Specification of unique Pit-1 activity in the hGH locus control region

Abstract

The human GH (hGH) gene cluster is regulated by a remote 5′ locus control region (LCR). HSI, an LCR component located 14.5 kb 5′ to the hGH-N promoter, constitutes the primary determinant of high-level hGH-N activation in pituitary somatotropes. HSI encompasses an array of three binding sites for the pituitary-specific POU homeodomain factor Pit-1. In the present report we demonstrate that all three Pit-1 sites in the HSI array contribute to LCR activity in vivo. Furthermore, these three sites as a unit are fully sufficient for position-independent and somatotrope-restricted hGH-N transgene activation. In contrast, the hGH-N transgene is not activated by Pit-1 sites native to either the hGH-N or rat (r)GH gene promoters. These findings suggest that the structures of the Pit-1 binding sites at HSI specify distinct chromatin-dependent activities essential for LCR-mediated activation of hGH in the developing pituitary.

The expression of the GH gene has been the focus of intense study over the past 20 years. A consistent feature of all GH promoters is a highly conserved pair of AT-rich binding sites for the pituitary-restricted trans factor Pit-1 (1, 2). Between these two Pit-1 elements is a conserved binding site for a ubiquitous DNA-binding protein, Zn15 (3). Additional, less conserved cis elements have been identified in the vicinity of the GH promoter in various species. Studies carried out in vitro and in cell-transfection models indicate that the Pit-1 binding sites are essential to the expression of the GH gene (4–6). However, studies of human (h)GH gene expression in transgenic settings demonstrate that promoter-proximal elements including the Pit-1 sites are not sufficient to activate hGH gene expression in vivo. For example, in transgenic mice, the hGH gene with its full promoter and as much as 7.5 kb of 5′-flanking sequences is either not expressed or expressed at very low levels and does not maintain somatotrope cell specificity (7, 8). These data suggest that regulatory sequences quite distant from the hGH gene are necessary to establish a transcriptionally active chromatin domain.

The hGH gene cluster spans 48 kb on chromosome 17q22-24 (for details see Fig. 1; ref. 9). This cluster contains the pituitary-expressed hGH-N and four genes expressed in the placenta: Chorionic somatomammotropin-like (hCS-L), hCS-A, hGH-variant (hGH-V), and hCS-B (Fig. 1). A search for distal regulatory determinants of hGH-N gene expression revealed DNaseI hypersensitive sites (HSs) in pituitary chromatin and a partially overlapping set of HSs in placental chromatin (ref. 8; Fig. 1). HSI and HSII, located between −14.5 kb and −15 kb relative to the hGH-N promoter, are specific to pituitary chromatin. HSIV, at −30 kb, is specific to the placenta and HSIII and HSV, at −27.5 and −32 kb, respectively, are shared in both expressing tissues. The five HSs together comprise a putative hGH locus control region (LCR). Significantly, when hGH-N transgene DNA is extended to encompass the entire set of HSs, hGH gene expression is rendered pituitary-specific, copy number-dependent, and robust in every line generated (8, 10). This establishment of an autonomous chromatin domain that supports high-level expression from a linked transgene irrespective of the site of integration in the mouse genome fulfills the most rigorous operational definition of an LCR (11, 12).

Figure 1.

The hGH gene cluster and its LCR. The pituitary GH gene (hGH-N) is shown as a black box, and the placentally expressed genes are shown as gray boxes. The unrelated but closely linked SCN4a and Igβ genes are shown as white boxes. The bent arrows indicate transcriptional orientations. The position of each HS is marked by a downward-pointing arrow. The 1.6-kb HSI,II BglII fragment (Bg) and its F14 subfragment are shown below the map. The positions of the three Pit-1 sites at HSI and two Pit-1 sites at the hGH-N promoter are indicated by open ovals.

Dissection of the hGH LCR revealed that a 1.6-kb BglII subfragment containing HSI and HSII (Fig. 1) is sufficient to activate hGH transgene expression (8, 13). Developmental studies in which the hGH transgene with its 500-bp proximal promoter (−0.5hGH) was linked directly to this HSI,II fragment demonstrated full retention of pituitary specificity and consistent transgene activation (13). Additional LCR determinants, specifically those encompassing HSIII–HSV, are necessary to impart copy-number dependence on transgene expression (8). The major gene-activating function within the HSI,II fragment has been resolved to a 404-bp subfragment coincident with HSI (“F14 fragment”; Fig. 1; ref. 13). In contrast to its robust action in transgenic pituitaries, the HSI,II fragment only mediates a minimal (2–3-fold) enhancing activity when assayed in cell culture (13). Thus the activity imparted by HSI seems to be most evident in a developmentally dynamic and fully chromatinized setting. A search for DNA–protein interactions within the active F14 fragment that might mediate this activity revealed three closely spaced Pit-1 binding sites (ref. 14; Fig. 1). Preliminary studies supported a positive role for these sites in hGH transgene activation (14). Thus, there seem to be two Pit-1 arrays involved in hGH gene activation: one at the promoter and a second located 14.5 kb distal at HSI of the hGH LCR.

hGH LCR action and long-range activation of the hGH gene seem to be mediated by alterations in core histone acetylation at the hGH locus. Chromatin immunoprecipitation (ChIP) assays revealed a 32-kb domain of histone hyperacetylation encompassing the hGH LCR. This hyperacetylated domain is specific to pituitary chromatin (15, 16). The levels of core histone acetylation in this domain peak centrally at HSI,II and extend in the 5′ direction to HSV and in the 3′ direction to encompass the hGH-N gene promoter (15). This positioning suggested that one or more determinants at this site are involved in the targeting of histone acetyltransferase coactivators to the locus. Consistent with this hypothesis, site-directed deletion of HSI core elements, including two of the three Pit-1 sites, resulted in a dramatic decrease in acetylation across the 32-kb LCR domain and a corresponding drop in hGH-N transgene expression levels (17). Thus, establishment of the acetylated chromatin domain at the hGH locus in the pituitary and corresponding transcriptional activation of the hGH gene both are linked to HSI function (17).

The dual role of HSI to target histone modification and to activate gene expression over an extensive distance confirmed its central importance in developmental control of the hGH locus. These data also emphasize the importance of identifying and characterizing the determinants of HSI action. Although Pit-1 elements identified within the 404-bp F14 subfragment seem to constitute core determinants of HSI activity, it remains unclear whether all three Pit-1 determinants in the array are functionally involved and whether additional as-yet-unidentified cis elements in this region also contribute to HSI activity. Furthermore, the identification of Pit-1 elements at HSI as necessary determinants in hGH-N gene activation is intriguing in light of the fact that a second set of Pit-1 elements is situated within the hGH promoter (5, 6). This organization raises the question of whether the two sets of Pit-1 binding sites act in an additive manner or, alternatively, whether they provide distinct and complementing functions. The present study demonstrates that the two sets of Pit-1 arrays are distinct in their functions and that the Pit-1 sites present at HSI of the hGH LCR are unique in their capacity to activate the hGH-N transgene in the mouse pituitary.

Materials and Methods

Construction of Recombinant Plasmids.

Mutant F14 DNA subfragments were generated and linked to the hGH-N gene as described (14). The rat (r)GHPit3-hGH and PitABC-hGH constructs were prepared by replacing the F14 fragment with duplex synthetic oligonucleotides containing the Pit-1 site trimer arrays. The arrays maintain all three Pit-1 binding sites in their native orientation relative to the hGH-N gene.

Transient Transgenic Assay.

Generation of transgenic mouse embryos and processing of embryonic day (E)18.5 transgenic founders were performed as described (13).

Immunofluorescent Staining.

Monoclonal antibody specific to hGH (mAb9) has been described (13). Monkey anti-rGH, which cross-reacts with mouse (m)GH but not hGH, and rabbit anti-mPrl were obtained from the National Hormone and Pituitary Program. Sections (4–5 μm) of paraffin-embedded mouse embryos or excised adult pituitaries were dewaxed, followed by antigen retrieval with antigen unmasking solution (Vector Laboratories) and washing in PBS. Sections were blocked with 10% goat serum in PBS and incubated with the primary anti-hGH antibody (1:2,000) at 4°C overnight, followed by washing in PBS. Next, the sections were incubated with goat anti-mouse IgGγ conjugated to biotin (Jackson ImmunoResearch) at 1:500 for 1 h at room temperature, washed with PBS, and incubated with Cy3-conjugated streptavidin (Jackson ImmunoResearch) for 30 min at room temperature. After washing with PBS, sections were incubated with the second primary antibodies (mGH or mPrl at 1:2,000) overnight at 4°C, followed by washing in PBS and incubation with the second secondary antibody. Secondary antibodies were FITC-conjugated goat anti-human IgG/IgM (Jackson ImmunoResearch) for detection of mGH or FITC-conjugated goat anti-rabbit IgG (Jackson ImmunoResearch) for detection of mPrl. Sections were finally washed in PBS, mounted with GelMount (Fisher), and visualized by fluorescence microscopy using standard methods.

Reverse Transcription (RT)-PCR Assay for hGH and mGH mRNA.

Levels of hGH and mGH mRNAs were determined by coamplification of hGH and mGH mRNAs by RT-PCR using primers complementary to both sequences as described (8). The amplified mGH and hGH cDNAs were distinguished by unique restriction sites and quantified relative to each other after resolution on a 5% nondenaturing polyacrylamide (19:1 acrylamide/bis-acrylamide) gel by PhosphorImager analysis (Molecular Dynamics).

Preparation of Recombinant Pit-1.

A glutathione S-transferase-linked rPit-1 cDNA (gift of Sally Radovick, University of Chicago) was expressed in bacteria and affinity-purified as described (18).

Electrophoretic Mobility-Shift Assays (EMSAs).

Binding reactions (20 μl) contained 5 ng of a ³²P 5′ end-labeled, double-stranded DNA fragment in buffer comprised of 10 mM Hepes (pH 7.9), 1 mM EDTA, 1 mM DTT, 32 mM KCl, 10% glycerol, 0.08 g/liter poly(dI·dC), and 0.3 g/liter BSA. Purified recombinant glutathione S-transferase-Pit-1 was added last, and the mixture was incubated for 20 min at room temperature. Samples were resolved on a 5% nondenaturing polyacrylamide gel and visualized by autoradiography.

Transient Transfection.

Presomatotrope GHFT1 cells (19) were transfected by calcium phosphate-DNA precipitation as described (20). Ten-centimeter culture dishes at ≈75% confluency were transfected with 5 μg of supercoiled test plasmid. Five micrograms of pRL-CMV (Promega), which encodes Renilla luciferase, was included in the transfection mix to control for transfection efficiency. After 48 h, hGH expression was determined by hGH ELISA (Roche Diagnostics, Indianapolis) by using aliquots of transfected cell-culture medium. Luciferase activity in transfected cell lysates was determined with the dual luciferase reporter assay system (Promega). hGH levels were normalized to luciferase activities to correct for transfection efficiency. The data were represented as the mean of the three independent transfections.

Results

The Gene-Activating Function of HSI Within the hGH LCR Reflects Contributions from All Three of Its Pit-1 Elements.

Functional mapping of HSI has demonstrated that a 404-bp subsegment (F14 fragment) encompassing an array of three Pit-1 binding sites (Fig. 1) is consistently effective in activating a linked and otherwise inactive −0.5hGH-N transgene in the somatotropes of the differentiating (E18.5) anterior pituitary (Fig. 2B, F14-GH). These studies also have demonstrated that the frequency of hGH transgene activation decreased to ≈50% when the Pit-1 B and C sites were inactivated individually and was essentially eliminated when both the B and C sites were mutated or deleted together (Fig. 2, F14ΔB,C-GH; see also ref. 14). These data suggested that full LCR action by HSI depended on both the B and C Pit-1 binding sites. The possibility that the Pit-1 A site also might contribute to overall LCR function has not been similarly established.

Figure 2.

All three Pit-1 elements within the F14 fragment are required for position-independent activation of the linked hGH transgene. (A) Sequences of HSI Pit-1 elements A, B, and C. Base substitutions introduced to block Pit-1 binding are underlined. (B) Transgene constructs and corresponding frequencies of transgene activation. The minimal −0.5hGH gene and each of the F14-derived segments are shown. Inactivated Pit-1 elements are indicated by Xs. The hGH-N exons are shown as black boxes, and introns are shown as dark gray boxes. The hGH expression frequency for the set of transgenic founders carrying each construct is indicated to the right of the corresponding construct map. An asterisk indicates published results (14).

The potential contribution of the Pit-1 A site to HSI function was evaluated by selectively inactivating this site either alone or in combination with mutations at Pit-1 sites B and C (Fig. 2). Pit-1 sites were inactivated by introducing sequence transversions that block Pit-1 binding (Fig. 2A). Each derived F14 fragment was linked to −0.5hGH-N. This −0.5hGH-N comprises the entire hGH gene and 500 bp of contiguous 5′-flanking sequence containing all known promoter elements and specifically including the two promoter-proximal Pit-1 sites (ref. 21; Fig. 2B). The resultant constructs were injected into fertilized mouse oocytes, and the reimplanted embryos were genotyped and assayed at E18.5. The frequency of hGH transgene activation was assessed by scoring hGH protein expression in each embryonic transgenic pituitary (“transient-transgenic” founder assay) (refs. 13 and 14; Fig. 3A). Selective inactivation of the Pit-1 A site (F14ΔΑ-GH) decreased the frequency of hGH-positive embryos to 57%. Combining inactivation of the A site with mutations of the B or C sites (constructs F14ΔΑΒ-GH and F14ΔΑC-GH) resulted in essentially the same frequency of hGH-expressing embryos (60 and 43%, respectively; Fig. 2B). These data indicated that Pit-1 element A is essential for consistent gene activation, but in contrast to sites B and C, this site cannot individually trigger transgene activation. Thus, although site A seems to have lower individual activity than sites B or C, the full ensemble of three Pit-1 elements is necessary to ensure consistent (100% frequency) activation of the hGH-N locus in the transgenic setting.

Figure 3.

Nonvariegated hGH expression in the somatotropes of F14-GH transgenic embryos. (A) mAb 9 (anti-hGH)-stained E18.5 sagittal pituitary sections from hGH-expressing (Left, hGH⁺) and nonexpressing (Right, hGH−) F14ΔA-GH transgenic embryos are shown. hGH(⁺) cells are indicated by the red cytoplasmic stain. The small stained bodies in the hGH(−) section are red blood cells that were stained nonspecifically with the secondary antibody. (B) Coimmunofluorescence for hGH (red) and mGH (green) in a single hGH⁺ E18.5 F14ΔA-GH transgenic pituitary section. The yellow color in the merged image indicates colocalized hGH and mGH.

Lack of Variegation of hGH-N Transgene Expression when Activated by F14 Fragment Determinants.

To define the roles of the Pit-1 binding sites in hGH gene activation further, the consistency of transgene expression within each pituitary was assessed on a cell-to-cell basis. hGH expression was detected in E18.5 embryos by indirect immunofluorescent microscopy using an hGH-specific monoclonal antibody (Fig. 3A); the somatotropes in the fields were identified positively by simultaneous staining for endogenous mGH. Merging these two stains defined the extent to which expression from each hGH transgene was somatotrope-restricted (Fig. 3B). Remarkably, in all the embryos in which a pituitary was scored as positive for hGH expression the staining revealed that all somatotropes (i.e., mGH⁺ cells) stained for hGH. Reciprocally, in all transgenic pituitaries scored negative for hGH, there was a total absence of hGH expression in somatotropes (Fig. 3A). These findings indicate that activation of the hGH transgene by one or more of the F14 Pit-1 determinants results in the establishment of a uniformly positive, nonvariegated expression pattern.

Assembly of the Three HSI Pit-1 Elements into a Single Array.

The current data, along with the preceding studies (13, 14), indicate that HSI activity maps to the F14 subregion and the full array of three Pit-1 binding sites within this subregion is required to mediate consistent hGH activation in the transgenic setting (ref. 14; see above). It remained to be determined whether these three Pit-1 sites, although essential to hGH gene activation, are fully sufficient for this purpose. To test for sufficiency, the three Pit-1 elements (A, B, and C; Fig. 2A) were fused to form a Pit-1 trimer array (Fig. 4A, PitABC). The spacing of these three Pit-1 sites paralleled exactly that of a previously reported trimerization of an rGH gene promoter Pit-1 site (4). This element is identical to the orthologous element in the hGH-N promoter except for a single nucleotide outside the Pit-1 interaction sites (ref. 22; Fig. 4A, rGHPit₃). This rGHPit₃ array effectively mediates Pit-1-dependent enhancer action when assayed in cell-transfection studies (4). The PitABC trimer was structured in a manner identical to the rGHPit₃ trimer to facilitate a direct comparison of these two arrays and determine whether the HSI-derived Pit-1 sites mediate qualitatively distinct actions relevant to LCR functions.

Figure 4.

The promoter-proximal and HSI Pit-1 elements bind Pit-1 with similar affinities. (A) Structures of the rGHPit₃ and PitABC Pit-1 element trimers. (B) EMSA analysis of Pit-1 binding. rGHPit₃ and PitABC trimers were end-labeled and used as probes in an EMSA with purified glutathione S-transferase (GST)-Pit-1. The probes are indicated below the autoradiogram. The levels of glutathione S-transferase-Pit-1 in the reactions are indicated above the lanes. The identities of the three possible complexes consisting of one, two, and three Pit-1 homodimers as well as the unbound probe band are indicated to the right.

Equivalent Activities of the rGHPit₃ and PitABC Arrays in Protein-Binding and Enhancer Function in Transfected Cells.

EMSA studies using recombinant Pit-1 protein demonstrated that the two trimer arrays bound Pit-1 with similar affinities and kinetics and formed sets of complexes with identical electrophoretic migration patterns (Fig. 4B). Incubation of the respective Pit-1 array probes with GH3 (rat somatotrope) nuclear extracts resulted in equivalent gel-shift patterns comprised of Pit-1 complexes as determined by Pit-1 antibody supershift analysis (data not shown).

To confirm that both arrays were able to form functional Pit-1 complexes, the PitABC-hGH and rGHPit3-hGH constructs were compared for their abilities to enhance the expression of the −0.5hGH-N gene in transiently transfected GHFT1 cells (Fig. 5A). This mouse presomatotrope cell line expresses substantial levels of Pit-1 (19). As shown in Fig. 5B, the two Pit-1 arrays mediated equivalent and modest enhancement of expression from the linked −0.5GH gene with its own promoter-proximal Pit-1 binding sites. Thus, the two Pit-1 trimer arrays could not be distinguished on the basis of their in vitro protein-binding properties or by their abilities to enhance the expression of the −0.5hGH gene in transfected cells.

Figure 5.

The HSI Pit-1 array is uniquely able to activate hGH-N transgene expression. (A) Schematic diagrams of the promoter (rGHPit₃-GH) and HSI (PitABC-GH) synthetic Pit-1 element trimer transgene constructs. The individual Pit-1 elements are represented by short boxes, whereas the two native hGH-N promoter-proximal Pit-1 sites are shown as ovals. The −0.5hGH fragment contains the hGH-N gene. (B) hGH expression levels of transient GHFT1 cell transfections. The values (arbitrary units), corrected for transfection efficiency, represent the mean (±SD) of three independent transfections. The identity of each transfected construct is indicated below the graph. (C) hGH transgene expression. RT-PCR was carried out with pituitary RNA (co-RT-PCR, Top). Each lane corresponds to an independent mouse line. The positions of coamplified hGH and mGH RT-PCR products are indicated to the right of the autoradiograph. The Southern blot hybridization signals for hGH-N (Middle) and mζ-globin (Bottom) probes are indicated also to confirm the presence of the hGH-N transgenes compared with an endogenous control. The hGH hybridization signal is proportional to hGH-N transgene copy number.

Selective in Vivo Activation of the hGH-N Transgene in the Mouse Pituitary by the HSI Pit-1 Array.

To test and compare the activities of the two Pit-1 trimer arrays for in vivo actions, each was linked to the −0.5hGH-N transgene and used to generate transgenic lines. These two constructs, rGHPit3-hGH and PitABC-hGH, were identical to those used in the preceding cell-transfection studies (Fig. 5 A and B). Each construct was released from vector sequences and microinjected into fertilized mouse oocytes. Four transgenic lines were established with rGHPit₃-hGH, and six lines were established with PitABC-hGH. Transgene expression was evaluated in the pituitaries of adult F₁ mice generated from each founder (Fig. 5C) by using an RT-PCR coamplification procedure (8). There was no evidence of transgene expression in the pituitaries of any of the rGHPit₃-hGH lines. In contrast, all six PitABC-hGH lines expressed hGH mRNA at levels within several fold of endogenous mGH mRNA. Copy number-dependent expression was not expected because it depends on the HSIII–HSV region of the LCR (8). Double immunofluorescence for hGH/mGH and hGH/mPrl in pituitary sections from the PitABC-hGH mice demonstrated that hGH transgene expression was nonvariegated and somatotrope-specific (Fig. 6A). The cell-type specificity was confirmed further by demonstrating that in addition to being present in all mGH-positive cells, hGH was absent from the majority of mPrl-positive lactotropes (Fig. 6B). Occasional hGH⁺/mPrl⁺ cells were observed, consistent with the small defined population of somatolactotropes. Thus, the array of three HSI-derived Pit-1 binding sites is sufficient to confer high-level, position-independent, somatotrope-specific hGH transgene expression. Importantly, the HSI Pit-1 elements are unique in this activity as compared with GH gene promoter-proximal Pit-1 elements.

Figure 6.

Pituitary hGH-N expression in a PitABC-hGH mouse line is nonvariegated and somatotrope-restricted. Shown are hGH/mGH (A) and hGH/mPrl (B) coimmunofluorescence in adult anterior pituitary sections from PitABC-hGH line 5. (A) Counterstained with 4′,6-diamidino-2-phenylindole (blue) to indicate nuclei. A indicates that hGH is present in all mGH⁺ cells (somatotropes, yellow merged stain). B indicates that the hGH⁺ and Prl⁺ (Pit-1-expressing lactotropes) are separate populations, evidenced by nonoverlapping red and green signals. In rare instances, hGH and mPrl are coexpressed in the same cell, which represents the somatolactotrope that expresses both GH and Prl (indicated by a blue arrow).

Discussion

We have described an LCR for the hGH gene cluster (Fig. 1). High-level, copy number-dependent, and position-independent expression of hGH-N transgenes in the mouse pituitary could be established consistently by linkage to the full LCR (8). HSI,II, located 14.5 kb 5′ to the hGH-N gene, was the only LCR component specific to pituitary chromatin. Transgenic analyses demonstrated that a 1.6-kb restriction fragment encompassing HSI,II was sufficient to establish high-level, somatotrope-specific, position-independent, and appropriately timed expression of the linked hGH-N gene (8). Subsequent studies localized this activity to the 404-bp F14 which co-maps to HSI (13) and demonstrated that this activity depended on an array of Pit-1 binding sites (ref. 14; Fig. 2). The HSI region therefore seems to encompass a set of Pit-1 sites that comprise the major somatotrope-specific activation determinants of the hGH LCR.

Analysis of chromatin modifications at the hGH locus revealed a broad 32-kb domain of pituitary-specific acetylation of histones H3 and H4 that peaks centrally at a position coincident with HSI,II. Remarkably, deletion of a 99-bp segment encompassing the high-affinity Pit-1 elements B and C of HSI (Fig. 1) from an otherwise intact 86-kb hGH locus transgene results in a generalized loss of histone acetylation throughout the domain, a concomitant loss of HSI formation in pituitary chromatin, and a dramatic loss of hGH-N expression (17). These findings strongly suggested that the Pit-1 elements at HSI constituted core LCR determinants, and that these sites link histone acetylation with transcriptional activation of the hGH-N gene. The demonstration by others of a specific interaction between Pit-1 and the histone acetyltransferase coactivator CBP/p300 in vitro (23) suggests that Pit-1 could directly target histone acetyltransferase activity to the hGH locus.

In the present study, we sought to resolve whether the three Pit-1 elements at HSI are the sole determinants of HSI activity. By designing a transgene in which the three isolated Pit-1 elements were joined in the absence of the surrounding sequences (Fig. 4A), we showed that these elements indeed are sufficient for driving high-level, position-independent (Fig. 5), somatotrope-restricted (Fig. 6) hGH-N transgene expression in the mouse. Thus, the activity of the HSI region of the hGH LCR does not seem to involve any cis elements outside the Pit-1 element array.

The necessity and sufficiency of the HSI Pit-1 array for hGH gene activation was surprising in light of the fact that two well documented Pit-1 elements in the hGH-N promoter clearly are insufficient for this purpose. The three HSI Pit-1 elements could be functioning in a simple additive manner with the two Pit-1 sites resident in the hGH-N promoter to generate a critical mass of factor complexes (24). Alternatively, the Pit-1 sites at HSI could be serving a distinct set of functions. We resolved these two possibilities by directly comparing the activities of the HSI Pit-1 element array with that of a parallel trimer of a promoter-proximal Pit-1 element (Fig. 4A). The two Pit-1 trimer arrays bound Pit-1 in an identical manner by EMSA (Fig. 4B), and both displayed a similar weak enhancement of hGH-N expression in transiently transfected cells (Fig. 5B). In marked contrast, the HSI-derived Pit-1 trimer array was unique in its ability to activate the −0.5hGH transgene (Fig. 5C). These results support the conclusion that the Pit-1 sites at HSI and within the GH promoter mediate distinct activities, with the former being central to LCR-mediated gene activation.

There is emerging evidence that the activity of a trans-acting factor can be modulated allosterically by the structure of its DNA-binding site. This allosteric effect would provide an additional mechanism for the specification of transcription factor function. The best-described examples of such allosteric control involve the POU homeodomain family of transcription factors, of which Pit-1 is a founding member. These proteins represent a highly flexible class of DNA-binding factors comprised of a classical homeodomain (POU_H) joined to a POU-specific homeodomain (POU_S) by a flexible, disordered linker. This bipartite DNA-binding activity allows for multiple conformations in the complex between POU protein homodimers and DNA (25, 26). These conformations have the capacity to alter the topology of bound protein with consequent alterations in the structure of its exposed protein–protein binding surface and downstream effector pathways. For example, investigation into the basis for the differential activities of two types of Oct-1 binding sites revealed that although both binding sites have equal affinities for Oct-1, only one of the complexes can recruit the coactivator OBF-1 (OCA-B, BOB-1) (27). Similar instances of allosteric specification have been observed also with Pit-1. In this case, an additional two base pairs between the POU_S and POU_H interaction sites in a rat prolactin gene Pit-1 element (Prl-1P) as compared with a rat GH gene element (GH-1) prevents the recruitment of the nuclear receptor corepressor (N-CoR) to the complex. In contrast, the GH-1 element complex is able to interact with N-CoR (22). These findings suggest that allosteric controls mediated by DNA-binding site structure may be a common mechanism for regulating the functions of POU domain proteins. The distinct activity of the HSI Pit-1 element array may be mediated by a topological specification of Pit-1 complexes that allows selective recruitment of a histone acetyltransferase-containing cofactor complex to the hGH chromatin locus.

The distinct in vivo functions of the HSI and the promoter Pit-1 sites also may be explained by differences in binding-site access. Despite nearly identical binding affinities of the HSI and promoter Pit-1 arrays to naked DNA (Fig. 4), the HSI Pit-1 elements may be uniquely accessible to Pit-1 in a natural chromatin context in vivo (Fig. 5). Remarkably, recent studies have shown that the promoter Pit-1 sites can be occupied only after HSI has activated the locus (17). This temporal dependence of promoter Pit-1 occupancy on HSI function suggests that Pit-1 can bind to its HSI sites in preactivated, hypoacetylated chromatin when access to the promoter Pit-1 sites is denied. Binding to the promoter sites depends on the preceding chromatin modifications initiated at HSI. The significantly higher levels of histone acetylation at HSI than at the promoter in the fully activated locus further suggest that Pit-1 bound at HSI is far more effective in targeting coactivator recruitment. This difference would imply that once the two sets of sites are occupied by Pit-1, allosteric differences in the corresponding complexes mediate distinct yet complementing downstream coactivator cascades. Thus trans-factor access and allostery models are not mutually exclusive, and aspects of both mechanisms may contribute to the unique contributions of the two Pit-1 arrays in activation of the hGH locus.

Abbreviations

h

human

r

rat

m

mouse

LCR

locus control region

HS

hypersensitive site

E

embryonic day

RT

reverse transcription

EMSA

electrophoretic mobility-shift assay