Address correspondence and reprint requests to J. Drouin, Laboratoire de génétique moléculaire, Institut de recherches cliniques de Montréal, 110 des Pins Ouest, Montréal Québec, Canada H2W 1R7. E-mail: firstname.lastname@example.org
Tyrosine hydroxylase (TH) is the rate-limiting enzyme of dopamine and (nor)adrenaline biosynthesis. Regulation of its gene expression is complex and different regulatory mechanisms appear to be operative in various neuronal lineages. Pitx3, a homeodomain-containing transcription factor, has been cloned from neuronal tissues and, in the CNS, mouse Pitx3 is exclusively expressed in midbrain dopaminergic (MesDA) neurons from embryonic day 11 (E11). TH appears in these neurons at E11.5, consistent with a putative role of Pitx3 in TH transcription. We show that Pitx3 activates the TH promoter through direct interaction with a single high-affinity binding site within the promoter and that this site is sufficient for Pitx3 responsiveness. In contrast, we did not observe an effect of Nurr1, an orphan nuclear receptor essential for normal development of MesDA neurons, on TH promoter activity. Pitx3 activation of TH promoter activity appears to be cell-dependent suggesting that Pitx3 action may be modulated by other(s) regulatory mechanism(s) and factor(s).
The complexity of the mammalian brain depends, in part at least, on the diversity of differentiation pathways taken by neuronal cells. Several genes are known to define large domains of the CNS during early development, acting on mitotic precursors. These genes pattern the neuroepithelium and specify the identity of neuronal precursors along dorso-ventral and antero-posterior axes (Lumsden and Krumlauf 1996; Tanabe and Jessell 1996). Final differentiation occurs once precursors have exited the cell-cycle and is characterized by expression of the enzymes or neurotransmitters required for neural functions. Few genes have been implicated in these late events of differentiation of specific neuronal populations (Edlund and Jessell 1999; Goridis and Brunet 1999).
The midbrain dopaminergic (MesDA) neurons have been actively studied because of their unique character, ease of identification and their involvement in the pathogenesis of Parkinson's disease. As other ventral neurons in the CNS, their early development is dependent of Sonic Hedgehog (Shh), which is secreted by notochord and floorplate (Echelard et al. 1993; Roelink et al. 1994; Chiang et al. 1996; Ericson et al. 1997). Their differentiation on the rostral side of the midbrain–hindbrain boundary (MHB) appears to be induced by FGF8, which is produced at the boundary. These two morphogens (Shh and FGF8) were sufficient to induce MesDA neuronal differentiation in explants taken from dorsal midbrain and telencephalon (Ye et al. 1998) and in neuronal stem cells originating from embryonic stem (ES) cells (Lee et al. 2000).
Two genes potentially required for the late differentiation of these neurons have been identified recently. Nurr1, an orphan nuclear receptor expressed widely in the mouse brain is the only member of its family [which includes Nur77 (NGFI-B) and NOR1] to be expressed in MesDA neuron precursors, from embryonic day 10.5 (E10.5) (Zetterstrom et al. 1996). This may explain why inactivation of the Nurr1 gene in mice specifically affected MesDA neurons. Indeed, MesDA neurons degenerate and die by apoptosis around day E18.5 in Nurr1 knock-out mice (Zetterstrom et al. 1997; Wallen et al. 1999). However, Nurr1 is not sufficient by itself to promote dopaminergic differentiation of neuronal stem cells (Wagner et al. 1999).
Pitx3 (Ptx3), an homeodomain-containing transcription factor, is expressed from day E11 in mouse MesDA neuron precursors. This is the only site of Pitx3 expression in the CNS and its expression is maintained throughout adult life in men and mice (Smidt et al. 1997). This expression is unaffected by Nurr1 knock-out, showing the independence of the two pathways (Saucedo-Cardenas et al. 1998). Pitx3 was hypothesized to be a regulator of MesDA tyrosine hydroxylase (TH) expression because of its appearance a half-day before TH during development. The Pitx subfamily of homeoproteins falls within the paired class of homeoproteins and together with the Otx genes, these subfamilies constitute a subgroup of mammalian proteins that share similar DNA binding specificity with Drosophila bicoid (Wilson et al. 1996). The mechanisms of action of these homeoproteins has been defined for their transcriptional role in expression of downstream target genes in the pituitary gland (Lamonerie et al. 1996; Szeto et al. 1996; Poulin et al. 1997, 2000; Tremblay et al. 1998, 1999; Tremblay and Drouin 1999; reviewed in Drouin et al. 1998).
We show here that the murine TH promoter contains several putative binding sites for Pitx3 and that its activity is stimulated by Pitx3 in transient transfection assays. Further, we show that only one of these sites is a high-affinity Pitx3 binding site and that it appears sufficient for Pitx3 activation of transcription. These results demonstrate a direct action of Pitx3 on the mouse TH gene and suggest a role for this homeoprotein during development and/or maintenance of MesDA neurons.
Materials and methods
Plasmids and oligonucleotides
The mouse tyrosine hydroxylase promoter has been kindly provided by Dr Richard Palmitter. Constructs containing subfragments of this promoter in front of the luciferase reporter gene were constructed in the pXP1 vector. Four-kilobase and 1.2 kb TH promoter fragments were subcloned with EcoRV-SalI and MscI-SalI, respectively. The − 762 bp construct was obtained by XhoI internal deletion of the 4 kb construct. The − 452 bp, − 305 bp and 157 bp TH promoter fragments were amplified by PCR, cloned as KpnI–SalI fragments in pXP1 and sequenced before use. Point mutations were made using pALTER (Promega) according to the manufacturer's recommendations. The oligonucleotide sequences used to insert mutations in site III, IV and V are, respectively, TGAACCCTTGTGGACTACTGCATGGGCGCT, TATGCCCTGGTATTAGTCGAGAGCTCTA, and TGCAATTAGATCGACTAGTACGGAGGCTCT (bold characters are the mutated nucleotides). Double mutation was performed in a single round of mutagenesis by using two mutant oligonucleotides at the same time. Mutant promoters were subcloned in pXP1 and have been fully sequenced prior their use. (POMC Pitx site)3-Luc construct was previously described (Lamonerie et al. 1996). Pitx, Otx and Nurr1 expression vectors were previously described (Smidt et al. 1997; Drouin et al. 1998; Maira et al. 1999; Tremblay et al. 1999).
Embryonic carcinoma P19S18 cells (Rossant and McBurney 1982) were cultured in α-MEM (Gibco-BRL) supplemented with 10% fetal bovine serum (FBS) or in α-MEM supplemented with 2.5% FBS and 7.5% donor bovine serum (DBS) during retinoic acid (RA) treatment. Neural differentiation of P19 cells (P19RA) was performed by treating P19 monolayers at 70% confluence with 10−6 M RA for 4 days. The sixth day, cells were harvested and replated for use. This step eliminated the few cells that failed to differentiate. CV-1 and PC12 cells were cultured in Dulbecco's modified Eagle medium (DMEM) (Gibco-BRL) supplemented with 10% FBS. All cultured cells were maintained at 37°C in humidified air containing 5% CO2.
For luciferase assays, cells were plated at 50 000 cells/well in 12-well plates and transfected the day after. P19, P19RA and CV-1 cells were transfected with the calcium phosphate co-precipitation method. One and a half micrograms of reporter plasmid were transfected in each well plus the indicated amount of expression vector(s). Empty expression vector and pBluescript were added to obtain the same total amount of expression vector and a total of 2.5 µg of DNA in each well. Eighteen hours after transfection, media were changed. Cells were harvested 40–44 h after transfection and luciferase activity measured as previously described (Lamonerie et al. 1996). PC12 cells were transfected with LipofectAMINE (Pharmacia) according to the manufacturer's recommendations using 500 ng of reporter and a total of 1.5 µg of DNA in each well of 500 000 cells. Transfected cells were harvested 24 h later for luciferase activity measurement.
For preparation of nuclear extracts, CV-1 cells were plated in 100-mm culture dishes (500 000 cells each) and transfected using the calcium phosphate co-precipitation method with 20 µg of DNA/dish. Media were changed 18 h after transfection and nuclear extracts were prepared 40–44 h post-transfection.
Transfected cells were washed once and harvested in cold phosphate-buffered saline (PBS) containing 6 m m EDTA. Cells were then centrifuged and resuspended in 400 µL of buffer A [10 m m HEPES (pH 7.9), 10 m m KCl, 0.1 m m EGTA, 0.5 m m phenylmethylsulfonyl fluoride (PMSF), 1 m m dithiothreitol (DTT) and 10 µg each of the protease inhibitors leupeptin, aprotinin and pepstatin/mL]. Cells were incubated on ice for 15 min before addition of 25 µL of NP-40 (10%) followed by vigorous vortexing. After centrifugation, the nuclear pellet was resuspended in 50 µL of buffer B [10 m m HEPES (pH 7.9), 0.1 m m EGTA, 0.4 m NaCl, 0.5 m m PMSF, 1 m m DTT and 10 µg each of protease inhibitors as above/mL] and shaken vigorously at 4°C for 1 h. The extracts were centrifuged and the concentration of the supernatant was estimated by the Bradford assay.
Sequences encoding amino acids 9–45 of rat Pitx3 were subcloned in frame with MBP and GST into their respective vectors. The fusion proteins were purified from Escherichia coli BL-21 according to the maltose and sepharose beads manufacturer's recommendations (New England Biolabs and Pharmacia Biotech, respectively). Antibodies were raised by injection of 100 µg of MBP-Pitx39−45 in New Zealand female rabbits; two booster injections were made at 4 and 6 weeks after initial injection. After assessment of immunological response by western blot, the rabbits were sacrificed and serum collected. The antiserum was purified on a GST-Pitx39−45 column to obtain an affinity-purified antibody preparation (Lanctôt et al. 1999).
Electrophoretic mobility shift assay
The following electrophoretic mobility shift assay (EMSA) probes were synthesized:
POMC site: CAGGATGCTAAGCCTCTGTCCA (coding) TGGACAGAGGCTTAGCAT (reverse), site I: GGAGATGATAATCAGAGGAATC (coding) GATTCCTCTGATTATCAT (reverse), site II: TAGAGGAATAATCTTTCTGAAA (coding) TTTCAGAAAGATTATTCC (reverse), site III: CCCTTGGGTAATCCAGCATGGG (coding) CCCATGCTGGATTACCCA (reverse), site IV: GAGCTCTCTAATCAAACCAGGG (reverse) CCCTGGTTTGATTAGAGA (coding), site V: ATTAGATCTAATGGGACGGAGG (coding) CCTCCGTCCCATTAGATC (reverse) and site VI: CTGACGTCAAAGCCCCTCTGGG (coding) and CCCAGAGGGGCTTTGACG (reverse). Probes were labeled by Klenow fill-in in presence of radioactive (32P) dCTP or dATP. 50 000 CPM of equivalent specific activity probes were used in each reaction with 5 µg of nuclear extracts. In supershift experiments, 5 µg of immunopurified immunoglobulins were used per reaction. Binding reactions were performed in 20 µL of binding buffer [25 m m HEPES (pH 7.9), 84 m m KCl, 10% glycerol and 5 m m DTT] and incubated on ice for 1 h. Binding reaction mixes were separated on non-denaturating 5% polyacrylamide gel in Tris-Glycine buffer (40 m m Tris and 195 m m glycine) at 200 V for 3 h, at 4°C. Gels were dried and exposed for autoradiography overnight at room temperature (20–22°C).
The mouse TH promoter contains several Pitx3 putative sites
The TH promoter was searched for putative Pitx3 binding sites. As for other homeoproteins with bicoid DNA binding properties, Pitx3 binds a consensus site similar to TAA(T/G) C(C/T) (Wilson et al. 1996; Smidt et al. 1997; Tremblay et al. 1998). Some mismatches within this consensus site are compatible with high-affinity DNA binding (Tremblay et al. 1998). This analysis revealed four putative sites in the first kb of the mouse TH promoter (Sites I, II, III and IV in Fig. 1a). We also considered another putative site of action (site V) because of previous work showing that this site is a target for another homeoprotein, Arix (Phox2A in mouse) in noradrenergic cells (Zellmer et al. 1995). Two of these sites (I and V) are perfectly conserved between mouse and rat. Two sites (II and III) are perfect consensus in mouse, but the homologous rat sequences lack the A at the third position and the T at the fourth position (site II) or the first T of the site (site III). We finally included in our analysis a VIth site because another group suggested a Pitx3-dependent effect through this site (Cazorla et al. 2000).
Pitx3 activates the TH promoter in P19 cells
In order to determine if Pitx3 activates the TH promoter, we performed transient transfections in P19 cells. These cells were derived from a mouse embryonic carcinoma and they will differentiate into different cell types when treated in vitro with different inducers. Upon RA treatment, P19 cells differentiate into neuronal cells. Most of these neuronal cells have a GABA-ergic phenotype and less than 1% are TH-positive (McBurney 1993). P19 cells do not express Pitx3 in their undifferentiated state nor when treated with RA, as determined by RT-PCR (data not shown). Thus, P19 cells are an appropriate model of undifferentiated cells with neuronal potential.
In P19 cells, Pitx3 activated (up to five-fold activation) transcription from a 4-kb TH promoter placed in front of the luciferase reporter gene and this effect is dose-dependent (Fig. 2a). In contrast with results obtained by others working with different cells (Sakurada et al. 1999; Cazorla et al. 2000), we did not observe any significant effect of Nurr1 alone or in combination with Pitx3 in these cells (Fig. 2b). Since others reported cell-specific activities for Pitx3 as well as for Nurr1, we tested a panel of cell lines for responsiveness of the TH promoter to Pitx3 (Fig. 3a). While Pitx3 activated the TH promoter in undifferentiated P19 cells (Fig. 2 and first two lanes of Fig. 3a), it had a mild repressive effect in P19 cells treated with RA for neuronal differentiation, although Pitx3 can activate a simple Pitx target reporter as well in both cell contexts (Fig. 3b). In addition, Pitx3 had no effect on TH promoter when cotransfected in pheochromacytoma cells (PC12 cells) or fibroblasts (CV-1 cells).
The three members of the Pitx family activated TH promoter similarly (Fig. 4). In contrast, the Otx transcription factors which are related to the Pitx subfamily (Lamonerie et al. 1996) did not significantly activate the TH promoter. These results suggested a unique property of the Pitx transcription factor(s) for TH transcription by contrast to the Otx factors which otherwise share similar DNA binding specificity (Drouin et al. 1998). This specificity is striking in the view of the likely expression of Otx2 in MesDA neurons (Simeone et al. 1993).
A single target site is responsible for Pitx3 action
In order to determine whether one or more Pitx3 binding site(s) (Fig. 1b) are required for Pitx3 activation of the TH promoter, we generated a serie of promoter 5′ deletions (Fig. 5a). Deletion of sequences between − 4 kb and − 762 bp did not affect the response to Pitx3, indicating that the target sequence is within the shorter promoter fragment. Whereas deletion of the upstream sites I and II had no effect, deletion of site III completely abolished the effect of Pitx3, suggesting that sites IV, V and VI are insufficient for Pitx3 responsiveness.
In order to ensure that the loss of Pitx3 responsiveness upon deletion of sequences between − 452 and − 305 bp resulted from deletion of the putative Pitx3 binding site III, but not of other essential regulatory sequences, we introduced point mutations in sites III, V or IV and V (Fig. 5b). In support of the promoter deletion analysis (Fig. 5a), mutagenesis of only site III prevented Pitx3 responsiveness. None of these mutations affected basal promoter activity. Thus, TH site III appears to be necessary for Pitx3 activation of this promoter.
TH site III is a Pitx3 high-affinity binding site
In order to test the relative affinity of Pitx3 for its putative binding sites in the TH promoter, we performed in vitro binding assays with probes derived from the six Pitx3 putative binding site (Fig. 6a). We used nuclear extracts from CV-1 cells transiently transfected with an expression vector containing Pitx3 coding sequences or with the empty vector. CV-1 cells do not express Pitx3 (RT-PCR, data not shown) and do not have endogenous binding activity for all probes, except site I probe (Fig. 6, lanes 4 and 5). As positive control for DNA binding, we used a probe containing the previously characterized (Lamonerie et al. 1996) Pitx binding site of the pro-opiomelanocortin (POMC) promoter (Fig. 6a, lane 3). Of the putative TH sites, only site III bound Pitx3 with similar affinity (lane 9). Sites I, II and VI exhibited faint binding (lanes 5, 7 and 15) and much longer gel exposure was required to detect even fainter binding to sites IV and V (data not shown). We confirmed by supershifting with a Pitx3 antiserum that the band observed with POMC and site III probes did indeed contain Pitx3 (Fig. 6b, lanes 5 and 11). Control IgGs had no effect on Pitx3 complexes (Fig. 6b, lanes 6 and 12), showing the specificity of the supershift.
In order to definitely support the requirement for Pitx3 binding to the TH promoter, we took advantage of the availability of a single amino acid mutant of the Pitx1 homeodomain which prevents DNA binding (Tremblay et al. 1999). Pitx1 activated the TH promoter as well as Pitx3 (Fig. 4 and Fig. 6c) and the Pitx1 K139A mutant no longer affected TH promoter activity (Fig. 6c), clearly showing the importance of DNA binding by these factors for TH promoter activation.
Taken together, these results show that Pitx3 induces mouse TH promoter activity in P19 cells and that it may do so through direct DNA binding of site III.
The present work clearly shows that Pitx3 can contribute to the control of TH promoter activity through direct interaction with the TH promoter. This transcriptional activation depends on the presence (Fig. 5) of a high-affinity promoter binding site (Figs 6a and b) and on Pitx3′s ability to bind DNA through its homeodomain (Fig. 6c). However, this ability of Pitx3 to activate TH promoter activity appears to be cell-dependent, suggesting that this function may be subjected to other(s) regulatory mechanism(s). Modulation of Pitx3 activity may play an important role during development, cell differentiation or maintenance of MesDA neurons.
Pitx3 activates TH promoter directly via a specific site
The Pitx3 binding site that we identified within the TH promoter is entirely comparable to other Pitx binding sites in target genes (Szeto et al. 1996; Amendt et al. 1998; Tremblay et al. 1998) and to binding sites for other paired proteins of bicoïd-related DNA binding specificity (Wilson et al. 1993, 1996). Indeed, all these high-affinity Pitx binding sites contain the consensus sequence TAA(G/T) C(C/T), although some active sites of unknown relative affinity may have a T at the second position (Tremblay et al. 1998). It is noteworthy that all these sites have perfectly conserved residues at 3rd and 5th positions of the consensus. In this context, it is not surprising that site VI of the TH promoter should be of low affinity (Fig. 6a) and that its presence in TH promoter constructs is not sufficient for Pitx3-dependent activation (Fig. 5a). The implication of this site by Cazorla et al. (Cazorla et al. 2000) in the mechanism of Pitx3 action on the TH promoter is surprising given the present findings. In contrast, the low Pitx3 binding affinity of site II was unexpected since this site has a consensus sequence that appears as good as site III. Differences in the immediate environment of this site in the TH promoter context may explain this fact. Finally, similarly to the site VI, site III of the rat TH promoter has a nucleotide difference with the mouse at the first position of the consensus (Fig. 1b): this difference decreased Pitx3 binding significantly (data not shown) and thus it is likely to render this sequence poorly responsive to Pitx3. Regulatory promoter sequences of the TH promoter appear to differ between species (Gandelman et al. 1990) and thus Pitx3 may still act on the TH promoter of other (than mouse) species through a different binding site. Indeed, such sites have been identified in the rat TH promoter (Schimmel et al. 1999). Further experiments will be needed to confirm the activity of these sites.
Ability to activate TH promoter is restricted to Pitx subfamily
In general, the three members of Pitx subfamily were found to have very similar activities (Smidt et al. 1997; Drouin et al. 1998; Tremblay et al. 2000). Consistent with these observations, the three Pitx factors activate the TH promoter and the closely related Otx homeoproteins do not (Fig. 4). Biologically, only Pitx3 can activate this promoter in MesDA neurons for the trivial reason that Pitx 1 and 2 are not expressed at this site. In the normal developing mouse, MesDA neurons appear to only express Pitx3 (Smidt et al. 1997) since Pitx1 is not expressed at all in the CNS (Lanctôt et al. 1997) and the brain expression of Pitx2 appeared to exclude the midbrain region. Pitx2 is nonetheless expressed both anteriorly and posteriorly of the MesDA system in the ventral brain (Kitamura et al. 1997). The only related homeoprotein known to be expressed in the midbrain is Otx2 (Simeone et al. 1993) but it did not significantly activate the TH promoter.
Nurr1 has no effect on mouse TH promoter
The presence of a Nurr1 consensus NBRE binding site (Maira et al. 1999) within the TH promoter suggested that this orphan nuclear receptor could influence the activity of this promoter. In addition, physical and transcriptional interactions have been already characterized between Pitx1 and an orphan nuclear receptor, SF-1 (Tremblay et al. 1998, 1999; Tremblay and Drouin 1999), suggesting that similar interaction could take place between Pitx3 and Nurr1 in the context of the TH promoter. However, we did not observe any significant effect of Nurr1 on TH promoter activity nor on the activation produced by Pitx3 (Fig. 2b). This is in apparent contradiction with results from other groups who reported activation of TH promoter by Nurr1 (Sakurada et al. 1999; Cazorla et al. 2000). These groups utilized the rat TH promoter in their studies, whereas the present work used the mouse promoter. In addition, the action of Nurr1 may depend on cellular context, although we found Nurr1 to potently activate a NurRE-Luc reporter in P19 cells (data not shown).
Pitx3 activates TH promoter specifically in undifferentiated P19 cells
Use of a specific cell line seems to be of particular importance in regard to Pitx3 activation of TH promoter activity. This can reflect the lack for a Pitx3 coactivator in other cell lines and in differentiated P19 cells. Pitx factors are known to interact with several transcription factors from different structural classes (Szeto et al. 1996; Tremblay and Drouin 1999; Tremblay et al. 1999; Poulin et al. 2000) and P19 cells may express a unique factor required for TH promoter activity. It is noteworthy that P19 cells express several genes that are not frequently active in other cell lines, such as Egr1, Gsc, Gbx2 and Wnt1 (Bouillet et al. 1995; Edwards et al. 1991; Papkoff 1994; Sawada et al. 2000). The pluripotent character of P19 cells may be important for Pitx3 activation of TH promoter activity. Similarly, differentiation of MesDA-like neurons in vitro was more efficient using ES cells (Lee et al. 2000) than with neuronal stem cells (Wagner et al. 1999), the latter having more restricted differentiation capacities compared to ES cells.
Role of Pitx3 in activation of TH promoter and MesDA neuron development or maintenance
Although the present data suggest that Pitx3 activates TH promoter activity, it is clear that TH can be expressed in the absence of this transcription factor since TH is expressed outside the midbrain where Pitx3 is not present. There are many examples of genes expressed in different tissues under control of different promoter elements and different transcription factors. Therefore, the true role of Pitx3 in MesDA transcription of TH may have to await gene inactivation (complete or conditional) studies of this gene. In the meanwhile, one mouse model with decreased midbrain Pitx3 expression has been reported. Indeed, lmx1b-deficient mice (lmx1b−/–) have few remaining Pitx3-positive cells in the midbrain, yet they still exhibit transient TH expression (Smidt et al. 2000). Although these TH-positive cells are not localized where normal MesDA neurons normally are, these results may be taken to suggest that Pitx3 is not essential for MesDA TH expression. Alternatively, the transient TH-positive neurons observed in lmx1b−/– mice may not truly be MesDA neurons since midbrain development appeared to be severely perturbed in this model. In summary, although different studies have documented the importance of Nurr1 (Zetterstrom et al. 1997; Saucedo-Cardenas et al. 1998), lmx1b and probably Pitx3 (Smidt et al. 2000) for MesDA development and/or maintenance, regulation of TH transcription in MesDA neurons (and in particular the precise role of Pitx3) remains difficult to investigate in the absence of an appropriate cellular model system.
Pitx3 could act in the same way as Phox2 in noradrenergic system
We are very thankful to Dr Richard Palmitter for providing the mouse TH promoter and to Dr Orla Conneely for the Nurr1 cDNA. We also thank Lise Laroche for expert secretarial assistance. This work was funded by the Parkinson Foundation of Canada and by a Glaxo Wellcome/Medical Research Council of Canada/PMAC research chair. ML is the recipient of a master's research scholarship from FCAR-FRSQ.