Journal of Reproduction and Development, Vol. 58, No 1, 2012 —Original Article— Molecular Cloning of LIM Homeodomain Transcription Factor Lhx2 as a Transcription Factor of Porcine Follicle-Stimulating Hormone Beta Subunit (FSHβ) Takako Kato1), Akio Ishikawa2), Saishu Yoshida2), Yoshiya Sano2), Kousuke Kitahara2), Michie Nakayama 2), Takao Susa2) and Yukio Kato1,2,3) 1)Institute of Reproduction and Endocrinology, Meiji University, Kanagawa 214-8571, Japan 2)Laboratory of Molecular Biology and Gene Regulation, Division of Life Science, Graduate School of Agriculture, Meiji University, Kanagawa 214-8571, Japan 3)School of Agriculture, Meiji University, Kawasaki, Kanagawa 214-8571, Japan Abstract. We cloned the LIM-homeodomain LHX2 as a transcription factor for the porcine follicle-stimulating hormone β subunit gene (Fshβ) by the Yeast One-Hybrid Cloning System using the upstream region of –852/–746 bases (b) from the transcription start site, called Fd2, as a bait sequence. The reporter assay in LβT2 and CHO cells revealed the presence of an LHX2-responsive region other than Fd2. A potential LHX2 binding sequence was confirmed as AATTAAT containing a consensus homeodomain binding core sequence AATT by Systematic Evolution of Ligands by Exponential Enrichment analysis. DNase I footprinting demonstrated three AATTAAT sequences located at regions –835/–829, –818/–812 and –806/–800 b in the Fd2 region and 12 binding sites in the distal and proximal regions mostly containing an AATT-core sequence. RT-PCR analysis of Lhx2 expression during porcine fetal and postnatal pituitary development showed a gradual increase from fetal day (f) 40 to postnatal day (p) 8 followed by a slight decrease to p230, suggesting that LHX2 may play its role largely in the late fetal and postnatal periods. The analyses of Lhx2 expression in pituitary tumor-derived cell lines showed their expressions in cell lines including αT31, LβT2 and others. Since LHX2 was previously identified as a transcription factor forCga and the in vitro experiments in the present study suggested that LHX2 regulated the expression of Fshβ, it is possible that LHX2 controls the synthesis of FSH at the transcription level. Key words: FSH, Gene regulation, Glycoprotein hormone, Lhx2, Pituitary (J. Reprod. Dev. 58: 147–155, 2012)

ollicle-stimulating hormone (FSH) is a member of the pituitary observed that GnRH significantly stimulates the gene expression Fglycoprotein hormone family that includes both a luteinizing of c-Jun and c-Fos, components of AP-1 factor [6]. Multiple pro- hormone (LH) and a thyroid-stimulating hormone (TSH). Each of gesterone response elements (PREs) have been identified in the these hormones shares a glycoprotein hormone common α subunit proximal region of ovine [7] and rat Fshβ [8]. Similarly, estrogen (αGSU) and contains a unique β subunit that confers biological response elements (EREs) were also identified in –105/–72 b of specificity on its respective hormone function. The synthesis and ovine Fshβ [9]. It was reported that the –270/–248 b region of rat secretion of FSH and LH are restricted to pituitary gonadotropes. Fshβ is responsive to activin-dependent activation mediating the Since the regulatory mechanisms of the three subunit that transcription factors Smad and PITX2 [10]. A similar region is form gonadotropins are of special interest, several approaches also responsible for Pitx1, the homologue of both PITX2 [11] and have already achieved and provided a better understanding of the FOXL2 [12]. Nevertheless, the mechanism underlying the cell/ molecular mechanisms and regulatory factors governing the basal tissue-specific control of Fshβ-expression remains unclear. and cell-specific expression of the αGSU gene Cga( ) and Lhβ [1]. We have previously reported that the Fd2 region (−852/−746 b On the other hand, our knowledge of the regulation of Fshβ expres- of porcine Fshβ promoter) is a target for many nuclear sion remains limited to the information from extracellular signals. of the porcine anterior pituitary [13] as well as cloned Prop1 [14] GnRH stimulation of FSH is mediated by the protein kinase C and Prx2 [15], both of which belong to the homeodomain tran- signaling cascade via AP-1 sites in the region of –120/–83 bases (b) scription factor family as an Fd2-binding protein. In addition, we from the transcription start site [2–4] and the distal region between have demonstrated that the −852/+10 b region is sufficient for the –4152/–2878 and –2550/–1089 b of ovine Fshβ [5]. We previously pituitary-specific expression of Fshβ [16]. PROP1 is essential for normal production of gonadotropins as well as for the development of PIT1 lineage cells in both humans and mice [17]. This paper Received: July 12, 2011 describes the molecular cloning of an additional transcription Accepted: October 31, 2011 factor for porcine Fshβ, LHX2, which is a member of the LIM- Published online in J-STAGE: December 2, 2011 ©2012 by the Society for Reproduction and Development homeodomain protein family and has been previously identified Correspondence: Y Kato (e-mail: [email protected]) as a transcription factor for Cga [18]. 148 KATO et al.

Materials and Methods been described previously [22]. The sequences were aligned using a TAAT motif as a landmark and were submitted for WebLogo One-hybrid cloning of pituitary transcription factors binding analysis (http://weblogo.berkeley.edu/), a graphical representation to the Fd2 region of a nucleotide sequence alignment able to display the relative The integration of Fd2 (–852/–746 b) of the porcine Fshβ pro- frequency of each nucleotide at any position [23]. moter into the of yeast YM4271 and the screening of Fd2-binding protein from an adult porcine pituitary cDNA library Cell culture, transfection and reporter gene assay constructed in pAD-GAL4 (Stratagene, La Jolla, CA, USA) has A transient transfection assay was carried out using LβT2 cells been described [14]. and Chinese hamster ovary cells (CHO) as already described [21]. LβT2 is the mouse pituitary gonadotrope lineage cell line [24] Production of recombinant protein endogenously expressing gonadotropin genes of Cga, Lhβ and Complimentary DNAs of a full-length porcine LHX2 and a trun- Fshβ [25] and was kindly provided by Dr PL Mellon (University cate LHX2 (ΔLIM-LHX2), which is deleted the N-terminal region of California, San Diego, CA, USA). After incubation for more (amino acid residues 1–203) containing two LIM domains, were than 48 h, an aliquot (5 µl) of cultured medium was assayed for cloned in-frame into the pET32a vector to generate a TrxA-His-Tag secreted alkaline phosphatase activity using the Phospha-Light fusion protein (Novagen, Madison, WI, USA). The cDNAs were Reporter Gene Assay System (Applied Biosystems) according then transformed into BL21-CodonPlus (DE3)-RIPL (Stratagene). to the manufacturer’s instructions with a MiniLumat LB 9506 The recombinant proteins were expressed and prepared using an luminometer (Berthold, Wildbad, Germany). All values were Overnight Express Autoinduction System 1 (Novagen), followed by expressed as means ± SD of quadruplicate transfections in two purification using MagExtractor-His-Tag- (Toyobo, Osaka, Japan). independent experiments. Statistical significance was calculated by Student’s t-test. The reproducibility and reliability of the reporter Electrophoretic mobility shift assay (EMSA) and DNase I assay without internal control used in this study was described in footprinting assay our previous paper [26]. The production of FAM-labeled DNA fragments was described previously [15]. EMSA and DNase I footprinting were also carried Quantitative real-time RT-PCR out as previously described [14, 19]. Total RNAs, which were extracted from the porcine anterior pituitaries of German Landrace pigs with intact gonads of both Construction of reporter vectors and expression vector sexes during the fetal (f40, f50, f65, f82, f95 and f110) and postnatal The following procedures were described previously [15]. periods (p8, p60, p160 (prepuberty) and p230 (sexually matured)), Serial truncated upstream regions of the porcine Fshβ (Accession were kindly supplied by Dr. F. Elsaesser and pooled for 1–6 individu- no. D00621) [20] were obtained by PCR and ligated to the se- als of the respective age and sex for cDNA synthesis as previously creted alkaline phosphatase (SEAP) plasmid vector, pSEAP2-Basic described [27]. Total RNAs were extracted from pituitary tumor (Clontech Laboratories, Inc., Mountain View, CA, USA), result- cell lines of the mouse (αT1-1, αT3-1, LβT2, LβT4, TαT1, MtT/S) ing in the reporter vectors FSHβ (–2320/+10), FSHβ (–1965/+10), and rat (AtT-20) and a nonpituitary mouse cell line, L929. The FSHβ (–985/+10), FSHβ (–596/+10), FSHβ (–238/+10) and FSHβ characteristics of each cell line and their cDNA syntheses were (–103/+10) for FSHβ promoter. A deletion mutant of −745/−104 described in a previous paper [28]. leaving Fd2 and the endogenous proximal promoter region (ΔFSHβ Real-Time PCR was performed on an ABI Prism 7500 Sequence (–985/+10)) was constructed. Detector (Applied Biosystems) using gene-specific primer sets and The expression vector of porcine Lhx2 cDNA was constructed in a TaqMan MGB probe for Lhx2 and cyclophilin A. Primer and probe the mammalian expression vector pcDNA3.1/Zeo+ (Invitrogen, San sequences were as follows: mouse and rat Lhx2 forward primer, Diego, CA, USA), resulting in Lhx2/pcDNA3.1. Reporter vectors 5′-TGAGAGCTTCCGTATTTTCAAAGA-3′; mouse and rat Lhx2 re- consisting of an LHX2-binding site were constructed by ligation of verse primer, 5′-CAATTCATCCAAAGTATGAAGATAAAACAG-3′; synthetic oligonucleotides with an AgeI cutting sequence on both mouse and rat Lhx2 TaqMan probe 5′-TGCCACGTGC­ sides of the minimum promoter of Cga fused in the pSEAP2-Basic CTTAG-3′; porcine Lhx2 forward primer, 5′-GGCTAGAGC­ vector [21]. Serial mutations in the upstream region were made by TTCTGTATTTTCAAAGAC-3′; porcine Lhx2 reverse primer, PCR with an oligonucleotide primer set composed of a replace- 5′-AATGAACAAGTCATCCAAAGTATGGA-3′; porcine ment (TAAT with GCCG or ATTT with CGGC, respectively) for Lhx2 TaqMan probe, 5′- TGCCACGTGCCTTAG-3′; mouse FSHβ (–985/+10). and rat cyclophilin A forward primer, 5′-ACGGAGAGA­ AATTTGAGGATGAGA-3′; mouse and rat cyclophilin A reverse prim- Systematic Evolution of Ligands by Exponential Enrichment er, 5′-ACATGCTTGCCATCCAGCCA-3′; mouse and rat cyclophilin A analysis (SELEX) TaqMan probe, 5′-TCCTGGCATCTTGTCCA-3′; porcine cyclophilin SELEX was performed to determine the LHX2-binding se- A forward primer, 5′-GATGAGAATTTTATCCTGAAGCATACG-3′; quence using ΔLIM-LHX2 and a 61-base oligonucleotide porcine cyclophilin A reverse primer, 5′-TGCCATCCAACCACTCAG­ pool containing randomized sequences of 15-mer-nucleotide, TCT-3′; and porcine cyclophilin A TaqMan probe, 5′- 5′-CCTCGAGGTCGACGGTATCCC-(N)15-CCGATCCAC­ TCCTGGCATCTTGTCCA-3′. Reaction mixtures included 900 TAGTTCTAGAGCGGCC-3′ (N=any nucleotide). The details have nM of each primer, a 250 nM TaqMan MGB probe, 1× TaqMan LHX2 MODULATES FSHb PROMOTER ACTIVITY 149 buffer A containing 10 mM Tris–HCl, 50 mM KCl, 0.01 mM EDTA, and the Fd2 sequence integrated into the yeast gene.

5 mM MgCl2, 200 µM dATP, 200 µM dCTP, 200 µM dGTP 400 Two of the 11 selected clones were identical and showed high µM dUTP 0.5 U AmpErase UNG and 1.25 U AmpliTaq Gold DNA identity with the nucleotide sequence of mammalian Lhx2s. Other Polymerase. The real-time PCR consisted of one cycle at 50 C for clones (Prop1 [14] and Prx2 [15]) are described elsewhere. The 2 min and 95 C for 10 min, followed by 40 cycles at 95 C for 15 nucleotide sequence of the cloned Lhx2 revealed the entire sequence sec and 60 C for 1 min. To estimate the relative gene copy numbers of 1764 bp (deposited in the DDBJ under Accession No. AB473486). with the same threshold line for Lhx2 and cyclophilin A, all data An open reading frame of 1221 bp coded 406 amino acids with 5′ were calculated by the comparative CT method (ΔΔCT method) and 3′ untranslated regions of 104 and 439 bp, respectively. Porcine [29] according to the instruction manual. LHX2 shows replacement of amino acids at 1 position for humans and 4 positions for rats and mice. No typical polyadenylation signal Results sequence (AATAAA) was present.

Cloning of Fd2 binding protein from a porcine pituitary cDNA Binding of LHX2 to the Fd2 region library Confirmation for the binding of LHX2 to the Fd2 region was One-hybrid cloning that was performed for 4.1 × 106 transfor- performed with an in vitro binding assay by EMSA. Addition of mants at an efficiency of 1 × 105 colony forming units per 1 µg the His-Tag-fused LHX2 (Fig. 1A, lane 2) generated shift bands DNA yielded several presumptive positive clones. Eleven puri- together with higher molecular size products trapped at the top of fied clones exhibited both specific amino acid requirements and the gel in the presence of a 250-fold excess amount of poly(dI-dC). β-galactosidase activity (data not shown) by being retransformed Since the TrxA-His-Tag peptide fused itself without binding to any into YM4271::[Fd2pHISi-1] [Fd2pLacZi] cells, confirming an DNAs (data not shown), the shift bands were formed by binding interaction between the products selected by one-hybrid cloning with LHX2. The binding was then decomposed by the addition

Fig. 1. Binding and transcriptional activity of LHX2 on the Fd2. (A) Electrophoretic gel mobility shift assay of LHX2 in the Fd2 region. The complex of recombinant LHX2 or ΔLIM-LHX2 (LIM domain-deleted LHX2) protein and FAM-labeled Fd2 was analyzed on a 4% poly- acrylamide gel followed by visualization with a fluorescence viewer. The composition of each binding mixture is indicated under the electrogram. (B) DNase I footprinting analysis of a LHX2-binding sequence in the Fd2 region. Labeling of Fd2 was performed by PCR using a FAM-labeled forward oligonucleotide. DNase I digests were prepared with or without recombinant porcine ΔLIM-LHX2 (lower panel and upper panel, respectively) as described in Materials and Methods. The nucleotide sequence of Fd2 corresponding to each signal is indicated below. A transient transfection assay of porcine Fshβ promoter in CHO (C) and LβT2 cells (D) was performed. Truncated Fshβ promoters fused with SEAP gene in pSEAP2-Basic vector (shown in the left panel) were transfected with either a pcDNA3.1 or Lhx2/ pcDNA3.1 expression vector. An aliquot of cultured medium was used for the SEAP assay. Reporter gene activities are indicated relative to the pcDNA3.1 vector. Open and shaded bars indicate values of pcDNA3.1 and Lhx2/pcDNA3.1, respectively. Data (mean ± SD) are means of quadruplicate transfections from two independent experiments. Asterisks indicate statistical significance by Student’s t test (P<0.01). 150 KATO et al. of an excess amount of unlabeled Fd2 (50-fold molar excess) to many faster migrating bands (lane 3). Since LHX3 formed a large complex by protein-protein interaction with its LIM domains [30], we produced ΔLIM-LHX2. The EMSA with ΔLIM-LHX2 mark- edly reduced the trapped molecules and increased the amount of shift bands (lane 4), which were also decomposed to faster bands by the addition of unlabeled DNA.

DNase I footprinting of the Fd2 region DNase I footprinting was performed to identify LHX2-binding sites in Fd2. Digests without ΔLIM-LHX2 showed signals along the entire region of Fd2 (Fig. 1B upper panel). In contrast, digestion in the presence of ΔLIM-LHX2 produced a marked change in the profile between –837/–798 b (Fig. 1B lower panel). In particular, most of the signals in three regions, –836/–827, –818/–810 and –808/–798 b, containing a characteristic common heptanucleotide sequence, AATTAAT, disappeared. Notably, a signal at position –797 b increased remarkably, indicating an increase in the nuclease susceptibility to DNA-bending caused by LHX2 binding.

Fig. 2. Transient transfection assay of porcine Fshβ promoter in CHO Transcriptional activity of porcine FSHβ promoter caused by and LβT2 cells. Truncated Fshβ promoters fused with either LHX2 via the Fd2 region SEAP gene in pSEAP2-Basic vector (A) were transfected in We investigated the transcriptional activity of Fd2 by cotrans- CHO (B) and LβT2 (C) cells with either pcDNA3.1 or Lhx2/ fection of the Lhx2-expression vector and reporter vector ΔFSHβ pcDNA3.1. An aliquot of cultured medium was used for the SEAP assay. Reporter gene activities are indicated relative to (–985/+10) produced by the fusion of Fd2 to FSHβ (–103/+10) into the pcDNA3.1 vector. Open and shaded bars indicate values of CHO and LβT2 cells. LHX2 significantly (P<0.01) stimulated the pcDNA3.1 and Lhx2/pcDNA3.1, respectively. Data (mean ± SD) transcriptional activity of ΔFSHβ (–985/+10) by 14-fold and 2.59- are means of quadruplicate transfections from two independent experiments. Asterisks indicate statistical significance by Stu- fold in CHO (Fig. 1C) and LβT2 (Fig. 1D) cells, respectively, while dent’s t test (*: P<0.01). FSHβ (–103/+10) and the basic vector were apparently unaffected.

Regulation of porcine Fshβ promoter activity by LHX2 To examine the effect of the regulatory activity of LHX2 on the mine the LHX2-binding sequence (Fig. 3B). In the distal re- porcine FSHβ promoter, a transfection assay was performed on gion over Fd2, 6 binding sites were identified. The signals of CHO cells. While the basal promoter activity over −239 decreased three regions, –2162/−2149 b (5′-TATAATGGTATGTT-3′), against that of the basic vector, LHX2 stimulated the activity of –1915/−1899 b (5′-TTCCATTCATTTGTGTC-3′) and −1880/−1868 each reporter vector by 2.0–2.5 fold (Fig. 2A). b (5′-ATAATTGTAATCA-3′), disappeared upon the binding of In contrast to CHO cells, the basal promoter activity of Fshβ rose LHX2, while increased signals (indicated by lower letters in higher than that of the long upstream in LβT2 cells (Fig. 2B). The Fig. 3B) were observed at –2168/–2165 b (ctga), −2143 b (c) and basal activity of the −985/−597 b region containing Fd2 increased –2136/–2134 b (ata); in a relatively widespread region, –1893/– from 11.0-fold to 29.7-fold for that of FSHβ (–596/+10). Similarly, 1884 b (ttagattcca); and at –1864/–1863 b (gg) around each bind- LHX2 exhibited opposite effects in LβT2 cells but not in CHO ing region. Three other regions were present at –1799/−1768 cells. LHX2 significantly (P<0.01) decreased the promoter activities b (5′-TTGACAATTACTTAAACACACACCAATCAGGG-3′), by 0.26-fold for FSHβ (–238/+10), 0.24-fold for FSHβ (–596/+10), –1441/−1422 b (5′-CATGCCAATTATACATTTTA-3′) and 0.19-fold for FSHβ (–985/+10), 0.18-fold for FSHβ (–1965/+10), and −1156/−1145 b (5′-CTTTGATTATTT-3′), each with some in- 0.18-fold for FSHβ (–2320/+10), respectively. creased signals at −1786 b (a), −1764 b (t), −1418/−1417 b (ag) and −1135 b (t). In the proximal region short of Fd2, LHX2 EMSA and DNase I footprinting showed 6 binding sites neighboring one another: −466/−454 b EMSA was carried out for FAM-labeled upstream fragments (5′-AATATAATTTACA-3′), −440/−433 b (5′-TATAATCA-3′), covering up to −2320 b to examine the LHX2-binding sites in −380/−367 b (5′-GTAACTTATTAACC3′), −298/−287 b the presence of a 250-molar amount of poly(dI-dC). The results (5′-CCCAAATTAAAT-3′), −260/−254 b (5′-CTTAATT-3′) and showed remarkable multiple shift bands (–2320/−1950, −1965/−1484, −218/−206 b (5′-AATTTAATTTgTA-3′), with increased signals −1521/−950 and −238/+10 b) and thick low mobility bands (−985/−620 at −457/−456 b (ta), −452 b (t), −366 b (a), −286 b (a), −252 b (g), and −596/−50 b) except for −706/−541 b, which had a faint shift −248 b (a) and −208 b (g), respectively. band (Fig. 3A). The 15 Lhx2-binding sequences identified are summarized A DNase I footprinting analysis was then performed by in Table 1, which shows that they consisted mostly of A/T-rich comparing the profiles with or without ΔLIM-LHX2 to deter- sequences. Notably, with the exception of region −1915/−1899 b, LHX2 MODULATES FSHb PROMOTER ACTIVITY 151

Fig. 3. LHX2-binding sites in the Fshβ promoter region. (A) An electrophoretic mobility shift assay was per- formed by analysis on 4% PAGE after binding of FAM-labeled porcine Fshβ promoter fragments with (+) or without (−) recombinant ΔLIM-LHX2. (B) DNase I digests were prepared with or without recom- binant porcine ΔLIM-LHX2 (upper panel and lower panel, respectively), followed by analysis using a capillary sequencer with a GeneScan system. Solid bars indicate the LHX2-binding region; nucleotide numbers are shown below. Signals that decreased and increased are indicated by upper- and lowercase letters, respectively.

they contained TAAT/ATTA, known as a core binding motif for Reporter assay of cis-acting element homeodomain transcription factors [31–34]. The cis-acting activity of the 12 binding sequences (Table 1) was examined with constructs fused to a minimal promoter vector Binding specificity of LHX2 and verified by its responsiveness to LHX2 (Fig. 5). To further We applied SELEX to verify the consensus binding sequence of confirm the activity, mutants of the LHX2-binding sequence were LHX2. After the 7th selection generation, the selected oligonucle- produced by nucleotide replacement, and their activities were as- otides were cloned using a plasmid vector and 72 clones, and their sayed. Since the −596/−104 b region showed a notable basal activity nucleotide sequences were ultimately determined. The TAATTA and responsiveness to LHX2 (Fig. 2A and B), we examined the motif was found in 54 of the oligonucleotides determined (data promoter activity of each LHX2-binding site using serially mutated not shown). The alignment of sequences determined by WebLogo vectors of FSHβ (−985/+10) in LβT2 cells (Fig. 6). Mutations at revealed that the LHX2 binding site has an AT-rich composition and –440/–433, –380/–367, –298/–287, –260/–254 and –218/–206 b that its high frequency was observed to be a TAATTA sequence at yielded significant (P<0.05 or P<0.01) decreases with a maximum the 5′ region of pentadecanucleotide sequence (Fig. 4). reduction of about 40%, except for –466/–452 b. 152 KATO et al.

Table 1. Binding sites of LHX2 and other homeodomain transcription in the porcine Fshβ promoter LHX2 LHX3 PRX2 −5057/−5030 −2162/−2149 5′-TATAATGGtATGTT-3′ −1915/−1899 5′-TTCCATTCATTTGTGTC-3′ −1880/−1868 5′-ATAATTGTAATCA-3′ −1799/−1768 5′-TTGACAATTACTTaAACACACACCAATCAGGG-3′ −1441/−1422 5′-CATGCCAATTATACATTTTA-3′ −1442/−1423 −1156/−1145 5′-CTTTGATTATTT-3′ −836/−827 5′-TAATTAATTA-3′ −838/−809 −838/−827 −818/−810 5′-AATTAATTG-3′ −824/−811 −808/−798 5′-TCAATTAATAA-3′ −808/−497 −538/−525 −509/−504 −466/−454 5′-AATATAATTtaCA-3′ −486/−458 −440/−433 5′-TATAATCA-3′ −444/−433 −380/−367 5′-GTAACTTATTAACC-3′ −383/−368 −298/−287 5′-CCCAAATTAAAT-3′ −295/−284 −298/−287 −260/−254 5′-CTTAATT-3′ −259/−253 −262/−253 −218/−206 5′-AATTTAATTTgTA-3′ −219/−209 1) This study examined up to about – 2.3 kb upstream. 2) LHX3 binding sites in –5.7 kb upstream were listed from the results of West et al. [49]. 3) PRX2 binding sites were listed from the results of Susa et al. [20]. 4) Serial mutant vectors were obtained by replacement of underlined nucleotides (A to C and T to G) and used to assay the promoter activity in Fig. 6.

Fig. 4. WebLogo analysis of SELEX data. Aligned data obtained by SELEX at the 7th generation were analyzed using WebLogo. The vertical scale indicates the relative frequency of each nucleotide expressed by the base logo at any position (hori- zontal scale).

Fig. 5. Transient transfection assay of LHX2-binding sites found in the porcine FSHβ promoter. Reporter vectors harboring LHX2- Lhx2 expression during porcine pituitary development binding sites (Table 1) were transfected in CHO with pcDNA3.1 The Lhx2 expression during porcine pituitary development was (open bar) or Lhx2/pcDNA3.1 (shaded bar). Data (mean ± SD) are means of quadruplicate transfections from two independent quantified by real-time PCR (Fig. 7). The results showed that the experiments and are indicated as relative values against those Lhx2 expression was extremely low (≤0.1% of cyclophilin A) at the of the basic vector. Asterisks indicate statistical significance by early fetal stage, f40, in both sexes, followed by a gradual increase Student’s t test (*: P<0.01). up to f110 (0.7–0.8%), and that it increased sharply to 1.6–1.8% at p8, and then slightly decreased at P230 a level similar to that at f110 in both sexes. On the other hand, the member of the Cga-expressing-cell lines (αT1-1, αT3-1, LβT2, LβT4 and TαT1 cells) expressed Lhx2 at more Expression of Lhx2 in pituitary tumor-derived cell lines than 5% of the cyclophilin A level. Lhx2 expression in pituitary tumor-derived cell lines was quan- tified by real-time PCR (Fig. 8). The values for MtT/S and AtT20 cells, in addition to nonpituitary cell line L929, were less than 2%. LHX2 MODULATES FSHb PROMOTER ACTIVITY 153

Fig. 6. Transient transfection assay of serial mutants in the proximal Fig. 7. Real-time PCR analysis of Lhx2 mRNA in porcine anterior promoter region of the FSHβ (–985/+10). Mutant reporter vec- pituitaries. Lhx2 mRNAs of male (A) and female (B) porcine tors prepared by replacement of the region indicated by nucleo- anterior pituitaries were quantified by real-time PCR, and tide number (A to G and T to C, respectively) were transfected the results are indicated as relative amounts of Lhx2 against in CHO with pcDNA3.1 or Lhx2/pcDNA3.1. Data (mean ± SD) those of cyclophilin A. f; fetal day. p; postnatal day. are means of quadruplicate transfections from two independent experiments and are indicated as values normalized by those of the basic vector without/with Lhx2/pcDNA3.1. Asterisks indi- cate statistical significance by Student’s t test (* and **: P<0.01 and P<0.05, respectively).

Discussion

Hitherto, we demonstrated that the Fd2 region functions as a target site for multiple transcription factors [13] and that the porcine Fshβ promoter (–852/+10 b) is sufficient for gonadotrope-specific expression by generation of transgenic rats [16]. Later, as binding proteins for the Fd2, we cloned two pituitary transcription factors, Fig. 8. Real-time PCR analysis of Lhx2 mRNA in cultured i.e., the pituitary specific paired-like homeodomain transcription cells. Real-time PCR was performed, and the results indicated the relative amounts of Lhx2 against those of factor, PROP1 [14], and the paired-related homeodomain factor, cyclophilin A. PRX2 [15]. Using the same method, we cloned and identified a LIM-homeodomain transcription factor, LHX2, as a novel tran- scription factor for the Fshβ. LHX2 binds to 15 sites within −2320 b including Fd2 and shows a cell type-dependent modulation. subunit genes [42]. RT-PCR of mRNA during porcine pituitary LIM-homeodomain transcription factors form a protein family development demonstrated that Lhx2 expression is relatively low and are present in several tissues with a restricted pattern and play at an early stage, f40, and gradually increases until neonatal p8, important roles in differentiation and development [35]. Though the followed by a slight decrease to a similar level by f110 in both sexes. presence of LHX2 and LHX3 in gonadotropes is yet unreported, we The expression pattern indicates that LHX2 is not involved in any recently observed the presence of LHX2 or LHX3 in rat pituitary important roles during early pituitary organogenesis and acquires gonadotropes using their antibodies (unpublished data). Previously, a certain degree of its role in the postnatal pituitary, differing from the critical roles of LHX3 and LHX4 for pituitary organogenesis those of LHX3 and/or LHX4. were demonstrated. Lhx3 and Lhx4 knockout mice showed a dis- A transfection assay demonstrated that LHX2 acts as a repres- ruption in the development of Rathke’s pouch, a primordium of the sor and stimulator in LβT2 and CHO cells, respectively (Fig. 2). pituitary [36–38]. LHX3 also reportedly regulates the porcine Fshβ These opposite roles (depending on the cell type) indicate that the promoter [39]. On the other hand, Lhx2 cDNA was first cloned from presence of any cell type-specific interactor with LHX2 modifies a gonadotrope cell line, αT3-1, and was observed to be involved in the effect of LHX2 action. Indeed, the LIM domain of LHX2 and the regulation of Cga in the mouse [18] and pig [40]. Cherrington other LIM family factors including LHX3 are widely known to et al. revealed that LHX2 participates in activation of the murine interact between several different classes of transcription factors GnRH receptor (GnRH-R) gene by binding to a 36-base boundary and their cofactors [43–46]. The nature of these complexes involves containing a tandem TAAT-motif in response to activin-mediated diverse biological functions. We previously cloned CLIM2 as the signaling [41]. However, in contrast to the essential roles of LHX3 LHX2-binding protein and demonstrated that the interaction of and LHX4, Lhx2 knockout mice showed no apparent effects from the CLIM2 represses the LHX2-dependent stimulation of porcine Cga pituitary development as well as in the regulation of gonadotropin expression [40]. CLIM2 is known to participate in transcriptional 154 KATO et al. activation by forming a large complex with several transcription of FSH at the transcriptional level. In addition, LHX3 is known to factors (TAL1, E47 and GATA-1), and with LIM-only protein participate in the modulation of porcine Cga [54] and Fshβ [39, 55] LMO2 in the erythroid cells [47–49]. Additionally, we have recently with common and unique target sites of LHX2. Since the presence demonstrated that CLIM2 interacts with several types of LIM- of mRNA and the protein for LHX3 in LβT2 cells [39, 55] and for only proteins [50] and also with single-stranded binding proteins LHX2 in αT3-1 cells [18, 52] have been reported, it is likely that 2, SSBP2 [51], in the pituitary. On the other hand, LHX2 has the LHX2 and LHX3 coexist and participate in a different regulation ability to recruit a coactivator, p300, and a TATA-binding protein of the same gene with characteristic binding specificity. Hence, through interaction with the LHX2-binding protein MRG1 [52]. it will be of interest to clarify how the two factors each perform Taken together, Fshβ as well as Cga may be controlled by a large their given roles in the same cells and in vivo. regulatory complex formed by a protein-protein interaction linkage of LHX2 and the LIM domain family. Acknowledgments A core binding motif, TAAT/ATTA, of the homeodomain tran- scription factors is well characterized [31–34]. As expected, 14 We would like to thank Dr PL Mellon of the University of Cali- out of the 15 binding sites for LHX2 identified in the porcine fornia, San Diego, for providing the αT1-1, αT3-1, LβT2, LβT4 and Fshβ promoter contain a TAAT/ATTA motif (Table 1). Although TαT1 cells and Dr K Inoue for the MtT/S cells. This research was the −1915/−1899 b region has no TAAT/ATTA motif, it shows partially supported by the Ministry of Education, Culture, Sports, an AT-rich composition. SELEX for LHX2 drew on an optimal Science and Technology of Japan, Grant-in-Aid for Scientific Re- binding sequence, TAATTA, which is similar to the tandem re- search (B) No. 21380184, and by a Grant-in-Aid for Research (A) peat of AATTAAT observed in the three parts of Fd2 (Fig. 1B). to YK from the Institute of Science and Technology of Meiji Uni- Meanwhile, a palindromic-like sequence, TACTTAGCTAATTA versity. Our study was also supported by the “High-Tech Research (known as PGBE; a pituitary glycoprotein hormone basal element Center” Project for Private Universities and a matching fund sub- at –343/–330 b), previously identified as an LHX2-binding site in sidy, 2008–2010, from the Ministry of Education, Culture, Sports, the upstream region of human Cga [18] was not present in any of Science and Technology of Japan. the 15 binding sites found in the porcine Fshβ upstream. We have confirmed the LHX2-dependent transcriptional activity References of the identified binding sites (Fig. 5). Notably, Westet al. reported that porcine LHX3 is a transcription factor for porcine Fshβ and 1. Jorgensen JS, Quirk CC, Nilson JH. Multiple and overlapping combinatorial codes found five binding sites in the upstream region [39]. Four of the orchestrate hormonal responsiveness and dictate cell-specific expression of the genes binding sites (−1442/−1423, −838/−809, −295/−284, −259/−253 encoding luteinizing hormone. Endocr Rev 2004; 25: 521–542. [Medline] [CrossRef] 2. Strahl BD, Huang H-J, Pedersen NR, Xu JC, Ghosh BR, Miller WL. Two proxi- and −219/−209 b) share common differences in nucleotide length mal activating protein-1-binding sites are sufficient to stimulate transcription of the in comparison with those of LHX2 (Table 1), while 10 binding ovine follicle-stimulating hormone-β gene. Endocrinology 1997; 138: 2621–2631. sites of LHX2 were absent in LHX3. We previously observed that [Medline] [CrossRef] 3. Strahl BD, Huang HJ, Sebastian J, Ghosh BR, Miller WL. Transcriptional activa- LHX2 and LHX3 participate through different cis-elements in the tion of the ovine follicle-stimulating hormone beta-subunit gene by gonadotropin- regulation of porcine Cga expression [40]. The homeodomain is releasing hormone: involvement of two activating protein-1-binding sites and protein known as composing a conserved 60-amino acid sequence with kinase C. Endocrinology 1998; 139: 4455–4465. [Medline] [CrossRef] three α-helixes, and it contains 13 and 5 amino acids important 4. Huang HJ, Sebastian J, Strahl BD, Wu JC, Miller WL. Transcriptional regulation of the ovine follicle-stimulating hormone-beta gene by activin and gonadotropin- for DNA binding and dimerization, respectively. In contrast to releasing hormone (GnRH): involvement of two proximal activator protein-1 sites for the paired-like class of homeodomain factors, LHX2 and LHX3 GnRH stimulation. Endocrinology 2001; 142: 2267–2274. [Medline] [CrossRef] act as monomers without homeodomain-dependent dimerization, 5. Vasilyev VV, Pernasetti F, Rosenberg SB, Barsoum MJ, Austin DA, Webster NJG, Mellon PL. Transcriptional activation of the ovine follicle-stimulating hor- the mechanism of which is discussed in a previous paper [53]. moneß gene by gonadotropin-releasing hormone involves multiple signal transduc- The homeodomains of porcine LHX2 and LHX3 show only 43% tion pathways. Endocrinology 2002; 143: 1651–1659. [Medline] [CrossRef] similarity in their amino acid sequences. Among 14 amino acids 6. Chung H-O, Kato T, Kato Y. Molecular cloning of c-jun and c-fos cDNAs from essential for DNA binding, LHX2 and LHX3 exhibit differences porcine anterior pituitary and their involvement in gonadotropin-releasing hormone stimulation. Mol Cell Endocrinol 1996; 119: 75–82 (Errata 120, 113–114). in 4 residues at position 24 (Gln and Lys), position 31 (Leu and 7. Webster JC, Pedersen NR, Edwards DP, Beck CA, Miller WL. The 5′-flanking Arg), position 42 (Lys and Met) and position 57 (Arg and Lys), region of the ovine follicle-stimulating hormone-β gene containing six progesterone for LHX2 and LHX3, respectively. We previously reported that response elements: three proximal elements are sufficient to increase transcription differences in the amino acids for DNA binding alter the nucleo- in the presence of progesterone. Endocrinology 1995; 136: 1049–1058. [Medline] [CrossRef] tide requirements around the core TAAT/ATTA motif [53]. Our 8. O’Conner JL, Wade MF, Prendergast P, Edwards DP, Boonyaratanakornkit V, preliminary SELEX experiment for LHX3 showed that this factor Mahesh VB. A 361 region of the rat FSH-beta promoter contains multiple T A T A progesterone receptor-binding sequences and confers progesterone responsiveness. has a high binding affinity for an AT-rich sequence, /A /T /AT / Mol Cell Endocrinol 1997; 136: 67–78. [erratum appears in Mol Cell Endocrinol TA (unpublished data), different from that of LHX2 (TAATTA). 1998, 142:215]. [Medline] [CrossRef] In the present study, we demonstrated for the first time by 9. Miller C, Miller W. Transcriptional repression of the ovine follicle-stimulating hor- transfection assay of cultured cells that LHX2 directly regulates mone-beta gene by 17 beta-estradiol. Endocrinology 1996; 137: 3437–3446. [Med- the Fshβ-expression. Taken together with previous studies show- line] [CrossRef] 10. Suszko MI, Lo DJ, Suh H, Camper SA, Woodruff TK. Regulation of the rat folli- ing that LHX2 is a transcription factor for Cga [40], the present cle-stimulating hormone beta-subunit promoter by activin. Mol Endocrinol 2003; 17: study suggested a possibility that LHX2 regulates the synthesis 318–332. [Medline] [CrossRef] LHX2 MODULATES FSHb PROMOTER ACTIVITY 155

11. Zakaria MM, Jeong KH, Lacza C, Kaiser UB. Pituitary 1 activates the main proteins is encoded by two sequentially expressed Drosophila genes from the rat FSHbeta (rFSHbeta) gene through both direct and indirect interactions with the 93D/E cluster. Nucleic Acids Res 1994; 22: 1202–1207. [Medline] [CrossRef] rFSHbeta gene promoter. Mol Endocrinol 2002; 16: 1840–1852. [Medline] [Cross- 34. Pomerantz JL, Sharp PA. Homeodomain determinants of major groove recognition. Ref] Biochemistry 1994; 33: 10851–10858. [Medline] [CrossRef] 12. Lamba P, Fortin J, Tran S, Wang Y, Bernard DJ. A novel role for the forkhead 35. Hobert O, Westphal H. Functions of LIM-homeobox genes. Trends Genet 2000; 16: transcription factor FOXL2 in activin A-regulated follicle-stimulating hormone beta 75–83. [Medline] [CrossRef] subunit transcription. Mol Endocrinol 2009; 23: 1001–1013. [Medline] [CrossRef] 36. Sheng HZ, Zhadanov AB, Mosinger B Jr, Fujii T, Bertuzzi S, Grinberg A, Lee 13. Kato Y, Tomizawa K, Kato T. Multiple binding sites for nuclear proteins of the ante- EJ, Huang S-P, Mahon KA, Westphal H. Specification of pituitary cell lineages by rior pituitary are located in the 5′-upstream region of the porcine follicle-stimulating the LIM homeobox gene Lhx3. Science 1996; 272: 1004–1007. [Medline] [CrossRef] hormone (FSH) β-subunit gene. Mol Cell Endocrinol 1999; 158: 69–78. [Medline] 37. Sheng HZ, Moriyama K, Yamashita T, Li H, Potter SS, Mahon KA, Westphal H. [CrossRef] Multistep control of pituitary organogenesis. Science 1997; 278: 1809–1812. [Med- 14. Aikawa S, Kato T, Susa T, Tomizawa K, Ogawa S, Kato Y. Pituitary transcription line] [CrossRef] factor Prop-1 stimulates porcine follicle-stimulating hormone β subunit gene expres- 38. Sheng HZ, Westphal H. Early steps in pituitary organogenesis. Trends Genet 1999; sion. Biochem Biophys Res Commun 2004; 324: 946–952. [Medline] [CrossRef] 15: 236–240. [Medline] [CrossRef] 15. Susa T, Ishikawa A, Kato T, Nakayama M, Kato Y. Molecular cloning of paired 39. West BE, Parker GE, Savage JJ, Kiratipranon P, Toomey KS, Beach LR, Col- related homeobox 2 (Prx2) as a novel pituitary transcription factor. J Reprod Dev vin SC, Sloop KW, Rhodes SJ. Regulation of the follicle-stimulating hormone β 2009; 55: 502–511. [Medline] [CrossRef] gene by the LHX3 LIM-homeodomain transcription factor. Endocrinology 2004; 145: 16. Cai LY, Kato T, Ito K, Nakayama M, Susa T, Aikawa S, Maeda KI, Tsukamura 4866–4879. [Medline] [CrossRef] H, Ohta A, Izumi SI, Kato Y. Expression of porcine FSHβ subunit promoter-driven 40. Susa T, Sato T, Ono T, Kato T, Kato Y. Cofactor CLIM2 promotes the repressive ac- herpes simplex virus thymidine kinase gene in transgenic rats. J Reprod Dev 2007; tion of LIM homeodomain transcription factor Lhx2 in the expression of porcine pi- 53: 201–209. [Medline] [CrossRef] tuitary glycoprotein hormone alpha subunit gene. Biochem Biophys Acta 2006; 1759: 17. Kioussi C, Carriere C, Rosenfeld MG. A model for the development of the hy- 403–409. pothalamic-pituitary axis: transcribing the hypophysis. Mech Dev 1999; 81: 23–35. 41. Cherrington BD, Farmerie TA, Lents CA, Cantlon JD, Roberson MS, Clay CM. [Medline] [CrossRef] Activin responsiveness of the murine gonadotropin-releasing hormone receptor gene 18. Roberson MS, Schoderbek WE, Tremml G, Maurer RA. Activation of the glyco- is mediated by a composite enhancer containing spatially distinct regulatory ele- protein hormone α-subunit promoter by a LIM-homeodomain transcription factor. ments. Mol Endocrinol 2005; 19: 898–912. [Medline] [CrossRef] Mol Cell Biol 1994; 14: 2985–2993. [Medline] 42. Wandzioch E, Kolterud A, Jacobsson M, Friedman SL, Carlsson L. Lhx2−/− mice 19. Kato Y, Koike Y, Ogawa S, Tomizawa K, Hosaka K, Tanaka S, Kato T. Presence develop liver fibrosis. Proc Natl Acad Sci USA 2004; 101: 16549–16554. [Medline] of activating transcription factor 4 (ATF4) in porcine anterior pituitary. Mol Cell En- [CrossRef] docrinol 1999; 154: 151–159. [Medline] [CrossRef] 43. Bach I. The LIM domain: regulation by association. Mech Dev 2000; 91: 5–17. [Med- 20. Hirai T, Takikawa H, Kato Y. The gene for the β subunit of porcine FSH: absence line] [CrossRef] of consensus oestrogen-responsive element and presence of retroposons. J Mol Endo- 44. Gill GN. The enigma of LIM domains. Structure 1995; 3: 1285–1289. [Medline] crinol 1990; 5: 147–158. [Medline] [CrossRef] [CrossRef] 21. Sato T, Kitahara K, Susa T, Kato T, Kato Y. Pituitary transcription factor Prop-1 45. Jurata LW, Pfaff SL, Gill GN. The nuclear LIM domain interactor NLI mediates stimulates porcine pituitary glycoprotein hormone α gene expression. J Mol Endocri- homo- and heterodimerization of LIM domain transcription factors. J Biol Chem nol 2006; 37: 341–352. [Medline] [CrossRef] 1998; 273: 3152–3157. [Medline] [CrossRef] 22. Nakayama M, Kato T, Susa T, Sano A, Kitahara K, Kato Y. Dimeric PROP1 bind- 46. Matthews JM, Visvader JE. LIM-domain-binding protein 1: a multifunctional co- ing to diverse palindromic TAAT sequences promotes its transcriptional activity. Mol factor that interacts with diverse proteins. EMBO Rep 2003; 4: 1132–1137. [Medline] Cell Endocrinol 2009; 307: 36–42. [Medline] [CrossRef] [CrossRef] 23. Crooks GE, Hon G, Chandonia JM, Brenner SE. WebLogo: a sequence logo gen- 47. Wadman IA, Osada H, Grutz GG, Agulnick AD, Westphal H, Forster A, Rab- erator. Genome Res 2004; 14: 1188–1190. [Medline] [CrossRef] bitts TH. The LIM-only protein Lmo2 is a bridging molecule assembling an ery- 24. Alarid ET, Windle JJ, Whyte DB, Mellon PL. Immortalization of pituitary cells at throid, DNA-binding complex which includes the TAL1, E47, GATA-1 and Ldb1/NLI discrete stages of development by directed oncogenesis in transgenic mice. Develop- proteins. EMBO J 1997; 16: 3145–3157. [Medline] [CrossRef] ment 1996; 122: 3319–3329. [Medline] 48. Xu Z, Huang S, Chang LS, Agulnick AD, Brandt SJ. Identification of a TAL1 tar- 25. Pernasetti F, Vasilyev VV, Rosenberg SB, Bailey JS, Huang HJ, Miller WL, Mel- get gene reveals a positive role for the LIM domain-binding protein Ldb1 in erythroid lon PL. Cell-specific transcriptional regulation of follicle-stimulating hormone-beta gene expression and differentiation. Mol Cell Biol 2003; 23: 7585–7599. [Medline] by activin and gonadotropin-releasing hormone in the LbT2 pituitary gonadotrope [CrossRef] cell model. Endocrinology 2001; 142: 62284–62295. [CrossRef] 49. Lahlil R, Lecuyer E, Herblot S, Hoang T. SCL assembles a multifactorial complex 26. Susa T, Kato T, Kato Y. Reproducible transfection in the presence of carrier DNA that determines glycophorin A expression. Mol Cell Biol 2004; 24: 1439–1452. [Med- using FuGENE6 and Lipofectamine2000. Mol Biol Rep 2008; 35: 313–319. [Medline] line] [CrossRef] [CrossRef] 50. Susa T, Ishikawa A, Cai LY, Kato T, Matsumoto K, Kitahara K, Kurokawa R, 27. Ogawa S, Aikawa S, Kato T, Tomizawa K, Tsukamura H, Maeda K-I, Petric N, Ono T, Kato Y. Highly related LIM factors, LMO1, LMO3 and LMO4, play different Elsaesser F, Kato Y. Prominent expression of spinocerebellar ataxia type-1 (SCA1) roles in the regulation of pituitary glycoprotein hormone common α subunit gene. gene encoding ataxin-1 in LH-producing cells, LßT2. J Reprod Dev 2004; 50: 557– Biosci Rep 2010; 30: 51–58. [CrossRef] 563. [CrossRef] 51. Kato Y, Kato T, Ono T, Susa T, Kitahara K, Matsumoto K. Intracellular localiza- 28. Aikawa S, Sato T, Ono T, Kato T, Kato Y. High level expression of prop-1 gene in tion of porcine single-strand binding protein 2. J Cell Biochem 2009; 106: 912–919. gonadotropic cell lines. J Reprod Dev 2006; 52: 195–201. [Medline] [CrossRef] [Medline] [CrossRef] 29. Lyons SR, Griffen AL, Leys EJ. Quantitative real-time PCR for Porphyromonas 52. Glenn DJ, Maurer RA. MRG1 binds to the LIM domain of Lhx2 and may function gingivalis and total bacteria. J Clin Microbiol 2000; 38: 2362–2365. [Medline] as a coactivator to stimulate glycoprotein hormone alpha-subunit gene expression. J 30. Bridwell JA, Price JR, Parker GE, McCutchan Schiller A, Sloop KW, Rhodes SJ. Biol Chem 1999; 274: 36159–36167. [Medline] [CrossRef] Role of the LIM domains in DNA recognition by the Lhx3 neuroendocrine transcrip- 53. Kato Y, Kimoto F, Susa T, Nakayama M, Ishikawa A, Kato T. Pituitary homeodo- tion factor. Gene 2001; 277: 239–250. [Medline] [CrossRef] main transcription factors HESX1 and PROP1 form a heterodimer on the inverted 31. Catron KM, Iler N, Abate C. Nucleotides flanking a conserved TAAT core dictate TAAT motif. Mol Cell Endocrinol 2009; 315: 168–173. [Medline] [CrossRef] the DNA binding specificity of three murine homeodomain proteins. Mol Cell Biol 54. Susa T, Ishikawa A, Kato T, Nakayama M, Kitahara K, Kato Y. Regulation of 1993; 13: 2354–2365. [Medline] porcine pituitary glycoprotein hormone alpha subunit gene with LIM-homeobox tran- 32. Damante G, Fabbro D, Pellizzari L, Civitareale D, Guazzi S, Polycarpou- scription factor Lhx3. J Reprod Dev 2009; 55: 425–432. [Medline] [CrossRef] Schwartz M, Cauci S, Quadrifoglio F, Formisano S, Di Lauro R. Sequence-spe- 55. Granger A, Bleux C, Kottler ML, Rhodes SJ, Counis R, Laverriere JN. The LIM- cific DNA recognition by the thyroid transcription factor-1 homeodomain. Nucleic homeodomain proteins Isl-1 and Lhx3 act with steroidogenic factor 1 to enhance Acids Res 1994; 22: 3075–3083. [Medline] [CrossRef] gonadotrope-specific activity of the gonadotropin-releasing hormone receptor gene 33. Jagla K, Stanceva I, Dretzen G, Bellard F, Bellard M. A distinct class of homeodo- promoter. Mol Endocrinol 2006; 20: 2093–2108. [Medline] [CrossRef]