Introduction

ES cells are continuously growing stem cells originally isolated from the inner masses of blastocysts and provide a potentially unlimited source for generating highly specialized cells and tissues in vitro. ES cells aggregate into embryoid bodies (EB), structures comprising three germ layers that interact with each other. By using embryoid body formation and other methods, ES cells can be differentiated into various cell lineages, for example, pancreatic cells (Lumelsky et al. 2001), hepatocytes (Kania et al. 2003; Yamamoto et al. 2003), motor neurons (Wichterle et al. 2002), and hematopoietic cells (Kyba et al. 2002; Daley 2003). However, few papers have described the induction of the kidney lineage from ES cells. Shuldiner et al. demonstrated that human ES cells cultured with growth factors, including hepatocyte growth factor and activin A, differentiated into cells expressing Wt1 and renin. Kobayashi et al. showed that Wnt4-transformed mouse ES cells cultured with the same growth factors differentiated into cells expressing aquaporin (AQP)2 (Kobayashi et al. 2005). Kim et al. revealed that mouse embryoid bodies treated with a combination of activin A, retinoic acid, and bone morphogenetic protein 7 (BMP7) expressed markers of the kidney lineage and formed epithelial structures in vitro (Kim and Dressler 2005). However, the induction efficiency in these experiments is still low and needs to be improved.

The kidney develops in three stages: the pronephros, mesonephros, and metanephros. The nephric duct (Wolffian duct) and nephrogenic mesenchyme develop from the intermediate mesoderm, which is located between the lateral and preaxial mesoderm. The Wolffian duct and nephrogenic mesenchyme elongate caudally towards the cloaca, and the Wolffian duct converts the adjacent mesenchyme into mesonephric tubules. Metanephric mesenchyme is formed in the caudal-most region of the nephrogenic mesenchyme. The metanephros is formed by a reciprocally inductive interaction between two precursor tissues, the metanephric mesenchyme and the ureteric bud. Upon induction by the ureteric bud, the metanephric mesenchyme differentiates into the epithelia of glomeruli and renal tubules.

Gene targeting studies have revealed many genes that are involved in these processes of kidney development, including Pax2, Lim1, Wt1, Sall1, Hoxa11, Pod1, Gdnf, and integrin α8 (Vainio and Lin 2002; Dressler 2006). We previously showed that Sall1, a zinc finger nuclear factor that is expressed in the metanephric mesenchyme, is essential for metanephros development and that mesenchyme expressing high level of Sall1 contains multipotent progenitors for the epithelia of glomeruli and renal tubules (Osafune et al. 2007). By contrast, the transcription factor Pax2, which is also essential for metanephros development, is expressed both in the mesenchyme and the ureteric buds (Torres et al. 1995). In addition, Pax2 is one of the most upstream regulators in the genetic cascade of kidney development. In the transient pro- and mesonephros of the mouse, Pax2 is expressed in the mesenchymal primordium as well as in the epithelial components and controls the initial differentiation from intermediate mesoderm to Wolffian duct in cooperation with Pax8; thus, mutant mice lacking both of these genes lack pronephros formation (Bouchard et al. 2002). Interestingly, retroviral overexpression of Pax2 in chick embryos leads to the ectopic formation of nephric structures in the intermediate mesoderm (Bouchard et al. 2002), suggesting that Pax2 may initiate the kidney formation program in vivo. Therefore, we hypothesized that Pax2 overexpression in ES cells may mimic the situation in vivo. In the present study, we generated ES cell lines harboring tetracycline-inducible Pax2 (Tet-off system) and examined their differentiation potential by embryoid body formation.

Materials and Methods

Plasmid construction. The Pax2 fragment was PCR-amplified with TaKaRa LA Taq (TaKaRa BIO, Otsu, Japan) using E13.5 whole embryo cDNA as a template and the primer pair 5′-CGGAATTCATGGATATGCACTGCAAAGC and 5′-GCGCGGCCGCCTAGTGGCGGTCATAGGCAGC. The amplified DNA was digested with EcoRI and NotI, cloned into pBluescript KS II(−) (pBlKs-Pax2), and sequenced. The exchange vector was created by inserting the 1.2-kbp XhoI-NotI fragment of pBlks-Pax2 into the XhoI and NotI sites of pPthC-Oct3/4 (Masui et al. 2005).

Culture of ES cells and embryoid bodies. MGZRTcH2 cells, in which gene expression can be controlled by the Tet-off system (Masui et al. 2005), were maintained on 0.1% gelatin-coated tissue culture plates in Glasgow minimal essential medium (Invitrogen, Carlsbad, CA), 10% fetal bovine serum, 0.1 mM β-mercaptoethanol, 1 mM sodium pyruvate (Invitrogen), 0.1 mM MEM non-essential amino acids solution (Invitrogen), 0.1 mg/ml G418, 0.05 mg/ml hygromycin B (Hyg) (Invitrogen), 0.01 mg/ml Zeocin (Invitrogen), and 103 units/ml mouse leukemia inhibitory factor (LIF; Chemicon, Pittsburgh, PA) at 37°C with 5% CO2. The ES cells were dissociated with 0.25% trypsin plus 1 mM EDTA (Invitrogen) and cultured in the same medium without LIF to induce embryoid body formation by the hanging drop culture method (5 × 103 ES cells/0.015 ml drop). The embryoid bodies in hanging drops were cultured for 5 d and then transferred to tissue culture plates coated with 0.1% gelatin in the same medium without LIF (designated as day 0). The adherent embryoid bodies were cultured for an additional 5 or 10 d (day 5 or 10).

Generation of ES cells with tetracycline-inducible Pax2. MGZRTcH2 cells were electroporated (250 V, 960 μF) with 20 μg of the Pax2-containing exchange vector and a Cre-expression plasmid (pCAGGS-Cre; Niwa et al. 1991) using a Gene Pulser (BioRad, Hercules, CA). The resultant ES cells were cultured in the presence of 1 μg/ml tetracycline (Tet; Sigma, St. Louis, MO), without Hyg. After 2 d, the medium was changed to medium containing 1 μg/ml Tet and 1.5 μg/ml puromycin (Puro; Sigma) without Hyg to expand the Puro-resistant clones. Eight days after electroporation, Puro-resistant clones were picked up and the recombination was confirmed by Southern blot analysis using the gene images random-prime labeling and detect system (GE Healthcare, Slough, UK). Seven out of 11 picked clones were recombinants.

Western blot analysis. ES cells and embryoid bodies were lysed in buffer (20 mM HEPES, pH 7.8, 500 mM NaCl, 1.5 mM MgCl2, 1% NP-40, 1% protease inhibitor cocktail (Sigma)). Protein aliquots (50 μg) were boiled for 3 min in sample buffer and electrophoresed on 0.1% SDS-7.5% polyacrylamide gels. The proteins were transferred to nitrocellulose membranes at 60 V for 1 h, probed with a polyclonal antibody against mouse Pax2 (Covance, Richmond, CA), and detected using horseradish peroxidase-conjugated goat anti-mouse IgG and the ECL plus system (GE Healthcare).

Reverse transcription-PCR. Total RNA was extracted using TRIZOL reagent (Invitrogen). One unit of DNase I (Roche, Mannheim, Germany) was added per microgram of RNA, and the mixture was incubated at 37°C for 30 min. Two micrograms of total RNA were reverse transcribed using Stratascript RT (Stratagene, La Jolla, CA) according to the manufacturer’s protocol. Two negative control reactions were also performed, with water instead of RNA or without reverse transcriptase. The primer pairs spanning at least one intron were as follows: Pax-2, 5′-AGGGCATCTGCGATAATGAC-3′, 5′-CTCGCGTTTCCTCTTCTCAC-3′; Lim-1, 5′-TGGACCGTTTCCTCTTGAAC-3′, 5′-TGTTCTCTTTGGCGACACTG-3′; Aqp1, 5′-CCTCCAGGCACAGTCTTCTC-3′, 5′-CAGTGGCCTCCTGACTCTTC-3′; Pod1, 5′-GGAACTTCCAACGAGAGCAC-3′, 5′-GTGGTCTTGAGCCTGGAGAA-3′; Sall1, 5′-CCCCATCCCTATTAGCCATT-3′, 5′-AGAGTACTGTTGCCCGCTGT-3′; Wt1, 5′-ACCCAGGCTGCAATAAGAGA-3′ 5′-GCTGAAGGGCTTTTGACTTG-3′; Gdnf, 5′-CCCGAAGATTATCCTGACCA-3′, 5′-TAGCCCAAACCCAAGTCA-3′; integrin α8, 5′-GGCGAAGTGCAGTCCTAAA-3′, 5′-GAAGGAGACATTCGGAGTGG-3′; and Hoxa11,5′-GGATTTTGATGAGCGTGGTC-3′, 5′-AGTAGCAGTGGGCCAGATTG-3′.

Immunocytochemistry. Embryoid bodies cultured for 10 d on chamber slides were fixed for 2 h with 4% paraformaldehyde at 4°C, washed with phosphate buffered saline and 0.1% Tween 20 (PBST), blocked for 2 h at room temperature with 2% goat serum in PBST, and incubated with rabbit anti-Pax2 (Covance) or mouse anti-AQP1 (Sigma) antibodies. After washing twice with PBST, fluorescence-conjugated secondary antibodies were added. Images were captured on a Nikon ES800 fluorescence microscope equipped with a SPOT digital camera. Staining without primary antibodies gave no signal.

Results

Generation of ES cell lines with tetracycline-inducible Pax2. The ROSA26 locus is widely used to drive ubiquitous expression of exogenous genes (Zambrowicz et al. 1997). MGZRTcH2 cells, an ES cell line, contain the ROSA26 locus in which the Tet-regulated transactivator (tTA) was inserted followed by loxP and lox PV sites, as shown in Fig. 1 A. The exchange vector carrying the Tet-responsive CMV*-1 promoter and Pax2 cDNA was introduced into this cell line along with the Cre-expression vector. The vector was efficiently incorporated into the ROSA26 locus upon Cre-mediated recombination, as confirmed by Southern blots (Fig. 1 B), and Pax2 cDNA, as well as the downstream Venus, was placed under the control of the Tet activator. In this Tet-off system, tTA driven by the endogenous ROSA promoter binds to the CMV*-1 promoter in the absence of Tet, thereby inducing expression of Pax2 cDNA-IRES-Venus. Indeed, Venus fluorescence was used to rapidly confirm the correct recombinants (Fig. 1 C), and Pax2 protein was detected only in the absence of Tet in the recombinants (Fig. 1 D). Similar results were obtained when embryoid bodies were cultured without tetracycline (data not shown).

Figure 1.
figure 1

Generation of ES cell lines with tetracycline-inducible Pax2. (A) The exchange vector containing Pax2 cDNA was introduced, along with the Cre-expression plasmid, into MGZRTcH2 cells containing the ROSA-TET locus. The correctly targeted clones were resistant to puromycin. (B) Southern blot analysis of untargeted (U) and targeted (T) clones. The genomic DNA of puro-resistant clones was digested with XbaI and hybridized with the probe indicated in A. (C) Venus expression in untargeted (U) and targeted (T) clones. Cells were cultured in the absence of Tet and Venus fluorescence was detected only in targeted clones. (D) Western blot analysis of Tet-inducible Pax2 in the targeted ES clones. Pax2 protein was detected only in the absence of Tet.

Full size image

Pax2 overexpression in embryoid bodies induced expression of Aqp1 and integrin α8. To investigate the differentiation potential of ES cells toward the kidney lineage, the expression levels of nine genes that are essential for kidney development or functions, specifically Pax2, Lim1, Aqp1, Pod1, Sall1, Wt1, GDNF, integrin α8, and Hoxa11, were examined (Fig. 2 A). RT-PCR analysis showed that nearly all genes were undetectable in the parental ES cells (MGZRTcH2). Sall1 was the exception, consistent with our previous report that Sall1 is expressed in ES cells (Sakaki-Yumoto et al. 2006). When cells were induced to form embryoid bodies for 5 d in hanging drops, followed by attachment to plastic dishes, the expression levels of Lim1 and Wt1 increased by day 5 after attachment, while that of Sall1 remained constant. As these genes are also expressed in nonrenal lineages, this does not imply differentiation toward a kidney fate. Pax2 was undetectable at all stages of differentiation.

Figure 2.
figure 2

Pax2 overexpression in embryoid bodies induces expression of Aqp1 and integrin α8. (A) RT-PCR analysis of the parental ES cell line (MGZRTcH2) and EB. Embryonic kidneys were used as a positive control. (B) RT-PCR analysis of clones containing Tet-inducible Pax2. In the presence of Tet (Tet+), gene expression patterns were similar to those in parental cells. In the absence of Tet (Tet-), Aqp1 and integrin α8 were upregulated when cells were induced to form embryoid bodies. Embryoid bodies were formed for 5 d in hanging drops, subsequently attached to the plastic dishes, and cultured for the indicated number of days.

Full size image

We next examined whether Pax2 overexpression induces differentiation of ES cells toward the kidney lineage (Fig. 2 B). In the presence of Tet, gene expression patterns were similar to those of the parental cells. In the absence of Tet, Pax2 was overexpressed both in ES cells and embryoid bodies, as expected. In ES cells, Pax2 overexpression did not lead to any phenotypic changes or to an increase in the expression levels of any genes examined, suggesting that Pax2 alone cannot trigger differentiation cues. During embryoid body formation, however, the expression level of integrin α8 was increased by day 5 after attachment and that of Aqp1 was upregulated by day 10. The expression kinetics of Lim1, Wt1, and Sall1 remained unchanged. Although Pax2 is reported to be an upstream regulator of Gdnf (Brophy et al. 2001), Gdnf was undetectable throughout all the experiments. These results were reproducible using three independent clones (data not shown). We further examined expression changes at the protein level. Embryoid bodies from ES clones containing Tet-inducible Pax2 were cultured with or without tetracycline and stained with anti-AQP1 and anti-Pax2 antibodies (Fig. 3). Pax2 protein was detected only in the absence of Tet, consistent with the results shown in Fig. 1 D. AQP1 protein was also upregulated in the absence of Tet, thus confirming the results shown in Fig. 2 B. Commercially available anti-integrin α8 antibodies did not give any signals, irrespective of the presence or absence of Tet; this was possibly due to antibody sensitivity, because we failed to detect signals using embryonic kidneys, which should express integrin α8. Thus, Pax2 can upregulate the expression of integrin α8 and Aqp1 in embryoid bodies with slow kinetics, and additional factors that are induced in the embryoid bodies may be required for this phenomenon.

Figure 3.
figure 3

Pax2 overexpression in embryoid bodies led to upregulation of AQP1 protein. Embryoid bodies derived from ES clones containing Tet-inducible Pax2 were cultured for 10 d after attachment to the slide chambers, with or without tetracycline, and stained with anti-AQP1 and anti-Pax2 antibodies.

Full size image

Discussion

Pax2 is one of the most upstream regulators in the genetic cascade of kidney development. By establishing tetracycline-inducible overexpression in ES cells, we showed that Pax2 can upregulate two genes that are possibly involved in kidney development (integrin α8 and Aqp1) in embryoid bodies, but not in ES cells. Together with the slow induction kinetics (5–10 d), our data suggest that Pax2 and additional factors that are induced in the embryoid bodies synergistically regulate the two targets.

Many integrins have been shown to be important for cell–matrix interactions during developmental processes. Integrin α8 is expressed in the metanephric mesenchyme surrounding the branching ureter tips, and mice deficient for this gene display kidney agenesis or dysgenesis (Muller et al. 1997). This is due to an impaired interaction of the mesenchyme with the ureteric epithelium that expresses the ligand for this integrin, nephronectin (Brandenberger et al. 2001; Linton et al. 2007). Although our data suggest that Pax2 upregulates integrin α8 in embryoid bodies, it is unknown whether Pax2 acts upstream of integrin α8 in vivo. Expression analysis using Pax2 mutant mice will help elucidate the answer to this question. Recently, it was shown that Hox11 paralogous proteins cooperate with Eya1 and Pax2 to activate Six2 and Gdnf, two genes that are essential for kidney development. Hox11, Eya1, and Pax2 proteins physically associate, bind to a metanephric mesenchyme-specific enhancer region, and synergistically activate reporter expression (Gong et al. 2007). Although Six2 and Gdnf expression was not induced by Pax2 in our experiments (Fig. 2 B and data not shown), it would be interesting to examine whether integrin α8 is also regulated by a Hox11/Eya1/Pax2 protein complex in vivo.

Aqp1 is a water channel expressed in epithelial cells of the proximal tubules and in the descending limb of Henle, but not in other nephron segments (Nielsen and Agre 1995; Agre and Nielsen 1996). Aqp1 is also expressed in red blood cells, lung, eyes, brains, and capillary endothelial cells (Agre et al. 1993). Several reports analyzing the proximal promoter of Aqp1 suggest that Aqp1 is regulated by thyroid transcription factor-1 and hypertonicity (Umenishi and Schrier 2003; Kim et al. 2007). Although it is not known whether Pax2-binding sites exist in the Aqp1 promoter, our data raise the possibility of a unique regulation of this water channel.

Pax2 overexpression did not induce kidney lineage differentiation or upregulate other genes, including Gdnf, which is known to be a downstream target of Pax2 in vivo (Brophy et al. 2001). Even an increase in the expression levels of integrin α8 and Aqp1 does not imply kidney differentiation, as these two genes are not restricted to the kidney. Thus, Pax2 alone is not sufficient to initiate the kidney development program in ES cells or embryoid bodies. Alternatively, conventional embryoid body formation may not be optimal for such a program. Kim et al. reported that the addition of activin A, retinoic acid, and BMP7 to embryoid bodies increased the expression of many genes involved in kidney development (Kim and Dressler 2005). It will be worth testing whether this condition plus Pax2 overexpression further enhances kidney induction from ES cells. Nonetheless, our rapid method for the generation of tetracycline-inducible ES cell lines will be useful for dissecting the genetic cascades functioning in the early stages of development, as well as for screening candidate genes that direct ES cells toward the desired lineages, which may be applicable to future cell therapies.