Polyisoprenylated Benzophenone, Garcinol, a Natural Histone Acetyltransferase Inhibitor, Represses Chromatin Transcription and Alters Global Gene Expression*

Histone acetylation is a diagnostic feature of transcriptionally active genes. The proper recruitment and function of histone acetyltransferases (HATs) and deacetylases (HDACs) are key regulatory steps for gene expression and cell cycle. Functional defects of either of these enzymes may lead to several diseases, including cancer. HATs and HDACs thus are potential therapeutic targets. Here we report that garcinol, a polyisoprenylated benzophenone derivative from Garcinia indica fruit rind, is a potent inhibitor of histone acetyltransferases p300 (IC50 7 M) and PCAF (IC50 5 M) both in vitro and in vivo. The kinetic analysis shows that it is a mixed type of inhibitor with an increased affinity for PCAF compared with p300. HAT activity-dependent chromatin transcription was strongly inhibited by garcinol, whereas transcription from DNA template was not affected. Furthermore, it was found to be a potent inducer of apoptosis, and it alters (predominantly downregulates) the global gene expression in HeLa cells.

The acetylation and deacetylation of histones play a key role in the regulation of gene expression in eukaryotic cells (1). The acetylation status of histones alters chromatin structure and thereby modulates gene expression. Two classes of enzymes can effect the acetylation of histones, histone acetyltransferases (HATs), 1 and histone deacetylases (HDACs) (1,2). Interestingly, these enzymes can also acetylate or deacetylate several non-histone substrates with functional consequences (1,3). Altered HAT and HDAC activities can lead to several diseases, ranging from cancer to neurodegenerative diseases (4 -7).
Several families of HATs have recently been identified, which includes the GNAT family (GCN5-related N-acetyltransferase), the MYST group, SAS2, TIP60, and p300/CBP families (1,3). The p300/CBP family of HAT is represented by two of the most widely studied HATs, p300 and CBP. These proteins share considerable sequence and functional homology. Several lines of evidence indicate that p300/CBP are involved in cell cycle progression and cellular differentiation (8 -11). Mechanistically, these proteins function as transcriptional coactivators through their direct interaction with a diverse group of transcription factors and the RNA polymerase II transcription machinery. The coactivation function is partially facilitated by their intrinsic HAT activity (12,13). Mutations in the HAT active site abolish transactivating function (1). The p300/CBPassociated factor, PCAF is one of the important HATs of the GNAT family. The C-terminal-half of PCAF has a highly significant sequence similarity to yeast GCN5 (14). In humans there are two GCN5 splice variants, hGCN5 and hGCN5-L (long form) synthesized from the same gene. The hGCN5-L is similar in length to PCAF and shares 75% amino acid sequence identity with PCAF. It also interacts with p300/CBP. The hGCN5-L is thus termed as PCAF-B (15). PCAF-B is an essential gene expressed ubiquitously early in development, whereas PCAF is expressed later in embryonic development and is not essential (16). In vivo PCAF exists in a large multiprotein complex, containing more than 20 different polypeptides (17). Unlike p300/CBP (which acetylates all the four core histones, predominantly H3 and H4) PCAF acetylates predominantly histone H3. For nucleosomal histone substrates, this specificity is quite exclusive. The acetylase domain of PCAF is required for MyoD-dependent coactivation and differentiation. Presumably the acetyltransferase activity of PCAF and PCAF-B is also involved in DNA repair (15). Both p300/CBP and PCAF also target non-histone protein substrates, which include, human transcriptional coactivators, PC4 (18), HMGB-1 (19), HMG17, HMGI/Y; transcription factors E2F, p53, GATA1 (3), and HIV Tat protein (20,21). The acetylation of these factors alters their DNA/nucleosome binding and/or protein-protein interactions and consequently influences their effect in regulating gene transcription.
It is thus evident that proper balance of acetylation and deacetylation is important for normal cell proliferation, growth, and differentiation. The dysfunction of these machineries leads to different diseases. Several lines of evidence indicate that HAT activity is associated with tumor suppression, and the loss or misregulation of this activity may lead to cancer. For exam-ple, viral oncogene proteins E1A target p300/CBP, disrupting its interaction with PCAF (14). E1A interaction with p300/CBP is essential for cellular transformation. Chromosomal translocations associated with certain leukemias indicate that gainof-function mutations in CBP is also oncogenic (22). Mutations in HATs cause several other disorders other than cancer. Mutations in CBP result(s) in the Rubinstein-Taybi syndrome (RTS) (23). It was found that a single mutation at the PHD domain of CBP causes this syndrome. Interestingly, this mutation (G to C at 4951) in CBP also abolishes its HAT activity (23,24). Degradation of CBP/p300 was found to be associated with certain neurodegenerative diseases (7). Proper HAT function is also essential for the replication of HIV. It was elegantly shown that treatment with HDAC inhibitors inhibits the latency of HIV, presumably by inducing acetylation of Tat protein and the nucleosomes on the LTR (25,26). These examples clearly indicate that histone acetyltransferases and deacetylases should be one of the potential targets for therapy. During the last decade, a number of HDAC inhibitors have been identified that induce apoptosis in cultured tumor cells (4). These inhibitors were also found to be potent anticancer agents in vivo. Furthermore, some of these inhibitors (e.g. SAHA) are already in human trial as antineoplastic drug (27). Although substantial progress has been made in the study of HDAC inhibitors, very little is known about HAT inhibitors. Initially, before the discovery of HATs, polyamine-CoA conjugates were found to inhibit the acetyltransferase activity of cell extracts (28). Availability of recombinant HATs (p300 and PCAF) made it possible to synthesize more targeted specific inhibitors, Lys-CoA for p300 and H3-CoA-20 for PCAF (29). However, these inhibitors could not permeate the cells and were found to be pharmacogenically poor (30). Recently, we have discovered a natural inhibitor anacardic acid from cashew nut shell liquid that potently inhibits both p300 and PCAF (31). Based on anacardic acid we have synthesized a small molecule activator of p300, CTPB. Interestingly, CTPB is specific for p300. Both anacardic acid and CTPB may serve as potential lead compounds for designing different drugs.
Here we report that a polyisoprenylated benzophenone, garcinol, isolated from Garcinia indica (an edible fruit) is a potent inhibitor of histone acetyltransferases p300 and PCAF. It also inhibits histone acetylation in vivo but has no effect on deacetylation of histones. Interestingly, though garcinol repressed the p300 HAT-dependent chromatin transcription, it had no effect on naked DNA transcription. Furthermore microarray analysis of gene expression in garcinol-treated HeLa cells showed that it represses transcription globally with an appreciable chromosome bias.

Purification of Human Core Histones and Recombinant Proteins-
Human core histones were purified from HeLa nuclear pellet as described previously (32). The FLAG epitope-tagged human histone deacetylase 1 (HDAC1), and PCAF were purified from the recombinant baculovirus-infected insect cell line Sf21 by immunoaffinity purification using M2 agarose (Sigma) (32). Full-length p300 was also purified from the recombinant baculovirus-infected Sf21 cells as a His 6-tagged protein through the nickel-nitrilotriacetic acid affinity column (Qiagen) as described previously (12). The His 6 -tagged nucleosome assembly protein 1 (NAP1) used for the in vitro chromatin assembly was purified from Escherichia coli cells as reported previously (32). The FLAGtagged chimeric activator Gal4-VP16 was expressed in E. coli and purified by immunoaffinity purification with M2 agarose (Sigma) (12). His 6 -tagged D-topoisomerase 1 (catalytic domain) was expressed in E. coli and purified as described earlier (33).
Purification and Structural Analysis of Garcinol-Garcinol was prepared from Garcinia indica fruit rind (46). In brief, G. indica dried fruit (Kokum) rind was extracted with ethanol, and the extract was fractionated by ODS (octadecyl silica) column chromatography eluted stepwise with 60 -80% aqueous ethanol. The fractions containing garcinol were concentrated and dried in vacuum. The residue was dissolved in hexane, and the solution was cooled at 5°C for 2 days. Yellow amorphous precipitate was collected from the solution and washed with cold hexane and recrystallized at room temperature. Pale yellow needle crystals were obtained from the solvent, which were identified as garcinol from the following spectral data: mp 126°C; Optical rotation at 30  HAT Assay-The protocol used for the HAT assays is described elsewhere (32). Indicated amounts of proteins (see figure legends) were incubated in HAT assay buffer containing 50 mM Tris-HCl, pH 8.0, 10% (v/v) glycerol, 1 mM dithiothreitol, 1 mM phenylmethylsulfonyl fluoride, 0.1 mM EDTA pH 8.0, 10 mM sodium butyrate at 30°C for 10 min in the presence and absence of garcinol followed by addition of 1 l of 4.7 Ci/mmol [ 3 H]acetyl-CoA and were further incubated for another 10 min. The final reaction volume was 30 l. The reaction mixture was then blotted onto P-81 (Whatman) filter paper and radioactive counts were recorded on a Wallace 1409 liquid scintillation counter. To visualize radiolabeled acetylated histones, the reaction products were resolved on 15% SDS-polyacrylamide gel and subjected to fluorography followed by autoradiography as described earlier (32). For the kinetic analysis of garcinol-mediated inhibition of HATs, a filter binding assay was performed as described in figure legends (Fig. 3).
HDAC Assay-The deacetylation assay was performed as described previously (31), Briefly 2.4 g of core histones were incubated in HAT buffer without NaBu, with 20 ng of p300 and 1 l of 4.7 Ci/mmol [ 3 H]acetyl-CoA for 30 min at 30°C. The activity of p300 was inhibited by incubating the reaction mixture with 10 nM p300-specific inhibitor lysyl-CoA (29) for 15 min at 30°C, after which 50 ng of HDAC1 was added in the presence or absence of garcinol and incubated further for 45 min at 30°C. The samples were analyzed as described above.
Analysis of in Vivo Acetylated Histones by Acid/Urea/Triton (AUT) Polyacrylamide Gel Electrophoresis-HeLa cells (3 ϫ 10 6 cells per 90-mm dish) were seeded overnight, and histones were extracted from the cells after 24 h of compound treatment as described elsewhere (34). Briefly, cells were harvested, washed in ice-cold buffer A (150 mM KCl, 20 mM HEPES, pH 7.9, 0.1 mM EDTA, and 2.5 mM MgCl 2 ) and lysed in buffer A containing 250 mM sucrose and 1% (v/v) Triton X-100. Nuclei were recovered by centrifugation, washed, and proteins were extracted for 1 h using 0.25 M HCl. Chromosomal proteins were precipitated with 25% (w/v) trichloroacetic acid and sequentially washed with ice-cold acidified acetone (20 l of 12 N HCl in 100 ml of acetone), and acetone, air-dried, and dissolved in the sample buffer (5.8 M urea, 0.9 M glacial acetic acid, 16% glycerol, and 4.8% 2-mercaptoethanol). The protein was quantified using a protein assay reagent (Bio-Rad). The histones were resolved on AUT gel as described elsewhere (35,36). Briefly, 8 cm of the separating gel (1 M acetic acid, 8 M urea, 0.5% Triton X-100, 45 mM NH 3 , 18% acrylamide mix, and 0.5% TEMED) was overlaid with 2 cm of an upper gel (1 M acetic acid, 8 M urea, 0.5% Triton X-100, 45 mM NH 3 , 3.3% acrylamide, 0.16% bisacrylamide, and 0.5% TEMED) and polymerization was aided with 0.0003% riboflavin. The gel was pre-electrophoresed for 3-4 h at 130 V in running buffer (1 M acetic acid) until the current no longer dropped. Fresh running buffer was added prior to loading the samples (0.2% methyl green was added as the tracking dye), and the gel was run overnight at 130 V and subsequently stained with Coomassie Brilliant Blue.
In Vitro Chromatin Assembly-Chromatin template for in vitro transcription experiments was assembled and characterized as described earlier (12).
In Vitro Transcription Assay-Transcription assays were essentially carried out as described previously (12) with some modifications. The scheme of transcription has been depicted in Fig. 5A. Briefly, the reconstituted chromatin template (containing 30 ng of DNA) or an equimolar amount of histone-free DNA was incubated with 50 ng of activator (Gal4-VP16) in a buffer containing 4 mM HEPES (pH 7.8), 20 mM KCl, 2 mM dithiothreitol, 0.2 mM phenylmethylsulfonyl fluoride, 10 mM sodium butyrate, 0.1 mg/ml bovine serum albumin, 2% glycerol. p300 was preincubated with indicated amounts of garcinol at 20°C for 20 min following which it was added to the transcription reaction and incubated for 30 min at 30°C. After acetylation, HeLa nuclear extract (5 l, which contains 8 mg/ml protein) was added to initiate the preinitiation complex formation. Transcription reaction was started by the addition of NTP mix and ␣-[ 32 P]UTP after the preinitiation complex formation. The incubation was continued for 40 min at 30°C. A sepa-rate reaction was setup with ϳ25 ng of supercoiled ML200 DNA, and the transcription assay was carried out as described above, without the addition of the activator (Gal4-VP16). 2 l of this reaction was added to each of the transcription reactions to serve as a loading control. The transcription reactions were terminated by the addition of 250 l of stop buffer (20 mM Tris-HCl, pH 8.0, 1 mM EDTA, 100 mM NaCl, 1% SDS, and 0.025 ng/l tRNA). Transcripts were analyzed by 5% Urea-PAGE and visualized by autoradiography. Quantification of transcription was done by phosphorimager (Fuji) analysis.
Apoptosis Assay-Garcinol-induced apoptosis was monitored by the extent of chromatin fragmentation. DNA was extracted from the untreated and garcinol-treated HeLa cells. The cells (3 ϫ 10 6 per 90-mm dish) were seeded and treated with the compound for 24 h. Harvested cells were washed with PBS and then lysed with lysis buffer containing 0.5% Triton X-100, 20 mM Tris, and 15 mM EDTA at room temperature for 15 min. The lysate was treated with RNase (0.1 mg/ml) and proteinase K (2 mg/ml) for 1 h, extracted with phenol/chloroform/isoamyl alcohol (25:24:1), and DNA was precipitated by incubating the upper aqueous phase with 0.1 volumes of 3 M sodium acetate (pH 5.2) and 1 volume of isopropyl alcohol overnight at Ϫ20°C. The pellet obtained on centrifugation was washed with 70% ethanol and dissolved after airdrying in 50 l of TE buffer. The extracted DNA was analyzed on a 1.8% agarose gel and visualized by ethidium bromide staining. Nuclei frag-mentation was also visualized by Hoechst staining of apoptotic nuclei. The apoptotic cells were collected by centrifugation, washed with PBS and fixed in 4% paraformaldehyde for 20 min at room temperature. Subsequently the cells were washed and resuspended in 20 l of PBS before depositing it on polylysine-coated coverslips. The cells were left to adhere on cover slips for 30 min at room temperature after which the cover slips were washed twice with PBS. The adhered cells were then incubated with 0.1% Triton X-100 for 5 min at room temperature and rinsed with PBS for three times. The coverslips were treated with Hoechst 33258 for 30 min at 37°C, rinsed with PBS, and mounted on slides with glycerol-PBS. Stained nuclei were analyzed by using Axioskop-2-plus upright microscope with epi-fluorescence equipment (Carl Zeiss), and the image was captured by Axiocam MRC camera and analyzed by AxioVision 3.1 software.
Microarray Analysis-The microarrays used in this study were procured from the Microarray center, University Health Network, Toronto, Ontario. Each array carries 19,200 spots from the human genome, arranged in 48 individual arrays of 400 spots each. Each of the 48 grids contains 3 Arabidopsis spots that serve as local controls. The total RNA was isolated from control and treated cells using RNaeasy kit (Sigma, 74103). The micromax indirect labeling kit (PerkinElmer Life Sciences, MPS521) was used to synthesize the labeled cDNA from 4 g of total RNA and further process the hybridized cDNA on the array by the tyramide signal amplification method (37). All steps were carried out according to manufacturer's recommendations (www.nen.com/pdf/ penen264-mmaxaminated_card.pdf). The array slides were scanned immediately using a GenePix Presonal 4100A Axon Scanner. The images were analyzed using the GenePix software and the Genowizard software (Genotypic Technology, Bangalore) was used for grid wise normalization of the array. Six arrays were used with two biological repeats of the treatment of cells and at least two dye swap experiments were included in the final analysis. The genes that were picked up as differentially regulated had a log mean of at least 1.27489 with S.D. less than 20% of the expression change in the case of up-regulated genes and a log mean of almost Ϫ1.75726 with S.D. less than 37% of the expression change in case of the down-regulated genes. Guidelines set by MIAME were followed, and the raw microarray data will be deposited in the GEO data base (www.ncbi.nlm.nih.gov/geo/).

RESULTS AND DISCUSSION
Dysfunction of histone acetyltransferases may lead to several diseases, predominantly cancer. Plant extracts or compounds known to have anticancer or cancer chemopreventive activities could be a source of small molecule modulators of HATs. By employing a highly purified recombinant HAT assay system, we have initiated a systematic effort to discover these molecules. Interestingly, a polyisoprenylated benzophenone from G. indica fruit rind has been found to be a potent inhibitor of histone acetyltransferases. Structural analysis identified it as the antioxidant and cancer chemopreventive agent garcinol (Fig. 1A). The HAT inhibitory activity was assayed using baculovirus-expressed recombinant histone acetyltransferase p300 and PCAF (Fig. 1B, lanes 3 and 4) and highly purified HeLa core histones as substrate (Fig. 1B, lane 1). Garcinol was found to be a highly efficient inhibitor of PCAF acetyltransferase activity with an IC 50 of ϳ5 M. Under similar conditions the IC 50 of the inhibitor for p300 acetyltransferase activity was ϳ7 M ( Fig. 2A and data not shown). These results suggest that although garcinol inhibits the HAT activity of both p300 and PCAF, it is relatively more potent as well as a faster inhibitor of PCAF compared with p300. In order to further confirm these results we analyzed HAT assay products on SDS-PAGE followed by fluorography. In agreement with the results of the p300 filter binding assay, it was found that the HAT activity of PCAF was almost completely inhibited by 10 M garcinol compared with the Me 2 SO control, (Fig. 2B, lane 3 versus 6) whereas even at 20 M concentration, 5-10% of 3 H-labeled histone H3 could be detected (Fig. 2B, lane 3 versus 7). Interestingly it was found that acetylation of histone H4 by p300 was more sensitive to inhibition by garcinol compared with that of H3 (Fig. 2B, top panel, lanes 4 -7). The p300-mediated acetylation of histone H4 was completely inhibited at a 1 M concentration of garcinol, while 20 M could not abolish the acetylation of H3 (Fig. 2B, top panel, lane 5 versus lane 7). After establishing garcinol as a strong inhibitor of HATs in vitro, we further investigated whether it could also affect the acetylation of histones in vivo. For this purpose HeLa cells were grown in monolayer (see "Experimental Procedures") and were treated with either Me 2 SO (the solvent for garcinol) or different concentrations of garcinol. Histones were extracted from the cell pellet and analyzed on an 18% acid/urea/Triton polyacrylamide gel electrophoresis. As seen from the profile of different histones (Fig. 2C), incubation with the compound alone did not alter the acetylation status of the cellular histones significantly. In agreement with previous reports (38) the bulk histones from HeLa cells are found to be largely unacetylated (Fig.  2C, lanes 1-3). Because the global acetylation of histones for asynchronous cells does not change significantly, it was not possible to determine the effect of HAT inhibitor on histone acetylation. In order to stimulate histone acetylation, cells were treated with the deacetylase inhibitors TSA and sodium butyrate. As expected deacetylase inhibitors enhance the acetylation of histone H4 as well as H2B dramatically (Fig. 2C, lane  4). The treatment of the cells with garcinol along with TSA and sodium butyrate significantly inhibits the enhanced acetylation of H4 as well as H2B (Fig. 2C, compare lane 4 versus 5 as indicated by an arrow). Taken together, these results establish that garcinol is a potent inhibitor of histone acetyltransferases in vitro and in vivo.
In order to understand the nature of inhibition as well as the mechanism of inhibition brought about by garcinol we analyzed the kinetics of inhibition for both p300 (Fig. 3A) and PCAF (Fig. 3B). The rate of the acetylation reaction at different concentrations of the inhibitor (and in its absence) was recorded with increasing concentrations of [ 3 H]acetyl-CoA and a constant amount of core histones as well as with increasing concentrations of core histones with constant amounts of [ 3 H]acetyl-CoA. The double reciprocal plot for each inhibitor concentration and in its absence was plotted as shown in Fig. 3. The kinetic results show that the inhibition patterns for p300 and PCAF are similar. When the concentration of acetyl-CoA was changed keeping the histone concentration constant, K m increases, whereas V max and k cat of the reaction decrease (Fig. 3, A and B, left panel and Table I).
On the other hand, increasing concentrations of histones with constant amounts of [ 3 H]acetyl-CoA increase K m but V max and k cat remain the same (Fig. 3, A and B, right panel and Table I), which indicates that in this context garcinol competes with histones for binding to the active site of the enzyme and thus acts as a competitive inhibitor.
The reaction mechanism for p300 and PCAF to acetylate the lysine residues is contrastingly different. The GNAT family members, PCAF and serotonin N-acetyltransferase, and GCN5 employ ternary complex mechanisms that involve the ordered binding and release of substrates and products (39). On the other hand, the p300/CBP family follows the double displacement (ping-pong) mechanisms (40). The dead end analogue of acetyl-CoA, desulfo-CoA was shown to be a linear competitive inhibitor versus acetyl-CoA but it behaves as a linear uncompetitive inhibitor versus peptide substrate. Garcinol-mediated inhibition kinetics (for both p300 and PCAF) shows that with changing concentrations of acetyl-CoA it behaves like an uncompetitive type of inhibitor whereas for core histones, as a competitive inhibitor. These differences in the inhibition pat- tern indicate the mechanistic uniqueness of garcinol.
In order to ensure enzyme specificity as well as substrate specificity we went on to check the effect of garcinol on the HDAC1 enzyme. The HDAC assay protocol was followed as described previously (31). Deacetylation of core histones in the presence or absence of the compound, garcinol (10 or 20 M) shows no difference whatsoever (Fig. 4, lanes 7 and 8 versus  lane 3). Addition of the solvent of garcinol, Me 2 SO, has no effect on the deacetylation of core histones by the recombinant HDAC1 (Fig. 4, lane 5). Therefore we can presume that garcinol is specific to HAT activity. In order to verify this HAT specificity we used the HAT-dependent in vitro chromatin transcription assay system as described previously (12). The chromatin template was assembled on pG5-ML-array (12) by employing the NAP1 assembly system. The assembled chromatin was characterized by DNA supercoiling and partial micrococcal (MNase) digestion assay (Fig. 5, B and C). As depicted in the figure, a substantial amount of relaxed DNA was found to be supercoiled upon deposition of nucleosome (Fig. 5B, lane 2  versus lane 3). Because the supercoiling assay does not assure the proper spacing of the histone octamer, partial micrococcal digestion was performed wherein we found 4 to 5 well resolved regularly spaced nucleosomes (Fig. 5C). The results of these assays suggest that the assembled chromatin is appropriate for in vitro transcription experiments. The transcription assay followed the protocol depicted in Fig. 5A. To establish the HAT-specific nature of garcinol, we tested its effect on transcription from DNA, which is not HAT-dependent (Fig. 5D). The chimeric transcriptional activator, Gal4-VP16 activates transcription around 10-fold compared with basal transcription without any activator (Fig. 5D, lane 2 versus 1). Addition of solvent (Me 2 SO) or 20 M and 50 M garcinol shows no effect on the activator-dependent transcription (Fig. 5D, lanes 3-5). The activator-independent transcription from the ML200 promoter was used as a loading control. As reported previously, transcription from the chromatin template shows complete dependence on acetylation (absolute requirement of acetyl-CoA), as depicted in Fig. 5E (lane 3 versus 4). Addition of Me 2 SO, marginally represses the transcription (Fig. 5E, lane 4 versus 5). Interestingly, increasing concentrations of garcinol (especially 50 M) dramatically represses HAT-dependent chromatin transcription (Fig. 5E, lane 5 versus 7). These data show that garcinol specifically inhibits HAT activity-dependent chromatin transcription but not transcription from the DNA template.
We have shown that garcinol is a potent inhibitor of HATs both in vitro and in vivo. Furthermore, it also inhibits the HAT-dependent transcription from chromatin template. In order to further understand its effect in vivo, we treated the HeLa cells with increasing concentrations of garcinol and performed the apoptosis assay. The effect of garcinol on chromatin fragmentation was investigated for this purpose. HeLa cells treated with hydrogen peroxide to induce the apoptosis were taken as   a positive control to test garcinol-mediated apoptosis. Fragmented chromatin was analyzed on a 1.8% agarose gel (Fig.  6A). The cells treated with buffer or solvent (Me 2 SO) did not show any obvious differences (Fig. 6A, lanes 1 versus 3), but treatment with hydrogen peroxide yielded huge amounts of faster moving species of DNA fragments (Fig. 6A, lane 2). Similar to hydrogen peroxide treatment, increasing concentrations of garcinol (30,70, and 100 M) also induce apoptosis and generate smaller DNA fragments (Fig. 6A, compare lane 2 with  4 and 6). To visualize the chromatin fragmentation in situ, compound-treated nuclei were stained with Hoechst (which stains the DNA). In agreement with the DNA fragmentation data Hoechst staining of the nuclei also shows that treatment with 50 and 100 M garcinol induces the fragmentation of nuclei, as indicated by arrows (Fig. 6B, panels c and d). Taken together these data show that the histone acetyltransferase inhibitor garcinol stimulates the apoptosis in HeLa cells.
The histone acetyltransferase specificity, induction of apoptosis, and more significantly the ability of garcinol to inhibit the histone acetylation in vivo prompted us to investigate its effect on global gene regulation. HeLa cells were treated with 100 M garcinol for 24 h and subjected to microarray analysis to investigate its effect on global gene regulation. Genome wide analysis of gene expression using microarrays indicates  that treatment of HeLa cells with garcinol causes the downregulation of a larger number of genes (1631 genes) compared with up-regulation (630 genes). As shown in the inset in Fig.  7, out of 2261 differentially regulated genes, 1445 genes have been annotated, and 816 genes are either ESTs or are unknown genes. We sorted out the annotated genes based on the chromosomal localization and represented this as a bar graph with differentially regulated genes shown per chromosome (Fig. 7). It is evident that on most chromosomes the number of down-regulated genes is higher except on chromosome Y, where there were no down-regulated genes. When normalized for the total number of genes known per chromosome, it turns out that ϳ6 -8% of the known genes in almost all the chromosomes were found to be differentially regulated on treatment with garcinol.
We classified the differentially regulated genes in various functional categories based on the available annotation in the public data bases and that supplied by the slide manufacturer. Some of the interesting categories and the genes that were either up-or down-regulated are listed in Table II. Among the up-regulated genes are those for caspase 4 and CED6, which are pro-apoptotic whereas anti-apoptotic genes like the BCL2 family members and the Fas inhibitory molecule are among the down-regulated genes. The ubiquitin-conjugating enzyme and the E3 ubiquitin ligase are up-regulated, which supports the observed death of treated cells by apoptosis. It was also found that the genes for the p53-induced protein PIGPC1 and p21 (CDKN1A)-activated kinase 6, which are down-regulated probably because they are targets of p53 and p21, respectively, which in turn are regulated by the p300/PCAF histone acetyltransferases. Proto-oncogenes form a class of genes of which more are down-regulated than up-regulated by this treatment, emphasizing the role of garcinol as a molecule with anti-cancer activity. Large numbers of differentially regulated genes involved in metabolism and those categorized as transcription factors or signal transducers have not been included in the table because of space constraints. The exact significance of the result is yet to be determined. A large number of unannotated genes that may have a significant role in cellular functioning are found to be differentially regulated.
As suggested in a recent review (1), the development of small molecular weight HAT inhibitors and activators as therapeutic targets is the next step, following the HDAC inhibitors; some of which are being tested in clinical trials. Here we show for the first time that garcinol, a polyisoprenylated benzophenone from G. indica fruit rind is a small molecule, HAT inhibitor that can be taken in by cells. There are very few HAT inhibitors known to date. The first reported HAT inhibitors were bisubstrate (29) types of inhibitors of p300 and PCAF, which contain CoA moieties. One of these compounds, Lys-CoA has proven useful for blocking the HAT activity of p300 specifically. Though it has been employed for in vitro transcription studies (12) and in cells via microinjection or with the use of cellpermeabilizing agents (41), Lys-CoA has generally been ineffective with simple addition to cell culture media. The cells were also found to be not permeable to a PCAF-specific inhibitor of the same group, H3-CoA-20, which contains CoA-conjugated to a 20-amino acid residue peptide from the N terminus of histone H3. Recently we have isolated the first naturally occurring HAT inhibitor, anacardic acid (AA), which inhibits the HAT activity of both p300 and PCAF very effectively (31). By using AA as a synthon we have synthesized an amide derivative of anacardic acid CTPB, which is the only known small molecule activator of any histone acetyltransferase (p300). Significantly, CTPB is exclusively specific for p300 HAT activity. However cells are not permeable or poorly permeable to both anacardic acid and CTPB. 2 We have demonstrated that garcinol not only inhibits the histone acetylation by p300 and PCAF in vitro (Fig. 2, A and B), it also represses the acetylation in vivo in HeLa cells (Fig. 2C). In correlation with this observation and earlier report (42), garcinol induces apoptosis of HeLa cells in a concentration-dependent manner. Garcinol is known to possess antioxidant and anticancer chemopreventive activity (Refs. 42 and 43 and references therein). Recently it has been shown that garcinol induces apoptosis in human leukemia cell lines (44). The present finding of garcinol as an inhibitor of histone acetyltransferases may help to further understand the mechanism of garcinol-induced apoptosis.
Presumably, hypoacetylation of histone is a prerequisite of apoptosis. Though the relationship between acetylation of histones and activation of gene expression is not as direct as it was believed to be, overall acetylation is a diagnostic feature of active genes. Thus inhibition of acetylation in vivo would repress the majority of the genes. Our microarray analysis of garcinol-treated HeLa cell gene expression indeed showed that more than 72% of genes (tested) were down-regulated. (Fig. 7 and Table II). The microarray data further revealed that several proto-oncogenes are down-regulated in the presence of garcinol, suggesting that garcinol may function as an anticancer compound. However, a systematic investigation using normal (untransformed) and different cancerous cell lines are essential to elucidate the specific role of garcinol for cancer prevention. Because alteration of histone acetylation also has a causal relation with the manifestation of other diseases, 2 R. A. Varier and T. K. Kundu, unpublished data. Transcription factor T74980 Basic leucine zipper nuclear factor 1 (JEM-1) H66228 Core-binding factor, runt domain, ␣-subunit 2; translocated to, 3 R39405 Dishevelled, dsh homolog 2 (Drosophila) AA101861 Heat shock transcription factor 4 R55134 HMG-box transcription factor TCF-3 AA043380 Homeobox D10 R09787 MBD2 (methyl-CpG-binding protein)-interacting zinc finger protein AA045325 msh homeobox homolog 1 (Drosophila) R78177 Paired-like homeodomain transcription factor 2 N92222 Putative homeodomain transcription factor 1 R71263 TAF4 RNA polymerase II, TATA box-binding protein (TBP)-associated factor, 135 kDa R67748 Transcriptional adaptor 2 (ADA2 homolog, yeast)-like R39430 Transcriptional intermediary factor 1 H14414 Zinc finger protein 195 T80906 Zinc finger protein 28 homolog (mouse) R23489 Zinc finger protein 354A a ACCID, GenBank TM accession identification number.
namely asthma (45) and AIDS, garcinol or its derivatives may serve as lead compounds for designing therapeutic targets for other diseases in addition to cancer.