Acute Non Lymphocytic Leukemia

Histone deacetylases in acute myeloid leukaemia show a distinctive pattern of expression that changes selectively in response to deacetylase inhibitors


Histone deacetylase inhibitors (HDIs) are a new class of drugs with significant antileukemic activity. To explore mechanisms of disease-specific HDI activity in acute myeloid leukaemia (AML), we have characterised expression of all 18 members of the histone deacetylase family in primary AML blasts and in four control cell types, namely CD34+ progenitors from umbilical cord, either quiescent or cycling (post-culture), cycling CD34+ progenitors from GCSF-stimulated adult donors and peripheral blood mononuclear cells. Only SIRT1 was consistently overexpressed (>2 fold) in AML samples compared with all controls, while HDAC6 was overexpressed relative to adult, but not neo-natal cells. HDAC5 and SIRT4 were consistently underexpressed. AML blasts and cell lines, exposed to HDIs in culture, showed both histone hyperacetylation and, unexpectedly, specific hypermethylation of H3 lysine 4. Such treatment also modulated the pattern of HDAC expression, with strong induction of HDAC11 in all myeloid cells tested and with all inhibitors (valproate, butyrate, TSA, SAHA), and lesser, more selective, induction of HDAC9 and SIRT4. The distinct pattern of HDAC expression in AML and its response to HDIs is of relevance to the development of HDI-based therapeutic strategies and may contribute to observed patterns of clinical response and development of drug resistance.


Acute myeloid leukaemia (AML) is an incurable disease in the majority of adults and few patients over 60 years are long-term survivors. New and more targeted therapies are urgently needed to reduce the toxicity and increase the efficacy of treatments. Mounting evidence indicates that mutant transcription factors, often resulting from chromosome translocations, contribute to the pathogenesis of AML by inappropriate association with complexes containing corepressors and histone deacetylases (HDACs), thus causing altered chromatin architecture and modified gene expression.1, 2 HDACs are key players in the control of gene expression through chromatin modification and recent studies have shown that treatment of AML cells with HDAC inhibitors (HDIs) resensitises them to signals for differentiation and or apoptosis, making HDIs particularly promising agents for AML therapy.3

The known HDAC enzymes can be divided into three classes on the basis of sequence homology.4, 5 Class I comprises HDAC1, 2, 3 and 8, all homologues of the yeast RPD3 protein; class II comprises HDAC4, 5, 6, 7, 9 and 10, homologues of the yeast Hda1 protein, while class III comprises the seven sirtuins, SIRT1–7, homologues of the yeast Sir2 protein.6 HDAC11 contains all the necessary features to be designated as an HDAC, but cannot be allocated to any one of the existing three classes as the overall sequence similarity is too low.5, 7 HDACs are usually part of multiprotein complexes whose other components both regulate HDAC activity and target the complex to the appropriate regions of the cell or the genome.8 Acetylation of lysines is a common protein modification, and it is now apparent that the primary in vivo substrates of some members of the HDAC family are not histones, but non-histone proteins.9, 10, 11, 12, 13, 14 It is not known which of the currently known family members are responsible for the antitumour effects of HDI, or which in vivo substrates are primarily involved.

Several classes of HDI have been identified and characterised, varying widely in structure and in the concentrations at which they are biologically active. These include salts of short chain fatty acids such as butyric and valproic acid; hydroxamic acids such as Trichostatin A (TSA), suberoylanilide hydroxamic acid (SAHA) and oxamflatin; cyclic peptides such as depsipeptide and benzamides such as MS-27-275 (reviewed in Marks et al15). Some of these inhibitors are currently in clinical trials against a variety of solid and haematological tumours.15, 16 From the limited amount of data so far available, it seems that HDIs show significant differences in activity against specific classes of HDAC, an effect that may contribute to their selective antitumour activity. For example, in vitro assays show that class III NAD-dependent deacetylases are less sensitive to both TSA and fatty acids than class I enzymes tested in parallel,6, 17 while deacetylases associated with the NCoR corepressor complex (isolated by immunoprecipitation) are more sensitive to inhibition by valproic acid than enzymes not associated with NCoR.18

The mechanisms by which HDI can selectively kill tumour cells remain unclear. One possible contributory factor could be differences between normal and tumour cells, in the expression levels of members of the HDAC gene family, or in the functional roles played by individual HDACs. To explore these possibilities, we have used real-time quantitative PCR (RTQ-PCR) to assay expression of all 18 known members of the HDAC gene family in mononuclear cells from AML patients and a variety of control cell types. We detect differences in HDAC expression that are characteristic of the disease state, prominent among which is consistent overexpression of HDAC6 and the NAD-dependent deacetylase SIRT1, and underexpression of HDAC5. In addition, leukaemic blasts treated in culture with HDI show a selective, several-fold increase in expression of HDAC11, a change mirrored in AML cultured cell lines. Overexpression of specific deacetylases in AML cells, whether intrinsic or induced by exposure to HDI, has implications for the effectiveness of therapies designed to promote selective killing of tumour cells.


Primary samples and cell culture

Primary AML blasts were obtained from bone marrow or high-count (>80% blasts) peripheral blood. Mononuclear cell preparations were made using Ficoll–Hypaque (Amersham-Pharmacia, UK) density separation according to the manufacturer's instructions. For tissue culture experiments, cells were plated at 1 × 106 cells/ml in RPMI 1640 with the following additives (all from R&D systems, Abingdon, UK), 1% (v/v) ITS (insulin/transferrin/selenium), 1 ng/ml IL3 and 10 ng/ml SCF (Stem Cell Factor, c-kit ligand).

Umbilical cord blood (UCB) cells were obtained from patients undergoing elective caesarean sections at the Birmingham Women's Hospital. Mononuclear cells were purified as above and CD34+ cells positively selected using anti-CD34+ magnetic beads (Milteny Biotech, UK). Cell pellets were frozen for use as noncycling cells. Cycling cells were obtained by culturing sorted cells in RPMI 1640, supplemented as above, for 4–5 days before resorting CD34+ cells and harvesting.

Peripheral blood mononuclear cells (PBMCs) were prepared by Ficoll–Hypaque gradient centrifugation, as above, either from untreated normal donors (quiescent cells) or from GCSF-mobilised normal adult donors (MPB, cycling cells). The latter were further purified by selection of CD34+ cells as above. All samples from patients and normal donors were taken under local ethical committee approval and with informed consent.

HL60, K562 and KG1a cell lines were cultured in RPMI 1640 medium supplemented with 10% heat-inactivated foetal calf serum, (FCS), 100 U/ml penicillin, 100 μg/ml streptomycin (Gibco BRL, UK) at 37°C, 5% CO2 in a humidified incubator.


HDIs were prepared as required from the following stock solutions; valproic acid (VPA), sodium salt, 1 M in water, TSA 100 μg/ml in ethanol, sodium butyrate 1 M in water (all purchased from Sigma) and SAHA 5 mM in water (a gift from Dr V Richon and Professor P Marks, Aton Pharma, NY, USA).

SDS gels and Western blotting

Proteins were resolved by SDS-polyacrylamide gel electrophoresis, and transferred onto nitrocellulose membrane, as previously described.19 Protein loadings for normalisation were measured on duplicate gels by Coomassie blue staining and laser densitometry (LKB Bromma Ultroscan XL Laser densitometer). Modified histones were detected with rabbit polyclonal antibodies raised and characterised as described previously.20 Anti-HDAC11 was from Abcam (Cambridge, UK). A mouse monoclonal antibody raised against HDAC3, but recognising HDAC1, 2 and 3, was from BD Transduction Laboratories; we have confirmed the unusual specificity of this antibody by comparison with our own antisera specific for HDAC1, 2 or 3.21 Bound antibody was detected either with peroxidase-conjugated anti-rabbit antibody and Enhanced ChemiLuminescence (ECL, Amersham-Pharmacia, UK) or with IRDye-800-conjugated, affinity-purified goat anti-rabbit secondary antibody followed by quantitation with an infra-red dye imaging system (LI-COR Biosciences Ltd, UK). The latter was used for all experiments with primary blast cells from patients.

RNA extraction and cDNA synthesis

The 23 samples used for HDAC expression analysis were provided as archived total RNA from bone marrow mononuclear cell preparations taken, pretreatment, from adult AML patients at the time of entry to the MRC AML 12 trial. cDNA was prepared from 1 μg total RNA using the Promega Reverse Transcription System (Promega, UK) and random hexamers. Total RNA was extracted from cell lines and primary cells using the Promega SV total RNA isolation kit according to the manufacturer's instructions. The protocol includes incubation with DNAse I to remove contaminating genomic DNA. For these samples, cDNA was synthesised from total RNA using Superscript II (Invitrogen). Briefly, 2 μg total RNA, 250 μg random hexamers and 10 pmoles dNTPs in 10.5 μl were denatured for 5 min at 65°C and placed on ice. The volume was made up to 20 μl with buffer (25 mM Tris-HCl (pH 8.3), 75 mM KCl, 3 mM MgCl2 final concentration), 10 mM DTT, 40 U RNaseOUT (RNase inhibitor, Invitrogen) and 200 U Superscript II. The reaction was incubated at 25°C for 10 min, 42°C for 90 min and then 70°C for 15 min. cDNAs were diluted five-fold and 1 μl was used in each PCR reaction.


RTQ-PCR was performed using primer/probe mixes (FAM-TAMRA labelled) purchased from ABI Biosystems. The assay (kit) numbers that uniquely identify each gene studied, accession numbers and other information for each gene can be obtained through the ABI website ( Reactions were performed using an ABI Prism 77000 sequence detector (Applied Biosystems, CA, USA) according to the manufacturer's instructions. Briefly, each reaction contained 1 × primer/probe mix, 1 × Mastermix (containing preoptimised dNTPs, MgCl2 and buffer concentrations; Eurogentec), and 1 μl diluted cDNA. Primers and probes to 18S rRNA (Applied Biosystems, CA, USA) were used as internal controls according to the manufacturer's instructions. Cycle threshold (Ct) values were obtained graphically for test genes and 18S internal standard using threshold values of 0.07044 and 0.0497, respectively. ΔCt values were calculated by subtracting 18S Ct from test gene Ct. Relative mRNA levels were determined by subtraction of normal control ΔCt values from AML ΔCt values to give a ΔΔCt value and conversion through 2 - ΔΔ C t .


HDAC expression levels in AML blasts

We have performed an RTQ-PCR screen of expression, in AML blasts, of all 18 known HDAC family members, three HDAC-associated proteins (RBBP4, RBBP7, SIN3A) and two histone methyltransferases (SET7, MLL). To validate the approach, we first screened expression in PBMCs from four normal healthy donors. There was a striking consistency in expression of all the target genes across all four samples, with standard errors consistently 1% or less of the mean ΔCt value and never more than 4% (Supplementary Table S1). These findings demonstrate a consistent pattern of HDAC mRNA levels in PBMCs from different individuals.

AML data were normalised to each of four different controls: (i) PBMCs, (ii) noncycling CD34+ cells from UCB processed directly at the time of sampling, (iii) cycling CD34+ cells from UCB (reselected for CD34 expression after exposure to IL3 and SCF) and (iv) cycling peripheral blood CD34+ cells from GCSF-mobilised adult donors (MPB). The complete data set, showing expression levels for each of the genes tested, averaged across 23 AML samples and related to each of the four different controls, is presented in Supplementary Table S2. Results for each mRNA expressed relative to MPB are shown graphically in Figure 1 (top left). The variability between AML samples is striking, as evidenced by the high standard deviations (Table S2). To illustrate this variability, the results are presented in Figure 2 as the number of samples, from the 23 tested, that are more than two-fold above (+ values) or below (− values), the mean for each target RNA in the chosen control. To facilitate comparison, RNAs in each panel are ranked on the basis of the frequency with which their expression levels are depressed or elevated when using cycling, CD34+ adult cells (MPBs) as the control (Figure 2a).

Figure 1

Expression of HDACs in primary AML cells and cultured cell lines. A total of 23 primary AML samples and leukaemic cell lines HL60, K562 and KG1a (as indicated) were analysed for expression of HDACs and other chromatin-modifying proteins by RTQ-PCR. Expression levels were calculated relative to expression in cycling CD34+ mobilised peripheral blood cells (MPB). Data are shown as mean±s.e. from a minimum of four experiments for each cell line. The order of the genes on the x-axis is the same as in Figure 2.

Figure 2

Frequency of elevated or depressed HDAC expression in primary AML cells relative to four different control cell populations. RNAs encoding HDACs and other proteins involved in chromatin remodelling were assayed in 23 primary AML samples by RTQ-PCR. Relative mRNA levels were calculated by comparison to either cycling mobilised peripheral blood cells (MPBs, n=2), noncycling peripheral blood, mononuclear CD34+ cells (PBMCs, n=4), noncycling umbilical cord blood CD34+ cells (UCBs, n=2) or cycling UCB CD34+ cells (n=4). Relative mRNA expression values greater than 2 × control levels or less than 0.5 × control levels were taken as elevated (+ve values) or decreased (−ve values), respectively. In all the graphs, the genes tested are arranged in the same order, determined by the proportion of samples showing increased expression with respect to the cycling MPB controls.

Some RNAs were underexpressed in most or all samples tested and never overexpressed (eg HDAC5, RBBP4), while others were overexpressed in the great majority of samples and rarely, if ever, underexpressed (eg SIRT1, HDAC6). This profile changed little when the values were normalised to noncycling PBMC (Figure 2b). The same four RNAs were the most frequently underexpressed (ie HDAC5, RBBP4, SIRT4 and HDAC4) and the same four were the most frequently overexpressed (ie HDAC6, SIRT1, SET7 and HDAC2). A few HDACs showed relatively small quantitative shifts as a result of changing from a cycling to noncycling control, among which MLL and HDAC8 stand out. These shifts presumably reflect cell-cycle-related changes in expression, although their significance remains unproven.

When normalised to cycling UBC cells (Figure 2c), the expression profiles remained similar, the only exception being a marked reduction in the frequency with which HDAC6 appeared to be overexpressed in AML. This was true also when using the noncycling UBC control (Figure 2d), and is attributable to a relative overexpression of HDAC6 in perinatal as compared to adult CD34+ cells. It seems that most AML samples show a level of HDAC6 expression at or around that of immature myeloid progenitor/blast cells.

AML blasts show widely varying sensitivities to VPA-induced killing in vitro

We obtained fresh bone marrow aspirates from four AML patients and determined both complete HDAC expression profiles and sensitivity to killing by VPA in culture. HDAC expression profiles were consistent with those from the 23 patient samples analysed previously, with individual HDACs showing varying expression from one patient to another and with consistent overexpression of HDAC6 and low expression of HDAC5 and SIRT4 (the complete data set is shown in Supplementary Table S3). Three of the four AML cell samples were resistant to prolonged exposure to 5 mM VPA in culture (0–20% killing after 6 days), while the fourth was sensitive (>90% killing). We find no relationship between cell killing and under- or overexpression of 15 of the 18 HDACs tested. However, the VPA-sensitive cells stand out in showing both elevated HDAC5 (almost six-fold) and slightly underexpressed SIRT1 and HDAC4. The elevated HDAC5 is unprecedented among the samples we have tested, while underexpression of SIRT1 is unusual (Figure 2, Table S3).

HDAC levels in cultured leukaemic cell lines

Figure 1 shows the overall, average HDAC expression profile, relative to cycling adult CD34+ cells, in the 23 primary AML samples and in HL60, KG1a and K562 cultured cell lines.22 Although relative expression levels of most HDACs varied from one line to another, the profiles showed consistent features. Thus, HDAC1, 2 and 6 were always among the most highly expressed HDACs in all three lines. Notably, prominent overexpression of SIRT1, common in AML patient samples, was not seen in any of the three cell lines tested.

Deacetylase inhibitors selectively induce HDAC expression in primary AML blasts and cell lines

We have assayed HDAC expression in HL60 cells after growth for 8 h in 1 and 5 mM VPA. Such treatment resulted in no significant loss of cell viability, as measured by dye exclusion or sub-G1 fragments on FACS analysis (not shown). As shown in Figure 3a, VPA at 1 mM caused an almost 200-fold increase in HDAC11 mRNA, with much smaller increases of HDAC9 and SIRT4 (10–20-fold). Increasing the VPA concentration to 5 mM caused only a slight increase in induction of HDAC11 and SIRT4, but had a major effect on HDAC9, which now showed a 77-fold increase over untreated (Figure 3a). We tested the effect of VPA on a second AML cell line, KG1a, and primary AML blasts (Figure 3b and c). In both cases, HDAC11 stood out as the most highly induced member of the HDAC family, although the level of induction was well below that seen in HL60 cells. This presumably reflects the higher level of HDAC11 expression in untreated KG1a cells (Figure 1). KG1a cells and AML blasts showed no significant change in HDAC9 expression after VPA treatment; once again this may reflect a higher expression level prior to treatment (Figure 2). The selective upregulation of HDAC11 was confirmed by Western blotting, which showed that HDAC11 protein was barely detectable in extracts of untreated HL60 cells, but readily visible after exposure to VPA (Figure 4, upper panel), although still at low levels relative to some other cell types, for example, HeLa (Figure 4, middle panel). The two bands typically detected by Western blotting are likely to be attributable to the splice variants noted previously.7 VPA caused no increase in the relatively high level of HDAC11 in HeLa cells, nor did it change levels of HDAC1, 2 or 3 in HL60 cells (Figure 4, lower panel). Loss of HDAC2 due to increased proteasomal degradation has been noted in cells exposed to VPA for 24 h or more.23 There was no loss of HDAC2 in our experiments, presumably due to the much shorter treatment times involved.

Figure 3

Effect of sodium valproate on HDAC expression in different cell types. All cells were grown with or without 1 or 5 mM (HL60 only) sodium valproate for 8 h and RNA extracted. Levels of mRNAs encoding HDACs and associated proteins were determined by QRT-PCR. Expression was calculated relative to untreated control cells. Results shown for HL60 are typical of several experiments (see also Figure 5) and for KG1a are the mean of two independent experiments.

Figure 4

Western blot analysis of HDAC expression in cells treated with sodium valproate. HL60 (top and bottom panels) or HeLa cells (middle panel) were grown with or without 5 mM valproic acid (VPA) for 6 h. Whole-cell extracts were separated by SDS-PAGE, Western blotted and immunostained with rabbit polyclonal antibody to HDAC11 (top and middle panels) or a mouse monoclonal antibody that recognises HDAC1, 2 and 3 (lower panel). These class I HDACs show no change after short exposure to VPA.

To study the timing of HDAC induction in HL60 cells, expression of all HDACs was examined at 15 min and 1, 4 and 8 h after addition of the inhibitor. We saw no change in any HDAC RNA after treatment with 1 or 5 mM VPA for 15 min or 1 h (expression levels relative to t0 ranged from 0.5–2.2-fold, with mean±s.d. values of 1.3±0.29 and 0.9±0.13, confirming the reproducibility of the procedure). After 4 h, changes in HDAC11, HDAC9 and SIRT4 were all detectable, but several-fold below the levels reached after 8 h. These findings are summarised in Figure 5a and b (note that a logarithmic scale is now used). Longer time points were not tested because of the complications introduced by altered cell cycle profiles and increasing numbers of apoptotic cells, and because bulk histone acetylation reaches a maximum after 6–8 h treatment (Figure 5f and results not shown). In the case of cells treated with TSA, levels of acetylation fall at longer treatment times, largely as a result of decay of the inhibitor (Figure 5f and results not shown).

Figure 5

Time course of induction of selected HDACs by various HDAC inhibitors. HL60 cells were treated with the HDIs shown (a–e) for 15 min, 1, 4 or 8 h (see key in (a)), before harvesting and analysis of HDAC expression by RTQ-PCR. All values are expressed (on a log scale) relative to an untreated cell sample prepared and analysed in parallel. Individual results are shown for HDAC9, SIRT4 and HDAC11, as indicated. Results labelled ‘Combined’ represent mean values (±2 × s.d.) for the 15 other HDAC and SIRTUIN genes tested (as listed in Table S3). (f) Extracts of HL60 cells treated with TSA or VPA as indicated were resolved by SDS-PAGE, Western blotted and immunostained with an antiserum to H4 acetylated at lysine 8 (H4K8ac).

As shown in Figure 5c–e, upregulation of HDAC11 (60–160-fold), and SIRT4 (9–15-fold), was seen also after growth of HL60 cells for 8 h in medium containing sodium butyrate (10 mM), TSA (50 ng/ml, 165 nM) or SAHA (2.5 μ M). Induction of HDAC11 would therefore seem to be a result of the common property of these inhibitors, namely deacetylase inhibition. In contrast, there were no significant changes in any other HDAC, and only a small increase in HDAC9, whose strong induction by sodium valproate (Figure 5b) would therefore seem to be peculiar to that inhibitor. We also noted that with both TSA and SAHA, significant increases in HDAC11 (ie more than two standard deviations above the mean for all HDACs, see error bars in Figure 5d and e) were seen 15 min and 1 h after addition of the inhibitor. No other HDAC showed such early increases in cells treated with SAHA or TSA, nor were they seen with HDAC11 after treatment with butyrate or valproate (Figure 5a–c).

HDIs increase methylation of H3 lysine 4

Selective changes in gene expression induced by exposure to HDIs are likely to result, at least in part, from increased levels of histone acetylation and consequent changes in chromatin structure at selected genomic regions. However, HDI-mediated changes in acetylation of transcription factors and other non-histone proteins, and possibly changes in other histone modifications,24 may also be involved. In the course of a wide-ranging study of the effects of HDI on patterns of histone modification and gene expression, we have found a consistent and robust increase in di- and trimethylation of H3 lysine 4 that parallels the overall increase in histone acetylation. Illustrative results with primary AML blasts treated for 18 h with 1 mM VPA are shown in Figure 6a. So far, the increased methylation is H3K4 specific, with no change in methylation at H3K9 (Figure 6a), H3K27 or H4K20 (not shown). ATRA added in combination with VPA produced no further increase in either acetylation or H3K4 methylation (Figure 6a). Across eight different AML samples tested, trimethylated H3K4 showed a stronger increase than the dimethylated isoform at both 1 and 5 mM VPA (Figure 6b).

Figure 6

AML blasts treated with valproate show increased methylation of H3 lysine 4. Mononuclear cells were isolated from AML bone marrow aspirates and placed in culture medium containing VPA with or without all-trans retinoic acid (ATRA, 100 nM), as indicated. Histones were prepared after 18 h, separated by SDS-polyacrylamide gel electrophoresis, Western blotted and immunostained with antibodies to acetylated or methylated histone isoforms as indicated. (a) Typical gel strips either stained with Coomassie blue (CB) to show the four core histones (H3, H2B, H2A and H4) or after immunostaining with antisera to H3 trimethylated at lysine 4 (H3K4me3), dimethylated at lysine 4 (H3K4me2), trimethylated at lysine 9 (H3K9me3) or H4 acetylated at lysine 8 (H4K8ac). (b) Immunostaining was quantified by laser densitometry and corrected for protein loading (CB staining). Results shown represent the means±s.d. of data from eight different AML samples assayed for levels of H3K4me3 and H3K4me2 after 18 h in 1 and 5 mM VPA.


Patterns of HDAC gene expression are altered in AML

The comprehensive analysis of HDAC expression presented here demonstrates that expression of these enzymes is consistently dysregulated in AML cells compared to normal haematopoietic progenitors. Previous studies of one, or a few, selected HDACs have noted changes in specific tumours. For example, there is frequent overexpression (>2-fold) of HDAC1 and HDAC6 in non-small-cell lung carcinoma, compared with cells from adjacent normal tissue25 and in human colon cancer explants, immunostaining has shown HDAC2 to be overexpressed relative to patient-matched normal tissue.26 Further, studies in mice have shown that loss of the tumour suppressor adenomatous polyposis coli (APC) leads to increased HDAC2 expression and that elevated HDAC2 is a key mechanism in prevention of apoptosis in APC-deficient cells, pointing towards HDAC2 as a particularly relevant potential target in colon cancer therapy.26 In the AML samples tested here, the pattern of dysregulation of HDAC2 is complex. Although HDAC2 expression averaged over all 23 AML samples was found to be increased 8–34-fold (depending on the control group), less than half the samples (10/23) had expression levels more than two times those of proliferating CD34+ cells, while several showed reduced HDAC2 expression.

Changes in HDAC expression in AML could be involved in mechanisms for abnormal cellular proliferation that operate through chromatin-independent pathways. The two HDACs that we identify as consistently overexpressed in AML cells, SIRT1 and HDAC6, both act on defined non-histone proteins. HDAC6 deacetylates tubulin9 while SIRT1 has been shown to deacetylate the tumour suppressor p53 and forkhead transcription factors such as Foxo3a.10, 11, 12, 13, 14 Acetylation of these proteins is an important regulator of their function and increased SIRT1 levels suppress both p53-dependent and forkhead-dependent apoptosis.13, 14, 27 SIRT1 knockout mice survive to adulthood only rarely, but survivors show hyperacetylation of p53 and increased apoptosis in thymocytes and spermatogonia.26, 28 In general, SIRT1 seems to promote cell survival and division in response to environmental stress or signals for terminal differentiation.29, 30 The presence of high levels of SIRT1 in AML cells may be an important contributor to their ability to survive toxic or differentiation-inducing, chemotherapeutic agents. Consistent with this, in AML blasts from four patients tested so far, the sample that was most sensitive to cell killing by VPA in culture also had an unusually low level of SIRT1, but not of any other HDAC, while HDAC5 was unusually high. Sensitivity could well be determined by the collective influence of several HDACs and further studies are required. However, it might now be worth exploring the therapeutic value of specific inhibitors of class III, NAD-dependent deacetylases,6, 29 particularly in those patients whose SIRT1 levels are strongly elevated.

Altered histone methylation induced by HDI

Exposure of primary AML cells to therapeutically achievable doses of VPA unexpectedly increased di- and trimethylation of H3 lysine 4, with similar kinetics to the changes observed in histone acetylation. H3 isoforms di- and trimethylated at lysine 4 have been associated with enhanced transcriptional activity31 and the parallel increases in H3K4 methylation and histone acetylation induced by HDI are potentially able to act synergistically in inducing transcription at selected loci. As histone lysine methylation is generally a more stable modification than acetylation, changes may persist for longer once inhibitor levels have fallen. Thus, modulating the levels of histone methylation may represent an important new therapeutic strategy in AML. Two mammalian histone methyltransferases (HMTs), SET7 and MLL, show specificity for H3K4 and so may mediate the observed effect.31 We have shown that both are expressed (though to variable extents) in AML primary cells and cell lines, and SET7 is among those genes that we find are most frequently overexpressed in AML samples. Both enzymes show only modest increases in expression (2-fold) in cells treated with HDI (unpublished results) and this is unlikely to account for the observed increases in H3K4 methylation. A more likely explanation is that acetylated H3 is preferentially methylated. This is consistent with experimental data showing enhanced MLL activity in vitro against H3 peptides acetylated at K9 or K14,32 and with our own findings that H3 molecules di- and trimethylated at K4 are found almost exclusively among the acetylated H3 isoforms, even prior to HDI treatment (unpublished results).

Changes in gene expression induced by HDI

Gene expression profiling has shown that HDIs, despite their ability to cause rapid and extensive histone hyperacetylation, cause changes in expression of only a small proportion of genes (10% or less), among which increased and decreased expression are equally frequent.33, 34, 35 In the present work, we have focused on the effect of HDI on expression of deacetylases themselves, with a view to deciding whether induction of one or more deacetylases might be a mechanism by which tumour cells could become resistant to HDI-based therapies. In the AML cell line HL60, and with all deacetylase inhibitors tested, HDAC11 was always the most highly induced deacetylase, with induction ranging from 60- to 200-fold after 8 h treatment. HDAC11 was also the HDAC that was most strongly induced by VPA treatment in both primary AML blasts and a second AML cell line, KG1a. Of the other 17 deacetylases, only SIRT4 showed consistently increased expression (9–16-fold) with all inhibitors. Selective induction of SIRT4 expression by HDI in cultured neuronal cells has recently been reported.36

Changes in HDAC expression patterns were influenced by inhibitor, cell type, concentration and length of treatment. For example, in contrast to the consistent induction of HDAC11, HDAC9 was strongly induced only by VPA (in a concentration-dependent manner) and not by sodium butyrate, a related short chain fatty acid, or by TSA or SAHA. Also, we consistently detected early induction of HDAC11 (15 min or 1 h) by TSA and SAHA, but never by butyrate or valproate. These findings indicate that there are subtle differences between even related HDI in their modes of action at the genome level and these differences are likely to determine how, or whether, the induced HDAC can influence the manner in which the cell responds to the drug.


  1. 1

    Moe-Behrens GH, Pandolfi PP . Targeting aberrant transcriptional repression in acute myeloid leukaemia. Rev Clin Exp Hematol 2003; 7: 139–159.

    CAS  PubMed  Google Scholar 

  2. 2

    Faretta M, Di Croce L, Pellici PG . Effects of the acute myeloid leukaemia-associated fusion proteins on nuclear architecture. Semin Haematol 2001; 38: 42–53.

    CAS  Article  Google Scholar 

  3. 3

    Melnick A, Licht JD . Histone deactylases as therapeutic targets in hematologic malignancies. Curr Opin Hematol 2002; 9: 322–332.

    Article  Google Scholar 

  4. 4

    De Ruijter AJM, Van Gennip AH, Caron HN, Kemp S, Van Kuilenburg ABP . Histone deacetylases (HDACs): characterization of the classical HDAC family. Biochem J 2003; 370: 737–749.

    CAS  Article  Google Scholar 

  5. 5

    Gregoretti IV, Lee Y-M, Goodson HV . Molecular evolution of the histone deacetylase family: functional implications of phylogenetic analysis. J Mol Biol 2004; 338: 17–31.

    CAS  Article  Google Scholar 

  6. 6

    Blander G, Guarente L . The Sir2 family of protein deacetylases. Annu Rev Biochem 2004; 73: 417–435.

    CAS  Article  Google Scholar 

  7. 7

    Gao L, Cueto MA, Asselbergs F, Atadja P . Cloning and functional characterisation of HDAC11, a novel member of the human histone deacetylase family. J Biol Chem 2002; 277: 25748–25755.

    CAS  Article  Google Scholar 

  8. 8

    Sengupta S, Seto E . Regulation of histone deacetylase activities. J Cell Biochem 2004; 93: 57–67.

    CAS  Article  Google Scholar 

  9. 9

    Matsuyama A, Shimazu T, Sumida Y, Saito A, Yoshimatsu Y, Seigneurin-Berny D et al. In vivo destabilization of dynamic microtubules by HDAC6 mediated deacetylation. EMBO J 2002; 21: 6820–6831.

    CAS  Article  Google Scholar 

  10. 10

    Langley E, Pearson M, Faretta M, Bauer UM, Frye RA, Minucci S et al. Human SIR2 deacetylates p53 and antagonizes PML/p53-induced cellular senescence. EMBO J 2002; 21: 2383–2396.

    CAS  Article  Google Scholar 

  11. 11

    Luo J, Nikolaev AY, Imai S, Chen D, Su F, Shiloh A et al. Negative control of p53 by Sir2 promotes cell survival under stress. Cell 2001; 107: 137–148.

    CAS  Article  Google Scholar 

  12. 12

    Vaziri H, Dessain SK, Ng EE, Imai SI, Frye RA, Pandita TK et al. hSIR2SIRT1 functions as an NAD-dependent p53 deacetylase. Cell 2002; 107: 149–159.

    Article  Google Scholar 

  13. 13

    Brunet A, Sweeney LB, Sturgill JF, Chua KF, Greer PL, Lin Y et al. Stress-dependent regulation of FOXO transcription factors by the SIRT1 deacetylase. Science 2004; 303: 2011–2105.

    CAS  Article  Google Scholar 

  14. 14

    Motta MC, Divecha N, Lemieux M, Kamel C, Chen D, Gu W et al. Mammalian SIRT1 represses forkhead transcription factors. Cell 2004; 116: 551–563.

    CAS  Article  Google Scholar 

  15. 15

    Marks PA, Richon VM, Miller T, Kelly WK . Histone deacetylase inhibitors. Adv Cancer Res 2004; 91: 137–168.

    CAS  Article  Google Scholar 

  16. 16

    Kelly WK, O’Connor OA, Marks PA . Histone deacetylase inhibitors: from target to clinical trials. Expert Opin Investig Drugs 2002; 11: 1695–1713.

    CAS  Article  Google Scholar 

  17. 17

    Barlow AL, van Drunen CM, Johnson CA, Tweedie S, Bird A, Turner BM . dSIR2 and dHDAC6: two novel, inhibitor resistant deacetylases in Drosophila melanogaster. Exp Cell Res 2001; 265: 90–103.

    CAS  Article  Google Scholar 

  18. 18

    Gottlicher M, Minucci S, Zhu P, Kramer OH, Schimpf A, Giavara S et al. Valproic acid defines a novel class of HDAC inhibitors inducing differentiation of transformed cells. EMBO J 2001; 20: 6969–6978.

    CAS  Article  Google Scholar 

  19. 19

    Turner BM, O’Neill LP, Allan IM . Histone H4 acetylation in human cells. Frequency of acetylation at different sites defined by immunolabeling with site specific antibodies. FEBS Lett 1989; 253: 141–145.

    CAS  Article  Google Scholar 

  20. 20

    White DA, Belyaev ND, Turner BM . Preparation of site specific antibodies to acetylated histones. Methods 1999; 19: 417–424.

    CAS  Article  Google Scholar 

  21. 21

    Johnson CA, White DA, Lavender JS, O’Neill LP, Turner BM . Human class I histone deacetylase complexes show enhanced catalytic activity in the presence of ATP and co-immunoprecipitate with the ATP-dependent chaperone protein Hsp70. J Biol Chem 2002; 277: 9590–9597.

    CAS  Article  Google Scholar 

  22. 22

    Drexler HG, Quentmeier H, MacLeod RAF, Uphoff CC, Hu Z-B . Leukemia cell lines: in vitro models for the study of acute promyelocytic leukaemia. Leukemia Res 1995; 19: 681–691.

    CAS  Article  Google Scholar 

  23. 23

    Kramer OH, Zhu P, Ostendorff HP, Golebiewski M, Tiefenbach J, Peters MA et al. The histone decetylase inhibitor valproic acid selectively induces proteasomal degradation of HDAC2. EMBO J 2003; 22: 3411–3420.

    Article  Google Scholar 

  24. 24

    Whitlock JP, Galeazzi D, Schulman H . Acetylation and calcium-dependent phosphorylation of histone H3 in nuclei from butyrate-treated HeLa cells. J Biol Chem 1983; 258: 1299–1304.

    CAS  PubMed  Google Scholar 

  25. 25

    Osada H, Tatematsu Y, Saito H, Yatabe Y, Mitsudomi T, Takahashi T . Reduced expression of class II histone deacetylase genes is associated with poor prognosis in lung cancer patients. Int J Cancer 2004; 112: 26–32.

    CAS  Article  Google Scholar 

  26. 26

    Zhu P, Martin E, Mengwasser J, Schlag P, Janssen K-P, Gottlicher M . Induction of HDAC2 expression upon loss of APC in colorectal tumorigenesis. Cancer Cell 2004; 5: 455–463.

    CAS  Article  Google Scholar 

  27. 27

    Araki T, Sasaki Y, Milbrandt J . Increased nuclear NAD biosynthesis and SIRT1 activation prevent axonal degeneration. Science 2004; 305: 1010–1013.

    CAS  Article  Google Scholar 

  28. 28

    Cheng HL, Mostoslavsky R, Saito S, Manis JP, Gu Y, Patel P et al. Developmental defects and p53 hyperacetylation in Sir2 homolog (SIRT1)-deficient mice. Proc Natl Acad Sci USA 2003; 100: 10794–10799.

    CAS  Article  Google Scholar 

  29. 29

    Fulco M, Schlitz RL, Iezzi S, King MT, Zhao P, Kishiwaya Y et al. Sir2 regulates skeletal muscle differentiation as a potential sensor of the redox state. Mol Cell 2003; 12: 51–62.

    CAS  Article  Google Scholar 

  30. 30

    Cohen HY, Miller C, Bitterman KJ, Wall NR, Hekking B, Kessler B et al. Calorie restriction promotes mammalian cell survival by inducing the SIRT1 deacetylase. Science 2004; 305: 390–396.

    CAS  Article  Google Scholar 

  31. 31

    Lachner M, O’Sullivan RJ, Jenuwein TJ . An epigenetic road map for histone lysine methylation. J Cell Sci 2003; 116: 2117–2124.

    CAS  Article  Google Scholar 

  32. 32

    Milne TA, Briggs SD, Brock HW, Martin ME, Gibbs D, Allis CD et al. MLL targets SET domain methyltransferase activity to HOX gene promoters. Mol Cell 2002; 10: 1107–1117.

    CAS  Article  Google Scholar 

  33. 33

    Van Lint C, Emiliani S, Verdin E . The expression of a small fraction of cellular gene is changed in response to histone hyperacetylation. Gene Exp 1996; 5: 245–254.

    CAS  Google Scholar 

  34. 34

    Suzuki H, Gabrielson E, Chen W, Anbazhagan R, van Engeland M, Weijenberg MP et al. A genomic screen for genes upregulated by demethylation and histone deacetylase inhibition in human colorectal cancer. Nat Genet 2002; 31: 141–149.

    CAS  Article  Google Scholar 

  35. 35

    Peart MJ, Smyth GK, van Laar RK, Bowtell DD, Richon VM, Marks PA et al. Identification and functional significance of genes regulated by structurally different histone deacetylase inhibitors. Proc Natl Acad Sci USA 2005; 102: 3697–3702.

    CAS  Article  Google Scholar 

  36. 36

    Kyrylenko S, Kyrylenko O, Suuronen T, Salminen A . Differential regulation of the Sir2 histone deacetylase genes by inhibitors of class I and II histone deacetylases. Cell Mol Life Sci 2003; 60: 1990–1997.

    CAS  Article  Google Scholar 

Download references


We thank Professor Paul Marks and Dr Victoria Richon (Aton Pharma Inc., New York) for generous provision of SAHA, Dorothy McDonald and Virginia Turner (National Blood Service, Birmingham) for invaluable help in the processing of primary samples, Christine James and Emma Yates for skilled technical assistance and the custodians of the National Cancer Research Network AML cell bank for access to archived material. This work was supported by the Leukaemia Research Fund (CMB, CC, BMT) and Cancer Research UK (BMT).

Author information



Corresponding author

Correspondence to B M Turner.

Additional information

Supplementary Information

Supplementary Information accompanies the paper on the Leukemia website (

Supplementary information

Rights and permissions

Reprints and Permissions

About this article

Cite this article

Bradbury, C., Khanim, F., Hayden, R. et al. Histone deacetylases in acute myeloid leukaemia show a distinctive pattern of expression that changes selectively in response to deacetylase inhibitors. Leukemia 19, 1751–1759 (2005).

Download citation


  • AML
  • valproic acid
  • HDACs
  • sirtuins
  • SIRT1
  • deacetylase inhibitors

Further reading


Quick links