LMO1, LMO3 and LMO4 were cloned from the adult porcine pituitary cDNA library. Amino acid sequences of porcine LMO1, LMO3 and LMO4 were highly conserved among mammalian species. Transfection assay of the pituitary-derived cell line LβT2 was carried out using the pituitary αGSU (glycoprotein hormone α-subunit) promoter (−1059/+12 b) fused to pSEAP2-Basic vector as a reporter gene. The results demonstrated that, whereas LMO4 showed no apparent effect, αGSU promoter activity was markedly repressed by LMO1 but activated by LMO3, indicating the different roles of the three highly homologous proteins, LMO1, LMO3 and LMO4. Knockdown assay by LMO siRNAs (small interfering RNAs) confirmed the above results for LMO1 and LMO3, whereas that by LMO4 siRNA increased the expression, indicating different modes of action. RT–PCR (reverse transcription–PCR) for total RNAs of several cell lines showed that LMO1 and LMO4 mRNAs were present ubiquitously in all cell lines, except for LMO1 in L929 cells. In contrast, LMO3 mRNA was abundant only in LβT4 and GH3 cells with only small amounts in LβT2 and MtT/S cells, indicating the cell-type-specific function of this protein. Real-time analyses of porcine pituitary ontogeny revealed that the three LMO genes are expressed during the fetal period and decline immediately afterwards, followed by a remarkably low level of LMO3 and LMO4 after birth. RT–PCR of the porcine tissues examined showed ubiquitous expression of LMO4, whereas LMO1 and LMO3 are expressed tissue specifically. Thus the present study demonstrated that three highly related LIM cofactors, LMO1, LMO3 and LMO4, have different effects on αGSU gene expression in the pituitary glands.
Pituitary αGSU (glycoprotein hormone α-subunit) is a common subunit for the heterodimeric hormones, LH (luteinizing hormone), FSH (follicle-stimulating hormone) and TSH (thyroid-stimulating hormone), all three of which contain a specific partner of β-subunit. Since the three glycoproteins are synthesized in gonadotropes (FSH and LH) and thyrotropes (TSH) respectively, the αGSU gene is controlled in a cell-type-specific manner in the two different cell types. Elucidation of the unique transcriptional control of the αGSU gene remains an interesting issue, although many investigators have reported a number of transcription factors that control its expression .
LIM homeodomain factor is well known as a crucial transcription factor for the αGSU gene. Lhx2 was first cloned from clonal pituitary tumour-derived αT3-1 cells as a specific protein binding to the regulatory element in the proximal region designated as the PGBE (pituitary glycoprotein hormone basal element), which is located 323 bp upstream of the murine αGSU promoter [2,3]. Thereafter, another related LIM homeodomain transcription factor, Lhx3, was identified as a regulator for the αGSU gene by binding to PGBE [4,5]. Concomitantly, Lhx3 was extensively investigated and found to play a crucial role in the early development of the pituitary primordium [6,7]. Meanwhile, we studied a novel potent activating region in the distal upstream region of the porcine αGSU gene , and reported that Lhx2 and Lhx3 similarly induce αGSU promoter activity in CHO (Chinese-hamster ovary) cells but modulate it in a reciprocal manner in LβT2 cells . These results prompted us to further investigate the molecular basis of the control of αGSU gene expression. Recently, we have succeeded in cloning a porcine orthologue of the CLIM2 (cofactor of LIM homeodomain protein 2), also known as NL1  and Lbd1  as a cofactor of Lhx2, and we have demonstrated that CLIM2 acts as a repressor of Lhx2 and Lhx3 actions on αGSU gene expression . CLIM2 is a pivotal factor in the formation of a protein–protein interaction with various factors such as GATA2, LMOs (LIM domains only), single-stranded DNA-binding protein and others in addition to the Lhx family . Therefore, an attempt to investigate interactants of CLIM2 in the pituitary glands may provide clues to a better understanding of the molecular mechanism of αGSU gene expression.
In the present study, we employed the yeast two-hybrid system to identify interacting protein(s) of CLIM2 in the anterior pituitary lobe. Ultimately, the cDNAs encoding porcine orthologues of the LMO protein were cloned. Transient transfection and knockdown assays demonstrated that three LMOs showed different effects on the αGSU gene expression with a reciprocal function. RT–PCR (reverse transcription–PCR) using RNAs from several cell lines and porcine tissues revealed cell-type-specific (LMO3) and tissue-specific (LMO1 and LMO3) expressions, whereas LMO4 was expressed ubiquitously. Moreover, in porcine pituitary glands, gene expression of LMOs markedly declined immediately before birth, especially in that of LMO3 and LMO4.
Cloning of CLIM2-interacting proteins by the yeast two-hybrid system
A yeast expression vector of porcine CLIM2  was constructed in the pBD-GAL4 vector (Stratagene, La Jolla, CA, U.S.A.) by ligating to the EcoRI/SalI site in frame (CLIM2/pBD-GAL4). The transformation of yeast YRG2 cells harbouring CLIM2/pBD-GAL4 with porcine anterior pituitary cDNAs/pAD-GAL4 (10 μg) was carried out using the YEASTMAKER Yeast Transfection System (Clontech Laboratories, Mountain View, CA, U.S.A.), and cells were grown on selective plates, followed by confirmation with the β-galactosidase assay and sequence analysis as described previously .
Construction of reporter and expression vectors
Construction of the reporter vector, α(−1059/+12)pSEAP2, containing upstream regions (−1059/+12) of porcine αGSU promoter in pSEAP2-Basic (Clontech), has already been described . Expression vectors were constructed by ligating with the coding region of cDNAs obtained by the yeast two-hybrid system in frame to the EcoRI/XhoI sites of the mammalian expression vector pcDNA3.1/Zeo+ (Invitrogen, San Diego, CA, U.S.A.).
Transfection assay and knockdown assay with siRNA (small interfering RNA)
LβT2 cells, which express the αGSU gene endogenously, were kindly provided by Dr P.L. Mellon (Department of Reproductive Medicine, University of California San Diego, La Jolla, U.S.A.) [14,15]. They were maintained in monolayer cultures, and trypsin-dispersed cells were plated on to a 96well plate at 1×104 cells/100 μl per well. At 24 h after seeding, 0.2 μl of Lipofectamine™ 2000 (Invitrogen)/30 ng of reporter vector was transfected with 20 ng of expression vectors in quadruplicate (n=4) according to the manufacturer's instructions. After incubation for more than 48 h, an aliquot of the medium (5 μl) was assayed for SEAP (secreted alkaline phosphatase) activity . The reliability and reproducibility of the results are described in our previous paper . For knockdown assay, Stealth™ Select RNAi (RNA interference) of mouse Lmo1 (Oligo ID: MSS246945) and mouse Lmo4 (Oligo ID: MSS237074) were obtained from Invitrogen. Stealth™ RNAi of mouse Lmo3 was synthesized by Invitrogen (5′-UGUCCAGUGCCUUUAGGAGAUACCG-3′ and 5′-CGGUAUCUCCUAAAGGCACUGGACA-3′). siRNAs (6 ng per well) were transfected with reporter vector (30 ng per well) and expression vector (20 ng per well) using Lipofectamine™ 2000 (0.2 μl per well) reagent as described above. At 48 h after transfection, the SEAP assay and preparation of total RNAs from transfected cells were performed. The synthesis of cDNA from total RNA was performed using PrimeScript reverse transcriptase (Takara, Osaka, Japan). Oligonucleotide primer sets of Lmo1 (5′-GGGCCCGAGACAATGTGTAT-3′ and 5′-AGACGGACAGATGGACCTGG-3′), Lmo3 (5′-CTGTAGACATTGAGCAGGCATACAA-3′ and 5′-TTTCCCGTTACACCAAACAGC-3′) and Lmo4 (5′-AAGAGCGGCATGATCCTTTG-3′ and 5′-GGGCTGTGGGTCTATCATGTTC-3′) were used for RT–PCR.
RT–PCR and real-time PCR of porcine pituitary and tissue RNAs and cultured cells
Porcine anterior pituitary glands of German Landrace male and female pigs were collected on fetal days (f40, f65, f82, f95 and f110) and postnatal days [p8, p60 and p230 (sexually mature)] . Total RNAs were extracted by using ISOGEN (Nippon Gene Co., Toyama, Japan) and were pooled from one to six individuals of the appropriate age and sex. RNAs from several porcine tissues were also prepared. Total RNAs were also extracted from 60–70% confluent cultured cells of pituitary- and non-pituitary-derived cell lines, the characteristics of which have been described in our previous paper . cDNA synthesis from total RNAs was carried out using ReverTra Ace reverse transcriptase (Toyobo, Tokyo, Japan) as described previously .
Oligonucleotide primer sets of LMO1 (5′-ACACCAAGGCCAACCTCATC-3′ and 5′-CACAAGATCATGTTGTTCTTCAGGA-3′), LMO3 for mouse and rat (5′-TACTGGCATGAGGACTGCCT-3′ and 5′-CTTCCTCGTAGTCTGTCTGGC-3′), LMO3 for pig (5′-GGACAATGTTTACCACCTCG-3′ and 5′-AGTCCGTCTGGCAAAGGATC-3′), LMO4 (5′-GCATGATCCTTTGCAGAAATGA-3′ and 5′-TGGCAGTAGTGGATTGCTCTGA-3′) and cyclophilin A (5′-TGGTGACTTCACACGCCATAATG-3′ and 5′ATTCCTGGACCRAAAACGCTCC-3′) were used for RT–PCR. Each cDNA was amplified in a reaction mixture (10 μl) containing the two required primers (10 pmol each) and 0.25 unit of AmpliTaqGold DNA polymerase (PE Applied Biosystems, Foster City, CA, U.S.A.) with 32 cycles of denaturation (94°C, 30 s), annealing (55°C, 30 s) and extension reaction (72°C, 1 min) steps. The resulting PCR products were analysed on 2% agarose-gel electrophoresis. Quantitative real-time PCR was performed using SYBR Green Real-time PCR Master Mix (Toyobo) and the 7500 Real-time PCR System (Applied Biosystems). The primer set and PCR conditions are described above.
Cloning of pituitary LMOs
Using the yeast two-hybrid system, we have cloned cDNAs encoding three types of LMO proteins, LMO1, LMO3 and LMO4 (respective accession numbers are AB304399 for LMO1, AB304400 for LMO3 and AB304401 for LMO4), as CLIM2-binding proteins from porcine pituitary cDNAs. Each porcine LMO comprises 156, 145 and 165 amino acids for LMO1, LMO3 and LMO4 respectively (Figure 1). The amino acid sequences show identities of more than 96% for LMO1 and 100% for LMO3 and LMO4 in comparison with the corresponding human and murine orthologues. Notably, between LMO1 and LMO3, the first and second LIM domains show 100 and 96% identities, whereas the homology between other LMOs or Lhx2 is extremely low (40–47%). Overall, a molecule of LMO3 shows 90% identity with LMO1 including low identity, with nine differences out of 11 residues in the C-terminus, and LMO1 has an additional 11-amino-acid extension in the N-terminus (Figure 2).
Nucleotide and deduced amino acid sequences of porcine LMO1, LMO3 and LMO4
Comparison of amino acid sequences among cloned porcine LMOs
Transfection assay of LMOs
Transfection assay using the reporter vector α(−1059/+12)/pSEAP2 in LβT2 cells was carried out in the absence or presence of the expression vector LMOs/pcDNA3.1. Overexpression of LMOs did not affect the empty reporter vector pSEAP2-Basic. Co-expression with reporter vector α(−1059/+12)/pSEAP2 and an empty expression vector pcDNA3.1/Zeo+ showed a high expression level of approx. 60-fold against the empty reporter vector pSEAP2-Basic in LβT2 cells (Figure 3A). The high promoter activity, however, was remarkably inhibited to approx. 0.24-fold by overexpression of LMO1. In contrast, LMO3 showed an approx. 1.27-fold significant activation of αGSU promoter. Finally, LMO4 showed a slight but not significant decrease, confirming that each LMO plays a different role in αGSU gene expression.
Transfection assay of LMOs in LβT2 cells
To confirm the above results, a knockdown assay with siRNAs of LMOs was carried out, and decreased mRNA levels were confirmed for LMO1 and LMO4, but not LMO3, probably due to its low content as revealed below. The knockdown assay demonstrated that siRNAs of LMO1 and LMO3 gave consistent results with that of the reporter assay, whereas that of LMO4 yielded an increase in reporter expression different from the result of overexpression, indicating that a decrease in saturated LMO4 level by siRNAs released its suppressive role that was not observed by overexpression.
LMO gene expression during porcine fetal and postnatal pituitary development
Real-time PCR was performed for the three LMOs using total RNAs prepared from porcine fetal and postnatal pituitaries. During the fetal period, expression of LMO1 increased at the middle stage. In contrast, expressions of LMO3 and LMO4 decreased at the middle stage or later (Figure 4). Expression of LMO1 decreased to moderate levels after birth and those of LMO3 and LMO4 fell markedly to almost undetectable levels.
RT–PCR for porcine anterior pituitary total RNAs
LMO gene expression in porcine tissues
RT–PCR was performed for total RNAs prepared from porcine tissues. Expression of LMO4 was observed in all tissues but with small differences in quantity, whereas LMO1 and LMO3 showed tissue-specific expression, not in heart, liver (LMO3 only), spleen, pancreas (LMO3 only), kidney and testis (Figure 5).
RT–PCR for total RNAs from porcine tissues
LMO gene expression in pituitary-derived cell lines
The cell-type-specific expression of LMOs was examined for pituitary tumour-derived cell lines (Figure 6). Ubiquitous expression in all cell lines examined including the non-pituitary cell line L929 was observed for LMO4. LMO1 was expressed in all pituitary cell lines examined, appearing at a somewhat abundant level in LβT4 cells, but not expressed at all in L929 cells. LMO3, unlike the other two LMOs, showed a distinct expression profile in which this gene was expressed in LβT4 and GH3 cells at high levels and in LβT2 and MtT/S cells at low levels, whereas there was no apparent expression in the other cell lines examined.
RT–PCR for total RNAs from cultured cells
We recently cloned CLIM2 as an Lhx2-binding protein . This protein is known to play various roles in the regulation of gene expression, since it contains not only the LIM-interacting domain but also other protein–protein interaction domains. The present study aimed to expand our search so as to better understand the network of transcriptional processes in the pituitary glands. By employing the yeast two-hybrid system using CLIM2 as a bait protein, three LMO cDNAs were cloned as CLIM2-binding proteins. Further study demonstrated that three types of porcine LMO cDNAs (LMO1, LMO3 and LMO4) are expressed in the porcine anterior pituitary glands. We identified LMO3 as a CLIM2-binding protein for the first time. In the present study, we showed the disparate effects of the three LMOs on the regulation of αGSU gene expression. Thus LMO1 and LMO3 clearly demonstrated opposite roles in αGSU gene regulation, whereas LMO4 expression was affected when knockdown by siRNA was employed (Figure 3). The expressions of these three LMO genes showed different profiles in the fetal and postnatal pituitaries (Figure 4) and in tissues (Figure 5). LMO3 gene expression was mostly confined to LβT4 and GH3 cells, with a low level in LβT2 and MtT/S cells (Figure 6), suggesting that LMO3 may play a crucial role co-operatively with LMO1 in those same particular pituitary cell types.
LMOs, one of the families of LIM domain-containing proteins, comprise two tandem LIM domains, which exhibit highly conserved cysteine-rich zinc-finger-like motifs [21,22]. LMO proteins are transcriptional cofactors that form regulatory protein–protein complexes with LIM-interacting domains or with many different protein-interacting domains through two internal LIM domains [23,24]. They consist of four members (LMO1, LMO2, LMO3 and LMO4) that play important roles in cell fate specification, differentiation, haematopoiesis and cytoskeletal organization [22,25]. LMO1 and LMO2 are likely to act as T-cell oncogenes by virtue of the fact that their translocations are linked to T-cell leukaemia [26,27]. LMO4, which is the most recently identified member , acts as an oncogene for breast cancer . On the other hand, LMO3 was cloned based on its sequence homology with LMO1 , and has been reported to act as an oncogene by interacting with the neural transcription factor HEN2 (helix–loop–helix protein 2) in neuroblastoma .
The above reports have indicated an interesting relationship between LMO1 and LMO3. Foroni et al.  demonstrated that both LMO1 and LMO3 show a high expression pattern in the restricted areas of the developing mouse brain but little in the lymphoid tissue. In addition, the combined null mutation in LMO1 and LMO3 genes that has caused perinatal deaths without anatomical defects indicates that both genes play crucial roles in fetal development, although the finding that a null mutation of either gene alone showed no apparent abnormality is puzzling . The overlapping expressions of LMO1 and LMO3 genes show that together they exert exquisite control of fetal development. Our observation of the different roles of the three LMOs in the αGSU promoter activity in LβT2 cells might provide a perspective about LMOs. Given the high similarity of the LMO domains of LIMs, the N- or C-terminal region, both of which are composed of different sequences, may explain the different actions of the three LMOs.
The two LIM domains of LMO1 and LMO3 show a high similarity (100 and 96%) in contrast with those between the other LMOs (49–66% and 38–55% respectively). LIM domains are notable for interacting with several proteins. The only interactors for LMO3 to date are HEN2  and CLIM2 (the present study), but LMO1 can interact with TAL1 (T-cell acute lymphocytic leukaemia 1) and LYL1 (lymphoblastomic leukaemia)  and CRP3/MLP (cysteine- and glycine-rich protein 3/muscle LIM protein) , in addition to CLIM2. The tertiary structures of LIM domains in LMO2 and LMO4 show an especially high similarity, despite the low homology of the amino acid sequence (49% for LIM1 and 38% for LIM2), and the first LIM domain similarly interacts with the LIM-interaction domain of CLIM2 . The first LIM domain of LMO4 can interact with BRCA1 (breast-cancer susceptibility gene 1), CLIM2 and CtIP (COUP-TF interacting protein) , and the mouse homologue of DEAF-1 (Drosophila deformed epidermal autoregulatory factor 1) . Furthermore, the interaction between CtIP and the first LIM is likely to be independent of the CLIM2 binding, suggesting that the first LIM, at least, contains many surfaces to facilitate the protein–protein interaction. Taken together, given the high sequence identity in the LIM domains of LMO1 and LMO3, they may well share interacting proteins and could readily switch on the expression of target genes in terms of their corresponding levels of expression. Another issue of considerable interest is how LMO1 and LMO4 respond to extracellular stimuli from the hypothalamus and/or gonadal organs.
We observed by RT–PCR that LMO3 is abundant in particular pituitary cell lines in LβT4 and GH3, and at low levels in LβT2 and MtT/S, whereas LMO1 and LMO4 are present in all pituitary cell lines examined (Figure 6). LβT2 and LβT4 are gonadotrope lineage cell lines established by immortalization using LHβ promoter-fused T-antigen [14,15]. They show properties of the gonadotrope cell line with expressions of αGSU, LHβ and GnRH (gonadotropin-releasing hormone) receptor. However, it remains unclear why the expression of the LMO3 gene is so markedly different between LβT2 and LβT4 cells. Since the amount of LMO3 is very different from those of LβT2 and LβT4, which are cells of the same gonadotrope lineage, the different characteristics of both cells may well be reflected. On the other hand, the expression of the LMO3 gene is abundant in GH3, but less so in MtT/S. GH3 is a somato-mammotrope cell line expressing both growth hormone and prolactin genes , whereas MtT/S is a somatotropic cell line expressing only the growth hormone gene . Taken together with our observation that LMO3 is expressed in a tissue-specific manner (Figure 5), it is of great interest to determine the role of LMO3 in those two cell lines.
We have demonstrated the roles of LMO1, LMO3 and LMO4 in the pituitary glands for the first time. The new finding in the present study is that three LMOs appear to play different roles in the regulation of αGSU gene expression, although their precise roles remain unclear. Exactly why LMO3 expresses in particular pituitary cell types in contrast with the ubiquitous expression of other LMOs remains an important issue for future study.
cofactor of LIM homeodomain protein 2
COUP-TF interacting protein
glycoprotein hormone α-subunit
helix–loop–helix protein 2
LIM domain only
pituitary glycoprotein hormone basal element
secreted alkaline phosphatase
small interfering RNA
This work was partially supported by the Ministry of Education, Culture, Sports, Science and Technology of Japan, a Grant-in-Aid for Scientific Research (B) [no. 18380177] and by a Grant-in-Aid for Research (A) to Y.K. from the Institute of Science and Technology, Meiji University. This work was also supported by the ‘High-Tech Research Center’ Project for Private Universities: matching fund subsidy, 2006–2008, from the Ministry of Education, Culture, Sports, Science and Technology of Japan. We declare that there is no conflict of interest that would prejudice the impartiality of this scientific work.