Evidence for a relief of repression mechanism for activation of the human telomerase reverse transcriptase promoter.

The transcriptional activation of human telomerase reverse transcriptase (hTERT) is an important step during cellular immortalization and tumorigenesis. To study how this activation occurs during immortalization, we have established a set of genetically related pre-crisis cells and their immortal progeny. As expected, hTERT mRNA was detected in our telomerase-positive immortal cells but not in pre-crisis cells or telomerase-negative immortal cells. However, transiently transfected luciferase reporters controlled by hTERT promoter sequences exhibited similar levels of luciferase activity in both telomerase-positive and -negative cells, suggesting that the endogenous chromatin context is likely required for hTERT regulation. Analysis of chromatin susceptibility to DNase I digestion consistently identified a DNase I hypersensitivity site (DHS) near the hTERT transcription initiation site in telomerase-positive cells. In addition, the histone deacetylase inhibitor trichostatin A (TSA) induced hTERT transcription and also a general increase in chromatin sensitivity to DNase treatment in telomerase-negative cells. The TSA-induced hTERT transcription in pre-crisis cells was accompanied by the formation of a DHS at the hTERT promoter. Furthermore, the TSA-induced hTERT transcription and chromatin alterations were not blocked by cycloheximide, suggesting that this induction does not require de novo protein synthesis and that TSA induces hTERT expression through the inhibition of histone deacetylation at the hTERT promoter. Taken together, our results suggest that the endogenous chromatin environment plays a critical role in the regulation of hTERT expression during cellular immortalization.

The transcriptional activation of human telomerase reverse transcriptase (hTERT) is an important step during cellular immortalization and tumorigenesis. To study how this activation occurs during immortalization, we have established a set of genetically related pre-crisis cells and their immortal progeny. As expected, hTERT mRNA was detected in our telomerase-positive immortal cells but not in pre-crisis cells or telomerasenegative immortal cells. However, transiently transfected luciferase reporters controlled by hTERT promoter sequences exhibited similar levels of luciferase activity in both telomerase-positive and -negative cells, suggesting that the endogenous chromatin context is likely required for hTERT regulation. Analysis of chromatin susceptibility to DNase I digestion consistently identified a DNase I hypersensitivity site (DHS) near the hTERT transcription initiation site in telomerase-positive cells. In addition, the histone deacetylase inhibitor trichostatin A (TSA) induced hTERT transcription and also a general increase in chromatin sensitivity to DNase treatment in telomerase-negative cells. The TSAinduced hTERT transcription in pre-crisis cells was accompanied by the formation of a DHS at the hTERT promoter. Furthermore, the TSA-induced hTERT transcription and chromatin alterations were not blocked by cycloheximide, suggesting that this induction does not require de novo protein synthesis and that TSA induces hTERT expression through the inhibition of histone deacetylation at the hTERT promoter. Taken together, our results suggest that the endogenous chromatin environment plays a critical role in the regulation of hTERT expression during cellular immortalization.
Telomeres are specialized nucleoprotein complexes that serve as protective caps of linear eukaryotic chromosomal ends and are essential for both chromosomal stability and long term cellular proliferation. Human telomeres consist of TTAGGG repeats that can be replenished by the action of telomerase (1), a ribonucleoprotein reverse transcriptase complex containing an RNA subunit (hTER) and a protein catalytic subunit (hTERT). 1 hTER RNA is ubiquitously expressed in both telom-erase-positive and -negative cells (2). Conversely, expression of the hTERT message correlates with the presence of telomerase activity (3,4). In most normal somatic cells, the hTERT gene is repressed and telomeres progressively shorten upon successive cell divisions, because of the absence of telomere maintenance, leading to permanent cell cycle arrest (known as M1 senescence) at the end of their life span (Hayflick limit). However, when normal cells are transformed by viral oncogenes, such as SV40 large T antigen, they are able to bypass M1 senescence and continue to proliferate despite telomere shortening (5). This leads to the loss of telomere capping activity and culminates in M2 crisis, a process of severe chromosomal instability and massive cell death. The few cells that survive crisis exhibit one of the two telomere maintenance pathways. One pathway is dependent on telomerase activity and involves activation or de-repression of hTERT gene expression. The second pathway is independent of telomerase activity and is referred to as alternative lengthening of telomere (ALT) (6). As suggested by studies in yeast, the ALT pathway presumably involves a recombination process (7).
The regulation of telomerase activity likely involves multiple mechanisms. The major regulatory step of hTERT expression is associated with transcription initiation (8). In cancer cells and immortalized cell lines that express telomerase, the level of hTERT transcripts has been shown to correlate with telomerase activity (3,4). In addition, the hTERT transcripts appear to include multiple alternatively spliced forms, several of which have been predicted to encode inactive or dominant negative isoforms of the hTERT protein (9). Post-translational modifications, such as phosphorylation, may also contribute to the regulation of hTERT activity (10).
Although the means by which hTERT gene expression is controlled remain an area of intense investigation, available evidence indicates that hTERT transcription is regulated by both positive and negative mechanisms. Positive regulation of hTERT gene expression has been suggested by results from analyses of the hTERT promoter. It has been reported that hTERT transcription can be activated by overexpression or activation of several transcription activators, including c-Myc, E6, and the estrogen receptor (11)(12)(13)(14)(15)(16). Experiments utilizing transiently transfected reporter constructs have led to the identification of a proximal promoter region that is apparently sufficient for maximum hTERT promoter activity. This region, located from Ϫ1 kb to ϩ60 bp relative to the transcription start site, contains putative binding elements for several transcription factors, including Sp1, c-Myc, E2F, NF-1, AP2, and the Wilms' tumor protein (16,17). It is known that c-Myc, E2F, and Sp1 proteins are expressed in normal proliferating cells that lack telomerase activity. Therefore, in addition to direct acti-vation of the hTERT promoter by transcription factors in many cancer cells, mechanisms of active repression must be present in normal somatic cells to prevent inappropriate expression of the hTERT gene.
Several lines of evidence suggest that the hTERT gene is subject to negative control. First, hTERT expression is repressed in most somatic cells. Its expression is induced in normal cells by treatment with the histone deacetylase (HDAC) inhibitor trichostatin A (TSA) (18,19), although this could be an indirect effect. Second, somatic cell fusion studies have revealed that hTERT expression in immortalized cells can be silenced upon their fusion with normal cells (20). Additionally, the introduction of several different normal chromosomes or chromosomal segments into immortal cells leads to telomerase suppression, telomere shortening, and a restored senescence/ crisis phenotype, implying that these chromosomal regions harbor genes involved in telomerase repression (21).
Although hTERT transcription has been studied extensively, mechanisms of its activation during cellular immortalization remain largely unknown. Using transiently transfected reporter constructs, a number of potential regulatory elements and their binding proteins have been identified (17,18,(22)(23)(24). Overexpression of some of these transcription factors, such as c-Myc, was shown to activate endogenous hTERT transcription in different cellular contexts; however, their roles in telomerase activation during cellular immortalization are still unclear. Durest et al. (25) recently reported that transiently transfected green fluorescence protein reporters controlled by hTERT promoter sequences were equally active in two unrelated lines of fibroblasts, the telomerase-negative GM847 ALT cell line and the telomerase-positive cell line GM639. This result indicates that transient transfection of plasmids containing the hTERT promoter may not be an appropriate model for studying activation of the endogenous hTERT transcription (25).
In most of the previous studies of the hTERT promoter, normal cells and cancer cell lines with unrelated lineage were used (16,17,26). Whereas these different cell lines have provided important information about hTERT promoter function, it is conceivable that cells with divergent backgrounds may regulate hTERT transcription differently because of the distinct complement of transcription factors they express. To avoid this problem, recent studies have utilized somatic cell hybrids (25,27). Durest et al. (25) showed that an extra chromosome 3 from normal cells was able to repress endogenous hTERT transcription but not transcription from transiently transfected hTERT reporters in the breast cancer cell line 21NT. In contrast, using the RCC23 renal cell carcinoma cell line, Horikawa et al. (27) showed that introducing an extra chromosome 3 repressed not only the endogenous hTERT promoter but also transiently transfected hTERT promoter reporters. The repression of hTERT promoter reporters required an E-box site downstream of the transcription start site (ϩ44 nucleotides), suggesting that Myc family proteins, or other E-box-binding proteins, may participate in the repression of the hTERT promoter. Thus, it remains to be resolved whether different genes on chromosome 3 are responsible for hTERT repression in different cell types (27). Whereas somatic cell hybrids have provided excellent models for understanding the regulation of hTERT expression because these cells are isogenic except for the transferred chromosome, the mechanisms by which the extra chromosome induces senescence are still obscure. It remains to be determined whether genetic mutations of genes on chromosome 3 are involved in the immortalization of human cells (28). It is also possible that overexpression of certain genes on this chromosome may lead to repression of the hTERT promoter by a mechanism that is not involved in cellular immortalization and tumorigenesis. Therefore, to understand the regulation of hTERT expression, it is necessary to examine the hTERT promoter activity in cells both before and after the immortalization event.
To address how telomerase is activated during cellular immortalization, we have established a set of clonal pre-crisis human fibroblast cell lines and their immortal progeny. The immortal cell lines were independently derived from initially clonal populations of pre-crisis cells, designated 3A or 3C, and included cell lines that expressed telomerase activity and ALT cell lines that did not express telomerase. All these lines have similar genetic backgrounds, exhibit similar morphologies, and proliferate at similar rates. They are therefore excellent models for studying hTERT promoter regulation during cellular immortalization. Using this set of cells, we demonstrate here that the endogenous hTERT promoter was expressed in telomerasepositive cell lines but repressed in telomerase-negative precrisis cells and ALT lines. The expression of hTERT mRNA in telomerase-positive cells correlated with the appearance of a major DNase I hypersensitivity site (DHS) near the hTERT transcription start site. However, transiently transfected hTERT promoter reporters were equally active in both telomerase-positive and -negative cells. In addition, the histone deacetylase inhibitor TSA induced endogenous hTERT expression in telomerase-negative cells. This induction did not require de novo protein synthesis and was accompanied by a general increase in chromatin sensitivity to DNase I digestion.
In pre-crisis cells, a DHS was also formed near the hTERT promoter upon TSA treatment. Our data suggest that the endogenous hTERT promoter is repressed by a native chromatin structure in telomerase-negative cells, and that activation of hTERT expression during cellular immortalization may involve relief of repression and chromatin remodeling at the hTERT promoter.

MATERIALS AND METHODS
Plasmids and Cell Culture-The hTERT reporter plasmids and the genomic clone pNSV4 were kindly provided by Drs. J. C. Barrett (NIEHS, National Institutes of Health) and Riccardo Dalla-Favera (Columbia University), respectively (16,17). 3A and 3C pre-crisis cells were established by transforming human lung fibroblast IMR90 cells with a plasmid expressing SV40 large T and small t antigens (29). Cells were cultured in minimal essential medium (Invitrogen) containing 10% fetal bovine serum (Hyclone).
pYF6 and pYF7 were derived from the genomic DNA clone pNSV4, a pUC19 derivative containing ϳ14 kb of genomic sequence of the hTERT promoter (16). The first and second codons of the hTERT open reading frame were converted to a PvuI site (CGATGCCG 3 CGATCGCG) by PCR-based mutagenesis. The mutagenized PCR product was subcloned into pCR-blunt II-TOPO (Invitrogen) and the mutation was confirmed by sequencing. The linker AAGCTTCTCGAGAGATCTAGAT was inserted into this new PvuI site. The HindIII-BamHI fragment of pGL3-Basic (Clontech), containing the firefly luciferase open reading frame and SV40 poly(A), was cloned into the HindIII and BglII sites of the linker. pNSV4 was digested with either XbaI or SphI and self-ligated to generate pNSV4-XbaI and pNSV4-SphI, respectively. The first exon and intron, located between two SacII sites, of these plasmids were replaced by their modified counterparts found in the pCR-blunt II-TOPO-derived construct, generating pYF6 and pYF7, respectively.
Northern Analyses-Total RNA was isolated from exponentially proliferating cells using Trizol reagent (Invitrogen). Ten micrograms of total RNA from various cell types was run on a 1.2% agarose/formaldehyde gel and transferred to a MAGNAGraph membrane (Osmonics) in 10ϫ SSC. Following UV cross-linking, the membrane was prehybridized and hybridized at 65°C in Church buffer (0.5 M NaPO 4 , pH 7.0, 7% SDS, 1 mM EDTA, 1% bovine serum albumin) in the presence of a 32 P-labeled probe of interest. The blot was then washed with 0.1ϫ SSC, 0.1% SDS twice for 15 min each at 65°C and subjected to autoradiography. To hybridize the blot with a new probe, the membrane was stripped by incubating in boiling water containing 0.5% SDS for 1 h to remove the old probe.
The probes were labeled with [␣-32 P]dCTP (3000 Ci/mmol, PerkinElmer Life Sciences) by the random priming method using Ready-To-Go DNA Labeling Beads (ϪdCTP) (Amersham Biosciences). Free, unincorporated ATP was removed from all labeled probes using MicroSpin G-50 columns (Amersham Biosciences). Luciferase Assay-Cells were transfected using FuGENE 6 (Roche Diagnostics) in 24-well plates with 0.2 g of firefly luciferase reporter DNA and 0.001 g of pRL-CMV (Promega). Cells were harvested 2 days post-transfection, and luciferase activities were measured using the Dual Luciferase Reporter (DLR TM ) assay system (Promega). The firefly luciferase activity was normalized to Renilla reniformis luciferase activity. Each data point represents an average of three independent transfections.
Telomerase Assay and Telomere Southern Blots-A modified telomeric repeat amplification protocol (TRAP assay) (30) was used to measure telomerase activity, as described previously (29). Telomere Southern analyses were performed as described previously (29). The oligo probe for telomere Southern blots was prepared by labeling the 5Ј end with [␥-32 P]ATP (3000 Ci/mmol) using T4 polynucleotide kinase (New England Biolabs).
RT-PCR Assay-One to 2 g of total RNA was reverse transcribed with an oligo(dT) primer using the SuperScript First Strand Synthesis System for RT-PCR (Invitrogen). The first strand cDNAs were amplified by PCR with primers specific for hTERT and CRR9 (cisplatinresistance related gene). The amplified products were analyzed on a 2% agarose, 0.5ϫ TBE gel. The sequences for the hTERT primers are: 5Ј-TTTCTGGATTTGCAGGTGAA-3Ј and 5Ј-CAGGAAAAATGTGGGG-TTCT-3Ј; CRR9 primers are: 5Ј-GCCATTGAGCTGTGGAAAGT-3Ј and 5Ј-TCCCCAAACTCGTTCACTCT-3Ј.
Real-time quantitative RT-PCR analysis (Taqman assay) was performed at the Genome Analysis Core, University of California, San Francisco, Comprehensive Cancer Center. Ten micrograms of total RNA was digested with RNase-free DNase I for 15 min at 25°C. DNase I treatment was terminated by the addition of 25 mM EDTA and incubation at 65°C for 10 min. cDNAs were synthesized with random hexamers using the SuperScript First Strand Synthesis System (Invitrogen). The hTERT primers for PCR reactions were 5Ј-CTGTGCCAC-CAAGCATTCCT-3Ј and 5Ј-GGCTGTCCTGAGTGACCCC-3Ј and the probe was 5Ј-FAM-ACTCGACACCGTGTCACCTACGTCCC-TAMRA-3Ј. A standard curve was created in each experiment using serial dilutions of positive template. The relative amount of hTERT mRNA was calculated by plotting the C t (cycle number above the threshold) against the standard curve and normalizing to the glyceraldehyde-3phosphate dehydrogenase message.
DNase I Hypersensitivity Assay-Cultured cells (ϳ1 ϫ 10 8 ) were harvested by scraping in phosphate-buffered saline and followed by two washes with phosphate-buffered saline. Nuclei were isolated by a method adapted from a protocol provided by Sergei Grigoryev (Penn State College of Medicine). Briefly, the pelleted cells were resuspended in 7 ml of RSB (3 mM MgCl 2 , 10 mM NaCl, 10 mM Tris-HCl, pH 7.6) plus 0.5% Nonidet P-40 and 0.5 mM Pefabloc. The cells were then homogenized in a 15-ml Dounce homogenizer (tight pestle) for 30 strokes over a 30-min period. After centrifugation at 1,000 ϫ g for 5 min, the pelleted nuclei were resuspended in 1.5 ml of RSB plus 0.5 mM Pefabloc and kept on ice until DNase I treatment.
Aliquots of 180 l of resuspended nuclei were placed into several microcentrifuge tubes. Serial dilutions of DNase I (RQ1-RNase I-free DNase, Promega) were prepared in DNase dilution buffer (10 mM HEPES-KOH, pH 7.9, 30 mM CaCl 2 , 30 mM MgCl 2 , 50% glycerol). Twenty microliters of the diluted DNase I was added to an aliquot of nuclei to obtain final concentrations of 0, 1, 2, 4, 8, and 16 units/ml. The samples were incubated at 37°C for 20 min, and reactions were terminated with 50 l of 0.5 M EDTA. Following DNase I treatment, the nuclei aliquots were collected by centrifugation in a microcentrifuge at 4,500 rpm for 5 min. Genomic DNA was isolated from these DNase I-treated nuclei using the Wizard genomic DNA purification kit (Promega). Ten micrograms of DNA was digested with EcoRI and SphI and separated on 0.7% agarose, 0.5ϫ TBE gels. Gels were sequentially treated in 1.5 M NaCl, 0.5 M NaOH and 1.5 M NaCl, 0.5 M Tris, pH 8.0, 20 min each, and transferred to MAGNAGraph membrane in 10ϫ SSC. Hybridization and washing conditions were the same as described above for the Northern blot procedure. The blots were exposed to autoradiography for 1 to 3 weeks.

Generation of Telomerase-positive and -Negative Immortal
Cell Lines from Pre-crisis Cells-Pre-crisis cells were obtained by transforming human lung fibroblast IMR90 cells with a plasmid expressing the SV40 early genes, large T and small t antigens (29). The growth curves for the two independently established, initially clonal cell populations 3A and 3C are shown in Fig. 1A. Both 3A and 3C cells proliferated at a similar rate and reproducibly underwent crisis after ϳ90 population doublings. In the representative experiment shown in Fig. 1A, all the pre-crisis 3C cells died during crisis, whereas spontaneous colonies arose from two of plates of pre-crisis 3A cells. Cells derived from these colonies have proliferated more than 100 population doublings beyond the crisis point and are thus considered immortal. The crisis could be averted by infecting the pre-crisis cells with a recombinant retrovirus expressing A, growth curves of two independent, initially clonal precrisis 3A and 3C cells. 5 ϫ 10 5 cells were re-seeded to a new 10-cm plate when they reached confluency. Multiple curves in each panel represent multiple passages of the same cells. Population doublings were based on estimated cumulative numbers from the initial IMR90 culture. B, telomerase activity measured by TRAP assays. 0.05, 0.25, and 1.00 g of cell extracts were used in each set of three assays. 3A/hTERT is a cell line immortalized by the recombinant retrovirus BABE-hTERT (29). C, telomere Southern analysis of the telomeric restriction fragments. DNA size markers (in kilobases) are shown on the left. "ϩ" or "Ϫ" indicates expression of telomerase in the cells examined. D, Northern analysis of hTERT detected by probes specific for hTERT, CRR9, and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) mRNA, respectively. The arrowhead points to the endogenous hTERT mRNA and asterisk indicates the message expressed from the integrated retrovirus BABE-hTERT.
hTERT cDNA (3A/hTERT), as described previously, indicating that crisis was induced by telomere defects (29). In other experiments, we were able to generate spontaneous immortal lines from both 3A and 3C pre-crisis cells. Whereas the precrisis cells themselves expressed no detectable telomerase activity as measured by TRAP assays, some of their immortalized progeny lines, such as 3C104a and 3C167b, were telomerasepositive (Fig. 1B). These telomerase-expressing cell lines had relatively short telomeres, ranging from 3 to 6 kb (lanes 3 and 5, Fig. 1C). On the other hand, some immortal lines (3A#96, 3C87a, 3C166a, and 3C4C) displayed no detectable telomerase activity and contained longer telomeres than both the parental pre-crisis cells and their telomerase-positive siblings (Fig. 1C,  lanes 1, 4, and 6). This suggests that these telomerase-negative immortal cells maintain their telomeres by the ALT pathway (6). These experiments also demonstrate that the pre-crisis cells can be immortalized either through activating telomerase expression or by the ALT pathway. The telomerase-expressing immortal lines and ALT lines had a similar morphology (data not shown) and proliferated at the same rate as their parental pre-crisis cells (Fig. 1A). Therefore, these cells are used as models for studying the molecular events that occur during immortalization.
To determine the mechanisms of telomerase regulation in these cells, we first measured the levels of hTERT mRNA in pre-crisis cells and several immortal cell lines by Northern blot analysis. The endogenous 4-kb hTERT message was detected in the telomerase-positive lines, 3C104a and 3C167b, but not in the pre-crisis cells and ALT lines (Fig. 1D). As expected, the 3A/hTERT cells expressed a 6-kb message from an integrated provirus carrying hTERT cDNA (29), but not the endogenous 4-kb hTERT mRNA. In contrast to the hTERT gene, CRR9, the gene located at the 5Ј side of hTERT, was expressed at a similar level in both telomerase-positive and -negative cells (Fig. 1D). Therefore, it is unlikely that the hTERT and CRR9 genes are regulated by the same mechanism.
To further quantify the hTERT expression level in various cell lines, hTERT mRNA was also measured by real-time RT-PCR, using a primer pair spanning intron 15 of the hTERT gene. The results from this RT-PCR assay showed that the levels of hTERT-specific signal were readily detected in telomerase-expressing cells (3C104a and 3C167b) but undetectable in pre-crisis cells and telomerase-negative ALT cells (3C4C, 3C166a, and 3A#96) ( Fig. 2A). This data demonstrates that the level of hTERT mRNA is correlated with telomerase activity, indicating that hTERT expression is regulated at the level of transcription.
The Activities of Transiently Transfected hTERT Reporters-Using pre-crisis cells and their independently immortalized derivatives, we have analyzed hTERT promoter activity. We first tested the luciferase reporter pBTdel-130 that contains a region of the hTERT promoter from Ϫ107 to ϩ28 bp relative to the transcription start site, the minimum sequence required for the full activity of hTERT promoter in several cancer cell lines (17). This core region contains several potential regulatory elements, including an E2F site (Ϫ98 bp), an Ets site (Ϫ22 bp), and four Sp1 sites (Ϫ90, Ϫ56, Ϫ36, and Ϫ9 bp) (22,23). pBTdel-130 was transfected into pre-crisis, telomerase-expressing, and ALT cells. Forty-eight hours post-transfection, luciferase activity was measured. As shown in Fig. 2B, this reporter expressed a level of luciferase activity that was 30 -50% of the activity of the control CMV promoter (pCMVLuc) in pre-crisis cells and the immortal cells (either telomerase-positive or -negative cells). In addition, this reporter was also active in 3A/hTERT, the immortal line that ectopically expresses a copy of hTERT cDNA and in which no endogenous hTERT expression has been detected. Thus, the luciferase activity expressed from this core promoter did not correlate with endogenous hTERT expression (compare Fig. 2, A to B). Instead, the core hTERT promoter is a relatively strong promoter in both telomerase-positive and -negative cells.
A possible explanation for the difference observed between the luciferase activity of the reporter and the level of endogenous hTERT expression is that a negative regulatory element may reside outside of the core promoter region. To look for such a regulatory sequence, we have examined a series of hTERT promoter-luciferase reporters in both hTERT-expressing cells and telomerase-negative cells. This series of reporters contain varying lengths of upstream sequences of the hTERT promoter inserted at the 5Ј end of the firefly luciferase gene in pGL3-Basic (Fig. 3A) (17). Interestingly, all of these reporters expressed essentially the same level of luciferase activity in precrisis 3C cells, the telomerase-expressing immortal 3C104a cells, and the telomerase-negative ALT 3C166a cells (Fig. 3A). Their activities ranged from 27 to 59% of that of the CMV promoter. Similar results were obtained using other telomer-  Fig. 1D). B, activity of the hTERT core promoter from pBTdel-130 measured by luciferase reporter assays. The firefly luciferase activity from pBTdel-130 was normalized to R. reniformis luciferase activity from a co-transfected pRL-CMV plasmid. The relative activity of the reporter is shown as the percentage of activity from a parallel set of plates transfected with pCMV-Luc. Each data point represents the average of three independent transfections. The experiment was repeated and the same results were obtained. The status of endogenous hTERT expression in each cell type is indicated below the bar graph.
ase-positive and -negative cell lines, such as 3C167b, 3C4C, 3A#96, and 3A/hTERT cells (data not shown). These results again indicate that activities of transiently transfected reporters do not correlate with endogenous hTERT promoter activity, i.e. there was no repression of the hTERT reporters in pre-crisis cells or ALT cells. Therefore, the upstream sequences (Ϫ1642 to ϩ28 bp) examined do not contain any negative regulatory elements that could account for the suppression of endogenous hTERT in telomerase-negative cells.
Potential repressive elements could be located further upstream or downstream of the hTERT promoter. To search larger genomic regions for such elements, we have constructed additional reporters, pYF6 and pYF7, by replacing the hTERT translation initiation codon in pUC19-derived genomic DNA clones with a firefly luciferase expression cassette (Fig. 3B). pYF6, containing the sequence Ϫ1319 bp to ϩ6.5 kb relative to the transcription start site, comprises a 1.3-kb upstream sequence, exons 1 and 2, intron 1, and approximately half of intron 2. pYF7 (Ϫ7369 to ϩ1682 bp) includes about 7.4 kb upstream sequence, exon 1, intron 1, and most of exon 2. Therefore, both pYF6 and pYF7 contain a firefly luciferase open reading frame surrounded by sequences upstream and downstream of the hTERT initiation codon (Fig. 3B). Whereas pYF7 expressed 3-7-fold higher luciferase activity than pYF6 in all cells examined, the expression of neither reporter correlated with the status of endogenous hTERT expression (Fig.  3B). The upstream sequence (Ϫ7369 bp to Ϫ1320 bp) does not contain any cis-activating elements because a reporter containing the hTERT sequence Ϫ1319 to ϩ1682 bp expressed a level of luciferase activity similar to pYF7 (data not shown). It remains to be resolved whether any negative regulatory elements are present in the second intron sequence (ϩ1683 bp to ϩ6.5 kb). Regardless, such elements are not sufficient for the differential regulation of hTERT reporters in telomerase-positive and -negative cells. Therefore, our results indicate that the 14-kb region (Ϫ7369 bp to ϩ6.5 kb) at the 5Ј side of the hTERT gene does not contain cis-regulatory elements capable of repressing the hTERT promoter in telomerase-negative cells in transiently transfected reporter assays.
As previously reported, overexpression of the proto-oncogene c-myc could induce telomerase expression (15). It has been suggested that E-boxes, the consensus binding sites for c-Myc and its related proteins, are directly involved in the regulation of hTERT expression (16,18,24,26). However, recent reports showed that the expression of several Myc family proteins, including c-Myc itself, did not correlate with the expression of endogenous hTERT transcription (25,27). In our studies, both pYF6 and pYF7 contain the two canonical E-boxes (CACGTG) at Ϫ165 and ϩ44 bp, respectively, in the proximal region. In addition, pYF6 also included a minisatellite sequence within the intron 2 sequence, consisting of 40 -100 copies of a 42-bp imperfect repeat (31). Interestingly, most of these repeats have a CACGTG canonical E-box and therefore this minisatellite region contains a large cluster of E-boxes (Fig. 3B). Based on restriction fragment analysis, we estimated that there were approximately forty 42-bp repeats in the pYF6 reporter (data not shown). Taken together, our results from the reporter assays indicate that neither the proximal E-boxes nor the E-box cluster in intron 2 are able to confer sufficient regulatory control to allow the transiently transfected reporters to mimic endogenous hTERT transcription. Additional distal negative regulatory elements outside of the 14-kb region may be required for the regulation of endogenous hTERT expression in telomerase-negative cells. It is also possible that the endogenous hTERT promoter may be suppressed in telomerase-negative cells by a mechanism that involves the endogenous chromatin environment.
Association of hTERT Expression with Chromatin Structure Alteration at the Endogenous Promoter-The methylation status of CpG dinucleotides has been shown to play an important role in the regulation of chromatin structure and gene expression (32). Although a CpG island has been identified at the 5Ј end of the hTERT gene, spanning from Ϫ900 bp into exon 2 (33), previous studies, using a variety of normal as well as immortalized cells and cancer cells, did not reveal a correlation between hTERT expression and CpG methylation (34,35). To determine the possible role of CpG methylation in the regulation of hTERT expression during immortalization, genomic DNA was isolated from pre-crisis cells, immortal telomerase-

FIG. 3. Analysis of hTERT promoter activity from reporters carrying various promoter sequences in pre-crisis and immortal cells.
A, luciferase activity from reporters containing hTERT upstream promoter sequences. These promoter sequences, shown on the left, were inserted into the vector pGL3-basic (17). Relative luciferase activity is shown as the percentage of activity from pCMV-Luc. B, luciferase activity from reporters pYF6 and pYF7 containing large hTERT promoter fragments. The structure of the hTERT genomic DNA is shown at the top. Black rectangles represent exons. Canonical E-box consensus sequences are represented by vertical lines below the diagrams. Numbers are nucleotide positions relative to the transcription start site. The firefly luciferase expression cassette was inserted into the hTERT initiation codon. pA, SV40 poly(A) signal. The lower part shows the luciferase activity of these two reporters relative to pGL3pro⌬, a pGL3-promoter (Promega) derivative that contains a defective SV40 origin of replication. The status of endogenous hTERT expression in the cell lines is indicated below the bar graph. Each data point represents an average of three independent transfections. The experiments were repeated and one representative experiment is shown. expressing cell lines 3C104a and 3C167b, and ALT lines 3A#96, 3C4C, and 3C166a. The genomic DNA was digested with both the CpG methylation-sensitive restriction enzyme HpaII and its methylation-insensitive isoschizomer MspI and analyzed on Southern blots. No consistent difference in overall or site-specific methylation patterns within the CpG island was observed between hTERT-expressing and non-expressing cells (data not shown). Consistent with this, we did not detect any induction of hTERT mRNA in telomerase-negative cells upon treatment with 10 M 5-azacytidine, a DNA methyltransferase inhibitor (data not shown). Therefore, CpG methylation within this CpG island does not seem to be essential for the silencing of the hTERT promoter in telomerase-negative cells.
One hallmark of induction of gene expression is the alteration of nuclease accessibility of the promoter, which reflects remodeling of the nucleosomal architecture. To understand the chromosomal context of hTERT gene regulation, the chromatin conformation of the hTERT promoter was examined by DNase I hypersensitivity assays. Nuclei from both telomerase-positive and -negative lines were prepared and digested with increasing concentrations of DNase I. Genomic DNA was isolated from these DNase I-treated nuclei, digested with EcoRI and SphI, and analyzed on Southern blots. As shown in Fig. 4, with probe a, which hybridized to the 5Ј end of the 5.6-kb restriction fragment, a major DNase I hypersensitivity band (3.9 kb) was detected on the hTERT promoter in telomerase-positive 3C167b cells but not telomerase-negative 3C166a cells. With probe b, which hybridized to the 3Ј end of the fragment, a 1.7-kb hypersensitive band was detected in 3C167b cells but not 3C166a cells (Fig. 4B). The sizes of these bands (3.9 and 1.7 kb from probes a and b, respectively) indicate that a major DHS is centered around the hTERT transcription start site (Fig. 4A). We have detected the same DHS on the hTERT promoter in telomerase-positive 3C104a cells but not in telomerase-negative pre-crisis cells or ALT 3A#96 and 3C87a cells (data not shown). The appearance of these hypersensitive bands coincided with the gradual disappearance of the fulllength 5.6-kb restriction fragment, and the bands eventually faded away at higher concentrations of DNase I. The size of the 1.7-kb hypersensitive band decreased slightly with increasing concentrations of DNase I (Fig. 4B), suggesting that a relatively broad region was hypersensitive to the nuclease digestion. These experiments indicate that a change in chromatin conformation at the hTERT promoter accompanies endogenous hTERT transcription, and this change may play an important role in the regulation of hTERT expression.
Reversible Induction of hTERT Expression-Covalent modifications of histone proteins, such as acetylation, have been implicated in the modulation of chromatin structure (36). To test whether the hTERT promoter is silenced by histone deacetylation in telomerase-negative cells, pre-crisis 3C cells and the ALT cell line 3C166a were treated with the HDAC inhibitor TSA. The expression of hTERT mRNA was measured by duplex RT-PCR. A pair of primers specific for the CRR9 gene, located 5Ј of the hTERT gene, was used to generate an internal control band in these RT-PCR reactions. As was demonstrated earlier, the CRR9 gene was expressed in both telomerase-positive and -negative cells (Fig. 1D). As shown in Fig.  5A, hTERT message was not present in 3C166a cells (lane 1). However, it was readily detected after cells were treated with 1 or 5 M TSA for 24 h (lanes 3 and 4). Similar results were obtained when pre-crisis 3C cells and other ALT cells were used (data not shown). The induction of hTERT by TSA was confirmed by directly measuring telomerase activity (Fig. 5B). These experiments indicate that histone deacetylation is involved in repressing hTERT expression in telomerase-negative cells and that relief of this repression leads to the expression of hTERT mRNA.
To test if the induction of hTERT promoter activity by TSA was reversible, pre-crisis cells were cultured in the presence of TSA for 24 h, followed by incubation in medium without TSA for 24 and 48 h. As shown in Fig. 5C, the expression of the hTERT message virtually disappeared 24 h after TSA removal, indicating that induction of hTERT expression by TSA was reversible in pre-crisis cells. In 3C166a ALT cells, the TSAinduced hTERT transcription was also reversible (data not shown). Therefore, expression of the hTERT gene induced by TSA was not associated with a permanent alteration of the chromatin structure at the hTERT promoter. These results suggest that repression is the default state of the hTERT promoter in telomerase-negative cells, and that hTERT expression can be induced even in this silenced state.
Evidence for Direct Derepression of the hTERT Promoter by TSA-The derepression by TSA described above could be a direct result of HDAC inhibition at the hTERT promoter. Alternatively, TSA could induce an unknown transcription activator that in turn activates the hTERT promoter. In the latter case, de novo protein synthesis would be required for the hTERT induction. To determine whether hTERT induction by TSA treatment requires new protein synthesis, pre-crisis cells and ALT cells were treated with TSA in the presence or absence of the protein synthesis inhibitor cycloheximide (10 g/ ml). Fig. 5D shows a representative RT-PCR experiment using pre-crisis 3A cells; similar results were obtained using several telomerase-negative ALT cell lines (data not shown). The hTERT transcript was not induced until 12 h after the start of TSA treatment. Co-treatment with cycloheximide neither reduced nor delayed the hTERT expression. Instead, the hTERT mRNA levels were higher in the presence of cycloheximide. The super-induction of hTERT mRNA by TSA and cycloheximide has been reported recently (37) and is consistent with the possibility that a labile repressor contributes to the inactivation of the hTERT promoter in telomerase-negative cells. Alternatively, inhibition of translational elongation by cycloheximide might stabilize hTERT mRNA and therefore increase its steady-state level (38). Our data suggest that induction of hTERT by TSA does not require de novo protein synthesis but is more likely a direct effect of HDAC inhibition at the hTERT promoter.
To determine whether TSA-induced hTERT transcription correlates with any chromatin changes at the hTERT promoter, nuclei isolated from TSA-treated telomerase-negative cells were subjected to a DNase I hypersensitivity assay. As an internal control, the DNase I sensitivity of the CRR9 promoter was also determined. The 3.9-kb CRR9 restriction fragment contained sequence between Ϫ188 (SphI) and ϩ3708 bp (EcoRI) relative to its transcription start site. As shown in Fig.  6, chromatin became more sensitive to nuclease digestion upon TSA treatment, suggesting that the hTERT promoter adopted a more open conformation upon HDAC inhibition. This open conformation may have permitted transcription from the hTERT promoter in both pre-crisis cells and 3C166a ALT cells. In pre-crisis cells, TSA treatment resulted in the formation of a major DHS near the hTERT transcription start site, and cotreatment of cycloheximide did not block the DHS formation (Fig. 6A). These data suggest that inhibition of histone deacetylation opened up the chromatin and allowed pre-existing transcription factors to assembly on the promoter. However, we did not observe such a DHS in TSA-treated 3C166a cells (Fig. 6B). Because DNase I hypersensitivity assays detect only relatively strong DHSs, it is unclear whether a similar but weaker chromatin alteration also occurred in 3C166a cells following TSA treatment but failed to be detected. As shown in Fig. 6, TSA treatment also increased the general DNase I sensitivity of the CRR9 promoter in both cell types, suggesting that the chromatin opening induced by TSA treatment was a rather nonspecific event and occurred in a broad region. Our results suggest that the hTERT promoter is repressed in pre-crisis and ALT immortal cells and that chromatin opening is an important step in the induction of hTERT transcription. DISCUSSION Extensive studies of the regulation of hTERT expression have revealed that the hTERT gene is subject to both positive and negative regulation. It is repressed in the majority of normal somatic cells and becomes activated in most immortal cell lines and cancer cells (22,37,39). However, conflicting results were obtained by comparing unrelated normal cells, immortal cell lines, and cancer cells. To create a better model to study the molecular mechanisms underlying telomerase regulation during immortalization, we have generated a set of human fibroblast cells with similar genetic backgrounds, the precrisis cells and their independently derived telomerase-positive and -negative immortal progeny lines.
Using Northern and real-time RT-PCR analyses, hTERT mRNA was readily detected in telomerase-positive immortal cells (Figs. 1D and 2A). In telomerase-negative pre-crisis cells and ALT lines, however, telomerase activity and hTERT mRNA were undetectable but could be induced by treating cells with HDAC inhibitor TSA. These results are consistent with those of previous studies and indicate that hTERT expression is transcriptionally regulated (9,25). Furthermore, hTERT expression is accompanied by a general increase in chromatin sensitivity to DNase I digestion and/or appearance of a major DNase I hypersensitivity site near the hTERT transcription start site. Together, our data are consistent with the hypothesis that the endogenous hTERT promoter in telomerase-negative cells is repressed by mechanisms involving local chromatin configuration and that cellular immortalization by telomerase activation may involve a relief of such repression.
In contrast to endogenous hTERT expression, transiently transfected hTERT promoter reporters containing sequences within a 14-kb region from Ϫ7369 bp to ϩ6.5 kb were active in both telomerase-positive and -negative cells (Figs. 2B and 3).
These results indicate that this 14-kb region, the largest examined thus far, does not contain regulatory elements that function to repress the hTERT promoter in transient reporter assays. Ducrest et al. (25) recently showed that reporters containing up to 7.4 kb of sequence upstream of the hTERT initiation codon were active in two unrelated immortal fibroblast lines, the telomerase-negative ALT line GM847 and the telomerase-positive line GM639. More importantly, the microcell-mediated transfer of chromosome 3 from a normal cell into the human breast carcinoma cell line 21NT suppressed endogenous hTERT expression, but failed to repress the transiently transfected hTERT promoter reporters (25). In contrast, another recent report by Horikawa et al. (27) indicated that an extra copy of chromosome 3 was able to repress both endogenous hTERT expression and the transiently transfected reporters in the telomerase-positive renal cell carcinoma line RCC23. Whereas this discrepancy remains to be resolved, our results are consistent with the results published by Ducrest et al. (27) with respect to the activities of transiently transfected reporters.
Previous studies have suggested a role for the two proximal E-boxes in the regulation of hTERT promoter (16,17,23). Of particular interest is the presence of a large cluster of E-box sites in a minisatellite sequence of intron 2. It was proposed previously that this E-box cluster might participate in hTERT regulation (16,31). However, we found that the presence of neither the two proximal E-boxes nor the downstream E-box cluster was linked to proper hTERT promoter regulation in transient reporter assays.
One explanation for the lack of correlation between the transiently transfected reporters and the endogenous hTERT promoter is that the specific transcription factors required for hTERT transcription are not limiting but that their access to the proximal region of the hTERT promoter is prohibited by the endogenous chromatin configuration. Non-replicating plasmids in transfected cells have been reported to form very different nucleosomal structures compared with native nuclear chromatin (40). It is thus conceivable that transcription factors can readily bind to the "naked" hTERT promoter on a plasmid, but not to the "compacted" endogenous promoter. This hypothesis is further supported by the finding that TSA could induce hTERT expression in telomerase-negative cells (Figs. 5 and 6).
The interplay between histone acetyltransferases and HDACs has been suggested to be critical in transcriptional regulation. The accessibility of a promoter to various nuclear factors is strongly influenced by the dynamic balance between competing acetylation and deacetylation reactions (41). Consistent with the repression model, the current study, along with other recent reports, showed that the HDAC inhibitor TSA could activate hTERT expression in telomerase-negative cells (18,19,35,39). Although the reversible induction of hTERT expression by TSA suggests that repression requires histone deacetylation, the possibility of indirect effects on the hTERT promoter upon TSA treatment cannot be ignored. As shown in Figs. 5D and 6, our results indicate that this induction does not require de novo protein synthesis and is likely a result of HDAC inhibition at the hTERT promoter because the TSA-induced hTERT expression and alteration of chromatin configuration were not blocked by the protein synthesis inhibitor cycloheximide.
Upon TSA treatment, the chromatin became more sensitive to nuclease treatment, indicating that the hTERT promoter adopted a more open conformation (Fig. 6). In pre-crisis cells, the opening of chromatin was associated with the appearance of a major DHS at the same position as the DHS identified in telomerase-expressing immortal cells. The DHS also appeared in the presence of cycloheximide, suggesting that pre-existing factors were able to assemble a functional transcription complex at the hTERT promoter following HDAC inhibition. In 3C166a ALT cells, the TSA-induced chromatin opening and hTERT expression, however, were not accompanied by the appearance of the DHS. Both the hTERT and CRR9 promoters seem to be more resistant to DNase I digestion in 3C166a cells than in pre-crisis cells (Fig. 6). Using the same blots, we found that chromatin at another locus (a gene encoding a human homolog of the Polycomb group protein) was equally sensitive to DNase I treatment (data not shown). This excluded the possibility that the differential chromatin nuclease sensitivity at the hTERT/CRR9 loci between pre-crisis cells and 3C166a ALT cells was because of variations in individual nuclei preparations. Furthermore, the level of hTERT transcripts induced by TSA was also lower in 3C166a cells (data not shown), suggesting that the hTERT locus is in a more repressive chromatin environment in 3C166a cells. Therefore, it may be difficult to detect TSA-induced alterations in nucleosomal configuration in such a repressive chromatin environment in these cells using DNase I hypersensitivity assays. A possible explanation for the more repressed chromatin state in 3C166a cells is that this entire chromosomal region is under the influence of telomere position effects (42). ALT cells have longer telomeres than pre-crisis cells (Fig. 1C) and the hTERT locus is located at the tip of chromosome 5p. Thus, it is conceivable that telomere position effects may contribute to hTERT repression in 3C166a cells (43). Taken together, our data are consistent with the hypothesis that the endogenous hTERT promoter is repressed by a chromatin-mediated mechanism in telomerase-negative cells.
DHSs generally result from either topologically altered DNA arising from sequence-specific DNA-protein interactions or nucleosome-free regions resulting from ATP-dependent nucleosome remodeling. The formation of a DHS at the transcription start site as a result of nucleosomal remodeling during transcriptional activation of the interferon-␤ promoter has been well characterized (44). An enhanceosome is first assembled in response to virus infection, leading to the recruitment of SWI/ SNF nucleosome remodeling complexes to the interferon-␤ promoter. SWI/SNF modifies histone-DNA contacts, allowing the binding of TBP (TATA-binding protein) to the TATA box and resulting in DNA bending. The TBP-induced DNA bending in turn causes the downstream nucleosome, which normally obstructs transcription, to slide further downstream and thereby exposes the TATA box and the transcription start site to different nuclear factors (44).
A similar chain of events may also occur at the hTERT promoter during the activation of hTERT transcription. The hTERT core promoter lacks a TATA box but instead contains an array of five GC boxes surrounded by two E-boxes (23,33). This arrangement is reminiscent of the TATA-less promoters found in many "housekeeping" genes (45). GC boxes, the binding sites for the constitutively expressed Sp1 transcription factors, are essential for promoter activity in many TATA-less promoters including the hTERT promoter, where mutations of these GC boxes completely abolished hTERT promoter activity in transient reporter assays (39). Furthermore, the binding of Sp1 family proteins induces an asymmetric bend in DNA (46). Sp1 binding may cause DNA bending and nucleosome sliding, consequently resulting in the formation of a DHS at the hTERT transcription start site.
Previous studies showed that the binding of Sp1 to GC boxes was severely affected by local chromatin structure (47). The affinity of Sp1 for its cognate binding site was diminished upon the formation of nucleosomes on naked DNA (48) but enhanced by the presence of SWI/SNF remodeling complexes (49). It is possible that SWI/SNF complexes may not be recruited to the hTERT promoter in telomerase-negative cells. As a result, Sp1 proteins may not have access to the GC boxes at the promoter. Because recruitment of SWI/SNF and stabilization of remodeled nucleosomes require histone acetyltransferase (44), inhibition of HDAC may shift the balance in favor of SWI/SNF recruitment, leading to the establishment of a DHS and activation of hTERT transcription. Therefore, it will be interesting to determine whether DNA bending and nucleosome repositioning occur at the hTERT promoter.
In conclusion, using a set of closely related pre-crisis cells and immortal lines, the present study provided evidence that the endogenous hTERT promoter was suppressed in telomerase-negative cells. The hTERT transcription in telomerasepositive cells is correlated with the appearance of a major DHS at the transcription start site. Inhibition of HDAC by TSA led to chromatin remodeling, as indicated by establishment of a DHS and activation of hTERT transcription in telomerasenegative cells. Our data suggest that derepression of the hTERT promoter and chromatin remodeling contribute to hTERT transcription during fibroblast cell immortalization. Further investigation of the chromatin structure and transcription factor recruitment at the endogenous hTERT promoter will shed light on the interaction between transcription factors and local chromosomal environment.