Mouse F9 teratocarcinoma cells (ATCC) were cultured in Dulbecco's Modified Eagle's Medium (DMEM; Lonza) enriched with 10% fetal bovine serum (FBS; Gibco) and 1% penicillin-streptomycin antibiotic (PS; Lonza). The cells were seeded in tissue culture (TC) treated 60 mm plates (BD Falcon) for subsequent protein and RNA isolation. When cells were ~40% confluent they were treated with either 10⁻⁷ M retinoic acid (RA; Sigma) overnight, and then subsequently with 10 μM Cyclopamine (Cyc; EMD Millipore) for luciferase assays or treated with either 5 or 10 μM Cyc simultaneously with 10⁻⁷ M RA for immunoblot and RT-qPCR analysis. To activate Hh signaling cells were treated with either 5 or 10 nM Smoothened agonist (SAG; EMD Millipore). To inhibit GSK3, cells were treated with either 5 or 10 nM 6-Bromoindirubin-3′-oxime (BIO, Sigma-Aldrich). As a vehicle control, cells were treated with 0.05% Dimethyl Sulfoxide (DMSO). All cells were incubated at 37°C and 5% CO₂ for a total of 96 h before immunoblot analysis, collected every 24 h for qRT-PCR analysis, or collected after 48 h for luciferase assays.

Cells were transfected with an empty vector or pcDNA3.1-Gata6 (a gift from Dr. M. Jaconi, University of Geneva) for overexpression studies, and pGL3-BARL (Promega) or pGL3-Gli (a gift from Philip Beachy, Stanford University) with pRL-TK (Promega) constructs for luciferase assays using Lipofectamine2000 according to the manufacturer's recommendations (ThermoFisher Scientific). Briefly, 10 μL of Lipofectamine2000 was mixed with a total of 4 μg of expression constructs to transfect cells grown to 60% confluence in 35 mm TC treated plates (BD Falcon); for co-transfection experiments, equal amounts of each construct were used with Lipofectamine2000 to DNA ratio of 10:4. Transfected cells were passaged 24 h later into 60 mm TC plates. Transfected cells were treated at the chemical concentrations described above. In overexpression studies pcDNA3.1 transfected cells were treated with either 0.05% DMSO or 10⁻⁷ M RA, and were collected with pcDNA-Gata6 transfected cells after 96 h.

To determine relative steady-state mRNA levels, total RNA was isolated from F9 cells at 24, 48, 72, and 96 h post-treatment/transfection using the RNeasy Mini Kit (Qiagen). RNA was reverse transcribed into first strand cDNA using the High-Capacity cDNA Reverse Transcription Kit (Applied Biosystems). Primers were designed to amplify the mouse Shh, Dhh, Ihh, Wnt6, c-Myc, Ccnd1, Dkk1, Dab2, Gata6, Smoothened, Patched1, Patched2, Sufu, Gli1, Gli2, and Gli3 nucleotide sequences (Supplementary Table 1). Primers to the constitutively expressed ribosomal gene L14 were used as controls. qRT-PCR was conducted under the following reaction conditions: 500 nM of each reverse and forward primer, SensiFAST SYBR Mix (FroggaBio), and 1 μL of cDNA template. Samples were analyzed with the CFX Connect Real-Time PCR Detection System (Bio-Rad) using the comparative cycle threshold (2^−ΔΔCt) method, where relative expression values were obtained from steady-state mRNA levels normalized to L14 mRNA. Relative expression values were subsequently normalized to DMSO/control plasmid treatment(s) to determine fold change.

Cells were fixed on glass-bottom 12-well plates (MatTek) with 4% paraformaldehyde (Electron Microscopy Sciences), permeabilized with 0.1% Triton X-100, and blocked in 1% bovine serum albumin. Cells were incubated overnight at 4°C with TROMA-1 antibody (1:50; Developmental Studies Hybridoma Bank) and Alexa488-conjugated anti-rat antibody (ThermoScientific) for 2 h, followed by DAPI (Molecular Probes) and Vectashield antifade reagent. All reagents were diluted in phosphate buffered saline. Cells were examined using a Zeiss LSM 5 Duo Vario Microscope with both ZEN and Image-Pro Premier 3D software.

Cells were lysed in 300 μL of 2% sodium dodecyl sulfate buffer containing 62.5 mM Tris-HCL pH 6.8, 10% glycerol, 5% Mercapto-2-ethanol, and 1X Halt Protease Inhibitor Cocktail (Thermo Scientific). Protein concentrations were measured using a Bradford assay (Bio-Rad), and 10–30 μg of total protein were separated on denaturing 10% polyacrylamide gels and transferred to Immunoblot PVDF membrane (BioRad). Membranes were incubated in Tris-buffered saline with 0.1% Tween-20 containing 5% skim milk for 1 h at room temperature. Blots were probed with primary antibodies to TROMA-1 (1:50), DAB2 (1:10,000; Santa Cruz) and β-actin (1:10,000; Pierce) overnight at 4°C. After extensive washes, membranes were probed with the appropriate HRP-conjugated secondary antibodies (1:1,000), and washed extensively with Tris-buffered saline. Signals were detected using the Luminata Classico Western HRP Substrate (Millipore), and imaged using a Chemi Doc Touch Imaging System (Bio-Rad).

Cells transfected with either pGL3-BARL (β-catenin activated reporter luciferase) or pGL3-Gli were treated with 0.05% DMSO (vehicle control), 10⁻⁷ M RA, 10 μM Cyc alone, 5 or 10 nM BIO, 5 or 10 nM SAG, or 10 μM Cyc with 10⁻⁷ M RA, were prepared 48 h after treatment for luciferase assays using the Dual Luciferase Assay Kit as per manufacturer's instructions (Promega). Cells transfected with either pGL3-BARL or pGL3-Gli, and then co-transfected with pcDNA3.1 (empty vector control) or pcDNA3.1-Gata6 were also prepared 48 h post-transfection for luciferase assays. Cells were co-transfected with pRL-TK to normalize luciferase levels. Luciferase expression was quantified using the GloMax System (Promega).

Data from all experiments are representative of at least three independent biological replicates performed on separate occasions. Data comparisons between the control and treated groups were performed using a one-way ANOVA with Tukey's honest significant difference (HSD) post-hoc analysis or Student's t-test (SPSS Statistics for Windows Version 19.0, IBM Crop. Released 2010, Armonk, NY). P-values were considered statistically significant at the 0.05 level.

F9 cells treated with RA differentiate into PrE, and this is accompanied by an increase in Ihh expression (Becker et al., 1997). Two other Hh genes are expressed in mouse and for this reason the expression profiles of Ihh, Shh, and Dhh were examined in undifferentiated F9 cells and those differentiated to PrE. Total RNA was collected from cells treated with DMSO (vehicle control) or 10⁻⁷ M RA from 24 to 96 h post-treatment. Results showed that by 24 h, RA caused a significant increase in Ihh expression relative to DMSO controls (Figure 1A). In contrast, the relatively high Ct values for Shh and Dhh would indicate that there were negligible changes in the levels of mRNA due to RA treatment (data not shown). Thus, Ihh was an RA-responsive gene and transcript levels increased during PrE differentiation, along with known PrE markers Gata6 (Supplementary Figure 2), Wnt6 (Figure 1K). The levels of Dab2 also increased (Figure 1L), which is significant as it is a Wnt target gene (Railo et al., 2009) that encodes a negative regulator of canonical β-catenin signaling in PrE (Golenia et al., 2017). Having identified Ihh as a candidate involved in RA-induced differentiation, we examined the expression of Hh pathway components including Sufu, Smo, Ptch1, Ptch2, Gli1, Gli2, and Gli3. Sufu levels were significantly higher than the DMSO control for the first 72 h of RA treatment (Figure 1B), however, no significant changes were detected in Smo expression (Figure 1C). Expression of Ptch1, a Hh target gene, also did not change significantly in response to RA (Figure 1D), whereas Ptch2 expression, also a Hh target gene, increased significantly with RA after 24 h post-treatment (Figure 1E). Gli1, another Hh target gene, was thoroughly interrogated as its expression was expected to increase because of activated Hh signaling. Despite the modest increase at 24 h, levels were not significantly different from DMSO, and levels dropped significantly after the 24 h time point (Figure 1F). While no significant differences in Gli2 expression relative to DMSO were observed at any time point (Figure 1G), Gli3 expression was found to increase significantly with RA after 48 h (Figure 1H). Confirmation of active Wnt signaling in response to RA came from examining the expression of the Wnt target genes c-Myc and Ccnd1, both of which whose levels increased significantly at 24 h post treatment (Figures 1I,J, respectively). Active Wnt signaling was also confirmed as evident by the RA-induced, significant increase in expression of another Wnt target gene Dickkopf-1 (Dkk1) that encodes a negative regulator of Wnt signaling (Supplementary Figure 3). This evidence for the RA-responsive increase in Ihh, and some, but not all, of the Hh signaling components prompted us to examine Ihh in more detail.

Given that GATA6 is involved in the RA induced differentiation of F9 cells to PrE (Hwang and Kelly, 2012), and with the identification of a putative GATA6 binding site in the Ihh promoter (Supplementary Figure 1), we sought to determine if overexpression of Gata6 would up-regulate Ihh expression. To test this, F9 cells were transfected with pcDNA3.1 (empty vector control), and were treated with DMSO or 10⁻⁷ M RA or were transfected with pcDNA3.1-Gata6. Total RNA was collected 96 h post-treatment for qRT-PCR analysis with L14, Gata6, Ihh, Wnt6, and Dab2 primers (Figure 2). As expected Gata6 expression in RA-treated cells was significantly greater than DMSO-treated cells, and there was no significant difference in the levels between RA and those due to Gata6 overexpression (Figure 2A). RA and Gata6 overexpression significantly increased both Wnt6 and Dab2 expression compared to DMSO controls (Figures 2B,C, respectively), indicative of PrE differentiation (Cho et al., 1999; Zhuang et al., 2003; Krawetz and Kelly, 2008). Gata6 overexpression also increased Ihh expression resulting in levels that were not statistically different from those induced by RA (Figure 2D). These results indicate that GATA6 signaling can directly or indirectly regulate the expression of Ihh and induce PrE, however, they do not address whether this increase in expression translates into activation of the Hh pathway.

To test for Hh pathway activation, F9 cells were co-transfected with pcDNA3.1 (empty vector control) or pcDNA3.1-Gata6, and pGL3-Gli and pRL-TK luciferase reporter constructs. Lysates were collected 48 h post-transfection and luciferase activity was compared between treatments. Results showed that pcDNA3.1-Gata6 overexpression not only caused a significant increase in Gli reporter activity (Figure 2E), but also caused a significant increase in the levels of Ptch1 and Ptch2 (Figures 2F,G, respectively). Thus, Gata6 overexpression served to regulate the Ihh gene, cause an increase in Gli-dependent transcription, and increase transcript levels of Hh responsive genes, which is accompanied by an increase in the expression of Dab2, a Wnt target gene and PrE marker.

Since overexpressing Gata6 stimulated the Gli reporter, and RA induces Gata6, then RA should also activate the pGL3-Gli reporter construct. To confirm this a pGL3-Gli luciferase reporter and a pRL-TK Renilla luciferase construct were co-transfected into F9 cells, which were then subsequently treated with DMSO (vehicle control) or 10⁻⁷ M RA. Results showed that treatment with RA led to a significant increase in luciferase activity relative to that in the DMSO control (Figure 3A). Having established the hierarchy between RA and GATA6, we then activated the Hh pathway with SAG, a SMO agonist, and tested the luciferase activity of the Gli reporter. SAG-treated cells showed a significant increase in luciferase activity compared to DMSO controls (Figure 3A). Thus, the data indicates that upstream control of Gata6 through RA and chemical activation of the Hh pathway are sufficient to induce Gli reporter activity. To further confirm this result, cells were treated with Cyc to block Hh signaling at the level of SMO, to determine that the RA-induced increase in Gli reporter activity was indeed due to activation of the Hh pathway. Treating F9 cells with 10 μM Cyc alone had no effect on the Gli reporter, however, 10 μM Cyc attenuated the RA response in Gli luciferase activity to levels comparable to controls cells (Figure 3A). Therefore, these results indicate that RA activates the Hh pathway in cells destined to form PrE, and the candidate ligand involved is IHH.

To test whether the attenuation of RA-induced increase in Gli reporter activity by Cyc (Figure 3A) had an effect on PrE differentiation, cells were treated with DMSO, 10⁻⁷ M RA, RA and 5 or 10 μM Cyc, or 5 or 10 μM Cyc alone, then collected after 96 h and processed for immunoblot analysis using antibodies to Disabled homolog 2 (DAB2), and TROMA-1, an antibody against KERATIN 8/18, which is a marker of PrE (Duprey et al., 1985). Results show that DAB2 was absent in DMSO controls, but was detected in RA-treated cells (Figure 3B). Cyc alone had no effect on DAB2, regardless of the concentration used, however, the intensity of the RA-induced signal decreased when cells were co-treated with 5 μM Cyc and RA, and more dramatically in cells co-treated with 10 μM Cyc and RA. TROMA-1 signals were detected in all treatment regimens, but the signal intensity was significantly increased with RA treatment (Figure 3B). Interestingly, these signals declined when cells were co-treated with RA and either concentrations of Cyc (Figure 3B). Densitometric analysis confirmed these results (Figure 3C), and confirmed that activation of the Hh pathway is necessary for RA-induced PrE. Dkk1 expression was also examined, and while levels in controls increased significantly due to RA, they dropped when cells were co-treated with RA and Cyc (Supplementary Figure 3). Since the data pointed to active Hh signaling being required during the induction of PrE, we predicted that activation of the Hh pathway alone was sufficient to induce PrE differentiation. To test for sufficiency, F9 cells were treated with DMSO, 10⁻⁷ M RA, RA co-treated with 5 or 10 nM SAG, or 5 or 10 nM SAG alone. Protein lysates were collected after 96 h for immunoblot analysis for DAB2 and TROMA-1 detection (Figure 3D). Results showed a weak DAB2 signal in DMSO-treated cells, which was comparable to those detected when cells were treated with 5 or 10 nM SAG alone. The intensity of the band increased, however, when cells were treated with RA, but there was no further increase in band intensity when RA-treated cells were co-treated with the highest SAG concentration. Similar results were seen with the TROMA-1 signal, and densitometric analysis confirmed that 10 nM SAG alone had not significantly altered DAB2 or TROMA-1 levels relative to the RA-alone control (Figure 3E). Thus, despite SAG activating Gli-mediated transcription (Figure 3A), and upregulating Gli1 and Ptch2 expression, but not Ptch1 (Supplementary Figure 4), the densitometric data confirmed that activating Hh signaling alone was not sufficient to induce PrE. As further proof, immunofluorescence microscopy using the TROMA-1 antibody was used to corroborate the immunoblot data. Results showed that KERATIN-8/18-positive intermediate filaments were only present in cells treated with RA alone (Figure 4). These results indicate the activation of the Hh pathway is necessary for RA-induced F9 cell differentiation; however, activating the Hh pathway alone is not sufficient for the cells to differentiate toward a PrE lineage.

Previous work has established that Wnt/β-catenin and Hh pathways form integrative signaling webs, however, the specifics of the crosstalk between them are poorly understood. Given this information, and the fact that activation of Hh was required, but not sufficient to induce PrE in a process requiring Wnt/β-catenin signaling (Liu et al., 2002), we next wanted to examine the relationship between the two pathways in PrE differentiation. Cells were transfected with either pGL3-Gli or pGL3-BARL reporter constructs, the latter as a readout of active Wnt signaling, and with a Renilla luciferase construct to normalize the data. To determine if the Hh pathway can signal to the Wnt pathway, F9 cells were first transfected with the pGL3-BARL construct and treated with 10 nM BIO, a GSK3 inhibitor, or with DMSO, 10⁻⁷ M RA, or 5 or 10 nM SAG. Results showed that the 5 and 10 nM concentrations of SAG were not sufficient to significantly activate pGL3-BARL activity (Figure 5A). RA, as expected, caused a significant increase in BARL activity relative to the control, and these results were not significantly different from those seen from BIO-treated cells (Figures 5A,B). The experiment was repeated, however, in this case the Hh pathway was chemically inhibited using 10 μM Cyc. Given that Cyc treatment with RA significantly decreased the levels of two Wnt signaling components, DAB2 (Figure 3B) and Dkk1 (Supplementary Figure 3), we expected Cyc would affect the activity of the pGL3-BARL reporter. However, results showed that when cells were co-treated with 10 μM Cyc and RA there was no significant difference in luciferase activity compared to control treatments (Figure 5B). This reduction contrasted the significant increase in reporter activity caused by RA alone (Figure 5B). Together, these results revealed that while active Hh signaling alone is not sufficient to induce markers of PrE differentiation or increase TCF/LEF transcriptional activity, the ability of Cyc to attenuate RA-induced TCF reporter activity and PrE differentiation, implicates active Hh signaling as being involved. Having identified a link between Hh and Wnt/β-catenin, one possible candidate serving as a node that could inhibit both pathways is GSK3β (Jia et al., 2002). If this is true in F9 cells, then inhibiting GSK3β activity should also stimulate the Gli reporter. To test this, F9 cells were transfected with the pGL3-Gli reporter and then treated with DMSO, 10 nM SAG (positive control), 10⁻⁷ M RA (positive control), or with 5 or 10 nM BIO (Figure 6). Results showed that the SAG treatment caused a significant increase in Gli reporter activity that was not significantly different from cells treated with RA. Interestingly, the 5 and 10 nM BIO treatments caused an increase, both of which were significantly different from DMSO controls, and comparable to the SAG and RA results. Thus, while activating the Hh pathway had no detectable effect on β-catenin-TCF/LEF-dependent transcription, inhibiting GSK3β increases the transcriptional activity of both pathways, and specifically the activity of Gli-mediated transcription.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest. The reviewer CC and handling Editor declared their shared affiliation.