HDACs (histone deacetylases) are considered to be among the most important enzymes that regulate gene expression in eukaryotic cells. In general, increased levels of histone acetylation are associated with increased transcriptional activity, whereas decreased levels are linked to repression of gene expression. HDACs associate with a number of cellular oncogenes and tumour-suppressor genes, leading to an aberrant recruitment of HDAC activity, which results in changes of gene expression, impaired differentiation and excessive proliferation of tumour cells. Therefore HDAC inhibitors are efficient anti-proliferative agents in both in vitro and in vivo pre-clinical models of cancer, making them promising anticancer therapeutics. In the present paper, we present the results of a medium-throughput screening programme aiming at the identification of novel HDAC inhibitors using HDAH (HDAC-like amidohydrolase) from Bordetella or Alcaligenes strain FB188 as a model enzyme. Within a library of 3719 compounds, several new classes of HDAC inhibitor were identified. Among these hit compounds, there were also potent inhibitors of eukaryotic HDACs, as demonstrated by an increase in histone H4 acetylation, accompanied by a decrease in tumour cell metabolism in both SHEP neuroblastoma and T24 bladder carcinoma cells. In conclusion, screening of a compound library using FB188 HDAH as model enzyme identified several promising new lead structures for further development.

INTRODUCTION

HDACs (histone deacetylases) are important enzymes in the regulation of gene expression in eukaryotic cells [1]. Together with HATs (histone acetyltransferases), HDACs maintain the delicate equilibrium between acetylated and deacetylated ϵ-amino groups of lysine residues in the N-terminal parts of histone proteins. In general, increased levels of histone acetylation are associated with increased transcriptional activity, whereas decreased levels of acetylation are associated with repression of gene expression [2]. Although histone deacetylation appears to have a fundamental role in regulating gene expression, HDAC inhibitors seem to directly affect transcription of only a relatively small number of genes [3,4]. Most of the genes regulated by HDAC inhibitors participate in the control of cell growth and survival, providing a mechanistic explanation for the anticancer properties of these inhibitors. In particular, activation of differentiation programmes, inhibition of the cell cycle and induction of apoptosis are key anti-tumour activities that make HDAC inhibitors promising anticancer agents. Several synthetic compounds and natural products have been reported to inhibit HDACs (reviewed in [57]). Furthermore, a small number of drug candidates are currently in Phase I–III clinical trials (reviewed in [8]), with the first drug, vorinostat (Zolinza; Merck & Co.), recently approved by the U.S. FDA (Food and Drug Administration). However, many currently available HDAC inhibitors are disadvantageous, e.g. with regard to their limited metabolic stability [9,10] or their cross-reactivity against other enzymes such as esterases [11]. Thus there is still a need for new HDAC inhibitors belonging to yet unexplored structural classes of compounds.

In the present paper, we report the identification of novel small-molecule HDAC inhibitors through medium-throughput screening of a compound library using a bacterial HDAH (HDAC-like amidohydrolase) [1214] as a model system. Several of the compounds specifically inhibit the bacterial enzyme. Other compounds proved to be pan-inhibitors that also inhibit eukaryotic HDACs and induce hyperacetylation in a neuroblastoma cell line (SHEP) and bladder carcinoma cells (T24). The novel inhibitor structures provide new lead structures for further development of improved anticancer drugs.

EXPERIMENTAL

Reagents

HDAH from Bordetella or Alcaligenes strain FB188 was expressed in Escherichia coli and was purified using Zn2+-IMAC (immobilized metal-ion-affinity chromatography) as described in [12]. The enzyme was stable for several weeks as 150 μg/ml solution in 10 mM potassium phosphate (pH 8.0) at 4 °C. Rat liver HDAC was purchased from Merck Biosciences, whereas human HDAC8 was expressed in E. coli and purified as described in [15]. Trypsin from porcine pancreas [Type IX-S, 13000–20000 BAEE units/mg (1 BAEE unit=ΔA253 of 0.001 per min with Nc-benzoyl-L-arginine ethyl ester as substrate at pH 7.6 at 25 °C)] was purchased from Sigma and, after solvation in trypsin buffer {50 mM Tris/HCl, pH 8.5, 100 mM NaCl and 0.1% PEG [poly(ethylene glycol)] 8000} to a solution of 10 mg/ml, was stable for several days at 4 °C. Boc-Lys(Ac)-AMC [where Boc is t-butoxycarbonyl, Lys(Ac) is acetyl-lysine, and AMC is 7-amino-4-methylcoumarin) (Bachem) and Tos-Gly-Pro-Arg-AMC (where Tos is tosyl) (Sigma) were dissolved to 30 mM in DMSO. Tos-Gly-Pro-Lys(Ac)-AMC was synthesized as described in [16]. Sodium valproate (Sigma) was dissolved to 1 M in PBS and stored at −20 °C.

Apart from RPMI 1640 medium and trypsin/EDTA (Cambrex), all cell culture reagents were purchased from Sigma.

Library screening

The compound collection was obtained from the Hans-Knöll-Institute (HKI), Jena, Germany, as solutions of approx. 10 μg/ml and was stored at −80 °C. The thawed substances were diluted 1:25 in assay buffer (10 mM potassium phosphate, pH 8.0). From each dilution, 20 μl was transferred into each well of black 96-well microtitre plates (Fluotrac200; Greiner). Screening of the compound library for inhibitors of the target enzyme HDAH was performed using the one-step version of the standard fluorigenic assay [16,17]. A robotic screening device (CyBi-Screen-Machine; CyBio) equipped with a Polarstar microtitre plate reader (BMG Labtech) was used.

To exclude false-positive candidates, control experiments were conducted with all hits from primary screening. For detection of compound autofluorescence, reactions without enzymes or substrate, but with the same concentration of each compound as used for screening, were measured. To discriminate real hits from substances predominantly inhibiting the reporter enzyme, trypsin-inhibition assays were performed for all primary hits at the same compound concentrations as used in primary screening. Samples of 150 μl of the mock reactions used for autofluorescence measurements were transferred to black 96-well microtitre plates, and 10 μl of trypsin (10 ng/ml diluted in trypsin buffer) was added. After 10 min at 30 °C, the reaction was started with 50 μl of trypsin substrate (200 μM Tos-Gly-Pro-Arg-AMC in trypsin buffer). Trypsin activity was recorded for 10 min as increasing AMC fluorescence at 30 °C in the plate reader at 460 nm after excitation at 390 nm. See Supplementary Figure S1 at http://www.BiochemJ.org/bj/413/bj4130143add.htm for the principle of the HDAC assay.

Determination of Z′-factor

For determination of the Z′-factor, a quality parameter used to judge assay performance, control enzyme inhibition experiments were carried out using the known HDAH inhibitor CypX (cyclopentyle-propionyle hydroxamic acid) [14]. In principle, the screening protocol was used, except that deacetylase activity was measured in the presence (48 wells) or absence (48 wells) of 10 μM model inhibitor instead of library compounds. The Z′-factor was calculated as described previously [18].

Determination of IC50 values

To assess the potency of hits, eight-point dose–response curves were measured in triplicate. For HDAH, the same protocol was applied as for primary screening. Reactions were monitored in the presence of 20 μl of hit compounds diluted in assay buffer to final concentrations of 1 nM–100 μM. The slopes of the activity curves obtained were corrected against blank reactions without HDAH. Dose–response curves for primary hits with rat liver HDAC and human HDAC8 were generated using a two-step HDAC assay protocol as described in [15,16]. IC50 values were calculated with the help of GraphPad Prism 3.0.

Cell culture

SHEP neuroblastoma cells were cultured in RPMI 1640 medium (Cambrex) supplemented with 10% FCS (fetal calf serum) (Sigma) at 37 °C under 5% CO2 [19]. PHFs (primary human fibroblasts) derived from a healthy donor were grown in DMEM (Dulbecco's modified Eagle's medium) containing 10% FCS and 1% non-essential amino acids at 37 °C under 5% CO2 as described previously [19].

Cellular metabolic activity

Metabolic activity was assessed with the tetrazolium derivative WST-1 (water-soluble tetrazolium salt 1), which is reduced to a soluble coloured formazan product by cellular mitochondrial dehydrogenases. The metabolic activity measured corresponds to the number of living cells and is used as a screening assay for cell viability [20]. Cells were plated at 2×103 cells/well into 96-well plates and were exposed to different concentrations of compounds 46F08 and 50B09 for 48 h. Compound 46F08 was dissolved in methanol to give a stock solution of 100 mM, and compound 50B09 was dissolved in DMSO to give a stock solution of 10 mM. Compounds were diluted further in cell culture medium to give the final concentrations (see Figure 2). In control experiments, cells were treated with 0.01% methanol and 0.25% DMSO corresponding to the highest solvent concentration applied in the compound-treated cells. WST-1 reagent was added to a final dilution of 1:10 to the wells, with those with medium only serving as blank control. After 60 min at 37 °C, absorbance at 450 nm was determined with a plate reader (BMG Labtech) and corrected for absorbance at 650 nm. Blank control values were subtracted from all results, and values for untreated cells were set at 100%.

Western blot analysis

The level of acetylated histone H4 was assessed as described previously [21]. Protein concentrations of the cell lysates were determined using the BCA (bicinchoninic acid) protein assay (Pierce). Briefly, 5 μl of cell lysate was incubated for 1 h at 37 °C with 200 μl of BCA mixture. Absorbance was measured with a plate reader (BMG Labtech). The following primary antibodies were used: anti-β-actin mouse monoclonal (clone AC-15; Sigma) and anti-(acetyl-histone H4) rabbit polyclonal (Upstate). Secondary antibodies were HRP (horseradish peroxidase)-conjugated goat anti-mouse (Dianova) and donkey anti-rabbit (Promega) IgG.

RESULTS

Medium-throughput screening and hit confirmation

For medium-throughput screening, the one-step version of the standard fluorigenic HDAC assay was used [16,17] because the enzyme HDAH is resistant against trypsin digestion. For the assay conditions used, a Z′-factor of 0.75 was determined as part of the overall screening program. A total of 3719 compounds containing mostly natural products were tested at a concentration of 0.04 μg/ml. A total of 38 compounds (1% of total tested samples) that displayed more than 75% inhibition were selected as primary hits (see Supplementary Figure S2 at http://www.BiochemJ.org/bj/413/bj4130143add.htm). From the selected primary hits, 22 were confirmed to be HDAH inhibitors and do not show autofluorescence or trypsin inhibition. Unfortunately, six out of the 22 hits were no longer available from the supplier. The results of all 16 remaining primary hits were validated by repeated measurements in the screening assay and subsequently by dose–response curves using final inhibitor concentrations ranging from 1 nM to 100 μM. On the basis of the selection criterion of IC50<10 μM for the HDAH assay, a total of ten compounds (0.26% of total tested samples) were selected as confirmed hits for follow-up studies (Table 1). The most efficient HDAH inhibitors exhibited IC50 values of 610 nM (45E09) and 550 nM (46F08).

Table 1
Most active hit compounds in the screening for HDAH inhibitors

NA, no activity at 50 μM inhibitor.

  IC50 (μM) 
Structure Number HDAH RL HDAC HDAC8 
graphic
 
21E08 1.08 NA (4)* 
graphic
 
22G03 9.24 NA (46)* 
graphic
 
23F10 1.12 (10)† (10)† 
graphic
 
26H05 1.61 (6)† (10)† 
graphic
 
33D02 2.71 (12)* (19)* 
graphic
 
33C08 3.66 (10)* (12)* 
graphic
 
40H03 6.18 (22)* NA 
graphic
 
45E09 0.61 (13)* (4)* 
graphic
 
46F08 0.55 0.64 0.10 
graphic
 
50B09 2.77 4.00 0.31 
  IC50 (μM) 
Structure Number HDAH RL HDAC HDAC8 
graphic
 
21E08 1.08 NA (4)* 
graphic
 
22G03 9.24 NA (46)* 
graphic
 
23F10 1.12 (10)† (10)† 
graphic
 
26H05 1.61 (6)† (10)† 
graphic
 
33D02 2.71 (12)* (19)* 
graphic
 
33C08 3.66 (10)* (12)* 
graphic
 
40H03 6.18 (22)* NA 
graphic
 
45E09 0.61 (13)* (4)* 
graphic
 
46F08 0.55 0.64 0.10 
graphic
 
50B09 2.77 4.00 0.31 
*

Percentage inhibition at 50 μM.

Percentage inhibition at 10 μM.

Selectivity of hit compounds

To determine whether the confirmed hit compounds selectively inhibit only HDAH, two other HDAC preparations were tested: rat liver HDAC (Calbiochem), containing mainly HDAC1, HDAC2 and HDAC3 [15], and recombinant human HDAC8 [15]. Interestingly, seven compounds (21E08, 23F10, 26H05, 33D02, 33C08, 40H03 and 45E09) showed a marked selectivity for HDAH (Table 1). Compound 22G03 exhibited a clear preference for HDAH, but also inhibited HDAC8 with a 5-fold increased IC50 value. Two compounds (46F08 and 50B09) proved to be rather unspecific pan-inhibitors. Both of these compounds showed a marked preference for human HDAC8.

Competition assay with selected compounds

For several hit compounds, sufficient material was available to perform competition binding assays as reported previously [22]. Briefly, a fluorescent HDAC inhibitor [FAHA (2-furylacryloylhydroxamate)], which shows FRET (fluorescence resonance energy transfer) with tryptophan residues of the enzyme was used. Upon competition with other HDAC inhibitors, the displacement of the fluorescent inhibitor is accompanied by a decrease in FRET. In our experiments, compounds 21E08, 22G03, 33D02 and 33C08 were able to displace fluorescent inhibitor FAHA from the enzyme complex (Figure 1).

FRET-based competition assay for selected hit compounds

Figure 1
FRET-based competition assay for selected hit compounds

The fluorescent HDAC inhibitor FAHA, which shows FRET with tryptophan residues of the enzyme, was used. Upon competition with other HDAC inhibitors, the displacement of the fluorescent inhibitor is accompanied by a decrease in FRET. Binding degrees and binding constants were calculated as described in [22]. SAHA, suberoylanilide hydroxamic acid.

Figure 1
FRET-based competition assay for selected hit compounds

The fluorescent HDAC inhibitor FAHA, which shows FRET with tryptophan residues of the enzyme, was used. Upon competition with other HDAC inhibitors, the displacement of the fluorescent inhibitor is accompanied by a decrease in FRET. Binding degrees and binding constants were calculated as described in [22]. SAHA, suberoylanilide hydroxamic acid.

Structure classes of inhibitors

The ten hit compounds listed in Table 1 fall into different structural classes. Surprisingly, only half of these inhibitors match the typical structure of many HDAC inhibitors, i.e. a cap and a zinc-binding moiety linked by a long spacer group [23]. Examples of this type are methyl ketones 45E09 and 23F10, trifluoromethyl ketone 46F08 and carboxylic acids 22G03 and 33C08 (Table 1). All other hit compounds belong to structural classes that have not been described as HDAC inhibitors so far.

The HDAH-selective inhibitors 45E09 and 23F10 (Table 1) showed IC50 values of 0.61 and 1.12 μM respectively and comprise a pyran-4-one or a p-benzoquinone ring as cap group, a C9-10 aliphatic spacer and a methyl ketone moiety. The last of these, in principle, may interact with the active-site Zn2+ ion, as this has been demonstrated for the trifluoromethyl ketone moiety [24]. Because an additional series of related compounds was available, we were able to identify structural elements that were important for the activity of this inhibitor class (Table 2). Simultaneous changes in the substituents of the pyran-4-one ring, i.e. replacement of the ethyl groups in the 3- and 5-position by methyl groups and the introduction of a hydroxy group replacing the 6-methoxy group, led to a 7-fold increase in the IC50 value for An2 compared with 23F10. The favourable effect of at least the ethyl group in the 3-position as compared with a methyl group was also observed in the context of the structural scaffold shared by compounds An3, An4 and An5. Changing the pyran-4-one ring (23F10) into a p-benzoquinone ring (45E09) obviously does not influence the inhibitory activity significantly. The opposite seems to be true for the length of the spacer. An inhibitor with a C6-spacer such as An1 appeared to be inactive, whereas inhibitors such as An2, 23F10 or 45E09 with a C9-10 aliphatic spacer show significant activity.

Table 2
Derivatives of compound 45E09
Structure Number IC50 (μM) (HDAH) 
graphic
 
45E09 0.61 
graphic
 
23F10 1.12 
graphic
 
 An1 ≥50 
graphic
 
 An2 7.76 
graphic
 
 An3 ≥50 
graphic
 
 An4 ≥50 
graphic
 
 An5 11.98 
Structure Number IC50 (μM) (HDAH) 
graphic
 
45E09 0.61 
graphic
 
23F10 1.12 
graphic
 
 An1 ≥50 
graphic
 
 An2 7.76 
graphic
 
 An3 ≥50 
graphic
 
 An4 ≥50 
graphic
 
 An5 11.98 

Compound 46F08 (Table 1) [25] is a pan-inhibitor with IC50 values for HDAH, rat liver HDAC and human HDAC8 of 0.55, 0.64 and 0.10 μM respectively. It belongs to the group of trifluoromethyl ketone-based HDAC inhibitors that have been described previously [10]. From a recent crystallographic study, it is known that the oxygen atom of the trifluoromethyl ketone moiety binds to the zinc atom of the enzyme and also that the fluorine atoms participate in various interactions with the enzyme [24]. In our screening, it was the most active inhibitor of FB188 HDAH, with an IC50 of 0.55 μM for HDAH. For the rat liver HDAC preparation, the IC50 was similar to that of HDAH. In contrast, the IC50 for human HDAC8 showed a 5-fold decrease, i.e. a value of 0.10 μM. Thus the potency of 46F08 is at least similar, if not slightly improved, compared with previously published trifluoromethyl ketone-based HDAC inhibitors [10], which were already the result of subsequent cycles of optimization.

In addition, resembling the more extended type of HDAC inhibitor, compounds 22G03 and 33C08 show a high selectivity for HDAH, with IC50 values of 9.24 and 3.66 μM respectively. As shown in Table 1, compounds 22G03 and 33C08 (isolated from yeast Candida bombicola A.T.C.C. 22214) [26] are marked by the presence of carboxylic acid moiety at the end of a long spacer, suggesting that this moiety may contact the active-site Zn2+ ion. 22G03 contains an epoxide group in the middle of an aliphatic chain reminiscent of depudecin [27,28], whereas 33C08 belongs to the class of sophorose lipids. The view that both compounds bind directly to the active-site pocket of the enzyme is supported by the finding that 22G03 and 33C08 are active in the FAHA competition assay (Figure 1).

However, the structural similarity between 33C08 and 40H03 (isolated from yeast C. bombicola A.T.C.C. 22214) [29] argues against the importance of direct binding to the active-site Zn2+ ion (see Table 1). Compound 40H03 shows an IC50 value of 6.18 μM for HDAH and contains a lipid chain without a polar end group, such as carboxylic acid functionality. In agreement with the view that a possible direct contact to the Zn2+ ion in 33C08 does not significantly contribute to binding, the IC50 of 40H03 (6.18 μM) shows only an increase by a factor of 2 compared with that of 33C08 (3.66 μM). Further support of this view comes from the fact that a cyclic ‘lactonic’ derivative of 33C08 termed 33D02 (isolated from yeast C. bombicola A.T.C.C. 22214) [26] was also among the most active hits. In this molecule, no free carboxylic acid end is present. In addition, it shows a second acetylation of the sophorose moiety. Similarly to the open-chain derivatives, it proved to be a selective HDAH inhibitor with an IC50 value of 2.71 μM.

Re-examination of the compound library indicated that it contained several compounds similar to 33C08 and 40H03, which appeared to be inactive under screening conditions (results not shown). Inactive variants of 40H03 included (i) derivatives with altered lipid chain length (±C2), and (ii) derivatives combining a second sophorose acetylation with either a longer lipid chain (±C2) or hydroxylation at position 12 of the lipid chain. Three types of inactive derivatives of 33C08 were identified: (i) a derivative with glucose replacing sophorose, (ii) a derivative with trehalose replacing sophorose and at the same time lacking the carboxylic acid moiety, and (iii) derivatives containing a saturated lipid chain in which the carboxylic acid moiety was replaced by butyl amide or higher amide moieties. In summary, it appears that the nature of the sugar moiety as well as the length and the nature, but not necessarily the cyclization, of the lipid chain have a significant influence on inhibitory activity.

Compound 26H05 (Table 1) is a chromone also termed pilopleurin originally isolated from the roots of Pilopleura kozo-poljanskii Schischk [30]. In our experiments, compound 26H05 proved to be rather selective for FB188 HDAH (IC50=1.61 μM), not inhibiting eukaryotic HDACs. Unfortunately, no closely related compounds were available for structure–activity-relationship studies.

Actinomycin X (compound 21E08; originally isolated by H. Zahner from Streptomyces michiganensis A.T.C.C. 14970; Table 1) also was among the most potent hit compounds, with an IC50 value of 1.08 μM for HDAH. Basically, no inhibitory activity was observed for rat liver HDAC and human HDAC8. Actinomycin X is a close relative of actinomycin D, an approved anti-tumour agent (Lyovac-Cosmegen®, Cosmegen®) [31,32]. In contrast with actinomycin X, actinomycin D carries an additional amino group in the phenoxazone system and an L-proline residue replacing one sarcosine to give rise to identical pentapeptidic cycles. Our results with actinomycin X prompted us to also test actinomycin D for possible HDAC-inhibitory activity. As a result, it appeared that actinomycin D inhibited FB188 HDAH with an IC50 of 81 μM and only weakly blocked rat liver HDAC (IC50 ≥100 μM).

Hit compound 50B09 (Table 1) belongs to the class of pyrazolopyrimidinones. It has been identified previously as a potent inhibitor of coxsackievirus B3 replication [33]. In the present study, it was among the two inhibitors of FB188 HDAH that also blocked eukaryotic HDAC preparations. For HDAH, rat liver HDAC and human HDAC8 IC50 values of 2.77, 4.00 and 0.31 μM respectively were measured.

Effects of hit compounds on tumour cells

Because 46F08 and 50B09 showed significant activity towards eukaryotic HDACs, we next focused our attention on the inhibitory activity of both compounds in cell culture models. Treatment of SHEP neuroblastoma cells with 46F08 (0.5–4 μM) indeed generated an increase in acetylation of histone H4, indicating HDAC inhibition. Similar results were obtained for T24 cells (see Supplementary Figure S3 at http://www.BiochemJ.org/bj/413/bj4130143add.htm). Acetylation, however, was weaker compared with that induced by trichostatin A (Figure 2B). In contrast, treatment with compound 50B09 did not show an increase in histone H4 acetylation at concentrations of 25–100 μM. In this case, it rather seems that 50B09 suppressed the increased level of histone H4 acetylation caused by DMSO [34,35], which was used as the solvent.

Cellular effects produced by hit compounds 46F08 and 50B09

Figure 2
Cellular effects produced by hit compounds 46F08 and 50B09

(A) Influence on cell metabolic activity. Upper panel: SHEP neuroblastoma cells and normal PHFs were treated with compound 46F08 for 48 h with the indicated concentrations. Cellular metabolic activity was determined using WST-1 assay. Vehicle: cells were treated with solvent only (0.01% methanol), corresponding to the methanol concentration applied in 10 μM 46F08-treated cells. Values for untreated cells were set to 100%. Lower panel: SHEP neuroblastoma cells were treated with compound 50B09 for 48 h with the indicated concentrations. Vehicle: cells were treated with solvent only (0.1% DMSO), corresponding to the DMSO concentration applied in 10 μM 50B09-treated cells. (B) Influence on histone H4 acetylation. SHEP cells were treated for 4 h with indicated compounds and harvested for Western blot analysis using an anti-(acetyl-histone H4) antibody (α-Ac-H4). Methanol (0.004%) was used as a solvent control for 4 μM 46F08-treated cells. DMSO (0.25, 0.5 and 1%) was used as solvent controls for 25, 50 and 100 μM 50B09-treated cells respectively. Note that DMSO, in contrast with methanol, harbours HDAC-inhibiting activity. β-Actin served as a loading control. TSA, trichostatin A.

Figure 2
Cellular effects produced by hit compounds 46F08 and 50B09

(A) Influence on cell metabolic activity. Upper panel: SHEP neuroblastoma cells and normal PHFs were treated with compound 46F08 for 48 h with the indicated concentrations. Cellular metabolic activity was determined using WST-1 assay. Vehicle: cells were treated with solvent only (0.01% methanol), corresponding to the methanol concentration applied in 10 μM 46F08-treated cells. Values for untreated cells were set to 100%. Lower panel: SHEP neuroblastoma cells were treated with compound 50B09 for 48 h with the indicated concentrations. Vehicle: cells were treated with solvent only (0.1% DMSO), corresponding to the DMSO concentration applied in 10 μM 50B09-treated cells. (B) Influence on histone H4 acetylation. SHEP cells were treated for 4 h with indicated compounds and harvested for Western blot analysis using an anti-(acetyl-histone H4) antibody (α-Ac-H4). Methanol (0.004%) was used as a solvent control for 4 μM 46F08-treated cells. DMSO (0.25, 0.5 and 1%) was used as solvent controls for 25, 50 and 100 μM 50B09-treated cells respectively. Note that DMSO, in contrast with methanol, harbours HDAC-inhibiting activity. β-Actin served as a loading control. TSA, trichostatin A.

HDAC inhibitors are known to have anticancer effects. Therefore the influence of hit compounds 46F08 and 50B09 on the metabolic activity of SHEP neuroblastoma cells was investigated using the WST-1 assay. This assay quantifies the enzymatic activity of mitochondrial dehydrogenases in the sample and correlates with the number of viable cells [20].

Treatment of cells with compound 46F08 resulted in a concentration-dependent pronounced reduction of cell metabolic activity even at low concentrations of 0.1–4 μM. Similar results were obtained for T24 cells (see Supplementary Figure S3 at http://www.BiochemJ.org/bj/413/bj4130143add.htm). Compared with the well-established HDAC inhibitor VPA (valproic acid) already used in clinical trials, 46F08 was approx. 104-fold more active (Figure 2). Because tumour-cell-selective activity of HDAC inhibitors has been described [36], the effect of 46F08 on the metabolic activity of normal non-transformed human skin fibroblasts was investigated. 46F08 did not obviously influence fibroblast metabolic activity, even at the highest concentration applied, suggesting tumour-cell-selective activity of the compound (Figure 2). In contrast, 50B09 elicited no effect on the metabolic activity of SHEP cells at concentrations up to 10 μM (Figure 2A).

DISCUSSION

We have successfully confirmed the feasibility of our screening assay to efficiently screen a large compound library. Using a robotic workstation (CyBi-Screen-Machine), the assay exhibited a very high Z′-factor of 0.75, indicating an excellent assay. The screening identified several compounds capable of inhibiting HDAH reporter enzyme activity. Here, ten out of 3719 compounds showed IC50 values <10 μM. The identification of inhibitors 46F08 and 50B09, both of which are highly potent inhibitors of human HDAC8, proved that FB188 HDAH indeed can be used as a model system to also find inhibitors of human HDAC. Altogether, hit compounds fall into different structural classes. Only half of the inhibitors identified followed the simple HDAC inhibitor architecture, realized, e.g., in SAHA (suberoylanilide hydroxamic acid), in which a cap and a zinc-binding moiety are linked by an extended spacer. For the group of ketone-based inhibitors with pyran-4-one or p-benzoquinone rings as cap moieties, a preliminary structure–activity study showed that structure elements such as alkyl substituents in the ring system, linker length and the potential zinc-binding moiety are important for inhibitory activity. However, on the basis of the relatively small number of derivatives examined so far, it is not justified to draw more specific conclusions.

One of our hit compounds, 46F08, belongs to the class of trifluoromethyl ketones. It has been shown that electrophilic ketones such as trifluoromethyl ketones are potent inhibitors of zinc-dependent enzymes [10,37,38]. In contrast with simple methyl ketones, trifluoromethyl ketones show an increase in electrophilicity at the carbonyl carbon with favourable effects on hydrogen-bonding capabilities of the carbonyl oxygen. The recently solved crystallographic structure of a complex of FB188 HDAH with a trifluoromethyl ketone inhibitor [24] also revealed that the fluorine atoms actively participate in direct protein binding. Compound 46F08 not only proved to be a potent inhibitor of FB188 HDAH, but also blocked rat liver HDAC and human HDAC8 with submicromolar IC50 values. Furthermore, it significantly reduced the metabolic activity of SHEP neuroblastoma cells, which correlates with a decrease in the number of viable cells. Interestingly, normal human fibroblasts were not affected by the compound, suggesting tumour-cell-selective activity of 46F08, as has been described for other HDAC inhibitors [36]. In T24 bladder carcinoma cells, a similar effect was observed. Thus 46F08 is an interesting candidate for further development. However, previously developed trifluoromethyl ketones had exhibited an unfavourably short half-life in animals [10]. It remains to be shown if compound 46F08 is also readily metabolized. In this regard, it will be interesting to test derivatives of 46F08 with substituents at the CH2- moiety next to the carbonyl group.

Compounds 40H03, 33C08 and 33D02 all belong to the class of sophorose lipids, a group of microbial glycolipids produced by yeasts [39]. Comparison with structurally similar, but inactive, derivatives of the compound library suggested that these molecules do not directly bind to the active-site Zn2+ ion. Interestingly, earlier work by Isoda et al. [40,41] and Scholz and co-workers [42,43] had already demonstrated that certain sophorolipids induce cell differentiation of human leukaemia cells by an ‘unknown mechanism’. We may now hypothesize that the anticancer activity of sophorolipids and maybe also of other glycolipids [40,44,45] is due to their inhibition of certain HDACs, particularly those with similarities to FB188 HDAH.

The same may be true for hit compound 26H05. In our experiments, it did show selectivity for HDAH and not for eukaryotic HDACs. However, several chromones have already been described as anticancer agents [46] with proposed mechanisms as diverse as inhibition of breast cancer resistance protein ABCG2 (ATP-binding cassette transporter G2) [47] or inhibition of phosphoinositide 3-kinase [48]. It cannot be excluded, however, that chromones may also work as inhibitors of certain HDACs, particularly those with similarities to HDAH.

Neither actinomycin X nor actinomycin D has been described as an HDAC inhibitor so far. On the contrary, actinomycin D has been shown to intercalate into DNA by virtue of its planar phenoxazone ring system [49,50]. As a consequence, it is believed that actinomycin D blocks DNA-dependent RNA polymerase [51] and poisons topoisomerase II, which results in double-strand DNA breaks [52]. In principle, the action of actinomycin D should not be restricted to a certain phase in the cell cycle. However, there is experimental evidence that it is by far most active in the S- and G2-phases, indicating that it also may act in a different way [53]. In the light of our data, it can be speculated that compounds such as actinomycin X and D may also work as inhibitors of certain HDACs.

With compound 50B09, another new class of HDAC inhibitors has been discovered in our screening programme. Previously, 50B09 and related pyrazolopyrimidinones had been identified as potent inhibitors of coxsackievirus B3 replication [33]. In this previous study, cytotoxicity was also determined. Although compound 50B09 did not show a significant cytotoxicity against HeLa cells or a pronounced growth inhibition in L-929 and K-562 cells, derivatives of 50B09 did show inhibitory activity on these cell lines [33]. In the present study, we also measured the effect of 50B09 on eukaryotic cells. Despite a good inhibitory activity against rat liver and human HDAC8, we detected neither a pronounced effect on the metabolic activity of SHEP neuroblastoma cells nor an increase in histone H4 acetylation. Structure–activity-relationship studies are under way to identify derivatives with improved anti-tumour cell activities.

In summary, several inhibitors have been discovered that block enzymes belonging to the HDAC family of proteins. Most of the identified inhibitor structures have not been described as HDAC inhibitors so far, and therefore may be a good starting point for further development towards future generations of HDAC inhibitors with possible applications in the fields of malignant and other diseases.

This research was in part supported by grant BioFuture 0311852 from the Bundesministerium für Forschung und Technologie, Germany, to A. Schweinhorst, and by a grant of the Nationales Genomforschungsnetzwerk (NGFN2) to O. W. A. Schweinhorst thanks R. Jansen (Helmholtz Centre for Infection Research, Braunschweig, Germany), H.-P. Fiedler (Mikrobiologisches Institut, Universität Tübingen, Tübingen, Germany), G.-V. Röschenthaler (Institute of Inorganic and Physical Chemistry, Universität Bremen, Bremen, Germany) and D. V. Sevenard (Institute of Inorganic and Physical Chemistry, Universität Bremen, Bremen, Germany) for providing hit compounds and for valuable discussions.

Abbreviations

     
  • AMC

    7-amino-4-methylcoumarin

  •  
  • BCA

    bicinchoninic acid

  •  
  • FAHA

    2-furylacryloylhydroxamate

  •  
  • FCS

    fetal calf serum

  •  
  • FRET

    fluorescence resonance energy transfer

  •  
  • HDAC

    histone deacetylase

  •  
  • HDAH

    HDAC-like amidohydrolase

  •  
  • Lys(Ac)

    acetyl-lysine

  •  
  • PHF

    primary human fibroblast

  •  
  • TBS

    Tris-buffered saline

  •  
  • Tos

    tosyl

  •  
  • WST-1

    water-soluble tetrazolium salt 1

References

References
1
Hildmann
C.
Riester
D.
Schwienhorst
A.
Histone deacetylases: an important class of cellular regulators with a variety of functions
Appl. Microbiol. Biotechnol.
2007
, vol. 
75
 (pg. 
487
-
497
)
2
Davie
J. R.
Spencer
V. A.
Signal transduction pathways and the modification of chromatin structure
Prog. Nucleic Acid Res. Mol. Biol.
2001
, vol. 
65
 (pg. 
299
-
340
)
3
Mariadason
J. M.
Corner
G. A.
Augenlicht
L. H.
Genetic reprogramming in pathways of colonic cell maturation induced by short chain fatty acids: comparison with trichostatin A, sulindac, and curcumin and implications for chemoprevention of colon cancer
Cancer Res.
2000
, vol. 
60
 (pg. 
4561
-
4572
)
4
Peart
M. J.
Smyth
G. K.
van Laar
R. K.
Bowtell
D. D.
Richon
V. M.
Marks
P. A.
Holloway
A. J.
Johnstone
R. W.
Identification and functional significance of genes regulated by structurally different histone deacetylase inhibitors
Proc. Natl. Acad. Sci. U.S.A.
2005
, vol. 
102
 (pg. 
3697
-
3702
)
5
Monneret
C.
Histone deacetylase inhibitors
Eur. J. Med. Chem.
2005
, vol. 
40
 (pg. 
1
-
13
)
6
Moradei
O.
Maroun
C. R.
Paquin
I.
Vaisburg
A.
Histone deacetylase inhibitors: latest developments, trends and prospects
Curr. Med. Chem. Anticancer Agents
2005
, vol. 
5
 (pg. 
529
-
560
)
7
Minucci
S.
Pelicci
P. G.
Histone deacetylase inhibitors and the promise of epigenetic (and more) treatments for cancer
Nat. Rev. Cancer
2006
, vol. 
6
 (pg. 
38
-
51
)
8
Riester
D.
Hildmann
C.
Schwienhorst
A.
Histone deacetylase inhibitors: turning epigenic mechanisms of gene regulation into tools of therapeutic intervention in malignant and other diseases
Appl. Microbiol. Biotechnol.
2007
, vol. 
75
 (pg. 
499
-
514
)
9
Veale
C. A.
Bernstein
P. R.
Bohnert
C. M.
Brown
F. J.
Bryant
C.
Damewood
J. R.
Jr
Earley
R.
Feeney
S. W.
Edwards
P. D.
Gomes
B.
, et al. 
Orally active trifluoromethyl ketone inhibitors of human leukocyte elastase
J. Med. Chem.
1997
, vol. 
40
 (pg. 
3173
-
3181
)
10
Frey
R. R.
Wada
C. K.
Garland
R. B.
Curtin
M. L.
Michaelides
M. R.
Li
J.
Pease
L. J.
Glaser
K. B.
Marcotte
P. A.
Bouska
J. J.
, et al. 
Trifluoromethyl ketones as inhibitors of histone deacetylase
Bioorg. Med. Chem. Lett.
2002
, vol. 
12
 (pg. 
3443
-
3447
)
11
Moreth
K.
Riester
D.
Hildmann
C.
Hempel
R.
Wegener
D.
Schober
A.
Schwienhorst
A.
An active site tyrosine residue is essential for amidohydrolase but not for esterase activity of a class 2 histone deacetylase-like bacterial enzyme
Biochem. J.
2007
, vol. 
401
 (pg. 
659
-
665
)
12
Hildmann
C.
Ninkovic
M.
Dietrich
R.
Wegener
D.
Riester
D.
Zimmermann
T.
Birch
O. M.
Bernegger
C.
Loidl
P.
Schwienhorst
A.
A new amidohydrolase from Bordetella or Alcaligenes strain FB188 with similarities to histone deacetylases
J. Bacteriol.
2004
, vol. 
186
 (pg. 
2328
-
2339
)
13
Nielsen
T. K.
Hildmann
C.
Dickmanns
A.
Schwienhorst
A.
Ficner
R.
Crystal structure of a bacterial class 2 histone deacetylase homologue
J. Mol. Biol.
2005
, vol. 
354
 (pg. 
107
-
120
)
14
Hildmann
C.
Wegener
D.
Riester
D.
Hempel
R.
Schober
A.
Merana
J.
Giurato
L.
Guccione
S.
Nielsen
T. K.
Ficner
R.
Schwienhorst
A.
Substrate and inhibitor specificity of class 1 and class 2 histone deacetylases
J. Biotechnol.
2006
, vol. 
124
 (pg. 
258
-
270
)
15
Riester
D.
Wegener
D.
Hildmann
C.
Schwienhorst
A.
Members of the histone deacetylase superfamily differ in substrate specificity towards small synthetic substrates
Biochem. Biophys. Res. Commun.
2004
, vol. 
324
 (pg. 
1116
-
1123
)
16
Wegener
D.
Hildmann
C.
Riester
D.
Schwienhorst
A.
Improved fluorogenic histone deacetylase assay for high-throughput-screening applications
Anal. Biochem.
2003
, vol. 
321
 (pg. 
202
-
208
)
17
Wegener
D.
Wirsching
F.
Riester
D.
Schwienhorst
A.
A fluorogenic histone deacetylase assay well suited for high-throughput activity screening
Chem. Biol.
2003
, vol. 
10
 (pg. 
61
-
68
)
18
Zhang
J. H.
Chung
T. D.
Oldenburg
K. R.
A simple statistical parameter for use in evaluation and validation of high throughput screening assays
J. Biomol. Screen.
1999
, vol. 
4
 (pg. 
67
-
73
)
19
Deubzer
H. E.
Ehemann
V.
Westermann
F.
Heinrich
R.
Mechtersheimer
G.
Kulozik
A. E.
Schwab
M.
Witt
O.
Histone deacetylase inhibitor Helminthosporium carbonum (HC)-toxin suppresses the malignant phenotype of neuroblastoma cells
Int. J. Cancer
2008
, vol. 
122
 (pg. 
1891
-
1900
)
20
Cook
J. A.
Mitchell
J. B.
Viability measurements in mammalian cell systems
Anal. Biochem.
1989
, vol. 
179
 (pg. 
1
-
7
)
21
Witt
O.
Monkemeyer
S.
Ronndahl
G.
Erdlenbruch
B.
Reinhardt
D.
Kanbach
K.
Pekrun
A.
Induction of fetal hemoglobin expression by the histone deacetylase inhibitor apicidin
Blood
2003
, vol. 
101
 (pg. 
2001
-
2007
)
22
Riester
D.
Hildmann
C.
Schwienhorst
A.
Meyer-Almes
F. J.
Histone deacetylase inhibitor assay based on fluorescence resonance energy transfer
Anal. Biochem.
2007
, vol. 
362
 (pg. 
136
-
141
)
23
Jung
M.
Inhibitors of histone deacetylase as new anticancer agents
Curr. Med. Chem.
2001
, vol. 
8
 (pg. 
1505
-
1511
)
24
Nielsen
T. K.
Hildmann
C.
Riester
D.
Wegener
D.
Schwienhorst
A.
Ficner
R.
Complex structure of a bacterial class 2 histone deacetylase homologue with a trifluoromethylketone inhibitor
Acta Crystallogr. Sect. F Struct. Biol. Cryst. Commun.
2007
, vol. 
63
 (pg. 
270
-
273
)
25
Pashkevich
K. I.
Sevenard
D. V.
Khomutov
O. G.
Reactions of 2-polyfluoroacylcyclohexanones with 1,2-diaminoarenes
Russ. Chem. Bull.
1999
, vol. 
48
 (pg. 
557
-
560
)
26
Asmer
H.-J.
Lang
S.
Wagner
F.
Wray
V.
Microbial production, structure elucidation and bioconversion of sophorose lipids
J. Am. Oil Chem. Soc.
1988
, vol. 
65
 (pg. 
1460
-
1466
)
27
Matsumoto
M.
Matsutani
S.
Sugita
K.
Yoshida
H.
Hayashi
F.
Terui
Y.
Nakai
H.
Uotani
N.
Kawamura
Y.
Matsumoto
K.
, et al. 
Depudecin: a novel compound inducing the flat phenotype of NIH3T3 cells doubly transformed by ras- and src-oncogene, produced by Alternaria brassicicola
J. Antibiot.
1992
, vol. 
45
 (pg. 
879
-
885
)
28
Privalsky
M. L.
Depudecin makes a debut
Proc. Natl. Acad. Sci. U.S.A.
1998
, vol. 
95
 (pg. 
3335
-
3337
)
29
Brakemeier
A.
Lang
S.
Wullbrandt
D.
Merschel
L.
Benninghoven
A.
Buschmann
N.
Wagner
F.
Novel sophorose lipids from microbial conversion of 2-alkanols
Biotechnol. Lett.
1995
, vol. 
17
 (pg. 
1183
-
1188
)
30
Avramenko
L. G.
Sklyar
Y. E.
Perel'son
M. E.
Pimenov
M. G.
The structure of pilopleurin: a new chromone from Pilopleura kozo-poljanskii
Chem. Nat. Compd.
1975
, vol. 
9
 (pg. 
15
-
16
)
31
Tan
C. T.
Dargeon
H. W.
Burchenal
J. H.
The effect of actinomycin D on cancer in childhood
Pediatrics
1959
, vol. 
24
 (pg. 
544
-
561
)
32
Shiba
S.
Studies on the therapy of reticulosarcoma(-tosis) with actinomycin
Acta Unio Int. Cancrum
1959
, vol. 
15
 (pg. 
264
-
266
)
33
Makarov
V. A.
Riabova
O. B.
Granik
V. G.
Dahse
H. M.
Stelzner
A.
Wutzler
P.
Schmidtke
M.
Anti-coxsackievirus B3 activity of 2-amino-3-nitropyrazolo[1,5-a]pyrimidines and their analogs
Bioorg. Med. Chem. Lett.
2005
, vol. 
15
 (pg. 
37
-
39
)
34
Forsberg
E. C.
Downs
K. M.
Christensen
H. M.
Im
H.
Nuzzi
P. A.
Bresnick
E. H.
Developmentally dynamic histone acetylation pattern of a tissue-specific chromatin domain
Proc. Natl. Acad. Sci. U.S.A.
2000
, vol. 
97
 (pg. 
14494
-
14499
)
35
Marks
P. A.
Breslow
R.
Dimethyl sulfoxide to vorinostat: development of this histone deacetylase inhibitor as an anticancer drug
Nat. Biotechnol.
2007
, vol. 
25
 (pg. 
84
-
90
)
36
Marks
P.
Rifkind
R. A.
Richon
V. M.
Breslow
R.
Miller
T.
Kelly
W. K.
Histone deacetylases and cancer: causes and therapies
Nat. Rev. Cancer
2001
, vol. 
1
 (pg. 
194
-
202
)
37
Walter
M. W.
Felici
A.
Galleni
M.
Soto
R. P.
Adlington
R. M.
Baldwin
J. E.
Frère
J.-M.
Gololobov
M.
Schofield
C. J.
Trifluoromethyl alcohol and ketone inhibitors of metallo-β-lactamases
Bioorg. Med. Chem. Lett.
1996
, vol. 
6
 (pg. 
2455
-
2458
)
38
Christians
F. C.
Loeb
L. A.
Novel human DNA alkyltransferases obtained by random substitution and genetic selection in bacteria
Proc. Natl. Acad. Sci. U.S.A.
1996
, vol. 
93
 (pg. 
6124
-
6128
)
39
Cameotra
S. S.
Makkar
R. S.
Recent applications of biosurfactants as biological and immunological molecules
Curr. Opin. Microbiol.
2004
, vol. 
7
 (pg. 
262
-
266
)
40
Isoda
H.
Shinmoto
H.
Kitamoto
D.
Matsumura
M.
Nakahara
T.
Differentiation of human promyelocytic leukemia cell line HL60 by microbial extracellular glycolipids
Lipids
1997
, vol. 
32
 (pg. 
263
-
271
)
41
Isoda
H.
Kitamoto
D.
Shinmoto
H.
Matsumura
M.
Nakahara
T.
Microbial extracellular glycolipid induction of differentiation and inhibition of the protein kinase C activity of human promyelocytic leukemia cell line HL60
Biosci. Biotechnol. Biochem.
1997
, vol. 
61
 (pg. 
609
-
614
)
42
Scholz
C.
Mehta
S.
Bisht
K.
Guilmanov
V.
Kaplan
D.
Nicolosi
R.
Gross
R.
Bioactivity of extracellular glycolipids: investigation of potential anti-cancer activity of sophorolipids and sophorolipid-derivatives
Proc. Am. Chem. Soc. Polym. Prepr.
1998
, vol. 
39
 (pg. 
168
-
169
)
43
Gross
R. A.
Bisht
K. S.
Scholz
C.
Mehta
S.
Guilmanov
V.
Nicolosi
R.
Treatment of cancer with chemically-modified sophorolipids
U.S. Pat. Application 60/092,446, 
1998
44
Sudo
T.
Zhao
X.
Wakamatsu
Y.
Shibahara
M.
Nomura
N.
Nakahara
T.
Suzuki
A.
Kobayashi
Y.
Jin
C.
Murata
T.
Yokoyama
K. K.
Induction of the differentiation of human HL-60 promyelocytic leukemia cell line by succinoyl trehalose lipids
Cytotechnology
2000
, vol. 
33
 (pg. 
259
-
264
)
45
Isoda
H.
Shinmoto
H.
Matsumura
M.
Nakahara
T.
Succinoyl trehalose lipid induced differentiation of human monocytoid leukemic cell line U937 into monocyte-macrophages
Cytotechnology
1995
, vol. 
19
 (pg. 
79
-
88
)
46
Aono
T.
Mizuno
K.
Chromone derivatives useful as antitumor agents
U.S. Pat. 4,904,690, 
1988
47
Boumendjel
A.
Nicolle
E.
Moraux
T.
Gerby
B.
Blanc
M.
Ronot
X.
Boutonnat
J.
Piperazinobenzopyranones and phenalkylaminobenzopyranones: potent inhibitors of breast cancer resistance protein (ABCG2)
J. Med. Chem.
2005
, vol. 
48
 (pg. 
7275
-
7281
)
48
Semba
S.
Itoh
N.
Ito
M.
Youssef
E. M.
Harada
M.
Moriya
T.
Kimura
W.
Yamakawa
M.
Down-regulation of PIK3CG, a catalytic subunit of phosphatidylinositol 3-OH kinase, by CpG hypermethylation in human colorectal carcinoma
Clin. Cancer Res.
2002
, vol. 
8
 (pg. 
3824
-
3831
)
49
Liu
X. Y.
Chen
H.
Patel
D. V.
Solution structure of actinomycin–DNA complexes: drug intercalation at isolated G-C sites
J. Biomol. NMR
1991
, vol. 
1
 (pg. 
323
-
347
)
50
Yu
C.
Tseng
Y. Y.
NMR study of the solution conformation of actinomycin D
Eur. J. Biochem.
1992
, vol. 
209
 (pg. 
181
-
187
)
51
Yu
F. L.
Selective inhibition of rat liver nuclear RNA polymerase II by actinomycin D in vivo
Carcinogenesis
1980
, vol. 
1
 (pg. 
577
-
581
)
52
Ross
W. E.
Bradley
M. O.
DNA double-stranded breaks in mammalian cells after exposure to intercalating agents
Biochim. Biophys. Acta
1981
, vol. 
654
 (pg. 
129
-
134
)
53
Wu
M. H.
Yung
B. Y.
Cell cycle phase-dependent cytotoxicity of actinomycin D in HeLa cells
Eur. J. Pharmacol.
1994
, vol. 
270
 (pg. 
203
-
212
)

Supplementary data