Canonical Wnt/β-Catenin Regulation of Liver Receptor Homolog-1 Mediates Pluripotency Gene Expression

Derivation of β-Catenin^-/- ES Cells

β-catenin^flox/flox male mice (Jackson Laboratories, Bar Harbor, ME, http://www.jax.org) were crossed with transgenic females expressing Cre-recombinase driven by the Zona Pelucida protein-3 (ZP3) promoter (Jackson Laboratories) triggering an early recombination event [22,30–32]. β-catenin^+/flox, ZP3-Cre/⁺ females were derived and backcrossed to wt males to yield β-catenin^-/+ offspring. Resultant β-catenin^-/+ offspring were isolated and maintained in a colony (Fig. 1A). β-catenin heterozygote mice were bred and blastocysts isolated at 3.5 dpc. ESC lines were derived by blastocyst outgrowth as previously described [12].

Genotyping of ES Cell Lines and Embryos

DNA was extracted from ES cell lines and mouse tails after overnight digestion in lysis buffer (0.5% sodium dodecyl sulfate, 50 mM Tris pH 7.5, 0.1 M NaCl, 5 mM EDTA, 0.2 mg/ml proteinase K), followed by phenol/chloroform/isoamyl alchohol extraction (Invitrogen, Carlsbad, CA, http://www.invitrogen.com). Genotyping was performed using the following specific primers for the β-catenin locus: RM41: 5′-AAGGTGGAGTGATGAAA- GTTGTT-3′, RM42: 5′-CACCATGTCCTCTGTCTATTC-3′, and RM43: 5′-TACA CTATTGAATCACAGGGACTT-3′ [22].

Cell Culture, In Vitro Differentiation, and Small Molecule Treatments

ES cell lines were maintained on plates treated with 0.1% gelatin (Sigma, St. Louis, MO, http://www.sigmaaldrich.com), under standard conditions described previously [12]. ES cell lines were differentiated by withdrawal of LIF from ESC media and addition of 1 μM all-trans-retinoic acid (RA; Sigma). Embryoid body differentiation was carried out by forming the aggregates in low-attachment plates (25,000 cells/ml) or by hanging drop (500 cells/20 micro l drop) for 3 days prior to transferring the embryoid bodies to gelatinized plates. ESC lines were treated with 2 micro M BIO (Calbiochem, Darmstadt, Germany, http://www.emdchemicals.com) reconstituted in dimethyl sulfoxide (DMSO) or 100 ng/ml mWnt3a (R&D Systems, Minneapolis, MN, http://www.rndsystems.com/). Recombinant Dickkopf-1 (Dkk-1; R&D Systems) was added at increasing concentrations from 1 ng/ml to 250 ng/ml. CHIR99021 (Stemgent, San Diego, CA, http://www.stemgent.com) was added at an increasing concentration (10 nM, 100 nM, 1 micro M, 2.5 micro M, and 5 micro M). All treatments were performed in the absence of LIF.

Transient Transfection and Plasmids

The Lrh-1 promoter regions were generously provided by You-Hua Xie [29] and were transfected alongside the TOP-FLASH reporter (Upstate, Billerica, MA, http://www.millipore.com) under the following conditions: 125,000 ESCs were transfected in suspension with 100 ng of reporter, 2 ng of Renilla luciferase expression vector (Promega, Madison, WI, http://www.promega.com), using 2.5 micro l Lipofectamine 2000 (Invitrogen), and plated in single well of a 12-well plate. Each transfection was assayed in triplicate. Promoter activity was measured using a Berthold Centro LB960 dual-luciferase luminometer and recorded in relative light units after normalizing to Renilla luciferase.

Immunofluorescence and Western Blot

Antibody information and conditions are listed in (Supporting Information Table 1). Western Blot analysis was performed under standard denaturing conditions.

Quantitative Reverse Transcription Polymerase Chain Reaction (qRT-PCR) Analysis

Total RNA was extracted from ES cells using Trizol reagent (Invitrogen). cDNA was generated using the Super Script III First Strand Synthesis Kit (Invitrogen) with Oligo dT primers following the manufacturer's protocol. Quantitative expression of endogenous genes was carried out using QuantiFast SYBR Green PCR (Qiagen, Valencia, CA, http://www.qiagen.com) on a Step 1 Plus Real Time PCR System (Applied Biosystems, Carlsbad, CA, https://products.appliedbiosystems.com). Target gene expression was normalized to β-actin expression in all experiments. For gene-specific primers see Supporting Information Table 2.

Chromatin Immunoprecipitation

Chromatin immunoprecipitation (ChIP) assays were performed following a protocol modified from previous reports [12,33]. Undifferentiated wt and β-catenin^-/- ESCs were treated for 6 hours with 2 micro M BIO or differentiated for 2 days with 1 micro M RA prior to fixation with 1.1% formaldehyde. Cross-linked DNA underwent enzymatic shearing using the ChIP-IT Express Enzymatic kit (Active Motif, Carlsbad, CA, http://www.activemotif.com) following the manufacturers protocol. Lysate protein concentration was determined postshearing and lysates diluted to 0.25 micro g/micro l. For antibody information and immunoprecipitation conditions see (Supporting Information Table 1). Recovered DNA was quantitatively amplified with gene-specific primers spanning the putative LEF/TCF sites using QuantiFast SYBR Green PCR (Qiagen) and normalized to 10% input. For ChIP primer sequences see Supporting Information Table 3.

Teratoma Formation

Teratomas were generated from wt and Lrh-1^-/- ESC after injection of approximately 10⁶ cells into the hind-quarters of nude SCID mice. Histological analysis was carried out on teratoma sections after standard H&E staining.

Generation of Myc-Tagged Lrh-1 β-Catenin^-/- Stable Cell Line

The previously described embryonic isoform of Lrh-1 was cloned into a pCMV-myc expression vector (Clontech) using PCR primers: forward 5′-TCGAATTCTCATGCTGCCCA AAG TGGAG-3′ and reverse 5′-TAGTCGACTTAGGCTCTT TTGGCATGCAGCA-3′ [29]. Myc-tagged Lrh-1 was then subcloned into the pEF1α-ORF-IRES-neo vector described in with PCR primers: forward 5′-TCGCTAGCATGGCA TCAA TGCAGAAGCT-3′, and reverse 5′- TCGCGGCCGCTTAGG CTCTTTTGGCATGCA-3′ [33]. The pEF1α-mycLrh1(ES)-IRES-neo vector was targeted into the β-catenin^-/- ESCs via electroporation (25 micro g linearized vector/10,000,000 cells). Heterologous recombinants were screened by selection with G418 for approximately 10 days, and clonal resistant ES colonies were isolated and expanded in culture.

Statistical Analysis

Statistical significance was determined by performing two-tailed t-tests where appropriate.

Generation and Characterization of β-Catenin^-/- ESC

To evaluate the role of canonical Wnt signaling in ESC pluripotency, β-catenin^-/- ESC were derived by blastocyst outgrowth following a germline Cre-lox recombination event yielding heterozygous mice [22] (Fig. 1A). Western blot analysis confirms the absence of β-catenin protein in the mutant cell line, however, Oct4 and Nanog are expressed (Fig. 1B). Although β-catenin^-/- ESC express these pluripotency factors, they are not pluripotent insofar as they exhibit pronounced defects in endoderm and mesoderm induction on embryoid body differentiation (Supporting Information Fig. 1). Furthermore, two separate attempts at forming teratomas with the β-catenin^-/- ESC yielded no detectable tumors. These findings are consistent with the phenotype of the β-catenin^-/- mouse, which exhibits a gastrulation defect in vivo [20]. Nevertheless, in culture, the program for self-renewal and the expression of pluripotency factors is intact making the β-catenin^-/- ESC a useful tool for studying canonical Wnt signaling in wt mESC.

Proper expression of pluripotency factors is paramount for ESC self-renewal [34], so β-catenin^-/- ESC were characterized on differentiation by embryoid body formation as well as on treatment with RA. Oct4 and Nanog exhibit significant misexpression in β-catenin^-/- ESC (Fig. 1C) and are more rapidly silenced on RA-mediated differentiation, which is apparent at the protein level (Fig. 1D and Supporting Information Fig. 2). Lrh-1 is expressed at basal levels in β-catenin^-/- ESC and is nonresponsive to RA treatment (Fig. 1D and Supporting Information Fig. 2), indicating that Lrh-1 is misregulated in the β-catenin^-/- ESC. Genetically, this casts Lrh-1 as a putative β-catenin target gene.

To support this finding, we employed a nongenetic approach to inhibit canonical Wnt signaling by treating wt ESC with recombinant Dkk-1. Dkk-1 is a secreted factor that prevents the association of Frizzled and Lrp5-6, a necessary step in the stabilization of β-catenin [35]. Oct4, Nanog, and Lrh-1 exhibit dose-dependent repression on treatment with Dkk-1 (Fig. 1E). The observation that Dkk-1 treatment is sufficient to repress pluripotency gene expression provides independent confirmation of the defect in pluripotency gene expression observed in the β-catenin^-/- ESC.

Lrh-1 Is Induced in a β-Catenin-Dependent Manner

In contrast to Dkk1 treatment, we sought to determine the effects of promoting canonical Wnt signaling in ESC. Promoting β-catenin stabilization with BIO, a pharmacological inhibitor of GSK3 [27], triggered a significant induction of Lrh-1 expression that was maintained for 5 days (Fig. 2A, upper graph). This effect was β-catenin-dependent, as no response was observed in the β-catenin^-/- ESC (Fig. 2A, lower graph). Additionally, β-catenin-dependent induction of Lrh-1 was recapitulated on treatment with CHIR99021, an alternate GSK3 inhibitor that is a component of the (2i/3i) GSK3/mitogen activated protein kinase Kinase (MEK) inhibitor cocktails that are reported to drive ground state pluripotency independent of LIF [7] (Supporting Information Fig. 3).

The moderate specificity that BIO has for GSK3 and the fact that GSK3 has other targets besides β-catenin led us to seek additional evidence to implicate canonical Wnt signaling in Lrh-1 induction [36,37]. Wnt stimulation was investigated as well by supplementing media with recombinant Wnt3a and Wnt5, regarded as “canonical” and “noncanonical” Wnts, respectively [38,39]. Although Wnt3a triggered a significant dose-dependent induction of Lrh-1, the noncanonical Wnt5a yielded only a modest induction that was not dose-dependent (Fig. 2B). Furthermore, Lrh-1 is induced in a β-catenin-dependent manner on stimulation with Wnt3a confirming that Lrh-1 is a novel target of the canonical Wnt signaling pathway (Fig. 2C). Wnt stimulation fully recapitulates the effects observed on GSK3 inhibition and validates BIO as a useful modulator of the pathway in our system.

Lrh-1 Is a Direct Target of β-Catenin in mESC

LEF/TCF transcription factors mediate Wnt induced changes in gene expression through an interaction with nuclear β-catenin [16]. The murine Lrh-1 gene contains an embryonic stem cell specific (ES) isoform driven by an alternate promoter downstream of the full-length (FL) promoter [29]. In silico analysis of each promoter region identified multiple LEF/TCF consensus sites (A/T A/T CAAAG) within each promoter [40]. Reporter constructs spanning these regions and described previously were transfected into wt and β-catenin^-/- ESC to compare activity [29]. Although the Lrh-1 FL promoter exhibits basal activity in both cell lines, the 1.4 kb ES promoter revealed significantly higher activity in wt cells compared with the mutant cell line, which was suppressed on differentiation with RA, and induced in a β-catenin-dependent manner on BIO treatment (Fig. 2D). Transfecting serial truncations of the ES promoter yielded decreased activity, and proved less responsive to BIO implying that the putative LEF/TCF sites identified are important for activity in vitro. The decreased activity and blunted response to BIO in β-catenin^-/- ESC closely mirrored TOP-FLASH activity [41] (Fig. 2E). Interestingly, TOP-FLASH activity is not significantly repressed on RA treatment in wt ESC. This is in contrast to the repressed activity observed on RA treatment in cells transfected with the 1.4 kb ES-specific Lrh-1 reporter, and may be due to promoter specific effects.

To determine if Lrh-1 induction is a primary effect of β-catenin stabilization, ChIP against β-catenin was performed. Of the eight putative LEF/TCF sites identified in the Lrh-1 promoter the highest degree of evolutionary conservation was observed at sites proximal to transcription start site in each promoter, here named: Ts(FL)-1, Ts(FL)-2, Ts(ES)-1, and Ts(ES)-2 (Fig. 3A). We therefore investigated these regions for β-catenin binding. Wild-type and β-catenin^-/- ESC were treated with BIO for 6 hours before fixation to ensure that only the immediate transcriptional effects of β-catenin would be assayed, whereas RA-treated cells were fixed after 2 days of treatment to reflect a differentiated ES cell. In the LIF-treated undifferentiated ESC, β-catenin exhibited little enrichment in both the FL and ES-specific Lrh-1 promoter, however, LEF/TCF sites: Ts(FL)-1, Ts(ES)-1, and Ts(ES)-2 became highly enriched after treatment with BIO (Fig. 3B). β-catenin enrichment at Ts(FL-2) exhibited negligible enrichment on each treatment, implying that Ts(FL)-2 may be irrelevant in mediating β-catenin regulation of Lrh-1. Any enrichment of β-catenin is lost on RA differentiation suggesting that the direct regulation of Lrh-1 by β-catenin is not a characteristic of RA-mediated differentiation. In controls for the ChIP, enrichment of β-catenin was observed at the Axin2 promoter after BIO treatment [42], whereas enrichment at an off-target Lrh-1 intronic region was meager (Fig. 3B).

β-Catenin Regulates Lrh-1 Expression Through Tcf3

β-catenin promotes target gene expression by interacting with members of the LEF/TCF family of transcription factors. Under nonstimulatory conditions, LEF/TCF factors mediate transcriptional repression of Wnt target genes through an interaction with Groucho [16]. The LEF/TCF factor, Tcf3, was previously reported to directly repress Nanog gene expression in ESC, and genetic ablation or knockdown of Tcf3 suggested that it acts as a limiter of self-renewal by counteracting effects of Nanog and Oct4 [43,44]. Others report that Tcf3 is an integral member of the Oct4/Sox2/Nanog pluripotency axis [45,46]. Considering these findings, we sought to determine if β-catenin regulation of Lrh-1 occurred through Tcf3. The Lrh-1 reporters described in Figure 2D were transfected into undifferentiated Tcf3^-/- ESC with an increasing concentration of Tcf3 expression vector [39]. As reported previously, the Lrh-1 FL promoter exhibited basal activity, however, the 1.4-kb ES-specific promoter yielded high promoter activity that was repressed by Tcf3 in a concentration-dependent manner (Fig. 3C). When serial truncations of this promoter were assayed, the ability of Tcf3 to repress activity was diminished on loss of the putative LEF/TCF sites in each promoter, implying that they are necessary for mediating Tcf3 repression of Lrh-1 in vitro (Fig. 3C).

ChIP of Tcf3 exhibited enrichment at Ts(ES)-1 and Ts(ES)-2 in the Lrh-1 ES-specific promoter, but enrichment was not observed in the FL-promoter (Fig. 3D). Tcf3 enrichment at the Oct4 and GAPDH promoters was evaluated and served as positive and negative controls for the ChIP assay [45,46]. Importantly, β-catenin and Tcf3 exhibit significant enrichment at the Lrh-1 ES-promoter, indicating a likely partnership between the two in the regulation of Lrh-1.

Lrh-1^-/- ESCs Express Decreased Levels of Oct4 and Nanog

If Lrh-1 is indeed a novel target gene of β-catenin in vivo, then we would expect the Lrh-1^-/- ESC to recapitulate the defect in pluripotency gene expression observed in the β-catenin^-/- ESC [12]. Wild-type and Lrh-1^-/- ESC were differentiated as embryoid bodies or by RA-mediated differentiation under the same conditions reported in Figure 1. As reported for the β-catenin^-/- ESC, the Lrh-1^-/- ESC express significantly decreased levels of Oct4 and Nanog, while expressing similar levels of β-catenin (Fig. 4A, Supporting Information Fig. 4). This defect was also apparent at the protein level in which Oct4 and Nanog underwent more rapid silencing in the Lrh-1^-/- ESC (Fig. 4B). Together, the Lrh-1^-/- ESC mirror the defect in pluripotency gene expression observed in the β-catenin^-/- ESC. In contrast to the β-catenin^-/- ESC, the Lrh-1^-/- ESC readily form teratomas composed of tissues of all three germ layers (Fig. 4C). Additionally, Lrh-1^-/- ESC can express markers of all three germ lineages on embryoid body differentiation (Supporting Information Fig. 5). In light of this finding, it seems that the gastrulation defect observed in β-catenin-null mice may go well beyond the misregulation of Lrh-1 alone.

Stabilizing β-Catenin Promotes Oct4 Expression in an Lrh-1-Dependent Manner

To explore whether Lrh-1 misexpression could account for the decreased pluripotency gene expression observed in the β-catenin^-/- ESC, we assayed Oct4 expression in our cell lines after stimulating Wnt signaling with BIO or Wnt3a in the absence of LIF. BIO treatment of wt ESC caused a significant induction of Oct4 and was able to maintain undifferentiated levels of expression out to 5 days of treatment (Fig. 4D, upper graph). Strikingly, the BIO-mediated induction of Oct4, and subsequent maintenance of expression was observed to be both β-catenin and Lrh-1-dependent (Fig. 4D, middle and lower graphs). Despite elevated transcript levels, BIO is not sufficient to maintain Oct4 protein levels in long-term culture, which is consistent with previous reports [7] (Supporting Information Fig. 6). Treating the ES cell lines with recombinant Wnt3a yielded approximately a twofold induction in Oct4, that was entirely β-catenin-dependent (Fig. 4E). Wnt3a yielded a modest induction of Oct4 in the Lrh-1^-/- ESC, however, it was significantly less than that observed in wt ESC, suggesting that full induction of Oct4 requires Lrh-1 and β-catenin (Fig. 4F).

Lrh-1 Regulates the Tbx3/Nanog Signaling Axis that Drives Self-Renewal

To investigate the significance of β-catenin regulation of Lrh-1 in ESC, we wanted to determine if Lrh-1 affects the Stat3/Klf4/Sox2 or Tbx3/Nanog signaling axes. In culture, LIF supports mESC self-renewal by promoting two parallel signaling axes [6]. One axis includes activation of Stat3, causing induction of Klf4, which in turn regulates expression of Sox2, whereas the parallel axis comprises activation of PI3K, causing induction of Tbx3, and subsequent regulation of Nanog expression. Intriguingly, LIF is not required in vivo to maintain pluripotency suggesting alternate mechanisms may promote these signaling axes [9]. To investigate if β-catenin regulation of Lrh-1 has a role in regulating these signaling axes, wt ESCs were treated with BIO to determine if stimulating Wnt signaling is sufficient to induce expression of these factors. Lrh-1, Oct4, Nanog, and Tbx3, all exhibit significant induction on BIO treatment in wt ESC, and are repressed on differentiation with RA (Fig. 5A). Importantly, the induction of these factors is both β-catenin-dependent and Lrh-1-dependent because neither of the mutant cell line exhibited a response to BIO. In contrast, BIO was ineffective at inducing expression of Klf4 in wt and mutant ESC (Fig. 5A).

To support the pharmacological data, we incorporated a genetic approach by stably overexpressing a myc-tagged ES isoform of Lrh-1 in a wt mESC [33] (Fig. 5B). Overexpression of Lrh-1 results in a considerable induction of Oct4, Nanog, and Tbx3 expression above the levels observed in wt ESC (Fig. 5C). Additionally, these factors all exhibit decreased expression in Lrh-1^-/- ESC suggesting that Lrh-1 is required to maintain wt levels of expression. Consistent with BIO treatment, Klf4 expression is nonresponsive to either Lrh-1 ablation or overexpression. ChIP of Lrh-1 was performed to determine whether Lrh-1 regulated Tbx3 and Nanog expression directly. Consistent with our previous findings, Lrh-1 enrichment was observed at the Oct4 PP and PE in ESC [12] (Fig. 5D). Lrh-1 was also enriched at a response element in the Nanog promoter (Fig. 5E) and in the Tbx3 enhancer region (Fig. 5F). Consistent with the previous data, Lrh-1 was not enriched at the Klf4 locus despite two nearly perfect consensus sites within the promoter (Fig. 5G). These data suggest that Lrh-1 directly regulates Tbx3, Nanog, and Oct4 expression in undifferentiated ESCs, and thereby selectively promotes the Tbx3/Nanog pluripotency axis.

Ectopic Expression of Lrh-1 Restores Oct4/Nanog Gene Expression in β-Catenin^-/- ESC

To further substantiate our hypothesis that misregulation of Lrh-1 in β-catenin^-/- ESC is in part causing the defect in pluripotency gene expression, Lrh-1 was stably expressed in a β-catenin^-/- ESC as described in (Fig. 5B), in an attempt to restore wt levels of pluripotency gene expression (Supporting Information Fig. 7). The stable clone, β-catenin^-/- myc-Lrh-1(B2), expressed Lrh-1 at levels fourfold higher than that observed in undifferentiated wt cells (Fig. 6A). Importantly, overexpression was able to restore Oct4 expression, as well as prolong expression during RA differentiation (Fig. 6B). Similarly, Nanog expression was restored to wt levels in undifferentiated cells (Fig. 6C). Canonical Wnt signaling was not restored in the stable cell line, which is apparent by the absence of β-catenin expression, and confirms that the observed effects on pluripotency gene expression are specific to ectopic Lrh-1 expression (Fig. 6D).