Efficient Inducible Cre-Mediated Recombination in Tcf21 Cell Lineages in the Heart and Kidney

Abstract

Tcf21 is a class II bHLH family member with essential roles in the formation of the lungs, kidneys, gonads, spleen, and heart. Here, we report the utility of a mouse line with targeted insertion of a tamoxifen-inducible Cre recombinase, MerCreMer at the Tcf21 locus. This mouse line will permit the inducible expression of Cre recombinase in Tcf21-expressing cells. Using ROSA26 reporter mice, we show that Cre recombinase is specifically and robustly activated in multiple Tcf21-expressing tissues during embryonic and postnatal development. The expression profile in the kidney is particularly dynamic with the ability to cause recombination in mesangial cells at one time of induction and podocytes at another time. These features make the Tcf21-driven inducible Cre line (Tcf21^iCre) a valuable genetic tool for spatiotemporal gene function analysis and lineage tracing of cells in the heart, kidney, cranial muscle, and gonads.

Members of basic helix-loop-helix (bHLH) transcription factor family have important roles in cell fate specification, differentiation and morphogenesis of several tissues during development. In higher eukaryotes, these factors regulate a multitude of key developmental processes including myogenesis, neurogenesis, hematopoiesis, and cardiac and pancreatic development (Weintraub et al., 1991; Lee et al., 1995; Bain et al., 1994; Porcher et al., 1996; Srivastava et al., 1997; Naya et al., 1997). While the class I bHLH proteins are widely expressed and capable of binding DNA on their own as homodimers, the class II proteins show a more tissue-restricted pattern of expression and require heterodimerization with a class I bHLH factor to bind at the target sites (Massari and Murre, 2000). Tcf21 (Capsulin/Epicardin/Pod-1) encodes a mesoderm-specific class II bHLH transcription factor that binds to the consensus E box sequence (CANNTG) as a heterodimer with the widely expressed class I bHLH protein E12 (Lu et al., 1998). In the developing embryo, Tcf21 is expressed in several cell lineages of the respiratory, digestive, urinogenital, and cardiovascular systems (Quaggin et al., 1998; Lu et al., 1998; Hidai et al., 1998).

Knowing the widespread expression and functional importance of Tcf21 in the development of multiple organs, we reasoned that mice expressing an inducible Cre recombinase from the Tcf21 locus would be a valuable tool for temporal and tissue-specific gene manipulation as well as lineage tracing studies. Here, we report the generation of a mouse strain expressing a Cre recombinase protein fused to two mutant estrogen-receptor ligand-binding domains (MerCreMer, Zhang et al., 1996) under the control of the endogenous Tcf21 locus (Tcf21^iCre) by homologous recombination (Fig. 1k). Heterozygous Tcf21^iCre mice have no apparent abnormalities, while homozygotes were indistinguishable from the previously reported Tcf21^LacZ null allele (Quaggin et al., 1999; Lu et al., 2000).

Figure 1. Whole mount R26R^LacZ tracing in Tcf21^iCre induced embryonic tissues.

(a,b and d) Whole mount views of β-galactosidase Cre reporter activity in E14.5 Tcf21^iCre embryos tamoxifen-induced at E10.5. (a) male gonads (b) female gonads and (d) heart. (c and e) Comparative β-galactosidase expression in E14.5 Tcf21^LacZ/+ ovary and heart respectively. (f–i) Cre reporter expression at E18.5 from embryos tamoxifen-induced at E10.5 (f) kidney (g) adrenal gland, (h) lung, and (i) spleen. (j) β-galactosidase activity in facial skeletal muscles of an E13.5 embryo tamoxifen-induced at E9.5. (k) Schematic representing wildtype and Tcf21^iCre targeted locus. The MerCreMer fusion cDNA was knocked into the Tcf21 locus replacing the entire coding region of the first exon. Open boxes and grey line represent 5′ UTR and vector sequences respectively. H, HindIII; RI, EcoRI; S, SacI; K, KpnI; pA, polyadenylation signal sequence.

To analyze if Cre-reporter activity faithfully recapitulated Tcf21 expression, Tcf21^iCre were first bred with R26R^LacZ mice (Soriano, 1999). As expected, no β-galactosidase reporter activity was observed in the absence of tamoxifen (data not shown). To examine Cre activity in the gonads, we isolated whole genital ridges following tamoxifen induction at E10.5. Both testes and ovary exhibited robust β-galactosidase activity (Fig. 1a,b). Cardiac expression of Tcf21 is first detectable around E9.5 in a discrete group of cells posterior to the heart that constitute the proepicardial organ (Lu et al., 1998; Hidai et al., 1998). At E10.5, Tcf21 distinctly marks the epicardial lining surrounding the heart and the endocardial cushions (Lu et al., 1998). Tamoxifen induction at this time point resulted in Cre activity in the epicardium at E14.5 (Fig. 1d). Moreover, the pattern of reporter expression in the heart and the ovary was very comparable to the β-galactosidase expression observed using Tcf21^LacZ/+ (Lu et al., 1998) allele (Fig. 1b–e). Examination of other tissues from an E10.5 tamoxifen treatment detected Cre reporter activity in kidney, adrenal gland, lung and spleen (Fig. 1f–i). Additionally, Cre induction at an earlier time point (E9.5) resulted in β-galactosidase reporter activity in a distinct set of facial skeletal muscles (Fig. 1j), where Tcf21 is reported to be redundantly required along with the related bHLH factor MyoR (Lu et al., 2002). Thus, administration of tamoxifen at multiple embryonic stages resulted in robust and specific induction of Cre activity in most of the expected Tcf21-expressing embryonic tissues (Table I).

Table I.

Summary of Cre reporter activity in different Tcf21^iCre tissues

	Heart	Kidney	Lung	Craniofacial	Testes	Ovary	Adrenal	Spleen
Cre Induction
E9.5	++	++	−	++	++	ND	++	++
E10.5	++	++	+	−	++	+	++	++
E14.5	++	++	ND	ND	++	+	++	++
P5, P7	++	++	ND	ND	++	ND	++	ND

Reporter activity (R26R^LacZ, R26R^{YFP and} R26R^tdTomato) and was noted in the following tissues:

Heart: epicardium, interstitial cells of the myocardium and atrioventricular valves; Kidney: Mesangial cells, podocytes, perivascular regions, ureter and interstitial cells; Lung: interstitial cells; Craniofacial: skeletal muscle; Testes: coelomic epithelium, peritubular myoid cells, interstitial cells of leydig and pericytes; Ovary: thecal cells; Adrenal: capsule and cells in the cortex; Spleen: interstitial cells. Refer to the text for more details.

‘−’: Not detected;

‘+’: detectable activity;

‘++’: good activity;

ND: not determined.

Heart

To gain better insight about the different cell populations that possessed Cre activity, inductions were performed at two embryonic time points (E10.5 and E14.5). Following tamoxifen treatment at E10.5, E18.5 hearts were analyzed for reporter activity using the R26R^YFP reporter (Srinivas et al., 2001), Use of this reporter line permitted us to costain tissues for vascular markers designating endothelial cells and vascular smooth muscle cells. YFP expression in the heart was detectable in a vast majority of epicardial cells and a subset of interstitial cells underlying the epicardium (Fig. 2a,b). Costaining with PECAM demonstrated that reporter expression was excluded from endothelial cells (Fig. 2b). Additionally, as previously observed with the Tcf21^LacZ/+ allele, reporter expression was also detectable in the atrioventricular valves and absent from the outflow tract valves (Fig. 2c). By contrast, hearts from E18.5 embryos that were induced at E14.5 had very little labeling of epicardial cells and showed reporter expression predominantly in cells scattered within the myocardium (Fig. 3a). This pattern of expression was likely caused by migration of cells from the epicardium into the myocardium following epithelial to mesenchymal transition. To demonstrate that a population of Tcf21^iCre lineage traced cells migrate into the subepicardial space, we performed an ex vivo migration assay. Adenovirus specifically transduces only the surface epicardial cells. Fig. 2j–l demonstrates a significant overlap of the adeno-GFP tagged cells with Tcf21^iCre lineage traced cells. Cre activity was also analyzed in 4-week old hearts induced at P5 and P7 using R26R^YFP line. Again, YFP expression was detected in a vast majority of interstitial cells that were distinct from PECAM-expressing endothelial or PDGFRβ-expressing vascular smooth muscle cells (Fig. 4a–c).

Figure 2. R26R^YFP and R26R^tdTomato tracing in Tcf21 ^iCre embryonic tissues induced at E10.5.

(a–c) YFP reporter expression in E18.5 Tcf21^iCre heart (green) and (d–i) R26R^tdTomato tracing in E18.5 Tcf21^iCre kidneys (red). (a) Epicardial and interstitial YFP reporter expression in embryonic heart costained with Dapi (blue) (b) Costain with PECAM (red) (c) Reporter expression in atrioventricular valves, costained with Vimentin (red). (d–i) Costainings (shown in green) with PDGFRβ(d–f) and endothelial marker Isolectin B4 (g–i) demonstrates labeling of mesangial cells of the glomeruli.. epi: epicardium; myo: myocardium; AoV: aortic valves; MV mitral valves marked by asterisk (*). Arrowheads in the kidney sections point to glomeruli. (j–l) Ex vivo migration assay showing Tcf21^iCre labeled cells in the myocardium (red) are derived from AdGFP transduced (green) epicardial cells. Scale bars represent the indicated magnifications.

Figure 3. R26R^LacZ tracing in E18.5 Tcf21 ^iCre embryonic tissues induced at E14.5.

(a–h)β-galactosidase stained sections of E18.5 Tcf21^iCre tissues tamoxifen- induced at E14.5. (a) heart, (b) spleen (c,d) testes, (e,f) kidney, and (g,h) ovary. (a) Reporter activity is seen in interstitial cells of the heart. (d) Magnified view of the testes showing reporter activity in peritubular myoid cells and interstitial cells of Leydig. β-galactosidase activity is seen in (f) glomerular podocytes. (h) magnified image of the ovary (region marked by asterisk (*) in (g)) showing β-galctosidase activity in thecal cells surrounding the follicles. Myo, myocardium; epi, epicardium; t, testes; m, mesonephrous; ce, ceolomic epithelium; tc, testis cords; pmc, peritubular myoid cells; ic, interstitial cells of leydig; bv, blood vessel; u, ureter; g, glomerulus. Scale bars represent the indicated magnifications.

Figure 4. R26R^YFP tracing in Tcf21 ^iCre tissues following postnatal induction.

(a–f) Cre activity detected by YFP reporter expression in postnatal day 28 (P28) Tcf21^iCre tissues tamoxifen-induced at P5 and P7. (a–c) postnatal heart (d–f) kidney. YFP expression is detected by immunostaining (green). Costainings (shown in red) with PECAM (b and e) and PDGFRβ (c and f) highlight the blood vessels in the heart and glomerular capillaries and mesangial cells in the kidney respectively. Scale bars represent the indicated magnifications.

Kidney

Published reports have documented high expression of Tcf21 in primarily two cell lineages of the developing kidney. These include the mesenchymal cells and visceral glomerular epithelial cells, also known as podocytes (Quaggin et al., 1998 and 1999). Inductions at E10.5 in the Tcf21^iCre kidneys with R26R^tdTomato (Madisen et al., 2010) showed reporter activity in mesangial cells of the glomerulus (Fig. 2d–i). This was confirmed by costaining with mesangial cell marker, PDGFRβ (Fig. 2d–f) and endothelial marker, Isolectin B4 (Fig. 2g–i). Cre reporter activity was also noted in regions surrounding the blood vessels and the ureter (data not shown), This is in agreement with previous studies using a Tcf21^LacZ/+ allele that indicated expression in vascular smooth muscle cell precursors as well as the smooth muscle of the ureter (Quaggin et al., 1999). No podocyte labeling was noted at this time point of induction. In contrast to the mesangial cell tagging at E10.5, induction at E14.5 resulted in robust β-galactosidase activity predominantly in interstitial and podocyte cell populations of the kidney (Fig. 3e,f). A very similar pattern of reporter expression was documented following postnatal inductions (Fig. 4d–f). Furthermore, cells positive for Cre reporter expression induced postnatally did not overlap with either the endothelial cells of the capillaries (PECAM⁺) or the mesangial cells (PDGFRβ⁺) in the glomerular apparatus (Fig. 4e,f). Thus, it appears that Tcf21 expression in the kidney follows a very dynamic expression profile allowing distinct labeling of multiple cells lineages at different times during development with the Tcf21^iCre line.

Gonads

Unlike other tissues, in gonads, Tcf21 follows a sex- and stage-dependent pattern of expression with far higher expression in embryonic testes than ovary (Tamura et al., 2001; Cui et al., 2004). Cells of the testes were efficiently labeled at all time points of Cre induction that we examined (Fig. 1a, 3c,d and data not shown). At E18.5, testes of Tcf21^iCre mice that were induced with tamoxifen at E14.5 showed strong β-galactosidase activity of coelomic epithelium, the peritubular myoid cells immediately surrounding the testis cords as well as Leydig cells in the interstitial region. Staining was also observed in a small number of pericytes surrounding the blood vessels. These cell lineages targeted in the Tcf21^iCre mouse line are notably distinct from the previously reported inducible Cre recombination system in the testes that utilizes the prion protein promoter allowing genetic recombination specifically in the germ cells (Weber et al., 2003). Cre induction in the ovaries at the same time point resulted in cells in the ovarian medulla staining positive for β-galactosidase activity (Fig. 3g,h). This staining was confined to the interstitial spaces between the follicles and was consistent with a pattern of theca cell progenitors (Fig. 3h).

Other tissues

In the adrenal gland, Cre recombination was best noted in E10.5 inductions that resulted in staining of random clusters of cells in the capsule as well as the adrenal cortex (Fig. 1g and data not shown). Despite the reported expression of Tcf21 in the lung, only a few mesenchymal cells stained positive for β-galactosidase activity when induction occurred at either E10.5 or E14.5 (Fig. 1h and data not shown). It is possible that tamoxifen induction at even later embryonic time points would be required to observe efficient Cre activity within the lung. Staining was also documented in a small splenic cell populations at most time points tested (Fig. 1i, 3b and data not shown).

Inducible Cre expression in the epicardium has been previously reported using the Wt1^CreERT2 mice that harbor the cDNA encoding a Cre-modified estrogen receptor ligand-binding domain (CreERT2) into the Wt1 locus (Zhou et al., 2008). Tamoxifen induction in these mice allowed identification of lineage traced epicardial progenitors that resulted in cardiomyocyte, VSMC, and interstitial cell tracing. In comparison, analysis of our Tcf21^iCre hearts showed reporter expression primarily in interstitial cells. Comparison of lineage tracing experiments with the Tcf21^iCre and Wt1^CreERT2 line may provide a better understanding of epicardial contribution to cardiac development.

Because mice deficient in Tcf21 die shortly after birth due to developmental defects in several tissues, postnatal functions of Tcf21 remain to be elucidated. In particular, nephrogenesis continues for a few days after birth, and knockouts of many podocyte-specific genes in mice including Tcf21, WT1 and Podocin are associated with embryonic and postnatal lethality, limiting detailed analysis in a functional adult kidney. Transgenic mice driving inducible podocyte-specific Cre recombinase under the control of human Podocin (NPHS2) promoter have been previously reported (Juhila et al., 2005; Yokoi et al., 2010). However, embryonic induction of Cre activity was not examined in the Podocin inducible Cre lines. Because Tcf21 exhibits an early embryonic podocyte expression, it is possible that the Tcf21^iCre line could be used to study gene function at very early timepoints in podocyte specification. Moreover, a substantial Cre activity in the Tcf21^iCre line is also noted in the interstitial cell population, making it a more desirable Cre line for genes expressed in both cell populations.

Thus far, our characterization of the Tcf21^iCre line has revealed efficient inducible Cre expression in the expected Tcf21 cell lineages in cranial muscle, heart, kidney and testes. In summary, this is the first report describing and validating inducible Cre expression from the Tcf21 locus. Our studies indicate that the Tcf21^iCre line holds great potential as a valuable genetic tool for gene ablation and lineage tracing studies in multiple tissues.

METHODS

Generation of the TCF21^ICRE mouse strain

The MerCreMer fusion cDNA (gift from Michael Reth, Max-Plank Institute, Freiburg, Germany) encodes the mutated estrogen receptor ligand-binding domain (amino acids 281 to 599, G525R) that allows induction of Cre activity specifically in the presence of 4-OH tamoxifen (Zhang et al., 1996). For generation of the Tcf21^iCre mouse line, the MerCreMer cDNA followed by the neomycin resistance gene was knocked into the Tcf21 locus replacing the entire first exon (+1 to 450 nt starting from the ATG). Nucleotides spanning 451-2433bp of the Tcf21 locus constituted the short arm for recombination at the 3′ end. Following homologous recombination in mice, the neomycin cassette was excised out. Mice were genotyped using southern blots and genomic PCR. Tcf21^iCre allele (500bp) was distinguished from the wild type allele (321bp) in a single PCR reaction using the following primers: Tcf21F: 5′ TTCTCCAGGCTCAAGACCAC 3′, Tcf21R: 5′ CAAACCCTAGCACAAATCACTCGC 3′ and MerP 5′ GCTTCCGATATCCAGATC CAGAC 3′.

Tamoxifen induction and β-galactosidase staining

Tamoxifen (T5648, Sigma or 156708, MP Biomedicals) was dissolved in 10% ethanol and 90% sunflower oil (S5007, Sigma) to a final concentration of 20mg/ml. To analyze Cre-recombinase activity, Tcf21^iCre knock-in mice carrying the R26R^LacZ, R26R^YFP or R26R^tdTomato reporter on a mixed 129/C57Bl6 background were crossed to wild type females. Pregnant females were gavaged with tamoxifen (0.1mg/g body weight) one time at E9.5, E10.5 or E14.5 and embryos harvested at E13.5, E14,5 or E18.5 to analyze for reporter expression. No Cre activity (as determined by ROSA26 reporter activity) was detected following inductions when sunflower oil was administered. We also did not detect any Cre activity in extraembryonic tissues at the time points of induction described (data not shown). For postnatal inductions, pups were induced twice (P5 and P7) by intraperitoneal injections (0.1mg/g body weight) and analyzed for R26R^YFP reporter expression at P28 by immunostaining. Inductions at the indicated time points for each tissue were performed multiple number of times (n>4) and confirmed using more than one reporter strain.

For whole mount β-galactosidase staining, tissues from E18.5 embryos were isolated in phosphate-buffered saline (PBS), fixed in 0.2% glutaraldehyde and 2% formaldehyde for 15min and incubated overnight in β-galactosidase staining solution containing X-gal substrate (5-Bromo 4-chloro 3-indoxyl beta-D-galactopyranoside, X4281C, Gold Biotechnology). Samples were washed with PBS twice followed by post-fixation in 10% formalin for 4 hours. Beta-galactosidase assay was also performed on frozen sections. Briefly, isolated tissues were either fixed as described above or in 0.4% paraformaldehyde overnight followed by 3 washes of PBS. Samples were then sequentially incubated for 1 hour each in 30% sucrose, 30% sucrose:OCT (1:1) and finally OCT before flash freezing with liquid nitrogen. Ten micron thick sections were cut with a microtome blade on a Leica CM-3050S cryostat and rinsed in PBS to remove the OCT. Sections were then stained overnight in β-galactosidase staining buffer, post-fixed in 10% formalin and counterstained with nuclear fast red (EM Science). Whole mount tissues were imaged using a Zeiss Stemi SV 11 Apo and an Olympus DP71 camera. Color images were captured using Zeiss Axiovert 200 with an Olympus DP71 camera.

Immunohistochemistry

Tissues were harvested and fixed in 4% PFA for 1 hour followed by processing for frozen embedding as described above. R26R^tdTomato reporter activity was directly visualized using fluorescence following frozen sectioning. Immunostaining was performed on embryonic and postnatal tissues to detect YFP reporter expression. Sections (10 μm thick) were immunostained with the following antibodies: anti-GFP (1:200, Molecular probes, Invitrogen), anti-PECAM (1:200, BD Pharmingen), anti-PDGFRβ (1:100, eBioscience), and anti-Vimentin (1:250, Sigma). Endothelial cells were also identified using biotinylated-Isolectin B4, followed by streptavidin-conjugated secondary antibody. Dapi was used for nuclear staining and fluorescent images were taken using Zeiss Axiovert 200 with a Hamamatsu ORCA-ER camera. All images and figures were edited and created in Photoshop CS4.

Migration assay

Ex vivo migration assay was performed as described previously (Mellgren et al., 2008) with a few modifications. Pregnant female mice resulting from a cross with Tcf21^iCre knock-in mice carrying the R26R^tdTomato reporter were induced with tamoxifen at E10.5. A day later (E11.5), hearts positive for reporter activity were isolated and cultured in 10% FBS 1:1 DMEM:M199 supplemented with basic fibroblast growth factor (2ng/ml, Sigma) in presence of green fluorescent protein-expressing adenovirus (AdGFP) for 12 hours. Hearts were then stimulated with TGFβ1 (10ng/ml) and PDGF-BB (20ng/ml) to promote epithelial to mesenchymal transition of the epicardium. After 48 hours, hearts were fixed in 4% PFA for 2 hours, frozen embedded, sectioned and visualized for fluorescence.