Report

Developmentally Regulated Activation of a SINE B2 Repeat as a Domain Boundary in Organogenesis

See allHide authors and affiliations

Science  13 Jul 2007:
Vol. 317, Issue 5835, pp. 248-251
DOI: 10.1126/science.1140871

Abstract

The temporal and spatial regulation of gene expression in mammalian development is linked to the establishment of functional chromatin domains. Here, we report that tissue-specific transcription of a retrotransposon repeat in the murine growth hormone locus is required for gene activation. This repeat serves as a boundary to block the influence of repressive chromatin modifications. The repeat element is able to generate short, overlapping Pol II–and Pol III–driven transcripts, both of which are necessary and sufficient to enable a restructuring of the regulated locus into nuclear compartments. These data suggest that transcription of interspersed repetitive sequences may represent a developmental strategy for the establishment of functionally distinct domains within the mammalian genome to control gene activation.

The growth hormone (GH) gene provides a well-studied transcription unit that is highly suited for defining how specific chromatin modifications (16) might be responsible for the spatial and temporal order of lineage specification events in the developing pituitary gland. The human GH locus is represented by a cluster of five GH-related genes that are regulated by a Pit-1– dependent locus control region (LCR) (710). The murine GH genomic locus is found on mouse chromosome 11 and encompasses five genes (Fig. 1A). In contrast to the human locus, the mouse locus does not contain tandem duplications of the GH gene, and there is no known murine LCR.

Fig. 1.

Developmental repositioning of the GH locus. (A) Schematic diagram of GH genomic locus murine chromosome 11, showing the five transcription units in the locus. Arrows indicate the direction of each transcription unit. Red bars represent PCR amplification regions in ChIP experiments. The regions later examined for function are labeled in letters (e.g., G-F). (B) ChIP analysis of isolated pituitary using antibodies specific for branched trimethyl K9H3 (Br-triMeK9H3), and diMeK9H3. Multistep ChIPs with αPit-1 then αN-CoR, or αPit-1 then αPol II for initial selection of Pit-1 lineage cells. (C) Immuno-FISH analysis of e12.5, e14.5, and e17.5 nuclei from murine pituitaries using a GH probe. (D and E) Conventional ChIP analysis across the GH locus using α-BrtriMe H3-K9 or α-diMe H3-K9 IgG on e12.5 and e14.5 murine pituitaries. PCR primers diagrammed in (A).

The GH gene is initially silenced at early stages of murine pituitary gland development but activated at embryonic stage 17.5 (e17.5) by the POU-homeodomain factor, Pit-1, in GH-producing cells (somatotropes) (fig. S1A). It remains actively repressed, or silenced, in the other pituitary cell types. Analysis of histone modifications was used to assess the chromatin state of the GH locus during development. Chromatin immunoprecipitation (ChIP) analysis (5, 1116) revealed high levels of triK9 methylation of histone H3, indicating a condensed heterochromatic state of the GH gene promoter, at e12.5, with a gradual decline to low levels by e14.5 and complete disappearance by e17.5 or in adult pituitary (Fig. 1B). Conversely, low levels of dimethylation of K9H3 on GH were observed at e12.5, but increased by e13.5 to e14.5, declined at 17.5, and disappeared in the adult gland. Based on complete loss of the condensed heterochromatin marks, we suggest that these chromatin events occurred and were sustained in all lineages. In dw/dw mice, containing an inactivating point mutation in the Pit-1 POU domain factor, the GH promoter lacked triMe K9H3 but contained elevated diMe K9H3, which suggests that these events occur in a Pit-1–independent manner (Fig. 1B). Two-step ChIP was performed using αPit and either αPol II or αNCoR immunoglobulin G (IgG). Together, the analysis of transcriptionally active or repressed Pit-1 bearing chromatin suggests that partitioning of these histone modifications occurs uniformly in the pituitary cell population in early pituitary development.

Fluorescence in situ hybridization (FISH) analysis was used to localize the GH gene within interphase nuclei at different stages of pituitary embryonic development (Fig. 1C) (12, 16). Both alleles of the GH gene localized to 4′,6′-diamidino-2-phenylindole (DAPI)–stained heterochromatin in virtually all of the clearly stained (>90%) e12.5 nuclei (Fig. 1C and fig. S1B). In contrast, at e14.5, of the total population of clearly stained nuclei (102 out of 109 nuclei), the GH gene had relocated to euchromatic territories outside the DAPI regions (Fig. 1C and fig. S1B). This is temporally coincident with the observed exchange of branched triMe K9H3 or diMe K9H3 methylation marks on the GH promoter (Fig. 1B). ChIP analysis (17) across a 30-kb genomic interval revealed a specific region within the GH locus (between primer pairs 4 and 6) at which the exchange of branched triMe K9H3 or diMe K9H3 occurs temporally between e12.5 and e14.5 in pituitary development (Fig. 1, D and E, and fig. S1, C and D).

Because the region of diMe K9H3 and triMe K9H3 transition is located –14 to –10 kb upstream of the GH transcriptional start site (Fig. 1, A, D, and E), weconsidered that a putative boundary element(s) might exist within the GH locus to establish these regulatory domains. We used an assay to locate DNA sequences that might possess putative insulator properties by assessing the enhancer-blocking activity of each DNA sequence. We placed the experimental boundary element between a known enhancer and a reference promoter driving a reporter gene (18). This strategy has been used for a number of known insulators (1921). The enhancer-blocking assay (EBA) was carried out with linearized constructs in transiently transfected HEK 293 cells (Fig. 2A). As controls, a 1.2-kb DNA fragment containing the 5′HS4 insulator of the chicken β-globin locus (22), and the FII/FIII core elements of this 5′HS4 boundary [wild-type (F8) and mutant (F6) forms] (18, 21) were used. Each element (Fig. 1A) was cloned between the cytomegalovirus (CMV) enhancer and minimal CMV promoter (XhoI site), or at the 5′ end of the CMV enhancer (PstI site), with enhancer-blocking occurring only when the experimental element is located between the enhancer and the promoter (Fig. 2A) (18, 19).

Fig. 2.

Identification of a SINE B2 repeat in the GH boundary region. (A) Enhancer-blocking assay performed using the pE Luc (CMVEmPLuc) vector in human embryonic kidney (HEK) 293 cells, placing the tested element either upstream (PstI) or downstream (Xhol) of the enhancer (map of the construct is given in fig. S2C). (B) Schematic diagram of the GH SINE B2 expression unit showing position of primers used for reverse transcriptase reaction (Y or 2 for Pol II and 3 or 4 for Pol III transcription, respectively) and PCR (Y, 2, 1, 3, 4). Assessment of Pol II and Pol III–generated transcripts from the reporter construct using RT-PCR analysis with primers indicated above. Designations of +1 and TATA indicate the transcription start site and TATA box, respectively. [Poly A refers to a stretch of A residues in the sequence (fig. S2A)]. (C) Effect of deletion of the SINE B2 element on function of the CD GH element in the enhancer-blocking assay. All data are ±SEM. (D) RFP expression in transgenic founder mice with wildtype mGH/RFP BAC (founders 1,2,3) or with deletion of the SINE B2 element (founders 4,5,6).

The maximal (by a factor of 54.7) reduction in activity was observed in the case of the CD region or any construct that included the CD element, and this region coincided with the region of the transition between developmentally regulated diMe K4H3 and the branched triMe K9H3 (Fig. 1F), conceptually consistent with the description of boundary activity. Further examination of the CD region revealed the presence of a short interspersed nuclear element (SINE) B2 retrotransposon repeat (Figs. 1A and 2B), derived from the tRNA gene, consisting of a 190-bp consensus sequence, including an RNA polymerase III promoter with its two conserved regulatory elements, A-box and B-box (23, 24). Recently, it was reported that SINE B2 expression can be driven in opposite strands by highly conserved Pol III or Pol II promoters (25, 26).

Assessment of the presence of the Pol II–and Pol III–generated transcripts from the GH SINE B2 reporter construct was performed using strand-specific RT-PCR analysis with sequence-specific primers used for extension to generate single-stranded DNA for PCR amplification on RNA samples recovered from transfected HeLa cells (Fig. 2B). SINE B2 was efficiently transcribed (Fig. 2B) from both Pol II and Pol III promoters independent of its insertion site with respect to upstream (PstI) or downstream (XhoI) (Fig. 2B) of the enhancer, and deletion of the SINE B2 repeat (ΔSine B2 CD) abolished the enhancer-blocking activity (Fig. 2C).

To confirm a functional role for this putative boundary element in vivo, a 220-kb bacterial artificial chromosome (BAC) encompassing a centrally located mGH gene was modified to replace the GH coding sequence with a red fluorescent protein (RFP) dimer coding sequence (fig. S3, A and B). This transgene was clearly expressed in three transgenic founder mice generated, recapitulating the physiological temporal and spatial pattern of GH locus expression (5) (Fig. 2D). Deletion of the 160-bp region encompassing the mGH SINE B2 repeat in this BAC caused either complete loss of RFP expression (two transgenic founder mice) or loss of RFP expression in all but a few residual cells (one transgenic founder mouse) carrying this BAC transgene (Fig. 2D). Therefore, the SINE B2 transcription unit can be suggested to be required for developmentally regulated GH gene activation in vivo. However, this assay does not establish that this reflects its boundary function in vivo.

RT-PCR analysis of RNA recovered from developing pituitary at embryonic day e14.5 reveals that both Pol II and Pol III transcripts were present in vivo (Fig. 3, A to E) (26). Pol III transcripts originating in the SINE B2 repeat were detected in the pool of the pituitary-specific RNA at all the times during embryonic development, including e12.5 (Fig. 3, D and E), but the transcript generated from the Pol II promoter (26) initially appeared only at e13.5 to 14.5, because it was not present in the pooled e9.5 to e12.5 sample (Fig. 3C) but was present in the pooled e12.5 to e15.5 sample (Fig. 3C) in the pituitary gland (Fig. 3, B and C), correlating both with the timing and establishment of differential domains of chromatin modifications within the GH locus and nuclear repositioning detected by FISH (Fig. 1, C and E).

Fig. 3.

In vivo expression of the GH SINE B2 transcripts during pituitary development. (A) Schematic diagram of primers used for reverse transcription, PCR, and primer extension analysis of the SINE B2 transcription units. Primers used for reverse transcription and ssDNA are shown in yellow, their PCR partnering primers and expecting PCR fragments are shown in black. (C72 bp = a 72-bp sequence conserved in human and mouse). (B and C) Identification of Pol II transcript by RT-PCR at e14.5, but not detected on e12.5, with actin-specific primers used as a positive control (see Actin lane). GH promoter–specific primers used as a negative control. (D) Demonstration of the Pol III transcript on e14.5 by RT-PCR using primers shown. (E) Ontogeny of expression of the Pol III transcripts, showing expression throughout pituitary development. The same RNA samples were used for RT-PCR as in (B) to (D).

We next inserted the SINE B2 from the GH locus into a 1.2-kb coding sequence from the Adam11 transcription unit that alone exhibited no activity in the enhancer-blocking assay. We evaluated deletion and substitution constructs in the enhancer-blocking assay (Fig. 4A) and found that insertion of SINE B2 transcription unit containing the minimal defined Pol II promoter (26) in Adam11 constructs resulted in clear enhancer-blocking activity (ctl+SINE B2) (Fig. 4A). However, insertion of a shorter version of the SINE B2 in which the Pol II promoter was deleted in the same position within Adam11 no longer exerted the enhancer-blocking activity. Substitution of nucleotide sequences representing the intragenic promoters for Poll III–mediated transcription (23) diminished enhancer-blocking activity of element CD (Fig. 4A). These results revealed that a bidirectional Pol III/Pol II-driven transcription of the GH SINE B2 repeat was both necessary and sufficient to establish boundary activity, independently of the surrounding sequences. Ongoing transcription of the SINE B2 repeat was required to mediate enhancer-blocking activity, rather than generation of SINE B2 transcripts themselves in trans (fig. S4).

Fig. 4.

Determinants of enhancer-blocking function of the GH SINE B2 transcription unit. (A) The enhancer-blocking assay was performed by inserting a minimal amount of SINE B2 transcription unit into 1.2 kb of Adam 11 coding region in the pE Luc vector. Mutations included removal of the Pol II promoter (Δ Pol II), mutation of Box A (mut Box A) or mutation of Box B (mut Box B). (B) Model of regulation of a GH boundary control element based on developmentally regulated expression of the Pol II transcription unit of the GH SINE B2 transcription unit, correlated with chromosomal repositioning. AcK14H3 is acetylated K14 on histone H3.

Together these data suggest that bidirectional, noncoding transcription from the murine GH SINE B2 transcription unit has the ability to contribute to establishing a putative boundary element by imposing a local perturbation in chromatin structure resulting in repositioning of the GH locus from a heterochromatic region to a more permissive euchromatic environment (Fig. 4B). Indeed, results obtained in fission yeast demonstrate a very similar organization of the boundary elements encompassing inverted repeats (27) and development changes in expression of SINE B2 reported for different systems (28), consistent with widespread use of SINE B2 elements in mammalian development. Our findings further support the notion that transcribed insulators/boundary elements may function as a dynamic infrastructure adapted to the transcriptional and/or developmental state of the cell, providing the required plasticity to respond to the developmental and environmental cues.

Supporting Online Material

www.sciencemag.org/cgi/content/full/317/5835/248/DC1

Materials and Methods

Figs. S1 to S4

Tables S1 and S2

References and Notes

View Abstract

Navigate This Article