Dioxin Silences Gonadotropin Expression in Perinatal Pups by Inducing Histone Deacetylases

Background: Dioxin attenuates the fetal production of luteinizing hormone (LH) to imprint impaired sexual behavior. Results: Fetal exposure to dioxin induces histone deacetylases to attenuate the acetylation status of histones twisted around the LHβ gene. Conclusion: Histone deacetylation in the pituitary contributes to the dioxin-induced damage to fetal/infant steroidogenesis. Significance: This study provides a new insight into the mechanism underlying sexual immaturity caused by dioxin. Maternal exposure to 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD) causes the impairment of reproduction and development in the pups. Our previous studies have revealed that maternal treatment with TCDD attenuates the fetal production of pituitary gonadotropins (luteinizing hormone (LH) and follicle-stimulating hormone) at gestational day (GD) 20, leading to the impairment of sexual behavior in adulthood. However, the mechanism underlying such a reduction has remained unknown until now. When pregnant rats at GD15 were given an oral dose of TCDD (1 μg/kg), the testicular expression of steroidogenic proteins was reduced between GD20 and postnatal days (PND) 2. In accordance with this, the pituitary expression of gonadotropin β-subunit and serum gonadotropin were also attenuated from GD20 to PND0 in a pup-specific fashion. To identify the target genes linked to a fetal reduction in gonadotropin β-subunit, we performed a DNA microarray analysis using the fetal pituitary and its regulatory organ, the hypothalamus. The results obtained showed that TCDD induced histone deacetylases (HDACs) in the fetal pituitary. In support with this, TCDD markedly deacetylated histones H3 and H4 twined around the promoter of the fetal LHβ gene. This effect was fetus- and LHβ-specific, and this was not observed in the maternal pituitary or for other pituitary hormone genes. Finally, an LHβ reduction caused by TCDD was completely restored by maternal co-treatment with valproic acid, an HDAC inhibitor. These results strongly suggest that the increased deacetylation of histone owing to HDAC induction plays a critical role in the TCDD-induced reduction in LHβ in the fetal pituitary.

Dioxins, such as 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD), 2 are an important class of representative environmental pollutants, and their harmful effects on humans and wild life continue to be of much concern (1, 2). In laboratory animals, maternal exposure to TCDD causes a number of developmental, and reproductive toxic effects in the offspring (2,3). These disorders are more serious because they are caused by TCDD at much lower doses than those needed to produce acute toxicity in the mother. Many investigations have been conducted to identify the mechanism whereby dioxin exerts its effect on reproduction and development. These studies have demonstrated a possible effect of dioxin on estrogen receptor signaling (4,5), the expression of steroid hormone receptors (6,7), the synthesis and secretion of sex steroids (8), and the expression of steroid-metabolizing enzymes (9). However, greater efforts are needed to understand how the above mechanisms or their combination contribute to this toxicity.
Our previous studies have demonstrated that maternal treatment with TCDD at the late gestational stage attenuates the fetal production of pituitary gonadotropins, the regulators of gonadal steroidogenesis (10 -12). For example, treating pregnant Wistar rats at gestational day (GD) 15 with 1 g/kg TCDD significantly reduces the fetal expression, at GD20, of luteinizing hormone (LH) and FSH in both pituitary mRNA and circulating hormone levels. Although both gonadotropins contain a common ␣-subunit (␣GSU) and a specific ␤-subunit, a reduction in the pituitary mRNA level was more sensitive to the ␤-subunit. The expression of ␣GSU was either little affected (10) or reduced slightly but significantly (12). In accordance with a change in circulating LH/FSH, TCDD causes a reduction in the expression of steroidogenic proteins, including steroido-genic acute-regulatory protein (StAR) and cytochrome P450 (CYP) 17 in the fetal gonads (10 -12). In one of the above studies, direct supplementation of equine chorionic gonadotropin (eCG), an LH-mimicking hormone, into the fetuses exposed to TCDD at GD15 was shown to restore not only the attenuated expression of fetal steroidogenic proteins but also the impairment of sexual behaviors after growing up (12). These observations strongly suggest that TCDD initially reduces gonadotropin biosynthesis in the fetal pituitary, leading to the imprinting of defects in sexual behaviors due to a shortage of sex steroids the production of which must take place in the fetal and newborn stages for proper sexual maturation (13,14).
The synthesis of gonadotropin subunits in the anterior pituitary is mainly stimulated by gonadotropin-releasing hormone (GnRH) secreted from neurons which extend to the pituitary by way of the hypothalamus (15)(16)(17)(18)(19). GnRH binds to its specific receptor on the membrane of pituitary gonadotropes, and activates several signaling pathways involving PKC, PKA, MAPKs, and calcium/calmodulin-dependent protein kinases (16). The activated kinases modulate the action of transcription factors and repressors located downstream to regulate the expression of gonadotropins in a subunit-specific manner (17)(18)(19). Our recent study has suggested that TCDD suppresses GnRH-mediated PKC and PKA and/or their downstream mechanisms in cultivated fetal pituitary to attenuate gonadotropin synthesis (20). In addition, TCDD reduces a number of fetal hypothalamic constituents, including neurotransmitters regulating GnRH secretion (21). Therefore, it would be reasonable to consider that gonadotropin production in the fetal pituitary is damaged by a combination of the TCDD effects on the pituitary and its regulatory organ, the hypothalamus, although many more studies are needed for a fuller understanding. More basically, the following question remains unanswered: what is the mechanism underlying a ␤-subunitand fetal age-specific reduction by TCDD in LH/FSH expression? Regarding the fetal age-specific nature, the details of its onset and termination should also be clarified, although we observed in a previous work that an LH␤ reduction by TCDD disappeared 1 week after birth (21).
To address the above issues, the present study thoroughly investigated the effect of TCDD on steroidogenesis governed by the hypothalamus-pituitary-gonadal axis in pups during the perinatal stages. Also, we conducted a DNA microarray analysis focusing on the fetal pituitary and hypothalamus to attempt to identify the target genes linked to a reduction in the expression of the gonadotropin ␤-subunit. As the DNA microarray analysis demonstrated that the induction of histone deacetylases (HDACs) may be the trigger to suppress LH␤ expression, we further investigated whether TCDD modifies the acetylation status of histones interacting with the LH␤ gene. In addition, DNA methylation of the promoter region of gonadotropin genes was also investigated.

EXPERIMENTAL PROCEDURES
Materials-TCDD was obtained from AccuStandard, Inc. (New Haven, CT). A rabbit antibody against rat StAR and mouse anti-␤-actin monoclonal antibody were purchased from Abcam Ltd. (Cambridge, UK) and BioVision Inc. (Mountain View, CA), respectively. Antibodies against acetylhistones H3 and H4 were purchased from Millipore Corp. (Billerica, MA). The other reagents were of the highest grade commercially available.
Animals and Treatments-All experiments were approved by the Institutional Animal Care and Experiment Committee of Kyushu University. Female (7-week-old) and male (10-weekold) Wistar rats were purchased from Kyudo Co. Ltd. (Kumamoto, Japan). All animals were bred on a standard chow (CE-2; CLEA Japan, Tokyo, Japan) and sterilized water ad libitum, and kept in an environmentally controlled room maintained at 22 Ϯ 5°C and 50 Ϯ 15% relative humidity under a 24-h light/dark cycle (light period, 7:00 a.m.-7:00 p.m.). Female rats were paired overnight with male rats. Next morning, the presence of sperm in the vaginal smears was checked by microscopy (ϫ400) to confirm pregnancy. When sperm was detected, the day was designated as GD0 of pregnancy, and the pregnant rats were housed alone before the experiment was started. At GD15, the pregnant rats were given an oral dose of TCDD (1 g/kg/2 ml corn oil). Then, tissues and blood were removed from the fetuses, neonates and dams at GD18, 19, 20, and 21 and postnatal day (PND) 0, 2, 4, and 7 (PND0 means birth day). In a separate experiment, pregnant rats were co-treated with TCDD and sodium valproate (VPA; Enzo life Sciences Inc., Farmingdale, NY), an HDAC inhibitor (22). In this case, TCDD was given to rats in the same schedule as described above, and VPA was administered once a day at a dose of 200 mg/kg/2 ml saline from GD16 to GD19. Fetal pituitaries were removed for analysis at GD20. When direct injection of reagent solution to fetuses was required, this was conducted according to the method reported previously (10). Briefly, the dam exposed to TCDD (1 g/kg, p.o., GD15) was anesthetized at GD17 with sodium pentobarbital, and her uterus was pulled out for the injection of one of the following reagents into the fetuses: eCG (5 I.U./5 l PBS: Sigma-Aldrich), GnRH (0.5 nmol/5 l PBS: Peptide Institute Inc., Osaka, Japan), and kisspeptin-10 (0.5 nmol/5 l PBS: Peptide Institute Inc.). Control fetuses were treated with vehicle alone. After treatment, the uterus containing the fetuses was returned to the abdomen, and the surgical wound was stitched up. The fetuses were removed again at GD20 for analysis.
RT-PCR-The expression of mRNAs was quantified by realtime RT-PCR according to the method described previously (21). In brief, total RNA was extracted from the testis, pituitary, and hypothalamus using an RNeasy Mini Kit (Qiagen GmbH, Hilden, Germany). The RNA (250 ng) obtained was treated with DNase I to digest the contaminated genomic DNA, and reverse-transcribed to its cDNA. Target mRNAs were amplified with Fast SYBR Green Master Mix (Invitrogen, Carlsbad, CA), using a StepOnePlus Real-time PCR system (Invitrogen). The primer designs are shown in the supplemental Table 1. The PCR conditions were as follows: 95°C for 20 s for 40 cycles (95°C for 3 s and 60°C for 30 s). The amount of quantified target mRNA was normalized by ␤-actin mRNA.
Immunoblotting-The expression of testicular StAR and ␤-actin proteins was analyzed by immunoblotting according to the method described earlier (12). The testes were removed from fetuses and infants during GD18 and PND7, and those from the all pups of one dam were pooled. The testes were homogenized, and centrifuged at 1,000 ϫ g for 10 min. The supernatant was centrifuged at 9,000 ϫ g for 25 min to obtain the mitochondrial fraction. This sample was electrophoresed, and detection was carried out using anti-StAR and anti-␤-actin IgGs as the primary antibodies. The amounts of proteins used for electrophoresis were; StAR: 20 g (GD18, 19, 20, and 21, and PND0 and 2) or 50 g (PND4 and 7); and ␤-actin: 5 g (GD18, 19, 20, and 21) or 2 g (PND0, 2, 4, and 7).
Measurement of GnRH, LH, and FSH Concentrations-The content of hypothalamic GnRH was quantified by enzyme immunoassay (LH-RH EIA kit, Phoenix Pharmaceuticals Inc., Burlingame, CA). Fetal hypothalamus removed from the dam was homogenized with 4 volumes, relative to the tissue weight, of 0.1 M HCl, and the homogenate was diluted five times with H 2 O before assay. The serum contents of LH and FSH were determined using commercial kits (Endocrine Technologies, Inc., Newark, CA). Fetal serum was used in the LH assay without dilution, and neonatal serum (PND0 -7) was diluted four times with the sample diluent supplied in the kit before assay. In the case of FSH, fetal serum (GD18 -21) and neonatal serum (PND0 -7) were diluted twice and four times, respectively, with H 2 O before assay.
DNA Microarray-Total RNA was isolated from the fetal pituitary (GD20) using an RNeasy Mini Kit (Qiagen). The RNA was purified by ethanol precipitation, and dissolved in RNasefree water. This preparation was electrophoresed on an Experion system (Bio-Rad), and its quality was judged by measuring the ratio of band intensity between 28 S and 18 S rRNA. The RNA passing the quality check was subjected to a DNA microarray analysis. The total RNA (250 ng) was converted to its biotinylated-cRNA according to the manufacturer's procedures (Illumina TotalPrep RNA Amplification Kit; Ambion, Austin, TX). Briefly, the reverse transcription of extracted mRNA to the first-strand cDNA was performed for 2 h at 42°C, using an oligo (dT) primer bearing a T7 promoter and reverse transcriptase (ArrayScript; Ambion). To prepare second strand cDNA, DNA polymerase I and ribonuclease H were added to the above reaction mixture, and the solution was further incubated for 2 h at 16°C. After cDNA purification by a cDNA filter cartridge, the eluted cDNA was used as the template for in vitro transcription. This reaction was performed at 37°C for 14 h in the presence of T7 RNA polymerase and NTP mix conjugated with biotin, yielding multiple copies of biotinylated antisense RNAs to each mRNA in the sample.
A total of 1500 ng biotinylated-cRNA was overlaid onto individual array spots of the rat microarray chip (RatRef12-v1 Bead-Chip; Illumina). The chip was hybridized at 58°C for 19 h, washed, labeled with fluorescent reagent, and scanned by an array reader (BeadArray Reader; Illumina), according to the protocol. The data on gene expression were compiled by Bead Studio software (Illumina). The microarray datasets were submitted to the GEO database (accession number: GSE32459; provided in the public domain by NCBI). The signal intensity for gene expression was transformed to logarithmic variables (log 2 ), and normalized by a quantile algorithm: for this, the preprocessCore library package (23) in Bioconductor software (24) was used. The genes agreeing with the criterion that their expression was detected in 4 or more samples out of a total of 6 samples (3 controls ϩ 3 TCDD-treated samples) at detection p values of more than 0.05 were selected and further analyzed. To identify any significant differences, the data were processed using a package in the Bioconductor, Linear Models for Microarray Analysis (limma) (25), and the criterion for a significant difference between the control and TCDD-treated groups was set at limma p Ͻ 0.01.
ChIP Analysis-The acetylation status of histones which twist around the promoter regions of gonadotropin genes was analyzed by procedures described below. Unless otherwise stated, the following procedures were carried out at 4°C. The pooled fetal pituitary (male, GD20) removed from one dam and the maternal pituitary were homogenized in 180 l of 10 mM Tris-HCl (pH 8.0) containing 10 mM NaCl, 0.2% Nonidet P40, and 1 mM PMSF. The homogenates were treated with 1% formaldehyde at 37°C for 5 min to cross-link DNA and chromatin protein. The excess formaldehyde was inactivated by adding 0.25 M glycine, and the mixture was centrifuged at 700 ϫ g for 5 min. The pellets were washed three times with 1 ml of PBS containing 1 mM PMSF, and suspended in 300 l lysis buffer: 50 mM Tris-HCl (pH 8.1) containing 167 mM NaCl, 1.2 mM EDTA, 1.1% Triton X-100, 0.01% SDS, and 1 mM PMSF. The sample obtained was sonicated for 40 min to cleave partially the DNA coupled with protein, and then centrifuged at 12,000 ϫ g for 10 min. The lysate (300 l) was diluted twice with 16.7 mM Tris-HCl (pH 8.1) containing 1.2 mM EDTA, 167 mM NaCl, 1.1% Triton X-100, 0.01% SDS, and 1 mM PMSF, and a portion (100 l) of the solution was saved as the input control for normalization. The remainder were pretreated with Dynabeads Protein G (Invitrogen; 450 g) for 2 h to remove nonspecific substances capable of binding to protein G. The Dynabeads were removed from the mixture using a DynaMag TM -Spin (Invitrogen), and the resulting solution was equally divided into two tubes (each 250 l). To each solution, either 2.5 g protein of anti-acetylhistone H3 antibody or 2.5 g of anti-acetylhistone H4 antibody and rabbit pre-immune IgG were added and the mixtures were gently rotated overnight. After that, the antibody-histone complex was gathered by rotating with Dynabeads Protein G (750 g) for 90 min at room temperature. The beads were washed five times with 1-ml portions of 1 mM Tris-HCl (pH 8.1) containing 0.25 M LiCl, 1% Nonidet P40, 1 mM EDTA, and 1% deoxycholate, and then washed twice with 1 ml of TE buffer (10 mM Tris-HCl (pH 8.0), 1 mM EDTA). To elute antibody complexes from the beads, an elution solution (150 l) consisting of 1% SDS, 0.1 M NaHCO 3 and 10 mM DTT was added to the beads, and the mixture was gently rotated for 15 min at room temperature. This step was repeated one more time. The elution solutions were pooled, and incubated with 0.2 M NaCl at 65°C for 4 h to reverse the cross-link between DNA and histone. After treatment with proteinase K (10 g) for 1 h at 45°C, the DNA was purified by phenol-chloroform extraction, and dissolved in 100 l of TE buffer. The promoter regions of the pituitary hormone genes in the prepared DNA were determined by real-time PCR as described above (see the section of RT-PCR). The primer designs are shown in supplemental Fig.  1A. The relative level of histone acetylation was calculated by normalizing with the input DNA. To prepare a gel image, the DNA sample diluted twice was amplified by ExTaq DNA polymerase (Takara-bio, Shiga, Japan) under the following con-ditions: 94°C for 2 min for 33 (acetyl-H4) or 36 (acetyl-H3) cycles (94°C for 30 s, 58°C for 30 s, and 72°C for 30 s) and 72°C for 3 min. Amplified DNA was electrophoresed in 1.5% agarose gel and stained with ethidium bromide.
DNA Methylation Analysis-We initially investigated whether there were CpG sites on the gonadotropin genes, using Methyl Primer Express Software v1.0 (Invitrogen) under the default setting (CpG island size: 300 -2000 bp, CϩGs/total bases: Ͼ50%, and CpG observed/CpG expected: Ͼ0.6). Examination of the 5Ј-upstream gene (10 kbp from the transcriptional start site) suggests the presence of one and three postulated CpG islands on ␣GSU and LH␤ genes, respectively. The methylation of ␣GSU and LH␤ promoters in the male fetal pituitary was investigated by methods using methylation-sensitive restriction enzyme (MSRE) (26) and bisulfite DNA sequencing (27). In the MSRE method, we used BstUI as the restriction enzyme. BstUI digests DNA at the CGCG sequence, while it cannot cleave its recognition site involving methylated cytosine. The genomic DNA in the fetal pituitary was extracted using a DNeasy blood and tissue kit (Qiagen). The DNA (1 g) was equally divided into two tubes, and incubated at 60°C for 1 h in the presence and absence of BstUI (10 units; New England BioLabs, Ipswich, MA). Both samples were subjected to realtime PCR (see the section of RT-PCR). The primers were designed so as to include the BstUI site in the amplified product. The primer sequences are described later together with the information concerning the location of the region sensitive to methylation (CpG island) in the gene promoter (see "Results"). The level of CpG methylation was quantitatively evaluated by dividing the amount of PCR product with BstUI treatment by that without such treatment. For the negative control in the above experiment, a separate preparation of DNA (1 g) was pretreated with sssI methylase (New England BioLabs) for 1 h at 37°C. As sssI methylase methylates all cytosines within the double-stranded 5Ј-CG-3Ј, BstUI cannot digest any CGCG site of sssI-treated DNA. We confirmed that the PCR of sssI methylase/BstUI-treated DNA yields an amplified product the quantity of which is the same as the product from the DNA template not treated with both enzymes.
In bisulfite DNA sequencing analysis, the genomic DNA extracted from the fetal pituitary was modified with sodium bisulfite, using an EpiTect Plus DNA Bisulfite Kit (Qiagen). The bisulfite-treated DNA (2 l) was amplified in 50 l of reaction mixture containing EpiTaq TM HS (2.0 unit; Takara-bio), 2.5 mM MgCl 2 , 0.3 mM dNTP mixture and each 0.4 M of forward and reverse primers (see supplemental Table 2 for their sequences). The conditions used for PCR amplification were as follows: 95°C for 3 min for 45 cycles (95°C for 30 s, 56°C for 30 s, and 72°C for 1 min) and 72°C for 5 min. The PCR products were electrophoresed in 1.5% agarose gel, and purified using a QIAquick Gel Extraction Kit (Qiagen). Each DNA fragment was inserted by TA ligation into a nicked pGEM-T Vector (Promega, Madison, WI), and this was transfected to JM109 competent cells for cloning. The plasmids were prepared from transformant colonies, and the sequences of their passenger DNAs were determined using a BigDye Terminator v3.1 Cycle Sequencing kit (Invitrogen) and the following primer: 5Ј-TCA-AGCTATGCATCCAACGC-3Ј. In this experiment of bisulfite DNA sequencing, a total of 6 fetuses at GD20 (3 from 3 control dams and another 3 from 3 TCDD-treated dams) were used for analysis. The plasmids obtained from 10 transformant colonies/one fetal sample were applied to sequencing.
Statistical Analysis-Data for the fetuses and neonates in one dam were averaged to become a single analytical unit. The statistical difference between the control and TCDD group was compared by Student's t test. The comparison among multiple groups was conducted by one-way analysis of variance with a post-hoc test (Tukey's multiple comparison test), using GraphPad Prism Version 5 software (GraphPad Software, Inc., SanDiego, CA). The statistical significance was set at p Ͻ 0.05.

RESULTS
The Perinatal Age Specificity of the TCDD Effect on Gonadal Steroidogenesis and Its Regulatory Mechanisms-To clarify age specificity in terms of a TCDD-produced reduction in gonadal steroidogenesis, we analyzed the expression of steroidogenic proteins and their regulatory hormones in the testes, pituitary and hypothalamus from GD18 to PND7. When pregnant rats were given an oral dose of TCDD (1 g/kg) at GD15, testicular mRNA coding for StAR, a cholesterol-transporting protein associated with a late-limiting process of steroidogenesis (28), was attenuated in the fetal (GD18, 20, and 21) and neonatal (PND2) testes (Fig. 1A). Similarly, the mRNA of CYP17, an essential enzyme for sex-steroid biosynthesis, was reduced from GD20 to PND2 (Fig. 1A). The reduced expression of StAR protein was also observed during the same period as the change in the mRNA (Fig. 1B). The gonadal expression of steroidogenic proteins is regulated by gonadotropin secreted from the anterior pituitary. In accordance with changes in StAR/CYP17, maternal exposure to TCDD attenuated the expression of pituitary gonadotropins in the litter in a perinatal stage-specific manner. For example, the treatment of dams with TCDD reduced the expression of LH␤ and FSH␤ mRNAs from the beginning at GD19 (20) to its termination at PND0 (Fig. 2A). Although the expression of ␣GSU mRNA was also reduced, its sensitivity to TCDD seems weaker than those of the ␤-subunits because a significant reduction in ␣GSU expression was seen only at GD21 (Fig. 2A). The ␤-subunit of thyroid-stimulating hormone (TSH␤), one of the other pituitary hormones, did not exert any change during the fetal and neonatal stages (supplemental Fig. 2). In parallel with LH/FSH mRNAs, TCDD produced a reduction in the serum levels of these hormones in perinatal pups, and the disorders returned to the normal situation at PND7 (Fig. 2B). In sharp contrast to perinatal pups, the expression of LH subunits in the maternal pituitary remained unchanged or increased (GD21) rather than decreased after TCDD treatment (supplemental Fig. 3). These results showed that maternal exposure to TCDD damages the pituitary-gonadal axis for steroidogenesis only during the last fetal and early postnatal periods. In addition, the data obtained suggest that such a disorder occurs mainly through initial targeting of LH␤/ FSH␤ subunits. The direct supplementation of eCG, an LHmimicking hormone, into fetuses exposed to TCDD at GD15 completely restored the TCDD-induced attenuation of StAR mRNA in the fetal testis (Fig. 2C). This observation agrees with the above mechanism.
As described before, hypothalamic GnRH is a main regulator of gonadotropin biosynthesis. However, neither the expression of GnRH mRNA nor its hormone level in fetuses/infants was altered even after TCDD exposure (Fig. 2, D and E), whereas pituitary mRNA coding for GnRH receptor was reduced by this treatment (Fig. 2F). In agreement with this, GnRH supplementation given to TCDD-treated fetuses failed to restore the reduced expression of testicular StAR (Fig. 2C). Similarly, kisspeptin, a stimulator of GnRH secretion, lacked the ability to improve the reduction (Fig. 2C). These pieces of evidence suggest that the mechanism whereby TCDD reduces the pituitary synthesis of LH/FSH cannot be explained by a change in the hypothalamic content of GnRH, while the pituitary GnRH receptor is one possible target to cause damage to the fetal pituitary-gonadal axis.
DNA Microarray Analysis to Identify the Target Gene for a Fetal Reduction in Gonadotropin ␤-Subunits-To identify the genes which are linked to the attenuated expression of gonadotropin ␤-subunits by TCDD, we performed a DNA microarray analysis, focusing on the fetal pituitary and hypothalamus at GD20. The result obtained showed that many changes in the FIGURE 1. The onset and termination of a TCDD-produced reduction in the expression of steroidogenic proteins during perinatal stages. Pregnant dams were exposed to TCDD (1 g/kg, p.o.) at GD15, and the fetal and neonatal testes were removed from GD18 to PND7. A, gonadal expression of StAR and CYP17 mRNAs was determined by real-time RT-PCR, and normalized by ␤-actin mRNA. Each plot represents the mean Ϯ S.E. of 12-20 fetuses (neonates), 2 of which were removed (born) from 6 -10 different dams. B, expression of StAR protein was analyzed by immunoblotting. The typical band patterns of each group are shown in upper panels. The relative level of StAR expression was calculated by normalization to ␤-actin. In these experiments, fetal (neonatal) testes in (from) one dam were pooled for analysis. The bars represent the means Ϯ S.E. of 6 -10 dams (the number of dams used is shown beneath each figure). Significantly different from the control group; *, p Ͻ 0.05; **, p Ͻ 0.01; and ***, p Ͻ 0.001. FIGURE 2. The effect of maternal exposure to TCDD on the expression of gonadotropins, GnRH, GnRH receptors, and StAR during fetal and neonatal stages. In all experiments, pregnant dams were exposed to TCDD (1 g/kg, p.o.) at GD15, and the fetal and neonatal pituitary (A and F), serum (B), and hypothalamus (D and E) were removed during GD18 and PND7. The mRNAs coding for pituitary gonadotropins (A) and GnRH receptor (F), and hypothalamic GnRH (D) were determined by real-time RT-PCR, and normalized by ␤-actin mRNA. Each inset is a graph which magnifies the data of GD18-21. The serum concentration of LH and FSH (B), and the hypothalamic content of GnRH (E) were measured by ELISA and EIA, respectively. Each value represents the mean Ϯ S.E. of 6 -10 fetuses (neonates) which were removed (born) from 6 -10 different dams. Significantly different from the control group; *, p Ͻ 0.05; **, p Ͻ 0.01; and ***, p Ͻ 0.001. In experiment C, the fetuses exposed to TCDD (maternal dose of 1 g/kg, GD15) were treated directly either with eCG (5 I.U.), GnRH (0.5 nmol), or kisspeptin (0.5 nmol) at GD17, and fetal testes were removed at GD20. Bars are the means Ϯ S.E. of 5-6 fetuses. Significant difference between the pair indicated; *, p Ͻ 0.05 and **, p Ͻ 0.01. gene expression occurred in the fetal pituitary following maternal exposure to TCDD: namely, 245 and 396 genes were up-and down-regulated significantly by TCDD, respectively (Fig. 3A: refer to GSE32459 by NCBI for details of the altered genes). Although TCDD also changed gene expression in the fetal hypothalamus (increase, 116 genes; and decrease, 104 genes) (Fig. 3B), the number of altered genes was rather fewer than that seen in the pituitary. In addition, the dioxin-responsive genes, such as CYP1A1, aryl hydrocarbon receptor (AhR) repressor and NAD(P)H quinone oxidoreductase 1, showed no significant change in the fetal hypothalamus. While we were unable to find any hypothalamic genes which would be expected to be linked to a reduction in the expression of pituitary gonadotropin, we found that TCDD induces HDAC5 and 11 in the pituitary. To validate the outcome of the microarray analysis, we performed real-time PCR. The results showed that HDAC1, 3, 4, 6, 7, and 9 as well as HDAC5 and 11 were induced by TCDD in the fetal pituitary at GD20 (Fig. 4). The induction of HDAC1, 5 and 7 was observed during late prenatal and early neonatal stages, when a TCDD-induced reduction in pituitary gonado-tropin took place concomitantly (Figs. 2 and 4). On the other hand, TCDD scarcely affected HDAC expression in the fetal hypothalamus and maternal pituitary (supplemental Fig. 4). Thus, HDAC induction emerged as a possible candidate explaining a TCDD-induced reduction in the fetal expression of gonadotropin ␤-subunits.
The Effect of TCDD on the Histone Acetylation of Gonadotropin Promoter-To assess the role of HDAC induction on gonadotropin suppression, we performed a ChIP analysis to determine the effect of TCDD on the acetylation level of fetal histones twinning around pituitary hormone genes. When we determined acetylated histones H3 and H4 associated with the promoter regions of LH␤, FSH␤, ␣GSU, TSH␤, and growth hormone (GH) genes, only the acetylation of histone on the LH␤ promoter was significantly reduced by maternal treatment with TCDD (Fig. 5). Although acetylated histones also tended to be reduced by TCDD for four other hormone genes, no significant change was detected in any of these cases (Fig. 5). Similarly, TCDD did not affect the acetylation status of histones H3 and H4 on the LH␤ promoter of the maternal pituitary (supplemental Fig. 1, B and C). These data suggest that TCDD deacetylates histones H3 and H4 associated with the LH␤ promoter to attenuate gonadotropin expression in an LH␤-and fetus-specific manner.
Absence of the TCDD Effect on DNA Methylation in the Promoter of Gonadotropin Genes-We then examined whether TCDD affects the status of methylation at the site 5Ј-upstream of the gonadotropin genes, using the MSRE and bisulfite DNA sequencing methods. Prior to the analysis, we firstly investigated the location of the CpG site in the gonadotropin genes (see "Experimental Procedures" for details). The computer-assisted survey of the 5Ј-upstream region (10 kbp from the transcriptional start site) found one and three CpG sites on the ␣GSU and LH␤ genes, respectively (Fig. 6A). On the other hand, the FSH␤ gene did not have any CpG site within 20 kbp. When the level of DNA methylation was examined in the CpG islands of ␣GSU and LH␤ genes, TCDD did not alter the methylation status at any of the sites (Fig. 6). These results suggest that TCDD has no influence on the DNA methylation of the gonadotropin gene in the fetal pituitary.
Recovery from a TCDD-induced Attenuation of LH␤ in the Fetal Pituitary-If a TCDD-induced reduction in LH␤ expression is caused by HDAC induction, the suppression of HDAC activity should block the reduction. To address this hypothesis, the pregnant dams exposed to TCDD at GD15 were co-treated with VPA, an HDAC inhibitor, from GD16 to 19. When the pituitary expression of fetal LH␤ mRNA was analyzed at GD20, a TCDD-induced reduction in LH␤ was completely restored by VPA co-treatment (Fig. 7A). In agreement with the result that the deacetylation of histones was specific to the LH␤ gene (see Fig. 5), maternal co-treatment with VPA failed to protect fetuses from an FSH␤ reduction (Fig. 7A). Moreover, VPA also rescued fetuses from TCDD-induced attenuation in serum LH content (Fig. 7B). These observations support the mechanism that TCDD induces fetal pituitary HDAC to reduce LH␤ expression.

DISCUSSION
This study proposed HDAC induction as an explanation of the following two questions: 1) how does TCDD attenuate the fetal expression of gonadotropin ␤-subunits? and 2) what is the mechanism for the fetal specificity of the damage? The observations that HDAC induction by TCDD occurred only in the fetal pituitary, and HDAC inhibition by VPA protected only LH␤ from TCDD-mediated suppression are persuasive evidence in support of this.
This study firstly examined the onset and continuation of TCDD-produced damage to the pituitary-gonadal axis. The results obtained showed that TCDD attenuated the expressions of gonadal steroidogenic proteins and pituitary gonadotropins only in fetuses/infants but not in their dams during the short period from GD20 to PND2. Since supplying gonadotropin (eCG) to fetuses exposed to TCDD restored the attenuated expression of gonadal StAR mRNA, this strongly suggests that a reduction in the gonadotropin ␤-subunit caused by TCDD trig- . Effect of TCDD on the pituitary expression of HDAC mRNAs in the male fetuses/neonates. Pregnant dams were exposed to TCDD (1 g/kg, p.o.) at GD15, and the fetal and neonatal pituitaries were removed from GD18 to PND7. The tissue level of mRNAs coding for HDAC isoforms was analyzed by real-time RT-PCR. Each plot represents the mean Ϯ S.E. of 6 fetuses (or neonates), which were removed (or born) from 6 different dams. Significantly different from the control group; *, p Ͻ 0.05; **, p Ͻ 0.01; and ***, p Ͻ 0.001.
gers the attenuated expression of steroidogenic protein in the fetal and neonatal testes. Conceivably, a reduction in gonadotropin content may be an outcome of the damage caused by TCDD to the upstream regulators governing gonadotropin expression. Regarding this issue, the fetal/infant content of hypothalamic GnRH, a regulator of gonadotropin synthesis, did not vary during the perinatal period even after TCDD treat-ment. In agreement with this, direct injection of GnRH and its regulator, kisspeptin, into TCDD-exposed fetuses did not have any beneficial effect on a reduction in the expression of steroidogenic proteins. On the other hand, TCDD attenuated the pituitary expression of GnRH receptor from GD20 to PND0, and this period is apparently consistent with that seen for a reduction in the expression of gonadotropin ␤-subunits. Therefore, it is reasonable to assume that TCDD reduces steroidogenesis by attenuating pituitary GnRH receptors rather than the effect on the content of hypothalamic GnRH. However, this mechanism may be inconsistent with the observation that the expression of ␣GSU is scarcely affected following TCDD treatment despite it being a GnRH-governed gonadotropin. While the LH␤ subunit is produced only in the gonadotropes, ␣GSU can be produced not only in the gonadotropes but also in the thyrotropes which mainly serve as a TSH producer (29,30). Such a difference is assumed to be one reason for distinct susceptibility of LH␤ and ␣GSU to TCDD. The data obtained in this study showing that TSH production in the fetal pituitary was resistant to TCDD would support this view. On the other hand, circulating LH is known to maintain its normal concentration, even if the GnRH input falls to below 20% of the normal level (31). This information suggests that a TCDD-produced reduction in the fetal expression of gonadotropins is the result of a combination of damages to not only the GnRH pathway but also other mechanisms.
From the results of the DNA microarray analysis, we found that TCDD has the ability to induce HDAC5 and 11. It has been suggested that HDAC plays an important role as a transcriptional repressor for gonadotropin ␤-subunits (19,32,33). In agreement with this, TCDD markedly deacetylated histones H3 and H4 twined around the promoter of the fetal LH␤ gene, but not the promoters of other pituitary hormone genes (FSH␤, ␣GSU, TSH␤, and GH). Such deacetylation was specific to fetuses, and did not occur in the maternal pituitary. Maternal co-treatment with VPA, one of the HDAC inhibitors, led to a recovery from a TCDD-induced reduction in the fetal expression of LH␤ but not FSH␤. These results strongly suggest that TCDD reduces the fetal expression of LH␤ by silencing its transcription through HDAC induction, while HDAC contributes little to the mechanism for a reduction in FSH␤. The reason why FSH␤ is insensitive to HDAC induction is thought to be due to the isoform specificity in the induction of HDAC by TCDD. It has been reported that while HDAC1, 5, and 7 are preferentially associated with the repression of LH␤ gene, HDAC2 and 3 mainly contribute to the suppression of FSH␤ transcription (33). In the present study, although HDAC1, 5, and 7 were induced in the fetal and neonatal pituitary by maternal exposure to TCDD, HDAC2 was not sensitive to TCDD treatment. Therefore, such a difference could reasonably explain an LH␤-specific reduction, and the induced expression of HDAC1, 5, and 7 seems to contribute to an LH␤ reduction. The difference in maturation between LH-and FSH-producing gonadotropes is also a possible reason for the specificity of the TCDD effect. A recent study has shown that FSH-producing cells acquire GnRH receptors during the last embryonic and early postnatal stages, while LH-producing cells already have a substantial level of GnRH receptors at the fetal stages (34). Pregnant dams were exposed to TCDD (1 g/kg, p.o.) at GD15, and the male fetal pituitary was removed at GD20. The acetylation of histones H3 and H4 in the male fetal pituitary was analyzed by the ChIP method. Briefly, genomic DNA associated with acetylhistones was co-immunoprecipitated with anti-acetylhistone antibodies, and the gene promoters of pituitary hormones in the complex were determined by real-time PCR. In real-time PCR, all the promoter regions, except for that of GH were amplified by dividing them randomly into 2 to 3 regions. For example, three regions (P1, 2 and 3) of the LH␤ promoter were separately amplified. The primers used in the PCR reaction are shown in supplemental Fig. 1. A, each DNA sample amplified by PCR was electrophoresed on 1.5% agarose gel. H3, H4, and NII represent the samples immunoprecipitated with anti-acetylhistone H3, anti-acetylhistone H4 and non-immune IgGs, respectively. B, relative level of histone acetylation was analyzed by real-time PCR, and normalized by the input sample. Bars  Indeed, our current study shows that GnRH treatment fails to induce FSH␤ in cultured fetal pituitary (GD20), although the expression of LH␤ is induced by the same treatment (20). Thus, FSH-producing cells in the fetal pituitary appear not to have enough GnRH receptors. Because the dissociation of HDAC from the promoter of the gonadotropin ␤-subunit gene is facilitated by GnRH stimulation (33), the shortage of GnRH recep-tors in fetal FSH-producing cells would give rise to their insensitivity to gene suppression by HDAC.
The molecular mechanism underlying an FSH␤ reduction caused by TCDD remains obscure. However, the microarray analysis carried out in this study suggests that TCDD suppresses the activin-mediated signaling pathway. Activin is one of the members belonging to the transforming growth factor-␤ FIGURE 6. Absence of the TCDD effect on the methylation of gonadotropin gene promoter in the fetal pituitary. A, location of the CpG island on the 5Ј-flanking region of LH␤ and ␣GSU genes is shown. This was estimated using Methyl Primer Expressா Software v1.0 (Invitrogen). Primer sequences for amplifying the CpG region are underlined, and the restriction sites recognizable by BstUI are in bold print. B, relative level of methylation on the LH␤ and ␣GSU gene was analyzed by MSRE method, and the bars represent the means Ϯ S.E. of 6 -7 fetuses from different dams. C, bisulfite DNA sequencing was performed to analyze the methylation status of LH␤ and ␣GSU genes. In short, genomic DNA was extracted from the fetal pituitary, treated with bisulfite, and then the 5Ј-upstream regions of LH␤/␣GSU genes were amplified by PCR. The amplified DNA inserted into a plasmid vector was transfected to JM109 competent cells for cloning, and the cloned plasmid was applied to sequencing (see "Experimental Procedures" for the details). Ten colonies/one fetal sample were taken, and their plasmids were sequenced. In this experiment, 3 fetuses (fetus No. 1-3) at GD20 were removed from different control dams (one fetus/one dam), and another 3 fetuses (fetus No. 4 -6) were also obtained from different dams treated with TCDD at GD15. In family, and plays a critical role together with GnRH in the production and secretion of FSH (19,(35)(36)(37). It is known that activin activates SMAD proteins, a family of transcription factors enhancing FSH␤ transcription (35,36)(the term SMAD comes from a combination of MAD: mothers against decapentaplegic (Drosophila protein) and SMA (Caenorhabditis elegans protein)). SMAD4 promotes FSH␤ transcription after forming active complexes with SMAD2/3 proteins. Our microarray data suggest that maternal treatment with TCDD reduced the expression level of SMAD4. Therefore, this downregulation is also possibly linked to the attenuated expression of FSH␤. In addition, the microarray data indicated that the fetal expression of pituitary follistatin was induced by TCDD treatment (p Ͻ 0.05, data not shown). Because follistatin prevents activin from binding to its receptors by associating with activin (37), the follistatin induction may inhibit activin signaling, leading to the attenuated expression of FSH␤ in the fetal pituitary.
TCDD had no effect on DNA methylation of the LH␤ promoter of fetal pituitary. This result suggests that the hypoacetylation of histone on the LH␤ promoter due to HDAC induction seems not to require any change in DNA methylation. In general, a number of toxic effects produced by TCDD are thought to be due to the activation of a nuclear receptor, AhR (38 -40). The AhR-dioxin complex binds to its cognitive sequence, the xenobiotic responsive element (XRE), present in the 5Ј-upstream region of the target genes to produce a change in the gene expression (41,42). Since there are several core XREs at the 5Ј-upstream regions (10 kbp) of HDAC genes, except for HDAC2 and 8 which were not induced by TCDD in this study, it is conceivable that while TCDD induces HDAC following accelerated AhR-XRE binding, HDAC2 and 8 are insensitive to TCDD because of an absence of XRE. Our recent study found that the fetal pituitary has the higher expression of AhR than do the fetal hypothalamus and maternal pituitary (data not shown). Taken together, AhR signaling is one possible mechanism for the isoform-, age-and tissue-specific induction of HDAC. However, this must remain uncertain until future studies are carried out to confirm it.
In conclusion, this study provides novel evidence that TCDD enhances histone deacetylation twined around LH␤ gene by inducing HDAC, leading to reduce LH expression in fetal stage. These pieces of evidence also suggest that pituitary HDAC is the target for dioxin-induced dysfunction to reproductive activity as well as the fetal synthesis of gonadotropins. Further studies are needed to clarify the mechanism whereby TCDD induces HDAC in the fetal pituitary.