Induction of Cell Cycle Arrest and Morphological Differentiation by Nurr1 and Retinoids in Dopamine MN9D Cells*

Dopamine cells are generated in the ventral midbrain during embryonic development. The progressive degeneration of these cells in patients with Parkinson's disease, and the potential therapeutic benefit by transplantation of in vitrogenerated dopamine cells, has triggered intense interest in understanding the process whereby these cells develop. Nurr1 is an orphan nuclear receptor essential for the development of midbrain dopaminergic neurons. However, the mechanism by which Nurr1 promotes dopamine cell differentiation has remained unknown. In this study we have used a dopamine-synthesizing cell line (MN9D) with immature characteristics to analyze the function of Nurr1 in dopamine cell development. The results demonstrate that Nurr1 can induce cell cycle arrest and a highly differentiated cell morphology in these cells. These two functions were both mediated through a DNA binding-dependent mechanism that did not require Nurr1 interaction with the heterodimerization partner retinoid X receptor. However, retinoids can promote the differentiation of MN9D cells independently of Nurr1. Importantly, the closely related orphan receptors NGFI-B and Nor1 were also able to induce cell cycle arrest and differentiation. Thus, the growth inhibitory activities of the NGFI-B/Nurr1/Nor1 orphan receptors, along with their widespread expression patterns both during development and in the adult, suggest a more general role in control of cell proliferation in the developing embryo and in adult tissues.

Nuclear receptors constitute a large family of ligand-regulated transcription factors including receptors for steroid hormones, retinoids, vitamin D, and thyroid hormone (1)(2)(3). In addition, a large number of so-called "orphan receptors" belong to the same gene family but lack identified ligands. Nuclear receptors play fundamental roles in adult physiology and during development where they are essential for cell fate specification and differentiation of various cell types both within and outside of the nervous system (3). Nurr1 (NR4A2) is an orphan nuclear receptor expressed in the developing and adult central nervous system (CNS) 1 (4,5). Nurr1 is highly related to two other orphan receptors, NGFI-B and Nor1 (NR4A1 and NR4A3) (6 -8). All three receptors are predominantly expressed in the CNS but are encoded by immediate early genes and can be strongly induced in both neural and non-neural tissues (see e.g. Refs. 9 -15). During development, Nurr1 is expressed at high levels in the ventral mesencephalon, a region where dopamine (DA) cells are being generated (5). These cells are critically involved in control of motor functions, reward mechanisms, and other neural processes. Importantly, the progressive degeneration of these cells is the underlying cause of Parkinson's disease. Analysis of Nurr1 null mutant mice has revealed that this gene is essential for the development of midbrain DA cells (16 -18). Neuronal induction occurs even in the absence of Nurr1, which is normally expressed as developing DA cells become postmitotic (19). However, newly generated, differentiating DA neurons fail to migrate to their normal lateral midbrain positions and are unable to extend axons towards their normal target areas (19). The mechanism whereby Nurr1 promotes differentiation has remained unknown, and the only Nurr1 target gene identified to date in developing DA cells is tyrosine hydroxylase, the ratelimiting enzyme in DA synthesis (20,21). Similar to other nuclear receptors, Nurr1 recognizes DNA by binding hormone-response elements (HREs) in the promoters of regulated target genes. In case of Nurr1, several such binding sites have been identified. The DNA sequence determines the type of protein complex interacting with the HRE. Accordingly, Nurr1 can bind as a monomer and activate transcription from a particular HRE referred to as NGFI-B-response element (NBRE) (22,23). Similar to NGFI-B (24,25), Nurr1 transactivation through the NBRE is mainly dependent on an activation function localized in the large amino-terminal domain of the receptor. In addition, in certain cell lines, Nurr1 activity can also be mediated in part by an activation function within its putative carboxyl-terminal ligand binding domain (LBD) (26). However, it remains unknown if endogenous ligands exist or if Nurr1 exerts its functions in the absence of ligands. Nurr1 can also form heterodimers with the retinoid X receptor (RXR) and bind to certain retinoic acid-responsive elements (27,28). Fi-* This work was supported in part by a grant from the Göran Gustafsson Foundation. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
§ Supported by a fellowship from the Gulbenkian Ph.D. Program in Biology and Medicine and Programa Praxis XXI.
¶ Supported by a fellowship from the National Network in Neuroscience.
ʈ Contributed equally to this work. § § To whom correspondence should be addressed. nally, a third type of binding site identified in the promoter of the pro-opiomelanocortin gene has been shown to be bound by Nurr1 homodimers (29,30).
Retinoids (vitamin A derivatives) are potent inducers of cell differentiation in a variety of tissues in vivo as well as in many different cell lines cultured in vitro. Two types of retinoid receptors, retinoic acid receptor (RAR) and RXR, mediate the effects of retinoids. RAR binds all-trans-and 9-cis-RA, whereas RXRs bind only 9-cis-RA (31). RXR may also utilize other naturally occurring ligands such as the recently identified brain-enriched RXR ligand docosahexaenoic acid (32). RXR forms heterodimers with RAR and with a number of other nuclear receptors including Nurr1 and NGFI-B. Whereas RXR is an inactive partner in complex with several other nuclear receptors (33)(34)(35)(36)(37), RXR can be very efficiently activated by its ligand in complex with Nurr1 or NGFI-B, suggesting that one function of these orphan receptors might be to promote ligandinduced signaling by RXR in vivo (27,28).
In vivo, retinoids are generated by oxidation of retinol into retinaldehyde and retinoic acid in two enzymatic steps requiring specific alcohol and aldehyde dehydrogenases, respectively. Interestingly, one of these enzymes, retinal dehydrogenase 1 (RALDH1) (reviewed in Ref. 38), is highly expressed in developing DA cells during embryogenesis (19,39). RALDH1 expression appears even before Nurr1 in proliferating dopaminergic progenitor cells and remains highly expressed in maturing DA cells and in the adult brain. Thus, it is likely that retinoids are exerting important functions in developing and/or mature DA cells.
To begin to elucidate the mechanisms whereby Nurr1 promotes cellular differentiation, it is desirable to identify suitable in vitro systems where Nurr1 would be actively engaged in the differentiation process. MN9D is a hybridoma cell line established by fusing embryonic primary cells from mouse ventral midbrain with cells from the mouse neuroblastoma cell line N18TG2 (40). MN9D cells show properties of immature neuronal cells that can mature and extend neurites under certain culturing conditions. In addition, these cells also have characteristic dopaminergic properties, e.g. they express high levels of tyrosine hydroxylase, synthesize and store DA, and are sensitive to N-methyl-4-phenylpyridinium ion, the active metabolite of the dopamine neuron-specific toxin N-methyl-4-phenyl-1,2,3,6-tetrahydropyridine (40,41). However, unlike more mature dopamine cells, MN9D cells do not express Nurr1. 2 In this paper we have addressed the functional consequences of expressing Nurr1 in transfected MN9D cells, and we have also analyzed the phenotype of these cells after exposure to retinoids. Our results demonstrate that both Nurr1 and retinoids promote cell cycle arrest at the G 1 phase and induce a mature and highly differentiated phenotype characterized by extension of long neurites. By expressing the related receptors Nor1 and NGFI-B, as well as mutated Nurr1 derivatives in MN9D cells, we have obtained additional insights into the mechanisms whereby Nurr1 and its close relatives are functioning in vivo.
Cell Culture and Differentiation Assay-MN9D, N18TG2, and 293 cell lines were maintained at 37°C and 5% CO 2 in Dulbecco's modified Eagle's medium/F-12 (Life Technologies, Inc.) (MN9D) or Dulbecco's modified Eagle's medium (Life Technologies, Inc.) (N18TG2 and 293), supplemented with 10% fetal calf serum (Life Technologies, Inc.), 100 units/ml penicillin, and 100 g/ml streptomycin (Life Technologies, Inc.). MN9D and N18TG2 cells were grown on poly-D-lysine-coated flasks and plated, 1 day before transfection, at a density of 25 ϫ 10 4 cells in 25-cm 2 flasks. Expression vectors for receptor, EGFP (CLON-TECH), and alkaline phosphatase were transfected at a 8:1:1 ratio, respectively, 4 g of total DNA per flask. Transfection was performed with LipofectAMINE (Life Technologies, Inc.) according to the manufacturer's protocol. After 8 h of incubation with DNA precipitate, complete medium containing 2% fetal calf serum was added. Transfection efficiencies were assessed 1 day later by assaying culture medium for alkaline phosphatase activity, and only samples with similar efficiency of transfection were compared. Based on EGFP expression, ϳ20 -40% of cells were transfected under these conditions. EGFP-expressing differentiated cells were blind-counted 3 days after transfection using an inverted fluorescence microscope (Eclipse TE300, Nikon, Japan). Cells were scored as differentiated when bearing at least one neurite greater DNA Binding Assays-Proteins were made by coupled in vitro transcription and translation in rabbit reticulocyte lysates (TNT, Promega). For gel mobility retardation assays (27), the indicated species of proteins were incubated with binding buffer. The buffer contained 10 mM Tris (pH 8.0), 40 mM KCl, 0.05% Nonidet P-40, 6% glycerol, 1 mM dithiothreitol, 0.2 mg of poly(dI-dC) (Amersham Pharmacia Biotech). Probe (0.2-0.5 ng), 32 P-labeled by a fill-in reaction with the Klenow fragment, with a specific activity of about 3-5 ϫ 10 8 cpm/mg, was added to the reaction and incubated on ice for 20 min. The mixtures were then loaded onto 4% non-denaturing polyacrylamide gels in 0.5ϫ TBE running buffer (0.045 M Tris borate, 0.002 M EDTA). After electrophoresis, gels were dried for autoradiography. The following oligonucleotide and its complement was 32 P-labeled and used as probe: ␤RE, agcttaa-ggGGTTCACCGAAAGTTCActcgcat.
Immunocytochemistry-Cells were plated and transfected as for the differentiation assay and fixed 3 days later with 4% paraformaldehyde in phosphate-buffered saline (PBS), for 30 min at RT. Blocking was performed for 1 h at RT with 3% bovine serum albumin in PBS, 0.1% Triton X-100. Cells were incubated overnight at 4°C with monoclonal anti-class III ␤-tubulin (TUJ1) (Babco), diluted at 1:250 in PBS, 0.1% Triton X-100. After washing with PBS, cells were incubated for 1 h at RT with Cy3-conjugated goat anti-mouse antibody (1:200, Jackson ImmunoResearch).
BrdUrd Incorporation Assay-MN9D cells were transfected as for the differentiation assay and 24 h later were given a 45-min pulse of 50 M BrdUrd (Sigma). After fixation with 4% paraformaldehyde in PBS for 30 min at RT, cells were rinsed in PBS, incubated for 20 min in 2 M HCl, rinsed in PBS, and blocked at RT for 30 min with 10% goat serum, 0.5% bovine serum albumin, and 0.5% Tween 20 in PBS. Incubation with monoclonal mouse anti-BrdUrd antibody (DAKO, Australia) (diluted 1:100 in blocking solution) was done overnight at 4°C. The following day, cells were incubated with Cy3-conjugated goat anti-mouse antibody (1:200, Jackson ImmunoResearch) for 1 h at RT. Five random fields from each sample were analyzed with a fluorescence microscope (Eclipse E1000M, Nikon, Japan). Percentage of EGFP-expressing cells incorporating BrdUrd was calculated as an average of four samples. Error bars represent S.D.
Cell Cycle Analysis-Cells were plated at a density of 4 ϫ 10 5 on poly-D-lysine-coated 25-cm 2 flasks and transfected using Lipo-fectAMINE as described above, with expression vectors for receptor and EGFP at a 50:1 ratio, with a total amount of 4 g of DNA per flask. The distribution of the cells in G 1 , S, and G 2 /M cell cycle phases was determined by DNA flow cytometry. 24 h post-transfection, cells were harvested, washed in PBS, fixed in 1% paraformaldehyde, 0.1% sodium azide for 1 h at 4°C, resuspended in 70% ice-cold ethanol, and stored at Ϫ20°C. Cells were stained with propidium iodide (50 g/ml) in presence of RNase A (50 mg/ml). Flow cytometric analysis was carried out on 10,000 gated EGFP-expressing cells using a FACSCalibur flow cytometer equipped with CellQuest software (Becton Dickinson).
Reporter Gene Assays-Transfections were performed in 24-well plates with LipofectAMINE, according to manufacturer's protocol (MN9D cells) or with calcium-phosphate, as described previously (26) (293 cells). Each well was transfected with 100 ng of reporter plasmid, 100 ng of receptor expression vector, and 200 ng of pCMX-␤gal reference plasmid containing a bacterial ␤-galactosidase gene. Additions to each well were adjusted to contain constant amounts of DNA and of pCMX expression vector. Cells were harvested 24 h after transfection and lysed, and extracts were assayed for luciferase and ␤-galactosidase activity in a microplate luminometer/photometer reader (Lucy-1; Anthos, Austria). Values shown are averages of quadruplicates, with error bars representing S.D.

Nurr1 Expression Induces Morphological Differentiation in MN9D Cells-Within 3 days after transient transfection of a
Nurr1 expression vector, a mature neuronal morphology, characterized by long, commonly bipolar neurites, was induced (Fig.  1A). This morphology could be clearly visualized by staining for a neuronal specific marker such as ␤-tubulin III or by cotransfection of enhanced green fluorescent protein (EGFP) expression vector (Fig. 1, B and C). Transient expression of several other nuclear receptors including the vitamin D-, thyroid hormone-, and peroxisome proliferator-activated receptor ␥ did not induce differentiation of MN9D cells (data not shown). In addition, Ptx3 and Lmx1b, two homeobox transcription factors also expressed in developing midbrain DA cells and implicated in their differentiation, did not promote differentiation in similar transfection experiments (data not shown) (45,46). Nurr1 Structural Requirements in MN9D Cell Differentiation-Because Nurr1 can form RXR heterodimers that are efficiently activated by ligands binding to RXR, the involvement of Nurr1-RXR heterodimers in the process of MN9D cell differentiation was analyzed. Culturing MN9D cells in the presence of the synthetic RXR ligand SR11237 (47) did not promote increased differentiation in either mock-or Nurr1-expressing MN9D cells (Fig. 2A). Although these results indicate that RXR activation is not essential for Nurr1-induced cell maturation, it remained possible that Nurr1 required heterodimerization with RXR for promoting differentiation of MN9D cells. To address this possibility, a dimerization-deficient Nurr1 mutant (Nurr1 dim ) was generated by introducing a 3-amino acid substitution in a region shown to be critical for dimerization in other heterodimers (see "Experimental Procedures"). Nurr1 dim did not heterodimerize with RXR either in a mammalian twohybrid assay in transfected cells or in vitro as demonstrated by a gel-shift experiment (data not shown, Fig. 2B). In contrast, Nurr1 dim was transcriptionally active as a monomer and activated a reporter gene containing three NBREs in transfected cells (Fig. 2C). Importantly, Nurr1 dim was fully active when tested in transfected MN9D cells for its ability to induce morphological differentiation (Fig. 2D). Overall, the results demonstrate that heterodimerization with RXR is not required in the differentiation process. Accordingly, a Nurr1 derivative lacking the entire LBD (Nurr1-(1-356)), which also encompasses the RXR dimerization interface, was also actively promoting differentiation, although to a somewhat lesser extent (Fig. 2E). In contrast, a mutation in the so-called A box (Nurr1 R334A ) (48) abolished DNA binding (data not shown) and was inactive in promoting MN9D cell differentiation (Fig. 2F). Thus, DNA binding by Nurr1 is required, whereas the integrity of the LBD is not essential in the differentiation process.
Retinoids Induce Morphological Differentiation in MN9D Cells-The previously reported (19,39) expression of RALDH1 in developing DA cells prompted us to test if retinoids could promote the differentiation of MN9D cells. After 3 days in the presence of all-trans-RA, cells had adopted a distinct morphology typical of highly differentiated neurons and indistinguishable from the appearance of Nurr1-overexpressing MN9D cells (Fig. 3A). Selective RAR and RXR ligands were tested either alone or in combination to assess their influence on the differentiation process. The RAR-selective agonist TTNPB (49) induced cell differentiation, whereas the RXR-specific agonist SR11237 did not promote maturation when added alone (Fig. 3,  C and D). However, when added together, the two ligands synergized to promote a mature phenotype suggesting that efficient differentiation is promoted by activating both subunits of RAR-RXR heterodimers in these cells (Fig. 3E). A constitutively active RAR derivative (RAR VP16 ), generated by fusing the transactivation domain from the herpes simplex virus protein VP16 to RAR (44), functions as a retinoid mimic in reporter gene assays (data not shown) and was also able to induce maturation in MN9D-transfected cells (Fig. 3F).
Nurr1 Acts Downstream of Retinoids in Promoting Morphological Differentiation-We wished to investigate if Nurr1 and retinoids promoted differentiation via similar or distinct path- Mock-transfected cells retained the usual round shape, bearing occasionally very short neurites, whereas Nurr1-expressing cells adopted a mature morphology, characterized by long, usually bipolar neurites. An indistinguishable morphology was evident after immunostaining for the neuronal specific marker ␤-tubulin III (B) or by visualizing EGFP expression by fluorescence microscopy in cells cotransfected with Nurr1 and EGFP expression vectors (C).
ways. Dominant negative derivatives of Nurr1 and RAR were used in these experiments. A Nurr1 dominant negative derivative was generated (Nurr1 EngR ) by replacing the Nurr1 LBD with the Drosophila Engrailed repressor domain (50,51). In a reporter gene assay Nurr1 EngR efficiently inhibited Nurr1induced activation of an NBRE reporter, even when coexpressed with Nurr1 at a 1:5 ratio (Fig. 4A). In contrast, a similar derivative containing a mutation in the A box, which disrupted DNA binding of Nurr1 EngR (Nurr1 EngR/R334A ), abolished the dominant negative activity (Fig. 4A). Importantly, Nurr1 EngR , but not Nurr1 EngR/R334A , was an efficient inhibitor of Nurr1-induced differentiation of MN9D cells (Fig. 4B, data not shown). Similarly, the previously characterized dominant negative derivative of RAR (RAR-(1-403)) (52) effectively blocked the differentiation induced by the retinoid mimic RAR VP16 (Fig. 4B). Interestingly, Nurr1 EngR blocked differentiation induced by RAR VP16 , whereas RAR-(1-403) was entirely unable to inhibit Nurr1-induced maturation (Fig. 4B). Retinoids do not induce the expression of either Nurr1, NGFI-B, or Nor1 as determined by reporter gene analyses (Fig.  4C) and reverse transcriptase-polymerase chain reaction (data not shown). In conclusion, RAR VP16 and Nurr1 are active at unique steps in a common differentiation pathway where Nurr1 is influencing a more downstream maturation event. However, the mechanism whereby retinoids influence maturation does not involve induction of Nurr1, NGFI-B, or Nor1. G 1 Arrest Induced by Nurr1 in MN9D Cells-Mechanisms that regulate cell differentiation are often intimately linked to the control of cell proliferation. Thus, proliferation of MN9D cells cotransfected with expression vectors for Nurr1 and EGFP was analyzed. Indeed, Nurr1 inhibited cell proliferation as the number of cells incorporating BrdUrd 24 h after transfection decreased in Nurr1/EGFP-expressing cells (Fig. 5A).
To determine whether the decrease in proliferation was caused by an arrest in cell cycle, the relative DNA content of MN9D cells transfected with Nurr1 was measured using FACS. Twenty four hours after transfection of Nurr1 expression vector, a substantial increase in the number of cells at the G 1 phase of cell cycle was detected (Fig. 5B). Thus, Nurr1 inhibits A gel mobility shift assay demonstrated that in vitro transcribed and translated Nurr1, but not Nurr1 dim , formed heterodimers with in vitro translated RXR and bound to a 32 P-labeled ␤-response element DNA probe. Bands were visualized by autoradiography. The positions of monomeric and dimeric complexes are indicated. Coincubation with an RXR antibody abolished dimer but not monomer binding to the ␤-response element probe. C, Nurr1 dim and Nurr1 are equally efficient in activation of an NBRE reporter gene. MN9D cells were transfected with a luciferase reporter plasmid containing three NBRE-binding sites (NBRE-tk-luc) and either Nurr1 or Nurr1 dim expression vectors, as indicated. Cells were harvested after 24 h, and cell extracts were assayed for luciferase and ␤-galactosidase activity. Relative light units (RLU) were computed after normalization to ␤-galactosidase activities. D, Nurr1 dim and Nurr1 are equally efficient in inducing MN9D cell differentiation. Cells were cotransfected with EGFP-and either Nurr1-or Nurr1 dim expression vectors, and the extent of differentiation was determined as in A. E, Nurr1 LBD is not essential for Nurr1-induced differentiation of MN9D cells. Cells were cotransfected with EGFP and either Nurr1 or Nurr1-(1-353) expression vectors, and the extent of differentiation was calculated after 3 days as in A. F, expression of the DNA-binding deficient mutant Nurr1 R334A does not differentiate MN9D cells. MN9D cells were cotransfected with EGFP and either Nurr1 or Nurr1 R334A expression vectors, and the extent of differentiation was calculated after 3 days in culture as in A.

FIG. 3. Retinoids induce morphological differentiation in MN9D cells.
A, after exposure to 1 M all-trans-RA for 2 days, MN9D cells adopted a differentiated morphology, indistinguishable from Nurr1-expressing cells. B-E, optimal differentiation requires both RAR and RXR activation. Cells were grown in the absence (B) or presence of the RAR agonist TTNPB (C), RXR agonist SR11237 (D), or both (E) for 2 days. Cells were analyzed by phase-contrast microscopy. F, RAR VP16 expression induces differentiation of MN9D cells. Cells were cotransfected with RAR VP16 and EGFP expression vectors and analyzed 3 days later under a fluorescence inverted microscope. proliferation by inducing G 1 arrest in MN9D cells. In addition, Nurr1 was able to induce G 1 arrest in the parental neuroblastoma cell line (N18TG2) without promoting morphological differentiation (Fig. 5B, data not shown). The dimerization-deficient mutant Nurr1 dim also induced G 1 arrest, whereas the DNA binding-deficient derivative Nurr1 R334A was inactive (Fig. 5C). These results demonstrate that Nurr1 promoted G 1 arrest through a DNA binding-dependent mechanism that was independent of dimerization with RXR.
Finally, we tested whether the Nurr1-related receptors Nor1 and NGFI-B were able to promote MN9D cell maturation. Interestingly, these experiments showed that Nurr1, Nor1, and NGFI-B were able to induce cell cycle arrest and MN9D cell differentiation with similar efficiencies (Fig. 6, A and B).
Nurr1 and Nor1 Expression Patterns in the Developing CNS-Because Nurr1, Nor1, and NGFI-B induce cell maturation and growth arrest, it was of interest to analyze how they are expressed relative to proliferating cells in vivo. Only Nurr1 and Nor1 are strongly expressed in the embryo, mainly in the developing CNS (5). In the spinal cord of mouse embryos at embryonic day 12.5, Nor1 mRNA was detected by in situ hybridization in the roof and floor plates as reported previously (5). In addition, Nor1 was also detected within the proliferative region in a zone immediately adjacent to the ventricle (Fig. 7A). Nurr1 was expressed in the spinal cord as well but in cells residing just outside of the ventricular zone (Fig. 7B). Interestingly, also in the brainstem Nor1 mRNA localized within the proliferative region marked by the expression of Nestin mRNA (Fig. 7, C and D). At embryonic day 12.5, Nurr1 expression in the midbrain occurs immediately after cells migrate out of the proliferating ventricular zone (Fig. 7, E and F). At embryonic day 16.5, both Nurr1 and Nor1 mRNA expression is detected in cortical cells located next to proliferative progenitors marked by the expression of Nestin mRNA (Fig. 7, G-I). In conclusion, along most of the neural axis, both Nurr1 and Nor1 are expressed immediately after or at the time when neural precursors are becoming post-mitotic. DISCUSSION We have begun to characterize the mechanisms whereby Nurr1 influences dopaminergic differentiation in vivo by using the dopaminergic cell line MN9D as a model for differentiating DA cells. Nurr1 expression in MN9D cells induces G 1 arrest and a differentiated phenotype characterized by extension of neurites. In addition, the closely related NGFI-B and Nor1 are equally efficient inducers of growth inhibition and differentiation. This similarity is presumably dependent on the strong conservation of the DNA binding domains of these nuclear receptors and on their ability to recognize similar DNA binding sequences. It should be noted, however, that among the members of the NGFI-B subfamily, only Nurr1 is expressed in the developing midbrain. Taken together, the results have implications for how Nurr1 influences developing DA cells, as well as for how Nurr1, NGFI-B, and Nor1 are involved in the control of differentiation and growth at other sites where these receptors are expressed in vivo.
An important issue concerns whether or not MN9D cells represent a valid model for differentiating DA cells. MN9D cells are derived from a fusion between a neuroblastoma cell line and primary embryonic DA cells. Several properties of these cells suggest that they are similar to embryonic DA cells. For example, the cells form cell aggregates when cultured in non-adherent conditions, suggesting that the cell line inherited embryonic properties from the primary dopaminergic neurons (40). This conclusion is also supported by the expression of RALDH1 in these cells, a highly specific marker for immature embryonic DA precursor cells (19). 3 Expression of mutated Nurr1 derivatives defined structural requirements for Nurr1-induced DA cell differentiation. Heterodimerization with RXR is not required for the ability of Nurr1 to induce maturation or cell cycle arrest. The results are consistent with the observation that Nor1, which unlike Nurr1 and NGFI-B is unable to heterodimerize with RXR (42), is equally potent in promoting MN9D cell maturation and growth arrest. Interestingly, a Nurr1 derivative that lacks the entire LBD is also competent, although somewhat less efficiently, to promote MN9D cell maturation. Thus, a requirement for an endogenous ligand in the process of MN9D cell differentiation can be ruled out. Nonetheless, it remains possible that putative natural or synthetic Nurr1 ligands, when identified, may modulate Nurr1-induced cell maturation.
MN9D cells respond to all-trans-RA and adopt a morphology that resembles that observed after Nurr1 expression. A strong synergy was observed when assessing effects induced by the 3 D. S. Castro, unpublished data.

FIG. 4. Nurr1 acts downstream of retinoids in promoting morphological differentiation. A, Nurr1
EngR is an efficient dominant negative derivative of Nurr1 in an NBRE-driven reporter gene assay. Human embryonic kidney 293 cells were transfected with a luciferase reporter plasmid containing three NBRE-binding sites (NBRE-tk-luc) and with Nurr1, Nurr1 EngR , or Nurr1 EngR/R334A expression vectors (numbers refer to relative levels of transfected plasmid DNA used in each transfection). Cells were harvested after 24 h in culture and lysed, and cell extracts were assayed for luciferase and ␤-galactosidase activities. Relative light units (RLU) were computed after normalization to ␤-galactosidase activities. B, Nurr1 EngR blocks RAR VP16 -induced differentiation, whereas RAR-(1-403) does not inhibit differentiation induced by Nurr1. Cells were cotransfected with EGFP and different combinations of Nurr1, Nurr1 EngR , RAR VP16 , and RAR-(1-403) expression vectors. The extent of differentiation was calculated 3 days later as described in Fig. 2. C, retinoids do not induce the expression of either Nurr1, NGFI-B, or Nor1. MN9D cells were transfected with a luciferase reporter plasmid containing three NBRE-binding sites (NBRE-tk-luc) and exposed to 1 M all-trans-RA (at-RA). Cells were harvested after 24 h in culture, lysed, and assayed as in A.
combined exposure to synthetic RAR-and RXR-selective retinoids. This is similar to what has often been observed in several other types of cells cultured in vitro (see e.g. Refs. 37, 53, and 54), suggesting that activation of both subunits of RAR-RXR heterodimers results in a more complete biological response in MN9D cells. Retinoids play crucial roles in patterning and differentiation of many cell types in vivo. The results presented here suggest that retinoids may play active roles in DA cell development. Alternatively, a retinoid-activated pathway, which does not necessarily play a role in vivo in differentiating DA neurons, may be triggered in retinoid-treated MN9D cells. Thus, it is interesting to note that RALDH1, an enzyme that can convert retinaldehyde into the active retinoid metabolite all-trans-RA, is expressed already in proliferating DA cell progenitors and continues to be expressed during later stages of dopaminergic development (19,39). Moreover, although competent to differentiate under certain conditions (55), the parental hybridoma cell line N18TG2 does not undergo differentiation by exposure to all-trans-RA indicating that this ability was inherited from the primary dopaminergic neurons. Taken together, these observations support the possibility that retinoids participate in signaling events important for DA cell maturation.
We used dominant negative derivatives of Nurr1 and RAR to investigate whether retinoid-and Nurr1-induced maturation pathways were similar or distinct. The Nurr1 dominant negative derivative was generated by fusing the Drosophila Engrailed repressor domain to Nurr1. Chimeras between a transcription factor and the Engrailed repressor domain have been used extensively in previous studies (56,57), in particular to inhibit gene functions during early Xenopus embryogenesis. The dominant negative activity appeared specific and did not inhibit retinoid-induced reporter genes (data not shown). Also, a mutation within the DNA binding domain of Nurr1 EngR abolished its dominant negative activity both in reporter gene assays and in blocking Nurr1-induced maturation (data not shown).
Our results suggest that Nurr1 and retinoids promote maturation by a similar mechanism but that Nurr1 is acting downstream of retinoid-induced events. A simple explanation would be that retinoids induce the expression of Nurr1, NGFI-B, or Nor1; however, this possibility was not supported by our experiments ( Fig. 4C and data not shown). Thus, it is more likely that a maturation-promoting gene can be activated by both retinoids and Nurr1, but only Nurr1 binds directly to the promoter of the gene. Such a model would be consistent with the FIG. 5. Nurr1 expression induces growth arrest in MN9D cells. A, Nurr1 expression decreases cell proliferation. MN9D cells were cotransfected with EGFP and Nurr1 or empty expression vectors. Proliferating cells were identified by BrdUrd incorporation and immunofluorescence staining using a BrdUrd-specific antibody. The diagram shows percentage of transfected cells that incorporate BrdUrd and is an average of four independent experiments. B, Nurr1 expression induces G 1 arrest in MN9D cells and in parental N18TG2 cells. Cells were cotransfected with EGFP and either Nurr1 or empty expression vectors, harvested after 24 h, and analyzed by FACS. Transfected cells were sorted by EGFP expression, and their distribution in different phases of cell cycle was determined by quantification of DNA by measuring the extent propidium iodide staining (see "Experimental Procedures"). The percentage of transfected cells in the G 1 phase of cell cycle is shown. C, G 1 arrest is mediated through a DNA binding-dependent mechanism, which does not require heterodimerization with RXR. Cells were cotransfected with EGFP and Nurr1, Nurr1 dim , or Nurr1 R334A expression vectors. Cells were harvested and analyzed as in B. Percentage of cells in G 1 , S, and G 2 /M phases were calculated.
fact that RALDH1 starts being expressed well before Nurr1 in developing DA cells.
Our result showing that Nurr1 can induce cell cycle arrest in the neuroblastoma cell line N18TG2 is consistent with a more general growth inhibitory role also in cells of non-mesencephalic origin. In the ventral midbrain, Nurr1 is highly expressed immediately lateral to the ventricular zone and is apparently induced in cells soon after they have become postmitotic (Fig. 7F). Interestingly, a similar correlation between cell proliferation and Nurr1 expression is observed along most of the neural axis (Fig. 7). Although Nurr1 is expressed in close vicinity to proliferative regions of the developing CNS, its expression is induced in cells that already have exited the cell cycle. The timing of Nurr1 expression is therefore inconsistent with an active role in inducing cell cycle exit of proliferating progenitors in vivo. Nonetheless, expression immediately outside of the ventricular zone in several regions of the developing CNS is intriguing and suggests a possible function in stabilizing the quiescent state, thereby protecting cells from undesired re-entry into the cell cycle. Indeed, previous studies of mice with targeted mutations in the cyclin-dependent kinase inhib-itors p19 Ink4d and p27 Kip1 have emphasized that control mechanisms preventing ectopic activation of cell proliferation in mature neuronal cells also in the postnatal brain are of critical importance (58). The NGFI-B/Nurr1/Nor1 group of receptors are rapidly induced in the CNS after stressful insults such as kainic acid-induced seizures, for example (9,10,12). Thus, an exciting hypothesis is that the function of stress-induced NGFI-B/Nurr1/Nor1 gene expression in post-mitotic neurons could be related to the growth inhibitory properties of these receptors and serve to stabilize the non-proliferative state after stressful insults.
In contrast to Nurr1, Nor1 is expressed within the zone of proliferating progenitors both in the spinal cord and in the brain stem. In these regions of the developing CNS, cells are localized at distinct positions depending on the exact phases of the cell cycle (59,60). Nor1 expression is confined to scattered cells in close vicinity to the ventricle suggesting that Nor1 may play a role in regulating cell cycle arrest in cells that are in the transition between M and G 1 phases of the cell cycle. Thus, an attractive possibility is that Nor1 regulates cell cycle exit of dividing neural progenitor cells in the spinal cord and brain stem.
In conclusion, our results have given important insights into the mechanisms whereby Nurr1 is functioning during development. Importantly, the MN9D cell system should provide opportunities to identify Nurr1-regulated target genes with functional relevance in the process of DA cell development. These in vitro based studies should also help to provide additional tools to be used in efforts to generate DA cells in culture, which in turn can be used in therapies aimed at replacing cells that are lost in patients with Parkinson's disease. Finally, because all three related orphans of the NGFI-B/Nurr1/Nor1 family were active in these processes, it is anticipated that the cellular activities unraveled in this study have implications also in non-dopaminergic cells in which these receptors are expressed in vivo.