Regulation of Human Telomerase Activity: Repression by Normal Chromosome 3 Abolishes Nuclear Telomerase Reverse Transcriptase Transcripts but Does Not Affect c-Myc Activity¹

Telomeres are specialized DNA-protein complexes at the end of eukaryotic chromosomes that protect chromosome ends from fusion and degradation (1, 2, 3, 4). The complete replication of telomeric DNA requires a specialized reverse transcriptase, telomerase (5, 6). Most normal somatic human cells lack this enzyme (7) and their telomeres shrink with each replication cycle by ∼30–100 bp (2, 8, 9). Because short telomeres induce cellular senescence in tissue culture (10), it has been proposed that telomere shortening may limit the replicative potential of normal cells providing a powerful tumor-suppressive mechanism (11). Cells of the germ line and certain stem cells, as well as 85% of tumor-derived immortal cells, contain telomerase, and their telomere length is stabilized (7). In a minority of tumor cells, however, an alternative non-telomerase-dependent mechanism (ALT) is responsible for telomere stabilization (12).

Telomerase is a ribonucleoprotein (RNP) enzyme that consists of an RNA moiety and several protein subunits. Of these, the RNA moiety and the catalytic subunit are essential for telomerase activity in vitro. The RNA subunit contains a short segment that serves as the template for telomeric repeat synthesis (13, 14, 15, 16). The catalytic protein subunit (hTERT)⁴ is related structurally and functionally to reverse transcriptases (17, 18, 19, 20, 21). Among the number of telomerase-positive and -negative cells thus far examined, the presence of hTERT mRNA is related to the presence of telomerase activity (19, 20). In contrast, the telomerase RNA subunit and other components implicated in telomere maintenance are present in both telomerase-positive and -negative cells. Furthermore, ectopic expression of hTERT in telomerase-negative fibroblasts or endothelial cells is sufficient to restore telomerase activity and to stabilize telomere length (10, 22, 23), whereas overexpression of dominant-negative mutants of hTERT in tumor cells can inhibit telomerase activity and induce growth arrest (24, 25).

The mechanisms that control hTERT gene expression may involve transcriptional regulation, RNA stability, processing and/or export to the cytoplasm. To date, a number of regulators of hTERT expression have been identified including the Wilms’ tumor suppressor gene (WT1) product that reduces hTERT RNA levels in 293 kidney cells (26). Retinoids were shown to down-regulate hTERT RNA in acute promyelocytic leukemia (27). Several activators of hTERT expression have also been identified. Estrogen induces hTERT RNA in estrogen receptor-positive cells (28, 29). The E6 oncoprotein of human papillomavirus type 16 induces telomerase activity in epithelial cells, but not in fibroblasts (30, 31). c-Myc directly acts on the hTERT gene, inducing hTERT expression (32, 33), whereas the c-Myc antagonist Mad down-regulates the expression of hTERT (34, 35). hTERT regulation involves histone acetylation because treatment of telomerase-negative cells with trichostatin A activates telomerase (36).

In cell hybridization experiments, the telomerase-negative state behaves like a dominant trait, given that hybrids between telomerase-positive and -negative cells are telomerase negative (37). Microcell transfer of human chromosomes 2, 7, 11 induced cellular senescence in some tumor-derived cells. However, telomere length and telomerase activity are retained in these cells, implying that several inducers of senescence function independently of telomerase (38). By microcell transfer of human chromosomes into breast and kidney tumor cell lines, a factor that directly or indirectly down-regulates telomerase activity has been mapped to a region on chromosome 3p (39, 40, 41). Recently, it was shown that transfer of human chromosome 6 into a HPV16-immortalized keratinocyte cell line (FK16A) and into a HPV16-containing cervical cancer cell line (SiHa) reduced hTERT RNA levels (42). Similar results were obtained after introduction of a fragment of human chromosome 10p into hepatocellular carcinoma cells (Li7HM; Ref. 43). It was also shown that transfer of human chromosomes 3 or 4 into HeLa cells abolished telomerase activity (44).

RNA processing has also been implicated in the regulation of hTERT. Several splice-variants of hTERT RNA that encode enzymatically inactive telomerases are expressed during embryonic development and are also detectable in some immortalized cells (45, 46, 47).

To study hTERT expression, we developed a quantitative RT-PCR assay and measured spliced and unspliced hTERT RNA levels in primary cells and immortal cell lines. We found that low levels of hTERT RNA are expressed in tumor cells, and that the level of immature nuclear hTERT RNA correlates with telomerase activity, which suggests a regulation of hTERT RNA levels in the nucleus. In addition, we demonstrated that reporters containing up to 7.4 kb of 5′ flanking region do not faithfully mimic expression of the endogenous hTERT gene. We showed that transfer of a normal chromosome 3 into the human breast cancer cell line 21NT results in complete silencing of endogenous hTERT (indicated by an absence of immature nuclear hTERT RNA) even in cells that are rescued from senescence by ectopic expression of hTERT cDNA construct. Moreover, we characterized the mechanism by which chromosome 3 represses hTERT RNA expression in the breast cancer cell line 21NT. We have provided evidence that the repressor does not act on regulatory elements in the immediate 5′ flanking region of the gene, and is independent of c-Myc or its coregulators.

HLFs (passage 6) were a gift from Urs Ziegler, Institute of Anatomy, University of Zurich (Zurich, Switzerland). The HT1080 fibrosarcoma-derived line was a gift from Ian Kerr, Imperial Cancer Research Fund, London. SV40-transformed telomerase-positive human fibroblasts GM639 were obtained from the Coriell Institute for Medical Research (Camden, NJ). SV40-transformed telomerase-negative human fibroblasts GM847 were obtained from Roger Reddel, Children’s Medical Research Institute (Sydney, Australia). HeLa cells were obtained from Beatrice Bentele, ISREC (Epalinges, Switzerland). SW480, a colon adenocarcinoma cell line, was obtained from Richard Iggo, ISREC. The above cells were maintained in high glucose DMEM with 10% FCS. The breast carcinoma cell line 21NT and its derivatives 21NT pCineohTERT (parental) and 21NT pCineohTERT HyTkchromosome 3 (21NT-chromosome 3 hybrids) were cultured as described previously (39). EREB 2–5 were obtained from Georg W. Bornkamm, GSF (Munich, Germany) and were cultured as described previously (48). HaCaT human adult skin keratinocytes (49) were obtained from Stephanie Lathion, ISREC, and were maintained undifferentiated, in medium A (1:3, DMEM to HAM-F12) containing 0.6 mm CaCl₂, 5% FCS, 8.3 ng/ml cholera toxin, 5 μg/ml insulin, 24 μg/ml adenine, 0.5 μg/ml hydrocortisone, and 10 ng/ml EGF. After growth to confluence, the cells were induced to differentiate in medium A containing 1.2 mm CaCl₂, 20% FCS, 8.3 ng/ml cholera toxin, 5 μg/ml insulin, 24 μg/ml adenine, and 0.5 μg/ml hydrocortisone for 14 days. For measuring RNA stability, HT1080 cells were treated with 2 μg/ml actinomycin D for 0.5–8 h. HLF-hTERT cells were generated by infection of HLF cells with pMSCV-puromycin-hTERT (50). HLF-cMyc cells were similarly generated using pBabe-puromycin-cMyc obtained from Bruno Amati (DNAX Research Institute, Palo Alto, CA; Ref. 51). Infections were performed as described previously (50).

pGRN121 contains hTERT cDNA (20) and was obtained from Geron Corporation, Menlo Park, CA. pNSV4 contains a genomic hTERT insert (Accession no. AF114847) encompassing 7.4 kb of the 5′ flanking region upstream of the hTERT translation start site, the first two exons, and part of the second intron (33). pSV2-Thy-1 expresses the mouse Thy-1.1 allele under the control of the SV40 enhancer and early promoter (52). We constructed hTERT-GFP plasmids using the following procedures. To generate the GFP reporter vectors: pG (basic vector), pSVG (promoter vector), and pSVEG (promoter/enhancer vector), we replaced the HindIII/XbaI fragment containing the luciferase gene of pGL3 (Promega) with the HindIII/XbaI fragment of pEGFP-N1 (Clontech) containing the EGFP gene. phTERT.1.3G contains a 1.3-kb fragment upstream of the translation start site of the hTERT gene. The 1.3-kb fragment was amplified from pNSV4 by PCR using oligonucleotides P1328f and P1r (see below in “DNA Oligonucleotides”) and subcloned into the NheI/BglII sites of the promoterless GFP vector. phTERT.5.1G, containing 5.1 kb of upstream sequence, was generated by cloning a 3.8-kb SacI/NheI fragment of pNSV4 into the SacI/NheI sites of phTERT.1.3G. phTERT.7.4G was generated by cloning a 2.3-kb SacI fragment of pNSV4 into the SacI site of phTERT.5.1G. phTERT.4.8G was generated by religation of phTERT.7.4G after digestion with SpeI, deleting a 2.6-kb fragment from the 5′ end of the hTERT promoter. phTERT.3.3G was generated by subcloning a 3.3-kb XhoI/HindIII fragment of phTERT.5.1G into pG. phTERT.4.4G was obtained by cloning a 1.1-kb SacI/XhoI fragment generated by PCR with primers P3061f and P4183r into the SacI/XhoI sites of phTERT.3.3G. We generated phTERT.0.9G, phTERT.0.6G, and phTERT.0.3G like phTERT.1.3G, except that oligonucleotides P951f, P602f, and P314f, respectively, were used as forward primers. To generate phTERT.1.3Δ0.1G and phTERT.1.3Δ0.3G, 108 bp and 160 bp, respectively, of the 3′ end of the 1.3-kb insert of phTERT.1.3G were removed by PCR using oligonucleotides P1328f and P108r or P260r, respectively.

The following DNA oligonucleotides were purchased from Microsynth (Balgach, Switzerland) and used for hTERT reporter constructs.

P1r: 5′-GGAACTAGTAGATCTCGCGGGGGTGGCCGGGG-3′; P108r: 5′-GGAACTAGTAGATCTGGGAGGCCCGGAGGGG-3′;P260r: 5′-GGAACTAGTAGATCTGTGCCCGCGAATCCACTG-3′; P314f: 5′-GGAGGATCCGCTAGCAGCTGCGCTGTCGGGG-3′; P602f: 5′-GGAGGATCCGCTAGCGCCTTCGTCCTCCCCTTC-3′; P951f: 5′-GGAGGATCCGCTAGCGGGCGGGATGTGACCAG-3′; P1328f: 5′-GGAGGATCCAGGGAGGGTGCGAGGCC-3′); P3061f: 5′-CATTTCCAGGAGCTCCCCGTCTC-3′; P4181r: 5′-TTGCAGGCCTGGGCTCGAGGC-3′.

Transient transfections with calcium-phosphate precipitates were performed according to the protocol described by Jordan et al. (53). Cells were cotransfected with 1 μg of pSV2-Thy-1.1 as reference plasmid and equimolar amounts of GFP reporters. The total amount of plasmid was kept constant (6 μg) by adding pUC19.

Transfected cells were harvested 48 h after transfection by trypsinization and were incubated for 30 min with a saturating concentration of either monoclonal phycoerythryn (PE)-labeled antimouse CD90.1 (Thy-1.1) antibody OX-7 (PharMingen, San Diego, CA) or APC-labeled anti-Thy-1 antibody III-5 (54) kindly prepared by Céline Maréchal, and washed once. We analyzed the cells on a FACS-scan or FACScalibur microflow cytometer (Becton Dickinson, Franklin Lakes, NJ). hTERT reporter gene expression was quantified by calculating the equivalent to the value used for enzyme reporter systems. For this, we considered the arithmetic mean of GFP expression and Thy-1 fluorescence. We subtracted the GFP background, obtained with cells transfected only with pSV2-Thy-1, from the arithmetic mean of GFP expression. To correct for transfection efficiency, this value was divided by the equivalent measure for Thy-1 expression in the same cells. The GFP expression of the reporter constructs was normalized to that of a plasmid containing the SV40 minimal promoter driving GFP expression (pSVG). The GFP-reporter assay will be described in detail elsewhere.⁵

Total RNA was extracted from different cell lines using the RNAeasy minikit (Qiagen). The quality of the RNA was determined on agarose gel electrophoresis. RNA was quantified with spectrophotometry at 260 and 280 nm (1 A_{260 nm}, ∼40 μg/ml). To perform RT-PCR with primer pairs that are not located in different exons or to quantify intron 2-containing hTERT RNA, a DNase I treatment was performed before the reverse transcription. Four μg total RNA was incubated in 20 μl with 10 units of DNase I (Roche Diagnostics Ltd) in 10 mm Tris-HCl (pH 8.0), 0.5 mm MgCl₂, 1 mm dithiotreitol, and 0.2 units/μl RNasin (Roche Diagnostics Ltd) for 1 h at 37°C, followed by 10 min at 65°C to inactivate the enzyme. We reverse-transcribed 100 ng of RNA in 20 μl using 100 ng of random hexamer primers and 20 units of MMLV-RT (Life Technologies, Inc.) according to the manufacturer’s protocol. To quantify intron 2-containing hTERT RNA, 1 μg of DNase I treated RNA was reverse transcribed as above using 10 pmol of primers 13156rv (E2-I2) and 10 pmol of primers 3407rv (GAPDH), respectively (Table 1). Quantitative PCR was performed using an ABI Prism 5700 Sequence Detection System (Perkin-Elmer Applied Biosystems, Foster City, CA). For each PCR run, a master mix was prepared with 1 × TaqMan master mix or 1 × SYBERGreen master mix (5.5 mm MgCl₂, 200 μm dATP, 200 μm dCTP, 200 μm dGTP, 400 μm dUTP, 0.01 units/μl AmpErase UNG and 0.025 units/μl AmpliTaq Gold DNA Polymerase ± SYBR Green I dye; Perkin-Elmer Applied Biosystem), 0.3 μm of each primer, and 0.1 μm TaqMan probe. Reverse transcription reaction (2.5 μl) was added to 22.5 μl of master mix. The thermal cycling conditions included an initial denaturation step at 95°C for 10 min followed by 50 cycles at 95°C for 15 s and 60°C for 1 min. Primers and probes for all of the RT-PCRs were chosen with the assistance of the computer program Primer express (Perkin-Elmer Applied Biosystems). BLASTN searches were used to check the gene specificity of the nucleotide sequences chosen for the primers and probes. PCR products were fractionated on agarose gel to confirm that their size corresponded to the expected length. Primers were purchased from Microsynth (Balgach, Switzerland), and TaqMan probes were from Eurogentec (Les Ulis Cedex, Belgium). To test the efficiency of the PCR primers, we carried out reactions with different concentrations of the appropriate template hTERT DNA (pGRN121, pNSV4) or GAPDH cDNA and plotted the cycle number at which the PCR signal rises above background (Ct) against the logarithm of the number of template molecules (L). The regression lines and correlation coefficients obtained were Ct = −3.29 L + 42.20 (R², 0.997); Ct = −3.21 L + 39.59 (R², 0.998); Ct = −3.62 L + 42.23 (R², 0.997); Ct = −3.33 L + 35.40 (R², 0.991) for hTERT E4–5, E9–10, E2-I2, and GAPDH primers, respectively. To test the efficiency of hTERT cDNA, synthesis we used an in vitro transcript as template. Different quantities of this synthetic template were mixed with total RNA of telomerase-negative HLF cells and were reverse-transcribed. Comparison of Ct values obtained on RT-PCR of the synthetic hTERT RNA with the Ct values obtained with known numbers of plasmid molecules showed that the efficiency of cDNA synthesis was 25%. The amount of total RNAs obtained from different cells was measured by alkaline hydrolysis as described previously (55). Per million cells, the following μg amounts of total RNA were present: HLF and HLF-cMyc, 23–25 μg; GM847, 20 μg; HT1080, 35 μg; 21NT and 21NT-chromosome 3 hybrids, 21–23 μg; HeLa, 35 μg; SW480, 20 μg; and EREB, 30 μg.

Nuclear and cytoplasmic extracts were either prepared by hypotonic swelling according to Schreiber et al. (56) or by dounce homogenization as described by Mirkovitch et al. (57). The cytoplasmic fractions were then mixed (1:1) with the lysis buffer from the RNAeasy minikit (Qiagen), and the nuclear pellets were resuspended in the same lysis buffer.

Live HT1080 cells were stained with 10 μg/ml DAPI (Fluka) for 30 min at 4°C in PBS containing 0.05% Triton X-100 and were washed with PBS. Cells were sorted according to DNA content on a FACS-sorter microflow cytometer (Becton Dickinson) and collected in lysis buffer from the RNAeasy minikit (Qiagen).

Total protein from four independent cultures of subconfluent 21NT parental and 21NT-chromosome 3 hybrids were extracted with 8 m urea, 0.5% Triton X-100, and 0.5% NP40. Fifty μg of protein was resolved on a 10% SDS-polyacrylamide gel and transferred to nitrocellulose membranes. c-Myc protein was detected with a mouse monoclonal IgG against human c-Myc (9E10, 1/1000; Santa Cruz Biotechnology), Mad 1 with a rabbit polyclonal antibody (C-19, 1/200; Santa Cruz Biotechnology), and actin with a goat polyclonal antibody (I-19, 1/200; Santa Cruz Biotechnology). Western blots were developed using the enhanced chemiluminescence system (ECL; Amersham) for actin and the SuperSignal West Pico kit (Pierce) for c-Myc and Mad1.

To measure the amount of hTERT RNA in tumor-derived cell lines and primary cells, we developed a quantitative RT-PCR assay using three different hTERT primer pairs (Table 1). Two primer pairs, E4–5 and E9–10, spanned the boundary between exons 4 and 5 and between 9 and 10, respectively, and amplified cDNA only from spliced RNA lacking intron 4 and/or 9, whereas the third primer pair, E2-I2, amplified cDNA from immature hTERT RNA containing the end of exon 2 and 256 nucleotides of intron 2. For comparison, we determined the amount of intron-free GAPDH RNA. The number of cDNA molecules present in a sample was calculated by plotting the corresponding Ct value onto the regression line of the Ct values obtained when graded amounts of precisely quantified hTERT plasmid or GAPDH PCR product were amplified. Because the efficiency of cDNA synthesis from an in vitro transcript of hTERT in the presence of total RNA of telomerase-negative cells was 25% (data not shown), the number of intron-free hTERT RNA molecules was assumed to correspond to the quadruple of the estimated number of cDNA molecules. We obtained the same estimates of hTERT RNA molecule numbers using either the E4–5 or the E9–10 primer pairs (data not shown).

Telomerase-positive tumor-derived cell lines contained between 0.2 and 6 spliced hTERT RNA molecules/cell (Fig. 1,A). No signal above the detection limit of 0.004 molecules/cell was obtained in telomerase-negative primary HLFs or in the telomerase-negative cell line GM847 (Fig. 1,A). Therefore, if telomerase-negative cells express any spliced hTERT RNA at all, its level is at least 50–1500 times lower than that of telomerase-positive cells. Primary fibroblasts that were transduced with an hTERT-retroviral construct expressed at least 100 times more spliced hTERT RNA than tumor-derived cells, but their telomerase activity was not higher than that of tumor-derived cells (data not shown). Fibroblasts expressing c-Myc contained spliced hTERT RNA at levels comparable with those in some telomerase-positive tumor cells (0.2 molecule/cell; Fig. 1,A). Quantification of GAPDH RNA demonstrated that each cell contained between 700 and 15,000 molecules (Fig. 1 A). Thus, by comparison with GAPDH, hTERT is a very rare RNA species, detectable exclusively in telomerase-positive cells.

Although it is generally assumed that hTERT expression is regulated primarily at the level of transcription, there is little direct evidence for this. In support of this notion, overexpression of c-Myc can directly induce hTERT expression (32, 33). Furthermore, hTERT run-on transcription signals (34) changed during differentiation of human hematopoietic U937 cells. We have been unable to detect run-on transcription signals from the hTERT gene with cells used in this study,⁶ which contain only 0.2–6 spliced hTERT RNA molecules/cell. To substantiate the assumption that hTERT expression is regulated in the nucleus, we compared the amounts of various spliced and unspliced hTERT RNAs in the nucleus and the cytoplasm. Nuclear and cytoplasmic fractions of HT1080 cells were prepared according to two different protocols (56, 57). Both methods produced very similar results for hTERT RNAs as well as for control RNAs (GAPDH and U3 snRNA; Fig. 1,B). The two controls were included to monitor contamination between nuclear and cytoplasmic fractions. Such contamination was low inasmuch as only 4% of GAPDH RNA was found in the nuclear fraction, whereas 93–98% of the U3 snRNA appeared in the nuclear fraction (Fig. 1,B). However, 20–30% of the telomerase RNA compound hTER, which was previously thought to be mostly nuclear, was also detected in the cytoplasm. On the other hand, we found a considerable fraction (35–75%) of hTERT RNAs in the nucleus. This included transcripts that still retained intron 2 as well as molecules that lacked intron 4 and/or intron 9. However, hTERT-negative cells lacked both intron-containing and intron-less hTERT RNA (Fig. 1 A).

We also analyzed the stability of spliced and unspliced hTERT RNA in HT1080 cells treated with Actinomycin D. Half-lives of intron 9-less and intron 2-retaining hTERT RNAs were 2 h and 2.5 h, respectively (data not shown). A slightly shorter half-life of 50 min for hTERT mRNA was found in human hematopoietic U937 cells (34). Because no form of hTERT RNA was detected in telomerase-negative cells, we conclude that hTERT regulation occurs in the nucleus.

To determine whether RNA levels of hTERT fluctuate during the cell cycle, exponentially growing HT1080 cells were stained with DAPI and sorted by fluorescence flow cytometry (FACS) according to their DNA content (Fig. 2,A). The FACS sorting gates were set sufficiently narrow to minimize cross-contamination of cells from different phases of the cell cycle. Total RNA was extracted, and hTERT RNA levels were measured by quantitative RT-PCR. We found no significant differences between spliced hTERT RNA levels in the different phases of the cell cycle (Fig. 2 B).

Using EREB cells, we investigated whether the proliferative state of the cells affected hTERT RNA levels. EREB cells are EBNA2-immortalized B-lymphocytes in which the EBNA2 gene is expressed as a chimeric fusion with the hormone-binding domain of the estrogen receptor. Thus, proliferation of the cells depends on the presence of estrogen (48). As expected, the number of cells in G₁ increased (from 60 to 90%; not shown) when EREB cells were cultured for 24 h in the absence of estrogen. Estrogen deprival lead also to a 5-fold reduction of spliced and intron-containing hTERT RNAs (Fig. 1,A). A very strong decrease in hTERT RNA levels was seen on terminal differentiation of immortalized human skin keratinocytes (HaCaT). HaCaT cells remain undifferentiated when cultured in low calcium. The addition of calcium to confluent cells is sufficient to induce cell differentiation. After 6 days in high-calcium medium, spliced and unspliced hTERT RNA levels had dropped at least 300-fold compared with undifferentiated HaCaT cells (Fig. 1 A). These results demonstrate that, although hTERT RNA levels do not vary during different stages of the cell cycle in tumor-derived cells, the levels of hTERT RNA are significantly reduced by cell cycle arrest and/or terminal differentiation. Diminution of spliced hTERT RNA during cell differentiation has been previously reported for U937 and HL-60 hematopoietic cells (19, 34).

The results above are consistent with the hypothesis that hTERT RNA levels are regulated by transcription. To examine the regulatory role of the hTERT 5′ flanking region, DNA fragments upstream of the start codon were fused to the GFP gene (Fig. 3,A). Expression of these reporter constructs was quantified by two-color fluorescence flow cytometry (FACS; see “Materials and Methods”). As shown in Fig. 3 B, reporters containing up to 7.4 kb of hTERT 5′ flanking region expressed GFP in HT1080 cells. In telomerase-negative HLF cells, the reporters containing either the 1.3 kb or the 3.3 kb of 5′flanking region expressed low GFP fluorescence, whereas the other constructs did not express GFP at levels that were significantly above that of the background observed with a promoter-less construct (pG; data not shown). Removal of 260 nucleotides immediately upstream of the start codon (phTERT.1.3Δ0.3G) completely abolished the weak GFP expression in HLF cells and strongly reduced it in HT1080 cells. Similar results were obtained using telomerase-positive HeLa cells (data not shown). Thus, as expected, the hTERT-GFP reporters activated GFP expression in tumor-derived cell lines but not in telomerase-negative primary HLFs.

The reporters also induced GFP expression to similar levels in two SV40-transformed fibroblast lines (Fig. 3,C). Of these, GM639 is telomerase positive and contains detectable hTERT RNA (Fig. 1,A). GM847, on the other hand, is telomerase negative and contains neither intron 2-containing or intron 9-less hTERT RNA (Fig. 1 A). Thus, the hTERT-GFP reporter activity in these fibroblast lines did not mimic hTERT RNA expression.

We investigated whether the hTERT-GFP reporter constructs could be used to identify regulatory elements in a breast carcinoma cell line, 21NT. Introduction of a normal human chromosome 3 into these cells by microcell fusion represses telomerase activity (39). This indicates that cis-acting targets of repression in the 21NT hTERT genes are intact. Repression of telomerase activity by chromosome 3 transfer is mediated by down-regulation of the hTERT RNA (Fig. 1,A). We tested whether the hTERT-GFP reporter constructs would also be repressed by chromosome 3. To avoid cell senescence as a result of hTERT extinction (39), hTERT was ectopically expressed in the parental and in the hybrids cells. Fig. 3 D shows representative results for 4 independent hybrids of the 10 tested. As expected, we detected no intron 2-containing hTERT RNA in the 21NT-chromosome 3 hybrids. In contrast, GFP expression in the hybrids was the same as in the parental cells. We obtained similar results after stable transfection of the reporter constructs (data not shown). Therefore, it appears that the regulatory elements required for repression of hTERT in GM847 cells and 21NT-chromosome 3 hybrids are not contained within the 7.4-kb region upstream of the hTERT start codon or that they do not function properly when removed from their endogenous location.

Previous studies showed that overexpression of c-Myc can induce hTERT expression in telomerase-negative cells (32, 33, 58). Therefore, we tested whether hTERT RNA down-regulation in 21NT-chromosome 3 hybrids was associated with changes in the c-Myc pathway. Similar levels of c-Myc and Mad1 proteins were detected in parental and hybrid cells (Fig. 4,A). Furthermore, using quantitative RT-PCR, we found that parental and hybrid cells expressed the same c-Myc and Mad RNA levels, which indicated that c-Myc is not a target of the putative repressor on chromosome 3 (Fig. 4,B). To determine whether the chromosome 3 repressor would act on other genes or gene-products of the c-Myc regulatory network, we measured the expression levels of five known c-Myc target genes: CAD, ODC, GADD45, eIF4E, and LDHA (59). The RNA levels of CAD, ODC, GADD45, eIF4E and LDHA in the 21NT parental cells and chromosome 3-containing hybrids were very similar, whereas hTERT RNA levels dropped at least 30-fold (Fig. 4 B). We conclude that the repressor on chromosome 3 defines a regulatory pathway controlling hTERT expression that does not involve c-Myc.

In this report, the transcripts of the gene coding for the catalytic subunit of human telomerase were quantified in different telomerase-positive and -negative cells. Intron 9-less and intron 2-containing transcripts of hTERT were detected in telomerase-positive cell lines but not in telomerase-negative HLF and GM847 cells. These results provide support for the critical role of hTERT RNA regulation for telomerase activity. Our data indicate that on average, a telomerase-positive cell contains less than six spliced hTERT RNA molecules, whereas spliced and intron-containing hTERT RNA levels in telomerase-negative cells, if present, are below the limit of detection (0.004 molecule/cell). We found that intron 9-less and intron 4-less hTERT RNAs were predominantly cytoplasmic, whereas intron 2-containing hTERT RNA was mainly nuclear. The relative levels of both RNA species correlated well with each other in all of the telomerase-positive and negative cells examined. These data suggest that hTERT RNA levels are controlled mainly prior to exit from the nucleus, by changes either in the rate of transcription or in the stability of nuclear RNA.

The low hTERT RNA levels combined with the intermediate RNA stability suggest that the rate of hTERT transcription is low. Assuming a polymerization rate of 2000 nucleotides/min by RNA polymerase II (60) and a half-life for hTERT RNA of 2 h, we estimate that 1–2 RNA polymerase complexes are transcribing the 40-kb gene (47) at any given time. This low RNA polymerase II density on the hTERT gene is consistent with our inability to detect hTERT transcription by run-on analysis (not shown). However, successful run-on analysis was reported by Günes et al. (34) using human myeloid leukemia U937 cells. Comparison of their data with ours suggests that the rate of hTERT transcription in U937 cells is much higher than in HT1080 fibrosarcoma cells.

In previous reports, several hTERT RNA splice variants had been described that are differentially expressed during embryonic development and could also be detected in some immortal cell lines (45, 46, 47). The splice-variants cannot encode enzymatically active telomerase because critical regions in the reverse transcriptase domain are missing. Because no hTERT transcripts were detectable in the telomerase-negative cells tested here (HLF, GM847, and 21NT-chromosome 3 hybrids), telomerase repression is likely to involve mechanism(s) preceding alternative splicing.

In arrested EREB cells and in terminally differentiated HaCaT cells, we observed down-regulation of hTERT RNA, whereas no change was detected during the cell cycle in proliferating tumor cells. This is reminiscent of the RNA levels of other DNA polymerases (α, δ, and ε) in proliferating cells (61, 62). It is unclear what factors increase hTERT RNA levels in proliferating cells and whether the same factors mediate hTERT up-regulation in tumors. c-Myc is known to trigger hTERT transcription when overexpressed, and it is expressed in proliferating but not in arrested cells. Thus, c-Myc may contribute to the activation of hTERT transcription in proliferating EREB and HaCaT cells. However, the levels of c-Myc present in proliferating fibroblasts are not sufficient to induce hTERT expression. The c-Myc protein is expressed at higher levels in many tumors and may contribute to the activation of hTERT expression. However, transfer of normal chromosome 3 into the breast cancer-derived cell line 21NT repressed hTERT expression without affecting c-Myc or Mad levels or expression of c-Myc target genes. This indicates that the gene(s) on chromosome 3 responsible for hTERT repression does (do) not act via changes in the Myc/Mad network. Genetic or epigenetic events other than changes in c-Myc levels must be required for hTERT activation in the tumor that gave rise to 21NT cells.

Our results strongly suggest that chromosome 3 acts to repress telomerase through transcriptional silencing of the gene that encodes hTERT. In our previous study (39), we were unable to obtain definitive proof that replicative senescence induced by chromosome 3 was exclusively attributable to telomerase repression. In the present study, we used, as recipients, 21NT cells that had previously been transfected with a hTERT cDNA expression construct in an attempt to prevent senescence resulting from repression of endogenous telomerase activity. The fact that chromosome 3 transfer did not induce senescence in these hTERT cDNA-transfected recipients, whereas endogenous hTERT immature RNA was down-regulated, clearly establishes the fact that the effect of the repressor on chromosome 3 in inducing senescence is attributable entirely to a specific silencing effect on hTERT expression.

Like others, we have developed hTERT constructs in which 5′ flanking segments of the hTERT gene drive expression of a reporter gene. Our data are in agreement with previous studies (26, 28, 32, 33, 34, 35, 36, 47, 63, 64, 65, 66) in that the hTERT promoter is active in telomerase-positive immortal cell lines, but barely so in telomerase-negative primary cells. However, we also describe examples in which hTERT reporter expression does not mimic expression of the endogenous gene. Firstly, the reporters are as active in the telomerase-negative ALT cell line GM847 as in another telomerase-positive SV40-transformed fibroblast line, GM639. Secondly, in microcell hybrids in which chromosome 3 turns off expression of endogenous hTERT, the activity of the reporter constructs is not affected. In contrast, Horikawa et al. (67) found that in RCC23-chromosome 3 hybrids, luciferase expression was abrogated using a reporter containing 1.7 kb of hTERT upstream region. The discrepancy between the reporter analysis in RCC23-chromosome 3 hybrids and in 21NT-chromosome 3 hybrids remains to be addressed. Endogenous hTERT RNA levels are influenced by the proliferative state of the cells (see Fig. 1 A, EREB and HaCaT cells; Refs. 27, 34, 68, 69). Different growth rates were observed for RCC23 cells and for RCC23-chromosome 3 hybrids (40), whereas 21NT and 21NT-chromosome 3 hybrids containing an hTERT transgene proliferated at the same rate (data not shown). Analyses of GFP reporters in 21NT-chromosome 3 hybrids and GM847 cells show that the region extending 7.4 kb upstream of the hTERT promoter is not sufficient to confer proper regulation outside its endogenous context.

The hTERT gene resides very close to the telomere of the short arm of chromosome 5 (70). Telomeric chromatin in yeast is transcriptionally silent (71), and recent evidence indicates that telomeric repression exists also in human cells (72). Thus, it is tempting to speculate that the chromatin structure near the telomere may play an important role in the repression of the hTERT gene in normal human somatic cells, and that the repressor gene on chromosome-3 may in part exert its effect through chromatin remodeling.

We thank Colleen Kelleher, Kenneth Raj, and Alexandre Roulin for critically reading the manuscript; Bruno Amati for discussion and for the pBabe-cMyc plasmid and ODC primer pairs; Geron Corporation for pGRN121; Urs Ziegler for HLF cells; Georg W. Bornkamm and Axel Polack for EREB cells; Roger Reddel for GM847 cells; Stephanie Lathion for culturing HaCaT cells; Martin Jordan for help with transfection protocols; Céline Maréchal and Anne Wilson for the preparation of APC-labeled αThy-1 antibody; and Pierre Zaech for cell sorting.