Mediation of Sonic Hedgehog-Induced Expression of COUP-TFII by a Protein Phosphatase

See allHide authors and affiliations

Science  12 Dec 1997:
Vol. 278, Issue 5345, pp. 1947-1950
DOI: 10.1126/science.278.5345.1947


A Sonic hedgehog (Shh) response element was identified in the chicken ovalbumin upstream promoter–transcription factor II (COUP-TFII) promoter that binds to a factor distinct fromGli, a gene known to mediate Shh signaling. Although this binding activity is specifically stimulated by Shh-N (amino-terminal signaling domain), it can also be unmasked with protein phosphatase treatment in the mouse cell line P19, and induction by Shh-N can be blocked by phosphatase inhibitors. Thus, Shh-N signaling may result in dephosphorylation of a target factor that is required for activation of COUP-TFII–, Islet1-, and Gli response element–dependent gene expression. This finding identifies another step in the Shh-N signaling pathway.

COUP-TFs belong to the orphan receptor subfamily within the steroid–thyroid hormone receptor superfamily and are found in all vertebrate species examined (1). In the mouse there are two COUP-TF members, COUP-TFI and COUP-TFII. Both are expressed in the neural tube during embryonic development; however, COUP-TFII is highly expressed and displays a restricted expression pattern that is coincident with motor neuron differentiation (1). Transplantation of a notochord to the dorsal side of the chick neural tube results in ectopic expression of COUP-TFII that coincides with the appearance of motor neuron markers such as Isl1 and SC-1 in these regions (2).

Sonic hedgehog (Shh) is a vertebrate homolog of theDrosophila segment polarity gene hedgehog (Hh) (3). The secreted Shh protein (Shh-N) from the notochord is required for induction of floor plate cells, motor neurons, and other axial midline structures (4-7). To investigate whether Shh activates COUP-TFII expression, we asked whether purified recombinant Shh-N expressed in Escherichia coli can induce COUP-TFII expression in P19 cells (8). Increased COUP-TFII expression is observed at Shh-N concentrations as low as 0.2 nM (Fig.1A, lane 2). This concentration is similar to the amount that is required for regulating other Shh-N target genes (3, 9) and for binding to its putative receptor, patched (ptc) (10, 11).

Figure 1

Shh-induced COUP-TFII mRNA in P19 cells. (A) P19 cells were treated for 24 hours with 0.2 (lane 2), 0.5 (lane 3), and 1 (lane 4) nM Shh-N; total RNA was obtained for a ribonuclease protection assay as described (13). Murine cyclophilin A was used as a control. (B) P19 cells were treated with or without Shh-N for 24 hours and nuclear extracts were used in an EMSA (20). Different amounts of unlabeled GliRE or ShhRE (lanes 3 and 7, 100-fold excess; lanes 4 and 8, 50-fold excess; lanes 5 and 9, 25-fold excess; lane 6, 10-fold excess) were used as a competitor.

To identify the target element or elements for Shh-N signaling, we used deletion analysis and identified a sequence between −1316 and −1298 nucleotides in the COUP-TFII promoter that increases reporter gene activity when linked to a heterologous herpes simplex virus thymidine kinase (tk) promoter (12) (Fig.2). Point mutations introduced within this element completely abolished the activation by Shh-N. Therefore, we defined this region as a Shh response element (ShhRE). Next, we examined whether this element could bind any factor or factors in response to Shh-N treatment. Nuclear extracts were prepared (13) from untreated P19 cells or from P19 cells treated with Shh-N and used in an electrophoretic mobility-shift assay (EMSA) (14). Shh-N treatment resulted in the enrichment of two specific complexes (Fig. 1B, lanes 1 and 2), which were blocked by excess ShhRE (lanes 7 to 9). Hence, Shh-N enhances the binding of a factor or factors to the ShhRE in the COUP-TFII promoter. Recent reports have identified a response element on Hh-induced target promoters that are bound by a member of the Ci/Gli family of transcription factors (15). The ShhRE in the COUP-TFII promoter does not resemble a Gli response element (GliRE). As expected, the GliRE is unable to compete for binding to the ShhRE in the COUP-TFII promoter (Fig. 1B, lanes 3 through 6). In contrast, in vitro translated human Gli1 can bind to the radiolabeled GliRE (16). Therefore, Shh signaling may not be restricted to the Ci/Gli family of transcription factors. The ShhRE includes an AT-rich motif followed by a GC core, and both are important for Shh-mediated activity. Because the binding site in the ShhRE has a TAAT motif, we predict that this element could bind to transcription factors containing a homeobox domain. Unlabeled oligonucleotide containing the Sox-9 recognition sequence is able to compete for binding to the ShhRE (16). Sox-9 belongs to a family of SRY-box–containing transcription factors that function in determining important cell fates during differentiation (17).

Figure 2

ShhRE-mediated activity is a direct result of Shh-N signaling. P19 cells were cotransfected with either three copies of the ShhRE (lanes 1 to 4) or mutant ShhRE (lanes 5 to 8) tk-CAT along with tk-LUC plasmid as an internal control. Total RNA was obtained from cells pretreated with 50 μM Chx for 2 hours before addition of 1.0 nM Shh-N and harvested 12 hours later. Sixty micrograms of total RNA from these cells was used in a ribonuclease protection assay.

To address the direct or indirect nature of this Shh-mediated target response at the ShhRE, the Shh-N–induced activity of the ShhRE-linked reporter was determined in the presence or absence of cycloheximide (Chx). The Shh-induced increase in reporter mRNA synthesis was not altered by the presence of Chx (Fig. 2, compare lanes 2 and 4). In contrast, reporter enzyme activities were both inhibited (16). Thus, the factor(s) that binds to and activates the ShhRE is a primary target of the Shh-N signal pathway, and its activation does not require protein synthesis.

Adenosine 3′,5′-monophosphate–activated protein kinase A antagonizes Hh signaling in Drosophila (1) and in vertebrates (18-21). Also, the product of a Drosophila gene, fused (fu), a serine-threonine kinase, is a downstream target of Hh and is required for Hh signaling (20,22). These results suggest that phosphorylation may be required for the Hh signaling pathway. Therefore, we incubated a nonspecific calf intestinal phosphatase (CIP) with both induced and uninduced nuclear extracts (Fig. 3A, lanes 5 to 8). Although incubation of CIP with Shh-N–treated extracts did not alter binding to radiolabeled ShhRE (compare lanes 7 and 8), CIP treatment of uninduced extracts was able to mimic the binding activity induced by Shh-N (compare lanes 1 and 5). Thus, the factor or factors that bind to this ShhRE are present in the nuclear extracts of untreated P19 cells, and Shh-N treatment specifically alters its ability to bind DNA by dephosphorylation.

Figure 3

Role of phosphatase in Shh-N–mediated activation of COUP-TFII. (A) Nuclear extracts from P19 cells treated with 1 nM Shh-N in the presence or absence of CIP or heat-denatured CIP (ΔCIP) (20 units for 15 min at room temperature) were used in an EMSA. Competitor oligonucleotides (lane 3, wild type; lane 4, mutant) were used in 100-fold molar excess. (B) Total RNA isolated from P-19 cells pretreated with (lanes 2, 4, 6, and 8) or without (1, 3, 5, 7, and 9) 1.0 nM Shh before addition of 5 nM OA (lanes 3 and 4), 0.5 nM CyA (lanes 5 and 6), or 5 nM NoA (lane 9) for another 12 hours. Twenty micrograms of total RNA was used in a ribonuclease protection assay (13). (C) Nuclear extracts from P19 cells treated with or without 1 nM Shh for 36 hours were used in a protein phosphatase assay system (Life Technologies, Bethesda) to quantitate the phosphatase activity with radiolabeled phosphorylase used as a substrate. In the case of inhibitors, 2 nM OA or 4 nM I-2 protein was incubated with the cells for 15 min at 30°C. (D) P19 cells were cotransfected with the ShhRE-SV40LUC and an expression plasmid for PP1, PP2A, PPIV, and PPV. In addition, an empty pBKRSV (RSV) or pCMV5 (CMV) vector was also cotransfected in the control wells. Cells treated with or without 1 nM Shh were harvested after 36 hours and assayed for LUC activity as described.

To characterize the protein phosphatases that might mediate this activity, we used the phosphatase inhibitors okadaic acid (OA) and calyculin A (CyA) (23). OA is a potent inhibitor of serine-threonine–specific protein phosphatases such as PP1, PP2A, PPIV, and PPV, but it has no effect on PP2B and PP2C. OA or CyA, when treated along with Shh-N (Fig. 3B, lanes 4 and 6), resulted in complete inhibition of the induced COUP-TFII mRNA (8). In contrast, the inactive congener norokadone (NoA) was unable to inhibit this induction (Fig. 3B, lane 8). Similarly, OA and CyA treatment abolished the Shh-induced binding to radiolabeled ShhRE (16). Finally, Shh-N–mediated induction of ShhRE reporter gene activity was also shown to be completely blocked by treatment with OA (16). Thus, OA and CyA specifically inhibit binding of a transcription factor to the ShhRE that results in lowered promoter activity and leads to decreased COUP-TFII transcripts. To assay the direct involvement of the type of protein phosphatase in the Shh signaling pathway, we used a phosphatase assay system that quantitates the PP1 and PP2A enzyme activity in control and in Shh-N–treated nuclear extracts in the absence or presence of OA. There is a net increase (twofold) in phosphatase activity after Shh treatment that was completely abolished by OA treatment (Fig. 3C). In contrast, cotreatment with Inhibitor-2 (I-2), a specific inhibitor of PP1, does not decrease the Shh-induced phosphatase activity. Thus, it is likely that a serine-threonine phosphatase such as PP2A, PPIV, or PPV is involved in the Shh-N signaling pathway.

We tested the role of specific protein phosphatases in this pathway by transfecting P19 cells with the ShhRE-luciferase (LUC) reporter plasmid along with expression plasmids that overexpress the catalytic subunits of PP1, PP2A, PPIV, or PPV. The catalytic subunit of PP2A, when overexpressed, can mimic the activity of Shh in untreated cells [compare lanes cytomegalovirus (CMV) and PP2A in Fig. 3D] (24). In contrast, neither PP1, PPIV, nor PPV could mimic this activity. Thus, a PP2A-like phosphatase can mediate this Shh-induced increase in target factor activity in vivo.

To determine whether phosphatase-mediated Shh signaling extends to genes other than COUP-TFII, we analyzed Shh activation of Isl1 in the presence of OA (25). P19 cells treated with Shh-N induced Isl1 mRNA (Fig.4A, lanes 1 and 2), and this induction was completely blocked by treatment with OA (Fig. 4A, lanes 3 and 4). Furthermore, 9.5 days postcoitum (dpc) mouse neural tube explants incubated with OA exhibited decreased amounts of immunodetectable Isl1 protein (Fig. 4B) (26). However, this concentration of OA had no effect on the neurofilament (NFL) staining within the neural tube. Finally, the involvement of protein phosphatases in GliRE-mediated activation by Shh was tested with the GliRE reporter from the hepatocyte nuclear factor 3β 3′ enhancer (27). As expected, in P19 cells, the GliRE-chloramphenicol acetyltransferase (CAT) reporter was enhanced in the presence of Shh expression plasmid (Fig. 4C). This enhanced activity was also sensitive to OA treatment. These results suggest a general role for protein phosphatase in the Shh signaling pathway.

Figure 4

Effect of phosphatase inhibitors on Isl1 expression and GliRE-mediated activity. (A) RNAs were isolated for reverse transcriptase (RT)–PCR analysis from P19 cells as described (Fig. 1A). A complementary RNA internal control (Int. std.) was used to standardize for RT activity. (B) Mouse embryos (8.5 to 9.0 dpc) were isolated and the neural plates were cultured in Matrigel in the absence (untreated) or presence (treated) of 5 nM OA as described (26). Whole-mount immunostaining (29) with Islet1 or NFL antibodies was used to analyze changes in Islet1 and NFL expression. (C) P19 cells were cotransfected with pBKCMV-Shh expression plasmid and a GliRE-CAT reporter plasmid (1 μg) as described (30). Cells cotransfected with pBKCMV-Shh (S) or pBKCMV (C) plasmid (200 ng) were treated with or without 5 nM OA for 14 hours.

Our results indicate that the Shh-mediated mechanism of activation of the COUP-TFII gene may involve dephosphorylation of a factor that leads to increased binding to the ShhRE in the COUP-TFII promoter. This increase in binding resulted in enhanced COUP-TFII promoter activity. Also, the phosphatase that mediates this dephosphorylation in response to Shh-N treatment is PP2A or is like PP2A. This particular response is channeled through a protein with DNA binding activity apparently unrelated to that of the Ci/Gli family of transcription factors previously implicated in Hh signaling. A similar protein phosphatase activity is also required in the Ci/Gli-mediated branch of the Drosophila Hh signaling pathway (28). Thus, activation of specific protein phosphatase activity appears to be a general feature of Hh signal transduction.

  • * These authors contributed equally to this report.

  • To whom correspondence should be addressed. E-mail: mtsai{at}


Stay Connected to Science

Navigate This Article