At birth, in vivo, most murine PCs are fusiform (bipolar shape, stage I; data not shown), as described for rats 20. When cultured at P0 and kept in organotypic cultures, PCs present first an immature morphology (bipolar fusiform, stage I), then retract their primitive dendrites to become stellate or atrophic (stage II), elongate numerous long and mature dendritic perisomatic protrusions (stage III), and finally develop their ultimate dendritic trees (stage IV) 12.

In organotypic cultures, after 7 days in a serum-containing medium, PCs were mostly in stage II (stellate or atrophic stage), whereas almost no stage III PCs were observed. In order to explore the involvement of T₃ in dendritic differentiation (that is, before the extension of the ultimate dendritic tree in stage IV), cultures from P0 animals were prepared and kept 7 DIV under serum-free conditions, with or without addition of T₃ at different concentrations. Cultures were then fixed and immunolabeled with an anti-calbindin (anti-CaBP) antibody to visualize PCs.

In P0 slices cultivated without T₃, most PCs (76%) displayed 'stellate or atrophic' dendrites after 7 DIV (stage II; Figure 1A,B). Adding T₃ at a concentration of 3 nM did not dramatically modify the repartition of PC classes (Figure 1C,D). In contrast, supply of 30 nM of T₃ led to a significant acceleration of the dendritic differentiation since we observed that only 27% of PCs were in stage II, whereas 39% displayed long dendritic perisomatic protrusions and were thus classified as stage III PCs, and 34% displayed an identified mature dendritic tree (stage IV; Figure 1E,F). Increased concentration of T₃ (100 nM) also led to an acceleration of the dendritic differentiation (compared to 0 nM and 3 nM of T₃), although this treatment was not as efficient as 30 nM T₃ since the proportion of stage IV PCs was lower (Figure 1G,H). Thus, these results show that, in organotypic culture, as already demonstrated for dissociated cell culture 1011, the addition of T₃ to the culture medium promotes PC dendritic development in a dose-dependent manner. In the following experiments, we have used the 30 nM concentration to assess the effects of T₃ since this concentration is the most efficient in our culture conditions.

To determine whether the T₃-induced acceleration of dendritic differentiation involves RORα, we first assessed RORα expression in cerebellar slices in response to T₃ treatment. By combination of promoter usage and alternative splicing, the Rora gene encodes two isoforms in the mouse (RORα1 and RORα4), which differ only in their amino-terminal modulator region 212223.

Western blots of P0 7 DIV cerebellar slices were performed with antibodies directed against the carboxyl terminus of RORα, which can detect both RORα1 and RORα4 isoforms. We detected an increase of 6.8-fold in the amount of RORα1 protein in slices treated with 30 nM of T₃ (Figure 2A).

In the cerebellum, RORα is known to be expressed only in PCs and interneurons 24. In order to determine in which cell type the upregulation of RORα expression occurs, we used immunofluorescence to detect and locate the RORα protein in organotypic cultures. Since only PCs in the cerebellum express CaBP, we used CaBP as a PC-specific marker, and we used parvalbumin as a marker of interneurons. Both mature interneurons and PCs express parvalbumin: interneurons were unambiguously identified as parvalbumin-positive and CaBP-negative cells.

To assess whether T₃ led to increased expression RORα1 in PCs, we quantified the fluorescence density of RORα labeling within the nucleus of PCs. We observed a significant increase in the fluorescence density in T₃-treated slices compared to control T₃-untreated slices (Figure 2B). Interestingly, following T₃ treatment, RORα labeling was observed in PCs and also in some CaBP-negative cells. Some CaBP-negative cells that express RORα also expressed parvalbumin, and were thereby identified as interneurons (Figure 2B). Therefore, T₃ treatment led to increased expression of RORα in both PCs and parvalbumin-positive interneurons.

To determine whether the increase in RORα protein levels in T₃-treated cerebellar slices is the consequence of increased expression of the Rora gene, and not stabilization of the protein, we analyzed by real time RT-PCR the mRNA level of the different Rora isoforms in the T₃-treated slices compared to untreated slices after 7 DIV (Figure 2C). A 3.3-fold increase was observed for Rora1 after 7 DIV. The Rora4 mRNA level was similar to untreated slices after 7 DIV. These results show that T₃ leads to increased expression of the Rora1 isoform after 7 DIV.

These results show that T₃ induced increased RORα1 protein levels in PCs in parallel with their dendritic differentiation after 7 DIV.

As we previously demonstrated that RORα is involved in early dendritic differentiation 12, we examined whether T₃ promotes this early change. We thus assessed the effect of T₃ treatment on cerebellar slices after 3 DIV, a time when PCs cultured without T₃ display mainly bipolar fusiform dendritic morphology (stage I; 97%) whereas very few are in a stellate or atrophic morphology (stage II; 3%; Figure 3A). In the presence of 30 nM of T₃ after 3 DIV, all PCs were still in stage I or II, but we observed an increased number of PCs in stage II (28%; Figure 3A) compared to the control without T₃ (3%; Figure 3A). From those experiments, we can conclude that T₃ promotes the first dendritic differentiation steps of PCs from stage I to stage II in organotypic cultures.

To determine whether T3 increases Rora expression in early stages of PC development, we analyzed the mRNA levels of the Rora isoforms in response to T₃ treatment during the first day of culture (Figure 3B). We observed a specific increase in the Rora1 mRNA level after 6 h and 24 h of T₃ treatment (6.5- and 5.7-fold increase, respectively). The Rora4 mRNA level was slightly and transiently increased in slices after 24 h of T₃ treatment (1.7-fold increase). These results show that T₃ leads to increased expression of the Rora1 isoform in fusiform PCs at P0 and, to a lesser extent, of the Rora4 isoform. Interestingly, our results also revealed that mRNA levels of both Rora1 and Rora4 isoforms were stable in cultures made at P0 and kept for 6 h, 24 h and 7 DIV in culture without T₃ (compare Figures 2C and 3B).

The experiments described above show that T₃ promotes dendritic differentiation (Figures 1 and 3) and leads to increased expression of RORα1 (Figures 2 and 3). We have recently shown that RORα is a crucial factor controlling the early steps of PC dendritic differentiation, and staggerer RORα-deficient PCs do not progress beyond early bipolar migratory morphology 12. To determine whether RORα is actually involved in the T₃-induced PC dendritic differentiation, we followed the progression of the dendritic differentiation of PCs from staggerer (Rora^sg/sg) and corresponding control Rora^+/+ cerebellar slices treated or not with T₃.

As previously observed in serum-containing cultures 12, PCs from Rora^sg/sg cultured in serum-free medium display the embryonic bipolar shape (stage I) after 7 DIV (Figure 4B), whereas most PCs in control Rora^+/+ cultures display 'regressive-atrophic' dendrites (stage II; Figure 4A). In the presence of 30 nM T₃, stage II, III and IV PCs were found (Figure 4C) with the same proportions as described in Figure 1. In contrast, PCs from Rora^sg/sg animals still display embryonic bipolar shape (Figure 4D) with long processes characteristic of stage I PCs 12, indicating that they were not responsive to T₃ treatment and remained in the very early stage of dendritic differentiation.

The absence of a functional RORα protein thus prevents T₃-induced accelerated dendritic differentiation of immature bipolar P0 PCs. This experiment shows that RORα is required in the T₃-induced dendritic differentiation-promoting process.

To gain further insight into the mechanism by which T₃ up-regulates Rora gene expression, we tested the effect of T₃ on the transcriptional activity of the p(-487)Rora-Luc construct in HepG2 cells, which shows 82.9% sequence homology with the murine sequence and has been previously used as a model to analyze the transcriptional regulation of the Rora gene 25 (Additional file 1). T₃ treatment resulted in a 3.6-fold increase in the activity of the p(-487)Rora-Luc construct compared to its basal activity in the absence of T₃ (Figure 5). Plasmid pDR4-TK-Luc, which contains a thyroid receptor response element (DR4), was used as a control for the effect of T₃ on transcriptional activity. As expected, the luciferase activity of pDR4-TK-Luc was strongly increased (12.1-fold) by T₃ treatment, indicating that T₃ is active under our experimental conditions.

Taken together, the results of these experiments indicate that the effect of T₃ on Rora expression is, at least in part, transcriptional and that the -487 to -45 Rora promoter region is involved in this regulation.

The role of TH in mammalian brain is well documented, particularly during cerebellar development (for review, see 126). Congenital hypothyroidism in humans leads to a syndrome termed cretinism 27, the apparent symptoms of which include ataxia and poor motor skills, indicating cerebellar defects. In PCs, TH is known to strongly promote differentiation of the elaborate dendritic tree and synapse formation. In contrast, little is known about its role in the events preceding the development of the ultimate dendritic tree, in particular the steps of neuritic regression and early extension of perisomatic protrusions, occurring in vivo in the rodent between P0 and P7.

To better understand the effect of TH action in the developing brain, the temporal patterns of initiation and cessation of hormone action need to be determined. Most in vitro or in vivo experiments explore the effects of hypo- or hyperthyroidism in the cerebellum from P15, or its equivalent age in culture. At this age, the characteristic shape of the dendritic arborization is already achieved, and extrinsic factors such as electrical activity 28 from granule cells, trophic factors 29303132 and TH modulate the growth of the dendritic arborization 4. Studies have shown a role for TH in the persistence of the external granular layer and the migration of granule cells into the internal granular layer 4533, in the proliferation and differentiation of interneurons 34, as well as a direct role of TH on PCs through TRα1 receptor activation 10. However, the effects of TH on PC dendritic differentiation during early steps that do not require cell-cell interaction have not been shown.

Using P0 slices after 3 DIV, we could specifically assess the role of TH in early development, and our results show that T₃ also plays a key role in the early dendritic differentiation of PCs in organotypic cultures, that is, before the formation of the elaborated dendritic tree. These data extend the well-known role of T₃ in the later stages of PC dendritic differentiation 68910 and identify a third molecule besides RORα and SCLIP 1235 involved in the first steps of PC dendritic differentiation. Interestingly, T3 is the first extrinsic factor described to play a role in these processes.

We also show that T₃ promotes the expression of both RORα and parvalbumin in interneurons, which corroborates recent results from Manzano et al. 34, who have shown that TH acts on the proliferation and differentiation of interneuron precursors in the cerebellum 34. Further studies will be required to determine whether the TH action on interneurons is mediated by RORα.

Interestingly, T₃ addition led to increased expression of RORα in the cerebellar PCs and interneurons. This result is in accordance with previous studies that showed decreased expression of RORα in the cerebellum of hypothyroid rats 36, whereas T₄ replacement led to increased expression 17.

Our results indicate that RORα1 expression is required for the T₃-induced effect on early dendritic differentiation. Further, we show that the activity of the Rora promoter was enhanced by T₃ treatment in culture, suggesting that TH acts on the process of early dendritic differentiation through increased expression of the Rora gene. TH binds to the nuclear TH receptor (TR), a ligand-regulated transcription factor, which then binds to a target DNA sequence known as a TH response element (TRE) within the promoter region of target genes. Further studies are needed to determine whether RORα is a target gene of TR, and whether the transcriptional effect observed in our study is under direct control. An additional level of interaction between RORα and TR has been demonstrated by Koibuchi and collaborators 3738, who showed that RORα1 increases TR-induced transactivation on several TREs. This could account for the RORα requirement in the T₃-mediated promotion of PC dendritic differentiation observed in this study. TR binds as a monomer, homodimer, or heterodimer (particularly with retinoid × receptors) to the TRE, which is composed of two half-site core motifs (AGGTCA) with specific nucleotide spacing and orientation. RORα binds as a monomer to a consensus motif composed of a 6-bp AT-rich sequence 5' to a half-site core motif, AGGTCA (ROR-response element, RORE), to activate transcription 23. Both TR and RORα are thus transcription factors that share the common core motif within their response elements. RORα1 is able to bind as a monomer to one of two core motifs (AGGTCA) of a TRE that is preceded by an AT-rich sequence 2337. This suggests that a subset of natural TREs containing appropriate AT-rich sequences could serve as dual-response elements for TR and RORα. Because of the high homology between the human and murine RORα1 coding and promoter sequences, it is possible that RORα mediates some TH actions in human. Beside its roles in the developing cerebellum, RORα has also been shown to play critical roles in many different tissues and systems, including immunity, cancer, cellular metabolism, circadian rhythm, development and ageing (for review, see 39). Understanding the roles of RORα could therefore provide further information about the pleiotropic effects of late prenatal or early postnatal hypo- and hyperthyroidism in humans.

RORα has been shown to be crucial for the progression of early differentiation of PCs in a cell-autonomous manner 12. In cerebellar slices, T₃ is likely to act on RORα expression within PCs. Our results extend previous studies of Heuer and Mason 10, which clearly demonstrated that PCs are a direct target of TH action through activation of TRα1: TH promotes the late stages of the elaboration of PC dendritic arborization, which is also dependent upon granule cell differentiation and synaptogenesis. Interestingly, RORα has been shown to control the expression of Sonic hedgehog (Shh) in PCs, which in turn promotes the proliferation of granule cells precursors in the external granular layer 40. Thus, a coordinate mechanism involving RORα and TH in cerebellar development can be proposed in which both T₃ and RORα act on PC dendritic differentiation directly as well as indirectly via the promoting effect on granule cell development. However, the later and direct effects of T₃ on early PC differentiation are unlikely to be mediated by RORα since we have shown that RORα does not influence this later step of differentiation 12.

In its homozygous state, the murine staggerer mutation of the Rora gene leads to cerebellar atrophy due to the degeneration of most PCs 1315414243. Several histological studies of the Rora^sg/sg cerebellum show that the remaining PCs are immature and display atrophic dendrites, devoid of spines 444546. These abnormalities of dendritic differentiation observed in homozygous staggerer mice are similar to, but worse than, those observed in hypothyroid rats. This implies that RORα acts on additional processes in cerebellar development, apart from those induced by THs. This hypothesis is strengthened by the recently demonstrated neuroprotective role of RORα at least partly through its control of oxidative stress mechanisms 1647.

In conclusion, our results show that RORα plays a critical role in the early T₃-induced dendritic differentiation of PCs.

Animal housing and all procedures were carried out in accordance with the guidelines of the French Ministry of Agriculture and the European Community. Swiss mice were obtained from Janvier (Le Genest-St-Isle, France). The staggerer Rora^sg/sg mutant mice were maintained on a C57BL/6J genetic background in our colony. Rora^sg/sg and their Rora^+/+ littermates were obtained by intercrossing fertile heterozygous Rora^+/sg animals, and were genotyped by PCR, as previously described 12.

Swiss mice at P0 were used. Organotypic cultures of cerebellum were prepared as described previously 48. Briefly, after decapitation, brains were dissected out into cold Gey's balanced salt solution (Sigma, Lyon, France) supplemented with 5 mg/ml glucose, and the meninges were removed. Parasagittal cerebellar slices (350 μm thick) were cut on a McIlwain tissue chopper (Stoetling Europe, Dublin, Ireland) and transferred onto 30 mm Millipore membrane culture inserts with a 0.4 μm pore size (Millicell CM, Millipore, Molsheim, France). Slices were maintained in culture in six-well plates containing 1 ml per well of medium containing basal medium with Earle's salts (BME), supplemented with Sigma I-1884 supplement (1:100 dilution, resulting in final concentrations of 5 μg/ml insulin, 5 μg/ml transferrin, and 5 ng/ml sodium selenite), 0.5 μg/ml BSA (Sigma), 4 mM L-glutamine (Invitrogen, GIBCO, Cergy Pontoise, France), 5 mg/ml glucose, with or without T₃ at 37°C in a humidified atmosphere with 5% CO₂. The medium was replaced every 2 days (after 2, 4 and 6 days in culture).

Mice obtained from Rora^+/sg intercrosses were also used in this study. In these litters, Rora^+/+, Rora^+/sg and Rora^sg/sg mice could be generated. For each animal, slices of each cerebellum were divided between two Millicells: half of the cerebellar slices served as controls and no T₃ was added and the other half were treated with T₃ (30 nM) in order to compare control (0 nM T₃) versus T₃-treated slices (30 nM T₃) from the same animals. The genotype was determined a posteriori by PCR on tail biopsy, in blind studies.

Immunostaining of CaBP, parvalbumin or RORα was performed as described previously 12. Briefly, cerebellar slices were fixed in 4% paraformaldehyde, and then incubated for 1 h in phosphate-buffered saline containing 0.25% Triton X-100, 0.2% gelatin, 0.1% sodium azide (PBSGTA) and 0.1 M lysine. Rabbit polyclonal or mouse anti-CaBP antibody (1:5,000 dilution; Swant, Switzerland) to visualize PCs, or rabbit polyclonal anti-parvalbumin (1:5,000 dilution; Swant) to visualize both PCs and interneurons, and goat polyclonal anti-RORα1 (sc-6062; 1:4,000 dilution; Santa-Cruz, Tebu-Bio SA, Le Perray en Yvelines, France) in PBSGTA were applied overnight. At this dilution, the intensity of RORα labeling was correlated to the RORα expression level 12. Specific labeling was detected with Cy3-conjugated donkey anti-rabbit antibody (1:500 dilution; Jackson Immunoresearch, Immunotech, Marseille, France) and FITC-conjugated donkey anti-goat antibody (1:2,000 dilution; Jackson Immunoresearch). The slices were analyzed with an inverted microscope (Nikon Eclipse TE 300). Immunofluorescence images were captured at 400× magnification using a Qimaging Retiga 1300 camera, and analyzed using Image-Pro Plus 4.1 software (Media Cybernetics, Bethesda, MD, USA). For RORα fluorescence intensity measurements, fluorescence density was measured in the nucleus of PCs (visualized by CaBP immunolabeling) using MetaMorph software.

Classification of PCs was assessed after CaBP immunostaining, as previously described 12. Briefly, fusiform PCs with a bipolar shape, reminiscent of embryonic migratory PCs, are defined as stage I and correspond to both 'simple' and 'complex' fusiform stages, observed from embryonic day 16 to P4 in vivo 20. Stage II comprises PCs with short processes all around the soma. This 'stellate' stage includes both 'regressive-atrophic dendrites' and 'stellate cell' stages described previously, from P2 to P6 in vivo. PCs with more than one long and mature dendritic protrusion are defined as stage III. They correspond to PCs around P5 to P10 in vivo. Finally, PCs with one well identified dendritic tree (defined as primary dendrites giving rise to additional side branches) are classified as stage IV. Images were taken from all slices, corresponding to at least 200 PCs in each experiment. Quantification was performed on three independent experiments.

Cultured slices were lysed in solubilization buffer (500 mM NaCl, 1 mM MgCl₂, 2 mM EGTA, 50 mM Bicine, pH 9.0, 50 mM NaF, 5 μM ZnCl₂, 100 μM Na₃VO₄, 1 mM dithiothreitol, 5 nM okadaic acid, 2.5 μg/ml aprotinin, 3.6 μM pepstatin, 0.5 μM phenylmethylsulfonyl fluoride, 0.5 mM benzamidine, 5.3 μM leupeptin) and dounced at 4°C. Insoluble materials were removed by centrifugation (13,000 g for 20 minutes at 4°C), supernatants were isolated and the samples were stored at -80°C. Proteins were dosed with the DC protein assay kit (Bio-Rad, Hercules, CA, USA). As previously described 49, cell-extracts containing equivalent amounts of protein were boiled for 5 minutes in sample loading buffer. After a 10% SDS-PAGE, proteins were transferred to a polyvinylidene difluoride membrane (ICN Biochemicals, Costa Mesa, CA USA). Non-specific sites were blocked with 5% skimmed dried milk for 2 h. Blots were then incubated overnight at 4°C with primary antibodies against RORα (1:2,000; Santa Cruz) and α-tubulin (1:10,000; Sigma) in 5% skimmed dried milk. They were then incubated with horseradish peroxidase-conjugated secondary antibodies in 5% skimmed dried milk for 1 h. The revelation was processed with enhanced chemoluminescence substrate (Amersham, Saclay. France). Quantification was performed using Densylab software (Microvision Instruments, Evry, France).

Total RNA from cerebellar slices from three animals was prepared according to the manufacturer's instructions using the RNeasy kit (Qiagen, Courtaboeuf, France) and cDNAs were synthesized from 1 μg of RNA (Promega, Charbonnieres-les-Bains, France) and avian myeloblastosis virus (AMV) reverse transcriptase, as per the manufacturer's instructions.

RT-PCR was performed using the ABsolute™ QPCR SYBR^® Green Mixes Kit (ABgene, Courtabeoeuf, France), as per the manufacturer's instructions. Reactions were performed in 25 μl of total volume containing ABsolute™ QPCR SYBR^® Green Mix with 8 ng of the first-strand cDNA and 300 nM of primers. The following primers were used: Rora1 sense, 5'-AGGCAGAGCTATGCGAGC-3', and antisense, 5'-TCAAACAGTTCTTCTGACGAGG-3'; Rora4 sense, 5'-GTCACATGGAGCCTCTTATGG-3', and antisense, 5'-TCAAACAGTTCTTCTGACGAGG-3'; 18 s sense, 5'-GGGAGCCTGAGAAACGGC-3', and antisense, 5' GGGTCGGAGTGGGTAATTT-3'. Amplification was performed on an iCycler (Bio-Rad) according to the manufacturer's instructions and cycle parameters were: 50°C (2 minutes) and 95°C (10 minutes), followed by 40 cycles of 95°C (15 s), 60°C (30 s) and 72°C (30 s). For expression quantification, a comparative C_T method was used 5051. The ΔC_T value was obtained by subtracting the C_T value of the 18 S (reference) from the C_T value of the gene of interest, where in each case the mean value of three reactions was used. For each gene, the fold change was calculated according to the formula , where ΔΔC_T was the difference between the ΔC_T of T3-treated cultures and the ΔC_T of untreated cultures as a calibrator value. To distinguish specific amplicons from non-specific amplifications, a dissociation curve was generated for each transcript. Quantification was performed on three independent experiments.

The plasmid p(-487)Rora-Luc contains the luciferase reporter gene placed under the control of the promoter region of the human Rora gene, from -487 to -45 relative to the Rora translation initiation site 25. The vector pTRα, containing mouse TRα1 cDNA, cloned in plasmid pSG5 and plasmid pDR4-TK-Luc, which contains a TRE in front of the promoter of the thymidine kinase gene of the herpes simplex virus controlling expression of the luciferase gene where kind gifts of Dr F Flamand (Ecole Normale Superieure, Lyon, France).

The promoter-less pGL3-basic luciferase reporter vector (pGL3-Luc) was from Promega. Transient transfection experiments were done with HepG2 human hepatoma cells using the calcium phosphate method. Twenty-four hours after the transfection, 30 nM of T₃ were added to the medium and the luciferase activity was assayed 24 h later, as described 25. Activities corresponding to cells cultured with 30 nM of T₃ were expressed relative to those of control cells cultured without T₃.

Independent experiments were performed with 10 to 12 cerebellar slices per sample and repeated three times using matched controls. For PC stage quantification, at least 200 PCs were analyzed in each sample. For the RORα RNA level quantification by real-time PCR, all slices of three animals were used in each experiment. Results are expressed in Figures as mean ± standard deviation. The statistical significance of differences between control and T₃-treated slices was assessed by a Student's t-test using Statview software (SAS Institute Inc., Berkeley, CA, USA).