Sodium Phenylbutyrate Enhances Astrocytic Neurotrophin Synthesis via Protein Kinase C (PKC)-mediated Activation of cAMP-response Element-binding Protein (CREB)

Background: Increase in neurotrophic factors in the brain is a possible therapeutic approach for different neurodegenerative disorders. Results: Sodium phenylbutyrate, an FDA-approved drug for hyperammonemia, increases neurotrophic factors in brain cells via the PKC-CREB pathway. Conclusion: These results delineate a novel neurotrophic property of sodium phenylbutyrate. Significance: Sodium phenylbutyrate may be of therapeutic benefit in neurodegenerative disorders. Neurotrophins, such as brain-derived neurotrophic factor (BDNF) and neurotrophin-3 (NT-3), are believed to be genuine molecular mediators of neuronal growth and homeostatic synapse activity. However, levels of these neurotrophic factors decrease in different brain regions of patients with Alzheimer disease (AD). Induction of astrocytic neurotrophin synthesis is a poorly understood phenomenon but represents a plausible therapeutic target because neuronal neurotrophin production is aberrant in AD and other neurodegenerative diseases. Here, we delineate that sodium phenylbutyrate (NaPB), a Food and Drug Administration-approved oral medication for hyperammonemia, induces astrocytic BDNF and NT-3 expression via the protein kinase C (PKC)-cAMP-response element-binding protein (CREB) pathway. NaPB treatment increased the direct association between PKC and CREB followed by phosphorylation of CREB (Ser133) and induction of DNA binding and transcriptional activation of CREB. Up-regulation of markers for synaptic function and plasticity in cultured hippocampal neurons by NaPB-treated astroglial supernatants and its abrogation by anti-TrkB blocking antibody suggest that NaPB-induced astroglial neurotrophins are functionally active. Moreover, oral administration of NaPB increased the levels of BDNF and NT-3 in the CNS and improved spatial learning and memory in a mouse model of AD. Our results highlight a novel neurotrophic property of NaPB that may be used to augment neurotrophins in the CNS and improve synaptic function in disease states such as AD.

Neurotrophins are a class of small, dimeric growth factors essential for the development, maintenance, and function of the vertebrate nervous system (1)(2)(3)(4). They are released in an activity-dependent manner from neurons, the major producers of neurotrophins (5)(6)(7) and from glial cells, where they are up-regulated under pathophysiological conditions (8,9). Neurotrophin knock-out animals are not viable, and alterations in neurotrophin levels have profound effects on regeneration, long term potentiation, and synaptic remodeling (10,11). In conditions such as Alzheimer disease (AD), 3 the most common progressive dementing disorder, trophic levels are reduced (12)(13)(14), probably contributing to characteristic synaptic loss, neuronal deterioration, and subsequent impairments in learning, memory, planning, and executive functioning (15)(16)(17).
To date, neurotrophin-based therapies for neurodegenerative diseases have produced variable results (18 -20) and are largely invasive. Because exogenous BDNF does not cross the blood-brain barrier without the aid of saturable transport systems (21) and intracerebroventricular injections of neurotrophins have many limitations (22), it is important to identify blood-brain barrier-permeable compounds that can induce neurotrophin production within the CNS. However, few compounds capable of directly modulating CNS neurotrophin production exist (20).
Here, we describe the ability of sodium phenylbutyrate (NaPB), a Food and Drug Administration-approved drug used to treat urea cycle disorders, to induce BDNF and NT-3 in cultured astrocytes and in a mouse model of AD via direct interaction between protein kinase C (PKC) and CREB. NaPB induced the activation of CREB, the proposed "final common pathway" of neurotrophin synthesis (23). In agreement with others (24), we found that NaPB readily crossed the blood-brain transgenic mice were used for experimentation. Animals were maintained, and experiments were conducted in accordance with National Institutes of Health guidelines and were approved by the Rush University Medical Center Institutional Animal Care and Use Committee.
Behavioral Analyses-Five-month-old 5XFAD mice or their non-transgenic littermates were fed 100 mg of NaPB per kg of body weight every day for 30 days, followed by testing on open field and Barnes maze tasks. For the open field test, mice were placed in the Digiscan Optical Animal Monitoring System (AccuScan Instruments) and allowed to explore the chamber for 30 min. Gross distance traveled was recorded every 60 s. Barnes maze testing was performed using a standard elevated platform (92-cm diameter) consisting of 20 5-cm (diameter) holes spaced 7.5 cm apart, one of which was the target hole (26). Visual cues were maintained beyond the perimeter of the maze for the duration of testing, and light was used as a weak aversive stimulus. For the spatial acquisition phase, animals were allowed to explore the maze for 180 s for 4 days with an intertrial interval of 24 h. On the fifth day, animals were not tested. Forty-eight h after the fourth spatial acquisition trial, the probe trial was conducted. Again, mice were allowed to explore the maze for 180 s and were monitored for primary latency (duration before the first encounter with the target hole), total latency (duration before all four paws were on the floor of the escape box), primary errors (incorrect responses prior to the first encounter with the target hole), total errors (incorrect responses before all four paws were on the floor of the escape box), primary time spent in the target quadrant (duration before the first encounter with the target hole), and total time spent in the target quadrant (duration before all four paws were on the floor of the escape box).
Cell Culture and Transfection-Astrocytes and neurons were obtained from postnatal day 7 and fetal (embryonic days 16 -18) mice, respectively. The optimized protocol for these procedures has been described by us (27). Hippocampal or cortical neurons were plated on poly-D-lysine (Sigma)-coated chamber slides for 5 min to allow neurons to adhere. The gliaenriched supernatant was aspirated and pelleted, and astrocytes were seeded in flasks for 12 days prior to experimentation. Neurons were seeded for varying time periods, and extended culture was achieved by replacing half of the culture volume every 3 days. Transfection of 0.25 g of CREB or scrambled siRNA was performed using Lipofectamine 2000 and siPORT NeoFX (Invitrogen) according to the manufacturer's instruction. After 48 h, cells were washed with Hanks' balanced salt solution and treated.
Isolation of Primary Human Astrocytes-Primary human astroglia were prepared as described (28,29). All experimental protocols were reviewed and approved by the Institutional Review Board of the Rush University Medical Center. Briefly, 11-17-week-old fetal brains obtained from the Human Embryology Laboratory (University of Washington, Seattle, WA) were dissociated by trituration and trypsinization. On the ninth day, mixed glial cultures were placed on a rotary shaker at 240 rpm at 37°C for 2 h to remove loosely attached microglia. On the 11th day, the flasks were shaken again at 190 rpm at 37°C for 18 h to remove oligodendroglia. The remaining cells were primarily astrocytes. These cells were trypsinized and subcultured in complete medium at 37°C with 5% CO 2 .
Immunocytochemistry-Slides plated at 50 -60% (astrocytes) or 20 -30% (neurons) confluence were fixed overnight with chilled methanol, blocked with 2% bovine serum for 30 min, and incubated with primary antibodies (see supplemental  Table 1) in PBST plus Triton X-100 (0.1%) for 2 h. Samples were then incubated with Alexa-fluor-tagged secondary antibodies (1:200) for 1 h, washed, and treated with DAPI (1:10,000; Sigma) for 4 -5 min. For negative controls, a slide was incubated under similar conditions void of primary antibodies. Slides were dehydrated in an EtOH and xylene gradient, mounted, and imaged using an Olympus BX41 fluorescent microscope equipped with a Hamamatsu ORCA-03G camera.
Immunohistochemistry-Following treatments, animals were sacrificed, and brains were fixed and embedded. 30-m sections were cut on a cryostat and placed in sucrose cryoprotectant at Ϫ20°C until use. Endogenous peroxidase inhibition was performed using 0.03% H 2 O 2 and 0.1% NaH 3 in PBS. Free floating sections were blocked (2% BSA); incubated with antibodies against BDNF, NT-3, and glial fibrillary acidic protein (GFAP) (see supplemental Table 1) in PBST plus Triton X-100 (0.1%) at 4°C overnight; washed (PBS); incubated with Alexafluor-tagged secondary antibodies (1:200) for 4 h; washed again; and treated with DAPI (1:10,000) for 5 min. Sections were then placed on slides, air-dried overnight, dehydrated in an EtOH/ xylene gradient, mounted with Cytoseal 60 (Richard-Allan Scientific), and dried for Ͼ3 h prior to imaging with an Olympus BX41 fluorescent microscope equipped with a Hamamatsu ORCA-03G camera.
Morphometric Analysis-To characterize and quantify astrocytes, the Cy3 (GFAP) channel was isolated from each image, converted to black and white, and normalized using a spatial threshold filter. 20 -40 cells in four serial sections (n ϭ 5, all groups) were traced using the MicroSuite FIVE TM Biological Suite. Based on area distribution, astrocytes were classified as small (less than quartile 1), medium (between quartile 1 and quartile 3), or large (more than quartile 3). The isolated Cy5 (BDNF) channel was then remerged with the same Cy3 (GFAP) images, and the proportion of BDNF-immunoreactive, GFAPpositive astrocytes was quantified for each quartile.
ELISA and Kinase Assays-BDNF and NT-3 were assayed using high sensitivity sandwich ELISA kits from Promega and Syd Labs, respectively, according to the manufacturers' recommendations. Plates were analyzed spectrophotometrically with a Thermo-Fisher Multiskan MCC plate reader. Assays for protein kinase A (PKA) and PKC were purchased from Enzo Life Sciences and performed according to the manufacturer's protocol using 20 l of supernatant from astrocytes cultured in 24-well plates. Values were normalized to purified kinase controls.
HPLC Analysis of NaPA-Mice were sacrificed by cervical dislocation after 30 days of NaPB treatment. Cortical and hippocampal sections were isolated, immediately frozen in dry ice, and stored at Ϫ80°C. On the day of the analysis, tissues were sonicated using an electronic pulse homogenizer in 0.2 M perchloric acid containing isoproterenol. Tissue homogenates were centrifuged at 20,000 ϫ g for 15 min at 4°C. After pH adjustment and filtration, 15 l of supernatant was injected onto an Eicompak SC-3ODS column (Complete Stand-Alone HPLC-ECD System EiCOMHTEC-500 from JM Science Inc., Grand Island, NY) and analyzed following the manufacturer's protocol.
Immunoblotting, Coimmunoprecipitation, and Densitometry-Western blotting was performed as described earlier (27). Briefly, cells and tissue were lysed in radioimmune precipitation assay buffer, and protein concentration was determined via the Bradford method (Bio-Rad). 50 -100 g of protein was electrophoresed at 200 V on 4 -12, 8 -16, or 4 -20% BisTris gels (Invitrogen) and transferred to nitrocellulose membranes (Bio-Rad) using the Thermo-Pierce Fast Semi-Dry Blotter (20 V/12 min). Membranes were blocked with blocking buffer (LI-COR Biosciences), incubated with primary antibodies (supplemental Table 1) overnight at 4°C under shaking conditions, washed, incubated with secondary antibodies (1:10,000; LI-COR Biosciences) for 1 h at room temperature, washed, and visualized with the Odyssey infrared imaging system (LI-COR Biosciences). Coimmunoprecipitations were performed as described before (31) with modification. Cells and tissue were lysed and homogenized in non-denaturing buffer (20 mM Tris-HCl, 137 mM NaCl, 10% glycerol, 1% Triton X-100, and 2 mM EDTA) by rotation at 4°C for 1 h. Lysates were centrifuged, and supernatants were incubated with antibodies overnight (4°C) and subsequently incubated with protein G-agarose beads (Millipore) for 5 h (4°C). Beads were washed three times with lysis buffer, and immunoprecipitated proteins were resolved by SDS-PAGE. Blots were analyzed using ImageJ (National Institutes of Health, Bethesda, MD) and normalized to their respective loading controls. Data are expressed as -fold change relative to control for three or four independent experiments.
In-cell Western Blot-Astrocytes seeded in opaque wall 96-well plates were fixed with 4% formaldehyde, washed for 20 min with PBS, blocked with blocking buffer (LI-COR Biosciences) for 2 h, incubated with primary antibodies in PBST plus 0.1% Triton X-100 for 2 h, washed, incubated with IR-Dye-labeled secondary antibodies (1:15,000; LI-COR Biosciences), and visualized/quantified with the Odyssey infrared imaging system. DNA Binding and Transcriptional Activities of CREB-DNA binding activity of CREB was analyzed from nuclear fractionation by non-radioactive electrophoretic mobility shift assay (EMSA) as described previously by us (27). After stimulation, cells were washed, centrifuged, and resuspended in a low salt buffer. The resulting pellet was digested in a high salt, nuclear envelope lysis buffer, rotated at 4°C for 30 min, and pelleted by centrifugation. The resultant supernatant was assayed for protein concentration via the Bradford method (Bio-Rad), complexed with a mixture of binding buffer (1 M Tris-HCl, 2.5 M KCl, 0.5 M EDTA, 1.0 mM DTT, 10ϫ Tris-glycine-EDTA, 10% glycerol, and 0.1% Triton X-100) and a CREB-specific IR-Dye probe (LI-COR Biosciences) for 20 min at room temperature, and resolved on 6% polyacrylamide-Tris-glycine-EDTA gels. For the supershift assay, we added 2 g of ChIP grade CREB antibody or IgG (Millipore) to the complex 15 min prior to the addition of probe. The shift was visualized using the Odyssey infrared imaging system (LI-COR Biosciences).
Transcriptional activity of CREB was analyzed using the protocol outlined previously by us (27) using pCRE-Luc (a CREBdependent luciferase reporter construct; Stratagene).
Morphological Characterization-Hippocampal neurons plated to 15-20% confluence were immunostained with anti-␤-tubulin (1:1,000; Millipore) and anti-PSD-95 (1:600; Abcam) and analyzed for dendritic morphology. For each treatment condi-tion, half of the culture medium was replaced every third day with fresh medium alone, medium from astrocytes grown in Neurobasal, or with medium supplemented with an anti-TrkB blocking antibody or recombinant human BDNF (rhBDNF). Every sixth day, the full culture volume was replaced. Supernatant BDNF and NT-3 levels were determined via ELISA and administered in accordance with the results shown in Fig. 1, A and C. Whole cell and dendritic regions were captured using ϫ40 and ϫ100 objective lenses, respectively, using identical contrast, gain, and exposure parameters for each treatment condition. Fluorophores were excited at 488 nm (FITC; PSD-95) and 633 nm (Cy5; ␤-tubulin). Autofluorescence was minimized by running slides in an EtOH and xylene gradient prior to fixation with fluoromount (Sigma). Images were standardized and pseudocolored to red (␤-tubulin) and white (PSD-95) using a Zeiss LSM Image Browser (Jena, Germany). Spine density was analyzed based on inclusion-exclusion criteria with ImageJ (National Institutes of Health). Minimum (0.2 m) and maximum (0.7 m) diameters for puncta were manually defined after Gaussian smoothing with respect to diffuse, nonspecific PDS-95 fluorescence present in the dendritic shaft. Only puncta within the criteria were quantified, and 50 representative images were taken from 10 -15 neurons.
Statistical Analysis-Unless otherwise indicated, one-way ANOVA followed by Tukey's or Scheffe's post hoc tests were applied for data analyses using SPSS 19. p Ͻ 0.05 was considered statistically significant.

RESULTS
NaPB Up-regulates BDNF and NT-3 mRNA in Pure Astrocytic Cultures-At first, we determined the purity of astroglial preparations. Cultures were double-labeled for GFAP and microtubule-associated protein-2 (MAP-2), a neuronal marker (Fig. 1A), or ionized calcium-binding adaptor molecule-1 (Iba-1), a microglial marker (Fig. 1B). Astrocytes were highly pure (Ն99%) because it was very difficult to find one Iba-1-positive cell among 100 GFAP-positive astrocytes ( NaPB Up-regulates Secreted and Cell-bound BDNF and NT-3-NaPB is known to be ␤-oxidized to NaPA, its bioactive metabolite (32). Therefore, primary mouse astrocytes were stimulated with both NaPB and NaPA. Both NaPB and NaPA significantly increased the production of BDNF (blue) and NT-3 (orange; Fig. 2, A and B) compared with unstimulated cells, as determined by ELISA. Lower concentrations and shorter treatment durations did not significantly elevate release of either neurotrophin. NaFO, similar in structure to NaPB but lacking the benzene ring, did not significantly alter BDNF or NT-3 secretion, suggesting the specificity of the effect. After 24-h exposure to NaPB and NaPA, increases in cell-bound BDNF ( and F) astrocytes, as determined by immunostaining. Immunoblotting for BDNF in cell lysates yields a faint band of ϳ30 kDa and a dark band at ϳ15 kDa (Fig. 2G). Increasing concentrations of NaPB significantly increased expression of the 14 kDa band corresponding to mature BDNF (Fig. 2, I and J), whereas no increases were observed in the ϳ30 kDa band corresponding to precursor BDNF (data not shown). Immunoblotting for NT-3 from cell lysates shows similar high and low molecular weight (MW) species as BDNF (Fig. 2H). The low molecular weight species (ϳ15 kDa), which corresponds to mature NT-3, was significantly elevated by NaPB (Fig. 2, K and L; 0.2 and 0.5 mM), whereas the high molecular weight species (ϳ29 kDa) corresponding to precursor NT-3 was unchanged (data not shown). Again, NaFO did not significantly alter the expression of either neurotrophin.
NaPB-mediated Neurotrophin Production Is Dependent upon CREB-Because neurotrophins have long been thought to be under the transcriptional control of CREB (33,34), we analyzed the effect of NaPB on CREB expression and activation. As analyzed by qPCR (Fig. 3, A and B, top) and RT-PCR (Fig. 3, A and B, bottom), NaPB significantly elevated CREB transcription dose-dependently (Fig. 3A) and time-dependently (Fig. 3B) in primary mouse astrocytes. Immunoblotting for total CREB expression in lysates from mouse astrocytes treated with various concentrations of NaPB also revealed significant increases with 0.1 and 0.2 mM concentrations (Fig. 3C, densitometry,  top). Additionally, analysis of cells treated with 0.2 mM NaPB for brief time periods showed significant increases in phosphorylated (Ser 133 ) CREB by immunoblot (Fig. 3D) and in-cell Western blot (Fig. 3F) without altering total CREB expression, suggesting a biphasic effect of NaPB on CREB. This increase in phosphorylated CREB was largely localized to the nucleus, as monitored by immunostaining (Fig. 3E). EMSA of nuclear extracts from astrocytes treated with NaPB revealed increased DNA-protein interactions in NaPB-treated cells (Fig. 3G). In support of this, astrocytes stimulated with various concentrations of NaPB and analyzed by a luciferase assay show increased reporter activity of CREB (Fig. 3H, blue). NaPB, however, did not significantly affect the transcriptional activity of either acti-vator protein-1 (AP-1; orange) or nuclear factor-B (NF-B; navy) (Fig. 3H). siRNA knockdown of CREB (confirmed in Fig.  3, I and J) abrogated the ability of NaPB to induce BDNF and NT-3 mRNA (Fig. 3, K and L) and protein (Fig. 3, M-O), suggesting an important role of CREB in NaPB-mediated neurotrophin expression.
PKC Phosphorylates CREB to Propagate NaPB-mediated Neurotrophin Synthesis-CREB activity can be regulated by a number of signaling pathways. For example, stimuli that activate PKA and PKC result in CREB phosphorylation (35,36). This prompted us to determine the upstream regulator of the NaPB-mediated CREB activation. At 30 min and 2 h, 0.2 mM NaPB produced ϳ120 and ϳ160% increases (navy), respectively, in active PKC relative to active PKA, and increases in PKA were not significant (Fig. 4A). NaPBtreated mouse astrocyte lysates immunoprecipitated with an anti-PKC antibody and immunoblotted for CREB show time-dependent increases in the PKC-CREB interaction (Fig. 4B). As determined by immunoblotting, increases in FIGURE 2. NaPB up-regulates BDNF and NT-3 in primary mouse and human astrocytes. A, NaPB dose-dependently induces BDNF and NT-3 release. Supernatants from primary mouse astrocytes stimulated with various concentrations of NaPB or 0.2 mM NaFO for 24 h (A) were analyzed for BDNF (blue) and NT-3 (orange) expression via ELISA. B, NaPB time-dependently induces BDNF and NT-3 release. Supernatants from mouse astrocytes stimulated with 0.2 mM NaPB for various time points (B) were analyzed for BDNF (blue) and NT-3 (orange) expression via ELISA. C and D, NaPB elevates cell-bound BDNF. Primary mouse cortical (E) and primary human astrocytes (F) stimulated with 0.2 mM NaPB, NaPA, or NaFO for 24 h were immunostained for BDNF (red), GFAP (green), and DAPI (blue). E and F, NaPB elevates cell-bound NT-3. Primary mouse cortical (G) and primary human astrocytes (H) stimulated with 0.2 mM NaPB, NaPA, or NaFO for 24 h were immunostained for NT-3 (red), GFAP (green), and DAPI (blue). G, anti-BDNF antibody detects two bands. 50 g of lysate from unstimulated astrocytes was resolved next to pure rhBDNF. I, NaPB induces expression of mature BDNF. Primary mouse astrocytes stimulated with various concentrations of NaPB or 0.2 mM NaFO for 24 h were subjected to immunoblotting with anti-BDNF and anti-␤-actin antibodies and subsequent densitometry (J). H, anti-NT-3 antibody detects two bands. 60 g of lysate from unstimulated astrocytes was resolved next to pure recombinant mouse NT-3 (rmNT-3). K, NaPB induces expression of mature NT-3. Primary mouse astrocytes stimulated with various concentrations of NaPB or 0.2 mM NaFO for 24 h were subjected to immunoblotting with anti-NT-3 and anti-␤-actin antibodies and subsequent densitometry (L). Each treatment condition is represented by two independent bands in I and K. ELISAs and densitometric analyses were analyzed for significance by one-way ANOVA (*, p Ͻ 0.01; **, p Ͻ 0.001). Scale bar, 20 m. MRGD, merged image; MW, molecular weight; MPA, mouse primary astrocytes; HPA, human primary astrocytes.
phosphorylated CREB in response to 0.2 mM NaPB were abrogated by the PKC inhibitor GF 109203X (GFX; 0.5 M) but not the PKA inhibitor H-89 (2 M) (Fig. 4C). In support of this, GFX, but not H-89, inhibited the NaPB-mediated increases in the DNA binding (Fig. 4D) and transcriptional activities (Fig. 4E) of CREB. As determined by qPCR, increases in the mRNA expression of CREB (Fig. 4F) and BDNF and NT-3 (Fig. 4G) in response to 0.2 mM NaPB were abated by GFX but not H-89. The mRNA results (Fig. 4G) were supported by Western blot data ( Fig. 4H; densitometry, top). Taken together, these results suggest an obligatory role for PKC in the NaPBmediated induction of BDNF and NT-3.

. PKC interacts with CREB to promote NaPB-induced synthesis of BDNF and NT-3 in primary mouse astrocytes.
A, NaPB induces PKC activity. Cells were stimulated with 0.2 mM NaPB for various times followed by monitoring activities of PKA (orange) and PKC (blue). Differences (⌬; navy) between kinase activations were calculated as a forward difference argument (⌬F(P) ϭ F(P ϩ ⌬P) Ϫ F(P)) ϫ 100 to yield the percentage variance between kinase activation levels. B, NaPB promotes the association between PKC and CREB. Cells treated with 0.2 mM NaPB for various times were lysed, cleared, incubated with an anti-PKC antibody, separated by SDS-PAGE, and probed with an anti-CREB antibody. Blots for PKC verified that equal amounts of protein were loaded. C, PKC is required for NaPB to induce phosphorylation of CREB. Cells were treated with 0.2 mM NaPB alone or after preincubation with inhibitors of PKA (H-89; 2 M) or PKC (GFX; 0.5 M) for 1 h before being subjected to immunoblotting with antibodies against CREB-Ser(P) 133 , CREB, and ␤-actin and subsequent densitometry (top). D, inhibition of PKC abrogates NaPB-induced increases in DNA binding of CREB. Cells treated with 0.2 mM NaPB for 30 min alone or after preincubation with 2 M H-89 or 0.5 M GFX for 1 h were subjected to EMSA. The DNA-bound nuclear extract was incubated with antibodies against CREB or IgG to supershift the complex. E, inhibition of PKC abrogates the NaPB-induced increase in CREB transcriptional activation. Cells were transfected with a luciferase reporter construct driven by CREB, treated for 1 h with various concentrations of NaPB alone or after preincubation with 2 M H-89 or 0.5 M GFX for 1 h, and assayed for transcriptional activation via the luminometry.  MARCH 22, 2013 • VOLUME 288 • NUMBER 12

JOURNAL OF BIOLOGICAL CHEMISTRY 8305
NaPB Does Not Alter Neurotrophin Receptor Expression-Because precursor neurotrophins and mature neurotrophins can bind p75 NTR and trigger apoptotic pathways (37) and upregulation of both neurotrophins and their receptors may contribute to the development of neuroblastomas (38,39), we monitored the expression of p75 NTR and TrkB, the selective BDNF receptor, in neurons. As determined by immunocytochemistry, neurons treated with 0.5 mM NaPB (Fig. 5A2) or NaPA (Fig. 5A3) for 24 h displayed no change in somatic p75 NTR (left panels) or TrkB (right panels) expression relative to control (Fig. 5A1). Receptor expression was equally unaltered in distal dendrites (data not shown). In support of this, neurons exposed to the same treatment conditions were subjected to immunoblotting for p75 NTR and TrkB (Fig. 5B). No significant change was observed in the expression of either receptor, as measured by densitometry (right).

Oral Treatment of NaPB Increases the Interaction between CREB and PKC and Up-regulates BDNF and NT-3 in a Mouse
Model of AD-To determine the effect of NaPB on neurotrophin induction in vivo, we used 5XFAD mice, an accelerated model of AD. At first, we examined if NaPB entered into the CNS of 5XFAD mice. NaPB is known to be ␤-oxidized into NaPA in the liver. Therefore, frontal-cortical homogenates from NaPB-treated mice were analyzed for NaPA by HPLC to confirm the bioavailability of NaPA in the brain (Fig. 7A). NaPA produced a spike just after 2 min in both the standard (top) and NaPB-treated (bottom), but not vehicle-treated (middle), sample. Hippocampal fractions isolated from 5XFAD mice, immunoprecipitated with antibodies against CREB and PKC and immunoblotted for PKC and CREB, respectively, indicate an increased interaction between PKC and CREB in mice fed with NaPB (Fig. 7B), supporting our cell culture finding (Fig. 4B). Immunoblotting for BDNF and NT-3 in cortical (top) and hip- FIGURE 5. NaPB treatment remains unable to modulate the expression of neurotrophin receptors in primary mouse cortical neurons. A, NaPB does not alter p75 NTR or TrkB expression in cortical neurons. Neurons seeded at ϳ20% were unstimulated (A1) or treated with 0.5 mM NaPB (A2) or NaPA (A3) for 24 h; fixed; immunostained with antibodies against ␤-tubulin (green), p75 NTR (left panels; red), and TrkB (right panels; red); and treated with DAPI (blue). Whole cell images and somatic images were captured using ϫ20 and ϫ60 objective lenses, respectively. Scale bar, 50 m. B, NaPB does not alter p75 NTR or TrkB expression. Neurons unstimulated or treated with 0.5 mM NaPB or 0.5 mM NaPA for 24 h were subjected to immunoblotting with antibodies against p75 NTR (blue), TrkB (orange), and ␤-actin and subsequent densitometry (right). Data (mean plus S.D. (error bars)) from three independent experiments were analyzed for statistical significance by one-way ANOVA. Each treatment condition is represented by two independent bands in B. MRGD, merged image; NS, not significant; Ctrl, control. pocampal (bottom) fractions from non-transgenic (non-Tg) (Fig.  7C) and 5XFAD (Fig. 7D) mice shows significant increases in neurotrophin expression in NaPB-treated mice (orange) as compared with vehicle-treated mice (blue). Each band represents independent animals. NaPB treatment also promoted PSD-95 expression in vivo in the hippocampus of 5XFAD mice (Fig. 6, E and F).
NaPB Up-regulates BDNF Independent of Astrocytic Morphology in CA1 of the Hippocampus-To determine colocalization of BDNF with astrocytes in vivo and verify immunoblotting data (Fig. 6), coronal hippocampal sections were double-labeled for GFAP and BDNF (Fig. 8A). The boxed area in CA1 indicates the dorsal anterior region magnified in Fig. 8B. BDNFimmunoreactive astrocytes from this area were then quantified based on rigorous morphological parameters. Administration of NaPB (100 mg of NaPB per kg of body weight every day for 30 days) or vehicle (H 2 O) to non-Tg and 5XFAD mice did not alter GFAP staining intensity (Fig. 8C), astrocyte area (Fig. 8D), or astrocyte size distribution (Fig. 8, E and F). NaPB also did not FIGURE 6. NaPB-treated astroglial supernatant propagates synaptic development. A, PSD-95 expression is increased by supernatant (sup) from primary mouse astrocytes treated with NaPB. Fetal hippocampal neurons were isolated, plated and seeded at ϳ20% confluence for 7 (left; DIV 7), 14 (middle; DIV 14), or 21 (right; DIV 21) days. After 3 days, neurons were either unstimulated (A1) or treated with glial supernatant from unstimulated astrocytes (A2), glial supernatant from NaPB-stimulated astrocytes (A3), glial supernatant from NaPB-stimulated astrocytes plus an anti-TrkB blocking antibody (A4), an anti-TrkB blocking antibody alone (A5), or 10 ng/ml rhBDNF as a positive control (A6). See "Materials and Methods" for culture and treatment details. Cells were then fixed and immunostained with antibodies against MAP-2 (red) and PSD-95 (white). Whole cell images and spine images were captured using ϫ40 and ϫ100 objective lenses, respectively. B, quantification of PSD-95 puncta in A. 50 representative images were taken from 10 to 15 neurons and counted based on exclusion-inclusion criteria (see "Materials and Methods"). Data are expressed as the number of puncta per 25 M dendritic shaft for DIV 7 (B1), DIV 14 (B2), and DIV 21 (B3). Blue, unstimulated; orange, control supernatant; navy, NaPB-treated supernatant; green, NaPB-treated supernatant plus anti-TrkB blocking antibody; black, anti-TrkB blocking antibody; gray, rhBDNF. C, supernatant from astrocytes stimulated with NaPB induces the expression of glutamatergic receptor subunits. Neurons stimulated with supernatant from NaPB-treated astrocytes or 10 ng/ml rhBDNF for 21 days were subjected to immunoblotting with antibodies against PSD-95 (blue), AMPA receptor subunit 1 (GluR1; orange), NMDA receptor subunit 2a (NR2A; navy), and ␤-tubulin, and subsequent densitometry is depicted in D. Data in C and D (mean plus S.D. (error bars)), representative of three independent experiments, were analyzed for statistical significance by one-way ANOVA (*, p Ͻ 0.05; **, p Ͻ 0.01). Each treatment condition is represented by two independent bands in C. sup, supernatant.
alter BDNF immunoreactivity in smaller astrocytes (less than quartile 1), but significantly elevated BDNF expression in medium (quartile 1 Ͻ x Ͼ quartile 3) and large (more than quartile 3) astrocytes in non-Tg and 5XFAD mice relative to vehicle-treated mice (Fig. 8G).
Oral Administration of NaPB Improves Spatial Memory in a Mouse Model of AD-Because neurotrophins are known to modulate mechanisms underlying learning and memory, we tested performance of NaPB-treated 5XFAD and non-Tg mice on the Barnes maze, a hippocampus-dependent cognitive task that requires spatial reference memory (26). NaPB did not alter gross motor activity or exploratory drive in either non-Tg or 5XFAD mice, as determined by open field testing (Fig. 9A). As expected, 5XFAD mice performed poorly on the Barnes maze task as compared with non-Tg mice (Fig. 9, B-D). However, oral treatment with NaPB for 30 days significantly reduced primary (F 3,19 ϭ 3.073, p Ͻ 0.05, p ϭ 0.035) and total (F 3,19 ϭ 9.978, p Ͻ 0.001, p ϭ 0.047) errors (Fig. 9B) and improved primary (F 3,19 ϭ 3.175, p Ͻ 0.05, p ϭ 0.034) latency (Fig. 9C) in 5XFAD mice relative to vehicle treatment. NaPB-treated 5XFAD mice also spent significantly more time in the target quadrant before the first encounter with the target hole (primary; F 3,19 ϭ 3.292, p Ͻ 0.05, p ϭ 0.043) as well as throughout the trial period (total; F 3,19 ϭ 3.989, p Ͻ 0.05, p ϭ 0.029) (Fig.  9D). Similarly, NaPB treatment also improved the performance of non-Tg mice on the Barnes maze as compared with vehicle treatment (Fig. 9, B-D).

DISCUSSION
Neurotrophins are secreted small proteins that are capable of signaling neurons to survive, differentiate, and grow. Furthermore, neurotrophins have also been suggested as rescuers of vulnerable neurons in many neurodegenerative diseases, including AD, Parkinson disease, and HIV-associated dementia, in which the levels of some neurotrophic factors are significantly reduced in the brain. For example, it has been shown that the levels of BDNF and NT-3 are significantly down-regulated in the brains of patients with AD. Accordingly, these neurotrophic factors (BDNF and NT-3) exhibit protective effects in cell culture as well as animal models of different neurodegen- , and gemfibrozil (gem) was used as an internal standard. B, NaPB propagates the interaction between CREB and PKC in vivo. Hippocampal homogenates from 5XFAD mice fed vehicle (blue) or NaPB (orange) were cleared; coimmunoprecipitated with anti-IgG, anti-PKC, or anti-CREB antibodies; immunoblotted with anti-PKC (top) or anti-CREB (bottom) antibodies; and analyzed by densitometry (bottom). Inputs were determined by immunoblotting for the same pull-down. C, NaPB induces BDNF and NT-3 expression in non-Tg mice. Cortical (top three panels) and hippocampal (bottom three panels) homogenates from mice fed vehicle (blue) or NaPB (orange) were subjected to immunoblotting with antibodies against BDNF, NT-3, and ␤-actin and subsequent densitometry (right). Bands correspond to independent animals. D, NaPB induces BDNF and NT-3 expression in 5XFAD mice. Cortical (top three panels) and hippocampal (bottom three panels) homogenates from mice fed vehicle (blue) or NaPB (orange) were subjected to immunoblotting with antibodies against BDNF, NT-3, and ␤-actin and subsequent densitometry (right). Bands correspond to independent animals. E, NaPB up-regulates PSD-95 in 5XFAD hippocampi. Hippocampal homogenates from 5XFAD and non-Tg (data not shown) mice fed vehicle (blue) or NaPB (orange) were subjected to immunoblotting with antibodies against PSD-95 and ␤-tubulin and subsequent densitometry (right). Bands correspond to independent animals. Data (mean plus S.D. (error bars)) representative of three or six independent experiments were analyzed for statistical significance by Student's t tests (*, p Ͻ 0.05; **, p Ͻ 0.01). WB, Western blot; HPC, hippocampus; Veh, vehicle; IP, immunoprecipitation. erative disorders (19). Although under physiological conditions, neurons produce neurotrophic factors, glial derived neurotrophins (44) assume significance under neurodegenerative conditions, when neurons die and do not produce these invaluable factors. Therefore, increasing the levels of neurotrophins in astrocytes, major glial cells in the CNS, is an important area of research. However, such mechanisms are poorly understood.
NaPB is a sodium salt of short chain fatty acid having multiple clinical interests. Known as Buphenyl and triButyrate in the United States or Ammonaps in Sweden, it is used as a medication against urea cycle disorders involving deficiencies of carbamylphosphate synthetase, ornithine transcarbamylase, or argininosuccinic acid synthetase. Although its major function is to scavenge ammonia and glutamine (45), recent studies have described NaPB as a potent inhibitor of HDAC (46). Due to its HDAC-inhibitory effects, it has been proposed as a drug against various forms of cancer (47,48). Accordingly, it has also been clinically tested as an anticancer drug (47). In addition to ammonia removal and HDAC inhibition, Ozcan et al. (49) have shown that NaPB can also function as a chemical chaperon during endoplasmic reticulum stress. Recently, we have also demonstrated anti-inflammatory and antioxidative properties of NaPB (50). Several lines of evidence presented in this study clearly support the conclusion that NaPB and its metabolite NaPA are capable of up-regulating neurotrophins in astrocytes. Our conclusion is based on the following observations. First, NaPB dose-dependently induced the expression of BDNF and NT-3 in primary mouse and human astrocytes. The metabolite of NaPB, NaPA, similarly up-regulated BDNF and NT-3 in astrocytes. This up-regulation was specific because NaFO, a compound structurally similar to NaPA but without having the benzene moiety, had no effect. Second, oral administration of NaPB increased the level of NaPA in the brains of mice and up-regulated BDNF and NT-3 in vivo in the brain.
Because the loss of these neurotrophic factors has been implicated in the pathogenesis of various neurodegenerative diseases (13) and these neurotrophins have been shown to be required for synaptogenesis and synaptic function (10,51,52), our results provide a potentially important mechanism whereby NaPB may improve synaptic function associated with different neurodegenerative conditions. In fact, we found that NaPB-treated astrocytic supernatant up-regulated the expression of PSD-95, NR2A, and GluR1 in cultured hippocampal neurons. Application of rhBDNF produced similar results, whereas blocking TrkB abrogated NaPB-mediated up-regulation of PSD-95 in hippocampal neurons. As expected, we observed decreased PSD-95 and poor memory performance in 5XFAD mice, an aggressive animal model of AD. However, oral administration of NaPB recovered characteristic deficiencies in PSD-95 expression and improved spatial learning and memory.
The signaling events required for the transcription of neurotrophic factors are becoming clear. Upon analysis of the BDNF promoter using the Genomatix Software Suite, we have found multiple binding sites for CREB and NF-B and at least one consensus binding site for C/EBP␤ and AP-1 in the BDNF promoter. Although these transcription factors (CREB, NF-B, C/EBP␤, and AP-1) may play a role in the expression of various trophic factors, activation of CREB seems essential for the transcription of these neurotrophic factors (23). Therefore, for a drug to exhibit neurotrophic effect, it is almost mandatory to stimulate the activation of CREB. Activation of CREB in cultured astrocytes by NaPB and abrogation of the ability of NaPB to induce neurotrophic factors by siRNA knockdown of CREB suggest that NaPB increases the levels of neurotrophic factors via CREB. Although the activation of NF-B has also been shown to be involved in the expression of neurotrophic factors, NaPB alone does not induce the activation of NF-B. Furthermore, recently, we have found that NaPB inhibits the activation  , 250 m). B, NaPB induces BDNF expression in mouse astrocytes in vivo. Free floating coronal sections from non-Tg mice fed vehicle (B1) or NaPB (B2) and 5XFAD mice fed vehicle (B3) or NaPB (B4) were immunostained for BDNF (red), GFAP (green), and DAPI (blue). Images represent the boxed region of CA1 in A (scale bar, 25 m). C-F, NaPB does not significantly alter astrocyte morphology. Three serial sections within CA1 from non-Tg mice fed vehicle (blue) or NaPB (orange) and 5XFAD mice fed vehicle (navy) or NaPB (green) (n ϭ 4, all groups) were analyzed for GFAP fluorescence intensity (C) and area (D) using MicroSuite FIVE TM Biological Suite. Box plots in D represent area quartile (quartile 1 (Q1), median, quartile 3 (Q3)) dispersion, whiskers represent area maximums/minimums, and white insets represent area mean Ϯ S.E. (error bars) for each group. Representative astrocytes, tracings, and areas from each quartile are represented in E. Astrocytes within each quartile range were quantified as a percentage of total cells counted in each group and presented as a 100% stacked column chart in F. G, NaPB induces expression of BDNF in astrocytes of various sizes. BDNFimmunoreactive (BDNFϩ) astrocytes were counted and quantified as a percentage of GFAP-immunoreactive (GFAPϩ) astrocytes for all quartiles. See "Materials and Methods" for detailed stereological morphometry information. Data were analyzed for statistical significance by one-way ANOVA (*, p Ͻ 0.05; **, p Ͻ 0.01). CA; cornu ammonis; DG, dentate gyrus; SO, stratum oriens; SP, stratum pyrimidale; SR, stratum radiatum.
of NF-B in activated glial cells (50). Therefore, NaPB does not up-regulate the expression of neurotrophic factors via NF-B.
However, it was unknown by which mechanisms NaPB induced the activation of CREB in brain cells. The PKC pathway, one of the most common and versatile signal pathways in eukaryotic cells, is involved in the regulation of multiple cellular functions, including cell growth, differentiation, and synaptic plasticity (53,54). Here we present evidence that NaPB induced the activation of CREB via PKC. Although both PKA and PKC are known to phosphorylate and activate CREB, NaPB specifically induced the activation of PKC and direct association between PKC and CREB. Accordingly, suppression of PKC by GF 109203X, but not of PKA by H-89, inhibited NaPB-induced activation of CREB and expression of neurotrophic factors in astrocytes. However, at present, we do not know the mechanisms by which NaPB induces PKC. In general, the PKC signaling pathway is transduced in different cell types, including brain cells, by diacylglycerol and/or Ca 2ϩ (55,56). Because translocation of PKC to the cell membrane is considered a hallmark of PKC activation and PKC interacts weakly or transiently with membranes in the absence of Ca 2ϩ or diacylglycerol, it is possible that NaPB may alter cytosolic Ca 2ϩ levels.
There are several advantages of NaPB over other proposed neurotrophic therapies. First, NaPB is fairly nontoxic. It is a Food and Drug Administration-approved drug against urea cycle disorders in children. It is rapidly metabolized in the liver and kidneys via ␤-oxidation to phenylacetate, the bioactive form of the drug (32). NaPB and its metabolite have been found to exhibit protection in animal models of multiple sclerosis (57), ischemic injury (58), and Parkinson disease (50) and to reduce Tau pathology in an AD mouse model (59). Second, NaPB can be taken orally, the least painful route. Third, NaPB, being a lipophilic molecule, readily diffuses across the bloodbrain barrier. For example, glutamine toxicity is a problem in urea cycle disorders. After treatment of patients with urea cycle disorders, NaPB is known to combine with glutamine to produce phenylacetylglutamine, a compound that is readily excreted in the urine. Simultaneous serum and CSF sampling in those patients showed comparable levels of phenylacetylglutamine in the CSF (32). Here we have also seen that oral administration of NaPB increases the level of NaPA, its active metabolite, in the brains of 5XFAD mice.
In summary, we have demonstrated that NaPB up-regulates neurotrophic factors in astrocytes via PKC-mediated activation of CREB. These results highlight undiscovered properties of NaPB and indicate that this Food and Drug Administrationapproved drug may be used to improve synaptic plasticity in neurodegenerative disorders as primary or adjunct therapy.