|Home | About | Journals | Submit | Contact Us | Français|
hTERT, the human telomerase reverse transcriptase, is highly expressed in stem cells and embryonic tissues but undetectable in most adult somatic cells. To understand its repression mechanisms in somatic cells, we investigated the endogenous hTERT gene regulation during differentiation of human leukemic HL60 cells. Our study revealed that silencing of the hTERT promoter was a biphasic process. Within 24 h after initiation of differentiation, hTERT mRNA expression decreased dramatically, accompanied by increased expression of Mad1 gene and disappearance of a nucleosome-free region at the hTERT core promoter. Subsequent to this early repression, nucleosomal remodeling continued at the promoter and downstream region for several days, as demonstrated by micrococcal nuclease and restriction enzyme accessibility assays. This later nucleosomal remodeling correlated with stable silencing of the hTERT promoter. Progressive changes of core histone modifications occurred throughout the entire differentiation process. Surprisingly, inhibition of histone deacetylation at the hTERT promoter did not prevent hTERT repression or nucleosomal deposition, indicating that nucleosomal deposition at the core promoter, but not histone deacetylation, was the cause of transcriptional repression. Our data also suggested that succeeding nucleosomal remodeling and histone deacetylation worked in parallel to establish the stable repressive status of hTERT gene in human somatic cells.
A central mechanism of cell specification and differentiation during development is transcription programming, the selective activation and silencing of specific genes (Muller and Leutz, 2001 ). This programming is largely achieved through highly regulated modulation of chromatin structures that package the eukaryotic genome. An important issue to resolve is how patterns of gene expression are established and stably maintained through subsequent cell divisions. On the one hand, activation of tissue-specific gene expression during cell differentiation has been studied in several models, which often involves temporal actions of transcription factors and chromatin-modifying complexes for covalent modifications of core histones and ATP-dependent nucleosomal remodeling at the promoters (de la Serna et al., 2006 ; Kaeser and Emerson, 2006 ). On the other hand, distinct and redundant chromatin modifications such as histone deacetylation and nucleosomal remodeling have been implicated in the transcriptional repression of several yeast genes (Zhang and Reese, 2004 ). However, little is known about the temporal processes of transcriptional repression during cell differentiation.
Telomeres are specialized nucleoprotein complexes capping chromosomal ends, and they play a key role in cellular aging and cancer (Morin, 1989 ; Kim et al., 1994 ). In stem cells and cancer cells, telomeres are replenished by telomerase, a ribonucleoprotein polymerase that maintains telomere ends by the addition of telomere repeat TTAGGG (Sharpless and DePinho, 2004 ). Telomerase consists of a catalytic reverse transcriptase subunit, encoded by the human telomerase reverse transcriptase (hTERT) gene, and an RNA component that serves as the template for telomere repeats. In most cells, the hTERT mRNA level parallels telomerase activity (Aisner et al., 2002 ). Both hTERT transcription and telomerase activity are down-regulated during cellular differentiation (Xu et al., 1999 ; Günes et al., 2000 ). Therefore, the hTERT gene is normally repressed in most postnatal somatic cells, resulting in progressive shortening of telomeres and proliferative senescence (Wright et al., 1996 ).
Studies of hTERT transcriptional regulation have largely been focused on factors that bind to the hTERT promoter. By binding to the E-box consensus sites in the hTERT promoter region, transcription factor c-Myc played a crucial role in the regulation of hTERT gene in both normal cells and cancer cells (Wu et al., 1999 ). It was shown that a switch of promoter occupancy by Myc/Max in proliferating cells to Mad/Max in differentiated cells correlated with hTERT repression during differentiation of HL60 cells (Xu et al., 2001 ). The mechanisms of transcriptional regulation by Myc/Max/Mad network proteins have been studied extensively (Eisenman, 2001 ; Luscher, 2001 ). It was shown previously that the activation domain of c-Myc protein bound to complexes containing histone acetylase (HAT) activities, such as TRRAP, p300/CBP, and TIP60 (Ikura et al., 2000 ; McMahon et al., 2000 ; Vervoorts et al., 2003 ). At the same time, c-Myc also associated with the SWI/SNF nucleosomal remodeling complex and the ATPase/helicases TIP48 and TIP49 (Cheng et al., 1999 ; Wood et al., 2000 ). Thus, recruitment of these two types of complexes is thought to be the mechanisms of transcriptional activation by c-Myc (Hooker and Hurlin, 2006 ). Conversely, Mad family proteins recruit mammalian Sin3 proteins (mSin3A and mSin3B) (Ayer et al., 1995 ), which are scaffold proteins that tether histone deacetylases (HDACs) (Knoepfler and Eisenman, 1999 ; David et al., 2008 ). In addition, histone modifications affect gene transcription by directly altering local nucleosomal architecture or through the recruitment of trans-acting factors that recognize specific histone modifications (Berger, 2002 ). ATP-dependent chromatin remodeling modifies histone-DNA contacts and results in insertion, eviction, and sliding nucleosomes along DNA, which change accessibility of DNA to transcription factors (Cairns, 2007 ). Hence, covalent modifications of core histones and ATP-dependent nucleosomal remodeling modulated by hTERT promoter binding proteins, such as Myc/Max/Mad network proteins, may play critical roles in the transcriptional control of the hTERT gene.
To understand the mechanisms of hTERT gene repression, we examined the chromatin structures of endogenous hTERT promoter during the differentiation of human HL60 cells. Upon stimulation by dimethyl sulfoxide (DMSO), HL60 cells underwent differentiation to granulocytic cells (Harris and Ralph, 1985 ), accompanied by a rapid and marked down-regulation of hTERT transcription. Our data revealed a dynamic process of transcriptional repression, as manifested by nucleosomal insertion and remodeling as well as core histone modifications. Interestingly, the initial cessation of hTERT transcription involved the deposition of a nucleosome and loss of c-Myc binding, but not histone deacetylation, at the core promoter region. Subsequent nucleosomal remodeling and histone deacetylation eventually led to a stably silenced state of hTERT promoter in fully differentiated cells. Therefore, our results demonstrated that nucleosomal remodeling/positioning and histone deacetylation played distinct roles in the establishment and maintenance of the stably repressive state of hTERT gene.
HL60 cells were cultured in RPMI 1640 medium supplemented with 10% fetal bovine serum. To induce differentiation, 2.5% DMSO was added to HL60 cells at a density of 0.5 × 106/ml. Cells were harvested at different time points after addition of DMSO.
Total RNA was isolated by TRIzol from 2 × 106 HL60 cells. cDNA was synthesized from 1 μg of total RNA, and PCR reactions were performed as described previously (Wang and Zhu, 2004 ). Real-time PCR was performed in triplicates using an ABI StepOnePlus system (Applied Biosystems, Foster City, CA). All experiments were repeated at least once. PCR primers and TaqMan probes are summarized in Supplemental Table 1.
ChIPs were performed as described previously (Wang et al., 2007 ). In brief, cells were cross-linked with 1% formaldehyde for 10 min at room temperature and harvested after quenching of formaldehyde with 125 mM glycine. After sonication, chromatin fragments of 200–800 base pairs were immunoprecipitated overnight at 4°C by using 0.5 μg of antibodies and amplified by PCR. For the hTERT core promoter region, TaqMan assays were used to quantify the ChIP results. Primers, probe, and a list of antibodies for ChIP are summarized in Supplemental Tables 1 and 2, respectively.
Nucleic preparation was performed as described previously (Wang and Zhu, 2003 , 2004 ). In brief, nuclei were prepared from 1 × 108 cells and incubated at 37°C with DNase I for 20 min, micrococcal nuclease (MNase) for 10 min, or MspI for 30 min. The nuclease digestion reactions were terminated by the addition of EDTA, and genomic DNAs were extracted. Genomic DNAs (10 μg) were digested with EcoRI and SphI (for DNase I assays) or with DraI (for MNase and MspI assays), followed by Southern blotting and indirect end labeling. For DNase I assays, probes a (−3.9 kb) and b (+1.7 kb) were used for 5′ and 3′ ends, respectively, as described previously (Wang and Zhu, 2003 ). For MNase and MspI assays, genomic fragments XbaI-DraIII (−1.3 to −0.9 kb, relative to transcription start site [TSS]) and PstI-PstI (+2.1 to +2.5 kb) were used as 5′ and 3′ probes, respectively.
Differentiation of human promyelocytic leukemia HL60 cells was accompanied by the repression of telomerase expression (Xu et al., 1999 ) and therefore investigated as a model of telomerase repression during cell differentiation. To determine the kinetics of repression, the hTERT mRNA level was examined by RT-PCR. Because the cisplatin-resistance-related gene 9 (CRR9), located immediately upstream of the hTERT gene, was constitutively expressed in all cell types examined and during cell differentiation (Wang and Zhu, 2003 ; Wang et al., 2004 , 2007 ), its mRNA, and also 18S rRNA, were used as normalizing controls. hTERT mRNA was readily detected in proliferating cells but was barely detectable after 24 h of differentiation (Figure 1A). The reduction of hTert mRNA was subsequently determined quantitatively by real-time PCR analysis. As shown in Figure 1B, the hTERT mRNA level decreased by 50- and 60-fold within 24 h of differentiation when normalized to 18S rRNA and CRR9 mRNAs, respectively. Four days after differentiation, the hTERT mRNA level decreased further to <100th of that in proliferating HL60 cells. Because it took at least 4 d for HL60 cells to completely differentiate into granulocytes (Birnie, 1988 ), these data indicated that down-regulation of hTERT expression was an early event of differentiation.
Myc/Max/Mad network transcription factors have been implicated in the regulation of hTERT gene, through binding to the E-box sites at the core promoter (Xu et al., 2001 ). Thus, the expression levels of c-Myc and Mad1 during cell differentiation were determined using semiquantitative RT-PCR analyses. As shown in Figure 1A, DMSO treatment of HL60 cells led to a gradual decrease of c-Myc mRNA during the differentiation process, consistent with the pattern of changes of the c-Myc protein, as determined by Western analysis (data not shown). This result indicated that the level of c-Myc expression did not correlate directly with that of hTERT expression during differentiation. Conversely, Mad1 expression showed a marked increase on the first day of differentiation and decreased gradually thereafter. In addition, the expression of Max, the binding partner of both c-Myc and Mad1, persisted throughout the differentiation process. These results suggested that increased Mad1 expression, but not the decrease of c-Myc expression, was crucial for the initiation of hTERT repression. Because the Mad1 mRNA level decreased during later phases of differentiation, its expression was unlikely to account for the stable maintenance of hTERT repression in fully differentiated cells.
Next, the physical associations of c-Myc and Mad1 proteins with the endogenous hTERT promoter during differentiation were examined by ChIP analysis. As shown in Figure 1C, the c-Myc protein was detected at the core promoter in proliferating HL60 cells. The binding of c-Myc to the core promoter decreased markedly after 4 d of differentiation. In contrast, binding of Mad1 protein to the core promoter increased in differentiated HL60 cells. Mad1 was shown previously to repress transcription through recruiting corepressor complexes such as mSin3A and HDACs to the chromatin (Rottmann and Luscher, 2006 ). Indeed, ChIP analysis showed that the association of mSin3A with the hTERT promoter, although detected in proliferating cells, was moderately increased in differentiated HL60 cells (Figure 1C). However, although this trend of increase was detected in multiple experiments, it was not statistically significant (p = 0.05~0.1). In short, these data were consistent with an earlier report by Xu et al. (2001) that c-Myc binding to the hTERT core promoter was replaced by promoter occupation of Mad family proteins during HL60 cell differentiation. The promoter association of mSin3A in proliferating cells was also consistent with our earlier observation that the hTERT promoter was subjected to repression even in telomerase-positive cells (Wang and Zhu, 2004 ; Wang et al., 2009 ).
To determine whether histone modifications were linked to the rapid reduction of the hTERT mRNA level that occurred after the initiation of differentiation, the time course of histone modifications was determined by quantitative ChIP analysis. As shown in Figure 2A, acetylation of histones H3 and H4 decreased progressively at the hTERT core promoter during differentiation. Similarly, levels of both di- and trimethylation of H3 lysine 4 (H3K4) at the core promoter were also reduced in differentiating cells. Whereas the trimethylated H3K4 became essentially undetectable in differentiated cells, a significant level of dimethylation of this residue remained, consistent with the notion that trimethylation, but not dimethylation, of H3K4 was correlated with active transcription. As a control, neither acetylation nor methylation of core histones was changed at the upstream CRR9 promoter (data not shown), which was constitutively active in both proliferating and differentiated cells (Figure 1A). Interestingly, dimethylation of H3K4 at the hTERT core promoter showed a transient increase on day 2 of differentiation. Although this H3K4 dimethylation on day 2 was lower compared with day 0, it was reproducible in multiple experiments. Finally, low levels of trimethylation of H3 lysine 27 (H3K27) were also detected at the hTERT promoter and displayed a gradual increase during differentiation. Thus, covalent modifications of core histones at the hTERT promoter underwent gradual changes during HL60 cell differentiation, whereas the decrease of hTERT mRNA expression was more rapid.
The gradual epigenetic changes over the hTERT promoter region raised the possibility that the initial rapid reduction of hTERT mRNA level during differentiation might be a posttranscriptional event. To address this issue, transcription in proliferating and differentiating HL60 cells was inhibited by actinomycin D treatment. The stability of the hTERT message in these cells was assessed by determining the hTERT mRNA levels at various times after the treatment. As shown in Figure 2B, the half-life of the hTERT message was 30 min to 1 h in both proliferating cells and cells that were induced for differentiation for 3 or 6 h, indicating that the stability of the hTERT mRNA was not significantly changed during differentiation. Thus, down-regulation of the hTERT mRNA level upon differentiation was unlikely to be a posttranscriptional event. In addition, as shown in Figure 2A by quantitative ChIP analysis, the association of RNA polymerase II (Pol II) with the hTERT promoter, and thus the formation of preinitiation complex (PIC), was greatly reduced within the first day of differentiation, correlated with the decreased level of hTERT mRNA. Together, our results indicated that transcriptional initiation of hTERT ceased early during differentiation.
DNase I sensitivity assays were performed to determine chromatin structural changes at the endogenous hTERT promoter during HL60 cell differentiation. Nuclei isolated from proliferating and differentiating cells were treated with increasing amounts of DNase I. Genomic DNAs were digested with restriction enzymes EcoRI and SphI, subjected to Southern blotting, and followed by indirect end labeling. The EcoRI–SphI full-length genomic fragment was shown in Figure 3 as a prominent 5.6-kb band. A 3.9-kb band was found in proliferating HL60 cells when the Southern blot was hybridized to a probe specific to the 5′ end of the full-length genomic fragment (Figure 3A). The intensity of this band initially increased with DNase I addition and subsequently decreased at higher nuclease concentrations, a characteristic of DNase I hypersensitive bands. The size of this band mapped a major DNase I hypersensitive site (DHS) to a region near the TSS. As HL60 cells started to differentiate, the intensity of this band decreased significantly within 24 h and the weakened band also shifted upward.
A higher resolution of DHS map was obtained, and multiple hypersensitive bands were detected when the same blot was hybridized to a probe specific to the 3′ end (Figure 3B). The first band is a diffused band around the position of 1.7 kb, corresponding to a major DHS at or immediately upstream of TSS. The intensity of this band decreased significantly within 24 h of differentiation. This DHS was probably the same as the DHS HS1 that we reported previously (Wang and Zhu, 2003 ). HS1 was found in all telomerase-expressing cells but undetectable in telomerase-negative cells (Wang and Zhu, 2004 ), indicating that HS1 was critical for hTERT transcription and probably a nucleosome-free region where Pol II and the transcription machinery were assembled. The rapid weakening of HS1 suggested that the nucleosome-free region was occupied by nucleosomes during an early stage of differentiation, correlating with the cessation of hTERT transcription. In addition, the 1.4- and 1.2-kb bands corresponded to DHSs in the first intron and the second exon of the hTERT gene, respectively. However, both of these DHSs appeared at higher concentrations of DNase I and were present throughout the differentiation process; therefore, they did not correlate with hTERT repression during cell differentiation.
For a more detailed analysis of chromatin structure at the hTERT promoter, the nuclei were subjected to digestion by MNase, which preferentially cleaves linker DNA between nucleosomes. Within the 1.6-kb region of −350 base pairs to +1300 base pairs relative to TSS, at least eight MNase cleavage sites were detected in both proliferating and differentiated HL60 cells when using a probe specific to the 5′ end of the DraI fragment (Figure 4A). The majority of these cleavage sites were also detected when the same blot was hybridized to a probe specific to the 3′ end (data not shown). In most cases, intervals between neighboring sites ranged from 150 to 250 base pairs. A similar pattern of cleavages was not obtained when naked genomic DNA was digested; therefore, these MNase cleavage sites were not intrinsic to the DNA sequences (Figure 4A, right). These features indicated that MNase sites were located at linker regions of nucleosomes and thus defined nucleosomal boundaries (Figure 4A, diagrams). A schematic interpretation of these data are shown in Figure 6. Although positions and relative intensities of most bands remained unchanged during differentiation, two clear differences of MNase digestion were identified between days 0 and 4. First, the region between two MNase sites within the first exon was also sensitive to the nuclease in proliferating cells (Figure 4A, left). This sensitivity (labeled by an open circle) persisted on days 0, 1, and 2 but not on day 4 (Figure 4A). The distance between these two MNase sites was <300 base pairs and was unlikely to accommodate two nucleosomes. Densitometric scanning of the cleavage profiles also confirmed increased MNase sensitivity at this region on days 0–2 (Figure 4B, gray triangles at the 1.5-kb position). Because this region of the first exon was not hypersensitive to DNase I, it was likely that it was packaged into a nucleosome but somehow still sensitive to MNase. Second, a 2.5-kb band was dominant in proliferating HL60 cells, corresponding to an MNase-hypersensitive cleavage site in the middle of exon 2. On differentiation, the band weakened progressively. A plausible interpretation is that an internucleosomal gap was present in the second exon in proliferating cells and this gap became smaller as HL60 cells differentiated. Overall, the MNase experiment allowed us to predict the positions of 10 nucleosomes (−3 to +7) within the 1.6-kb region at the 5′ end of the hTERT gene (see Figure 6).
To provide independent evidence about nucleosomal structures of the hTERT promoter, nuclei were also treated with restriction enzyme MspI. Within the 1.6-kb region, there are 10 single or clustered MspI sites (Figures 5 and and6).6). Due to the use of indirect end-labeling technique, the abundance of larger radiolabeled fragments was diminished by cleavages closer to the labeling end. Hence, cleavages were more accurately assessed at sites near the labeling end. In addition, relative sensitivities of adjacent cleavage sites could be compared. Three MspI clusters (1–3) are located within 300 base pairs upstream of TSS, a region that was hypersensitive to DNase I treatment (Figure 6). All three MspI clusters were accessible to the restriction enzyme in proliferating HL60 cells (Figure 5, lanes 1–3). Cluster 3, consisting of three MspI sites (−44 to −66 nt, relative to TSS; Figure 6), was especially sensitive to MspI digestion, and this hypersensitivity was not observed when naked genomic DNA was used (Figure 5, A and B, lanes 14 and 15), suggesting that the hTERT core promoter was readily accessible in proliferating cells. The accessibility of cluster 3 decreased significantly within 24 h of differentiation (Figure 5, lanes 4–6), consistent with rapid incorporation of a nucleosome (−1) to the core promoter (Figure 6). This loss of accessibility also correlated with the cessation of hTERT transcription early during differentiation. MspI site 2, overlapping with an MNase site, remained sensitive to MspI throughout the differentiation process, suggesting that this site was in the linker region before and after differentiation. MspI cluster 1, in contrast, became progressively more resistant to MspI digestion as cells differentiated. Because cluster 1 was also in the DNase I hypersensitive region, it was possible that this site was in a nucleosome-free region in proliferating cells and a nucleosome was recruited to this region more slowly than the downstream (−1) nucleosome. Alternatively, MspI cluster 1 was loosely wrapped around a nucleosome (−2) in proliferating cells and the packaging became gradually more compact as cells differentiated (Figure 6).
It was proposed that the first (+1) nucleosome downstream of TSS had a major impact on transcriptional regulation (Agalioti et al., 2000 ; Chen and Yang, 2001 ). Our data indicated that the first 200 base pairs of exon 1 seemed to be packaged into a nucleosome (+1) in proliferating cells, because it was resistant to digestion by both DNase I and MspI (Figures 3 and and5).5). However, the same sequence was also sensitive to MNase cleavage, as indicated by an open circle in Figure 4 (also see Figure 6). These results suggested that the (+1) nucleosome was in a loose configuration and/or relatively mobile, covering an extended DNA segment. A similar feature was also found in the (+1) nucleosome downstream of the HPRT promoter, which was called “split nucleosome” (Lee and Garrard, 1991 ; Chen and Yang, 2001 ). Such an open nucleosomal configuration is probably crucial for transcription to occur. However, this open conformation persisted through the first 2 d of differentiation, long after hTERT transcription ceased. This phenomenon also coincided with increased accessibility of MspI sites 4 and 5 on day 4 (Figures 5A and and6),6), suggesting that the (+1) nucleosome underwent remodeling and became more compact at a late stage of differentiation and long after cessation of hTERT transcription.
In addition, site 10 was also hypersensitive to MspI and its sensitivity decreased as cells differentiated, consistent with its position in a nucleosomal gap in proliferating cells (Figures 5B and and6).6). Finally, nucleosomes became more evenly distributed over the hTERT promoter region 4 d after differentiation (Figure 4, ,4d).4d). This ordered nucleosomal array was probably present in most human somatic cells and important for stable repression of the hTERT promoter.
We have demonstrated that down-regulation of hTERT transcription occurred within the first 24 h of differentiation, whereas nucleosomal remodeling and histone modifications at the hTERT promoter continued throughout the entire differentiation process. In fact, the hTERT mRNA level was significantly decreased within 3–6 h after the induction of differentiation, as shown in Figure 2B. To further dissect the repression process in the early phase of differentiation, DNase I sensitivity of the hTERT promoter was determined at multiple time points within the first 24 h of differentiation. The experiment was performed in the same way as in Figure 3, except that HL60 cells were treated with DMSO for 0, 3, 6, and 24 h. When the blots were probed with the upstream probe a, the major DNase I hypersensitive band became weakened and shifted upward even at 3 h of differentiation (Figure 7A). Higher resolution of DNase I cleavage patterns at the hTERT promoter was revealed by hybridizing the same blots with the downstream probe b (Figure 7B). The 1.7-kb hypersensitive band corresponding to the HS1 at the hTERT core promoter virtually disappeared after 3 h of DMSO treatment, suggesting that the postulated nucleosomal deposition (the −1 nucleosome) at the core promoter occurred very early during differentiation. ChIP experiments showed that loss of c-Myc binding and dissociation of Pol II from the hTERT promoter also took place within this initial phase of differentiation (Figure 7C). Conversely, acetylation and methylation of histones decreased gradually (Figure 7C). Together, these results demonstrated that the initial phase of hTERT repression occurred almost immediately upon the initiation of differentiation and involved dissociation of c-Myc and Pol II from the core promoter and concurrent occupation by a de novo nucleosome.
Covalent modifications of histones and nucleosomal remodeling at the hTERT promoter seemed to be temporally distinct events during differentiation, leading to a stably silenced hTERT gene in differentiated cells. To determine when stable repression of the hTERT promoter was established, HL60 cells were treated with DMSO for various times, and hTERT expression was measured by real-time RT-PCR 24 h after DMSO removal. As shown in Figure 8A, the level of hTERT mRNA returned to ~70% of the pretreatment level after DMSO was withdrawn at 1 d of differentiation. Two days after the initiation of differentiation, reversal of hTERT repression still occurred but the recovered level was <20% of that in proliferating cells. After 3 and 4 d of differentiation, the hTERT transcriptional repression became essentially irreversible. Because c-Myc and Mad1 were probably involved in the transcriptional regulation of the hTERT gene, their expression levels were also determined by RT-PCR analysis. As shown in Figure 8B, DMSO withdrawal resulted in significant increases of c-Myc expression and marked decreases of the Mad1 mRNA level at all time points. Interestingly, the up-regulation of c-Myc and down-regulation of Mad1 did not lead to substantial increase of hTERT expression at 3 or 4 d of differentiation. Therefore, these results indicated that repression of the hTERT promoter became stabilized on days 2–3 during differentiation, correlating with nucleosomal remodeling at the hTERT core promoter, especially the remodeling of +1 nucleosome.
Our results showed that deacetylation of core histones and nucleosomal remodeling occurred at the hTERT promoter in differentiating HL60 cells, suggesting that both processes contributed to hTERT repression. To determine whether the two processes were interdependent, HDACs were inhibited by trichostatin A (TSA) during HL60 cell differentiation. In this experiment, HL60 cells were only allowed to differentiate in the presence or absence of TSA for 24 h because longer TSA treatment resulted in increased cytotoxicity. Surprisingly, the hTERT mRNA level only slightly increased at higher concentrations of TSA (2 and 4 μM) after 1.5 h of differentiation, but its level showed no difference with or without TSA treatment at 3 h of differentiation (Figure 9A). ChIP experiments showed that TSA treatment completely prevented deacetylation of histones H3 and H4 at the hTERT core promoter during the first 24 h of differentiation (Figure 9B). In addition, examination of the chromatin structures by DNase I analysis revealed that HDAC inhibition by TSA resulted in a significant increase of general sensitivity of the hTERT promoter region to DNase I digestion but did not prevent the disappearance of HS1 (Figure 7, A and B). A closer look at the chromatin structures of TSA treated cells at 24 h of differentiation confirmed that MspI clusters 1–3 in differentiating cells became much less accessible to MspI digestion than the control proliferating HL60 cells despite the lack of histone deacetylation (Figure 9C). Comparing with the proliferating cells, intranucleosomal MspI sites 4 and 6 in differentiating cells were also more resistant to MspI digestion, whereas the internucleosomal MspI 5 was unaffected, consistent with a more stable nucleosomal array in this region. These results demonstrated that histone deacetylation was dispensable for nucleosomal remodeling at the hTERT promoter during the early stage of cell differentiation. The cessation of hTERT transcription was more closely linked to nucleosomal remodeling than histone deacetylation at the hTERT promoter.
Telomerase activity and hTERT mRNA expression is undetectable in the majority of adult somatic cells (Wright et al., 1996 ; Horikawa et al., 2005 ). How the repression of hTERT gene is initiated and maintained in somatic cells remains to be an important and yet unresolved question. Here, using HL60 cells as a model and exploiting multiple nuclease mapping techniques, we delineated the chromatin structures of the hTERT promoter during its transcriptional repression in differentiating cells. Our results, for the first time, revealed that transcriptional repression of the hTERT gene underwent at least three different states: transcriptionally active, nascently repressed, and stably silenced states.
The active hTERT promoter in proliferating HL60 cells contained a strong DHS (HS1), most likely a nucleosome-free region, at the core promoter sequence immediately upstream of TSS, as revealed by its hypersensitivity to DNase I and restriction enzyme MspI. This nucleosome-free region was surrounded by an upstream (−2) and a downstream nucleosome (+1). Because intranulceosomal DNA sequences in both adjacent nucleosomes were sensitive to MNase and/or MspI, the boundary nucleosomes were probably in modified and/or mobile configurations. Alternatively, these two nucleosomes might be absent in a subpopulation of cells, rendering the underlining sequence susceptible to nuclease cleavage. Although the exact status of these two nucleosomes remains to be elucidated, it was suggested previously that boundary nucleosomes were enriched with certain histone variants or covalent modifications, such as H3.3, H2A.Z, and/or ubiquinated histones (Jin and Felsenfeld, 2007 ; Weake and Workman, 2008 ). The maintenance of this active transcriptional state probably involved binding of the c-Myc protein, or related transcription activators, to the hTERT promoter, recruiting HAT and ATP-dependent nucleosomal remodeling complexes (Vervoorts et al., 2003 ), as the promoter region contained hyperacetylated histones H3 and H4.
Transcription of the hTERT gene ceased within the first few hours of differentiation. At this nascent repression state, the promoter contained a newly assembled nucleosome (−1) at the previously nucleosome-free region of the core promoter, as indicated by the loss of the major DHS HS1 as well as the decreased accessibility by MspI. However, the upstream (−2) and downstream (+1) nucleosomes remained largely unchanged at this time of differentiation. This nascent repression state coincided with the increase of expression of Mad family genes (Figure 1A; data not shown) and the switch of promoter occupancy from c-Myc to Mad1 proteins, resulting in the recruitment of mSin3A-containing complexes with HDAC activity (Figure 1C; Xu et al., 2001 ). In addition, recent data from the Xu laboratory showed that the histone demethylase RBP2 was also recruited to the hTERT promoter by Mad1 upon HL60 differentiation (Ge et al., 2009 ). As a result, acetylation of core histones H3 and H4, as well as di- and trimethylation of lysine 4 of H3, decreased throughout the entire differentiation process. In spite of this, the conversion from the state of active transcription to the nascent repression state did not require histone deacetylation at the hTERT promoter (Figure 9). Thus, it was likely that removal of the c-Myc transcription factor and the associated nucleosomal remodeling complexes from the promoter resulted in the incorporation of a de novo nucleosome to the core promoter or vice versa. However, the repression at this nascent state remained fragile and reversible, because hTERT transcription was almost fully recovered upon DMSO withdrawal (Figure 8A).
During the later stages of differentiation (days 2–4), nucleosomes at the hTERT promoter continued to remodel and the repression ultimately became irreversible (Figure 8). In particular, nucleosomes (−2) and (−1), the two nucleosomes overlapping with the hTERT core promoter, became more compact and stable, as indicated by the progressively inaccessible MspI clusters 1 and 3, respectively (Figure 5). In both proliferating and early differentiating cells (days 0, 1, and 2), nucleosome (+1) covered an extended region within the first exon and its DNA was susceptible to MNase attack (Figure 4), suggesting that this nucleosome was mobile and/or missing in a subpopulation of cells. Continuing remodeling resulted in stabilization of nucleosome (+1), as suggested by its decreased MNase sensitivity (Figure 4A, lane d, and Bd). This nucleosome was also more compact and occupied less DNA sequence, as indicated by increased accessibility of MspI sites 4 and 5 (Figures 5 and and6).6). It has been reported previously that the nucleosome (+1) also adopted such an altered configuration in the active, but not inactive, HPRT promoter (Chen and Yang, 2001 ). The molecular nature of this type of altered nucleosomal configuration remains to be determined, although it was proposed previously that incorporation of linker histones and histone variants, as well as covalent modifications of core histones, might exert a major influence on nucleosomal mobility and chromatin accessibility (Ura et al., 1995 ; Jin and Felsenfeld, 2007 ; Weake and Workman, 2008 ). Our result indicated that remodeling of nucleosome (+1) was not required for the initiation of repression but might be important to lock the hTERT promoter into a stably repressive state. In this stably repressed state of the promoter, nucleosomes were orderly positioned over the 5′ end of hTERT gene. Furthermore, the maintenance of this silenced state did not seem to depend on continued expression of Mad1, Mad3, and Mad4, which increased on day 1, peaked on days 1–3, and decreased after day 4 (Figure 1A; data not shown). In addition, in this silenced state, histones H3 and H4 at the core promoter were hypoacetylated. Although not required for the initial repression, histone deacetylation might be important for stable maintenance of the repressive state of the hTERT promoter.
Three parallel mechanisms of repression have been proposed from previous studies: nucleosome positioning, recruitment of HDACs, and direct interference with the general transcription machinery or activators. Using DNA damage-inducible genes, Zhang and Reese (2004) observed that genetically inactivating at least two of the three processes led to a corresponding increase in the expression of RNR3 and HUG1 in Saccharomyces cerevisiae. Although such genetic manipulations are difficult to achieve in a mammalian cell culture system, we believe that similar mechanisms are also important for repressing the hTERT promoter. Our experiments further demonstrated that nucleosome positioning and histone deacetylation might play temporally distinct roles in the course of repression of the hTERT gene.
Based on our experiments, transcription occurred at the hTERT promoter that contained a nucleosome-free region upstream of TSS. This nucleosome-free region, as revealed by HS1, has been detected in all telomerase-expressing cells (Wang and Zhu, 2004 ), indicating that it is essential for hTERT transcription. Maintaining this nucleosome-free region probably involves the binding of activators such as c-Myc protein, which recruits HAT and ATP-dependent nucleosome remodeling complexes. Although histone acetylation facilitates transcription factor binding and eases nucleosomal displacement, ATP-dependent nucleosomal remodeling complexes may be responsible for nucleosome removal. Indeed, moderate inhibition of histone acetylation by curcumin, which was shown to inhibit the p300 CBP (cAMP response element-binding protein binding protein) HAT (Balasubramanyam et al., 2004 ), led to a significant reduction of hTERT transcription in proliferating cells (Supplemental Figure S1). This inhibition also occurred in the presence of protein synthesis inhibitor cycloheximide, indicating a direct role of histone acetylation in maintaining the active state of the hTERT promoter (Supplemental Figure S1B). Therefore, during the initial phase of differentiation, binding of negative transcription factors such as Mad proteins to the hTERT promoter may indirectly result in nucleosomal deposition by displacing the activator c-Myc (Figure 7C) and thus its associated HATs and other chromatin modifiers. In addition, negative factors may also recruit nucleosome assembly factors/histone chaperones, which bind to histones and facilitate their assembly into nucleosomes (Akey and Luger, 2003 ). This nucleosomal deposition alone may be sufficient to shut down transcription because nucleosome structure is incompatible with the assembly of transcription machinery due to the large amount of DNA sequence directly contacted by the Pol II/PIC complexes (Kornberg and Lorch, 1999 ). As proposed by Zhao and colleagues, although histone acetylation is involved in nucleosomal eviction, nucleosomal deposition does not necessarily require histone deacetylation (Schones et al., 2008 ). Instead, histone deacetylation may be more closely linked to chromatin stabilization after nucleosomal deposition.
Although histone deacetylation has been generally linked to gene repression, its precise role in transcriptional regulation remains unresolved. During HL60 cell differentiation, early repression of the hTERT promoter did not require histone deacetylation. Although surprising, this result was consistent with studies in other model systems. For example, repression of RNR3 remained when genes encoding histone deacetylases HDA1 and RPD3 were either deleted alone or together (Zhang and Reese, 2004 ). In contrast, significant derepression was observed when Δhda1 mutation was combined with another mutation Δisw2, which disrupted nucleosome positioning. However, histone deacetylation is probably essential for stable hTERT silencing, because HDAC inhibition led to hTERT transcription in many human somatic cells. The important role of histone deacetylation may lie in the subsequent remodeling of nucleosome (+1) and/or stabilization of the repressive state of the hTERT promoter in differentiated cells.
Several questions need to be addressed to further understand the mechanisms of hTERT silencing in somatic cells. First, besides the Myc/Max/Mad network proteins, other transcription factors, such as Sp1, E2F1, Ets family protein, and AP1(activator protein-1), also have been implicated in hTERT regulation (Crowe et al., 2001 ; Goueli and Janknecht, 2004 ; Takakura et al., 2005 ; Wooten and Ogretmen, 2005 ). Thus, it is still not resolved whether the switch of promoter occupancy by the c-Myc protein to Mad repressors was sufficient for transcriptional repression. Second, factors that facilitate nucleosomal deposition and subsequent nucleosomal remodeling at the hTERT core promoter remain to be identified. Third, it is of significant interest to understand the detailed molecular characteristics of the boundary nucleosomes (−2) and (+1) before and after remodeling that eventually led to the stably repressed state of the hTERT promoter. Finally, although the current work provided significant insight into the mechanisms of hTERT repression in differentiating HL60 cells, future studies will be needed to determine whether these mechanisms are also involved in hTERT silencing in other somatic cells. To this end, our recent study also showed that hTERT repression was associated with decreased DNase I accessibility at the hTERT promoter during TPA (12-O-tetradecanoylphorbol-13-acetate)–induced U937 cell differentiation (Wang and Zhu, 2004 ; Wang et al., 2009 ).
In summary, the current study identified several spatial and temporal steps of hTERT repression during the differentiation of HL60 cells. Our data demonstrated that the initial transcriptional repression of the hTERT promoter was independent of histone deacetylation but correlated with the deposition of a nucleosome to the core promoter. Subsequent remodeling of nucleosomes, in particular the nucleosome (+1) downstream of TSS, resulted in a stably silenced hTERT promoter.
We thank Dr. Sergei Grigoryev for insightful discussion and advice on chromatin analyses and Longgui Chen for technical assistance. The study was supported by National Institutes of Health grant GM-071725. J. Z. is a Research Scholar of American Cancer Society.
This article was published online ahead of print in MBC in Press (http://www.molbiolcell.org/cgi/doi/10.1091/mbc.E09-06-0456) on January 6, 2010.