Natural products are an important resource for the discovery of new leads for cancer therapies. Although garcinol has been shown to have cancer chemopreventive properties in animal models , its biological action remains poorly understood. While garcinol may have pleiotropic effects in cells due to its moderate antioxidant properties , the discovery that it can directly inhibit histone acetylation by p300 [16, 17], indicates that it may impact directly on global histone modifications, and thus gene regulatory processes in tumour cells. However, significant gaps remain in our understanding of the biological effects of garcinol and related molecules on cell function.
A recent report showed that garcinol can block the proliferation of MCF7 breast cancer cells in culture . Using concentrations of garcinol in excess of 25 μM, significant MCF7 cell apoptosis was observed . In this study we confirmed the growth inhibitory effects of garcinol against MCF7 cells (Figure 1A), and established that garcinol is cytotoxic to these cells when used at concentrations in excess of 20 μM, inducing substantial loss of cell adherence and cell lysis (data not shown). As this precludes accurate measurement of the effects of garcinol on histone PTMs by western blot and immunocytochemical analyses, we performed experiments at subcytotoxic levels of garcinol (= or < 20 μM) to better understand its effects on lysine acetylation targets in MCF7 cells.
The reduced proliferation of MCF7 cells observed in MTT assays was confirmed by a dramatic decrease in the number of cells entering S-phase, as detected by BrdU incorporation in flow cytometry analyses (Figure 1B&C). However, the data suggested that the biological effects of curcumin and garcinol/LTK14 in these cells may be distinct. Curcumin-arrested cells showed an enhanced accumulation in G2/M, whereas cells treated with garcinol-related compounds arrested in G1. Similarly, HEPG2 cells have also been reported to arrest in G2/M after treatment with curcumin . It is also worth noting that curcumin differed from garcinol in that it appeared to stimulate the growth of MCF7 cells at the lowest concentration tested (2 μM) (Figure 1A). This may be consistent with reports that low levels of curcumin can stimulate proliferation of other cell types, including neural progenitors  and 3T3-L1 preadipocytes . Consistent with its anti-proliferative effects on a range of other cancer cell lines, curcumin blocked MCF7 cell growth at higher doses (10-20 μM) (Figure 1A). These results highlight the importance of considering the bioavailability of HAT inhibitor compounds to select dose ranges that inhibit rather than promote the growth of malignant cells.
Targeting of histone modifying enzymes is an area of emerging interest in the development of anticancer drugs. Pan-inhibitors of deacetylases (HDACs) have shown promise in preclinical models and have entered clinical trials. The involvement of CBP, p300, MOZ and MORF genes in chromosomal translocations associated with leukaemia  suggests that inhibitors of acetyltransferase enzymes may also have cancer chemopreventive properties. However, little is known regarding the biological effects of currently available lysine acetyltransferase inhibitors, such as garcinol. Consistent with previous studies on other cell types [16–18], we have shown here that treatment of MCF7 cells with curcumin or garcinol can lead to a dose-dependent reduction in bulk levels of histone acetylation, as determined using pan-acetylH3 and pan-acetylH4 antibodies (Figure 2A). Remarkably however, these compounds were found to have differential effects on bulk levels of selected PTMs encountered in histones H3 and H4 (Figures 2,3 ,4). While curcumin had no obvious negative effect on H3K18 acetylation at the concentrations tested, garcinol treatment resulted in H3K18 hypoacetylation in three cancer cell lines tested (Figure 2B,D&E). In contrast, neither compound was found to substantially affect H3K9 acetylation (Figure 2B&D). Acetylation of H3K9 has been shown to be catalysed by GCN5  which is insensitive to garcinol [17, 18]. Interestingly, recent studies have shown that acetylation of H3K18 by CBP/p300 is required for the activation of S phase in quiescent fibroblast cells [22, 38]. Thus, garcinol inhibition of CBP/p300-mediated acetylation of H3K18 may be a contributary factor in the failure of MCF7 cells to proceed through S phase.
Although pan-acetylation of H4 was observed to be reduced by both curcumin and garcinol at 10 μM, we noted that at 20 μM this inhibitory effect was not as clear (Figure 2A). This anomalous result suggested differential dose-dependent effects of HAT inhibitors on H4 acetylation, and highlights the disadvantage of using pan-acetyl H3/H4 antibodies in that effects on specific histone PTMs can be masked. However, as shown in Figure 2B&C, H4K16 acetylation, which is known to be reduced in cancer cell lines , was barely detectable in control MCF7 cells at the concentration of antibody used. However, acetylated H4K16 was readily detected after treatment with curcumin (Figure 2B) or garcinol (Figure 2B&C). Acetylation of H4K16 is normally established by hMOF [39, 40] although in conditions of cell stress other HATs can target this modification, e.g. the DNA damage-associated TIP60. We did not detect any change in the expression levels of hMOF after treatment with garcinol, whereas TIP60 expression appeared to be elevated (Figure 3C&E). Thus increased expression of TIP60 or other HATs may account for the increase in H4K16 acetylation. Although we also attempted to knock down TIP60 transcripts using siRNA in garcinol-treated cells, we did not observe a reduction in TIP60 protein levels by western blotting over the time course of the experiment (data not shown), thus we were unable to establish definitively whether TIP60 is responsible for the observed increase in H4K16Ac.
The observed elevation of γH2A.X foci in MCF7 cells exposed to garcinol (Figure 3A&B) is consistent with an increased incidence of DNA double strand breaks, likely associated with replicative stress . Interestingly, acetylation of H2A.X by TIP60 has been reported to be required for phosphorylation of H2A.X S139 in response to DNA damage [41, 42]. TIP60 is also responsible for acetylation of the DNA binding domain of p53 at K120 . Our observations that garcinol induces expression of p53 and TIP60 in MCF7 cells (Figure 3C-E), accompanied by increased acetylation of p53K120 and its accumulation in the cytoplasm (Figure 3E&F), suggests that TIP60 may drive this switch in p53 function. This is consistent with other studies revealing that p53K120 is acetylated by TIP60 and that this is important for the apoptotic functions of p53 in response to DNA damage . It has also been shown that p53K120Ac is enriched in the cytoplasm and associated with mitochondria where it impacts on apoptotic pathways . We conclude that the garcinol-induced blockade of CBP/p300 inhibits acetylation of the p53 C-terminus and coupled with upregulation of TIP60 or other HATs, is likely to promote an acetylation-mediated switch in p53 function.
A surprising observation in our study was that garcinol also impacts on histone methylation, specifically trimethylation of H4K20 (Figure 4A-C). We have shown that this is due to the induced expression of SUV420H2 (Figure 4C-F), one of the major enzymes targeting H4K20 for multiple methylation. Like H4K16Ac, H4K20Me3 has been implicated in the repair of DNA damage  and cell senescence , both PTMs impact on chromatin structure [44, 45], and a recent study has demonstrated their interdependence in gene transcription . However, the consequences of reduced incidence of H4K20Me3 and H4K16Ac in breast tumours  remains to be determined.