Structure-based artificial screening with computer-based molecular docking simulations was conducted by InnoPharmaScreen Inc. (Asan, Korea). The X-ray crystal structure of the HAT domain of human PCAF bound to coenzyme A (CoA; Protein Data Bank ID: 1CM0) was used for the initial structure of the target protein. CoA was removed and atomic hydrogens were added to produce a complete molecular model of the target protein for use in the molecular docking simulations. Docking simulations were performed between the target model and the ChemBridge chemical library that consists of 450,000 unique compounds. Binding efficiency was based on the binding force between the target molecule and the test compound as calculated by the Gold 4.0.1 software package (Cambridge Crystallographic Data Centre, Cambridge, UK). The top 10 docking positions for each compound were calculated, and the library search efficiency was 100%.

Murine BV-2 cells and Neuro-2A cells were obtained from the American Type Culture Collection (ATCC; Manassas, VA, USA; CRL no. 2270). Fetal bovine serum (FBS), trypsin-EDTA, penicillin and streptomycin were purchased from Gibco-BRL (Gaithersburg, MD, USA). 3-(4,5-Dimethylthiazol-2-ly)-2,5-diphenyl tetrazolium bromide (MTT) was purchased from Sigma-Aldrich (St. Louis, MO, USA). Aβ_1–42 was purchased from Bachem (Bubendorf, Switzerland). Other chemicals used were purchased from Sigma-Aldrich. BV-2 cells were cultured in Dulbecco’s modified Eagle’s medium (Gibco-BRL) containing 5% heat-inactivated endotoxin-free FBS, 2 mM glutamine, 100 μg/ml streptomycin and 100 U/ml penicillin in a humidified 5% CO₂ atmosphere at 37°C. Neuro-2A cells were cultured in modified Eagle’s medium (Gibco-BRL) containing 10% heat-inactivated endotoxin-free FBS, 2 mM glutamine, 100 μg/ml streptomycin, and 100 U/ml penicillin in a humidified 5% CO₂ atmosphere at 37°C. Aβ peptides were dissolved in phosphate-buffered saline (PBS) and pre-incubated at 37°C for 5 days to allow for fibril formation. The peptides were stored at -20°C until use. Compound C-11 was dissolved in dimethyl sulfoxide (DMSO) and later diluted with distilled water (the final concentration of DMSO was <0.5%). For the co-culture experiments, BV-2 cells were treated with various concentrations of C-11 (0.1–1.0 μM final concentration) for 12 h prior to stimulation with 5 μM aggregated Aβ_1–42 (the Aβ + C-11 group), 5 μM non-aggregated Aβ_1–42, or medium only (control) for 24 h. Conditioned medium (CM) from the BV-2 cells and primary microglial cells was collected, centrifuged and transferred to the Neuro-2A cells for a further 24 h. CM from the medium-only treated cells was used as the control. Following incubation, cell viability was measured by MTT assay and western blot analysis was performed.

HeLa cell nuclear extract was prepared as previously described (26). HAT and HDAC activity assays were performed with nuclear extracts using commercially available kits according to the manufacturer’s instructions (BioVision, Inc., Milpitas, CA, USA). For in vitro HAT activity assays, recombinant HAT proteins were incubated with HAT assay buffer [50 mM HEPES pH 8.0, 10% glycerol, 1 mM DTT, 1 mM phenylmethanesulfonyl fluoride (PMSF), 10 mM sodium butyrate and 1 μl [³H]-acetyl-CoA], 5 μg histone H4 tail peptides, and/or indicated concentration of inhibitors at 30°C for 1 h. Samples were separated on a 15% sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) gel and analyzed by autoradiography. HMT activity assays were performed using the assay buffer and core histones from the HMT Assay Reagent kit (Upstate Biotechnology, Inc., Lake Placid, NY, USA) according to the manufacturer’s instructions. Core histones were incubated for 1 h at 30°C in methyltransferase buffer (final concentration 50 mM Tris pH 8.0, 1 mM PMSF, and 0.5 mM DTT), 1 μl (0.55 μCi) S-adenosyl-L-[methyl-³H]-methionine (Amersham Pharmacia Biotech, Inc., Piscataway, NJ, USA) and 10 μg/ml HeLa nuclear extract. The total volume of the reaction mixture was 30 μl. The reaction was spotted on P-81 paper for scintillation counting. The P-81 paper was washed 3 times with 10% trichloroacetic acid (Upstate Biotechnology, Inc.) for 15 min, washed with 95% ethanol for 5 min at room temperature and allowed to dry. Dry P-81 papers were counted with a multi-purpose LS 6500 scintillation counter (Beckman Coulter, Indianapolis, IN, USA). SIRT1 activity was assayed using a SIRT1/Sir2 Deacetylase Fluorometric assay kit (CycLex Co., Ltd., Nagano, Japan) according to the manufacturer’s instructions.

For silencing PCAF expression, 2 pairs of oligonucleotides that encoded PCAF-specific shRNA were purchased from MISSION shRNA (Sigma-Aldrich). Lentiviral particles were prepared using pLKO.1-PURO PCAF shRNA via a 3 plasmid co-transfection according to the instructions of the manufacturer (Invitrogen, Carlsbad, CA, USA). The BV-2 cells were transfected with the lentivirus. After 3 days of incubation, the lentivirus from the culture medium was collected and concentrated in a Centricon-plus-20 centrifugal filter device (Millipore, Billerica, MA, USA). Lentivirus PURO shRNA was generated as the control.

Cell viability was measured to determine the cytotoxicity of the Aβ peptides on neuronal cells. Neuro-2A cells were seeded at 1×10⁴ to 1×10⁵ cells in a 96-well plate. Following 12 h of incubation, the medium was replaced with CM from the BV-2 cell cultures treated with or without Aβ and different concentrations of C-11. After 24 h, the CM was replaced with non-CM, 15 μl MTT solution (2 mg/ml final concentration) was added for 90 min at 37°C, and the absorbance was recorded at 570 nm. A reference was recorded at 630 nm using a microplate reader (Model 550; Bio-Rad, Hercules, CA, USA).

The treated cells were washed with cold PBS, scraped off the culture dishes and harvested. The cells were incubated for 20 min in lysis buffer containing 0.5% Triton X-100, 20 mM HEPES (pH 7.4), 150 mM NaCl, 2 mM DTT and 1 mM PMSF. The lysates were centrifuged at 20,000 × g for 10 min at 4°C. The protein concentrations in the clarified lysates were determined with the Bradford assay using bovine serum albumin as a reference. Total cell lysate proteins were separated on 8 or 12% SDS-PAGE gels and transferred onto nitrocellulose membranes. The membranes were blocked by incubating for 12 h in 5% (w/v) non-fat Difco™ skim milk blocking buffer. The blocked membranes were incubated overnight at 4°C with primary antibodies that recognize iNOS (1:1,000 dilution), COX-2 (1:1,000), IL-1β (1:500), NF-κB (p65; 1:500) or β-actin (1:5,000). The antibody against acetyl-NF-κB (K122) was raised against the synthetic peptide, CIHSFQNLGIQCV_ACKKRDLE, by GenScript (Piscataway, NJ, USA). After washing 3 times with PBS and 0.1% Tween-20, the membranes were incubated with secondary horseradish peroxidase-conjugated antibody (1:1,000) for 1 h. The bands were detected using the enhanced chemiluminescence system (Amersham Pharmacia Biotech, Inc.) according to the manufacturer’s instructions.

Total RNA from the Neuro-2A, BV-2 and primary microglial cells was extracted using TRIzol reagent (Invitrogen) according to the manufacturer’s instructions. The mRNA levels of iNOS, COX-2 and IL-1β were determined by real-time PCR (ABI PRISM 500 Sequence Detection System, Applied Biosystems, San Jose, CA, USA). cDNA amplification was performed in duplicate in 20-μl reaction mixtures containing 2X SYBR-Green Master Mix (Roche, Indianapolis, IN, USA) and 10 pM forward and reverse primers. The initial denaturation step was 5 min at 95°C followed by 40 amplification cycles of 30 sec at 95°C, 30 sec at 58°C and 30 sec at 72°C with a final 10 min extension at 72°C. The results were analyzed using ABI sequence detector software version 2.3. Relative mRNA expression of the target genes was calculated after normalizing to GAPDH expression and was expressed as the fold induction. The primers used for amplification were as follows: human IL-1β gene forward, 5′-GTTGACGGACCCCAAAAGAT-3′ and reverse, 5′-AAGG TCCCACGGGAAAGACAC-3′; human COX-2 gene forward, 5′-GAGTGGGAGGCACTTGCATT-3′ and reverse, 5′-TGGA GGCGAAGTGGGTTTTA-3′; human iNOS gene forward, 5′-TCTTGGAGCGAGTTGTGGAT-3′ and reverse, 5′-GGGT GGTAATGTCCAGGAAGT-3′; and human GAPDH gene forward, 5′-GTGTTCCTACCCCCAATGTGT-3′ and reverse, 5′-AGGAGACAACCTGGTCCTCAGT-3′. All reactions were performed in triplicate. Relative expression levels and standard deviations (SDs) were calculated using the comparative value method.

To develop PCAF-specific inhibitors, we performed structure-based artificial screening using computer-based molecular docking simulations (Fig. 1A). We initially selected 50 compounds with a binding force of at least 70 after the molecular docking simulation (Table I). Among these, compound number 11 (C-11) displayed the most significant inhibitory effect against the PCAF enzyme (Fig. 1B). Furthermore, C-11 selectively inhibited PCAF, but not p300 and GCN5 (Fig. 1C). Thus, we selected C-11 as the putative PCAF inhibitor for our experiments.

To examine the enzyme specificity of C-11, we examined whether C-11 inhibits other epigenetic enzymes, including HDAC, HMT and SIRT1. As shown in Fig. 2A, C-11 failed to inhibit the enzyme activities of HDAC and SIRT1. Furthermore, C-11 had no effect on HMT activity (Fig. 2B). These results demonstrate that C-11 specifically inhibits PCAF acetyltransferase activity but not the activity of other epigenetic enzymes. As several HAT inhibitors (HATi) have recently been shown to possess anti-PCAF activity (12,23,27), we compared the relative anti-PCAF activity of C-11 with other HATi. As shown in Fig. 2C, C-11 had the highest anti-PCAF activity among the reported HATi with a half-maximal inhibitory concentration of approximately 0.25 μM (Fig. 2C and D).

Since PCAF was found to activate NF-κB-mediated transcription in response to cytokines (6), we first examined whether the inhibition of PCAF by C-11 suppresses NF-κB activity by inhibiting the acetylation of NF-κB. Since PCAF is known to acetylate NF-κB at Lys-122 (6), we first generated an acetyl-specific NF-κB antibody that recognizes acetylated NF-κB at Lys-122. Polyclonal antibodies were generated against the NF-κB acetyl-peptide, ¹⁰⁹CIHSFQNLGIQCV_ACKKRDLE¹²⁷, and the antisera were examined by western blot analysis following in vitro acetylation reactions with recombinant PCAF enzyme and GST-p65 substrate (Fig. 3A, upper panel). A peptide competition assay clearly demonstrated the specificity of the NF-κB Lys(Ac)-122 antibody, as the complete blocking of NF-κB acetylation was observed by the antibody raised against the acetyl-peptide, but not by an antibody raised against a non-acetyl-peptide (Fig. 3A, lower panel). In addition, non-acetyl-NF-κB antibody failed to detect the acetylation that resulted from an in vitro HAT reaction and thus confirmed the specificity of the acetyl-specific NF-κB antibody for the PCAF-mediated acetylation of NF-κB at Lys-122 (Fig. 3A, lower panel).

We then examined whether Aβ treatment induces NF-κB activation through the acetylation of NF-κB at Lys-122. Aβ treatment efficiently induced the activation of NF-κB, and this was inhibited by the knockdown of PCAF (Fig. 3B and C). Importantly, Aβ treatment also increased the acetylation of NF-κB at Lys-122, which was verified by 2 antibodies that recognize acetylated NF-κB at Lys-122 (Fig. 3D). The acetylation of NF-κB at Lys-122 was specifically mediated by PCAF, as shRNA against PCAF, but not control shRNA, abrogated the Aβ-induced acetylation of NF-κB. Collectively, these data suggest that PCAF mediates the Aβ-induced activation of NF-κB through the acetylation of NF-κB at Lys-122.

We then investigated the inhibitory effects of PCAF on NF-κB activity and cytokine production. We first assessed the effect of knocking down PCAF on cytokine production by BV-2 cells. Consistent the results from our previous study (10), Aβ treatment significantly increased the expression of cytokine genes in BV-2 cells (Fig. 4A). Of note, treatment with shRNA against PCAF markedly diminished the Aβ-induced cytokine production in BV-2 cells when compared with control shRNA (Fig. 4A). Importantly, PCAF inhibition by C-11 markedly reduced the Aβ-induced cytokine production in BV-2 cells. Furthermore, we also observed similar results with LPS treatment (Fig. 4B). These data collectively demonstrate that the inhibition of PCAF abolishes NF-κB-mediated cytokine production by blocking the acetylation of NF-κB at Lys-122.

Since PCAF inhibition suppresses the expression of pro-inflammatory cytokines, we examined the neuroprotective effects of PCAF inhibition on BV-2 microglial-mediated Aβ neurotoxicity. Consistent with the results from our previous study showing the dose-dependent cytotoxicity of gallic acid in the same system (10), treatment with Aβ-conditioned medium (CM) from BV-2 microglial cells reduced the viability of Neuro-2A cells, and C-11 reversed the Aβ-CM-mediated neuronal cell death in a dose-dependent manner (Fig. 4C). Real-time PCR demonstrated that C-11 consistently suppressed the Aβ-CM-induced cytokine production in Neuro-2A cells (Fig. 4D). Collectively, these data demonstrate that the inhibition of PCAF by C-11 inhibits NF-κB acetylation, which in turn suppresses Aβ-induced neuroinflammation and Aβ-mediated neurotoxicity.