Retinoic Acid Induces Neurogenesis by Activating Both Retinoic Acid Receptors (RARs) and Peroxisome Proliferator-activated Receptor β/δ (PPARβ/δ)*

Background: RA regulates transcription through the CRABP-II/RAR and FABP5/PPARβ/δ pathways, but the contributions of these pathways to RA-induced neuronal differentiation are unknown. Results: RA signaling switches from CRABP-II/RAR to FABP5/PPARβ/δ during differentiation. The switch is controlled by transient up-regulation of RARβ and the CRABP-II/FABP5 ratio. Conclusion: Both RA pathways are employed during neuronal differentiation. Significance: The data contribute insights into RA-induced neurogenesis. Retinoic acid (RA) regulates gene transcription by activating the nuclear receptors retinoic acid receptor (RAR) and peroxisome proliferator-activated receptor (PPAR) β/δ and their respective cognate lipid-binding proteins CRABP-II and FABP5. RA induces neuronal differentiation, but the contributions of the two transcriptional pathways of the hormone to the process are unknown. Here, we show that the RA-induced commitment of P19 stem cells to neuronal progenitors is mediated by the CRABP-II/RAR path and that the FABP5/PPARβ/δ path can inhibit the process through induction of the RAR repressors SIRT1 and Ajuba. In contrast with its inhibitory activity in the early steps of neurogenesis, the FABP5/PPARβ/δ path promotes differentiation of neuronal progenitors to mature neurons, an activity mediated in part by the PPARβ/δ target gene PDK1. Hence, RA-induced neuronal differentiation is mediated through RAR in the early stages and through PPARβ/δ in the late stages of the process. The switch in RA signaling is accomplished by a transient up-regulation of RARβ concomitantly with a transient increase in the CRABP-II/FABP5 ratio at early stages of differentiation. In accordance with these conclusions, hippocampi of FABP5-null mice display excess accumulation of neuronal progenitor cells and a deficit in mature neurons versus wild-type animals.

has been an especially useful model for studying RA-induced neuronal differentiation. Treatment of these cells with RA results in the formation of embryonic bodies resembling the blastula stage. Re-plating of these cell aggregates then gives rise to neuronal and glial cells (2,3). The biological activities of RA originate from its ability to activate several members of the nuclear receptor family of transcription factors as follows: the classical RA receptors RAR␣, RAR␤, and RAR␥ (4) and the peroxisome proliferator-activated receptor ␤/␦ (PPAR␤/␦) (5)(6)(7)(8)(9). The partitioning of the hormone between its receptors is regulated by two intracellular lipid-binding proteins that deliver it from sites of synthesis in the cytosol to cognate receptors in the nucleus; cellular RA-binding protein II (CRABP-II) transports RA to RARs, and fatty acid-binding protein type 5 (FABP5) shuttles the ligand to PPAR␤/␦. Hence, the spectrum of target genes activated by RA and the biological responses to the hormone are determined by the relative expression of these binding proteins in specific cells; RA controls expression of RAR target genes in cells that display a high CRABP-II/FABP5 ratio, but it regulates PPAR␤/␦ target genes in cells in which this ratio is low (7, 8, 10 -14). An additional RA-binding protein that may be involved in regulating the activities of the hormone is CRABP-I. It was thus reported that CRABP-I dampens the response of F9 teratocarcinoma cells to RA and that it does so by enhancing the degradation of the hormone (15,16).
It is well established that activation of RAR by RA is essential for induction of neuronal differentiation, and various RAR target genes were reported to be involved in the process (1,17). These include primary targets, e.g. Hoxa-1, Hoxb-2, Sox6, and Wnt-1, and indirect targets such as Mash-1, Ngn-1, NeuroD, N-cadherin, and Pbx. Components of the alternative RA pathway, FABP5 and PPAR␤/␦, are highly expressed in embryonic brain. In rat brain, both proteins appear at mid-term, around day E10.5. FABP5 expression peaks at birth and gradually decreases to attain a lower sustained expression in adult brain. PPAR␤/␦ expression peaks at E13.5-15.5 days and then decreases slightly but remains high through development and adult life (18,19). It was reported that activation of PPAR␤/␦ induces oligodendrocyte differentiation and enhances neuronal maturation in cultured cell models of neurogenesis (20,21) and that, by exerting anti-apoptotic and anti-inflammatory functions, PPAR␤/␦ displays neuroprotective activities (22). Available information thus raises the possibility that both RAR and PPAR␤/␦ are involved in mediating RA-induced neuronal differentiation. However, the contributions of the two paths to the overall process are poorly understood.
Here, we show that the CRABP-II/RAR path mediates the ability of RA to induce commitment of stem cells to neuronal progenitors and that the FABP5/PPAR␤/␦ pathway is critical for subsequent progenitor cell differentiation into mature neurons. The shift in RA signaling during neurogenesis is accomplished by a transient increase in the CRABP-II/FABP5 ratio in stem cells undergoing differentiation to neuronal progenitors. In accordance with the conclusion that the RA-activated FABP5/PPAR␤/␦ pathway is critical for promoting differentiation of neuronal progenitor cells to mature neurons, the hippocampus of FABP5-null mice displays an excess of neuronal progenitor cells and a deficit in mature neurons.
Cell Culture and Differentiation-P19 cells were obtained from ATCC. All experiments were performed with cells within 10 passages. Transfections were carried out using Lipofectamine 2000 (Invitrogen). Cells stably expressing shRNAs were selected using puromycin (2.5 g/ml).
P19 cell were induced to undergo neuronal differentiation according to established procedures (23). Briefly, cells were cultured on 1% agarose-coated 10-cm dishes at 3.0 ϫ 10 5 cells/ml in ␣-minimal essential medium (Hyclone) supplemented with 10% FBS. Differentiation was induced by addition of RA (1 M) and medium containing RA replaced 2 days later. On day 4, cell aggregates were collected by centrifugation, separated to single cells by trypsin/EDTA treatment, replated onto poly-L-lysinecoated plates, and cultured in ␣-minimal essential medium supplemented with 10% FBS. On day 6, medium was replaced with neurobasal medium containing B27 supplement (Invitrogen) and 2 mM GlutaMAX. Medium was replaced every 2 days for an additional week.
Mouse Studies-WT and FABP5-null C57BL/6 littermate mice were maintained on a 12-h light and dark cycle on a normal chow diet. Mice were housed according to ARC protocol. The breeding diet (5P76 from LabDiet) contained 25,000 -29,000 IU of vitamin A per kg. Mice had access to water and diet ad libitum.

Neuronal Differentiation Is Accompanied by a Transient Shift in RA Signaling toward the CRABP-II/RAR Pathway-
The involvement of RA and its binding proteins and nuclear receptors in neurogenesis was examined using P19 cells, a well established cultured cell model of neuronal differentiation (2,3,24). P19 cells were cultured in agarose-coated plates and induced to differentiate by treatment with RA. Four days post-induction, cell aggregates were collected, dissociated, and cultured in poly-L-lysine-coated plates in medium containing 10% FBS but not supplemented with RA. On day 6, the medium was changed to neurobasal medium containing B27 supplement and GlutaMAX (see "Experimental Procedures"). The progress of the differentiation program was evident by loss of the stem cell marker Oct3/4 at day 2, transient up-regulation of the neural progenitor marker nestin, whose mRNA level peaked at day 6, appearance of the immature neuronal marker ␤3-tubulin at day 4, and finally, appearance of the mature neuronal markers MAP2 and NeuN on days 6 -12 ( Fig. 1, a and b).
Undifferentiated P19 cells express the RA receptors RAR␣, RAR␤, RAR␥, and PPAR␤/␦, as well as the RA-binding proteins CRABP-II and FABP5 (Fig. 1c). Induction of differentiation by treatment of cells with RA resulted in transient up-regulation of CRABP-II and down-regulation of FABP5 that were observed at the level of both the respective proteins ( Fig. 1d) and mRNAs ( Fig. 1, e-g). Following the initial decrease, the level of both FABP5 protein (Fig. 1d) and mRNA (Fig. 1f) increased to attain a 2-2.5-fold higher level in mature neurons as compared with undifferentiated P19 cells. Induction of differentiation did not markedly affect the levels of either RAR␣ or PPAR␤/␦. The level of RAR␥ mRNA decreased by about 5-fold by day 4 and remained low in mature neurons (data not shown). Interestingly, concomitantly with the transient up-regulation in expression of CRABP-II, induction of differentiation resulted in up-regulation of RAR␤ mRNA whose expression peaked at day 4 and subsequently decreased (Fig. 1h).
The observations that expression of both CRABP-II and its cognate receptor RAR␤ is up-regulated early during the neuronal differentiation process suggest that initial events in the process are driven by RA through the CRABP-II/RAR path. In agreement with this notion, the synthetic RAR agonist TTNPB, but not the PPAR␤/␦ agonist GW0742, mimicked the ability of RA to increase the expression of CRABP-II and decrease the expression of FABP5 (Fig. 2, a and b). Treatment with TTNPB also induced the expression of the neuronal progenitor markers nestin and ␤3-tubulin (Fig. 2, c and d) and decreased the expression of the stem cell marker Oct3/4 (Fig. 2d).

RA Signaling through the FABP5/PPAR␤/␦ Path Inhibits the Formation of Neuronal Progenitors but Enhances Progenitor
Maturation-The shift in RA signaling away from the FABP5/ PPAR␤/␦ path in the early stages of neuronal differentiation may signify that the path is detrimental for the process. The effects of varying the expression of FABP5 and PPAR␤/␦ on the formation of neuronal progenitors were thus investigated. P19 cells in which the expression of FABP5 and PPAR␤/␦ is stably reduced using corresponding shRNAs were generated (Fig. 3, a  and b). Interestingly, down-regulation of FABP5 also reduced the expression of PPAR␤/␦, suggesting that the receptor is subject to auto-regulation by the FABP5/PPAR␤/␦ path. Cells were treated with RA, and their differentiation to neural progenitors was followed by monitoring the expression of nestin. Nestin expression reached a higher level in cells with decreased expression of either FABP5 or PPAR␤/␦ (Fig. 3, c and d). Conversely, ectopic overexpression of either PPAR␤/␦ or FABP5 (Fig. 3e) decreased the expression of nestin (Fig. 3f). Notably, as transfection efficiency in these experiments, assessed by fluorescence microscopy, was ϳ30% (data not shown), the observed changes in nestin expression upon varying the expression of FABP5 and PPAR␤/␦ underestimates the magnitude of the effect. The data thus suggest that the FABP5/PPAR␤/␦ path inhibits the differentiation of stem cells to neuronal progenitors.
Surprisingly, immunohistochemistry showed that although cells with decreased expression of FABP5 or PPAR␤/␦ formed nestin-expressing neuronal progenitors more readily, they displayed significantly lower expression of the later neuronal markers ␤3-tubulin (Fig. 4a). Immunoblot analysis showed further that down-regulation of either FABP5 or PPAR␤/␦ resulted in decreased expression of both ␤3-tubulin (Fig. 4b) and the mature neuronal marker NeuN (Fig. 4c). These observations suggest that the FABP5/ PPAR␤/␦ path, which inhibits early events in the differentiation process, may be necessary for enabling neuronal maturation.
Notably, although it is well established that RA triggers neuronal differentiation, it is not clear whether the presence of the hormone is necessary throughout the differentiation program.  Standard protocols prescribe removal of RA on day 4 but retain vitamin A in cell media throughout differentiation. Specifically, following removal of RA, cells are usually cultured in medium supplemented with 10% FBS, which contains retinol, followed by culturing in neurobasal medium with B27 supplement that includes 0.35 M retinol and 0.3 M retinyl acetate (Invitrogen). To examine whether retinoids are necessary throughout the differentiation process, cells were cultured in retinoid-depleted (charcoal-treated) medium post day 4. Depletion of vitamin A suppressed the appearance of MAP2 on day 12 (Fig. 4, d and e) and supplementation of media with either retinol and retinyl acetate at concentrations corresponding to the standard B27 supplement or with RA reversed the inhibition and recovered MAP2 expression (Fig. 4, d and e). Similarly to RA, supplementation of retinoid-depleted media with the PPAR␤/␦ ligand GW0742 also recovered MAP2 expression (Fig. 4, d and e). The data thus show that RA is necessary throughout the differentiation program and suggest that activities of the hormone in late stages of differentiation are mediated by PPAR␤/␦.
Neurogenesis Inhibitors SIRT1 and Ajuba Are Direct PPAR␤/␦ Target Genes-To gain insight into the mechanism by which FABP5 and PPAR␤/␦ suppress the ability of RA to support early events in neuronal differentiation, we considered two proteins that were reported to inhibit the transcriptional activity of RAR in neuronal precursors. One of these is SIRT1, an NAD ϩ -dependent class III histone deacetylase that was shown to inhibit RA-induced differentiation of P19 cells by competing with RAR for the coactivator SKI-interacting pro-tein (25). The other is Ajuba, a member of the Ajuba/Zyxin family of LIM proteins (26), which was shown to inhibit RAinduced P19 cell differentiation by interacting with RAR␣ to repress its transcriptional activity (27). In agreement with these reports, ectopic overexpression of either SIRT1 or Ajuba (Fig.  5a) decreased the expression of nestin at day 4 post-induction of differentiation (Fig. 5b). Note again that, due to low transfection efficiency in these experiments, the data underestimate the magnitude of the effects.
RA induced the expression of both Ajuba and SIRT1 (Fig. 5, c  and d). RA also induced the expression of the PPAR␤/␦ target gene ADFP (28) and the RAR target gene RAR␤ and of PPAR␤/␦ itself (Fig. 5d). The increase in the level of PPAR␤/␦ mRNA was modest but functionally meaningful as demonstrated by the observations that it was accompanied by a significant increase in PPAR␤/␦ protein (Fig. 1c). Similarly to RA, the synthetic PPAR␤/␦-selective agonist GW0742 also induced the expression of Ajuba, SIRT1, ADFP, and PPAR␤/␦ (Fig. 5d,  inset). Hence, the data indicate that RA can signal through both of its pathways in undifferentiated P19 cells and support the existence of a positive feedback loop through which PPAR␤/␦ up-regulates its own expression.
Pretreatment of cells with the protein synthesis inhibitor cycloheximide did not block the ability of RA to up-regulate the expression of ADFP, Ajuba, SIRT1, or PPAR␤/␦ (Fig. 5d). Notably, treatment with cycloheximide alone up-regulated the expression of Ajuba and SIRT1. The ability of cycloheximide to increase the levels of some mRNAs has been previous reported (29 -31). The basis for the phenomenon is not well understood, although it is often ascribed to stabilization of some mRNAs by the compound. Nevertheless, the increase in expression of Ajuba and SIRT1 upon cotreatment with RA and cycloheximide significantly surpassed that observed in the presence of cycloheximide alone (Fig. 5d). Hence, RA-induced, PPAR␤/␦mediated up-regulation of ADFP, Ajuba, SIRT1, and PPAR␤/␦ did not require de novo protein synthesis, indicating that these genes are direct targets for the receptor in P19 cells.
Examination of the promoter of the mSIRT1 gene (NUBIScan Version 2.0) revealed the presence of a DR-1 response element (GGTCCAgAGAGCA) at Ϫ606 bp and a DR-2 sequence (ACAGGAtcACTTGG) at Ϫ1.1 kb upstream from the transcription start site. Chromatin immunoprecipitation (ChiP) assays indicated that neither PPAR␤/␦ nor RXR bound to the DR-1 region (data not shown), but in agreement with the reports that a DR-2 element may function as PPRE in some promoters (32)(33)(34)(35), ChiP assays revealed that both PPAR␤/␦ and RXR associate with this region in P19 cells (Fig. 5, e and f). Functional PPREs, which efficiently recruit both PPAR␤/␦ and RXR, were also found in the promoters of the mAjuba (ACGGGGaACTGGA at Ϫ312 bp) and mPPAR␤/␦ (AGGTCAgAGGACA at Ϫ2.826 kb) genes (Fig. 5, e and f). In accordance, overexpression of either PPAR␤/␦ or FABP5 potentiated the ability of GW0742 to induce the expression of SIRT1 and Ajuba (Fig. 5g).

Enhancement of Neuronal Maturation by PPAR␤/␦ Is Mediated by PDK1
-To understand the mechanism by which activation of PPAR␤/␦ promotes the transition of neuronal progenitors to mature neurons, we considered the well established direct PPAR␤/␦ target gene 3-phosphoinositide-dependent kinase-1 (PDK1) (36). This kinase was recently reported to be closely involved in neuronal differentiation both in cultured cells and in vivo (37,38). The PPAR␤/␦ agonist GW0742 induced the expression of PDK1 (Fig. 6a). In agreement with the involvement of PPAR␤/␦ in regulating PDK1 expression, decreasing the expression of the receptor markedly reduced the expression of the kinase (Fig. 6b), and ectopic overexpression of either PPAR␤/␦ or FABP5 potentiated the ability of GW0742 to induce PDK1 (Fig. 6c).
Attesting to its regulation by RA, PDK1 expression during late phase differentiation was reduced upon removal of retinoids from the media and recovered by replenishing media with either RA precursors or RA itself (Fig. 6, d and e). The involvement of PDK1 in promotion of neuronal maturation by PPAR␤/␦ was then examined. Decreasing the expression of PPAR␤/␦ in P19 cells hampered neuronal maturation, reflected by lower expression of ␤3-tubulin. Notably, ectopic expression of PDK1 in these cells rescued the inhibition and allowed maturation to proceed (Fig. 6f). The data thus indicate that promotion of neuronal maturation by PPAR␤/␦ path is mediated, at least in part, by PDK1.

Ablation of FABP5 Results in Excess Neuronal Progenitor Cells and in Deficit in Mature Neurons in Vivo-
The localization of FABP5 in C57BL/6 mouse brain was examined by immunofluorescence microscopy (Fig. 7a) and by immunoblots of extracts of dissected brain regions (Fig. 7b). The protein was found to be expressed in the hippocampus, thalamus, hypothalamus, cerebral cortex, and brain stem and to display low expression in the olfactory bulb and cerebellum. Hippocampi of WT and FABP5 Ϫ/Ϫ mice were then compared to examine whether the lack of the binding protein affects neuronal differentiation in vivo. In agreement with the observations that FABP5 inhibits the formation of neuronal progenitor cells in culture, hippocampi of FABP5 Ϫ/Ϫ mice expressed higher levels of the neural progenitor markers nestin and SOX2 as compared with WT mice (Fig. 7, c and d). Furthermore, in agreement with the observations that the FABP5/PPAR␤/␦ path promotes neuronal maturation, histological and biochemical analyses (Fig. 7e and Fig. 8, a-c) showed that expression of the mature neuronal markers MAP2 and NeuN are significantly lower in hippocampi of FABP5 Ϫ/Ϫ versus WT mice. In contrast, expression of the non-neuronal marker glial fibrillary acidic protein is not altered in FABP5 Ϫ/Ϫ hippocampi (Fig. 8, a and d). Hence, FABP5 inhibits progenitor formation and enhances neuronal maturation in vivo. Hippocampi of FABP5 Ϫ/Ϫ mice also showed decreased expression of PDK1 (Fig. 8, a, e, and f), further supporting the identification of PDK1 as a FABP5/ PPAR␤/␦ target gene in neurons.

DISCUSSION
RA potently induces neuronal differentiation, but the mechanism through which it exerts this activity and the contributions of the two nuclear receptors that are activated by the hormone, RAR and PPAR␤/␦, to the process are incompletely understood. The partitioning of RA between RAR and PPAR␤/␦ in cells is controlled by the relative expression levels of their respective cognate intracellular lipid-binding proteins, CRABP-II and FABP5. Consequently, multiple biological activities of RA critically depend on maintaining a proper balance between these intracellular lipid-binding proteins. For example, it was reported that shifts in RA signaling, brought about by alterations in the CRABP-II/FABP5 ratio, underlie the ability of the hormone to inhibit the growth of some cells while enhancing proliferation and survival in others (7,8). It was also demonstrated that a shift in the CRABP-II/FABP5 ratio is critical for regulation of adipocyte differentiation by RA (6,39) and that interplay between its two paths underlie the ability of RA to regulate energy homeostasis and insulin responses (5,39). The findings of this study demonstrate that coordinated alterations in the ratio of the RA-binding proteins and the ensuing shifts in activation of RA nuclear receptors are also critical for neurogenesis.
The observations show that RA-induced neurogenesis requires contributions from both the CRABP-II/RAR and the FABP5/PPAR␤/␦ paths, but that the two pathways are differentially employed in different stages of the process. The data indicate that early steps in RA-induced neurogenesis, entailing differentiation of stem cells into neuronal progenitors, are driven by CRABP-II and RAR and can be inhibited by the FABP5/PPAR␤/␦ path. The data further show that inhibition of the commitment of stem cells to the neuronal lineage by the FABP5/PPAR␤/␦ path is mediated by the direct PPAR␤/␦ target genes Ajuba and SIRT1, known to suppress the transcriptional activity of RAR (25,26). In contrast with their inhibitory activity in this early stage, FABP5 and PPAR␤/␦ promote the completion of neurogenesis by supporting differentiation of neuronal progenitor cells to mature neurons. The data indicate that this activity is mediated to a large extent by PDK1, a well established direct PPAR␤/␦ target gene (36) previously reported to be involved in neuronal differentiation (37,38).
The model that emerges from the observations (Fig. 9) suggests that neuronal differentiation critically relies on shifts in RA signaling that minimize activation of PPAR␤/␦ in early stages but allow activation of this pathway in late stages of neurogenesis. The shift is accomplished by a transient up-regulation of RAR␤ and CRABP-II and down-regulation of FABP5 immediately following differentiation induction. Hence, in undifferentiated P19 cells, the CRABP-II/FABP5 ratio is low, enabling RA to activate both RAR and PPAR␤/␦. Following induction of differentiation, the CRABP-II/FABP5 ratio rapidly rises, effectively blocking RA signaling through PPAR␤/␦ and   of neuronal differentiation by RA, but they did not address a potential role of the third RA-binding protein CRABP-I in the process. It has been reported in regard to this that CRABP-I expression is induced by RA in P19 cells (40). Although its function in these cells remain to be examined, the observations that CRABP-I expression in F9 teratocarcinoma cells decreases the transcriptional activity of RA (15,16) suggest that its induction during RA-induced neuronal differentiation may serve to modulate the process.
Similarly to the localization of FABP5 in monkey brain (41), FABP5 is expressed in most regions of the mouse brain, including the hippocampus and cerebral cortex, and displays a low expression level in cerebellum and olfactory bulb. In agreement with the conclusions that the FABP5/PPAR␤/␦ path inhibits stem cell differentiation into neuronal progenitor cells but is necessary for differentiation of these cells to neurons, hippocampi of FABP-null mice were found to contain a higher complement of neuronal progenitor cells but fewer mature neurons. Ongoing studies aim to delineate the functional consequences of these alterations.