Identification of small molecules inhibitors of GCN 5 histone acetyltransferase activity

Starting from a yeast phenotypic screening performed on 21 chemically different substances we described the discovery of two small molecules as GCN5 inhibitors. The 2-methyl-3carbethoxyquinoline 9 and its 2-desmethyl analogue 18 were able to significantly reduce the yeast cell growth, thus miming the effect of GCN5 deletion mutant. Tested to evaluate their effect on GCN5-dependent transcription of the HIS3 gene, 9 and 18 showed high inhibitory activity of gene transcription, more evident in activated conditions. Compound 9 was also able to reduce the acetylation levels of H3 and, to a lesser extent, H4 in yeast at 0.6 mM. In human leukemia U937 cells, at 1 mM concentration 9 showed 27% apoptosis induction, while 18 had just a little effect in the same conditions. Further studies on 9 and 18 will be performed to deepen their effects on GCN5-related phenomena.


Introduction
The reversible process of histone acetylation occurring at the ε-amino group of lysine residues in the N-terminal tails of core histones mediates conformational changes in nucleosomes.Two classes of enzymes are involved in such process: histone acetyltransferases (HATs) and histone deacetylases (HDACs), which catalyze the addition to and removal, respectively, of acetyl units to histones.][15][16] The anticancer effects of HDAC inhibitors (HDACi) are well known, and a number of them are in clinical trials, 17,18 while the chemotherapeutic potential of HAT targets has been less validated.Missense and deletion mutations in the p300 gene have been found in colorectal, gastric, and epithelial cancers, and the loss of heterozygosity of p300 gene has been related to glioblastoma. 19,20p300 and PCAF have been reported to play an important role in MyoD dependent cell cycle arrest, 21 and dysregulation of GCN5/PCAF in genetic diseases and cancer has led to the supposition that selective inhibitors of these HATs may have therapeutic applications. 22,23  date, a small number of HAT inhibitors has been reported.Among them, the bi-substrate analogues Lys-CoA and H3-CoA-20 are selective for p300 and PCAF, respectively. 24Some natural products such as anacardic acid, 25 garcinol, 26 and curcumin 27 have been reported as potent p300 and PCAF inhibitors, and recently the γ-butyrolactone MB-3 and a series of isothiazolones have been disclosed as inhibitors of both p300 and PCAF HAT activities 28,29 (Figure 1).Pursuing our searches on design, synthesis, and biological evaluation of small molecules as epigenetic tools for regulating gene expression and transcription, [30][31][32][33][34][35][36][37][38][39][40] we performed a yeast phenotypic screening on a set of both newly synthesized and commercially available molecules (1-14) (Figure 2), to find among them one or more modulators of HAT activity.Between tested compounds, derivatives 1-4 were chosen because they resemble some structural similarities with the ε-acetyl-lysine, the product of acetylation reaction.Anacardic acid (AA) 5, 25 was used as the template for design and synthesis of several analogues 6-10, in which the AA benzene ring was replaced by pyrimidine (6, 7) or quinoline (8) moiety, including the two simpler quinolines 9 and 10.All-trans retinoic acid (ATRA) 11, 41 a well known transcriptional modulator, and the corresponding hydroxamate 12 also show some structural analogies with AA, and were included in the assay.Apicidin 13, 42 and HC-Toxin 14, 43 were two HDACi belonging to the cyclic tetrapeptide family, and were comprised in the cellbased test to study an eventual positive modulation of GCN5.Indeed, a amide analogue of AA (N-(4-chloro-3-trifluoromethylphenyl)-2-ethoxy-6-pentadecylbenzamide, CTPB) 25 has been reported to be a p300 activator, and we expected that HDACi could show in cell-based assay a phenotype similar to that of HAT activators.

Lys-CoA
Between tested compounds 1-14, only the 3-carbethoxy-2-methylquinoline 9 was found active in inhibiting yeast cell growth.In Saccharomyces cerevisiae we have assessed that the 9 inhibitory effect is equivalent to a GCN5 loss-of-function mutation. 44in the presence of (a) di-tert-butyl dicarbonate and (b) diisopropyl azodicarboxylate (DIAD)/triphenylphosphine followed by acidic removal of the protecting tert-butoxycarbonyl group (Scheme 1).Acylation of potassium ethyl malonate with pentadecyl imidazolide in the presence of magnesium dichloride and triethylamine furnished the corresponding β-oxoester 24, which was in turn condensed with thiourea and sodium ethoxide in ethanol to give the 2-thiouracil 6.Following treatments of 6 with (a) methyl iodide, and (b) N-bromosuccinimide (NBS) afforded the 5-bromopyrimidine 25, that was converted into the 4-hydroxy-2-methylthio-6tetradecylpyrimidine-5-carboxylic acid 7 by reaction with n-butyl lithium and carbon dioxide at -78 °C (Scheme 2).
The 4-hydroxyquinolines 8 and 15 were obtained by reaction of ethyl 3-oxooctadecanoate (for 8) or ethyl acetoacetate (for 15) with isatoic anhydride in the presence of sodium hydroxide as catalyst (Scheme 3).The synthesis of 9 was accomplished by a single-step variant of the Friedländer synthesis involving o-nitrobenzaldehyde and ethyl acetoacetate in the presence of SnCl 2 and ZnCl 2, 44 and the corresponding carboxylic acid 10 46 was obtained from 9 by standard procedure (Scheme 3).
The N-hydroxyretinamide 12 47 was prepared by reaction of the commercial all-trans-retinoic acid 11 with ethyl chloroformate, followed by addition of O-(2-methoxy-2propyl)hydroxylamine, 48 and subsequent acidic treatment in the presence of the Amberlyst ® 15 ion-exchange resin (Scheme 4).
The commercial 3-quinolinecarboxylic acid 20 was the starting material for the preparation of both the corresponding ethyl ester 18 (by standard method) 49 and the carboxyamide 21, 50 (by two-step, one-pot reaction with ethyl chloroformate/2.5 M ammonia solution in dioxane).Finally, the ethyl 2-naphthoate 19 was prepared from 2-naphthoyl chloride by standard reaction. 51

Results and Discussion
Compounds 1-14 were tested to evaluate the effect of growth inhibition of yeast cells.Because yeast strains lacking GCN5 (gcn5∆) showed a slow growth and alteration in cell cycle phases, 52 we expected that compounds slowing down the yeast cell growth could mimic the effects produced by the deletion of GCN5.Among the tested compounds 1-14 (Figure 4), only 9 gave a significant slow of cell growth at 0.2 to 1 mM concentration.Tested on gcn5∆ yeast strain, 9 had just little effect of cell growth (Figure 5, left panel).To determine if there was a direct link between 9 activity and the HAT catalytic activity of GCN5, we tested the effect of 9 on growth in the mutant F221A yeast strain, that contains the F221A point mutation in the HAT minimal catalytic domain of yGCN5.Compared with the effect of 9 on the respective isogenic wild-type strain yMK839, the reduction of F221A growth was significantly lower (about 25%) than that observed with MK839 (about 50%) (Figure 5, right panel), thus providing the evidence that the 9 inhibition is exerted on the catalytic activity of GCN5 but not on the whole protein.
yeast cell growth  Furthermore, we demostrated that the action of 9 on wild-type Saccharomyces cerevisiae is equivalent to a GCN5 loss-of-function mutation.The HAT activity of GCN5 is required for transcriptional activation of target genes in vivo. 52In yeast, GCN5 was first described as a transcriptional coactivator of amino acid biosynthetic genes, HIS3 being one of the most affected.Tested on HIS3-lacZ, GCN5-dependent activated transcription in wt strain under basal (SD medium) and activated (SD + 3-aminotriazole (3AT)) conditions, 9 strongly inhibited the yeast β-galactosidase activity, while failed in inhibiting the GCN5-undependent GAL10-CYC1-lacZ transcription (Figure 6).To ascertain that acetylation reaction is the primary target for the inhibiting activity of 9, a in vivo assay was performed by measuring the acetylation level of histones H3 and H4 N-terminal tails in protein extracts from wt (W303) and gcn5∆ (yPO4) cells, grown for 16 h in YPD medium alone and in the presence of 0.6 mM 9. In these conditions, 9 clearly inhibited H3 and, to a lesser extent, H4 histone tail acetylation (Figure 7).Prompted by these results, we prepared and tested in our cell-based phenotypic screen seven novel analogues of 9 (compounds 15-21), to acquire SAR information about the inhibiting activity.Among 15-21, only the 3-carbethoxyquinoline 18 was able to inhibit the growth of yW303 cells, it being less effective in inhibiting the gcn5∆ strain cell growth.The insertion of a hydroxyl group at the C4 position of 9 (compound 15) as well as the replacement of the quinoline of 9 with the pyridine (compound 16) or the deletion of the C3-carbethoxy function (compound 17) abolished the GCN5 inhibiting activity.Differently, compound 18 lacking the C2-methyl substituent retained the GCN5 inhibitory activity, but the replacement of its quinoline with naphthalene (19), or the introduction of a carboxyl (20) or carboxamide (21) function at the quinoline C3-position instead of the 3-carbethoxy moiety gave inactive products (Figure 8).The 3-carbethoxyquinoline 18 and the 3-carbethoxy-2-methylpyridine 16 (as negative control) were tested in HIS3-lacZ transcription assay, to evaluate the functional inhibition of GCN5-dependent gene transcription.In this test, the reporter activity was highly reduced in the presence of 18 both in basal (SD medium) and, more drastically, in activated (SD +3AT) conditions, while the treatment with 16 was ineffective (Figure 9).To investigate on human cell activities of 9 and 18, the effect on apoptosis induction in human leukemia U937 cell lines wase determined.After 24 h, at 1 mM concentration 9 was able to induce 27% of apoptosis, while 18 had just little effect in the same conditions (Figure 10).The two compounds also showed different capabilities in cell cycle arrest, compound 9 being the strongest (data not shown).Further studies on 9 and 18 will be performed in human cancer cell lines to deepen their effects on GCN5-related phenomena.

CFigure 4 .
Figure 4. Inhibiting effect of 1-14 on yeast cell growth.The numbers of yeast cells were evaluated as arbitrary units by optical density.

Figure 5 .
Figure 5. Inhibition of yeast cell growth in the presence of 0.2 to 1 mM concentrations of 9 in wt (W303 and MK839), gcn5∆, and F221A mutant strains.

Figure 6 .
Figure 6.Effects of 9 on GCN5-dependent (left panel) and GCN5-undependent (right panel) transcription.Data represent mean values of at least three separate experiments.

Figure 7 .
Figure 7. Effect of 9 on acetylation levels of yeast histones H3 and H4.

Figure 9 .
Figure 9. Effects of 18 and 16 on GCN5-dependent transcription.Data represent mean values of at least three separate experiments.

Figure 10 .
Figure 10.Apoptosis induction by 9 and 18 in human leukemia U937 cells.Data represent mean values of at least three separate experiments.