Peroxisome Proliferator-activated Receptor β Regulates Acyl-CoA Synthetase 2 in Reaggregated Rat Brain Cell Cultures*

Peroxisome proliferator-activated receptors (PPARs) are nuclear hormone receptors that regulate the expression of many genes involved in lipid metabolism. The biological roles of PPARα and PPARγ are relatively well understood, but little is known about the function of PPARβ. To address this question, and because PPARβ is expressed to a high level in the developing brain, we used reaggregated brain cell cultures prepared from dissociated fetal rat telencephalon as experimental model. In these primary cultures, the fetal cells initially form random aggregates, which progressively acquire a tissue-specific pattern resembling that of the brain. PPARs are differentially expressed in these aggregates, with PPARβ being the prevalent isotype. PPARα is present at a very low level, and PPARγ is absent. Cell type-specific expression analyses revealed that PPARβ is ubiquitous and most abundant in some neurons, whereas PPARα is predominantly astrocytic. We chose acyl-CoA synthetases (ACSs) 1, 2, and 3 as potential target genes of PPARβ and first analyzed their temporal and cell type-specific pattern. This analysis indicated that ACS2 and PPARβ mRNAs have overlapping expression patterns, thus designating the ACS2 gene as a putative target of PPARβ. Using a selective PPARβ activator, we found that the ACS2 gene is transcriptionally regulated by PPARβ, demonstrating a role for PPARβ in brain lipid metabolism.

Peroxisome proliferator-activated receptors (PPARs) 1 are members of the nuclear hormone receptor superfamily. Three PPAR isotypes, ␣, ␤ (also called ␦, FAAR, and NUC1), and ␥, have been cloned from Xenopus, rodents, and human. PPARs regulate gene expression by binding as heterodimers with retinoid X receptors to peroxisome proliferator response elements in the promoter of genes involved in lipid metabolism (1)(2)(3).
The biological roles of PPAR␣ and PPAR␥ are relatively well understood, not least because specific ligands for these isotypes have been identified (4). PPAR␣ regulates genes involved in peroxisomal and mitochondrial ␤-oxidation as well as lipoprotein metabolism (2,5). PPAR␣ also suppresses apoptosis in cultured rat hepatocytes (6) and reduces inflammatory responses (7,8). PPAR␥ stimulates adipogenesis, enhances insulin sensitivity, and is involved in cell cycle control and regulation of tumor growth (9 -13).
In contrast, the functions of PPAR␤ are poorly understood, due partly to its ubiquitous expression and the lack of a selective ligand. The objective of this study was to start unraveling PPAR␤ functions using an experimental model that is easy to manipulate and that expresses high levels of PPAR␤ compared with PPAR␣ and PPAR␥. The nervous system seemed an appropriate target for this, as PPAR␤ is abundantly expressed in brain from embryogenesis to adulthood, whereas PPAR␣ and PPAR␥ are barely detectable (14 -17). The brain is the organ with the highest lipid concentration in the body, second only to adipose tissue. Brain lipids serve primarily in modifying the fluidity, structure, and functions of the membranes, and both anabolic and catabolic pathways of lipid metabolism are important in brain development (18,19). Fatty acids need to be activated to their acyl-CoA by acyl-CoA synthetases (ACSs), the activities of which have been found in the brain (20,21). Moreover, ACS1, ACS2, and ACS3 mRNAs have been analyzed in the postnatal rat brain, and the levels of ACS2 and ACS3 vary during its development (22). Given the key role of ACSs in fatty acid utilization, we speculated that they could be potential targets of PPAR␤ in the brain.
We chose cultures of reaggregated neural cells prepared from the telencephalon of rat embryos, which provide a three-dimensional network of different neural cell types that progressively acquire a tissue-specific pattern resembling that in the brain (23,24). We first determined the expression pattern of PPARs during maturation of brain aggregates, providing evidence for PPAR␤ being the prevalent isotype. Then, using a selective activator for PPAR␤, we demonstrate that the ACS2 gene is regulated by PPAR␤ at the transcriptional level.

EXPERIMENTAL PROCEDURES
Cell Culture-Reaggregated brain cell primary cultures were prepared from mechanically dissociated telencephalon of 16-day rat embryos as described previously (24). Cultures were initiated and grown in serum-free, chemically defined medium consisting of Dulbecco's modified Eagle's medium with high glucose (25 mM) 24). Gentamicin sulfate (25 g/ml) was used as antibiotic. Culture media were replenished by exchange of 5 ml of medium (of a total of 8 ml/flask) every third day until day 14, and every other day thereafter because of increased metabolic activity in the cells. The cultures were maintained under constant gyratory agitation (80 rpm) at 37°C and in an atmosphere of 10% CO 2 /90% humidified air.
Northern Blot Analysis-PPAR␤ mRNA was probed with the 1.32-kb BamHI fragment of the plasmid pSG5-rPPAR␤ containing the coding region of the rPPAR␤. 2 The different ACS mRNAs were probed with DNA fragments isolated from their respective rat cDNA containing plasmids (22). ACS1 mRNA was probed with the 0.96-kb EcoRV fragment of the pRACS15 plasmid. ACS2 mRNA was probed with the 2-kb EcoRI fragment of the pBACS9 plasmid. ACS3 mRNA was probed with the 2.2-kb BglII/SacI fragment of the pACS3 plasmid. L27 mRNA, which was used as an internal control, was probed with the 0.22-kb XbaI/EcoRI fragment of the pKS ϩ /L27 plasmid (25). The above-mentioned gel-purified DNA fragments of PPAR␤ (100 ng), the three ACSs (100 ng each), and L27 (25 ng) were used as templates to synthesize [␣-32 P]dCTP-labeled probes using the High Prime kit (Roche Molecular Biochemicals). Probes were purified on Elutip-d columns according to the manufacturer's protocol (Schleicher and Schuell).
Brain cell aggregates from each flask were carefully transferred to round-bottomed 14-ml Falcon tubes, washed with PBS, and frozen rapidly by plunging the capped tubes into liquid nitrogen after aspirating the residual amounts of PBS in the samples. Frozen aggregates were stored at Ϫ80°C until use. Aggregates were thawed rapidly at 37°C in 5 ml of TRIzol (Life Technologies, Inc.), and total cellular RNA was prepared according to the protocol supplied by the manufacturer.
RNA (25 g/well) was electrophoresed in a 2.2 M formaldehyde-1% agarose gel in MOPS buffer and transferred onto Zeta Probe GT-membrane (Bio-Rad) by capillary blotting in 10ϫ SSC for about 18 h. The Northern blot membranes were baked for 30 min at 80°C, prehybridized at 65°C for 2-6 h in 0.25 M Na 2 HPO 4 (pH 7.2) containing 7% SDS, and hybridized at the same temperature with ␣-32 P-labeled probes (2-3 ϫ 10 6 cpm/ml) for 18 h. After hybridization, the membranes were sequentially washed at 65°C for 30 min each in 20 mM Na 2 HPO 4 (pH 7.2) containing 5% SDS and in 20 mM Na 2 HPO 4 (pH 7.2), 1% SDS and then exposed to preflashed films at Ϫ70°C for 2-3 days for PPAR␤ and ACSs and 18 -24 h for L27. Autoradiographs were quantified in a Elscript 400 AT/SM densitometer (Hirschmann), and the mRNA signals were normalized for the loading error using the L27 mRNA signal as internal control, as described previously (26).
RNase Protection Assay-[␣-32 P]UTP-labeled riboprobes were transcribed in vitro with T7 RNA polymerase, and the RNase protection assay was carried out on 15 g of total RNA as described previously (25). The PPAR␣ riboprobe was synthesized from the 717-bp TaqI fragment of the pKS ϩ /PPAR␣ plasmid and the L27 riboprobe from the pBS-L27(150) plasmid linearized with EcoRI. Samples were electrophoresed on 6% polyacrylamide gels and quantified using a Storm 840 phosphorimager (Molecular Dynamics).
Subcloning of the Acyl-CoA Synthetase DNA Fragments-Short DNA fragments of the three ACS cDNAs (22) were subcloned into the pBluescript II KS Ϫ vector (pBSII KS Ϫ ) from Stratagene to use as probes for in situ hybridization. For ACS1, the 305 bp HindIII fragment corresponding to nucleotides 1484 -1789 in the plasmid pRACS15 was subcloned into the HindIII site of the vector, and the recombinant plasmid was named pBSII KS Ϫ -ACS1. For ACS2, the 368-bp BamHI/EcoRI fragment corresponding to nucleotides 143-511 in the plasmid pBACS9 was subcloned into the BamHI/EcoRI site in the bluescript vector, and the recombinant plasmid was named pBSII KS Ϫ -ACS2. For ACS3, the 364-bp SpeI-XbaI fragment corresponding to nucleotides 682-1047 in the plasmid pACS3 was subcloned into the vector digested with SpeI-XbaI, and the recombinant plasmid was named pBSII KS Ϫ -ACS3. The ACS1, ACS2, and ACS3 recombinant plasmids contained the insert in such an orientation that the antisense probe was synthesized with T3 RNA polymerase and the sense probe with T7 RNA polymerase.
In Situ Hybridization-Digoxigenin-labeled PPAR␣ and PPAR␤ riboprobes were transcribed in vitro from linearized, gel-purified plasmids using T7 RNA polymerase as described previously (17). Antisense probes were synthesized from the plasmids pKS ϩ /PPAR␣ and pSK ϩ / PPAR␤, whereas the sense probes were synthesized from the plasmids pSK ϩ /PPAR␣ and pKS ϩ /PPAR␤. Digoxigenin-labeled ACS riboprobes were in vitro transcribed from the plasmids pBSII KS Ϫ -ACS1, pBSII KS Ϫ -ACS2, and pBSII KS Ϫ -ACS3 linearized with XbaI for antisense probes of ACS1 and ACS2 and NotI for ACS3. Sense probes were synthesized from the same plasmids linearized with XhoI for ACS1 and HindIII for ACS2 and ACS3.
Aggregates washed with PBS were embedded in tissue freezing medium (Jung, Nussloch), frozen in isopentane cooled with liquid nitrogen as described previously (24), and stored at Ϫ80°C until use. Cryostat sections (12 m thickness) were hybridized for 40 h at 58°C with digoxigenin-labeled probes (400 ng/ml) in 5ϫ SSC containing 50% formamide and 40 g/ml salmon sperm DNA. Sections were washed, and the mRNAs were visualized by alkaline phosphatase staining as described previously (17) and then dehydrated, mounted, and photographed on an Axiophot microscope (Carl Zeiss).
Immunohistochemistry-Monoclonal antibodies against cell typespecific proteins were used for immunohistochemical analysis of the aggregates. Antibodies against glial fibrillary acidic protein (GFAP), microtubule-associated protein 2 (MAP2) (Roche Molecular Biochemicals), and myelin basic protein (Boehringer-Ingelheim) were used to identify astrocytes, neurons, and oligodendrocytes, respectively. Cryosections (12 M) were fixed for 1 h in 4% paraformaldehyde-PBS at room temperature, washed in PBS (three times for 5 min each), and incubated overnight at 4°C with the primary antibody diluted 1:100 in dilution buffer (PBS containing 1% bovine serum albumin and 0.3% Triton X-100). After washing away the primary antibody, sections were first incubated at room temperature for 1 h with an anti-mouse IgG (Sigma) bridging antibody diluted 1:100 in dilution buffer, and then incubated at room temperature for 1 h with the alkaline phosphatase/ anti-alkaline phosphatase complex diluted 1:100. Sections were then washed with PBS, stained for alkaline phosphatase for 20 min, dehydrated, mounted, and photographed on an Axiophot microscope.
Reporter Gene Assay in HeLa Cells-Cells were cultured at 37°C, 5% CO 2 in Dulbecco's modified Eagle's medium supplemented with antibiotics and 5% fetal calf serum. For transient transfection by the calcium phosphate method, 2.5 ϫ 10 5 cells/well were plated on six-well plates. After 24 h, DNA mixture (200 l) containing 0.1 g of PPAR expression plasmid, 2 0.5 g of the internal control plasmid CMV-␤gal (CLON-TECH), 2 g of the reporter plasmid Cyp2XPalCAT (27), and 5 g of sonicated salmon sperm DNA was added to each well, and cells were transfected for 12 h. The medium was then replaced with Dulbecco's modified Eagle's medium supplemented with 5% charcoal-treated fetal calf serum, activators were added, and cells were incubated at 37°C for 24 h. Cells were then scraped off the dishes, and total cell extracts were prepared by three cycles of freezing in liquid nitrogen and thawing at 37°C. ␤-Galactosidase and chloramphenicol acetyltransferase (CAT) activities were measured in these extracts by standard methods (28). Relative CAT activity was calculated as the fold increase in CAT activity over the basal activity obtained with the empty vector in untreated cells.
Treatment with Activators and Marker Enzyme Assays-Stock solutions of bezafibrate (Sigma) and L-165041 compound (a gift from Parke-Davis) were prepared in ethanol and dimethyl sulfoxide (Me 2 SO), respectively. The effect of activators on marker enzyme activities were measured on reaggregated cultures, equivalent to one-fourth of the original cultures, treated on day 34 or 35 with either 0.05% ethanol (control-bezafibrate), 0.05% Me 2 SO (control-L-165041), 10 M bezafibrate, or 5 and 10 M L-165041 for 24 h. After treatment, the samples were homogenized in ground glass homogenizers. The marker enzymes used were glutamine synthetase for astrocytes, glutamic acid decarboxylase for GABAergic neurons, and choline acetyltransferase for cholinergic neurons, and they were measured using radiometric methods as described previously (24). Determination of protein amounts by the Lowry method was done on replicate cultures representing one-fourth of the original cultures.

PPAR mRNAs in Reaggregated Brain Cell Cultures-As
PPARs have not been studied so far in the developing postnatal brain, we first characterized the aggregates for the basal level of PPAR mRNAs. We examined these cultures at four ages representing different stages of maturation of neurons and glia (henceforth referred to as differentiation). At day 7 (d7), the cells were undifferentiated; at d21, differentiation of neuron and glia subtypes was ongoing, d35 cultures reached the steady state of differentiation, and d49 represented late cultures with the highest level of neuron-specific parameters, although some demyelination may have already occurred (29,30).
The basal levels of the PPAR isotype mRNAs were different in these aggregates. PPAR␣ mRNA was barely detectable by Northern blot analysis but was detected by RNase protection assay (Fig. 1A). In contrast, PPAR␤ mRNA was abundant and could be detected easily by Northern blot analysis of total RNA ( Fig. 1B). Total cellular RNA from adult rat liver and brain was used as a positive control for PPAR␣ and PPAR␤, respectively. PPAR␥ mRNA was not detected in the aggregates even by reverse transcription-PCR (data not shown), and therefore, this isotype was not analyzed in subsequent experiments. There was also a different age-dependent pattern of expression of both PPAR isotypes. On the one hand, the PPAR␤ mRNA basal level was low in undifferentiated cultures (d2 and d7), increased 3-5-fold during differentiation (d14, d21, and d28), and remained high in differentiated (d35) and late (d42 and d49) cultures (Fig. 1C). On the other hand, the level of PPAR␣ mRNA remained low at the four developmental stages examined (Fig. 1, A and C, inset). Thus, of the three PPARs, only the two isotypes ␣ and ␤ were present in the aggregates, and the levels of PPAR␤ increased with the onset of cell differentiation.
Cell Type-specific Expression of PPAR Isotypes in Reaggregated Brain Cell Cultures-In order to identify the cells expressing PPARs in the aggregates, we used in situ hybridization to detect PPAR transcripts in cells ( Fig. 2A), as well as immunohistochemical staining for cell type-specific markers to identify these cells (Fig. 2B). The marker antigens used for immunohistochemistry were MAP2 for neurons, GFAP for astrocytes, and myelin basic protein for oligodendrocytes. Because MAP2 and GFAP are cytoskeletal proteins, immunostaining for these antigens is seen in the cell bodies and processes, and the in situ hybridization signal of PPAR mRNAs is cytoplasmic. Furthermore, we used thionine staining in parallel sections to visualize the nuclei in the aggregates (Fig. 2, A,  q-t, and B, i and k). At d7, a faint signal of PPAR␣ mRNA was detected in most cells of the aggregates (Fig. 2A, compare a with q). At d21, the signal was slightly stronger and restricted to cells with small nuclei (Fig. 2A, compare b with r), which were identified primarily as astrocytes by immunohistochemistry (Fig. 2B, compare c with d, h, and l). The pattern was similar at d35 ( Fig. 2A, c). At d49, the same cell type was positive for PPAR␣ but with a decreased signal intensity ( Fig.  2A, d). The sense probe for PPAR␣ did not stain the sections ( Fig. 2A, e-h), indicating that the signal detected with the antisense probe was specific.
The in situ hybridization signal for PPAR␤ was higher than that for PPAR␣ in all developmental stages of the aggregates ( Fig. 2A, compare i-l with a-d). At d7, the PPAR␤ signal was of medium intensity and was present in most cells of the aggregates ( Fig. 2A, i). At d21, the signal was stronger and exhibited a cell type-specific expression pattern. A signal of very high intensity was observed in cells with large nuclei (Fig. 2A,  compare j with r), which formed a broad layer in the aggregates. Immunostaining with MAP2 revealed that these cells were neurons (Fig. 2B, compare g with h). A PPAR␤ signal of medium and high intensity was also observed in the same region of the aggregates in cells with smaller nuclei, which were neurons, astrocytes, and oligodendrocytes (Fig. 2B, compare g with d, h, and l). At d35 and d49, the pattern of expression of PPAR␤ mRNA was similar to that of d21, but the signal intensity was lower ( Fig. 2A, k and l). The peripheral layer of glial cells were positive for PPAR␤ at all developmental stages except d7 (Fig. 2B, d and l). Again, the sense probe barely stained the sections ( Fig. 2A, m-p), confirming the specificity of the signal obtained for PPAR␤. These results showed that PPAR␤ mRNA was ubiquitous in the aggregates and high in some neurons, whereas PPAR␣ mRNA was mainly detected in the astrocytes.
Acyl-CoA Synthetases in the Aggregates-We used a threestep approach to establish whether ACSs were target genes of PPAR␤ in this model of the developing rat brain. First, we determined the basal levels of the three ACSs in the aggregates. We then examined their cell type-specific expression pattern, and finally studied whether PPAR activators regulate their expression.
All three ACS mRNAs were easily detectable by Northern blot analysis of total RNA (Fig. 3). ACS1 mRNA was present at a low level in the four developmental stages examined. At d21, a small but significant increase of 1.8-fold over the level at d7 was observed (t test, p Յ 0.05). The basal level of ACS2 mRNA was very low at d7, but it increased 5-7-fold in d21 and d35 cultures and remained steady in the late cultures. In undifferentiated cultures, the basal level of ACS3 mRNA was the highest among the three ACSs, and this increased further 2-3-fold until d49. The increase over the basal level at d7 was statistically significant for ACS2 and ACS3 mRNAs (t test, p Յ 0.01).
The comparison of the temporal expression pattern of ACS mRNAs with that of PPARs revealed similarities between ACS1 and PPAR␣ mRNAs on one hand and ACS2 and PPAR␤ mRNAs on the other hand ( Figs. 1 and 3).
Identification of the Cell Types Expressing acyl-CoA Synthetases in the Aggregates-We examined by in situ hybridization the cell type-specific pattern of expression of ACS mRNAs and compared it with the immunohistochemical analysis shown in Fig. 2B. ACS1 mRNA was barely detectable in d7 cultures (Fig. 4A, a). At d21, a signal of medium intensity was observed in small cells in most regions of the aggregates, which were predominantly astrocytes (Fig. 4A, b and Fig. 2B, d). The intensity of this signal decreased at d35 and d49 but was still primarily astrocytic (Fig. 4A, c and d). The glial cells at the periphery of the aggregates did not stain for ACS1 mRNA. ACS2 mRNA signal was higher than that of ACS1 mRNA in the aggregates and was observed in most cells at d7 (Fig. 4A, e). At d21, the signal was of high intensity in neurons with large nuclei, which form a broad layer in the aggregates. In the same region, a signal of medium intensity was observed in other neurons, astrocytes, and oligodendrocytes (Fig. 4A, f). In aggregates of d35 and d49, the ACS2 mRNA signal decreased in intensity (Fig. 4A, g and h) but was present in the same cell types as observed at d21. ACS3 mRNA was found in most cells of the aggregates at days 7 and 21 (Fig. 4A, i and j), with a few cells showing a very strong signal. At d35 and d49, the ACS3 mRNA signal decreased in intensity and was observed in neurons, astrocytes, and oligodendrocytes (Fig. 4A, k and l). Large cells, which stained intensely at d7 and d21, were not apparent in these differentiated cultures. The peripheral layer of glial cells was positive for ACS2 and ACS3 mRNAs in cultures of all developmental stages except d7. No signal was observed with the sense probes for ACS1, ACS2, and ACS3 mRNAs (data not shown).
The in situ hybridization analysis of the three ACSs in d21 aggregates was compared with that of PPAR␤ and is shown at a higher magnification in Fig. 4B. The expression pattern of ACS2 mRNA was similar to that of PPAR␤ (Fig. 4B, b and c). This finding, taken together with their temporal pattern of expression, suggested that ACS2 might be a target gene of PPAR␤.

Activation of Rat PPARs in a Cell-based Reporter Gene
Assay-To test whether ACS2 was a PPAR␤ target gene, a PPAR activator was required that preferentially activates rPPAR␤. In a recent study from our laboratory, several compounds were screened with the coactivator-dependent receptor ligand assay. Fatty acids, eicosanoids, and hypolipidemic drugs were identified as ligands for Xenopus PPARs (27). Most of the ligands identified for xPPAR␤ by this assay exhibited similar or greater activity toward xPPAR␣. As bezafibrate, which was strong on xPPAR␤, was the least active on xPPAR␣, we tested this hypolipidemic drug on rPPARs in a reporter gene assay in HeLa cells. Transfection of the PPAR␣ expression vector resulted in an apparent ligand independent activation of the reporter gene (Fig. 5A), which has been observed previously in HeLa cells (31). In cells treated with 10 and 100 M bezafibrate, PPAR␣-mediated transcription of the reporter gene was enhanced. In contrast, expression of PPAR␤ did not affect the basal activity of the reporter, and treatment with the 100 M bezafibrate activated the reporter gene only weakly. Thus, at a concentration of 10 M, bezafibrate can be used as a specific activator of rPPAR␣.
L-165041 compound has been recently identified as a human PPAR␤ agonist (32). In order to characterize its activity profile on rPPARs, we tested this compound in the reporter gene assay in HeLa cells. In cells treated with 5 and 10 M L-165041, there was only a small increase in PPAR␣-mediated transcription of the reporter gene, indicating a weak effect of the compound on this isotype (Fig. 5B). After a similar treatment, PPAR␤-dependent activity of the reporter gene increased dramatically above its basal level (Fig. 5B). This strong activation contrasts with the absence of background activity of rPPAR␤ and provides strong evidence for the high L-165041 responsiveness of this isotype. We therefore used this compound as a PPAR␤ activator on the aggregates.
Effect of Bezafibrate and L-165041 on the Aggregates-We next tested whether the two activators had any strong and general effect on the aggregates. These cultures are an established model for neurotoxicological studies (33), and the major subtypes of neurons present in the differentiated aggregates are cholinergic, glutamatergic, and GABAergic neurons. In the aggregates, the lactate dehydrogenase release assay underestimates the cytotoxic effects of test compounds, and therefore, estimation of cell type marker enzyme activities is a more accurate indicator of the effect of chemical treatments. Routine biochemical analysis for the cellular effects of compounds involves measuring glutamine synthetase, choline acetyltransferase, and glutamic acid decarboxylase, representing glial cells, excitatory neurons, and inhibitory neurons, respectively, whereas the total protein content is used as an indicator of general cytotoxicity. Treatment of the differentiated aggregates with bezafibrate or L-165041 for 24 h did not cause any visible alteration in their size or glucose metabolism. Similarly, there was little effect on the marker enzyme activities and on the total protein content of the aggregates (Table I). Therefore, we conclude that treatment with the two activators did not modify the neuron and glial cell parameters and did not have general cytotoxic effects on the aggregates. This also suggested that the marker enzymes were probably not regulated by either PPAR.
Acyl-CoA Synthetase 2 Is a Target Gene of PPAR␤ in Reaggregated Brain Cell Cultures-Specific activators or ligands of PPAR␣ and PPAR␥ have been successfully used to identify their target genes. Therefore, it was reasonable to expect an effect of L-165041 on ACS2 expression if it was a target gene of PPAR␤. Treatment of differentiated aggregates for 24 h with bezafibrate had no effect on either the ACS mRNAs or PPAR␤ mRNA (Fig. 6A). However, a similar treatment of the aggregates with 5 and 10 M L-165041 increased the ACS2 mRNA levels significantly (2.4-and 3-fold) but did not affect the ACS1 and ACS3 mRNAs (Fig. 6B). We conclude that ACS2 is a target gene of PPAR␤ in the postnatal rat brain because its temporal and cell type-specific expression pattern correlated with that of PPAR␤, and its mRNA is induced after treatment with a selective activator of this PPAR isotype. The 2-fold increase in PPAR␤ mRNA level itself after L-165041 treatment (Fig. 6B) was statistically significant (t test, p Յ 0.05).
L-165041-mediated Induction of ACS2 mRNA Occurs by Transcriptional Activation and Requires Protein Synthesis-Next, we tested whether the L-165041 effect on ACS2 mRNA expression was at the transcriptional or posttranscriptional level by treating aggregates for 6 and 30 h with 10 M L-165041 in the presence of 0.1 M of the transcriptional inhibitor actinomycin D (Fig. 7, ActD). In the absence of actinomycin D, the expected L-165041-dependent induction was observed after a 30-h treatment. However, treatment with actinomycin D prevented induction by L-165041, ruling out an effect of the compound on the stability of the ACS2 mRNA. If L-165041 stabilized the ACS2 mRNA, then higher levels of this mRNA would have been observed in presence of the compound in actinomycin D-treated samples. Therefore, regulation is at the transcriptional level, and it was of interest to test whether it requires de novo protein synthesis. This possibility was examined in aggregates incubated in the presence of a 5 M concentration of the protein synthesis inhibitor cycloheximide (Fig. 7, CHX).
In the presence of cycloheximide, induction of the ACS2 gene by L-165041 was blocked, providing evidence that de novo protein synthesis was required for ACS2 mRNA stimulation. The requirement for ongoing protein synthesis suggests either that the PPAR␤ protein or another key factor is turned over rapidly or that L-165041 induces the de novo synthesis of a missing regulatory protein.

DISCUSSION
This is the first report on PPAR expression in reaggregated rat brain cell cultures, which is an in vitro experimental model representative of the developing postnatal brain in vivo. The finding that PPAR␤ was the prevalent isotype in these cultures allowed us to test whether ACS1, ACS2, and ACS3 were target genes of this PPAR. Using L-165041 as a selective activator of rPPAR␤ in the aggregates, we have identified ACS2 as the first PPAR␤ target gene.
The Aggregates as an in Vitro Experimental Model of the Developing Postnatal Brain-Morphological and biochemical studies have shown that the reaggregated cultures of mechanically dissociated fetal rat telencephalon are able to mimic several morphogenetic events occurring in vivo, including cell migration, synaptogenesis, and myelination. These three-dimensional cell cultures provide maximum intercellular contacts and interactions, which facilitates the maturation-dependent expression and deposition of extracellular matrix components (34). The distinct stages of cell proliferation and differentiation observed are comparable to the tissue in vivo. The aggregates also exhibit cell type-specific and developmentdependent expression of neuronal and glial cytoskeletal proteins and reproduce the developmental pattern of Na ϩ -K ϩ -ATPase gene expression observed in vivo (35,36). Therefore, the observations made in this model can be considered to reflect the in vivo situation. Because these aggregates were analyzed between day 2 and day 49, this study covers a period of time corresponding to the development of the brain from embryonic day 18 to about postnatal week 6.
PPARs in the Aggregates-The differential expression of PPARs in the aggregates, taken together with their cell typespecific expression pattern, suggests that the neurons contain a low level of PPAR␣ and a high level of PPAR␤, whereas the glia contain a low level of both PPARs. This is in concert with the expression of PPAR␣ and ␤ in monolayer cultures of neonatal neurons (37). However, in astrocyte-enriched monolayer cultures, it was observed that PPAR␤ mRNA levels were higher than those of PPAR␣ mRNA. Furthermore, in both neuron-and astrocyte-enriched monolayer cultures, a very low level of PPAR␥ was present (37). It has also been reported that PPAR␤ was abundant in monolayer cultures enriched for oligodendrocytes but present only at low levels in primary astrocyte cul-  tures (38). Our in situ hybridization results indicate that in the aggregates, PPAR␤ is present at a low level in both these cell types. These differences in PPAR expression between monolayer and aggregate cultures most likely reflect the effects of these two modes of culture on gene expression.
The general differential expression of PPAR mRNAs in d49 aggregates (representative of the 6-week-old rat brain) is consistent with that observed in the adult rodent (14,17,39) and human (40) brain in vivo. Furthermore, the cell type-specific expression of PPAR␤ in these aggregates is also consistent with that found in the rat brain (16,17). Thus, PPAR␤ expression in the aggregates has more similarities with that in vivo than in monolayer cultures. This confirms that reaggregated cultures provide a better model to study the biological role of this PPAR isotype.
Acyl CoA Synthetases in the Aggregates-The differential expression of ACSs in the aggregates, with ACS2 and ACS3 being abundant and ACS1 at a very low level, is consistent with that observed in the developing postnatal rodent brain (22,41). Because the cell type-specific expression of ACS1, ACS2, and ACS3 mRNAs has not been studied in vivo, we analyzed the distribution of these mRNAs in the aggregates and found that ACS1 mRNA is localized mainly in the astrocytes and ACS2 mRNA in astrocytes and neurons. These expression patterns of ACS1 and ACS2 correlate well with PPAR␣ and PPAR␤ expression, respectively. ACS3 mRNA was present in most cells of the aggregates, a pattern of expression quite different from that of PPAR␣ or PPAR␤. The similarity in the ACS2 and PPAR␤ expression patterns identified the former as the product of a potential target gene of the latter. We could confirm this in the differentiated aggregates, which have a high level of PPAR␤, using an activator that exhibited a preference for this PPAR isotype.
Activators of PPAR␤-One of the key issues in analyzing PPAR␤ function is that of the availability of PPAR isotype selective ligands. All compounds that have been reported to activate PPAR␤ so far in various types of assays are also active on PPAR␣ (7,14,27,(42)(43)(44). Species differences in PPAR ligand affinity are another factor that needs to be taken into account. For example, bezafibrate was identified as a ligand for xPPAR␤ (27), but in reporter gene assay in HeLa cells, we found that bezafibrate was an activator of rPPAR␣ but not of rPPAR␤. In the same type of assay, L-165041 preferentially activated rPPAR␤, which enabled us to identify ACS2 as the first target gene of this isotype.
ACS2 as a PPAR␤ Target Gene-In this study, we provide evidence that ACS2 is a target gene of PPAR␤, indeed the first one identified. The L-165041-mediated induction of ACS2 mRNA occurs by transcriptional activation of the gene and is not a consequence of stabilization of the mRNA itself. In the presence of a ligand, a direct target gene of PPAR␤ would be activated via a PPRE present in its promoter, even in the absence of protein synthesis, whereas an indirect response gene would require protein synthesis for its activation. It is not known whether the promoter of the ACS2 gene contains a PPRE. However, it is noteworthy that a functional PPRE has been identified in the promoter of the ACS1 gene (45). The requirement of de novo protein synthesis to achieve L-165041mediated induction of ACS2 mRNA suggests that the ACS2 gene might be an indirect target of PPAR␤. However, at this stage we cannot exclude that a short half-life of the PPAR␤ protein itself is the limiting parameter. This possibility will be tested when a PPAR␤ selective antibody becomes available.
Possible Biological Functions of ACS2 in the Aggregates-ACS2 mRNA levels increase significantly in the differentiating aggregates, which coincides with the maturation of neurons, onset of myelination, and high metabolic activity (29). Because no specific roles have yet been attributed to ACS2 in the brain, we present below several functions in the modulation of which a regulated expression of ACS2 might be beneficial. Maturation of neurons (that is, their cytodifferentiation and the formation of neuronal connectivity) involves extensive lipid biosynthesis and turnover. Biochemical analysis of ACS2 has revealed that its preferred substrates are arachidonic, eicosapentaenoic, and docosahexaenoic acids (46), which are major components of the neural membranes. Another characteristic of neuron maturation is the production of neurotransmitters and their vesicular transport, storage, and release. Recently, ACS1 has been localized in Glut 4-containing vesicles prepared from rat adipocytes (47), and palmitoyl-CoA appears to be required for vesicular transport in vitro (48). With respect to signal transduction pathways that play key roles in the brain, there are several examples of the involvement of acyl-CoA esters (reviewed in Ref. 49), such as in the stimulation of Ca 2ϩ release from intracellular compartments (50 -52) and stimulation of ion channels (53,54). Finally, acylation is a common posttranslational modification of myelin proteins (reviewed in Ref. 55), the bound fatty acids of which are turned over very rapidly (3). ACS activity has been reported in myelin (21), but the type involved not yet determined. It will be the aim of future work to investigate whether and how ACS2 is involved in the processes described above.
Conclusions-In reaggregated brain cell cultures, PPARs are expressed differentially, with a cell type-specific pattern. PPAR␤ is the prevalent isotype, and it is ubiquitous in the aggregates. PPAR␣ is present at a low level and is mainly astrocytic, whereas PPAR␥ is absent. Among the three ACSs analyzed in this study, the temporal and cell type-specific expression pattern of ACS2 correlates very well with that of PPAR␤. The identification of this ACS as a PPAR␤ target gene demonstrates a role of this receptor in lipid metabolism in the brain. The diversity of the cellular processes in which acyl-CoA esters are known to be involved may partly explain why PPAR␤ appears early during development and is ubiquitous in adult tissues. A systematic analysis to correlate PPAR␤ directly with the cellular processes regulated by acyl-CoA esters will help to identify other target genes of this receptor.