and adult heart disease. These Cre alleles, in combination with complementary labeling approaches in avian embryos, have shown that EPDCs migrate into the underlying myocardium to differentiate into fibroblasts and smooth muscle cells.^{1, 3} They also contribute to vascular endothelial cells, although quantitative studies in mammals suggests that the fraction of vascular endothelial cells that arise from epicardium is small.^{10, 11, 14} EPDCs also potentially contribute to cardiomyocytes^{10, 11}, although this result has remained controversial.¹⁵

Recently, Drs. Rudat and Kispert reported limitations of using Wt1-based Cre tools in dissecting epicardial cell fate.¹⁶ In this Research Commentary we refine some of the observations made in this study, and add additional data on the properties of Wt1-based Cre alleles. By understanding the properties of these tools, they can be productively used to selectively label epicardium and epicardium-derived cells in the embryonic and adult heart.

Methods

Wt1^GFPCre (named Wt1^creEGFP in the Rudat and Kispert study¹⁶), Wt1^CreERT2, Rosa26^fsLz, Rosa26^mTmG, and Nkx2-5^LacZ were described previously.^11,17-19 Immunostaining and immunohistochemistry were performed as previously described^{20, 21}. For tamoxfen (Tam) induction, we dissolved Tam (Sigma, T5648) in ethanol and then emulsified it in sesame oil with sonication to a final concentration of 12.5 mg/ml. 200 μl to 300 μl freshly emulsified Tam were injected by gavage at indicated time point to induce Cre. For neonatal CreERT2 induction, we injected 0.2 mg Tam directly into the stomach at P4. The CreERT2 induction for adult heart was described previously.⁴ Mice were used in accordance with protocols approved by the Institutional Animal Care and Use Committee at Children’s Hospital Boston.

The extent of Cre recombination catalyzed by Wt1^CreERT2 was quantitated in histological sections. Cells in the epicardium were scored for expression of the Cre-activated reporter in more than 5 images from each of three Wt1^CreERT2/+::Rosa26^mTmG hearts and reproduced in a second set of embryos treated, processed, and analyzed by a different investigator.

Results and Discussion

Cre-loxP genetic fate mapping requires that the Cre is active in progenitors but not in their descendants. The results of Cre-loxP fate mapping are strongly influenced by level of Cre expression and by Cre-loxP reporter sensitivity. Lack of Cre reporter activity might indicate low promoter activity of the reporter (as sometimes seen in adult tissues where the reporter is driven by endogenous Rosa26), insensitive reporter gene detection, or resistance of the reporter to recombination by the level of Cre being expressed.²² On the other hand, Cre-recombination detected by very sensitive methods and reporters may conceal important and functional differences between levels of Cre activity, such as the difference between higher levels of tissue-specific Cre activity and trace expression without interpretable biological function.

The choice of Cre-activated reporter is important when using the constitutive Wt1^GFPCre line (Table 1). We observed that Wt1^GFPCre broadly activates Rosa26^LacZ generated by P. Soriano²³ in a pattern consistent to that reported by Rudat and Kispert for Rosa26^mTmG.¹⁶ However, using the Rosa26^LacZ line generated by S. Orkin,¹⁷ we observed specific recombination in proepicardium without detectable recombination in the myocardium at E9.5 (Fig. 1A-C; representative of 3 embryos studied at that stage). In addition, strong and specific X-gal staining was observed in the developing urogenital region in Wt1^GFPCre/+;Rosa26^LacZ/+ embryos (Fig. 1D). Likely, a thresholding effect inherent in Cre-loxP experiments²² magnified differences in reporter gene recombination catalyzed by low levels of Cre activity in the early Wt1^GFPCre embryo,^{24, 25} so that the more sensitive reporters became broadly recombined and hence uninformative. These data suggest that there are circumstances when Wt1^GFPCre can be used to selectively recombine certain targets, if control experiments indicate that broad, early recombination has not occurred. Additional experiments using independent approaches, such as dye labeling, multiple independent Cre transgenes, or selective viral Cre delivery, should be performed to confirm results. However, because of the potential for Wt1^GFPCre-mediated recombination outside of the mesothelial lineage, we generally no longer favor the use of this reagent for epicardial fate-mapping or gene inactivation studies unless it can be shown that the floxed target of interest undergoes epicardium-restricted recombination. It does remain useful as a WT1 knockout allele, and for genetically marking cells that actively express WT1 with GFP^{4, 6, 20, 21}

Inducible Cre alleles allow one to activate recombination at defined times, when the domain of Cre expression can be carefully defined. The Wt1^CreERT2 allele thereby circumvents the potential for widespread recombination in the early embryo that can occur using the constitutive allele in combination with some floxed constructs. The Wt1^CreERT2 labeling described by Rudat and Kispert¹⁶ by Tam administration at E11.5 was less efficient than what we have observed and reported^{11, 21}. The reasons for this inefficiency, and for the inability to deliver Tam prior to E11.5, are unclear to us. We have found that strain background (outbred superior to inbred) and administration route (gavage preferable to intraperitoneal injection) influence Tam side effects and efficacy. We do observe Tam toxicity in late gestation (past ~E16.5) and impaired delivery of fetuses, requiring reduced Tam dose, delivery by cesarian section, and fostering to surrogate mothers. The problems in Tam administration by Rudat and Kispert likely contributed to their lack of detection of cardiomyocyte labeling. We find the Wt1^CreERT2 line is useful for epicardial lineage tracing and gene ablabtion in both embryonic and adult epicardium, and also mesothelium of other organs like liver and testis.^{4, 6, 21, 26-28}. We achieve efficient labeling (70.2 ± 10.4% of epicardial cells labeled by Tam administration at E10.5 (n=3); Fig. 2A). This degree of recombination is sufficient for most fate-mapping studies, and for many gene inactivation studies, but incomplete inactivation should be considered when interpreting results. Higher levels of recombination might be achieved with multiple Tam doses. We reproducibly observe labeling of a small subset of cardiomyocytes.¹¹ Using independent epicardial Cre lines (Sema3d^Cre, Scx^Cre and Wt1^IRES-GFPCre), other groups have also observed cardiomyocyte labeling, although these experiments also require cautious interpretation.^{9, 13}

A more difficult issue appropriately raised by Rudat and Kispert is whether or not cardiomyocytes express Wt1 (or Cre driven by Wt1). The correct interpretation of Cre fate mapping results depends on clear definition of the Cre expression map. However, obtaining expression maps with the appropriate spatiotemporal resolution and sensitivity can be problematic. First, Cre fate maps represent the accumulation of continuous Cre-mediated recombination, while expression maps represent expression at a discrete time point and have the potential risk of missing detection of expression in certain cell populations at other time points. Second, immunostaining or in situ hybridization have imperfect sensitivity and specificity, and their dynamic range may not match well with that of the Cre-loxP method. In examining whether or not cardiomyocytes express WT1, antibody constraints precluded Rudat and Kispert from reaching a definitive conclusion¹⁶. We¹¹ and others²⁹ did not observe WT1 expression in cardiomyocytes in co-staining experiments, with strict staining controls (Fig 1E-I). However, given immunostaining limitations and potential mismatch between immunostaining compared to Cre-loxP detection, we cannot absolutely exclude potential cardiomyocyte expression of WT1 in the developing heart. The basis for any Cre-loxP fate map is clear and definitive expression map, therefore whether epicardium contriubtes to cardiomyocytes requires further investigation. Additional experiments using complementary strategies that do not hinge on expression or lack of expression of Cre in a cell population of interest are needed to resolve this and similar controversies.

The origin of coronary endothelial cells remains controversial. We and others showed little contribution of Wt1 or Tbx18 lineages to coronary vessels in mouse heart.^{10, 11, 14, 30} Rudat and Kispert detected WT1 expression in endothelial cells as early as E9.5,¹⁶ although the frequency was not reported. Tamoxifen induction of Wt1^CreERT2 at E10.5 led to rare labeling of coronary endothelial cells, suggesting that few E10.5 cells destined to become endothelial cells express Wt1-driven CreERT2. This result suggests that most coronary endothelial cells do not arise from the epicardial Wt1 lineage, and they may arise from an epicardial subpopulation not labeled by Wt1^CreERT2, as recently suggested by Tabin and Epstein, or from a non-epicardial source.^{13, 30} However, activation of Wt1^CreERT2 at E14.5 or P4 revealed substantial labeling of endothelial cells at E18.5 and P7 respectively (Fig. 2B). Consistent with expression of Wt1 in endothelial cells at these stages of development, as reported by Rudat and Kispert¹⁶ and as shown in Fig. 2C, Wt1^CreERT2 labeling of endothelial cells at these stages likely reflected expression of Cre within these cells rather than their origin from epicardial cells. Surprisingly, Wt1^CreERT2 did not substantially label endothelial cells in the neonatal lung (Fig. 2D), suggesting stage- and tissue-specific regulation of endothelial cell Wt1 expression. Unlike the fetal epicardium,^{10, 11, 31} neonatal epicardium did not continue to contribute to coronary smooth muscle cells (Fig. 2E-F). The large majority of endothelial cells of the adult heart does not express WT1,^{4, 32} and consistent with this result very few endothelial cells were labeled by adult induction of Wt1^CreERT2 (Fig. 2B).⁴ In summary, the data indicate that Wt1-expressing cells at E10.5 do not substantially contribute to coronary endothelial cells. At late fetal and neonatal stages, the expression of Wt1 (and Wt1^CreERT2) within endothelial cells precludes using this tool at these stages to assess epicardial contribution to the endothelial lineage. Different tools or strategies are needed to address the origin of endothelial cells during this time window.¹³

One additional Cre allele useful for inducible recombination in the epicardium, Tcf21^MerCreMer, was recently reported.¹² Tcf21 is selectively expressed in epicardial cells and endocardial cushion mesenchyme. When induced at E10.5, this Tcf21^MerCreMer knockin allele efficiently recombines the epicardium and in the atrioventricular valves. Cre-activated lineage tracer was detected in vascular smooth muscle and fibroblast lineages, but not in endothelial or cardiomyocyte lineages.³³ When induced at E14.5 or P5-7, little epicardial labeling was observed. Rather, label was predominantly found in cardiac fibroblasts but not smooth muscle, endothelial, or cardiomyocyte lineages.³³ These data suggest that Tcf21 initially labels most epicardial cells, but later is confined to the cardiac fibroblast lineage. Thus this allele appears to be useful for lineage tracing of the epicardium prior to EMT (e.g. ~E10.5 and prior) and of a subset of cardiac fibroblasts at subsequent fetal and postnatal stages.

The Cre-loxP system is powerful, but clearly has pitfalls. Detailed understanding of the technology and the properties of Cre alleles and Cre-activated reporters is essential to properly interpret Cre-based gene inactivation and lineage tracing studies. The Wt1^GFPCre allele has limitations, as we noted in the Jackson labs database and as pointed out by Rudat and Kispert.¹⁶ Proper choice of Cre-activated reporters can reduce but not eliminate the problem, and as a result we no longer favor use of this allele for epicardial-specific gene inactivation or genetic fate mapping experiments. We find that Wt1^CreERT2 is a useful and specific tool for labeling epicardium and its derivatives. This allele can be induced by a single dose of Tam with moderately high efficiency, and has proven to be specific for epicardium and its derivatives when activated by Tam between E9.5-E11.5. Tam-treated embryos survive well to late gestation, but postnatal study of Tam-treated fetuses is complicated by problems in labor and delivery. Tam treatment in later gestation or in the neonatal period is complicated by recombination within coronary endothelial cells, precluding fate mapping into this cell population. As with most techniques, complementary approaches, such as selective cell labeling with dyes or with viruses, are needed to independently support lineage tracing results obtained with Wt1-based Cre tools. In some cases, resolution of controversies awaits further studies using complementary approaches that do not hinge on whether or not one categorizes Cre as expressed or not in a specific cellular compartment.