| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Physiologie de l’Axe Gonadotrope (A.G., C.B., R.C., J.-N.L.), Centre National de la Recherche Scientifique, Unité Mixte de Recherche 7079, Physiologie et Physiopathologie, Université Pierre et Marie Curie-Paris6, 75252 Paris, France; Centre Hospitalier Universitaire de Caen, Hôpital Georges Clemenceau, Département Génétique et Reproduction (M.-L.K.), 14033 Caen, France; and Department of Cellular and Integrative Physiology (S.J.R.), Indiana University School of Medicine, Indianapolis, IN 46202
Address all correspondence and requests for reprints to: Jean-Noël Laverrière, Centre National de la Recherche Scientifique, Unité Mixte de Recherche 7079, Physiologie et Physiopathologie, Université Pierre et Marie Curie-Paris6, 4 place Jussieu, 75252 Paris cedex 05, France. E-mail: jean-noel.laverriere{at}snv.jussieu.fr.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
-subunit and the specific ß-subunit of LH and FSH. In addition to this action at the gene level, GnRH stimulates the release of LH and FSH, which in turn, through the systemic circulation promote gametogenesis and modulate endocrine functions such as sex steroid secretion. The GnRH-R thus plays a crucial role in mediating brain control on the reproductive function at the pituitary level (1). The expression level of GnRH-R at the cell surface is a determinant for the amplitude and the specificity of gonadotrope responsiveness. Indeed, the GnRH-R number is subject to fine-tuned control during a number of physiological processes including pituitary development, estrous cycle, pregnancy, and lactation. These modulations take place, at least in part, at the transcriptional level. In an attempt to characterize the mechanisms that underlie the tissue-specific and regulated expression of the GnRH-R gene in both adults and developing animals, several investigators have isolated and characterized the 5' regulatory sequences of the human, ovine, mouse, and rat genes (2, 3, 4, 5, 6, 7). From these studies, a few transcription factors involved in the tissue-specific and regulated expression of this gene have been identified; some of them, notably steroidogenic factor 1 (SF-1), are common to all species studied (see review in Ref. 8).
Using transgenic mice that have integrated the human placental alkaline phosphatase (hPLAP) reporter gene under the control of the 3.3-kb rat GnRH-R promoter, we have demonstrated that this promoter could direct the specific expression of the reporter gene in gonadotrope cells in the anterior pituitary (9). More importantly, we have shown that the transgenic GnRH-R promoter was already activated at embryonic d 13.5 (E13.5) in the developing pituitary, thus long before the LH and FSH ß-subunit terminal markers (9). These latter data indicated that GnRH-R was an early marker of the gonadotrope cell lineage and raised the question of the signaling molecules and/or transcription factors involved at the onset of GnRH-R gene expression in the developing pituitary. Several homeodomain transcription factors are possible candidates because they play important roles in the initial events in pituitary organogenesis. These include paired class Hesx-1 (Rpx) protein; Ptx1 and Ptx2 bicoid-related homeodomain proteins; LIM-homeodomain (LIM-HD) proteins Lhx3, Lhx4, LH-2, and Isl-1; and proteins of the Six family Six-1, Six-3, and Six-6 (see review in Refs.10 and 11).
By transient transfection in T3-1 and LßT2 gonadotrope cell lines (12, 13), we have also shown that the gonadotrope-specific expression of the rat promoter depends on two principal domains located within 1 kb upstream of the ATG codon (7). The most proximal domain contains cis-acting elements that include an activator protein 1 (AP-1) binding site (–352/–346), a SF-1 element (–245/–237), and an imperfect cAMP response element (CRE) (–110/–103). Both the SF-1 element and the CRE contribute in mediating the pituitary adenylate cyclase-activating polypeptide-regulated expression of the GnRH-R gene (14). Proximal elements within the –275/–225 region, notably SF-1, are functionally linked to a distal specific enhancer called GnRH-R specific enhancer (GnSE), itself composed of two active and independent regions located in its 5' and 3' domain. The 5' domain (–983/–962) appears to bind factor(s) related to the GATA family, whereas the 3' domain, including a motif extending from –871/–862, is suspected to bind an as-yet-unidentified LIM-HD protein(s). Both the distal enhancer and the proximal promoter domain are required for full gonadotrope-specific activity of the rat promoter (15).
In the present study, we have focused on the 3' domain of the GnSE that accounts for 40–60% of enhancer activity (15). Our main objectives were to precisely delineate the active domain of the 3' GnSE, to identify the transcription factors involved and to determine the type of interactions between factors that may account for the functional relationship apparently linking the distal and proximal domains of the promoter. To achieve this, we took advantage of the T3-1 and LßT2 gonadotrope-derived cell lines. In our hands, these cell lines appear identical regarding the expression of the rat GnRH-R gene (15). Consequently, the main processes related to GnRH-R gene regulation should be similar in both cell lines. In addition, the transgenic mouse model described above was used to monitor in vivo promoter activity. The data collected using these different models strongly suggest that the pair of LIM-HD proteins composed of Lhx3 and Isl-1 contributed together with SF-1 to the gonadotrope-specific activity of the rat, as well as the human GnRH-R gene promoter in vitro and in vivo.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
The Proximal Domain Contains Four cis-Regulatory Elements, Two of Which Mediate GnSE Activity
A CRE-like element at position –110/–103 and a motif 5' adjacent to the SF-1 element that bound an unidentified factor, hereafter referred to as SF-1 adjacent protein (SAP) (previously referred to as AB factor in Ref. 14), were found to be required for pituitary adenylate cyclase-activating polypeptide and cAMP-regulated activity of the rat GnRH-R gene promoter (14). To determine whether these elements might also mediate constitutive promoter activity together with AP-1 (–352/–346) and SF-1 (–245/–237), we designed new constructs containing the luciferase reporter gene under the control of the 1.1-kb GnRH-R gene promoter with single or combined site-specific mutations. The constructs were then analyzed by transient transfection into gonadotrope-derived T3-1 and LßT2 cells (Fig. 1A
). As expected, single mutations affecting either the AP-1 or SF-1 elements decreased promoter activity by 45 and 70–80%, respectively, in agreement with our previous data (15). Interestingly, mutations affecting either the CRE-like or the SAP elements also decreased constitutive promoter activity (30–45 and 75%, respectively). The greatest attenuation in promoter activity was observed with mutations affecting the SAP element and was probably due to simultaneous abrogation of the activity of the distal enhancer. We have indeed previously shown that the distal enhancer activity required the presence of proximal elements located in a 50-bp region encompassing the SAP and SF-1 elements (14). By contrast, single mutations affecting either the AP-1 or the CRE-like elements were less efficient. The respective importance of the CRE-like and SAP elements was further investigated in the context of multiple mutations. The results showed that simultaneous mutations of three or four elements within the proximal domain abrogated promoter activity (93–98%). These data, together with those previously published, corroborated the existence of two classes of elements in the proximal promoter. The first group synergized with the distal enhancer such as the SF-1 and SAP elements, and the second appeared to act independently such as the AP-1 and the CRE-like elements.
|
). First, truncation of the 5' and 3' region of the 50-bp module containing the SF-1 and SAP had no effect on promoter efficiency either in the absence or in the presence of the GnSE, demonstrating that the resulting 24-bp DNA fragment was sufficient to synergize with the GnSE. Second, the presence of either the SF-1 or SAP element with the GnSE was insufficient to generate full promoter activity. Nevertheless, these constructs displayed similar activity, which was significantly higher than in absence of GnSE, indicating that the GnSE-interceding function of SAP and SF-1 was equivalent, independent, and thus most likely supported by two distinct factors.
The Distal Enhancer Displays Two Functional Homeodomain Response Elements
Next, we focused on the 3' region of the distal enhancer (3'GnSE). To precisely delineate the response element present within this domain, T3-1 and LßT2 cells were transiently transfected with artificial promoter constructs containing various deletions of the 3'GnSE fused to two 50-bp SAP/SF-1 modules and the minimal PRL promoter (Fig. 2A). As previously mentioned (15), duplicating the 50-bp module caused enhanced promoter activity and thus increased the sensitivity of the transfection assay. The promoter consisting of a 20-bp element extending from –873 to –852 had the same activity (80–90%) as the complete 3'GnSE (–896/–825). A more reduced DNA sequence extending from –871 to –862 failed to synergize with the SAP/SF-1 modules.
|
). This effect appears to be specific to gonadotrope cells, because all of these constructs were inactive when transfected in the heterologous Chinese hamster ovary (CHO) cell line (data not shown). In the absence of the SAP/SF-1 module, the 20-bp element was inactive, even under the octameric form, in agreement with the dependence of GnSE on SF-1/SAP for activity. Together, these data demonstrate that the –873/–852 element retained both the functionality and the specificity of the entire 3'GnSE.
Inspection of the –873/–852 sequence revealed two conserved TAAT motifs (underlined in Fig. 3A) that resembled the core motif of homeodomain response elements (16). To evaluate the functional importance of these motifs, we designed different double-stranded oligonucleotides with 2-bp mutations referred to as M1 to M6 (Fig. 3A). Wild-type and mutant oligonucleotides were inserted into artificial constructs similar to those depicted in Fig. 2A. Transient transfection studies revealed that mutations within (M2 and M5) or adjacent to (M3) the TAAT motifs markedly decreased promoter activity, by 84, 63, and 74%, respectively, in T3-1 cells or by 77, 55, and 50%, respectively, in LßT2 cells.
|
T3-1 cells that were displaced by an excess of the homologous unlabeled probe but not by excesses of heterologous GATA or Pitx probes (15). Similarly, the wild-type –873/–852 element was forming shifted complexes that were displaced by an excess of unlabeled homologous probe (Fig. 3B). Three specific shifted complexes were observed in T3-1 nuclear extracts (Fig. 3B) and two in LßT2 nuclear extracts (Fig. 3C). The two upper migrating complexes identified in T3-1 cells were similar to complexes I and II found in LßT2 cells. Regarding these complexes, the M4 probe appeared as efficient as the wild-type probe in complex formation, indicating that the GC dinucleotide located in the center of the sequence was not involved in the DNA/protein interaction. In contrast, disruption of one of the two TAAT motifs (mutations M2 and M5) as well as mutation affecting the neighboring bases (M3) resulted in a strong decrease or complete abrogation of the upper migrating complex I. Because the binding site contained two AT-rich motifs, the complex I formation might involve at least two homeobox proteins that could form homo- or heterodimers. Conversely, the complex II might correspond to the binding of only one homeobox protein. According to this assumption, the M1 and M6 mutations, although moderately affecting promoter activity, modified in an opposite manner the pattern of the shifted complexes, by increasing and decreasing complex II, respectively. This suggested that the nucleotides surrounding the TAAT motifs such as those affected by M1 and M6 mutations were important for influencing dimer vs. monomer binding.
Isl-1 and Lhx3 Interact with the Homeodomain Response Elements of the Distal Enhancer
Given that LIM-HD proteins are known to bind TAAT-containing motifs, we analyzed the binding specificity of the 3'GnSE. Competition experiments were done using two unlabeled oligonucleotides corresponding to the pituitary glycoprotein basal element (PGBE) of the mouse glycoprotein hormone -subunit promoter and the A3/A4 element of the rat insulin promoter (Fig. 4A). The former was previously shown to bind Lhx2 and Lhx3 (17, 18) and the latter to bind Isl-1 and Lmx-1 in vitro (19, 20). Homologous competition with increasing amounts of unlabeled probe resulted in the progressive abrogation of the specific complexes formed with LßT2 nuclear extracts (Fig. 4B, left panel). Likewise, heterologous competitions with either unlabeled A3/A4 or PGBE probes were strongly effective. In contrast, mutated M2 and M3 oligonucleotides were inefficient for competing with the wild-type probe (data not shown). Similar results were obtained with T3-1 nuclear extracts (data not shown). These data implied that the –873/–852 element was able to bind LIM-HD proteins and prompted an evaluation of the direct implication of Isl-1 and Lhx3. The addition of increasing amounts of an Isl-1 antibody resulted in a progressive abrogation of the shifted complexes obtained with either LßT2 (Fig. 4B, right panel) or T3-1 nuclear extracts (data not shown). Because the monoclonal 39.4D5 antibody is directed against the C terminus of the rat Isl-1 protein, in close vicinity with the DNA binding homeodomain, the abrogation of complexes suggested that the antibody prevented Isl-1 binding to DNA. We called this 20-bp sequence LIRE for LIM-related element. In contrast, an antibody directed against Lhx3 was ineffective in abrogating the formation of the complexes, suggesting that either Lhx3 was not involved in complex I formation or the antibody was inappropriate for EMSA (data not shown).
|
). Lhx3 and Isl-1 were immunodetected in crude LßT2 nuclear extracts (input), and, more specifically, they were also detected in samples eluted from the magnetic beads. The unrelated transcription factor Egr-1 that is primarily expressed in the gonadotrope and somatotrope cell lineages (21) was, as expected, also present in the crude LßT2 nuclear extracts but, in contrast to Isl-1 and Lhx3, was not detected in the eluted sample. Compared with Egr-1, the enrichment in Lhx3 and Isl-1 in the eluted sample was approximately 100-fold, clearly indicative of the binding specificity of LIRE. We also weakly detected SF-1 in the eluted sample (Fig. 4C, third panel) suggesting that this protein may be recruited by the heterodimer Lhx3/Isl-1 in absence of its DNA response element or may bind to the LIRE with low affinity.
Lhx3 and Isl-I, But Not SF-1, Physically Interact in Vivo
To determine whether SF-1 might interact physically with Lhx3 or Isl-1, we performed in vivo coimmunoprecipitation. The monoclonal anti-Isl-1 antibody was incubated with nuclear extracts of T3-1 cells and used as the precipitating antibody together with protein A-agarose beads. The bound materials was then eluted from the beads and analyzed by Western blot using the polyclonal rabbit anti-SF-1 antibody. No positive immunoreactive signal was detected in three independent experiments, suggesting that SF-1 does not interact directly with Isl-1 (data not shown). Similarly, to determine whether Lhx3 might interact physically with SF-1, T3-1 cells were first transiently transfected with Myc epitope-tagged Lhx3a, and nuclear extracts were prepared. Indeed, because the anti-Lhx3 and anti-SF-1 antibodies were both raised in rabbit, they cannot be used together. The anti-Lhx3 IgG heavy chain migrates at the same level as SF-1 and leads to dramatic background when revealing SF-1 with the anti-rabbit secondary antibody coupled to peroxidase. The nuclear extracts from transiently transfected cells were thus used in immunoprecipitation with a monoclonal anti-Myc antibody. Again, no positive signal was detected in Western blot using the anti-SF-1 antibody, although control experiments showed that Myc epitope-tagged Lhx3a was strongly precipitated by the anti-Myc antibody (data not shown). Together, these data suggested that neither Lhx3 nor Isl-1 interacted physically with SF-1 in vivo. However, the possibility that the antibodies used were not appropriate for such experiments cannot be formally excluded.
By contrast, a similar approach using the polyclonal anti-Lhx3 antibody as the precipitating antibody and the monoclonal anti-Isl-1 to reveal the immunoblot demonstrated a potent interaction between Isl-1 and Lhx3 using nuclear extracts from gonadotrope cells (Fig. 4D).
Coexpression of Isl-1, Lhx3a, and SF-1 Results in Promoter Activation in CHO Cells
To assess the involvement of these LIM-HD proteins in GnRH-R gene promoter activity, expression vectors encoding Lhx3, Isl-1, and SF-1, together with the artificial luciferase construct containing eight copies of the LIRE element were cotransfected (Fig. 2B) into nongonadotrope CHO cells that do not express this set of transcription factors (Fig. 5A). Consistent with our previous results (15), SF-1 stimulated the activity of the artificial promoter construct. In contrast, Lhx3a as well as Isl-1 had no significant influence, alone or in association singly with SF-1. However, transactivation was potentiated when both LIM-HD proteins were combined together. Furthermore, under these conditions, the addition of SF-1 resulted in a marked increase in promoter activity. In our model, Lhx3b that differs from Lhx3a in NH2-terminal sequence that prevents DNA binding and gene activation (22) failed to synergize with Isl-1 and the promoter displayed the same activity as with SF-1 alone. We also investigated the role of other closely related LIM-HD proteins. For instance, Lhx9 that is expressed in gonads and highly homologous to Lhx2, and Lhx5 that is expressed in the hippocampus, failed to synergize with Isl-1. In contrast, transactivation was restored with Cs Isl-2 (from Chinook salmon) (Fig. 5B), suggesting that the interaction between Lhx3 and Isl-1 may be evolutionarily conserved among species. This effect was entirely abrogated with Cs Isl-2 DLIM, a truncated protein lacking LIM domains, consistent with a crucial function of these domains in mediating protein-protein interactions.
|
). In agreement with data obtained with the artificial construct, SF-1 and the heterodimer Lhx3/Isl-1 together stimulated the activity of the 1100-bp GnRH-R promoter.
|
and data not shown). However, maximal activity required a complete combination that involved SF-1 and the LIM-HD protein pair as observed for the rat promoter. Combinations involving only one or two of these transcription factors led to intermediate levels of stimulation. Deletions of up to 122 bp abrogated the stimulating activity of SF-1 and LIM-HD proteins, indicating that the LIM response element of the human promoter was localized in the –306/–122 region.
Synergistic Stimulation of the GnRH-R Promoter by SF-1 and the LIM-HD Proteins Is Dependent on the Cell Context
In CHO cells, both LIM-HD proteins Lhx3a and Isl-1 were required for promoter activation (Fig. 5A); however, only additive effects were observed with SF-1 (Figs. 5A and 6A). From data obtained in gonadotrope cells with mutant promoters and artificial constructs (Figs. 1 and 2B) (15), we expected the synergistic rather than additive effects of SF-1 together with the LIM-HD proteins on the GnRH-R gene promoter. Therefore, we performed experiments to assess whether such synergistic or additive effects were dependent on the cell context. The wild-type luciferase promoter construct was thus transfected in T3-1 cells together with expression vectors for SF-1, Isl-1, Lhx3a, individually or in various combinations (Fig. 6C). Although the stimulation induced by Isl-1 alone was not significant, Lhx3a as well as SF-1 were capable alone to stimulate the activity of the cotransfected wild-type promoter construct, however to a limited extent. Cotransfection of any two of the three expression vectors was more efficient and led to activity levels that were closer to those expected from additive stimulations. In contrast, simultaneous coexpression of the three factors resulted in synergistic increase in promoter activity, indicating that the three partners must be present to cooperate and generate optimal efficacy in the activation of the rat GnRH-R gene promoter. These data were therefore different from those obtained in CHO cells, suggesting that the cell context, and likely additional transcription factors, played a significant role.
Additional LIM Response Elements Are Present within the Rat GnRH-R Promoter
In the FSHß promoter, six elements were able to bind Lhx3 and the three most proximal were tested positive in transactivation (23). To determine whether an analogous situation occurred in the rat GnRH-R promoter, the 20-bp sequence corresponding to LIRE was deleted from the wild-type promoter construct and the dose-dependent effect of cotransfected Lhx3a and Isl-1 expression vectors were evaluated in T3-1 cells (Fig. 7). Transfection of vectors expressing Lhx3a and Isl-1 led to a dose-dependent stimulation of the cotransfected wild-type construct pLuc1.1GnRH-R (Fig. 7B), and, as expected, deletion of the LIRE element within the 1100-bp promoter (pLuc1.1DeltaLIRE) resulted in a significant decrease in promoter activity. Nevertheless, the overexpressed LIM proteins remained able to stimulate the LIRE-deleted construct in a dose-dependent manner (Fig. 7B). These data were suggesting that supplementary LIM response elements were present within the GnRH-R gene promoter. We then evaluated three other constructs in their ability to respond to Lhx3a and Isl-1: the construct containing the mutated SAP and SF-1 site above described (see Fig. 1) referred to as pLuc1.1SAPmSFm, the same construct further deleted from the characterized LIRE (pDeltaLIRE SAPmSFm), and a construct truncated at –475 (pLuc0.44GnRHR) and therefore deleted from the distal part of the promoter that included the LIRE (Fig. 7A). The pLuc1.1SAPmSFm and pDeltaLIRE SAPmSFm constructs were again stimulated by the cotransfected LIM-HD expression vectors in a dose-dependent manner. In contrast, the promoter activity of the truncated –475 construct was not significantly increased under the same conditions (Fig. 7B). Together, these data suggested that additional LIM response elements were present within the promoter, and, in contrast with the characterized LIRE, they might act independently of the SAP and SF-1 elements. They were likely located within the distal part of the promoter, between –1135 and –475. Whether such additional elements were actually active in absence of overexpressed LIM-HD proteins or were only operative under these cell stress conditions remains to be determined.
|
). The same experiments were then conducted with the LIRE-deleted promoter construct and the LIRE-deleted, SAP/SF-1 mutated construct. Obviously, the dominant-negative derivative induced a decrease in the activity of the LIRE-deleted construct, corroborating the existence of additional LIM response elements within the rat GnRH-R promoter (Fig. 7C). Most importantly, the inhibitory effect induced by KRAB-Lhx3 was not cumulative with that of the LIRE deletion, indicating that the LIRE element in the wild-type promoter bound KRAB-Lhx3 and was actually implicated in its negative action. From these experiments, the relative importance of the characterized LIRE element was evaluated to account for 60% of the LIM-dependent promoter activity. In addition, under these conditions, the mutant promoter deleted from the LIRE element and containing the mutated SAP and SF-1 elements was not significantly inhibited, suggesting that, under physiological concentrations of LIM-HD proteins, the activity of the additional LIM response elements was also dependent on SAP and SF-1. This contrasted with the SAP/SF-1-independent activity of these elements observed in the Lhx3a and Isl-1 overexpression experiments (Fig. 7B). Furthermore, these data suggested a novel property for the LIM-HD proteins that might switch over from a SAP/SF-1-dependent to a SAP/SF-1-independent activity. The switch itself could be controlled by the concentration levels of the LIM-HD proteins.
SF-1 Binding and LIM-HD Protein Activity
We then turned on the potential mechanisms that underlie the apparent dependency of LIM-HD factor upon SF-1 binding. To this end, we designed several wild-type and mutated-SF-1 proteins fused to the Gal4 DNA binding site (Fig. 8A) and tested their activity on an artificial promoter that contained eight copies of LIRE response elements placed upstream of five copies of Gal4 response elements in the context of the minimal PRL promoter/luciferase construct. After transient transfection in T3-1 cells, this promoter appeared to be highly responsive to Gal4-VP16 fusion protein (data not shown) and significantly stimulated by cotransfected Lhx3- and Isl-1-expressing vectors (Fig. 8C). Surprisingly, overexpression of Gal4-SF-1 fusion protein was unable to activate the promoter and further led to inhibition of the stimulation induced by overexpression of the LIM-HD proteins. Similar situation was observed with several Gal4-mutated SF-1 fusion proteins, especially with the N-terminal and C-terminal deleted mutant Gal-SF N78-C334. Significant promoter activity was restored with overexpression of the Gal-SF N171-C334 and Gal-SF N278-C334 fusion proteins. In this promoter context, a limited region within the ligand binding domain of SF-1 located between amino acids 278 and 334 appeared thus sufficient to elicit high transactivation capacity. In addition, the LIM-HD proteins remained able to further stimulate promoter activity.
|
The LIM-HD Proteins Isl-1 and Lhx3 Colocalize with in Vivo GnRH-R Promoter Activity in the 3.3GnRH-R-hPLAP Transgenic Mice
At this point in the study, it appeared crucial to determine whether the combination of the LIM-HD proteins Isl-1 and Lhx3 might be compatible with GnRH-R promoter activity in vivo. A classical chromatin immunoprecipitation assay could not be performed, because the endogenous mouse promoter in T3-1 and LßT2 cells, unlike the rat promoter, does not contain a functional GnSE, all the essential elements being localized within 600 bp upstream of the ATG codon. This was duly reestablished by Jeong et al. (24). Furthermore, McGillivray et al. (25) have shown the presence of a LIM response element at –298 in the proximal region of the mouse promoter, notably using a chromatin immunoprecipitation assay. To answer this question, we then took advantage of our transgenic mouse model that contains the hPLAP coding sequence under the control of the 3.3-kb promoter of the rat GnRH-R gene (9). In this transgenic model, hPLAP expression can be considered as a cell-specific marker of the gonadotrope lineage as well as an index of in vivo promoter activity in the adult animal and during pituitary ontogenesis from E13.5 to postnatal d 0 (9). Transgenic mouse embryos were thus removed at E13.5 and E15.5 and analyzed by immunofluorescence with antibodies directed against either Lhx3 or Isl-1 and by histochemistry for evaluation of hPLAP activity (Fig. 9). At E13.5, the transgene was expressed in the cytoplasm of a few cells localized in the ventral region of the developing pituitary, in agreement with our previously published data (9). At this time, confocal analysis showed that a large number of cell nuclei were immunoreactive for Lhx3 with a maximal expression in the dorsal region of the pituitary, namely in the pars intermedia. In contrast, Isl-1 immunoreactivity was restricted to a small number of cell nuclei localized in the ventral region of the pituitary (Fig. 9A). These expression patterns of Isl-1 and Lhx3 were consistent with those reported by others (26). At E15.5 (Fig. 9B), Lhx3 expression appeared to be less intense than at E13.5, because a significant number of cell nuclei were negative [
40% compared with 4',6-diamidino-2-phenylindole (DAPI)-labeled nuclei in the section shown]. Simultaneous analysis with the anti-Isl-1 antibody showed that Isl-1 expression was restricted to a limited number of cells coexpressing Lhx3 (Fig. 9B). To further analyze the profiles of transcription factors with respect to transgene expression, histochemical detection of hPLAP activity was subsequently followed by immunofluorescence using antibodies directed against either Isl-1 or Lhx3 on the same section (Fig. 9C). One apparent observation was the cytoplasmic expression of hPLAP that colocalized with nuclei displaying Lhx3 immunofluorescence. A similar approach also allowed a demonstration of the colocalization of hPLAP activity and Isl-1 expression. Together, these data suggested that both Isl-1 and Lhx3 might be colocalized with transgene expression. To verify this hypothesis, the three markers were simultaneously analyzed on the same section. Only a small number of cells that expressed hPLAP without assessable levels of Isl-1 and/or Lhx3 were observed. Indeed, analysis of multiple sections of several E15.5 embryos, and consistent with the double-label staining experiments, two major categories of cells coexisted with respect to transgene expression and LIM-HD proteins. The first expressed both Lhx3 and Isl-1, whereas the second expressed Lhx3 and Isl-1 as well as hPLAP (Fig. 9D). Because SF-1 colocalized with hPLAP (data not shown and Ref. 9), the former might correspond to SF-1-negative cells, whereas the latter was likely related to SF-1-expressing cells. In conclusion, the in vivo promoter activity was positively correlated, in a majority of transgene-expressing cells, with the presence of both Lhx3 and Isl-1, reinforcing the hypothesis of a gonadotrope-specific LIM code operating together with SF-1 to activate the expression of the GnRH-R gene during the development of the pituitary.
|
). |
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
Consistent with its crucial role in pituitary ontogenesis, Lhx3 is involved in the constitutive expression of several specific genes, including gonadotrope-expressed genes such as the glycoprotein hormone -subunit and the FSHß genes (17, 23, 36). In contrast, the potential target genes for Isl-1 in the anterior pituitary remain to be defined. Here, we show that the GnRH-R gene could be the first described possible candidates, not only in rat species but most probably also in humans. Reminiscent of the LIM code deciphered in the developing spinal cord (29), we also demonstrate that Isl-1 acts together with Lhx3 to transactivate the GnRH-R gene promoter. This statement is supported by both in vitro and in vivo experimental data. First, we show that coexpression of Lhx3 and Isl-1 is necessary for activation of the GnRH-R gene promoter in CHO cells and, second, that these LIM-HD proteins interact physically in the gonadotrope-derived cell lines as shown by coimmunoprecipitation. Third, we demonstrated that the overlapping expressions of Isl-1 and Lhx3 mostly correlate with in vivo promoter activity in our transgenic mouse model. The role of Isl-1 and Lhx3 in GnRH-R promoter activity is further substantiated by the observed decrease in promoter activity induced by LIRE deletion or by overexpression of the dominant-negative derivative of Lhx3. In addition, these experiments would allow us to detect additional LIM-HD response elements accounting for 40% of the LIM-HD-dependent activity of the GnRH-R gene promoter. This is reminiscent of data reported by West et al. (23) who have identified six LIM-HD response elements within the porcine FSHß promoter, among which the three most proximal being involved in the gonadotrope specific activity of the promoter in LßT2 cells. This redundancy of the LIM-HD response elements in both the FSHß and the GnRH-R promoters is consistent with a key role of the LIM-HD protein in gonadotrope-specific gene expression.
In addition to their ability to form a binary code, LIM-HD proteins may interact with transcription factors of other families. For instance, in the case of the developing spinal cord, synchronization of neuronal subtype specification during neurogenesis is achieved by direct interactions between bHLH and LIM-HD proteins mediated through the adaptor protein NLI (37). In the anterior pituitary, synergistic transcriptional activation of the promoters of the thyrotropin ß-subunit, Pit-1, and PRL genes results from direct interaction between the LIM domain of Lhx3 and the POU domain of the pituitary transcription factor Pit-1 (36). Likewise, we show here that in gonadotrope-derived cells the LIM-HD proteins are unable to transactivate the GnRH-R promoter in the absence of the SF-1 response element, strongly suggesting that SF-1 is an obligatory partner in this specific transactivation process. However, data from coimmunoprecipitation experiments does not favor a physical interaction between SF-1 and Lhx3 or Isl-1. Our unpublished in vitro experiments with GST-Lhx3 fusion protein and in vitro-labeled SF-1 protein contrastingly show significant interactions. In this context, the experiments with the Gal4-SF-1 fusion proteins bring important information. They indeed suggest that conformational change in SF-1, consequential to binding to its appropriate response element, is likely required for coactivator recruitment and transcriptional activation. In the absence of the SF-1 response element, SF-1 may bind to low-affinity binding sites, including the LIRE as suggested by the results of in vitro experiments using multimerized biotinylated LIRE (Fig. 4C). We may then speculate that, under these conditions, conformational change is prevented and corepressors such as DP103 or N-CoR via DAX-1 are recruited. These are expressed in gonadotrope cells and are known to interact with SF-1, leading to transcriptional repression (38, 39). By this mean, they partially prevent transactivation by the LIM-HD proteins as observed with artificial promoter deleted of the SAP/SF-1 module (15) (Fig. 2B) as well as in the case of the LIM-Gal4-responsive promoter (interacting with the full-length Gal4-SF-1 fusion protein). With this hypothesis, the LIM-HD proteins may thus exert their own effect on promoter activation without necessitating interaction with SF-1, but the response element for SF-1 must be present in the promoter.
The overlapping expression of Isl-1 and Lhx3 colocalizes with an in vivo promoter activity at E15.5. At the same time, SF-1 is also detected in hPLAP-expressing cells (Ref. 9 and data not shown), indicating that at this stage of pituitary ontogenesis, at least three transcription factors involved in GnRH-R promoter activity in vitro are also present in vivo, strongly arguing for a functional role of these factors in the cell-specific expression of the GnRH-R gene. In fact, when our paper was in its final phase, a role for Lhx3 in the mouse GnRH-R promoter activity was demonstrated (25), indicating that both the mouse and rat promoters may be controlled by LIM-HD proteins. Furthermore, the role of these factors does not seem to be restricted to rodents but might be extended to other species such as human as indicated by our in vitro experiment. Consistent with this hypothesis, three mutations in the human Lhx3 gene have underlined its crucial role in anterior pituitary functions. The affected individuals display a syndrome of combined pituitary hormone deficiency and a rigid cervical spine (40, 41). These mutations consist of a missense mutation (Y116C) in the second LIM domain, an intragenic deletion predicting a truncated protein lacking the DNA-binding homeodomain, and a single base pair deletion in exon 2, resulting in the product of short, inactive LHX3 proteins. One nonsense mutation has been identified (Q310X) in the human Isl-1 gene (42) in a heterozygous Japanese patient with type II diabetes that leads to a decreased activity of Isl-1. However, no default in reproduction has been reported so far.
The LIM-HD proteins may be factors critical at the onset of GnRH-R gene expression. This is suggested by the role of these proteins in other developmental models (see above) but also by data obtained with our transgenic mice. Indeed, at E13.5, SF-1 was not detected in the developing pituitary, despite its obvious expression in the ventral diencephalon (9). It cannot be excluded, however, that this is due to insufficient sensitivity of the immunofluorescence procedure but using the same approach, Isl-1 and, a fortiori, Lhx3 are actually detected in the pituitary. This LIM-HD protein pair may thus be involved at the onset of GnRH-R gene expression, acting at this stage of development independently of SF-1 as observed in the nongonadotrope CHO cells. The SF-1 dependence of the GnSE in the GnRH-R promoter could be acquired at latter stages of development, between E13.5 and E15.5. Further analyses combining both in vitro and in vivo promoter studies would be necessary to highlight these particular points and delineate the specific combination of transcription factors involved in the adult pituitary as well as at different stages of development including the postnatal physiology.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
Oligonucleotides, Enzymes, and Antibodies
Oligonucleotides and enzymes were obtained from Eurobio (Courtaboeuf, France) and Fermentas (St. Leon-Rot, Germany), respectively, except when otherwise stated. The monoclonal antibodies anti-Isl-1 (39.4D5) and anti-myc (9E10) were obtained from the Developmental Studies Hybridoma Tissue Bank, which was developed under the auspices of the National Institute of Child Health and Human Development and maintained by the University of Iowa (Department of Biological Sciences, Iowa City, IA). Rabbit anti-LIM-3 polyclonal antibody (AB3202) was purchased from Chemicon International (Temecula, CA), rabbit anti-Egr-1 polyclonal antibody (sc-110) from Santa Cruz Biotechnology (Santa Cruz, CA), and rabbit anti-SF-1 polyclonal antibody from Upstate Biotechnology (Charlottesville, VA).
Plasmid Constructs and Expression Vectors
The nucleotide sequence of the rat (GenBank accession nos. Z99955 and Z99956) and human GnRH-R gene is numbered relative to the translational initiation codon. The 1228-bp fragment of the human promoter used in this study was isolated in our laboratory (our unpublished data) and displays a nucleotide sequence completely homologous to that registered in the human genome resource data bank of the National Center for Biotechnology Information.
To generate mutant promoters, simple and combined targeted mutations were inserted according to the three-step PCR protocol previously described using the adequate mutant and wild-type primers (15). A similar approach was applied to generate the deleted LIRE constructs. The 3'GnSE domain (–896/–825) was generated by PCR. To obtain artificial promoter constructs, synthetic sense and antisense oligonucleotides corresponding to sequences –871/–862, –873/–852, and mutants M1 to M6 were annealed. The resulting double-stranded fragments with 5'-HindIII- and 3'-KpnI-recessed ends were inserted into the previously described artificial luciferase construct that contained the minimal PRL promoter under the control of two 50-bp modules (–275/–225) including the SAP and the SF-1 response elements (15).
Directional multimerization of the LIRE sequence was performed as described by Ouwerkerk and Memelink (43). Briefly, the synthetic monomer LIRE fragment with XbaI and KpnI sites at its 5' and 3' end and a NheI site close to the KpnI site was first inserted into the XbaI and KpnI sites of the pGEM-3Zf(–) vector (Promega, Charbonnières, France). After amplification in bacteria, the monomer LIRE sequence was excised with XbaI and KpnI, and then reintroduced into the NheI and KpnI sites of the same construct resulting in a head-to-tail dimer. Repeating the procedure with the dimer-containing plasmid led to a tetramer. LIRE multimers, from dimer to octamer, were then excised with HindIII/KpnI and reinserted into the luciferase construct containing two 50-bp modules and the minimal PRL promoter.
The hLhx3a and pcDNA3-Myc-His hLhx3b were previously described (22). The pCMV5-mSF-1, pCS2-rat Isl-1, pcDNA3 Cs Isl-2, and the pcDNA3 Cs Isl-2 
LIM were kindly provided by Drs. K. L. Parker (University of Texas Southwestern Medical Center, Dallas, TX), S. L. Pfaff (The Salk Institute for Biological Studies, La Jolla, CA), and G. Salbert (Université de Rennes I, Rennes, France), respectively. Dr. S. Réteaux (Institut de Neurobiologie Alfred Fessard, Gif-sur-Yvette, France) and Dr. H. Westphal (National Institute of Child Health and Human Development, Bethesda, MD) kindly provided the vectors pGEM-mLhx9 and pBlueScript-mLhx5, respectively. The complete coding sequence of each vector was amplified by PCR with Expand High Fidelity PCR System (Roche Molecular Biochemicals, Mannheim, Germany) using specific primers, and the resulting amplified products were inserted into the BamHI/XbaI sites of the pcDNA3-Myc-His. The Gal4-SF-1 fusion proteins were generated in the pcDNA3 vector as follows: the HindIII-BamHI fragment corresponding to the Gal4 DNA binding domain (1–147) was excised from the pSG424 vector and inserted into the pcDNA3 vector. The full-length SF-1 coding sequence as well as the N terminus deletion mutants (N78, N171, and N278) were amplified by PCR using appropriate primers containing a BamHI (sense primer) and a XbaI (antisense primer) recognition site, and inserted into the pcDNA3 vector with the Gal4 coding sequence in frame. The C terminus C334 SF-1 deletion mutants were generated using the BlpI restriction site located at this position in the SF-1 coding sequence. The pcDNA3-Gal4-SF-1 construct was thus subjected to BlpI and XbaI digestion, followed by filling in with Klenow fragment of DNA polymerase I, purification, and autoligation of the pcDNA3 Gal4-SF-1 deleted vector. All constructs were verified by sequencing.
Cell Culture and Transfections
Mouse gonadotrope T3-1 and LßT2 cells, generously provided by Dr. P. Mellon (La Jolla, CA), and CHO cells were maintained in monolayer cultures as previously described (15). For transient transfection, cells were plated in 48-well plates at a density of 4 x 104 cells/well and transfected with LipofectAMINE/PLUS reagents (Life Technologies, Gaithersburg, MD) as previously described for luciferase reporter constructs (15). All quantities and volumes were scaled down 2-fold except LipofectAMINE. Where appropriate, cells also received 10 ng expression vector/well or an equivalent amount of control pCMVß expression vector (BD Biosciences Clontech, Palo Alto, CA). Cells were harvested 18 h after transfection and firefly and renilla luciferase activities (Promega) were measured as described (15).
Preparation of Nuclear Extracts and EMSA
Nuclear extracts were prepared from cells and used in EMSA as described (15). Monoclonal antibody 39.4D5, polyclonal anti-Lhx3, or the control antibody anti-myc (9E10) were incubated with nuclear extracts for 1 h at 4 C before addition of the labeled probes.
Western Blotting
Nuclear extracts (25–40 µg) were analyzed by Western blot as previously described (44). Primary antibodies directed against either Lhx3, Isl-1, SF-1, or Egr-1 were diluted at 1:3000, 1:1500, 1:2000, and 1:400, respectively, and secondary antibodies rabbit antimouse IgG coupled to horseradish peroxidase (DAKO, Glostrup, Denmark) or goat antirabbit IgG peroxidase were diluted at 1:4000 and 1:7000, respectively. Immune complexes were visualized using the enhanced chemiluminescence system (Amersham Biosciences, Saclay, France).
DNA Affinity Purification
The pGEM-3Zf(–) vector containing the LIRE tetramer was digested by XhoI and labeled by filling-in the recessed 3'-termini with biotinylated dUTP using the Klenow fragment from Escherichia coli DNA polymerase I. The labeled LIRE tetramer was then removed from the plasmid by XbaI digestion, separated by electrophoresis on 3% (wt/vol) NuSieve agarose gel (Cambrex Bio Science Rockland, Wokingham, UK), and purified by phenol/chloroform extraction and ethanol precipitation. Biotinylated tetramers were incubated with Dynabeads M-280 streptavidin (Dynal Biotech, Compiègne, France) in 10 mM Tris-HCl (pH 8), 1 mM EDTA, and 100 mM NaCl at room temperature for 2 h. At least 5 pmol LIRE tetramers were bound per milligram of beads. For the in vitro DNA-protein interaction, 1 mg activated beads was washed three times in binding buffer [20 mM Tris-HCl (pH 8), 10% (vol/vol) glycerol, 1 mM EDTA, 1 mM dithiothreitol (DTT), 100 mM NaCl, 0.01% (vol/vol) Triton X-100] at 4 C, and then incubated with 1 mg LßT2 cells nuclear extracts at room temperature for 30 min with rotation. Poly(dI-dC) was added to the binding reaction as a nonspecific competitor. Beads were then washed twice with binding buffer, and bound proteins were eluted in 30 µl of 20 mM Tris-HCl (pH 8), 10% (vol/vol) glycerol, 1 mM EDTA, 1 mM DTT, 500 mM NaCl, and 0.01% (vol/vol) Triton X-100 at 4 C for 15 min with rotation. Eluted samples were analyzed by Western blotting.
Coimmunoprecipitation Assays
A preclearing step was performed on 500 µg LßT2 nuclear extracts that was combined with 15 µl protein G-agarose (Roche Molecular Biochemicals) and 2.5 µl rabbit preimmune serum in buffer A [20 mM Tris-HCl (pH 8), 0.2 mM EDTA, 1.5 mM MgCl2, 0.5 mM DTT, 150 mM NaCl, 0.1% Nonidet P-40] and incubated for 6 h at 4 C on a roller shaker. Protein G-agarose beads were then pelleted by centrifugation. The clarified supernatant was subsequently immunoprecipitated with either anti-Lhx3 antibody diluted at 1:200 or 2.5 µl rabbit preimmune serum for 1 h at 4 C. Immune complexes were then collected with 30 µl protein G-agarose overnight at 4 C on a roller shaker. After three washes with buffer A, beads were boiled for 3 min in 2-fold concentrated SDS-loading buffer, and samples were submitted to Western blot analysis using the anti-Isl-1 antibody (39.4D5).
Immunocytochemistry
Experiments were performed as previously described (9). Briefly, transgenic mouse embryos were fixed in 4% paraformaldehyde. After two washes with PBS, the embryos were incubated sequentially in 12, 15, and 18% sucrose in PBS at 4 C for 24 h each, and then frozen at –60 C. Sixteen-micrometer cryostat sections of embedded embryos were mounted onto Super Frost Plus Slides. Immunostaining was performed with antibodies directed against Lhx-3 or Isl-1 at a 1/500 dilution. The final revelation was performed with Alexa Fluor antibodies directed against rabbit or mouse IgG, respectively, diluted at 1/1000.
|
ACKNOWLEDGMENTS |
|---|
T3-1 and LßT2 cell lines; to Drs. K. L. Parker (University of Texas Southwestern Medical Center, Dallas, TX), S. L. Pfaff (The Salk Institute for Biological Studies, La Jolla, CA), and G. Salbert (Université de Rennes I, Rennes, France) for kindly providing expression vectors; and to Dr. S. Réteaux (Institut de Neurobiologie Alfred Fessard, Gif-sur-Yvette, France) and Dr. H. Westphal (National Institute of Child Health and Human Development, Bethesda, MD) for providing cDNA-containing vectors. We thank the Institut Fédératif de Recherche 83 for imaging facilities (PhosphorImager and confocal microscopy) and animal husbandry. The unrelenting secretarial support of Marie-Claude Chenut is greatly appreciated. We gratefully acknowledge the contribution of Dr. Lisa Oliver (Unité 601, Institut National de la Santé et de la Recherche Médicale, Nantes, France) for the correction of English text and editorial assistance. |
FOOTNOTES |
|---|
The authors have nothing to declare.
First Published Online April 13, 2006
Abbreviations: AP-1, Activator protein 1; CHO, Chinese hamster ovary; CRE, cAMP response element; DAPI, 4',6-diamidino-2-phenylindole; DTT, dithiothreitol; E13.5, embryonic d 13.5; GnRH-R, GnRH receptor; GnSE, GnRH receptor specific enhancer; hPLAP, human placental alkaline phosphatase; LIM-HD, LIM homeodomain; LIRE, LIM-related element; PGBE, pituitary glycoprotein basal element; PRL, prolactin; SAP, steroidogenic factor 1 adjacent protein; SF-1, steroidogenic factor 1.
Received for publication May 9, 2005. Accepted for publication April 4, 2006.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
