Teratogenic Effects Mediated by Inhibition of Histone Deacetylases

Andrew G. Foley , Andrew W. Cassidy , Ciaran M. Regan ... Ding Wang , Caiyun Fang , Nobel C. Zong , David A. Liem , Martin Cadeiras , Sarah B. Scruggs...
0 downloads 0 Views 230KB Size
Download PDF
272

Chem. Res. Toxicol. 2006, 19, 272-278

Teratogenic Effects Mediated by Inhibition of Histone Deacetylases: Evidence from Quantitative Structure Activity Relationships of 20 Valproic Acid Derivatives Daniel Eikel,† Alfonso Lampen,‡ and Heinz Nau*,† Department of Food Toxicology and Chemical AnalysissFood Toxicology, Center for Systemic Neuroscience HannoVer, Center for Food Science, UniVersity of Veterinary Medicine HannoVer, Foundation, Bischofsholer Damm 15, D-30173 HannoVer, Germany, and Department of Food Safety, Federal Institute for Risk Assessment (BfR), Thielallee 88-92, D-14195 Berlin, Germany ReceiVed August 12, 2005

The widely used antiepileptic drug valproic acid (VPA), which is also used in migraine prophylaxis and the treatment of bipolar disorders, is also under trial as an anticancer agent. Despite its wide range of therapeutic applications, VPA also has two severe side effects: acute liver toxicity and teratogenicity. The mechanism of action for all these properties is unknown to date, but recently, it was shown that VPA is able to inhibit the enzyme class of histone deacetylases (HDACs), proteins with a fundamental impact on gene expression and therefore possible molecular targets of VPA-induced signaling cascades. The purpose of this study was to determine if teratogenic side effects of VPA could be linked to its HDAC inhibition ability by studying a large set of structurally diverse derivatives based on the VPA core structure. We demonstrate that only VPA derivatives with a teratogenic potential in mice are able to induce a hyperacetylation in core histone H4 in teratocarcinoma F9 cells. We also demonstrate that this marker of functional HDAC inhibition occurs almost immediately (15 min) after exposure of F9 cells to VPA, whereas no influence on the HDAC protein levels (HDAC 2 and HDAC 3) could be detected even after 24 h of treatment. Further measurement of the IC50(HDAC) values of VPA derivatives in a human HDAC enzyme test system revealed an activity range from 10 to 10 000 µM; in some derivatives, HDAC inhibition ability was 40 times that of VPA. We also show a quantitative correlation between the IC50(HDAC) and the teratogenic potential of VPA derivatives, which clearly points toward HDACs as the formerly described teratogenic receptors of VPA-induced neural tube defects (NTDs). Introduction (VPA)1

Valproic acid is one of the antiepileptic drugs most frequently prescribed (1); it is also used clinically in a variety of other pathologies including bipolar disorders (2) and migraine prophylaxis (3). Currently, VPA is in clinical trials and under investigation as an anticancer agent (4). In addition to its exciting broad spectrum of properties, VPA is generally well-tolerated (5) but exhibits two rare but severe side effects: liver toxicity (6) and teratogenicity (7). Aside from malformations of the heart (8), the predominant VPA-induced teratogenic effects in humans are due to a failure of the neural tube to close (neural tube defects, NTDs) leading to conditions such as spina-bifida-aperta, anencephaly, and exencephaly (9). These teratogenic effects can also be induced in mice models by differential administration of VPA during the sensitive time of gestation and have been described as the * Corresponding author: Prof. Dr. Dr. h.c. Heinz Nau, University of Veterinary Medicine Hannover, Foundation, Center for Systemic Neuroscience Hanover, Center for Food Science, Department of Food Toxicology and Chemical AnalysissFood Toxicology, Bischofsholer Damm 15, 30173 Hannover, Germany. E-mail, [email protected]; tel., 0049-511856-7600; fax, 0049-511-856-7680. † University of Veterinary Medicine Hannover. ‡ Federal Institute for Risk Assessment (BfR). 1 Abbreviations: HDAC(s), histonedeacetylase(s); VPA, valproic acid; NTD(s), neural tube defect(s); H4, core histone 4; AcH4, acetylated core histone 4; NMRI, Naval Medical Research Institute; IC50(HDAC), substrate concentration with half-maximum HDAC enzyme activity; TSA, trichostatin A.

NMRI-exencephaly-mouse model (10) and the NMRI-spinabifida-aperta-mouse model (11). Our group has used these mouse models to establish a structure-activity relationship with a variety of derivatives based on the VPA core structure, and important structural prerequisites for VPA-induced NTDs have been discovered (12). VPA itself has been shown to be teratogenic, and some of its plasma metabolites also exhibit teratogenic effects (13). Elongation of one side chain and introduction of a triple bond in position C4 in the second side chain resulted in VPA analogues with increased induction of exencephaly in NMRI mice, while a further branching of a side chain diminished the teratogenic effects (14, 15). Derivatization of the carboxylic acid to the corresponding ester, amides, or hydroxamic acids also decreased or completely abolished the teratogenic potency (13, 16-18). In addition to these structural prerequisites, the most interesting and striking structural factor of VPA teratogenicity is the R hydrogen atom at position C2. On one hand, teratogenic effects were minimized or completely prevented in the NMRI-exencephaly-mouse model by substitution of the R hydrogen at C2 with a methyl group (15), hydroxyl group (M. Radatz, unpublished results), or a fluorine atom (19) or by introduction of a double bond between C2 and C3 (20). On the other hand, structure-activity relationship studies (SARs) have also demonstrated that there is differentiation between enantiomeric VPA analogues if a chiral center is position C2 (21). These findings ultimately led to the theoretical prediction of a stereoselective receptor of VPA-induced teratogenic effects (22).

10.1021/tx0502241 CCC: $33.50 © 2006 American Chemical Society Published on Web 01/11/2006

HDAC Inhibition by Teratogenic VPA DeriVatiVes

Chem. Res. Toxicol., Vol. 19, No. 2, 2006 273

Figure 1. Chemical structures of the 20 VPA derivatives investigated in this study (derivatives were numbered with Roman numerals, and their respective teratogenic potency are on the arbitrary scale from 0 (no detectable teratogenic potential) to +++++ (very high teratogenic potential)).

Whereas the antiepileptic potential of VPA is much less sensitive to changes in the chemical structure, even slight changes in the molecular structure of VPA derivatives (e.g., enantiomers) can completely prevent malformations in the NMRI-exencephaly-mouse model without substantially influencing the pharmacokinetic properties (23, 24). This phenomenon represents a unique possibility for the studying of the underlying molecular mechanisms of VPA-induced teratogenicity by investigating a proper set of structurally diverse derivatives. Such a screening approach covering VPA derivatives with both much higher and much lower teratogenic potency has already been applied by our group to demonstrate the involvement of peroxisome proliferation-activated receptors (PPARs) in VPA-induced NTDs. While PPAR R and γ were activated nonstructurally, specifically the PPAR β (PPAR δ) isoform was activated only by VPA analogues with high teratogenic potency (25, 26). Since it was not possible to demonstrate a direct binding of VPA by PPARs (27), it was suggested that PPAR β (PPAR δ) is a molecular marker for VPA-induced NTDs rather than a molecular target of VPA (28). It was recently shown that the enzyme class of histone deacetylases (HDACs) is inhibited by VPA, and it was proposed that HDACs were a possible enzyme target structure of both anticancer and teratogenic properties of VPA (29-32) due to

the fundamental importance of HDACs in the chromatin remodeling of cells and therefore in gene expression and function of the cell collective. HDAC inhibition might therefore lead to cellular differentiation or apoptosis, both events that could ultimately also lead to embryonic malformations. Trichostatin A (TSA), a classical HDAC inhibitor, is also controversially discussed as a possible teratogen as it leads to malformations similar to those of VPA if investigated in vitro (27, 33, 34) but not in vivo (35); this discrepancy might be due to metabolism and possible detoxification of TSA in the mice (36). In addition, the well-known teratogen carbamazepine is also proposed as an HDAC inhibitor (37, 38), which is further indication for HDACs as interesting molecular target structures in the field of reproductive toxicology. In addition to functional inhibition of HDACs, it was also reported that HDAC inhibitors can alter the cellular protein level of histone deacetylases, an effect that might also have an influence on the sensitive balance of acetylation and deacetylation of core histones (39, 40). In this study, we used a structurally diverse set of 20 VPA derivatives (Figure 1, VPA and its derivatives, coded with Roman numerals) that had been extensively investigated for reproductive toxicity in the NMRI-exencephaly-mouse model by our group. The analogues used here have both higher and

274 Chem. Res. Toxicol., Vol. 19, No. 2, 2006

Eikel et al.

Table 1. Decision Criteria for Teratogenic Potency Grading of VPA Derivatives teratogenic potency

dose range (mmol/kg)

exencephaly rate (%)

description

0 + ++ +++ ++++ +++++

>3.0 2.0-3.0 2.0-3.0 2.0-3.0 1.0-2.0 0.25-1.0

0 1-5 5-25 25-60 40-60 40-60

no teratogenic potency detectable low teratogenic potency lower teratogenic potency than VPA equal teratogenic potency to VPA higher teratogenic potency than VPA very high teratogenic potency

lower teratogenic potency than VPA and represent the bestcharacterized set of test compounds for VPA-induced NTDs known so far. The derivatives used in this study cover all of the known structural aspects of VPA-induced malformations such as carboxylic acid derivatization, side chain saturation, side chain length, and especially the chirality at position C2. We show here that there is yet a quantitative correlation between functional HDAC inhibition and the teratogenic potency of the corresponding VPA analogues, thus indicating HDAC inhibition to be a crucial aspect of VPA-induced teratogenicity but also demonstrating the possibility to utilize HDAC inhibition as a prediction system for teratogenic side effects on an even broader selectivity.

Experimental Procedures Materials and Valproic Acid Derivatives. All chemicals used were of analytical grade if not stated otherwise. Valproic acid (VPA) and trichostatin A (TSA) were obtained from Fluka-Sigma-Aldrich GmbH (Germany); the valpromide (VPD) was a kind gift from Katwijk Chemie (The Netherlands). Valproic acid derivatives were synthesized as described in detail elsewhere (14-17, 19-21, 23, 41). Standard GC-MS analysis showed that the chemical purity of the derivatives was >95%. The optical purity of chiral compounds was measured after suitable derivatization with chiral reagents by standard GC-NPD analysis and found to be >95% ee (enantiomeric excess). All VPA derivatives used in the cell culture assays were dissolved in dimethyl sulfoxide (DMSO) to give 1 M stock solutions. Teratogenic Potency Measurement. The exencephaly rates used as the model parameter for teratogenicity were derived from previous publications of our group. In these studies, the exencephaly rates had been measured in the NMRI-exencephaly-mouse model (7) at one or more dose levels (13-21). As a result of certain differences in the experimental procedures of these publications (e.g., pH of injected solutions, sc versus ip application, etc.), the

exencephaly rates were grouped into an arbitrary range of teratogenic potency according to the decision criteria shown in Table 1 in a range from 0 (no teratogenic potency detectable) to +++++ (very high teratogenic potency). The resulting rating of the 20 VPA derivatives is given both in Figure 1 and Table 2. Cell Culture. The teratocarcinoma mouse cell line F9 (American Type Culture Collection, Rockville, MD) was cultured in Ham’s F-12/DMEM medium containing 2 mM L-glutamine, 10% (v/v) fetal bovine serum, 0.145 mM 2-mercaptoethanol, and 100 U/mL penicillin/streptamycin (medium and supplements from Invitrogen, Germany). For the experimental setup, 106 cells were treated in triplicate in 6-well plates by incubation at 37 °C in a humid atmosphere of air and 5% (v/v) CO2. After the indicated time, treated cells were scraped from the bottom of the wells, washed twice with PBS, dissolved in 100 µL of lysis buffer (62.5 mM Tris/ HCl, pH 6.8, 2% (w/v) sodiumdodecyl sulfate, 1% (v/v) glycerin, 2.5 µM dithio-DL-threitol, 250 µM phenylmethansulfonyl fluoride, 0.05 µg/mL bestatin, 2 µg/mL aprotinin, and 0.05 µg/mL leupeptin), and boiled immediately for 5 min at 90 °C. Western blot analysis of acetylated histone 4 (AcH4), histone deacetylase 2 (HDAC 2), histone deacetylase 3 (HDAC 3), and β-actin was generally made directly on 10 µL of cell lysate, whereas the total cellular protein content was measured by the bicinichonic acid method (42) for cell samples treated longer than 6 h in order to compensate analogue-dependent proliferation of cells. Western Blot Analysis of the Acetylated Core Histone H4. A total of 10 µL of the cell lysate was separated by 15% SDSpolyacrylamide gel electrophoresis and transferred to a nitrocellulose membrane by semi-dry electroblotting. The blotted membrane was washed with TBS buffer (2.4 g/L Tris/HCl and 8 g NaCl, pH 7.6) and blocked with TBS buffer containing 3% nonfat dry milk (TBSM) for 1 h at room temperature. The nitrocellulose membrane (Amersham Bioscience, Germany) was incubated with a 1:2000 dilution of anti-acetyl histone H4 antibody (Upstate/Biomol, Germany) in TBS-M at 4 °C for 12 h. The membrane was washed once with TBS buffer and incubated again with a 1:5000 dilution of an anti-rabbit antibody (Amersham Bioscience ECL detection kit) in TBS-M for 1.5 h at room temperature. Blots were washed three times with TBS buffer, once with 0.05% (v/v) Tween 20 in TBS buffer, and again three more times with TBS buffer before antibodies were detected with the ECL detection kit (Amersham Bioscience) according to the manufacturer’s instructions. Western Blot Analysis of HDAC 2, HDAC 3, and β-Actin. Western blot analysis of HDAC 2, HDAC 3, and β-actin was performed as described above but with the following changes: after measurement of the protein content of the whole cell lysate, 10 µg of proteins was separated by 8% SDS-polyacrylamide gel elec-

Table 2. Summary of the Measured Properties of 20 Valproic Acid Derivatives (Teratogenic Potential, Hyperacetylation of Core Histone 4 in Treated F9 Cells, and Concentration of Half-Maximum Effect in the HDAC Enzyme Inhibition Assay) Sorted by HDAC Inhibition Potential VPA derivative

teratogenic potential

(()-2-heptyl-4-pentynoic acid (XVII) (()-2-hexyl-4-pentynoic acid (XVI) (()-2-propyl-octanoic acid (XI) (()-2-pentyl-4-pentynoic acid (XV) S-2-pentyl-4-pentynoic acid (XXI) (()-2-butyl-4-pentynoic acid (XIV) (()-2-propyl-heptanoic acid (X) (()-2-propyl-hexanoic acid (IX) valproic acid (I) R-2-pentyl-4-pentynoic acid (XX) (()-2-propyl-4-pentenoic acid (VI) valproic hydroxamic acid (III) 2-propyl-2-hydroxy-pentanoic acid (IV) S-2-propyl-4-hexynoic acid (XIX) R-2-propyl-4-hexynoic acid (XVIII) valpromide (II) (()-2-isobutyl-4-pentynoic acid (XII) (()-2-propyl-4-hexynoic acid (XIII) 2-propyl-2E-pentenoic acid (V) (()-2-isobutyl-4-pentenoic acid (VII) (()-2-ethyl-4-methyl-pentanoic acid (VIII)

+++++ +++++ ++++ ++++ +++++ ++++ +++ +++ +++ +++ ++ 0 + ++ 0 + + + 0 0 0

AcH 4

(0 to ++) ++ ++ ++ ++ ++ + ++ + + + ++ + 0 + 0 0 0 0 0 0 0

IC50(HDAC) ( SE (µM) 12 ( 2 13 ( 2 25 ( 4 35 ( 10 48 ( 12 98 ( 18 103 ( 21 144 ( 34 398 ( 50 869 ( 183 2620 ( 3170 5040 ( 6740 5300 ( 365 5840 ( 19380 7360 ( 10050 >10000 >10000 >10000 >10000 >10000 >10000

HDAC Inhibition by Teratogenic VPA DeriVatiVes

Chem. Res. Toxicol., Vol. 19, No. 2, 2006 275

Figure 2. Mathematical enzyme inhibition function used for data fitting. E, enzymatic effect; C, concentration of test compound; Emax, maximal enzymatic effect with no test compound (negative control); E0, minimal enzymatic effect (positive control); δ, flexion of the sigmoid enzyme inhibition curve; and IC50, concentration of the test compound with half-maximum enzymatic effect.

trophoresis. The blotted membranes were incubated either with the anti-HDAC 2 or anti-HDAC 3 antibody (Upstate/Biomol GmbH, Germany) in a dilution of 1:1000 or with the anti-β-actin antibody (Dunn Labortechnik GmbH, Germany) in a dilution of 1:2000 in TBS-M at 5 °C overnight. Human HDAC Enzyme Assay. HDAC activity was measured by using an HDAC fluorescence activity assay kit (Biomol, Germany). Because of the pH dependency of the enzymatic test system, all compounds measured were first dissolved in water and neutralized before preparation of further dilution series with HDAC assay buffer. The dose activity was assayed according to the manufacturer’s instructions with at least three repeats. In short, HeLa nuclear extracts (1 µL of between 6 and 9 mg/mL) were incubated with 500 µM acetylated Fluor-de-Lys substrate in 50 µL of assay buffer in the presence or absence of the respective valproic acid analogue. The HDAC-inhibitor trichostatin A (TSA) at a concentration of 5 µM served as positive control. The deacetylation reaction was carried out at 37 °C for 4 h and stopped by addition of 50 µL of Fluor-de-Lys developer solution containing 2 µM trichostatin A. After 15 min, fluorescence activity was measured with a Victor 1420 fluorescence reader (Perkin-Elmar LAS GmbH, Germany) at 355 nm excitation and 535 nm emission. The enzyme activity was calculated relative to the measured fluorescence activity of four negative controls (HDAC assay buffer only) on each 96-well plate. The IC50(HDAC) value was determined by computational fitting of at least six dose HDAC inhibition data to a mathematical enzyme inhibition function (Figure 2) with the pharmacodynamic module of the WinNonLin 4 software package (Pharsight Corporation, USA) to yield IC50 values in µmol/L with a standard error (SE) representing the goodness-of-fit between the computational model and the experimental data.

Results Teratogenic Potency Grading of Valproic Acid Derivatives. Data mining was conducted on previously published reproductive toxicity studies of our group. VPA-based substances had been intensively characterized in the NMRIexencephaly-mouse model, and animal experimental data were obtained for all VPA derivatives used in this study. These were then transformed into the arbitrary range of teratogenic potential from 0 (no detectable teratogenic potency) to +++++ (very high teratogenic potency) by means of the grading criteria given above. Teratogenic ratings for all VPA derivatives used in this study are summarized in Figure 1 and Table 2. Induction of Hyperacetylation of Core Histone H4 as a Cellular Marker for Functional HDAC Inhibition. Teratocarcinoma F9 mouse cells were first treated for different periods of time with valproic acid (I, +++) and S-2-pentyl-4-pentynoic acid (XXI, +++++) for characterization of the rate and intensity of the cellular response (Figure 3). Hyperacetylation of the core histone 4 (AcH4) served as the marker for inhibition of enzymatic HDAC function and could be detected as soon as 15 min (VPA at 0.25 mM), 2 h (S-2-pentyl-4-pentynoic acid at 0.25 mM), or 60 min (Trichostatin A at 200 nM, data not shown) after treatment of cells. The intensity of hyperacetylation was not slowly increasing with time but appeared abruptly between two time points without further increase at later time points. Induction of AcH4 was concentration-dependent as was shown for both valproic acid (I,+++) and S-2-pentyl-4-pentynoic acid

Figure 3. Time-dependent hyperacetylation of core histone 4 after F9 cell treatment with (A) 0.25 mM valproic acid (I, +++) and (B) 0.25 mM S-2-pentyl-4-pentynoic acid (XXI, +++++). Positive control (Pos) represents F9 cell treatment for 6 h with 200 nM trichostatin A; negative control (Neg) represents 6 h treatment with 1% (v/v) DMSO in F9 cell medium.

Figure 4. Concentration-dependent hyperacetylation of core histone 4 after F9 cell treatment for 6 h with (A) valproic acid (I, +++) and (B) S-2-pentyl-4-pentynoic acid (XXI, +++++). Positive control (Pos) represents F9 cell treatment with 200 nM trichostatin A; negative control (Neg) represents treatment with 1% (v/v) DMSO in F9 cell medium.

Figure 5. Hyperacetylation of core histone 4 after treatment of F9 cells for 6 h with 1 mM each of VPA derivatives (A) side chain elongated and further branched derivatives and (B) analogues with the chiral center at C2. Positive control (Pos) represents F9 cell treatment with 200 nM trichostatin A; negative control (Neg) represents F9 cell treatment with 1% (v/v) DMSO in F9 cell medium; E represents an empty lane.

(XXI, +++++) after F9 treatment for 6 h with concentrations raging from 0.05 to 3.00 mM (Figure 4). The grade of H4 acetylation was different from control samples at concentrations as low as 500 µM (VPA) and 50 µM (S-2-pentyl-4-pentynoic acid) with respect to differences in band intensity of the H4 acetylation state of control samples. These differences represents the basal acetylation of the core histone H4 due to normal cell activity and are mainly due to slight differences in the developing process of the Western blot film. A second band sometimes occurring in the trichostatin A-treated positive controls is likely to be another hyperacetylated core histone; because of the kilodalton range of this band, it is likely to be AcH . 3 All 20 VPA derivatives were screened for induction of AcH4 in the F9 teratocarcinoma mouse cell line. There was a clear correlation between induction of AcH4 and the teratogenic potential of valproic acid analogues. The high sensitivity of the cellular test system is indicated by differentiation between VPA derivatives with only minor structural changes such as side chain elongation and further side chain branching (Figure 5 A). While chain elongation (I, IX, X, XI) leads to consecutive higher levels of acetylated H4, further branching at position C4 (VIII) completely averted this effect.

276 Chem. Res. Toxicol., Vol. 19, No. 2, 2006

Eikel et al.

Figure 6. Protein levels of F9 whole cell lysate (10 µg) of (A) β-actin (42 kDa), (B) HDAC 3 (48 kDa), and (C) HDAC 2 (55 kDa) after 24 h treatment with 1 mM concentrations of VPA derivatives with no teratogenic potential (III, V), intermediate teratogenic potential (I, XII, XX), and very high teratogenic potential (XXI). Negative control (Neg) represents treatment of F9 cells for 24 h with 1% (v/v) DMSO in F9 cell medium.

The high selectivity of the F9 cell system was demonstrated by its ability to distinguish between enantiomeric VPA derivatives with the chiral center at position C2 (Figure 5B). Both S-enantiomers studied here (XIX, XXI) showed higher levels of acetylated core histone 4 than the corresponding R-enantiomers (XVIII, XX). All structural aspects of the induction of histone hyperacetylation correlate qualitatively with the structural prerequisites for teratogenic effects of VPA derivatives known so far. Influence of VPA Derivatives on the Cellular HDAC 2 and HDAC 3 Levels. F9 mice cells were treated for 6, 12, and 24 h with selected VPA derivatives with different grades of teratogenic potential: high (XXI), comparable to that of VPA (I, XX, XII), and undetectable (III, V). The cellular HDAC 2 and -3 levels were measured via Western blot analysis in direct comparison to β-actin as control protein (Figure 6, data not shown for 6 and 12 h). The VPA derivatives had no influence on cellular HDAC 2 and HDAC 3 levels after any of the three incubation times (24, 12, and 6 h), although some VPA derivatives (I, XXI) did influence the morphology and proliferation of the F9 cells after 12 and 24 h treatment (data not shown; see ref 25). These results suggest that a functional HDAC inhibition and not HDAC protein degradation is correlated with the teratogenic effects of VPA. Measurement of the IC50(HDAC) Values of the 20 VPA Analogues. The IC50(HDAC) values of the 20 VPA derivatives were measured using an HDAC inhibition assay with human HeLa cell nuclear extract as the scource of HDAC activity. A typical set of computational fits of concentration effect curves is shown in Figure 7. Side chain elongation of VPA derivatives with a triple bond in position C4 resulted in decreasing IC50(HDAC) values, while further branching of one side chain is diminishing the HDAC inhibition effect. The calculated IC50(HDAC) values range from 10 to 10 000 µM, with some VPA derivatives exceeding the HDAC inhibition potential of VPA itself 40-fold. One can also detect a 20-fold increase in IC50(HDAC) inhibition potential between VPA steroisomers having a different teratogenic potential. This clearly shows a stereoselective interaction with HDAC enzymes. The resulting IC50(HDAC) values of all tested VPA derivatives are summarized in Table 2, and these measurements confirm quantitatively the aforementioned qualitative correlation of teratogenic potential and HDAC inhibition abilitiy of VPA derivatives. Correlation of HDAC Inhibition Properties and Teratogenic Potential of VPA Derivatives. The correlation between the IC50(HDAC) values and the graded teratogenic potential of the 20 VPA derivatives clearly indicates that these two compound properties are not only related but also quantitatively connected with each other (Figure 8). The most potent HDAC inhibitors, with IC50(HDAC) values between 10 and 50 µM (XI,

Figure 7. Concentration effect curves fitted to the experimental HDAC inhibition data of side chain elongated and further branched VPA derivatives with a triple bond in position C4 (XII-XVII). All fitted dose-response curves are based on at least six concentrations with at least three independent measurements of each concentration. Data points shown represent the mean of these measurements with error bars showing the standard deviation of the mean (SD).

Figure 8. HDAC enzyme inhibition ability correlated with the teratogenic potency of the 20 investigated VPA derivatives. A linear regression and the 95% confidence interval visualize the quantitative relationship of these two compound properties.

XV, XVI, XVII, XXI), were also the VPA derivatives with the highest teratogenic potential. This correlation strengthens the hypothesis that VPA-induced NTDs are mediated by HDAC inhibition and points toward HDAC inhibition as a potential endpoint in screening systems of reproductive toxicity.

Discussion Valproic acid (VPA) has recently been shown to bind and inhibit the enzyme class of histone deacetylases (HDACs), which are important regulators of the chromatin remodeling of cells and therefore have a great impact on gene expression and cell function (29, 30). Inhibition of enzymatic HDAC function can lead to differentiation, apoptosis, or interruption of cell proliferation (43, 44), all cellular events that can possibly cause embryonic malformations. HDAC inhibition might therefore be part of a molecular signaling cascade which results in VPAinduced neural tube defects (NTDs), and it can be hypothesized that HDACs are the theoretical “teratogenic receptors” for VPAinduced NTDs first described by our group (22). This hypothesis has been strengthened by further reports of chemicals with some

HDAC Inhibition by Teratogenic VPA DeriVatiVes

structural similarities to the VPA core structure that are also not HDAC inhibitors when being nonteratogenic (31, 32). However, these studies are mostly based on VPA-like structures which are less potent teratogens than VPA. To prove the hypothesis that HDACs are the hypothesized teratogenic receptors of VPA, we measured the HDAC inhibition ability of a structurally diverse set of 20 VPA derivatives with various structural modifications such as derivatization of the carboxylic function, side chain elongation and further branching, introduction of double and triple bonds, and enantiomeric derivatives with the chiral center at position C2. These compounds had been extensively studied by our group in the NMRI-exencephaly-mouse model before and represent one of the best-characterized set of structurally diverse reproductive toxicants so far. The VPA derivatives used in this study cover the arbitrary range of teratogenic potency from undetectable teratogenic potency (0) to very high teratogenic potency (+++++). Therefore, this set is the first investigated set of VPA derivatives that also considers structures with a much higher teratogenic potency than VPA itself. It also consists of stereoisomers which could prove the enantioselective interaction with HDAC enzymes and therefore represents a unique opportunity for the study of qualitative and quantitative relationships between HDAC inhibition and teratogenic potency. In fact, we found that VPA almost immediately (15 min) induced hyperacetylation of core histone 4 (AcH4) in F9 cells, which shows that HDAC inhibition is a very early cellular event in the response to VPA. It is therefore reasonable to assume that HDAC inhibition is not only part of, but might be the first step in, a molecular signaling cascade leading toward neural tube defects. The structurally different VPA derivative S-2pentyl-4-pentynoic acid (XXI) leads to AcH4 only after 2 h of cell treatment; this slower hyperacetylation rate might be due to a slower passage of this congener through the cellular membrane rather than to a delayed cellular response. Only teratogenic VPA derivatives induced AcH4 in F9 cells, which is in accordance with previously published results (29-32). VPA is reported not only to inhibit the function of HDACs but also to induce a selective degradation of HDAC 2 protein levels both in vitro and in vivo (38). To date one, another HDAC inhibitor (suberoyl anilide hydroxamic acid, SAHA) has also been reported to decrease HDAC protein levels, but is selective for HDAC 3 (39). To determine whether degradation of HDAC proteins could be involved in VPA-induced NTDs, we investigated the cellular levels of both HDAC isoform 2 and 3 in F9 cells after treatment with selected VPA derivatives covering the whole teratogenic potential range from 0 to +++++. Unlike the investigation mentioned above, we did not detect any degradation of either HDAC 2 or HDAC 3 after 6, 12, and even 24 h of cell treatment. It is noteworthy that we were not able to reproduce the degradation of HDAC 2 after VPA treatment in the ostensible same F9 cell culture used by Kraemer et al. (38). However, after 24 h of treatment, we already detect major changes in the cellular morphology of the F9 cells, and proliferation had already been strongly altered by both VPA and XXI, both effects described by our group before (25, 33). This VPA derivative, although possessing a much higher teratogenic potency than VPA, also did not induce HDAC 2 degradation. It is also noteworthy that HDAC inhibition occurs as soon as 15-60 min after cell exposure, whereas HDAC degradation did not occur up to 24 h of treatment. Taking also into account that SAHA had been reported to induce HDAC degradation of isoform 3 instead of isoform 2 and with respect to the small data set, it seems likely that there

Chem. Res. Toxicol., Vol. 19, No. 2, 2006 277

is no correlation between HDAC inhibition and HDAC degradation. In this regard, our results clearly demonstrate that the functional inhibition of HDACs is related to the VPA-induced NTDs, and degradation of HDAC seems not to be linked to this severe drug side effect. The HDAC inhibition ability of the 20 VPA derivatives investigated here was measured quantitatively in a commercial HDAC inhibition assay with HeLa nuclear extracts as the source of HDAC activity. There was a wide range of activity at IC50(HDAC) concentrations from 10 to 10 000 µM with some derivatives exceeding the VPA HDAC inhibition potential 40fold. Most interestingly, testing VPA stereoisomers, we detected an up to 20-fold difference in HDAC inhibition potential, thus, demonstrating a stereoselective interaction with HDAC enzymes. Here, too, the teratogenic potential of VPA derivatives corresponded to their IC50(HDAC) values. The correlation between the teratogenic potential of the investigated 20 VPA derivatives and their IC50(HDAC) is excellent when the teratogenic potential of the derivatives is equal or even higher than that of VPA, but the correlation was slightly smaller if the teratogenic potential of the VPA analogues was lower than that of VPA. This also demonstrates that to disclose molecular targets of VPA-induced NTDs it is important to investigate a proper set of VPA derivatives consisting of analogues with both higher and lower teratogenic potential than VPA. The correlation was surprisingly good, although no consideration was taken of differences in metabolism (activation or deactivation) or pharmacokinetics in vivo. Our results clearly indicate that HDAC inhibition is a molecular target for VPAinduced NTDs. We therefore conclude that HDACs are the theoretically hypothesized “teratogenic receptors“ that can trigger a VPA-induced molecular signaling cascade resulting in embryonic malformations. Furthermore, there are strong indications that trichostatin A (TSA), a classical HDAC inhibitor, can also cause embryonic malformations (33, 34), and there are controversial reports that carbamazepine, a known teratogenic drug, inhibits HDAC activity (36, 37). We can therefore assume that HDAC inhibition not only mediates VPA-induced NTDs but might also trigger other chemically induced embryonic malformations. In this regard, our finding that the correlation between HDAC inhibition and VPA-induced NTDs can be measured quantitatively demonstrates that HDAC inhibition might be a suitable molecular endpoint in screening systems of reproductive toxicology. Ongoing studies by our group on both known HDAC inhibitors as reproductive toxicants in the NMRI-execephalymouse model as well as screening of known teratogens for HDAC inhibition will reveal if HDACs can be successfully utilized as endpoints in reproductive toxicity screenings. Acknowledgment. We thank the Deutsche Forschungsgemeinschaft (DFG-NA 104/2-1), the European Research Training Network (RTN2-2001-00370), the European Commission (6th Framework Program: ReProTect), the Federal Ministry for Education and Research (BBF) (Project 0313070D), and the Academy for Animal Health (ATF) for generous financial support. We also thank Mrs. J. McAlister-Herman for valuable suggestions and for the critical reading of this manuscript.

References (1) Perucca, E. (2002) Pharmacological and therapeutic properties of valproate: a summary after 35 years of clinical experience. CNS Drugs 16 (10), 695-714. (2) Goodwin, F. K., Fireman, B., Simon, G. E., Hunkeler, E. M., Lee, J., and Revicki, D. (2003) Suicide risk in bipolar disorder during treatment with lithium and divalproex. JAMA, J. Am. Med. Assoc. 290 (11), 1467-1473.

278 Chem. Res. Toxicol., Vol. 19, No. 2, 2006 (3) Frediani, F. (2004) Anticonvulsant drugs in primary headache prophylaxis. Neurol. Sci. 25 (Suppl. 3), S161-S166. (4) Blaheta, R. A., Michaelis, M., Hernaiz-Driever, P., and Cinatl, J., Jr. (2005) The evolving anticancer drug valproic acids: insights into the mechanism and clinical studies. Med. Res. ReV. 25 (4), 383-397. (5) Lo¨scher, W. (2002) Basic pharmacology of valproate: a review after 35 years of clinical use for the treatment of epilepsy. CNS Drugs 16 (10) 669-694. (6) Clarkson, A., and Choonara, I. (2002) Surveillance for fatal suspected adverse drug reactions in the UK. Arch. Dis. Child. 87 (6), 462-466. (7) Nau, H., Rating, D., Koch, S., Hauser, I., and Helge, H. (1981) Valproic acids and its metabolites: placental transfer, neonatal pharmacokinetics, transfer via mother’s milk and clinical status of neonates of epileptic mothers. J. Pharmacol. Exp. Ther. 219 (3), 768-777. (8) Lindhout, D., Meinardi, H., Meijer, J. W., and Nau, H. (1992) Antiepileptic drugs and teratogenesis in two consecutive cohorts: changes in prescription policy paralleled by changes in pattern of malformations. Neurology 42 (4 Suppl. 5) 94-110. (9) Cabrera, R. M., Hill, D. S., Etheredge, A. J., and Finnell, R. H. (2005) Investigations into the etiology of neural tube defects. Birth Defects Res., Part C 72, 330-344. (10) Nau, H., Zierer, R., Spielmann, H., Neubert, D., and Gansau, C. (1981) A new model for embryotoxicity testing: teratogenicity and pharmacokinetics of valproic acid following constant-rate administration in the mouse using human therapeutic drug and metabolite concentrations. J. Life Sci. 29 (26), 2803-2814. (11) Ehlers, K., Sturje, H., Merker, H. J., and Nau, H. (1992) Valproic acid-induced spina bifida: a mouse model. Teratology 45 (2), 145154. (12) Nau, H., Hauck, R. S., and Ehlers, K. (1991) Valproic acid induced neural tube defects in mouse and human: aspects of chirality, alternative drug development, pharmacokinetics and possible mechanisms. Pharmacol. Toxicol. 69 (5) 310-321. (13) Lo¨scher, W., and Nau, H. (1985) Pharmacological evaluation of various metabolites and analogues of valproic acid. Anticonvulsant and toxic potencies in mice. Neuropharmacology 24 (5), 427-435. (14) Hauck, R. S., and Nau, H. (1990) Structure-teratogenicity relationships of valproic acid and related branched chain carboxylic acids. Teratology 41 563-567. (15) Bojic, U., Elmazar, M. M. A., Hauck, R. S., and Nau, H. (1996) Further branching of valproate-related carboxylic acids reduces the teratogenic activity, but not the anticonvulsant effect. Chem. Res. Toxicol. 9 (5), 866-870. (16) Radatz, M., Ehlers, K., Yagen, B., Bialer, M., and Nau, H. (1998) Valnoctamide, valpromide and valnoctic acid are much less teratogenic in mice than valproic acid. Epilepsy Res. 30 (1), 41-48. (17) Levi, M., Yagen, B., and Bialer, M. (1997) Pharmacokinetics and antiepileptic activity of valproyl hydroxamic acid derivatives. Pharm. Res. 14 (2), 213-217. (18) Volland, J. (2002) Embryotoxicity, anticonvulsant action, sedation and adipocyte differentiation: structure-activity-studies with valproic acid derivatives in mice and in C3H/10T1/2-cells. Ph.D. Thesis, University of Veterinary Medicine Hannover, Foundation, Hannover, Germany, http://elib.tiho-hannover.de/dissertations/vollandj_2002.pdf. (19) Ellerbeck, U., Radatz, M., and Nau, H. (1998) Structure-teratogenicity relationship of valproic acid analogues and development of new antiepileptic drugs. Naunyn-Schmiedeberg’s Arch. Pharmacol. 30, R772. (20) Nau, H., and Lo¨scher, W. (1986) Pharmacologic evaluation of various metabolites and analogues of valproic acid: teratogenic potencies in mice. Fundam. Appl. Toxicol. 6 (4), 669-676. (21) Hauck, R. S., Nau, H., and Elmazar, M. M. A. (1991) On the development of alternative antiepileptic drugs. Lack of enantioselectivity of the anticonvulsant activity, in contrast to teratogenicity, of 2-n-propyl-4-pentenoic acid and 2-n-propyl-4-pentynoic acid, analogues of the anticonvulsant drug valproic acid. Naturwissenschaften 78 (6), 272-274. (22) Bojic, U., Ehlers, K., Ellerbeck, U., Bacon, C. L., O’Driscoll, E., O’Connell, C., Berezin, V., Kawa, A., Lepekhin, E., Bock, E., Regan, C. M., and Nau, H. (1998) Studies on the teratogen pharmacophore of valproic acid analogues: evidence of interactions at a hydrophobic centre. Eur. J. Pharmacol. 354 (2-3), 289-299. (23) Hauck, R. S., Nau, H., and Elmazar, M. M. A. (1991) On the development of alternative antiepileptic drugs. Lack of enantioselectivity of the anticonvulsant activity, in contrast to teratogenicity, of 2-n-propyl-4-pentenoic acid and 2-n-propyl-4-pentynoic acid, analogues of the anticonvulsant drug valproic acid. Naturwissenschaften 78 (6), 272-274. (24) Hauck, R. S., Elmazar, M. M. A., Plum, C., and Nau, H. (1992) The enantioselective teratogenicity of 2-n-propyl-4-pentynoic acid (4-ynVPA) is due to stereoselective intrinsic activity and not differences in pharmacokinetics. Toxicol. Lett. 60 (2), 145-153.

Eikel et al. (25) Lampen, A., Siehler, S., Ellerbeck, U., Go¨ttlicher, M., and Nau, H. (1999) New molecular bioassays for the estimation of the teratogenic potency of valproic acid derivatives in vitro: activation of the peroxisomal proliferator-activated receptor (PPARdelta). Toxicol. Appl. Pharmacol. 160 (3), 238-249. (26) Werling, U., Siehler, S., Litfin, M., Nau, H., and Go¨ttlicher, M. (2001) Induction of differentiation in F9 cells and activation of peroxisome proliferator-activated receptor delta by valproic acid and its teratogenic derivatives. Mol. Pharmacol. 59 (5), 1269-1276. (27) Lagace, D. C., McLeod, R. S., and Nachtigal, M. W. (2004) Valproic acid inhibits leptin secretion and reduces leptin messenger ribonucleic acid levels in adipocytes. Endocrinology 145 (12), 5493-4503. (28) Lampen, A., Carlberg, C., and Nau, H. (2001) Peroxisome proliferatoractivated receptor delta is a specific sensor for teratogenic valproic acid derivatives. Eur. J. Pharmacol. 431 (1), 25-33. (29) Phiel, C. J., Zhang, F., Huang, E. Y., Guenther, M. G., Lazar, M. A., and Klein, P. S. (2001) Histone deacetylase is a direct target of valproic acid, a potent anticonvulsant, mood stabilizer, and teratogen. J. Biol. Chem. 276 (39), 36734-36741. (30) Gottlicher, M., Minucci, S., Zhu, P., Kramer, O. H., Schimpf, A., Giavara, S., Sleeman, J. P., Lo Coco, F., Nervi, C., Pelicci, P. G., and Heinzel, T. (2001) Valproic acid defines a novel class of HDAC inhibitors inducing differentiation of transformed cells. EMBO J. 20 (24), 6969-6978. (31) Gurvich, N., Tsygankova, O. M., Meinkoth, J. L., and Klein, P. S. (2004) Histone deacetylase is a target of valproic acid-mediated cellular differentiation. Cancer Res. 64 (3), 1079-1086. (32) Eyal, S., Yagen, B., Shimshoni, J., Bialer, M. (2005) Histone deacetylase inhibition and tumor cells cytotoxicity by CNS-active VPA constitutional isomers and derivatives. Biochem. Pharmacol. 69 (10) 1501-1508. (33) Lampen A., Grimaldi P. A., and Nau H. (2005) Modulation of peroxisome proliferator-activated receptor delta activity affects neural cell adhesion molecule and polysialyltransferase ST8SiaIV induction by teratogenic valproic acid analogs in F9 cell differentiation. Mol. Pharmacol. 68 (1) 193-203. (34) Svensson, K., Mattsson, R., James, T. C., Wentzel, P., Pilartz, M., MacLaughlin, J., Miller, S. J., Olsson, T., Eriksson, U. J., and Ohlsson, R. (1998) The paternal allele of the H19 gene is progressively silenced during early mouse development: the acetylation status of histones may be involved in the generation of variegated expression patterns. DeVelopment 125 (1), 61-69. (35) Nervi, C., Borello, U., Fazi, F., Buffa, V., Pelicci, P. G., and Cossu, G. (2001) Inhibition of histone deacetylase activity by trichostatin A modulates gene expression during mouse embryogenesis without apparent toxicity. Cancer Res. 61 (4), 1247-1249. (36) Elaut, G., Torok, G., Vinken, M., Laus, G., Papeleu, P., Tourwe, D., and Rogiers, V. (2002) Major phase I biotransformation pathways of Trichostatin a in rat hepatocytes and in rat and human liver microsomes. Drug Metab. Dispos. 30 (12), 1320-1328. (37) Eyal, S., Yagen, B., Sobol, E., Altschuler, Y., Shmual, M., and Bialer, M. (2004) The activity of antiepileptic drugs as histone deacetylase inhibitors. Epilepsia 45 (7) 737-744. (38) Beutler, A. S., Li, S., Nicol, R., and Walsh, M. J. (2005) Carbamazepine is an inhibitor of histone deacetylases. J. Life Sci. 76 (26) 3107-3115. (39) Kra¨mer, O. H., Zhu, P., Ostendorf, H. P., Golebiewski, M., Tiefenbach, J., Peters, M. A., Brill, B., Groner, B., Bach, I., Heinzel, T., and Go¨ttlicher, M. (2003) The histone deacetylase inhibitor valproic acid selectively induces proteasomal degradation of HDAC 2. EMBO J. 22 (13) 3411-3420. (40) Xu, Y., Voelter-Mahlknecht, S., and Mahlknecht, U. (2005) The histone deacetylase inhibitor suberoylanilide hydroxamic acid downregulates expression levels of Bcr-abl, c-Myc and HDAC 3 in chronic myeloid leukemia cell lines. Int. J. Mol. Med. 15 (1) 169-172. (41) Gravemann, U. (2002) Synthesis of achiral, racemic and enantiomerically pure valproic acid derivatives with anticonvulsant, neurotoxic and teratogenic potency. Ph.D. Thesis, University of Hannover, Hannover, Germany, http://edok01.tib.uni-hannover.de/edoks/e01dh03/ 359589863.pdf. (42) Smith, P. K., Krohn, R. I., Hermanson, G. T., Mallia, A. K., Gartner, F. H., Provenzano, M. D., Fujimoto, E. K., Goeke, N. M., Olson, B. J., and Klenk, D. C. (1985) Measurement of protein using bicinchoninic acid. Anal. Biochem. 150 (1), 76-85 (erratum in (1987) Anal. Biochem. 163 (1), 279). (43) Marks, P. A., Richon, V. M., and Rifkind, R. A. (2000) Histone deacetylase inhibitors: Inducers of differentiation or apoptosis of transformed cells. J. Natl. Cancer Inst. 92 (15) 1210-1216. (44) Gershall, L. J. (2001) Targeting histone acetylation as a novel approach to cancer therapy. P&S Med. ReV. 7 (2) 21-26.

TX0502241