Adenovirus-Mediated Increase of HNF-3 Levels Stimulates Expression of Transthyretin and Sonic Hedgehog, Which Is Associated With F9 Cell Differentiation Toward the Visceral Endoderm Lineage

Adenovirus Vector Preparation

Two 1.6-kb EcoRI fragments containing the entire coding sequence for either the rat HNF-3α or rat HNF-3β protein were subcloned into the adenovirus recombination vector, pAdCL-XEB. This vector contained the cytomegalovirus (CMV) promoter, a multiple cloning site (including EcoRI), the SV40 poly A site, and a splice donor/acceptor site between human adenovirus type 5 genome sequences 1–355 and 3333–5788 (see also Fig. 1). We used the overlap recombination procedure (14) to construct adenoviruses that overexpressed rat HNF-3α and rat HNF-3β genes. The resulting plasmids, denoted pAdHNF3α or pAdHNF3β, were linearized with AscI and co-transfected with type 5 adenovirus dl324 DNA into HEK (human embryonic kidney) 293 cells. dl324, which contains a deletion in the E1 region (genome positions 1334–3639 are missing) and in the E3 region, was digested with ClaI at genome position 918. We transfected 5 μg of each plasmid (HNF-3α or HNF-3β) together with 1 μg of cut virus per 35-mm cell culture plate. After providing sufficient time for recombination to take place (between 5 and 14 days), 293 cells were lysed and potential recombinant viruses were detected by performing standard plaque assays on 293 cells (35). Following plaque purification, viral DNA was isolated and used for the preparation of Southern blots. HNF-3-containing viruses were detected by probing the blots with probes specific for HNF-3α and HNF-3β, followed by PCR with HNF-3-specific primers. Homologous recombination between the linearized plasmids and dl324 replaced portions of the adenovirus E1 region with the 1.6-kb HNF-3α or HNF-3β gene fragments, and created the AdHNF3α or AdHNF3β, respectively (see Fig. 1). Using the procedure described above, we also generated a control virus (AdLacZ), which contained the prokaryotic LacZ (β-galactosidase) gene instead of the HNF-3 cDNAs.

Purification of Recombinant Adenoviruses

For large-scale preparations, 20 tissue culture plates (150 × 25 mm) of HEK 293 cells (1–2 × 10⁷ cells/plate) were infected with AdHNF3α, AdHNF3β, or AdLacZ at a multiplicity of infection (MOI) of 1–5 plaque forming units (pfu)/cell. After an additional 2 days of culture, the cells were harvested and washed once with PBS (8 mM Na₂HPO₄, 1.5 mM KH₂PO₄, 135 mM NaCl, 2.5 mM KCl). The cells were then resuspended in lysis buffer (10 mM Tris-HCl, pH 7.8, 20 mM MgCl₂, 0.5% DOC). Freeze thawing lysed the samples and DNase was added to a final concentration of 10 μg/ml. After 30 min at 37°C, an equal volume of fluorocarbon (1,1,2-tricholorotrifluoroethane, Sigma) was added and the samples were incubated on a rocker for 10 min at room temperature. Following a centrifugation (10 min at 1000 × g), the supernatant was layered onto a cesium chloride gradient (1.4 g/ml and 1.2 g/ml in 50 mM Tris-HCl, pH 7.8, 10 mM MgCl₂) and centrifuged at 90,000 × g for 3 h at 4°C (SW 28 rotor). Viral bands were harvested and mixed with an equal volume of loading buffer (50 mM Tris-HCl, pH 7.8, 10 mM MgCl₂). The samples were again layered onto a cesium chloride gradient and centrifuged overnight. Viral bands were removed, mixed with four volumes of freezing solution (0.1% BSA, 50% glycerol, 10 mM Tris-HCl, pH 7.8, 100 mM NaCl), and stored at −20° C.

Cell Culture Experiments

Mouse F9 embryonal carcinoma cells, mouse P19 embryonal carcinoma cells, and HEK 293 cells were cultivated in Dulbecco’s modified Eagle’s medium (DMEM) containing 7.5% ultra calf serum and 2.5% fetal calf serum (Inovar). For viral infections, cultures were grown on tissue culture dishes to 60–70% confluence. Cells were washed once with serum-free DMEM and then exposed to serum-free DMEM containing AdHNF3α, AdHNF3β, or AdLacZ at 20 pfu/cell. After an incubation of 1 h at 37°C, the medium was replaced with DMEM containing 7.5% ultra calf serum and 2.5% fetal calf serum and cells were cultured for various times at 37°C.

To measure infection efficiency, which was represented by the β-galactosidase activity in AdLacZ-infected cells, they were fixed with 0.05% glutaral-dehyde in 0.1 M phosphate buffer, pH 7.3, for 5 min at room temperature. The cells were then washed with ice-cold PBS (8 mM Na₂HPO₄, 1.5 mM KH₂PO₄, 135 mM NaCl, 2.5 mM KCl) and stained with X-gal solution [1 mg/ml 5-bromo-4-chloro-3-in-doyl-β-d-galactopyranoside, 0.1 M phosphate buffer, pH 7.3, 3 mM K₃Fe(CN)₆, 3 mM K₄Fe(CN)₆, 1.3 mM MgCl₂] for 4–6 h at 37°C.

RNase Protection Assay

Nuclear Protein Extraction and Gel Shift Assay

Nuclear proteins were extracted from cell cultures according to the protocol of Andrews and Faller (1). Binding of HNF-3 proteins to DNA oligonucleotides was initiated by mixing the following components: 1 ng of kinased double-stranded oligonucleotide, binding buffer (20 mM Tris, pH 7.5, 5% glycerol, 40 mM KCl, 1 mM MgCl₂, 0.5 mM DTT, 1 mM EDTA), 3 μg of poly(dI-dC), and 10 μg of extract. One microliter of HNF-3α-specific or HNF-3β-specific antiserum (15) was added where indicated in a total reaction mixture of 30 μl. After 30 min at room temperature, the reaction mixture was loaded onto a 5% polyacrylamide gel (acrylamide/bis 29:1). Electrophoresis was performed for 120 min at 150 V (room temperature). Subsequently, the gel was dried and subjected to autoradiography as described above. The strong affinity HNF-3 binding site from the TTR promoter (6,29,40) was employed by hybridizing the following double-stranded DNA oligonucleotides: 5′-gttgactaagtcaataatcagaatcagca-3′ and 3′-caactgattcagttattagtcttagtcgt-5′.

Transfections, Adenovirus Infection, and CAT Assay

Coinfection experiments with an HNF-3-dependent chloramphenicol acetyltransferase (CAT) reporter construct demonstrated that the adenovirus-transduced HNF-3α and HNF-3β proteins were transcriptionally active. We first transfected an HNF-3-dependent CAT reporter plasmid (30) into P19 cells, then infected them with either AdHNF3α, AdHNF3β, or AdLacZ and 24 h later we performed CAT assays with protein extracts prepared from these cells. On the day before transfection, P19 embryonal carcinoma cells were split at a density of ∼10⁵ cells per ml in 10-cm tissue culture dishes in 10 ml medium. Transfections of the HNF-3-dependent CAT reporter construct method and SV40 promoter-driven β-galactosidase (β-gal) constructs (pSV-LacZ, included as internal control) were done by the calcium phosphate method as described previously (30). The transfected P19 cells were then infected with either AdHNF3α, AdHNFβ, or AdLacZ at a multiplicity of infection (MOI) of 20 pfu/cell. The cells were harvested after another 24 h of culture, lysed by freezing and thawing, and the extract was used for the β-gal assay (38) and the CAT assay. The β-gal assay was performed by reacting a specified amount of protein with substrate ONPG (O-nitrophenyl-β-d-galactopyranoside) and measuring the absorbance at 420 nm. Extracts containing equal β-gal activities were used for CAT assay. For this purpose, the extract was incubated at 65°C for 15 min to inactivate the deacetylating enzymes and subsequently incubated with 20 μl 4 mM acetyl CoA and 3 μl [¹⁴C]chloramphenicol at 37°C for 2 h with repeated addition of acetyl CoA. CAT activity was monitored by thin-layer chromatography on silica plates in CHCl₃/CH₃OH (95:5) followed by autoradiography.

AdHNF3α and AdHNF3β Give Rise to Functional Protein Products

In order to determine whether the HNF-3 transcription factors alone are sufficient to induce expression of downstream target genes in F9 cells without reti-noic acid differentiation, we have used adenovirus-based expression vectors to increase expression of HNF-3α and HNF-3β proteins. We utilized the overlap recombination procedure (14) to construct adenoviruses that overexpress rat HNF-3α and HNF-3β genes, as described in Materials and Methods (Fig. 1). In order to verify that AdHNF3α and AdHNF3β generated functional transcription factors, we first infected mouse P19 embryonal carcinoma cells with AdHNF3α or AdHNF3β at a MOI of 20 pfu/cell. Control adenovirus AdLacZ infections detected by β-gal enzyme staining with the X-gal substrate determined that greater than 90% of the cells were infected using an MOI of 20 pfu/cell. RNA was isolated from P19 cells at different intervals following infection with either AdHNF3α or AdHNF3β and used for RNase protection assays to determine HNF-3α or HNF-3β mRNA levels (see Materials and Methods). Infection of P19 cells with AdHNF3α led to increased HNF-3α mRNA levels between 2 and 12 h postinfection (PI) (Fig. 2A, lanes 6–9). No HNF-3α mRNA was detected in P19 cells that were either infected with control LacZ-expressing adenoviruses (AdLacZ) or mock infected cells (Fig. 2A, lanes 1–5). Likewise, infection of P19 cells with AdHNF3β caused induction of HNF-3β mRNA at 2 h PI, with levels gradually increasing through the 12-h PI time point (Fig. 2B, lanes 6–9).

In order to demonstrate that AdHNF3α and AdHNFβ infections led to increased HNF-3α and HNF-3β protein levels, we prepared nuclear extracts from adenovirus-infected cells and gel shift experiments were carried out to visualize the HNF-3 protein–DNA complexes (see Materials and Methods). HNF-3α protein–DNA complex formation was detected between 6 and 72 h PI (Fig. 3A, lanes 4–8), and disruption of this protein–DNA complex by addition of the HNF-3α antibody in the binding reaction confirmed the presence of HNF-3α protein in the complex (Fig. 3A, lane 9). Furthermore, no abundant HNF-3α protein–DNA complex was found in binding reactions using nuclear extracts from either mock-infected or AdLacZ-infected controls (Fig. 3A, lanes 1–3). Likewise, infection of P19 cells with AdHNF3β resulted in the increased HNF-3β protein–DNA complexes by 6 h PI, with continued gradual increase until the 24-h PI time point (Fig. 3A, lanes 1–7). Furthermore, supershifting of the adenovirus-derived HNF-3β protein–DNA complex by addition of the HNF-3β-specific antibody confirmed the presence of HNF-3β protein there (37).

Finally, coinfection experiments with an HNF-3-dependent reporter construct demonstrated that the adenovirus-transduced HNF-3α and HNF-3β proteins were transcriptionally active. We first transfected an HNF-3-dependent CAT reporter plasmid (30) into P19 cells, then infected them with either AdHNF3α, AdHNF3β, or AdLacZ, and then 24 h later we performed CAT assays with protein extracts prepared from these cells (see Materials and Methods). Both AdHNF3α and AdHNF3β were potent activators of HNF-3-dependent CAT reporter gene expression, whereas no transcriptional activation was found with AdLacZ infection, indicating that this stimulation was specific (Fig. 4). In summary, these studies demonstrated that adenovirus infections could increase levels of transcriptionally active HNF-3α and HNF-3β proteins.

HNF-3α Induces Expression of HNF-3β, Shh, and TTR in F9 Cells

We have previously shown that retinoic acid-mediated differentiation of F9 cells causes a cascade of gene expression in which increased HNF-3α levels are observed first, followed by elevated HNF-3β, which precedes activation of TTR (15,37). Because HNF-3β Shh and TTR promoter regions contain HNF-3 binding sites (4,6,12,23,31), we next examined whether HNF-3α could mediate activation of these genes in the absence of retinoic acid stimulation. Infections of F9 cells with AdHNF3α (20 pfu/cell) caused a large increase in rat HNF-3α levels between 12 h and 3 days PI, with subsequent, gradual decrease of these levels between 3 and 5 days PI (Fig. 5, HNF-3α panel). HNF-3α-mediated induction of endogenous mouse HNF-3β, however, was not observed until 4 days PI and elevated HNF-3β levels were sustained throughout the time points examined (Fig. 5A, HNF-3β lanes 7–9). These studies demonstrate that although HNF-3α is involved in stimulating endogenous HNF-3β expression in F9 cells, sustained HNF-3α levels are not necessary to maintain HNF-3β expression. Furthermore, HNF-3α mediated a later increase of TTR expression at 6 days PI, which is 2 days after the induction of endogenous HNF-3β levels (Fig. 5A, TTR lane 9). Control reactions performed with either F9 stem cell mRNA or AdLacZ-infected F9 cell mRNA (Fig. 5A lanes 1–2) demonstrated that induction of HNF-3β and transthyretin were specific for AdHNF3α infection. Moreover, AdHNF3α infection of F9 cells stimulated a transient increase in Shh expression between 12 h and 1 day PI during a period when HNF-3α levels are maximal (Fig. 5B, Shh, lanes 3–4). By comparison, AdHNF3α infection of fibroblasts caused elevated HNF-3α levels, but it was unable to stimulate Shh expression in this cell type (Fig. 5B, lanes 11–13). Taken together, these results suggest that HNF-3α mediates the induction of the endogenous mouse HNF-3β, Shh, and TTR genes in F9 cells without retinoic acid stimulation. Moreover, our studies indicate that ectopic expression of HNF-3α stimulated HNF-3β expression followed by TTR, which indicates that the viral system recapitulates the expression pattern observed during retinoic acid-mediated F9 cell differentiation (15,37).

The HNF-1α, HNF-1β, and HNF-4α promoters contain HNF-3 binding sites and are therefore putative targets for HNF-3α in F9 cells during retinoic acid-mediated F9 cell differentiation (10,21,22,46,49). Furthermore, expression of these transcription factors is activated after the appearance of HNF-3α following retinoic acid stimulation. We therefore examined whether expression of HNF-1α, HNF-1β, or HNF-4α could be affected by the HNF-3α transcription factors in F9 cells. F9 cells were infected with AdHNF3α to elevate HNF-3α protein levels and RNase protection assays indicated that HNF-3α alone was insufficient to induce the expression of these genes (data not shown). These studies demonstrate that transcriptional activation of the HNF-1α, HNF-1β, and HNF-4α genes requires other regulatory proteins induced by the retinoic acid differentiation program.

Ectopic HNF-3β Levels in Mouse F9 Cells Are Sufficient to Induce Expression of TTR and Shh

Retinoic acid differentiation of F9 induces HNF-3α first, followed by delayed stimulation of HNF-3β expression (15,37). Because these HNF-3 proteins are closely related, we next examined whether HNF-3β could substitute for HNF-3α function and stimulate TTR and Shh levels in the absence of retinoic acid stimulation. In order to test our hypothesis, we infected F9 cells with AdHNF3β, which provided maximal HNF-3β levels at 2 days PI and gradually declined until the 5-day time point (Fig. 6A, HNF-3β lanes 3–8). Longer exposure of the HNF-3β signal demonstrated detectable HNF-3β levels throughout the time points examined (Fig. 6B). AdHNF3β infection mediated an increased in TTR expression between 4 and 6 days PI, suggesting that high HNF-3β levels were not required to maintain TTR transcription (Fig. 6, lanes 7–9). Moreover, elevated HNF-3β levels provided greater stimulation of Shh expression in F9 cells than had been observed with the HNF-3α protein (compare Fig. 6B with Fig. 5B). This result suggests that the HNF-3 proteins are not equivalent in their ability to activate target genes and that HNF-3β is the preferred transcriptional regulator for the Shh gene. Overall, this result is consistent with our gene cascade model, in which HNF-3α induces HNF-3β and HNF-3β induces expression of the differentiation-specific Ttr and Shh genes.

Although HNF-3α is a primary target gene for retinoic acid action (16), its promoter region contained an HNF-3 recognition site (34). We therefore examined whether elevated HNF-3β levels in F9 cells could stimulate expression of HNF-3α. RNA was prepared from F9 cells at different time points following AdHNF3β infection, and HNF-3α activation was assessed by RNase protection. Despite achieving elevated HNF-3β levels, this transcription factor was not able to stimulate expression of the endogenous mouse Hnf3α gene (Fig. 6, HNF-3α panel). This result indicates that increased transcription of the Hnf3α gene requires additional factors that are stimulated by the retinoic acid differentiation program. Similar to HNF-3α experiments in F9 cells, elevated HNF-3β levels were also insufficient to stimulate expression of HNF-1α, HNF-1β, or HNF-4α (data not shown).