Chromatin status and transcription factor binding to gonadotropin promoters in gonadotrope cell lines

To establish how chromatin status on key gonadotrope genes changes during gonadotrope cell maturation, we studied three model immortalized mouse gonadotrope cell lines αT1–1, αT3–1, and LβT2, as well as two control cell lines, the mouse thyrotrope cell line TαT1, and the mouse fibroblast cell line, NIH3T3 (ATCC). All cell lines were cultured in DMEM with 4.5% glucose (Mediatech Inc., Herndon, VA), 10% fetal bovine serum (Gemini Bio, West Sacramento, CA), and 1× penicillin-streptomycin (Life Technologies, Inc./Invitrogen, Grand Island, NY) in 5% CO₂ at 37 °C. Cells were seeded on 10 cm dishes (Nunc, Roskilde, Denmark) and harvested at subconfluency. For hormone treatment of LβT2 cells, cells were serum-starved for 16 h in DMEM containing 4.5% glucose, 1× penicillin-streptomycin and 0.1% bovine serum albumin. Cells were then treated for 4 h with 100 ng/ml GnRH (Sigma-Aldrich, St. Louis, MO) ± 25 ng/ml activin (Calbiochem, La Jolla, CA) before they were harvested for the DNase sensitivity assay.

Actively transcribed chromatin is characterized by being in an open conformation, allowing easy access of transcription factors to their binding sites. This open chromatin increases the sensitivity of the chromatin to DNaseI. To establish to what degree the gonadotrope promoters of model gonadotrope cell lines would increase their sensitivity to DNaseI treatment during maturation we performed a DNase sensitivity assay. DNase sensitivity assay was performed as previously described [37]. In brief, αT1–1, αT3–1, LβT2, and NIH3T3 cells were lysed in hypotonic buffer and nuclei were isolated by centrifugation at 2200 g for 5 min at 4 °C. Intact nuclei were resuspended in 1X DNaseI Reaction Buffer (Promega, Madison, WI) containing 2% glycerol. Equal amounts of nuclei were added to increasing quantities of DNaseI (Promega) in 1X DNase Reaction Buffer, ranging from 0 units (U) to 7.5 U, and incubated at 37 °C for 5 min. DNaseI was inactivated using DNaseI Stop Solution (Promega) and incubation at 65 °C for 10 min. Treated nuclei were lysed in Nuclei Lysis Buffer followed by RNase A and Proteinase K digestion. Genomic DNA was then isolated by extraction twice with phenol/chloroform/isoamyl alcohol and once with chloroform. DNA was ethanol precipitated and resuspended in Tris-EDTA buffer followed by 55 °C incubation for one hour. qPCR was performed using SYBR Green supermix and an iQ5 real-time PCR machine (BioRad). Primer sequences are detailed in Table 1. Forty ng of DNA from each treatment condition was quantitated relative to a standard curve of dilutions of undigested DNA. Data from each primer set were normalized to the active gene, Actb, or to the vehicle treatment as stated in the figure legends [37].

To evaluate transcription factor binding to key gonadotrope gene promoters, we performed ChIP assays. ChIP assays were performed as previously described [38]. Chromatin was sonicated to an average length of 300–500 bp using a Branson Sonifier 250 (Branson Ultrasonics Corp., Danbury, CT). Antibodies recognizing specific histone modifications were: anti-acetyl-Histone-H3 (06–599; Millipore, Temecula, CA), anti-trimethyl-Histone H3-Lys4 (07–473; Millipore), anti-dimethyl-Histone H3-Lys9 (ab1220; Abcam), all of which are ChIP-grade antibodies. To recognize phosphorylated polymerase, ChIP-grade anti-RNA polymerase II CTD repeat YSPTSPS (phospho S5) (ab5131; Abcam, Cambridge, MA), was used. Immunoprecipitated DNA and DNA from input chromatin were analyzed for sequences of interest by qRT-PCR using promoter-spanning primers specified in Table 1. For ChIP assays comparing αT1–1, αT3–1, and LβT2 chromatin, the percentage of enrichment of antibody signal over IgG was calculated for each primer set. IgG was the same species and source as the comparison antibody. Then, values for activating chromatin marks were normalized to those for the inactive gene Gnrh1. For repressive chromatin marks, values were normalized to the highly active gene Actb [37‐39].

Antibodies were ChIP grade or previously validated for ChIP assays: anti-LHX3 (L2202, US Biological, MA) [25], anti-PITX1 (sc-18,922X, Santa Cruz, ChIP grade). Immunoprecipitated DNA was analyzed for sequences of interest by qRT-PCR using promoter-specific primers shown in Table 1. For ChIP assays comparing αT1–1, αT3–1, and LβT2 chromatin, the percentage of enrichment of antibody signal over IgG was calculated for each primer set. Values were then normalized to those for the inactive gene Gnrh1 [37, 38].

To confirm that the immortalized pituitary cell lines expressed the key pituitary lineage markers and provide direct quantitation of expression levels, we performed RNA sequencing. We found that all five of the cell lines expressed high levels of β-actin (Fig. 1a, Actb). As expected, the common gonadotrope hormone α subunit, Cga, was expressed in all the gonadotrope lineage cell lines, as well as the thyrotrope cell line, TαT1 [15, 18], but not in NIH3T3 cells (Fig. 1a). The immature (αT3–1) and mature (LβT2) pituitary cell lines both expressed Gnrhr, whereas the gonadotrope marker, Lhb, was only expressed by LβT2 cells. The thyrotrope cell line TαT1 [18] was the only cell line studied expressing the thyrotrope marker Tshb, and did not express any of the gonadotrope-specific genes, other than the common α subunit, Cga (Fig. 1a). We did not detect Fshb in LβT2 cells, although these cells are known to express this mRNA as detected by the more sensitive method of qRT-PCR. Fshb is found at low levels that can be induced by activin treatment [19, 40, 41]. NIH3T3 cells were used as a negative control for pituitary gene expression (Fig. 1a).

To determine if chromatin status correlates with expression of gonadotrope genes, we investigated the chromatin status by DNaseI sensitivity assay [42]. We focused on chromatin accessibility of Cga, Gnrhr, Lhb, and Fshb in the gonadotrope lineage cell lines αT1–1, αT3–1, and LβT2, which represent different stages of gonadotrope development. NIH3T3 cells were used as a negative control. Actb (β-actin) is ubiquitously expressed in all the above cell lines, reflecting an open chromatin status, and was therefore used to normalize the DNaseI data (Fig. 1a and b, Actb). Three classes of chromatin accessibility have been determined by this assay: closed (repressed), poised, and open (active) [43, 44]. One of the characteristics of chromatin is how dynamic its conformation is, and its capacity to compact and unfold. Thus, degrees of compaction and opening of the DNA are possible. Normalizing our data to the highly transcribed Actb allowed us to determine the relative DNaseI sensitivity of the studied gonadotrope genes. To reveal the degree of accessibility of the DNA, we used increasing amounts of DNaseI. As expected, we found that Cga, Lhb, and Gnrhr were much less sensitive to DNaseI treatment than Actb (Fig. 1b), which correlated with gene expression levels (Fig. 1a). Unexpectedly, Fshb was sensitive to DNaseI treatment, although to a lesser extent than Actb (Fig. 1b). However this did not correlate with gene transcription (Fig. 1a), supporting the importance of using more than one approach to study chromatin status to obtain a complete image of chromatin accessibility to transcription factors. Cga is the earliest gonadotrope expressed gene (E11.5 in the mouse embryo) and has high expression in the gonadotrope progenitor αT1–1 cell line (Fig. 1a). Consistent with this, we found an open chromatin structure in the promoter region of Cga in this cell line (Fig. 1b, αT1–1). The chromatin states of all other later chromatin modifications of gonadotrope cell lines were repressed as evidenced by their structures being closed and resistant to DNaseI degradation (Fig. 1b, αT1–1, Fshb, Lhb and Gnrhr). The immature gonadotrope cell-line αT3–1 expresses both Cga and Gnrhr (Fig. 1a). In agreement with this, we determined the chromatin state of these genes to be open (Fig. 1b, αT3–1). Lhb and Fshb are not expressed in the αT3–1 cells and have closed chromatin in this cell (Fig. 1a). Our data show that the chromatin of Lhb and Fshb started to shift from closed to poised in the LβT2 cells (Fig. 1b, LβT2, Two-way ANOVA, followed by Sidaks multiple comparison, p > 0.01 at 7.5 Units of DNaseI). The only partial opening of Lhb and Fshb in LβT2 cells is not surprising, as cells in culture are not in the tridimensional environment they experience in vivo and do not receive hormonal stimulation, as gonadotropes would in vivo, allowing them to fully activate expression of Lhb and Fshb.

Numerous modifications of chromatin and histones allow for successful recruitment of phosphorylated RNA polymerase II (phospho-Pol II), which is required to initiate transcription [10]. To determine if the opening of the chromatin in these cell lines correlates with a change in the acetylation and methylation signatures on the histones binding to the promoters and allows occupation with phospho-Pol II, we next performed ChIP assays. We used antibodies recognizing the activating histone modification marks histone H3 acetylation (H3Ac), and H3K4 trimethylation (H3K4Me3, Fig. 2), as well as recruitment of phospho-Pol II [38, 44]. Although, Cga is highly expressed by all three gonadotrope lineages (Fig. 1a), and the promoter is sensitive to DNaseI treatment (Fig. 1b), the active histone marks in all three cell lines were surprisingly little enriched as compared to Gnrh1, despite recruitment of phospho-Pol II and active transcription (Fig. 1a and 2a). This is unexpected and is possibly due to the chosen region of study of the Cga promoter. In agreement with both transcription levels of Gnrhr, and its promoter’s sensitivity to DNaseI treatment, the Gnrhr regulatory region went from possessing very few active histone marks in αT1–1, to highly enriched in these marks in αT3–1 and LβT2 cells (Fig. 2). In agreement with our findings using the DNaseI sensitivity assay, which found the Fshb promoter to be in a rather closed conformation (Fig. 1b), this promoter exhibits only one of the chromatin marks of an open promoter (H3Ac), but does not bind phosphorylated polymerase II (Fig. 2). Finally, the chromatin status of Lhb is enriched in H3Ac and recruits phosphorylated polymerase II to the regulatory region somewhat in αT3–1 and more strongly in LβT2, which correlated with transcription in LβT2 cells where it also exhibits enrichment of H3K4Me3 (Fig. 1a and 2).

To assess the epigenetic role of repression, we performed ChIP assays to analyze the repressive histone modifications H3K9Me2 and H3K27Me3 [45]. We found that one or both modifications were reduced with developmental maturation on the promoters of Gnrhr and Lhb (Fig. 3). Gnrhr loses both repressive marks in αT3–1 and LβT2 cells, while Lhb has progressively reduced H3K27Me3 from αT1–1 to αT3–1 and finally LβT2 and trends toward lower H3K9Me2 in LβT2 cells (Fig. 3). Whereas no changes were observed on the Fshb promoter (Fig. 3).

Opening of the chromatin, along with activating histone modifications, promotes access to the chromatin of transcription factors. It has previously been shown that the known gonadotrope regulatory proteins PITX2 (Pitx2), SF1 (Nr5r1), and LHX3 (Lhx3) control expression of the mature gonadotrope markers Lhb, Fshb, and Gnrhr [25, 46, 47]. Using RNA sequencing, we show that these transcription factors are expressed in the gonadotrope lineage and thyrotrope cell lines, but not in NIH3T3 cells (Fig. 4a). Appropriately, the gonadotrope-specific transcription factor SF1 is not expressed in thyrotropes (TαT1) or in the gonadotrope progenitor cells (αT1–1). To test whether the gonadotrope-specific regulatory factors, LHX3 and PITX1, are associated with regulatory regions of gonadotrope promoters in any of the studied cell lines, we performed a ChIP analysis using LHX3 and PITX1 antibodies. ChIP analysis revealed that both transcription factors can bind on the gonadotrope terminal target genes (Fig. 4b, c). LHX3 is able to bind the promoter of Cga and Gnrhr, but neither Lhb nor Fshb exhibit binding by this method, at any of the stages of development (Fig. 4b, LHX3). PITX1 binds the gonadotrope-specific promoters Cga and Lhb, but not Gnrhr nor Fshb in this assay (Fig. 4c). This suggests the program of differentiation in these cells is not regulated simply by deacetylated histones [48], but more likely requires a timed balance of transcription factors, co-factors, histone modifications, DNA methylation, and/or networks to control the developmental gonadotrope gene expression program.

To extend this further, we asked if the final differentiation of the gonadotrope lineage cells required GnRH and/or activin treatment for the proper expression of mature gonadotrope genes. We serum starved LβT2 cells for 16 h then treated them for 4 h with 100 ng/ml GnRH ±25 ng/ml activin and assayed the chromatin status of the Lhb promoter for DNase sensitivity. Indeed, activin with or without GnRH treatment promoted Lhb promoter opening (Fig. 4d, Two-way ANOVA compared to vehicle, p > 0.05 for activin at 2 and 5 units of DNaseI, and p > 0.01 for activin + GnRH at 2 and 5 units of DNaseI, p < 0.05 when comparing activin to activin + GnRH). Lhb is known to be induced by either GnRH or activin [23]. In contrast, these treatments did not change chromatin opening of the already opened Cga and Gnrhr promoters and could not change the chromatin status of Fshb (data not shown).