Distinct Chromatin Configurations Regulate the Initiation and the Maintenance of hGH Gene Expression

Abstract

For many mammalian genes, initiation of transcription during embryonic development must be subsequently sustained over extensive periods of adult life. It remains unclear whether maintenance of gene expression reflects the same set of pathways as are involved in initial gene activation. The human pituitary growth hormone (hGH-N) locus is activated in the differentiating somatotrope midway through embryogenesis by a multicomponent locus control region (LCR). DNase I-hypersensitive site I (HSI) of the LCR is essential to full developmental activation of the hGH-N locus. Here we demonstrate that conditional deletion of HSI from the active hGH locus in the adult pituitary effectively silences hGH-N expression. Analyses of chromatin structure and locus positioning demonstrate that a specific subset of the HSI functions active in the embryo retain their HSI dependence in the adult pituitary. These functions sustain engagement of the hGH locus with polymerase II (Pol II) factories, histone acetylation at the hGH-N promoter, and looping of the LCR to its target promoter. These data reveal that HSI is essential to both the maintenance and the initiation phases of gene expression. These observations contribute to our mechanistic understanding of how stable patterns of mammalian gene expression are established in a terminally differentiated cell.

INTRODUCTION

Major shifts in transcriptional profiles during organismal development and cell type specification reflect complex sets of controls that mediate alterations in chromatin structure as well as specific positioning of loci within the nucleus (1). Gene activation is generally linked to acetylation of histones H3 and H4, to the distancing of the gene from the nuclear membrane, and to colocalization of the gene with punctate concentrations of polymerase (Pol) II known as “transcription factories” (2, 3). Reciprocally, gene silencing often correlates with deacetylation of the locus, positioning of the locus proximal to the nuclear periphery (4), and disengagement from Pol II factories (5). These dynamic changes in locus structure and positioning play a pivotal role(s) in stabilizing the gene in active or silenced states (6) and are triggered by interactions between trans factors and their corresponding sets of cis-acting regulatory determinants (7–9).

In metazoans, cis-regulatory elements critical to gene activation can be located hundreds of kilobases from target promoters (10). One class of these remote regulatory elements is the locus control region (LCR). An LCR comprises a set of determinants that can modify chromatin domains, augment transcription, establish subnuclear localization of target genes, and insulate a locus from the surrounding chromatin environment (8, 11).

The human growth hormone (hGH) gene cluster contains the pituitary-specific hGH-N and four placenta-specific paralogs, hCS-L, hCS-A, hGH-V, and hCS-B (Fig. 1A) (12). hGH-N is transcriptionally activated by a multicomponent hGH LCR located between 14.5 and 32 kb upstream of the hGH-N promoter (13). DNase I-hypersensitive site I (HSI), the major enhancer determinant of the hGH LCR, is positioned 14.5 kb upstream of the hGH-N gene (13). The multiple functions of HSI include recruitment of histone-modifying complexes (14, 15), activation of a localized domain of noncoding transcription (16, 17), and higher-order “looping” of chromatin between itself and the hGH-N promoter (18). HSI activity is dependent on an array of three binding sites for the pituitary-specific POU homeodomain transcription factor, Pit-1 (19). Deletion of these Pit-1 binding sites in the germ line results in loss of HSI formation, loss of the full array of HSI actions, and a corresponding dramatic decrease in hGH-N gene activation in the developing pituitary (15). Thus, the Pit-1 array within HSI plays an essential role in the robust activation of hGH-N expression during embryonic development.

Fig 1.

Cre-mediated deletion of the HSI Pit-1 array from the hGH transgene locus. (A) Structures of the native hGH locus, the 87-kb hGH/P1 transgene, and the hGH/P1-derived transgene containing a floxed Pit-1 array hGH/P1(FloxHSI). The genes, their sites of expression, and the direction of their transcription are all indicated, as are the positions of the four HSs that constitute the hGH LCR in the pituitary. The two loxP sites that flank a 300-bp region containing an array of three Pit-1 binding sites at HSI are indicated by arrowheads in the simplified schematic of the hGH/P1(FloxHSI) transgene. An expanded view of the floxed region of the hGH/P1(FloxHSI) transgene is shown at the bottom. (B) Cre-mediated deletion of HSI from the hGH/P1 transgene. A mouse carrying the hGH/P1(FloxHSI) transgene (A) was crossed with a mouse carrying a transgene encoding Cre recombinase under the transcriptional control of the CMV promoter. The transgenic mice generated from this cross are designated hGH/P1(loxpΔHSI) transgenic. (C) Germ line deletion of HSI results in a dramatic loss of hGH-N expression. RNA was extracted from the pituitaries of mice carrying the hGH/P1(FloxHSI) or the hGH/P1(loxpΔHSI) transgene. hGH-N mRNA was quantified relative to the endogenous mGH mRNA by a co-RT-PCR analysis (13). A representative autoradiograph of the co-RT-PCR is displayed, with the bands corresponding to the mGH mRNA and the hGH mRNA (major and alternatively spliced forms) indicated by the arrows. The average hGH-N mRNA expression levels, normalized to copy number of hGH/P1(FloxHSI) transgene, are shown below the respective lanes. (D) Western analysis of pituitary extracts from the hGH/P1(FloxHSI) and the hGH/P1(loxpΔHSI) transgenic mice. The membranes were probed with antibodies specific for hGH, mGH, and Pit-1.

Studies in multiple model systems support the critical role(s) of remote cis-regulatory elements in developmental activation of gene expression (20). What remains less clear is whether these same elements participate in long-term maintenance of gene activity. Studies with Drosophila melanogaster suggest that once a gene is activated, it can be locked into chromatin configurations that sustain gene expression in an autonomous fashion (21, 22). These data suggest that such “memory modules” may function independently of the original gene-activating signals (23, 24). Although compelling in certain settings, the generality of these findings and their extension to mammalian systems have not been directly tested. Studies of long-term gene expression maintenance are of particular relevance to mammalian species in light of their substantial longevity.

Initial activation of hGH-N expression is critically dependent on the actions of the Pit-1 array in HSI. This expression must be subsequently maintained at robust levels throughout adult life to sustain normal physiological and metabolic functioning (25). Here we demonstrate that the Pit-1 array essential for embryonic activation of hGH-N is also required for its sustained expression throughout adult life. Inactivation of HSI in the adult pituitary by conditional deletion of this HSI Pit-1 array results in substantial decay of hGH-N expression. This decay is associated with a complex spatial reconfiguration of the locus in which it is released from Pol II transcription factories while retaining its “active” positioning in relationship to the nuclear periphery. Furthermore, the conditional inactivation of HSI results in a selective loss of H3 and H4 histone acetylation at the hGH-N promoter and adjacent 5′ flanking region, reverses the looping between LCR and promoter, and leaves LCR acetylation intact. Thus, the HSI determinant critical to activation of the hGH-N gene during embryogenesis is also essential to the maintenance of robust hGH-N expression throughout adult life. These data reflect a previously unappreciated complexity in gene regulation as a locus is configured for stable, high-level expression in the adult organism.

MATERIALS AND METHODS

Generation of transgenic mice.

The hGH/P1(FloxHSI) transgene was generated by RecA-mediated recombination in Escherichia coli using the hGH/P1 clone as the template as described previously (15). The hGH/P1 clone encompasses 87 kb of the hGH locus and all the regulatory elements for somatotrope-specific expression of hGH-N (26, 27) (Fig. 1A). The three Pit-1 binding sites in the HSI region of the hGH LCR (19) were flanked in hGH/P1 by the 34-bp loxP sequences, and the correct insertion was confirmed by DNA sequencing and Southern blotting (data not shown). The hGH/P1(FloxHSI) plasmid was linearized with NotI digestion and purified as described previously (15). The DNA was microinjected into the pronuclei of C57BL/6J × SJL mouse zygotes. Founders were identified by dot blot analyses of tail DNA using a 1.37-kb SmaI fragment from hGH-N as a probe (15). The transgene copy number was determined by Southern blotting using DNA from previously established hGH/P1 transgenic mouse lines as a reference (27).

The HSI,II-rtTA transgene (Fig. 2A) was constructed by placing the tetracycline-controlled trans activator (rtTA) (28) 3′ to the 1.6-kb DNA fragment encompassing the somatotrope-specific HSI,II of the hGH LCR and a 500-bp fragment containing the hGH-N promoter. The generation of transgenic mice carrying the HSI,II-rtTA was as described above.

Fig 2.

HSI is essential to the maintenance of hGH-N expression in the adult pituitary. (A) Conditional deletion of the HSI Pit-1 array in the adult mouse. hGH/P1(FloxHSI) transgenic mice were crossed with mice carrying the HSI,II-rtTA transgene and the tetO-Cre transgene. Triply transgenic mice generated in this manner were identified by genomic DNA analysis. Treatment of these mice with DOX resulted in activation of the rtTA, expression of Cre, and deletion of the Pit-1 array from the hGH/P1(FloxHSI) transgene in the pituitary somatotropes to generate hGH/P1(loxpΔHSI). (B) Analysis of pituitary RNA from adult mice after deletion of the Pit-1 array from HSI. The triply transgenic adult mice (as in panel A) were treated with doxycycline (+) in drinking water containing sucrose or with sucrose water alone (−). RNA was analyzed for hGH-N and mGH RNAs after 14 days of treatment by the co-RT-PCR assay (as in Fig. 1C). (C) Histogram representing the quantification of two separate studies as in panel B. In each case, the expression was normalized to copy number of hGH/P1(FloxHSI). The numbers of mice tested in each group are indicated below the corresponding bars. The means and standard errors are shown along with the results of the Student t test.

The CMV-Cre transgene (kind gift of Klaus Kaestner, University of Pennsylvania) contains a 1.0-kb fragment containing the cre recombinase open reading frame under the control of the human cytomegalovirus (CMV) enhancer and promoter.

A transgenic mouse carrying the tetO-P_hCMVCre transgene was a gift from Jeffrey Gordon, Washington University School of Medicine (28). The construction contains seven copies of the tet operator sequence that are located upstream of a minimal CMV promoter.

The hGH/P1(loxΔHSI) transgenic mouse line was generated by crossing the hGH/P1(FloxHSI) mouse with a CMV-Cre mouse. A compound transgenic founder was identified by PCR using primer sets located within each transgene. The successful deletion of HSI was confirmed by PCR using two sets of amplimers flanking and within the deleted sequence (data not shown). The founder was then outbred to establish an hGH/P1(loxΔHSI) transgenic line with germ line inactivation of HSI.

Conditional deletion of the Pit-1 binding site array from HSI in adult hGH/P1(FloxHSI) mice.

Triply transgenic mice used for conditional inactivation of HSI were generated by crossing hGH/P1(FloxHSI) mice with compound transgenic mice carrying HSI,II-rtTA and tetO-P_hCMVCre transgenes (Fig. 2A). The triply transgenic mice were identified by PCR using tail DNA. Eight- to 10-week-old mice were used for the studies. Deletion of the Pit-1 array from HSI was induced by a 14-day administration of doxycycline (DOX; 2 mg/ml) and 5% (wt/vol) sucrose in the drinking water given to the mice. Mice in the control group were treated with 5% sucrose for the same amount of time. After doxycycline or control treatment, mice were sacrificed and pituitary glands harvested for the studies. Doubly transgenic mice carrying hGH/P1(FloxHSI) plus HSI,II-rtTA or tetO-P_hCMVCre were treated with doxycycline to serve as additional controls.

RT-PCR assays.

RNA extracted from the pituitaries of transgenic mice was assayed for hGH-N gene expression by targeted reverse transcription-PCR (RT-PCR) as described previously (13, 15). The ratio of hGH-N to mGH RNA was divided by the corresponding transgene copy number to establish hGH-N transgene expression per copy in each mouse.

Western blotting.

Pituitaries were dissected from transgenic mice and washed with cold phosphate-buffered saline (PBS) followed by lysis in 20 μl of 2× SDS-PAGE loading buffer. The proteins were separated by 12% or 15% SDS-PAGE and transferred to membranes. hGH protein was detected using a species-specific mouse anti-hGH antibody (ab7905; Abcam Cambridge, MA). mGH was detected by rabbit anti-mGH antibody (National Hormone and Peptide Program, NIH). A Pit-1-specific antibody (sc16288) was purchased from Santa Cruz (Santa Cruz, CA).

Immunofluorescent staining.

Preparation of single-cell suspensions from mouse pituitaries, immunofluorescent staining, and image analyses were all performed as described previously (26). Briefly, four pituitaries from each experimental group of mice were pooled and the pituitary cells were dissociated with cell dissociation buffer (Invitrogen). The cells were fixed on slides with 4% formaldehyde in PBS at room temperature for 10 min. The slides were washed with PBS, and the cells were permeabilized by treatment with 0.5% saponin and 0.5% Triton X-100 for 10 min at room temperature. The cells were then washed with PBS and blocked in 4× SSC (1× SSC is 0.15 M NaCl plus 0.015 M sodium citrate), 2% bovine serum albumin (BSA), and 0.1% Tween 20. Mouse GH was detected using monkey anti-rat GH that cross-reacts with mGH but not hGH (National Hormone and Peptide Program, NIH) (14). The antibody specific for hGH (mAb9) has been described previously (26). After a washing with 4× SSC and 0.1% Triton X-100, the cells were mounted with fluorescent mounting medium (KPL, Gaithersburg, MD) and visualized by Olympus IX70 inverted microscopy. The nuclei were stained with 4′,6-diamidino-2-phenylindole (DAPI). The images were collected and processed using Deltavision Softworx software (Applied Precision, WA). The images were analyzed by using ImageJ software (NIH). Only the cells with a clearly intact cytoplasmic hormone-staining pattern over background were scored as positive for the hormone markers. More than 200 mGH-positive cells in each group were analyzed.

Semiquantitative PCR detection of conditional HSI deletion.

Pituitary DNA samples from triply transgenic mice treated with DOX or with sucrose only (control) were extracted with a DNeasy kit (Qiagen). Thirty nanograms of the DNA from each sample was subjected to PCR. The sequences for the HSI amplicon used are as follows: 5′-ATTCCAATGAACTGAACATCTGACAGC-3′ and 5′-TCCGTTTTCCAGTCTGTGCT-3′. The sequences for the p2 primers were described previously (15). The PCR products were resolved on a 1.5% agarose gel, and the gel was then stained with SYBR gold (Invitrogen). The band intensities were quantified on a Typhoon phosphorimager (GE).

Immuno-DNA FISH (3D FISH).

Three-dimensional (3D) DNA fluorescent in situ hybridization (FISH) was performed as described previously (8), with modifications. Briefly, dissociated pituitary cells were fixed and treated with 0.5% saponin and 0.5% Triton X-100 as described above. The slides were incubated for 10 min in 0.1 M HCl at room temperature, treated with RNase (100 μg/ml) at 37°C for 1 h, and washed with 2× SSC prior to hybridization. The probe used to detect the hGH locus is the 87-kb hGH/P1 clone (27). The hGH/P1 probe was labeled with digoxigenin-11-dUTP (Roche) by nick translation. One hundred nanograms of labeled probe and 5 μg of human cot-1 DNA (Invitrogen, Grand Island, NY) were used for hybridization. The cells on the slides were denatured at 76°C in 70% formamide–2× SSC, pH 7.0, for 3 min and 50% formamide–2× SSC, pH 7.0, for 1 min prior to hybridization. The hybridization was carried out at 37°C overnight. The slides were washed three times with 50% formamide–2× SSC at 42°C for 5 min each time. Immunostaining was carried out as described above. The hGH locus was detected using fluorescein isothiocyanate (FITC)-conjugated mouse antidigoxigenin antibody (Sigma). The primary antibody used to detect RNA Pol II recognizes the phosphorylation at serine 5 of the C-terminal domain of Pol II (ab5131; Abcam) and the secondary antibody was Cy3-conjugated donkey anti-rabbit IgG (Jackson ImmunoResearch Inc.). The somatotrope was identified by the detection of cytoplasmic mGH using the monkey anti-rGH antibody as described above. The secondary antibody used to detect the mGH was Cy5-conjugated donkey anti-human IgG (Jackson ImmunoResearch Inc.).

3D image analysis.

Images stacks (z sections spaced 0.25 μm apart) were collected on an Olympus IX70 inverted microscope using a cooled charge-coupled-device (CCD) camera and deconvoluted using Deltavision Softworx software (Applied Precision, WA). The distance of the hGH locus from the nuclear periphery and association of the hGH locus with Pol II transcription factories were measured on 3D reconstructed images using ImageJ (NIH) or Volocity software. The shortest distance between the middle of the transgene signal and the nuclear periphery (defined as the sharp drop in DAPI staining) was measured. The measurements were normalized with respect to the radius of the nuclei to compensate for differences between preparations. When nuclei had an ovoid shape, the longest radius was measured to avoid biasing the results. Partial overlapping of hGH locus with Pol II signal as detected on merged images was scored as association in each case.

ChIP assays.

Chromatin immunoprecipitation (ChIP) assays were performed as described previously (17), with minor modifications. Four or five pituitaries from each group were pooled for preparation of the pituitary chromatin in these assays. Twenty micrograms of pituitary chromatin was used for each assay. The antibodies specific for the acetylated histone H3 and H4 were from Millipore (Billerica, MA). Normal rabbit serum (Santa Cruz) was used as a control for the ChIP assays. The input and bound DNAs were amplified by PCR. PCR products were quantified by Southern blotting as described previously (17). The sequences and position of the primers used for PCR were as described previously (17). A series of dilutions of the chromatin input fraction was used to determine the linear region for the PCR amplification. The signals of bound fractions were normalized to the signals of 0.1% of the input (the second dilution in Fig. 7) after subtracting the signal from the normal rabbit IgG control. The signals of bound fractions were normalized to the signals of 0.1% of the input for AcH3 (the second dilution in Fig. 7) or 1% of the input for AcH4 (the first dilution in Fig. 7) after subtracting the signal from the normal rabbit IgG control. The sequences of the HSII primers used for PCR analyses were 5′-CATGACCTGGTAGTCCCAGCT-3′ and 5′-TCCGTTTTCCAGTCTGTGCT-3′. The sequences of the remaining primers used for the PCR analyses have been described previously (17).

Fig 7.

Conditional deletion of HSI from the active hGH locus in the adult pituitary results in a selective loss of histone acetylation at the hGH-N promoter and adjacent 5′ flanking region. A map of the hGH/P1(FloxHSI) transgene is shown (top). The vertical arrows indicate the positions of the four pituitary-specific DNase I HSs. The short horizontal lines indicate the sites of the six sets of PCR amplimers used in the ChIP assays. ChIP assays were performed using antibodies specific for acetylated histone H3 (AcH3) and histone H4 (AcH4) and pituitary chromatin from doxycycline-treated and control mice. The input and bound DNAs were assessed by semiquantitative PCR followed by Southern blotting as described previously (17). The relative enrichment for acetylated histone was calculated as the ratio of signal from bound fraction to 1% of input (first dilution) for AcH4 and to 0.1% of input (second dilution) for AcH3 after subtracting the background signal detected with control IgG. The full data set, representing four replicate studies, is shown on the histogram (bars represent the means and standard deviations). A representative ethidium bromide-stained agarose gel of the analytic PCR studies is shown at the bottom. Each experiment was performed four times (*, P < 0.05; **, P < 0.01).

3C assay.

The chromatin conformation capture (3C) procedure was performed as described previously (18). The frequencies of ligation between hGH-N promoter fragment and the other four 5′ BglII fragments at the hGH locus were assessed by semiquantitative PCR analysis. The sequence for the hGH-N promoter BglII fragment was 5′-AAAGATGCCCTGTCCAGCCA-3′. The sequences for the primers of the other BglII fragments were as described previously (18). The PCR products were separated on 1.5% agarose gels. The gels were stained with SYBR gold and quantified using the ImageQuant program (GE Healthcare). The ligation frequencies were calculated as a ratio of ligated products to the corresponding random ligation of digested hGH/P1(FloxHSI) plasmid DNA. This ratio was then normalized to the 3C analysis at an endogenous basal transcription factor locus (ercc3) (29) using a pair of identically oriented primers located within two adjacent BglII fragments of the ercc3 locus. Thus, the ligation efficiency was calculated as (signal_pituitary/signal_{random control})/signal_{ercc3 ligation}.

RESULTS

Germ line deletion of the Pit-1 array at HSI blocks hGH transgene expression.

To assess the contribution of HSI to the maintenance of hGH-N expression, we established a system in which HSI could be conditionally deleted from a previously characterized and validated hGH/P1 transgene (Fig. 1A). The functionally critical array of three Pit-1 binding sites within HSI (19) was flanked by loxP recombination sequences [hGH/P1(FloxHSI)] in the context of an 87-kb hGH/P1 transgene (Fig. 1A). A mouse carrying a single copy of this hGH/P1(FloxHSI) transgene was identified and maintained as a line for subsequent studies.

As an initial test of the loxP recombination, the hGH/P1(FloxHSI) mouse was crossed with a mouse carrying the ubiquitously expressed CMV-Cre transgene (Fig. 1B). This cross should result in the deletion of the Pit-1 array early in embryogenesis and inactivate hGH-N expression (15). This effect was confirmed (Fig. 1C); in comparison to the robust levels of hGH mRNA in the hGH/P1(FloxHSI) mouse pituitary, hGH-N mRNA in the pituitary of a mouse carrying the germ line hGH/P1(loxpΔHSI) transgene was reduced to trace levels. Western analysis confirmed this observation (Fig. 1D). These data are consistent with the essential role of the HSI Pit-1 array in developmental activation of hGH-N expression (15).

Deletion of the HSI Pit-1 array from the hGH transgene locus in the adult mouse pituitary effectively silences hGH-N expression.

We next assessed the role of the HSI Pit-1 array in maintenance of hGH-N expression in the adult. A set of two transgenes were assembled that allowed for doxycycline (DOX)-controlled Cre recombinase expression in the pituitary somatotropes (Fig. 2A; see also Materials and Methods). Triply transgenic mice carrying these two transgenes and the target hGH/P1(FloxHSI) transgene, hGH/P1(FloxHSI) × HSI,IIrTA × tetO-Cre, were divided into experimental and control groups. The experimental group was supplied with drinking water containing DOX (2 mg/ml) and 5% sucrose for 14 days. The control group was provided 5% sucrose water over the same 14-day course. At the end of the treatment period, hGH-N mRNA levels were assessed in the corresponding pituitaries by co-reverse transcription-PCR (co-RT-PCR) (Fig. 2B and C). The data revealed a significant reduction in hGH-N mRNA in the DOX-treated group compared to the sucrose-only controls (Fig. 2B and C) and compared to DOX-treated binary transgenic mice [hGH/P1(FloxHSI) plus HSI,II rtTA or hGH/P1(FloxHSI) plus Tet-Cre] (data not shown). These results led us to conclude that the Pit-1 array in HSI is essential to the maintenance of hGH-N expression in the adult pituitary.

Conditional deletion of HSI from the fully activated hGH/P1(FloxHSI) locus in the adult mouse pituitary resulted in a 60% decrease in hGH-N mRNA levels in the pituitary (Fig. 2C). Although substantial, this decrease was less than the 98% decrease observed in mice with germ line inactivation of HSI (compare Fig. 2B and C with Fig. 1C) [hGH/P1(loxpΔHSI)]. This difference could reflect two distinct and nonexclusive mechanisms. It is possible that the DOX-dependent Cre-mediated deletion of HSI occurred in only a fraction of the somatotropes, resulting in full silencing in this subpopulation. Alternatively, the deletion may have occurred in all somatotropes but resulted in a partial inactivation of the affected hGH locus. To distinguish between these two possibilities, we examined the impact of the Cre-mediated deletion in individual somatotropes using an immunostaining approach (Fig. 3A). hGH-N is restricted to the somatotropes in pituitary of transgenic mice carrying the intact hGH locus (26). Disaggregated pituitary cells were prepared from triply transgenic mice supplied with sucrose water alone. The somatotropes in these cell preparations were identified using antibody specific for the endogenous mGH, while hGH expression was visualized with a separate, species-specific antibody (Fig. 3A, images a and d). hGH-N was clearly visualized in 86% of the somatotropes (mGH⁺ cells in Fig. 3A). In contrast, parallel analysis of the triply transgenic mice treated with Dox revealed that only 30% of the pituitary somatotropes were hGH⁺ (Fig. 3A). The intensity of hGH-N staining in hGH⁺ cells in these DOX-treated mice was equivalent to that observed in the control animals (sucrose water only). Thus, 70% of somatotropes had lost hGH expression subsequent to Cre-mediated recombination. This percentage correlated well with the 60% decrement in hGH-N mRNA expression in the DOX-treated mice (Fig. 2 and 3). These data support the conclusion that deletion of the Pit-1 binding site array from HSI in the adult pituitary results in a nearly complete loss of hGH-N expression, comparable to that seen in the germ line deletion.

Fig 3.

Expression of the hGH-N transgene is markedly decreased in somatotropes of the adult pituitary after conditional deletion of the HSI Pit-1 array. (A) Immunofluorescent staining was performed on disaggregated pituitary cells isolated from triply transgenic mice treated for 2 weeks with doxycycline (top) or sucrose alone (bottom). The antibodies used in the studies were specific for the endogenous mGH or the transgene-encoded hGH-N. The merges of the two images are shown (Merge). Nuclei were stained with DAPI (blue). Two somatotropes (mGH⁺ cells) are indicated by the arrows in the top images; the cell that is also positive for hGH is indicated by the orange arrow, and a somatotrope lacking appreciable levels of hGH-N is indicated by the white arrow. In the bottom images, both somatotropes costained for hGH-N. The numbers to the right of the two rows represent the percentages of somatotropes that costained for hGH-N. The data represent analysis of a minimum of 600 somatotropes isolated from four pituitaries from both the DOX- and sucrose-treated mice. Bars in the lower right of each frame correspond to 5 μm. (B) Assessment of DOX-induced HSI deletion in adult pituitary somatotropes. DNA was isolated from the pituitaries of triply transgene mice that had been fed sucrose (control) or DOX and from a mouse with a germ line HSI deletion. The deletion of the Pit-1 array at HSI was assessed by PCR using two amplicons, the first targeting the HSI region and the second located outside the floxed segment. The efficiency of DOX-induced HSI deletion was detected by a decrease in the ratio of the HSI-dependent PCR product relative to that in the flanking region. The average ratio for the DOX-treated mice was 76% that of the sucrose control. As an additional control, the data demonstrate that the DNA from the mouse carrying the HSI germ line deletion is entirely lacking the HSI-specific PCR fragment.

The efficiency of the doxycycline-induced HSI deletion was also assessed by targeted PCR (Fig. 3B). Pituitary DNA was isolated from control and DOX-treated triply transgenic mice. This DNA was amplified with two primer pairs, one detecting the intact HSI region and the second detecting an adjacent region outside the floxed segment (Fig. 3B, HSI and p2 amplimers, respectively). The intensity of the HSI-specific amplification product was normalized to that of the p2 amplimer. The results showed that the average intensity of the PCR product at HSI in pituitary DNA from the DOX-treated mouse decreased to 76% of that of the sucrose control. Previous published results have shown that 31% of primary pituitary cells are mGH⁺ somatotropes (26). Thus, the 24% decreased in the HSI region in pituitary DNA of DOX-treated mice is consistent with the conclusion that HSI deletion occurred in a fraction of somatotropes.

Transcriptional activity of the hGH-N transgene is linked to its subnuclear localization.

Studies in a number of model systems link the subnuclear localization of specific loci with their transcriptional state (6, 8, 30–34). In particular, defined positive regulatory elements have been demonstrated to contribute to transcriptional activity by excluding loci from repressive regions adjacent to the nuclear membrane (7–9, 35). Based on these studies, we asked whether the decay of hGH-N expression in the adult pituitary following HSI inactivation altered the subnuclear positioning of the hGH transgene locus relative to the nuclear periphery.

To establish a baseline for this study, we determined the nuclear positioning of the active hGH/P1(FloxHSI) transgene. The nucleus was segmented into five concentric zones, zone 1 being the outermost and zone 5 the innermost (see Materials and Methods) (Fig. 4A). The position of the transgene locus was determined by 3D fluorescent in situ hybridization (immuno-FISH or 3D FISH) (Fig. 4B). The nuclear position of the intact hGH/P1(FloxHSI) locus was determined in somatotropes (mGH⁺ cells) and nonsomatotropes (mGH⁻ cells) from the same pituitary cell preparation. In nonsomatotropes, 76% of the hGH loci were positioned within the outermost zone (zone 1). In somatotropes, there was a significantly lower frequency of localization of the hGH-N locus to the outer zone (52%) (Fig. 4C and D). A reciprocal relationship was observed in zone 2, where the percentage of hGH-N loci was significantly higher in somatotrope lineage cells than in the nonsomatotrope (mGH⁻) population (34% versus 22%, respectively) (Fig. 4C and D). These data were consistent with other model systems in demonstrating that the hGH locus in expressing cells (somatotropes) is preferentially positioned farther from the nuclear periphery than the same locus in nonexpressing (nonsomatotrope) cells.

Fig 4.

Transcriptional activity of the hGH locus is associated with its subnuclear localization. (A) Immuno-FISH was performed on single-cell suspensions from adult transgenic mouse pituitaries, and the positions of the transgene locus were measured relative to the nuclear periphery. Each cell nucleus, identified by DAPI staining, was divided into five circular zones based on the relative distance to the nuclear periphery as illustrated. The position of the signal representing the hGH locus was assigned to a zone in each nucleus based on a three-dimensional analysis (3D FISH). (B) A representative single image of a 3D FISH study. The frame at the left displays the identification of a somatotrope using an antibody to mGH (red). The frame on the right displays immuno-FISH detection of the unique hGH transgene locus using the hGH/P1 DNA as a probe (green). The nuclei were stained with DAPI (blue). The arrow indicates the somatotrope in this field. The bars represent 5 μm. The measurements were performed as described in Materials and Methods. The z stacks were collected at a distance of 0.25 μm apart. (C) Distributions of the hGH/P1(FloxHSI) transgene loci within the nuclei of somatotropes or nonsomatotropes. The data were collected from somatotrope (mGH⁺) and nonsomatotrope (mGH⁻) cells on the same slide. A minimum of 70 cells was studied in each group. The plot compares the position of the transgene locus (zones 1 to 5) relative to the percentage of loci at that position. Each cell studied contained a single transgene locus. (D) Compilation of data on the localization of the hGH/P1(FloxHSI) transgene loci in zones 1 and 2 of somatotrope and nonsomatotrope cells monitored on the same slide. The significance of the indicated comparisons is shown above the respective bars. The experiments were performed at least five times, and at least 90 cells were analyzed in each individual study (**, P < 0.01; *, P < 0.05).

Inactivation of HSI in the adult pituitary fails to relocalize the hGH locus to the repressive nuclear periphery.

We next determined the positioning of the hGH locus carrying the germ line deletion of the HSI Pit-1 array [hGH/P1(loxpΔHSI)] compared to the intact locus [hGH(FloxHSI)] (Fig. 5). From the 3D FISH analysis, 68% of the hGH/P1(loxpΔHSI) loci were located in the most peripheral zone (zone 1) of the somatotrope nuclei; the remaining 32% were located in the more central zones (Fig. 5A and B). This distribution was indistinguishable from that of the intact hGH locus [hGH/P1(FloxHSI)] in nonsomatotropes (compare Fig. 4C and 5A) and showed a significant shift toward the periphery compared to the intact locus in somatotropes (Fig. 5A and B). These data support a linkage between transcriptional activity of the hGH-N transgene and its nuclear positioning. They further demonstrate that the positioning of the locus distal to the nuclear periphery is dependent on the HSI Pit-1 array.

Fig 5.

The distribution of the hGH transgene in somatotrope nuclei fails to shift following deletion of the HSI Pit-1 array in the active adult pituitary. (A) Positioning of the intact and HSI-inactivated hGH transgene loci in somatotropes. 3D FISH was performed on somatotropes isolated from hGH/P1(FloxHSI) and hGH/P1(loxpΔHSI) mice. Three mice from each of the two groups were used for the study, and a minimum of 70 cells was analyzed by 3D FISH in each category. The data were collected and displayed as in Fig. 4. (B) Compilation of data on the localization of the two transgenes in zones 1 and 2, as shown in panel A. The significance of the indicated comparisons is shown above the respective bars. The experiments were performed at least five times, and at least 90 cells were analyzed each time (**, P < 0.01; *, P < 0.05). (C) Compilation of data on the localization of the hGH/P1(FloxHSI) transgene loci in triply transgenic mice (as described for Fig. 2A) treated with DOX or with sucrose alone for 14 days. The positioning of the transgene locus was determined by 3D FISH, and the data were collected and displayed as in Fig. 4. (D) Compilation of data on the localization of the transgene loci in zones 1 and 2 as shown in panel C. The comparisons failed to reveal significant shift after inactivation of HSI in the adult pituitary.

With these studies in place, we proceeded to determine the impact of conditional deletion of HSI in the adult pituitary on positioning of the hGH transgene locus (Fig. 5C and D). Surprisingly, the study failed to reveal a significant shift in the positioning of the hGH locus subsequent to DOX treatment. The distributions of the hGH loci in the DOX-treated group and in the sucrose-only group were essentially identical. These data indicated that inactivation of HSI in the adult pituitary silences hGH-N expression but fails to reposition the locus to a more peripheral zone as is characteristic of the germ line inactivated locus.

Inactivation of HSI in the adult pituitary disengages the hGH transgene locus from Pol II factories.

RNA polymerase II (Pol II) can be detected as distinct accumulations (transcription factories) within the nucleus using indirect immunostaining approaches (6). Studies in a number of systems have identified a positive correlation between the transcriptional activity of a locus and its close association with Pol II factories (6, 8). This relationship was explored in the context of hGH-N expression by 3D FISH. The Pol II factories were detected using antibody specific for the serine-5-phosphorylated C-terminal domain (CTD) of RNA Pol II. Prior studies have revealed that this Pol II isoform is associated with the initiation of transcriptional elongation (36). Therefore, the distinct foci enriched for phosphorylated serine-5 Pol II are considered engaged transcription factories (8). The active hGH/P1(FloxHSI) transgene (triply transgenic mouse fed sucrose water only) was positioned in close proximity to a Pol II factory in 55% of the somatotropes (GH⁺ cells) (Fig. 6). This compared with a 2% frequency for the same locus in nonexpressing cells (nonsomatotropes) in the same sample of disaggregated pituitary cells (Fig. 6). The association with the Pol II factories in somatotropes of the germ line HSI deletion was 14% (Fig. 6A), consistent with the low expression level of hGH-N in these mice (Fig. 1C and D). Remarkably, the DOX-induced deletion of the Pit-1 array from the hGH/P1(FloxHSI) transgene in the adult pituitary resulted in a substantial decrease in the association with Pol II factories (22%) (Fig. 6A). The magnitude of this decrease (from 55% to 22%) approximates the 60% decrease in hGH-N mRNA expression in the pituitary subsequent to deletion of the Pit-1 (Fig. 2 and 6A). We conclude from these data that there is a tight correlation between the transcriptional activity of the hGH-N transgene and its proximity to Pol II factories. This relationship is maintained in the adult pituitary by HSI activity.

Fig 6.

HSI deletion from the hGH/P1(FloxHSI) transgene in the adult pituitary disengages the hGH transgene locus from its close proximity to RNA Pol II transcription factories. (A) Three-dimensional FISH was performed on pituitary cells of triply transgenic mice treated with sucrose alone (sucrose control) or sucrose with DOX. RNA Pol II was detected using an antibody specific to the C-terminal domain of the Ser5-phosphorylated isoform of Pol II. The detection of the mGH protein and the hGH transgene locus was as described for Fig. 4. The histogram shows the percentage of each of the hGH transgene loci that were associated with Pol II transcription factories in each setting, as indicated on the x axis. (B) A representative 3D immuno-FISH analysis of the nucleus stained for the hGH-N transgene locus (green) and for Pol II transcription factories (red). In each case, the identity of the cell as somatotrope was established by positive mGH staining of the cytoplasm (as in Fig. 4B; data not shown here). The left images show the detection of the Pol II (red) inside the DAPI-stained nuclei, and the right images show the costaining for PolII (red) and hGH (green). The top right image (with expanded window) shows a representative analysis of a somatotrope containing the intact hGH/P1(FloxHSI) transgene with close juxtaposition between hGH-N and Pol II factories. The bottom right image shows a nonsomatotrope from the same preparation with separation between the hGH-N locus and the Pol II factories. The bars in all frames represent 1 μm.

Inactivation of HSI in the adult pituitary results in selective deacetylation at the hGH-N promoter and the adjacent 5′ flanking region.

HSI plays an essential role in establishing a broad domain of histone acetylation at the hGH locus in the developing pituitary. This 32-kb domain encompasses the hGH LCR and extends to the hGH-N promoter (14). Germ line deletion of the HSI Pit-1 array effectively eliminates this modification, with a commensurate absence of hGH-N expression (15). To characterize mechanisms of HSI action in the maintenance of hGH-N expression in the adult pituitary, we assessed histone H3 and H4 acetylation subsequent to the conditional deletion of the critical Pit-1 array (Fig. 7). Histone acetylation was monitored by ChIP analyses at six sites spanning the LCR and at the hGH-N promoter. As expected, these sites in the triply transgenic mice treated with sucrose water (control animals) were all enriched for histone H3 and H4 acetylation (Fig. 7). Inactivation of HSI in adult somatotropes subsequent to DOX treatment of the triply transgenic mice resulted in a significant loss of histone H3 and H4 acetylation at the hGH-N promoter and its adjacent 5′ flanking region (probe p1) (Fig. 7). This loss correlates with loss of hGH-N expression. Surprisingly, there was no significant change in the levels of H3 or H4 acetylation at the sites monitored within the LCR (Fig. 7). These data demonstrated that histone acetylation within the hGH LCR, which is established by an HSI-dependent pathway during pituitary development, is maintained in the adult by an HSI-independent pathway. In contrast, acetylation at the promoter and adjacent 5′ flanking region remains HSI dependent.

Inactivation of HSI in the adult pituitary results in loss of long-range interaction between the hGH LCR and the hGH-N promoter.

We next investigated the effect of conditional HSI deletion on the higher-order chromatin configuration at the hGH locus. Our previous studies have demonstrated that HSI,II of the LCR is positioned in close proximity to the hGH-N promoter in the active hGH locus by looping and that the establishment of this configuration is dependent on the actions of HSI (18). For example, this interaction fails to be established in the pituitary of a transgenic mouse carrying a germ line HSI deletion (18).

The impact of the conditional HSI deletion on the higher-order configuration of the active hGH-N locus was explored by chromosome conformation capture (3C) (37). As in our prior 3C studies, we used the restriction enzyme BglII to segment the hGH locus and isolate the HSI,II region on a relatively small (1.6-kb) restriction fragment. A primer specific for the hGH-N promoter was used as the anchor primer for the PCR amplification of the ligation products (Fig. 8). Using this approach, the chromatin conformations of the hGH loci in primary pituitary cells from the DOX-treated and sucrose control triply transgenic mice were compared in two independent experiments (Fig. 8). The signal of the PCR products from the ligation were normalized to a random ligation control [ligation of BglII-digested hGH/P1(FloxHSI) plasmid DNA] and to the ligation frequency between fragments generated from the endogenous ercc3 locus (18). A high ligation frequency between HSI,II and the hGH-N promoter was detected in the pituitary chromatin of the sucrose control mice (Fig. 8). This result is consistent with our previous finding that the hGH-N promoter is proximal to HSI,II in adult pituitary (18). In contrast, the 3C results from the pituitary of the DOX-treated mice showed that the ligation frequency of the hGH-N promoter fragment to each of the 5′ restriction fragments increased as an inverse function of the distance between the fragments and specifically lacks enrichment for ligation products between the HSI,II region and the hGH-N promoter (Fig. 8; note that the primer in the HSI,II fragment is positioned outside the floxed segment). Of particular note, the frequency of ligation between the hGH-N promoter fragment and the HSI,II fragment decreased 6-fold (from 1.2 to 0.2) in the DOX-treated mice in comparison with sucrose-treated controls (Fig. 8). This set of 3C analyses leads us to conclude that the long-range looping interaction between the hGH LCR and the remote hGH-N promoter is dependent on the continued action(s) of the HSI LCR determinant in the adult.

Fig 8.

Conditional deletion of HSI from the active hGH locus in the adult pituitary abolishes the long-range interaction (looping) between hGH LCR and the hGH-N promoter. (A) BglII map and amplimer positions for 3C analysis. In these studies, the fragment encompassing the hGH-N promoter was used as the anchor site and the four upstream primers were positioned to detect the BglII fragments that encompass the HSV, HSII, CD79b, and p1 regions. (B) HSI-dependent association between the HSI,II region and hGH-N. Levels of chromatin 3C ligation products were assessed by semiquantitative PCR. The data were normalized to parallel analyses of the endogenous ercc3 locus (see panel C). Each histogram bar represents the average of two independent assays of the indicated chromatin samples, control pituitary chromatin and DOX-treated pituitary chromatin. The panels on the right of the histogram show representative ethidium bromide-stained agarose gels containing the indicated PCR amplification products. Lane C shows a PCR analysis of ligation products of BglII-digested hGH/P1 plasmid DNA and represents random ligation of equimolar quantities of each of the BglII fragments in the locus under the conditions of the assay. In order to measure the intensities of the PCR product, the agarose gel was stained with SYBR gold (Invitrogen) and scanned using a Typhoon phosphorimager (GE). The ligation efficiency, shown on the y axis, was calculated using the equation (signal_pituitary/signal_{random control})/signal_{ercc3 ligation}. The average of the two biologically replicated experiments is shown on the histogram (asterisks represent the ligation efficiency of each experiment). (C) 3C analysis of the ercc3 locus (internal control). The indicated head-to-head ligation frequency of two adjacent BglII fragments released from the ubiquitously expressed ercc3 locus during the 3C analysis was determined and used as an internal control for each chromatin conformation capture assay.

DISCUSSION

HSI of the hGH LCR is both essential and sufficient for activation of hGH-N gene expression in the course of mouse embryonic development (15, 19). In vivo functional mapping of HSI pinpointed a critical array of three Pit-1 binding sites (19). Our previous studies demonstrated that this Pit-1 array is essential for the formation and activation of HSI. HSI itself is a multifaceted determinant, necessary for structural modification and activation of the hGH locus during somatotrope differentiation (15, 17, 18).

While it is clear that HSI plays an essential role in the activation of hGH-N during embryonic development (19), its role in the subsequent maintenance of hGH-N expression throughout adult life remained unexplored. The data presented here support a model in which the maintenance of hGH-N expression is dependent on continuous actions by the HSI Pit-1 array. Deletion of this array in the adult pituitary results in a decrease in steady-state hGH-N mRNA (Fig. 2). Analyses of individual pituitary cells from targeted mice suggest that the deletion of the HSI Pit-1 array from the active adult locus essentially ablates hGH-N expression (Fig. 3). Residual expression of hGH-N in these mice most likely reflects incomplete Cre-mediated deletion within the somatotrope population. We conclude from these data that the impact of HSI inactivation on hGH-N expression is of equal and substantial magnitude whether it occurs in the germ line or in the context of the adult pituitary.

The mechanisms underlying the maintenance phase of hGH-N expression were subsequently explored. In the adult pituitary, HSI is critical to the stable engagement of the hGH locus with Pol II factories (Fig. 6), to maintenance of the looping between the LCR and the hGH-N promoter (Fig. 8), and to maintenance of acetylation at the hGH-N promoter and adjacent 5′ flanking region (Fig. 7). In contrast, the acetylation of the LCR chromatin domain and the positioning of the locus distal to the nuclear periphery, two functions that are dependent on HSI during initial locus activation, are converted to HSI independence in the adult pituitary. We conclude from these findings that the active contribution of HSI to the maintenance of hGH-N reflects a subset of the full range of HSI functions critical to the initial phase of gene activation.

The finding that the Pit-1 array at HSI is critical for both initial activation and subsequent maintenance of hGH-N expression contrasts with the controls over the Pit-1 gene itself. Pit-1 is a key determinant in the differentiation of somatotropes, lactotropes, and thyrotropes in the pituitary (38). In mice, Pit-1 is activated within Rathke's pouch, the precursor of the anterior pituitary, at embryonic day 13.5 (e13.5), 2 days prior to induction of mGH. This initial induction of Pit-1 transcription is mediated by an early enhancer located 5.9 kb upstream of the Pit-1 promoter (39). In contrast, the subsequent maintenance phase of Pit-1 gene expression is under the control of a distinct late enhancer located 10 kb upstream of the Pit-1 gene (40, 41). The continued input of the early enhancer is not essential to the stable maintenance of Pit-1 expression in adult life (39, 40). Of central importance to this switch between the early and late enhancers is the finding that the activity of the late enhance is dependent on the actions of Pit-1 itself (40). Thus, the stable maintenance phase of Pit-1 expression represents a self-sustaining “feed-forward” pathway. This sort of pathway may represent a durable memory mechanism and that may be particularly suited for genes encoding transcription factors.

The concept of self-sustaining memory modules that support long-term stable expression has been most extensively studied in the context of the homeotic genes in Drosophila melanogaster (22, 24). In this system, transiently expressed regulators that are themselves induced by developmental signaling factors activate gene expression. The subsequent maintenance of homeotic expression patterns is mediated by the polycomb group (PcG) and trithorax group (trxG) complexes (22) that bind to a distinct set of cellular “memory modules” and mediate posttranslational modifications of the core histones at the target genes (42). As in the case of the Pit-1 locus, specific determinants appear to be utilized to maintain durable expression of critical genes subsequent to, and independent of, the transcriptional interactions involved in the initiation of gene activity.

Covalent and noncovalent modifications of chromatin structures are important in the regulation of gene expression, and these activities appear to be accompanied by relocation of genes within nuclear subcompartments (43). A linkage between LCR function and positioning of a target locus relative to the nuclear periphery (6, 8, 30) was initially reported for the intensively studied mouse β-globin gene. On the basis of multiple comparisons of subnuclear localization and transcriptional activity of β-globin in erythroid cells at different developmental stages, Ragoczy and coworkers concluded that the relocation of β-globin away from the nuclear membrane is a consequence of its transcription (8). This conclusion appears to contrast with our current report. Our observations reveal that the maintenance of the hGH locus at a distance from the nuclear periphery is sustained following HSI inactivation in the adult and is independent of the concomitant dissociation of the locus from Pol II factories and the loss of transcription. These results suggest that the repositioning of the hGH locus away from nuclear periphery in somatotropes is not a readily reversible characteristic defined by its transcriptional activity. Although the relationship of the hGH locus to Pol II factories appears to be direct, the current data suggest that control of its proximity to the nuclear periphery may be more complex.

In general, gene silencing correlates positioning of the locus proximal to the nuclear periphery (4). Of particular note, we observed that 48% of the hGH loci are peripherally located in somatotropes (Fig. 4A), while hGH protein is detected in 86% of the somatotropes of a transgenic mouse carrying the intact hGH locus (Fig. 3A). These results suggest that a fraction of these peripherally located hGH loci are actively transcribed in somatotropes in the presence of HSI. These observations suggest that the major impact of HSI on hGH-N expression is to maintain the active chromatin structure at the hGH-N promoter and guide the promoter to Pol II factories regardless of the subnuclear localization. Consistent with these data are RNA FISH analyses which reveal that a fraction of peripheral β-globin loci are actively transcribed (8).

The critical role of HSI in establishing an extensive domain of histone acetylation at the hGH locus during embryonic activation of the hGH-N gene appears to be partially lost during the maintenance phase of hGH-N expression (see model in Fig. 9). Mice with germ line inactivation of HSI have a global absence of histone acetylation throughout the hGH locus, extending from the LCR to the hGH-N promoter. In contrast, inactivation of HSI in the adult pituitary, while silencing hGH-N, selectively decreases histone acetylation at the hGH-N promoter. In contrast, acetylation is fully maintained within the LCR itself subsequent to the conditional deletion of HSI in the adult pituitary. These data suggest that the maintenance of histone acetylation within the LCR is supported in the adult by a pathway that has switched from HSI dependent to HSI independent and is functionally unlinked to the target hGH-N promoter. The observed loss of looping between the LCR and the hGH-N promoter subsequent to conditional deletion of HSI in the adult pituitary fits with this model. The LCR, while remaining in an active configuration (i.e., acetylated), is no longer juxtaposed to the target promoter.

Fig 9.

Inactivation of HSI in the adult pituitary silences hGH-N expression and triggers a unique reconfiguration of the hGH chromatin locus. (A) Configuration of the wild-type (WT) hGH transgene chromatin locus in pituitary somatotropes. HSI of the hGH LCR establishes an active chromatin domain (hyperacetylation and looping; heavy orange line). This activation excludes the locus from the repressive environment adjacent to the nuclear membrane (gray oval), places it in close proximity to Pol II factories (aqua oval), and correlates with robust transcription from the hGH-N promoter (heavy arrow). The five genes in the hGH cluster are indicated by rectangles (gene substructure is not shown), with the hGH-N gene highlighted in red and the four placentally expressed genes in gray. The gold stars represent the HSs of the hGH LCR in somatotrope chromatin. Pituitary-specific HSI is indicated by the thin arrow. (B) HSI is essential to establishing the epigenetic modifications and subnuclear positioning linked to transcriptional activation of the hGH locus during somatotrope differentiation. If HSI is inactive early in development (germ line ΔHSI), the hGH locus fails to be acetylated and is positioned close to the nuclear membrane. This transgene lacks significant engagement with Pol II factories (data not shown), the looping between HSI and the hGH-N promoter fails to assemble, and transcription is limited to trace levels. (C) Conditional deletion of the HSI Pit-1 array from a fully active hGH locus in the adult pituitary silences hGH-N transcription and reconfigures the locus in a unique manner. This conditional deletion results in selective loss of histone acetylation at the hGH-N promoter, disengagement of the locus from Pol II factories, and a corresponding loss of gene transcription. Remarkably, this inactivated locus retains the active positioning away from the nuclear membrane and retains full levels of histone acetylation within the LCR. Taken together, these data demonstrate that HSI is critical to the initial activation of the hGH locus in the embryonic pituitary and to its subsequent robust maintenance in the adult. Remarkably, a subset of the HSI-dependent alterations at the hGH locus that track with its developmental activation are converted to an HSI-independent state once the maintenance phase has been established.

The findings of the current study, along with our prior studies, indicate that the function of the Pit-1 array in HSI plays an essential role both in the initiation of hGH-N expression in the embryo and in its durable maintenance in the adult (Fig. 9). The conditional deletion of the Pit-1 array is shown to eliminate the long-range interaction between the LCR and the hGH-N promoter and the association with Pol II factories while leaving intact its roles in LCR acetylation and locus subnuclear localization. Thus, the last two attributes appear to be sustained in the adult by HSI-independent pathways. These observations demonstrate that developmental initiation of gene expression and its subsequent robust maintenance throughout adult life share a dependence on a common long-range enhancer, while the functions of this enhancer at these two developmental periods are linked to distinct patterns of chromatin structure and locus positioning within the somatotrope nucleus. The generality of these findings to other mammalian loci can be considered in the light of these findings.