Oncogene (2003) 22, 131–146 & 2003 Nature Publishing Group All rights reserved 0950-9232/03 $25.00 www.nature.com/onc hTERT associates with human telomeres and enhances genomic stability and DNA repair

Girdhar G Sharma1,2, Arun Gupta1,2, Huichen Wang3, Harry Scherthan4, Sonu Dhar1,2, Varsha Gandhi5, George Iliakis3,7, Jerry W Shay6, Charles SH Young2 and Tej K Pandita1,2,*

1Radiation and Cancer Biology Division, Washington University School of Medicine, St Louis, MO 63108, USA; 2College of Physicians and Surgeons, Columbia University, New York, NY 10032, USA; 3Thomas Jefferson University, Philadelphia, PA 19107, USA; 4Max-Plank-Institute for Molecular Genetics, Ihnestraße 73, D-14195 Berlin, Germany; 5University of Texas, MDACC, Houston, TX 70330, USA; 6University of Texas Southwestern Medical Center, Dallas, TX 75390, USA

Ectopic expression of in telomerase-silent cells a function of age in cells derived from normal human is sufficient to overcome senescence and to extend cellular blood, colonic mucosa and skin (Hastie et al., 1990; lifespan. We show here that the catalytic subunit of human Allsopp et al., 1992; Levy et al., 1992; Vaziri et al., telomerase (hTERT) crosslinks telomeres. This interac- 1994). It has been suggested as a result of this shortening tion is blocked by the telomere repeat binding factor 1, but that transcriptional profile of critical at the ends not by a dominant negative form of this protein. It is also of the may be altered (Levy et al., abolished by destruction of the RNA component of 1992; Olovnikov, 1992) or undergo a change in telomerase as well as by mutations in the hTERT protein. expression (Wright and Shay, 1992), leading to cell Ectopic expression of hTERT leads to transcriptional growth arrest. Biochemical and genetic studies have alterations of a subset of genes and changes in the established an association between telomere mainte- interaction of the telomeres with the nuclear matrix. This nance and extended lifespan mediated through the is associated with reduction of spontaneous expression of TERT. Telomerase is a reverse transcrip- damage in G1 cells, enhancement of the kinetics of DNA tase that synthesizes telomeric DNA thereby compen- repair and an increase in NTP levels. The effect on DNA sating for telomere loss that occurs with each replication repair is likely indirect as TERT does not directly affect cycle. DNA end rejoining in vitro or meiotic recombination in Ectopic expression of hTERT prevents replicative vivo. The observed effects of hTERT occurred rapidly senescence in several cell types including fibroblasts and before any significant lengthening of telomeres was epithelial cells (Bodnar et al., 1998; Vaziri and Bench- observed. Our findings establish an intimate relationship imol, 1998; Ouellette et al., 2000; Wood et al., 2001). It between hTERT–telomere interactions and alteration in may also exert an antiapoptotic action at an early stage transcription of a subset of genes that may lead to of the cell death process prior to mitochondrial increased genomic stability and enhanced repair of genetic dysfunction and caspase activation (Fu et al., 2000). It damage. These novel functions of telomerase are distinct has been proposed that telomere shortening during from its known effect on telomere length and have human replicative aging generates antiproliferative potentially important biological consequences. signals that mediate p53-dependent G1 arrest as is Oncogene (2003) 22, 131–146. doi:10.1038/sj.onc.1206063 observed in senescent cells (Vaziri et al., 1997). In support of this idea, Wong et al. (2000) reported that Keywords: hTERT; telomeres; transcription; DNA re- telomere dysfunction in mTerc null mice impairs DNA pair; ionizing radiation repair and subsequently leads to cell growth arrest; and Goytisolo et al. (2000) reported radiosensitivity of the late generation telomerase knockout mice. Choi et al. Introduction (1997) demonstrated that telomerase expression sup- presses senescence-associated genes in Werner syndrome The mammalian telomeres are composed of TTAGGG cells. However, it is not known how the TERT protein arrays bound by a complex of specialized proteins that influences and subsequently extends the protect chromosome ends from exonucleolytic attack, lifespan of a cell. Here we show for the first time that fusion and incomplete replication. Telomeres shorten as hTERT interacts with the telomeres and influences the interaction of telomeres with the nuclear matrix. It leads to transcriptional alteration along with increased *Correspondence: TK Pandita, 4511 Forest Park, Suite 411, St Louis, MO 63108, USA; E-mail: [email protected] G1 chromosome stability, enhanced DNA repair and 7Present address: University of Essen, D-45122 Essen, Germany increased NTP pools. These effects seem to be Received 23 July 2002; revised 18 September 2002; accepted 24 independent of the effect of telomerase on telomere September 2002 length. Associating telomeres GG Sharma et al 132 Results ELISA assay (Sawant et al., 1999) (Figure 1b). Im- hTERT associates with the telomeres munoprecipitates from HFF+hTERT and HeLa cells had telomerase activity. However, no activity was found To investigate whether hTERT associates with telo- in the immunoprecipitates of HFF+Ad.DhTERT or meres, human foreskin fibroblasts (HFF) with control HFF cells. These results confirmed the specifi- (HFF+hTERT) and without (HFF) ectopic hTERT city of the hTERT antibody for detection of hTERT expression were examined. Stable human fibroblasts protein and functional telomerase complex. (HFF+hTERT) were generated by infection with an To determine the association of hTERT with telo- hTERT-expressing retrovirus as previously described meres, formaldehyde-crosslinked chromatin from HFF, (Wood et al., 2001). To determine the levels of HFF+hTERT and HeLa cells was immunoprecipitated ectopically expressing hTERT in HFF+hTERT cells, using anti-hTERT and anti-p21 antibodies. By using the Western blotting was performed to detect hTERT by chromatin-immunoprecipitation (ChIP) procedure using an anti-hTERT antibody. The antibody recog- (Levy et al., 1992; Braunstein et al., 1993), we found nized hTERT in protein extracts of HFF+hTERT and that hTERT crosslinks with bulk genomic DNA as well HeLa cells while no such band was detected in HFF cells as telomeric DNA (Figure 1c). To rule out the (Figure 1a). Further, it was investigated whether possibility that hTERT interaction with the telomeres hTERT antibody recognizes the mutant hTERT is because of nonspecific binding, we investigated (DhTERT). This was achieved by infecting adenovirus- whether a negative regulator of telomerase, TRF1, expressing hTERT with deletions in the reverse tran- influences hTERT binding to telomeres. Association of scriptase (RT) motif in HFF cells (HFF+Ad.Dh- hTERT with telomeres was examined in the cells that TERT). hTERT antibody recognized the mutant overexpress intact or truncated version of TRF1. This protein in HFF+Ad.DhTERT cells (Figure 1a). The was achieved by infecting HFF+hTERT cells with hTERT antibody also immunoprecipitates (IP) func- adenovirus-expressing TRF1. The levels of TRF1 tional telomerase complex as determined by a TRAP- protein were analyzed by Western blots. TRF1 increased

Figure 1 Human telomeric DNA coimmunoprecipitated with hTERT antibody after in vivo crosslinking. (a) Western analysis of hTERT with anti-hTERT antibody. Note hTERT antibody recognized both the wild (HFF+hTERT and HeLa) as well as mutant [HFF+hTERT(m), i.e, HFF+Ad.DhTERT] hTERT protein and endogenous hTERT in HeLa cells, whereas no such band was observed in HFF cells. (b) Telomerase activity of telomerase complex immunoprecipitated by anti-hTERT antibody. Note that the telomerase activity was found in IPs with anti-hTERT antibody in HFF+hTERT and HeLa cells, whereas no such activity was observed in HFF and HFF+Ad.DhTERT cells. (c) In vivo crosslinking of HeLa, HFF+hTERT and HFF cells, followed by immunoprecipitation with anti-hTERT or anti-a-p21 antibody. The following probes were used for hybridization: telomeric DNA, total human genomic DNA and Alu DNA; no formaldehyde treatment (ÀF) or formaldehyde treatment for 1 h (+F). The same blots were used for hybridization with different probes after the stripping of the probe from the blot. Note that no telomeric signal is detected in DNA immunoprecipitated with a-p21 antibody, whereas only DNA coimmunoprecipitated by hTERT antibody after formaldehyde crosslinking showed telomeric signals. Immunoprecipitation by hTERT in HFF cells resulted in no telomeric signals

Oncogene Associating telomeres GG Sharma et al 133 at least four-fold 48 h postinfection (Figure 2a). Inter- HFF+hTERT cells were treated with an antisense estingly, cells overexpressing wild-type TRF1 showed oligonucleotide directed against hTR linked to a 2–5A reduced interaction of hTERT with telomeres (50-phosphorylated 20–50-linked oligoadenylate) mole- (Figure 2b). The degree of interaction correlated cule (2–5A-anti-hTR) as described by Kondo et al. inversely with the level of TRF1 (Figure 2a, b). (1998). A 2–5A-antisense-hTR oligonucleotide is tar- However, truncated TRF1 (lacking the Myb DNA- geted against the predicted loop of hTR ( 76– binding domain) has no effect on the interaction of 94). HFF+hTERT cells treated with 4 mm 2–5A hTERT with telomeres (data not shown). These antisense oligonucleotide for 48 h showed loss of hTR observations suggest that hTERT directly interacts with (Figure 2c). However, hTERT protein levels in 2–5A- telomeres and such interaction is blocked by TRF1 (a anti-hTR-treated cells were comparable to control (data protein that specifically interacts with TTAGGG not shown). Interestingly, 2–5A-anti-hTR-treated cells repetitive sequences). These studies are consistent with showed markedly decreased interaction of the hTERT the view that the shielding of telomeric ends and their with telomeres, but hTERT binding to bulkgenomic elongation by telomerase are dependent on telomere- DNA was unaffected (Figure 2d). These results suggest binding proteins especially TRF1 (Smith and de Lange, that association of hTERT with telomeres is dependent 1997; Pandita, 2002). upon hTR. In the second approach, primary HFF cells We next investigated if the association of telomerase with telomeres is dependent on the RNA component of telomerase (hTR). To determine whether hTR is required for the association of hTERT with telomeres, two different approaches were applied. First,

Figure 2 Factors that influence hTERT interactions with telo- meres. (a,b) Interaction of hTERT in cells overexpressing TRF1 (Ad.TRF1). (a) Western analysis of TRF1 from cells infected with adenovirus expressing TRF1 at different time points postinfection. Note that there is a substantial increase in the TRF1 levels 48 h postinfection. (b) In vivo cross-linking of HFF+hTERT cells with ectopic expression of TRF1. In vivo crosslinking of cells was performed at different time points after infection. Note the decrease in signal of telomeric DNA but not in bulkgenomic bulk DNA. (c–f) Interactions of hTERT protein with telomeres is influenced by abolishing hTR. Two different approaches were used to determine the influence of hTR for hTERT interactions with telomeres. First approach: HFF+hTERT cells were treated with antisense oligonucleotides for hTR. (c) Northern analysis for hTR. Lanes labeled with ‘C’ represent hTR from control cells and lanes labeled with ‘T’ represents hTR from the treated cells. Note that the hTR was abolished in cells after 48 h treatment with the antisense hTR oligonucleotides. (d) HFF+hTERT cells were treated with antisense oligonucleotides for hTR and in vivo cross- linking was performed at different time points after treatment. Note a decrease in the amount of telomeric DNA as compared to the total genomic DNA. Second approach: HFF cells were first treated with anti-sense oligonucleotides for hTR and then subsequently infected with adenovirus expressing hTERT. (e) Western analysis for hTERT protein determined after 72 h of infection with adenovirus expressing hTERT. Lane 1 represents HFF cells treated with 2–5-anti-hTR but without ectopic expres- sion of hTERT; lane 2 represents HFF cells treated first with 2–5- anti-hTR and then infected with adenovirus expressing hTERT; lane 3 represents HFF cells infected with adenovirus expressing hTERT and; lane 4 represents HeLa cells. Note that the hTERT protein is expressed in cells with and without treatment with 2–5- anti-hTR. (f) In vivo crosslinking of hTERT was performed 72 h postinfection with adenovirus expressing hTERT. Control repre- sents HFF cells expressing hTERT and treated represents HFF cells were first treated with 2–5-anti-hTR and then infected with adenovirus expressing hTERT. Immunoprecipitation by hTERT in control (untreated) resulted in both total as well as telomeric DNA signals, whereas 2–5-anti-hTR treated resulted in only total DNA signals and no signal was observed for telomeric DNA. (g) Influence of mutations in hTERT for its interaction with telomeres. HFF+hTERT cells were infected with DhTERT in adenoviral construct and in vivo crosslinking was performed at different time points after infection. Note a decrease in the amount of telomeric DNA as compared to the total genomic DNA. The results are representative of four experiments done separately

Oncogene Associating telomeres GG Sharma et al 134 (population doubling, PD ¼ 40) were first treated with Table 1 The list of known genes that showed altered expression by 2–5A-anti-hTR and then followed by infection with a more than three fold in cells with ectopic hTERT expression recombinant adenovirus vector expressing hTERT Gene/s Fold change (HFF+Ad.hTERT). These cells were examined 72 h in expression after infection for hTERT protein and its interactions Superoxide dismutase 2a +7.3 with telomeres. Treatment with 2–5A-anti-hTR did not Guanine -binding protein HMB89 +6.8 affect the ectopic expression of hTERT (Figure 2e); NF-kappa-Bb +6.5 however, it did eliminate detectable telomerase activity Inositol hexabisphosphate (IP6) +6.3 as determined by TRAP-ELISA (data not shown). In Ku80 +5.3 Peroxisome proliferator-activated receptor gamma; +5.2 cells treated with 2–5A-anti-hTR, hTERT protein PCNA crosslinked with total genomic DNA but did not bind MLH1; MSH6 +4.3 to telomeres (Figure 2f). In contrast to cells that were PMS2L4; XRCC9 +4.1 not treated with 2–5A-anti-hTR, hTERT crosslinked to HSP-70b +4.0 Translational elongation factora +3.9 total genomic DNA as well as telomeres (Figure 2f). GTF2H4; RPA1 +3.8 Data obtained from two different approaches suggest DK7; MMS19 +3.7 that hTR is essential for hTERT interaction with Inositol-1,4,5-triphosphate receptor (type 3) +3.6 telomeres. RAD51L3; XRCC3a +3.6 We next asked the question as to whether a MOF; POLQ +3.5 PgR; POLL; TRF4-2a +3.4 catalytically active telomerase complex at telomeres is Erb-B3; TGF b receptor type 3 +3.3 required for the interaction of hTERT with telomeres. DNA II +3.2 To do so, adenovirus expressing mutant hTERT (Ad. FGF receptor 4; Jak1 +3.1 DhTERT), with deletion in the Activin B-c chain; endoglin +3.0 Stromelysin-1a À6.7 (RT) motif (from nucleotides 2250 to 2270), were GROa; Fibronectinb À6.5 infected in HFF+hTERT (retroviral expressing hTERT IL-6 À6.3 stable cell line) cells. Ectopic expression of these mutants RAB-3A À5.8 abolished the interaction of hTERT with the telomeres. Tenascin À5.4 Telomerase activity in such cells after 96 h of infection junB; prostacyclin-stimulating factor À5.1 Keratin II 7; transglutaminase 1 À4.6 with adenovirus-expressing hTERT mutant was not Extracellular SOD À3.8 detected by TRAP-ELISA assay (data not shown). CYP1B1a À3.2 However, as was the case with the hTR-deficient cells described above, mutant hTERT protein still associated +, elevated expression; À, is decreased expression. Expression in some with the bulkgenomic DNA (Figure 2g). Thus, the genes was confirmed by RT-PCR or Northern analysis. aGenes whose expression was confirmed by RT-PCR. interaction of hTERT with telomeres requires a bGenes whose expression was further confirmed by Northern analysis. functionally active hTERT protein.

Influence of hTERT on gene expression confirmed by Northern blot analysis. mRNA from cells with and without hTERT expression was used for Besides the obvious function of telomerase in telomere confirmation of microarray results by RT-PCR, for synthesis, no other functions for telomerase at the 30% of genes as described in Table 1. When the telomeres are known. One possibility is that hTERT interaction of telomerase with telomeres in may modify telomere interactions with the nuclear HFF+hTERT cells was blocked, either by overexpres- matrix and subsequently induce alterations in transcrip- sion of TRF1 or expression of mutant hTERT,as tion of certain genes. To investigate the influence of described earlier, the transcription profile of these cells hTERT on global transcription, gene expression profiles became similar to that of HFF cells. of proliferating HFF and HFF+hTERT cells were compared according to the procedure described recently hTERT interaction with telomeres precedes alteration in (Vafa et al., 2002). The transcription profiles were gene expression analyzed by monitoring 5776 expressed sequence tags (ESTs) on oligonucleotide arrays. Each experiment was Next, we wanted to determine whether the association repeated four times. The first series of experiments were of hTERT with telomeres precedes alteration in gene conducted on cells infected with a retroviral hTERT and expression. Thus, we devised a model to examine the then after selection for drug resistance, control HFF and hTERT–telomere interaction and gene expression pro- HFF+hTERT cells were analyzed at various popula- files at different time points after infection. The retro- tion doublings. Several possible pairings for hybridiza- virus-infected cells (HFF+hTERT) described above are tion for each of the cultures (HFF, HFF+hTERT) at not suitable for such experiments because of the lower PDs of 40, 50 and 60 were performed. Only known genes infection efficiency, and thus the time that would be with greater than three-fold difference in expression required to select cells resistant to drug markers. levels are listed in Table 1. In all, 67 (about 1.1%) ESTs Therefore, we used a recombinant adenovirus vector reproducibly exhibited significant changes in expression. expressing hTERT, infecting primary human fibroblasts These results were confirmed by RT-polymerase chain at PD 50. The multiplicity of infection for adenovirus reaction (RT-PCR) (Figure 3) and some of them were was such that >99% of cells were infected. In vitro

Oncogene Associating telomeres GG Sharma et al 135 but did not bind to telomeres (Figure 2f) and also did not show any change in transcription (data not shown), suggesting that functional telomerase is required for hTERT interaction with telomeres as well as transcrip- tional alteration. We next addressed the question whether the enhance- ment of transcription of some genes in hTERT- expressing cells is caused by due to the binding of hTERT with the promoters of the genes with higher levels of expression. The interaction of hTERT with the promoters of superoxide dismutase 2 (SOD-2) and NF- kappa-B was examined. Chromatin immunoprecipita- tion with hTERT, followed by PCR of the promoter regions of SOD-2 and NF-kappa-B, did not reveal hTERT binding with these promoters (data not shown). It is thus more likely that hTERT protein, either through its interaction with telomeres or with unknown target/s, generates a series of signals to regulate the expression of a variety of genes required for extended lifespan through indirect mechanisms. Figure 3 Representative picture showing comparison of gene expression between HFF and HFF+hTERT cells for four different genes analyzed by RT-PCR. Expression of NF-kappa-B and hTERT expression influences telomere–nuclear matrix XRCC3 is higher in HFF+hTERT cells than HFF cells, and the association expression of fibronectin and stromelysin-1 is lower in HFF+hTERT than HFF cells. Note the results obtained from Although hTERT interactions with telomeres precede RT-PCR for gene expression comparison between HFF and transcriptional alterations, the detailed mechanisms of HFF+hTERT cells are similar to that obtained by microarray how hTERT interacts with telomeres or chromatin and analysis influences gene expression is not clear at present. One possibility is that hTERT interactions with telomeres telomerase activity in extracts of such cells was may influence the association between telomeres and measured by a TRAP-ELISA procedure (Sawant et al., nuclear matrix. Recently, we and others have shown 1999). Telomerase activity appeared 48 h after infection that hTERT is present throughout the nucleus but in and peaked by 96 h (data not shown). The association of certain instances hTERT is concentrated in the nucleo- hTERT protein with the telomeric DNA sequences lar organizing region (Tahara et al., 1999; Vaziri et al., started to appear 48 h after infection (Figure 4a). 1999; Kawakaim et al., 2000; Wood et al., 2001; However, alterations in gene expression appeared 60 h Etheridge et al., 2002). More directly, telomeres have after infection (Figure 4b, c). The gene expression been shown cytologically as well as biochemically to be profiles of hTERT-expressing cells with retroviral tethered to the nuclear matrix (Smith and de Lange, (HFF+hTERT) infection and passage for 40 PD or 1997; Pandita, 2002). The nuclear matrix is a proteinac- adenoviral (HFF+Ad.hTERT) infection and passage eous scaffold in the interphase nucleus isolated by for 72 h were almost identical (data not shown). When removing most of the nuclear DNA and RNA, along HFF cells were infected with empty vectors or mutants with histones and loosely bound proteins. Since of hTERT gene, their expression profile did not change, telomeres are associated with nuclear matrix (Smith suggesting that the changes are not because of non- and de Lange, 1997; Pandita, 2002) and such associa- specific effects of infection stresses to cells. Many of the tions could influence DNA metabolism (Struhl, 1998), genes, whose expression in hTERT-expressing cells is we next examined whether hTERT expression influences decreased, are linked with senescence (Table 1). Some of telomere–nuclear matrix associations. the genes that showed elevated expression in hTERT- Logarithmically growing HFF (control) and expressing cells are involved in chromatin modification HFF+hTERT cells were processed using a lithium and DNA damage repair processes (Table 1). diiodosalicylate (LIS) procedure, and the resulting To determine whether transcriptional changes ob- nuclear matrix halos were cleaved with Sty1 (Smilenov served are because of hTERT alone or require an active et al., 1999). The nuclear matrix halos are the insoluble telomerase complex, hTR was abolished, which is nonchromatin scaffolding of the interphase nuclei. The required for the binding of hTERT with telomeres and nuclear remnant and associated DNA were separated by telomerase activity. This was achieved by treating centrifugation (Smilenov et al., 1999). For genomic primary HFF cells (PD ¼ 45) with 2–5A-anti-hTR and blotting analysis, equal volumes representing DNA then followed by infection with a recombinant adeno- from an identical number of halos were fractionated virus vector expressing hTERT. These cells were on 1.5% agarose gels. HFF cells have 6774% of the examined 72 h after infection for hTERT interactions telomeric DNA associated with the nuclear matrix with telomeres and change in gene expression profile. In (attached) (P) fraction and 3372% in the soluble (free) these cells, hTERT crosslinked with total genomic DNA (S) fraction (Figure 5a). In contrast, HFF+hTERT cells

Oncogene Associating telomeres GG Sharma et al 136

Figure 4 Kinetics of hTERT interaction with telomeres and alteration of gene expression. (a) Time course occurrence of hTERT at the telomeres and p21 in total genomic DNA in HFF cells infected with adenovirus containing hTERT. Note the levels of hTERT bound to total or telomeric DNA increase after infection with adenovirus containing hTERT, while there is no change in p21 binding to total DNA, suggesting p21 binding is not affected by hTERT expression. (b) Quantitative RT-PCR after 35 cycles to determine the levels of SOD-2 at different time points postinfection of adenoviral construct containing hTERT in HFF cells. Note the appearance and subsequent increase of SOD-2 message after 60 h of infection, whereas there is no significant change in expression of alpha-actin. (c) The relative increase of hTERT binding to telomeres and SOD-2 expression. The estimation of hTERT binding to telomeres is relative to the p21 binding to genomic DNA and the estimation of SOD-2 gene expression is relative to alpha-actin expression. Note an increase in telomeric DNA levels coimmunoprecipitated with the increase in postinfection time, suggesting that TERT protein binds to telomeres. The intensities were quantitated by the procedure described in the Methods section. Note that the increase in the interaction of hTERT with telomeres precedes the increase in SOD-2 gene expression

have 5472% of the telomeric DNA associated with the associations concomitant with the expression of hTERT nuclear matrix and 4673% in the soluble fraction may contribute towards the alteration of gene expres- (Figure 5a). In both instances, summation of the P and S sion required for extending cellular lifespan. values is equal to the total telomeric DNA (T), To determine whether alterations in telomere–nuclear suggesting that essentially no telomeric DNA was lost matrix association with ectopic hTERT expression are during the extraction. The ratio of P/S fractions in because of the change in the mean telomere length, we HFF+hTERT cells is 1.17, which is 73% less than the compared the mean telomere length by Southern blotting ratio of P/S (2.03) found in HFF cells. Further, it was at 0, 48, 72 and 96 h postinfection of cells with the determined whether such changes in the interaction of adenoviral vector expressing hTERT. No dramatic telomeres with the nuclear matrix occurred immediately change in the mean telomere length was found in such after ectopic expression of hTERT. This was achieved cells (Figure 5c). These observations suggest that altera- by examining telomere-nuclear matrix interactions in tions in the telomere–nuclear matrix association because HFF cells infected with adenovirus expressing wild-type of functional hTERT protein are not dependent upon hTERT, mutant hTERT and empty vector controls. The telomere length. These results also suggest that change in change in telomere–nuclear matrix interactions was seen telomere length is not a prerequisite for the alteration in only in cells infected with wild-type hTERT after 48 h of telomere–nuclear matrix interaction or transcription. infection (Figure 5b). We found no change in telomere– Hemann et al. (2001) and Steinert et al. (2000) have nuclear matrix interactions in cells infected with mutant demonstrated in mouse and human cells that the hTERT (data not shown). These results suggest that shortest telomere, not average telomere length, is critical ectopic hTERT expression influences the association of for cell viability and chromosome stability. Therefore, telomeres with the nuclear matrix and this occurs very the results showing no change in the average telomere rapidly. This reduction in telomere–nuclear matrix length are in agreement with the maintenance of the

Oncogene Associating telomeres GG Sharma et al 137 cells infected with the adenoviral vector expressing hTERT. However, this analysis only yields an appraisal of the population of TRFs generated, and does not monitor ends of individual chromosomes. To determine the presence of telomeres of each chromosome, fluores- cence in situ hybridization (FISH) for telomeric repeats in metaphase cells was performed by using a telomere- specific Cy3 labeled (CCCTAA)3 peptide nucleic acid probe. In total, 50 metaphase chromosome spreads from each cell type were included and analyzed. No sig- nificant change in telomere-specific fluorescent signals was observed in cells examined 72 h postinfection with adenoviral vector expressing hTERT (Figure 6). Based on the function of telomerase on telomeres (Blackburn, 2001; Chan and Blackburn, 2002), we believe that there may be a gain in the size of telomeres by about 100–300 nucleotides in a period of 72 h after infection with adenoviral vector expressing hTERT. However, this increase in size cannot be determined by the currently available techniques. Therefore, these results suggest that a drastic change in telomere length is unlikely to be a prerequisite for the alteration in gene expression, at least in HFF cells. Although ectopic expression of hTERT did not result in any drastic changes in telomere length, however, such expression did influence telomere–nuclear matrix asso- Figure 5 Autoradiograph of telomeric DNA in HFF cells and HFF+hTERT cells. (a, b) Determination of associations of ciations. Whether alteration in telomere nuclear–matrix telomeric DNA with the nuclear matrix in HFF cells with and associations correlates with the telomere stability, without ectopic expression of hTERT. (a) Telomeric DNA from chromosome end associations in G1 phase as well as HFF and HFF+hTERT (retroviral infected) cells and (b) metaphase cells were examined. To examine chromo- Telomeric DNA from HFF and HFF+Ad.hTERT (adenovirus infected) cells 48 h after infection. Halo preparation was digested some end associations in G1 phase cells, we used the with StyI and centrifuged to separate free supernatant (S) and combination of the premature chromosome condensa- pelleted (P) DNA. Lanes T represent total telomeric DNA; lanes S, tion (PCC) technique with in situ hybridization (Pandita soluble fraction; lanes P, nuclear matrix fraction. Note that the et al., 1995), which allows direct examination of summation of P and S values is equal to total (T). P and S lanes of chromosomes in the interphase cells, whether hTERT HFF and P and S lanes of HFF+hTERT represent telomeric DNA from a similar number of halos. Note the difference in the expression stabilizes telomeres and reduces sponta- ratio of P versus S fractions of telomeric DNA between HFF and neously induced chromosome damage (Figure 7). Loga- HFF+hTERT cells. The results are representative of three rithmically or exponentially growing human cells were separate experiments. (b) Determination of associations of fused with Chinese hamster ovary (CHO) mitotic cells telomeric DNA with the nuclear matrix in HFF and HFF+hTERT (adenovirus infected) cells 48 h postinfection. The by the procedure described in the Methods section. results are representative of three separate experiments. Lanes T Human chromosomes were detected by using either all represent total telomeric DNA; lanes S, soluble fraction; lanes P, human genomic probe or human centromeric probe. A nuclear matrix fraction. (c) Determination of mean telomere total of 200 PCC spreads were microscopically examined length. For genomic blotting analysis of telomere length, DNA for frequencies of chromosome end associations and was digested with RsaIorHinfI and analyzed by Southern hybridization using a TTAGGG repeat probe. Molecular sizes chromosome damage. Cells with ectopic hTERT ex- are indicated on the left. Lane 1, DNA from HFF with retroviral pression had only 0.04 telomeric associations per G1 hTERT (HFF+hTERT cells at PD 96); lane 2, DNA from cell, which is 12-fold less than the cells without ectopic HFF+hTERT cells after 0 h of infection with adenoviral vector hTERT expression (0.5 telomeric associations per G expressing hTERT; lane 3, DNA from HFF+hTERT cells after 1 48 h of infection with adenoviral vector expressing hTERT; lane 4, cell). The differences in telomeric associations in DNA from HFF+hTERT cells after 72 h of infection with interphase cells with and without hTERT were statisti- adenoviral vector expressing hTERT; lane 5, DNA from cally significant as demonstrated by the Student’s t-test HFF+hTERT cells after 96 h of infection with adenoviral vector (Po0.001). Spontaneous chromosome breaks in G1- expressing hTERT; lane 6, DNA from HFF cells without ectopic phase cells (HFF) without ectopic hTERT expression expression of hTERT. Note that no dramatic difference in the mean telomere length is found in the DNA of lanes 2–6; however, a occur at 0.2 breaks per cell, which is 20-fold higher than significant increase in the mean length is seen in the DNA of lane 1. the cells with ectopic hTERT expression The results are representative of three experiments done separately (HFF+hTERT) with 0.04 breaks per cell (Po0.001). shorter telomeres preventing chromosome instability hTERT expression improves DNA damage repair and cell senescence. By Southern blotting, we found no significant differences in telomere repeat fragment The higher mRNA levels of some of the DNA repair (TRF) sizes, when comparing DNA derived from HFF genes (Table 1) suggest that genome stabilization in cells

Oncogene Associating telomeres GG Sharma et al 138

Figure 6 Telomere FISH analysis of metaphase chromosomal spreads. Detection of telomeres on metaphase from (a) HFF cells without ectopic expression of hTERT (HFF) and (b) HFF cells infected with adenoviral construct expressing hTERT (HFF+hTERT) after 72 h of infection was done by FISH using a telomere-specific probe. Note the presence of telomere signals at the chromosome end- to-end associations in HFF cells as indicated by arrows and the lackof differences in the amount of telomere FISH signals between HFF and HFF+hTERT cells

in cells with and without ectopic hTERT expression, DNA repair was examined after treatment with ionizing radiation and cisplatin. Ionizing radiation produces DNA strand breaks, whereas cisplatin produces DNA adducts. Exponentially growing isogenic cells with and without hTERT expression were treated with 15 Gy of gamma rays and were allowed to repair at 371C for various time periods prior to the analysis by neutral filter elution for the residual levels of DNA strand breaks (Pandita and Hittelman, 1992). Cells in logarith- mic growth without ectopic hTERT expression (HFF) exhibited a slow rate of DNA repair along with higher residual DNA damage like the ataxia telangiectasia cell line GM5823 (A-T) cells, whereas cells with ectopic hTERT expression (HFF+hTERT) had a fast rate of DNA repair and less residual DNA damage (Figure 8a). The most frequent damage that occurs in vivo is DNA adduct formation and deficiency in its repair is linked with the age of an individual. DNA interstrand cross- links produced by the incubation of human peripheral blood mononuclear cells with cisplatin (CDDP) are repaired faster in young donors (age 25 years) as compared to individuals of an elderly group (age above 75 years) (Rudd et al., 1995). Those from the elderly group have significantly higher mean levels of cross- linking after CDDP treatment and this could be indicative of impaired DNA repair capacity in the cells Figure 7 Visualization of chromosome end associations and from the elderly group. Since telomere shortening also spontaneous chromosome damage in G1 cells using PCC technol- correlates with advanced age and hTERT expression ogy in combination with FISH. In (a) and (b), human chromo- stabilizes telomeres, we investigated whether hTERT somes were detected by using whole genomic painting probe to expression has an influence on DNA adduct repair. analyze chromosome end associations in G1-phase cells. Human Thus, DNA adduct repair was examined by alkaline chromosomes fluoresce yellow in color due to hybridization with FITC-labeled human genomic probe and the unlabeled CHO elution in cells after treatment with CDDP (Nishikawa mitotic chromosomes are red in color. (a)G1 PCC of HFF cell at et al., 1993). The extent of DNA interstrand crosslinking 50 PD having frequent chromosome end associations as shown by was measured up to 24 h after treatment with 5mg/ml arrow. (b)G1 PCC of HFF+hTERT cell having telomere defect CDDP (1 h treatment with drug) (Figure 8b). The rate of corrected as there are no chromosome end-to-end associations. In (c) and (d), human centromeric probe was used, to determine the formation of DNA crosslinks reached a peak level 11 h dicentrics in G1 chromosome. (c)G1 PCC of HFF cell having after treatment in cells with and without ectopic hTERT dicentrics as indicated by arrow. (d)G1 PCC of HFF+hTERT expression. However, a sharp contrast in the repair cells having reduced spontaneous chromosome damage process was observed (Figure 8b). DNA damage was rapidly repaired in cells with ectopic hTERT expression with ectopic hTERT expression could be because of after 15 h following exposure to CDDP, while less enhanced repair capacity of the cells. To determine change in interstrand crosslinking was observed in cells whether there is any difference in DNA repair efficiency without hTERT expression. These results suggest that

Oncogene Associating telomeres GG Sharma et al 139 (Plunkett et al., 1980; Sherman and Fyfe, 1989; Rodriguez et al., 2000). Data from three separate experiments are summarized in Table 2. When the dNTP pools were compared, the values were statistically similar. As expected, the intracellular levels of NTP pools were much higher: ATP being the highest of all four NTPs. Interestingly, cells with ectopic hTERT expression (HFF+hTERT) had relatively higher ribo- nucleotide pools as compared to the cells without ectopic hTERT expression (HFF) (Table 2) or those expressing a mutant hTERT (data not shown). When comparing the ATP levels vs dATP levels, cells with hTERT expression (HFF+hTERT) had relatively low- er levels of dATP but higher levels of ATP, suggesting that ectopic hTERT expression does influence ATP metabolism. Thus, an increased level of ATP may be an important factor responsible for the fast kinetics of the DNA repair. To test whether differences in ATP between HFF and HFF+hTERT cells were because of the differences in frequency of different phases of the cell cycle, we determined the frequency of G1,SandG2/ M by procedures described previously (Pandita and Hittelman, 1992). HFF+hTERT cells have similar cell cycle phase distribution to HFF cells, suggesting that the differences in the ATP levels are not due to differences in frequency of different phases of the cell cycle. The above results established a possible role of hTERT in DNA repair. To determine whether hTERT protein is directly involved in the process of DNA end- Figure 8 Determination of DNA damage repair. (a) DNA joining, we used a well-characterized in vitro assay that double-strand breaks (DSB) repair: The level of DNA DSB in involves the rejoining of DNA ends of pSP65 plasmid in cells was estimated as described in the material and method section. HeLa cell extracts (Iliakis et al., 1999). The pSP65 (&) represents HFF cells with hTERT in retroviral construct plasmid was digested with SalI and plasmid end-joining [HFF+hTERT] (PD ¼ 55 postinfection); (’) represents HFF cells with hTERT in adenoviral construct (72 h postinfection) was performed as described previously (Iliakis et al., [HFF+Ad.hTER] and; (m) represents primary HFF cells without 1999). hTERT protein in HeLa cell extracts was ectopic expression of hTERT. A-T (n) was used as a control to depleted by immunoprecipitation by using different indicate the kinetics of DNA DSB repair and residual DNA damage. Note that the rate of DNA repair is slow in primary cells (HFF) and such cells have higher residual DNA damage. The mean and standard deviation of triplicate experimental points is shown. The differences in residual damage in HFF cells with and without Table 2 Comparison of dNTP and NTP concentrations in HFF- ectopically expressing hTERT are statistically significant as hTERT and HFF+hTERT cells demonstrated by Student’s t-test (Po0.05). (b) DNA interstrand crosslinking repair: DNA interstrand crosslinking was estimated by Mean concentration7s.d. the alkaline elution as described in the Materials and methods HFF hTERT HFF+hTERT section. The HeLa (.) cell line was used as a control to indicate the À kinetics of DNA DSB repair and residual DNA damage. The mean dNTPs and standard deviation of four experimental points is shown. Note dATP 6.673.5 5.270.7 that the primary cells (HFF) (m) have high DNA crosslinking dCTP 3.871.9 3.871.5 as compared to cells (HFF+hTERT) with ectopic hTERT dGTP 1.470.6 1.670.8 expression dTTP 18.4710.4 20.275.8

NTPs* hTERT expression has a pronounced effect on the ATP 25337115 37317170 CTP 1917007 2807078 repair of DNA adducts. GTP 4557030 6157028 One possible reason for faster DNA repair kinetics in UTP 6277046 8527134 hTERT-expressing cells could be because of the increased NTP pool levels. To determine whether there Note, no significant difference in dNTPs was found in HFF+hTERT is a correlation between the levels of NTP and the and HFF cells. However, a significant difference in NTPs was found in kinetics of DNA repair, we measured the levels of HFF+hTERT and HFF-hTERT cells. HFF-hTERT cells were analyzed at PD of 48, 50 and 55. HFF cells with empty vector did intracellular deoxynucleotides (dNTP) and ribonucleo- not show any increase in the NTP pool levels. tides (NTPs) in cells with and without ectopic hTERT *These differences are statistically significant as determined by expression utilizing previously described procedures Student’s t-test (Po0.05).

Oncogene Associating telomeres GG Sharma et al 140

Figure 9 Influence of TERT protein on DNA end joining in vitro and in vivo.(A) In vitro plasmid DNA end joining: (a) Immunodepleted HeLa cell extract (50 mg) was separated on 10% SDS-PAGE and probed by anti-hTERT antibody (500-fold dilution). Lane 1: Untreated control sample; lane 2: sample treated with normal rabbit serum; lane 3: sample immunodepleted of hTERT. Note that the depletion of hTERT is complete. (b) Cell extract (20 mg) was used for plasmid end joining. Lane 1: no extract; lane 2: untreated control extract; lane 3: extract treated with normal rabbit serum. Lane 4: extract immunodepleted for hTERT. Note that immunodepletion of hTERT does not affect plasmid end joining in this assay. (c) Plasmid end joining in HeLa whole-cell extract supplemented with 1 ml of the indicated dilution of the anti-hTERT antibody (lower set), or of a normal rabbit serum (upper set). Note that no significant reduction in the plasmid end joining was observed in reactions incubated with anti-hTERT antibody. (B) Spreads of spermatocytes I immunostained for SCP3 (green) and TRF1 (red): (a) Control mice and (b) Tert null mice. The ends of the SCs (green) are in both genotypes capped by telomeric TRF1 signals (red). The spread sex body is fluorescing DAPI-bright (blue) and contains the unpaired XY axes. Note that there was no difference in SC formation as well as for TRF1 staining, suggesting that TERT does not influence meiotic telomeres and chromosome pairing

concentrations of hTERT antibody. Complete immu- detected in the depleted extracts (data not shown). The nodepletion of hTERT was confirmed by Western hTERT immunodepleted extracts were used to assay for analysis and by conducting a functional telomerase the DNA end-joining activity. No difference in DNA activity assay. Figure 9A(a) shows the hTERT protein end joining was found in extracts with and without band in control lanes 1 and 2, while the hTERT band in hTERT protein (Figure 9A(b)). We further analyzed the immunodepleted sample was undetected (lane 3). whether the addition of hTERT antibody could Telomerase activity was measured by TRAP-ELISA influence the DNA end joining kinetics. An insignificant procedure (Sawant et al., 1999) and no activity was reduction in the plasmid end-joining was seen in the

Oncogene Associating telomeres GG Sharma et al 141 presence of anti-hTERT antibody (Figure 9A(c)). Simi- immunohistochemistry results described earlier (Martin- lar results were observed in HFF+hTERT cells (data Rivera et al., 1998; Tahara et al., 1999; Vaziri et al., not shown). These results suggest that hTERT protein 1999; Wood et al., 2001) suggest that hTERT does bind itself does not directly interact with the DNA repair to total genomic DNA and the present data suggest that machinery. However, these observations do not rule out this binding is independent of hTR or the RT motif of the possible role of TERT in chromosomal recombina- hTERT. It is yet to be determined whether hTERT tion. One way to determine the influence of TERT on binds to any specific sequence of total genomic DNA. recombination in vivo is to compare the formation of However, Etheridge et al. (2002) have demonstrated that bouquets (also known as telomere clustering), synapto- hTERT localizes to the nucleolus. In contrast to this, nemal complexes, sex bodies and chiasmata frequency in hTERT binding to telomeres is dependent on hTR as meiocytes of mTert null and control mice. well as the RT motif, which suggests that hTERT must Inactivation of mTert leads to telomere instability in be functionally reconstituted in order to interact with mice but only after several generations (Liu et al., 2000). telomeres so that it could provide the function of To determine if such inactivation influences chromoso- replenishing the telomeric DNA. mal recombination, we examined telomere clustering, It is thought that telomerase exerts its function on synaptonemal complexes, sex bodies and chiasmata telomeres by extending telomere length. In yeast, as well frequency in meiocytes of mTert null and control mice as in human cells, it has been shown that telomere length according to the standard procedure (Pandita et al., influences transcription (Gottschling et al., 1990; Kyrion 1999). It has been suggested that meiotic telomere et al., 1993; Renauld et al., 1993; Baur et al., 2001). clustering might support the alignment of homologs However, it is yet to be determined as to what prior to their synaptic pairing (Scherthan 2001). We constitutes the optimum size for a telomere to be found no difference in the bouquet formation between functional in mammalian systems. Here, we show that the spermatocytes of mTert null and control mice (data the expression of hTERT influences transcription 60 h not shown). The synaptonemal complex formation as after infection with an adenovirus vector expressing well as sex body formation (Figure 9b) was nearly hTERT. This change in transcription occurs immedi- identical in mTert null and control mice. We anticipated ately after hTERT is found associated with telomeres. that the absence of mTert might indirectly influence the Whether such alteration in transcription could be linked binding of telomere repeat-binding factor (TRF1) to to any change in telomere length is not clear, because of telomeres, but immunostaining with the antibody the fact that within 60 h of infection with adenovirus against TRF1 was identical between the spermatocytes expressing hTERT, the cells may undergo two or three from mTert null and control mice (Figure 9b). Further, replications, which could add about 200 nucleotides at chiasmata frequency determined in mTert null and the end of telomeres. Such small changes in telomere control mice was identical. These results suggest that length in mammalian cells could not be determined mTERT does not have a direct influence on the in vivo either by Southern or FISH analysis using telomere meiotic pairing or recombination process. specific probes. It is yet to be determined whether such small changes in the telomere length would have any influence on the gene expression in mammalian cells. Recently, Baur et al. (2001) have shown that telomere Discussion position effect is operative in the human system. The fact that no major changes in telomere size are detected The human telomerase holoenzyme contains a catalytic in such cells, after 60 h of infection with adenovirus protein, hTERT and a 451-base integral RNA, hTR expressing hTERT, suggests that change in transcription (Feng et al., 1995), that are essential for assembling may be because of alteration of telomere–nuclear matrix telomerase activity in vitro and in vivo (Weinrich et al., associations. 1997). In this report, we demonstrated for the first time Mammalian telomeres are packaged in telomere- that hTERT associates with telomeres and that this specific chromatin (Smith and de Lange, 1997). Human association is dependent upon both hTR as well as the and mouse cells have their telomeric tracts attached to RT motif of hTERT protein. The association of hTERT the nuclear matrix, which is a proteinaceous subnuclear with telomeres was found in ectopically expressing fraction (Luderus et al., 1996; Smilenov et al., 1999). In hTERT cells (HFF+hTERT, HFF+Ad.hTERT) as mammals, a nuclear matrix-binding site occurs at least well as in HeLa cells with endogenously expressed once in every kilobase of the telomere tract (Luderus hTERT. Consistent with these results is the inhibition of et al., 1996). This indicates that mammalian telomeres hTERT association with the telomeric DNA by TRF1. have frequent multiple interactions with the nuclear However, overexpression of TRF1 did not have any matrix. Such interactions are influenced by hTERT influence on the association of hTERT protein with bulk protein as the cells with ectopic expression of hTERT genomic DNA, suggesting the specificity of hTERT were found to have less telomeric DNA associated with binding to telomeres. Further, abolishing the expression the nuclear matrix. This is based on the fact that of hTR or deletion in the RT motif of hTERT protein alteration of telomere–nuclear matrix association was also did not influence the hTERT association with found after 48 h of infection and alteration in transcrip- genomic DNA but did affect the binding to telomeres. tion was found after 60 h of infection with adenovirus The ChIP results described in this manuscript and expressing hTERT in HFF cells. These observations

Oncogene Associating telomeres GG Sharma et al 142 suggest that a change in the interaction of telomeres higher in the G1 phase than in the G2 phase, with the with the nuclear matrix may influence the pattern of fewest associations observed at metaphase (Pandita transcription. The cell cycle profile of control HFF and et al., 1995). It is probable that the end associations adenoviral vector hTERT expressing HFF cells after seen at mitosis reflect a continuation of interphase 48 h of infection were similar (data not shown) and thus chromosome behavior, indicating interactions or lin- changes in proliferation are unlikely to account for the kages between chromosome ends and the nuclear differences described above. Thus, it is possible that a matrix. Thus, hTERT modifying the interactions of novel function of hTERT may be to set the proper telomeres with the nuclear matrix may help to resolve conditions for the change in interactions of telomeres the chromosome end associations seen at metaphase. with nuclear matrix, as a signal for transcriptional Since we observe changes in the interactions of alterations in gene regulation. Although there is a link telomeres with nuclear matrix occur within 48 h after between the alteration of the interaction of telomeres infection with adenovirus expressing hTERT, one of the with the nuclear matrix and change in transcription functions of hTERT might be to influence associations induced by ectopic expression of hTERT, it is yet to be of the telomeres with nuclear matrix and thus to determined how a change in telomere–nuclear matrix stabilize the telomeres. could influence transcription. The changes in the Besides the stabilization of telomeres, the frequency of interaction of telomeres with nuclear matrix as well as spontaneous chromosome breaks was also influenced by change in the transcription induced by ectopic hTERT the ectopic expression of hTERT. Cells with ectopic expression do not appear to be dependent upon major expression of hTERT (HFF+hTERT) have a signifi- changes in the telomere length in HFF cells. cantly lower number of chromosome breaks as com- There is mounting evidence for the fact that the pared to the cells without ectopic expression of hTERT telomere participates in processes of chromosomal (HFF). The reduction in the spontaneous chromosome repair, as evidenced by the ‘capture’ or de novo synthesis breaks could possibly be the result of better repair. This of telomeric repeats at double-stranded breaks and by is consistent with the findings that cells expressing the capacity of yeast telomeres to serve as repositories of hTERT (HFF+hTERT) were found to have higher essential components of the DNA repair machinery, mRNA levels of several DNA repair and chromatin- particularly those involved in nonhomologous end- modifying genes as compared to cells without ectopic joining (NHEJ) (Blackburn, 2001; Chan and Blackburn, expression of hTERT (HFF). HFF+hTERT cells have 2002; de Lange, 2002). Wong et al. (2000) demonstrated better repair kinetics of DNA DSBs as well as DNA the role of telomerase and telomere function on the crosslinks along with less residual damage as compared cellular and organismal response to ionizing radiation. to the cells without ectopic expression of hTERT They reported that loss of telomerase activity per se had (HFF). The depletion of hTERT from cell extracts did no discernible impact on the response to ionizing not influence the DNA end-joining processes, nor did radiation, but cells with telomere dysfunction in late- the inactivation of Tert in mice affected meiotic generation TercÀ/À mice imparted a radiosensitivity recombination. Overall, this indicates that TERT may syndrome associated with accelerated mortality. They not be directly involved in DNA damage–repair process. also reported that the radiosensitivity of telomere The ability of hTERT to alter gene expression, dysfunctional cells correlated with defective DNA stabilize the telomeres and enhance DNA repair repair. Goytisolo et al. (2000) reported that short prompted us to postulate that one important pathway telomeres result in organismal hypersensitivity to ioniz- that impacts DNA repair may involve ATP metabolism. ing radiation in mammals. Radiosensitivity has been DNA repair requires ATP and its hydrolysis by several linked with higher residual DNA damage. It is possible repair proteins (Van Komen et al., 2000). An ATP- that the defective DNA repair could be because of the dependent branch migration and Holliday junction alteration or deficiency of components in the repair resolution during recombination and DNA repair is machinery. In this context, our gene expression compar- conserved from prokaryotes to mammals (Constantinou ison in isogenic cells with and without hTERT expres- et al., 2001). Nucleosomes are organized into ordered sion provided some valuable clues. hTERT expression arrays along the DNA by ATP-dependent chromatin enhances mRNA levels of several DNA repair genes and assembly and spacing factors such as ATP-utilizing chromatin-modifying factors. This increase in the chromatin assembly and remodeling factor ACE (Krude expression levels of such genes could be an important and Keller, 2001).Consistent with the role of ATP in component in the cascade of events leading to genome DNA repair, HFF+hTERT cells with ectopic expres- stabilization. In the present study, we show that sion of hTERT have relatively higher levels of ATP as HFF+hTERT cells have 12-fold less telomere fusions compared to those of HFF cells without ectopic and 20-fold reduced spontaneous chromosome breaks in expression of hTERT. G1 cell as compared to the parent HFF cells. Interestingly, ectopic expression of hTERT protein is What factors influence chromosome end associations? associated with enhanced genome stability and DNA One possible factor is loss or shortening of telomeres, as repair. How hTERT protein induces alteration in gene suggested by Counter et al. (1992); another is altered expression is not clear at present. One possibility is that chromatin structure (Smilenov et al., 1999; Dhar et al., hTERT interaction with telomeres leads to altered 2000). In our previous studies, we reported that the telomere–nuclear matrix associations, and such changes frequency of cells with chromosome end associations is may generate signals to down regulate senescence-

Oncogene Associating telomeres GG Sharma et al 143 associated gene expression and to upregulate DNA were resuspended in lysis buffer and telomerase activity was repair-associated genes. Our findings suggest that determined using the Telomerase PCR–ELISA kit (Boehringer upregulation of DNA repair genes coupled with the Mannheim) as described previously (Sawant et al., 1999). increase in the NTP pool levels could lead to telomere stability and the observed decrease in spontaneous Chromatin immunoprecipitation chromosome damage. Therefore, we propose that the Human telomeric DNA coimmunoprecipitated with hTERT functional interaction of hTERT with telomeres repre- antibody (Vaziri et al., 1999; Fu et al., 2000; Kharbanda et al., sents a key biological activity of telomerase in human 2000; Wood et al., 2001) on in vivo crosslinking was performed cells. by the standard procedure described previously (Braunstein et al., 1993; Hsu et al., 1999). Immunoprecipitated (IP) DNA was placed onto a membrane by using a dot-blotting apparatus and then hybridized with 32P-labeled DNA probe. The Materials and methods following probes were used for hybridization: telomeric DNA, total human genomic DNA and Alu DNA. The blots Cell culture were reused for hybridization with different probes after Normal human primary foreskin fibroblasts without (HFF) stripping. Quantitations of the DNA signals were achieved by and with (HFF+hTERT) ectopic hTERT expression and phosphorimaging. The relative increase of hTERT binding to GM5823 cells were cultured according to the procedure telomeres is determined by comparison with p21 binding to described recently (Vaziri and Benchimol, 1998; Vaziri et al., total genomic DNA. 1999; Wood et al., 2001). For most of the experiments, cells were examined at the PD of 46–50, unless specified. The Northern blot analysis maximum PD for HFF cells is 69, whereas HFF+hTERT cells have passed 150 PD. HeLa, 293 and CHO cells were Total RNA was isolated from cells using RNeasy Kit cultured according to the described procedure (Pandita and (QIAGEN Inc. Valencia, CA, USA). A total of 10 mg were Hittelman, 1992). resolved on 1.5% agarose/formaldehyde gels and transferred to a nylon filter by using standard methods (Gene Screen; DuPont). Hybridizations with hTR probe were done in Construction of the recombinant hTERT and TRF1 adenovirus Quikhyb (Stratagene) according to the manufacturer’s instruc- The adenovirus vector expressing hTERT (Ad.hTERT) or tions. Filters were exposed to autoradiographic film for up to 3 mutant hTERT (Ad.DhTERT) or TRF1 (Ad.TRF1) or mutant days. To test the uniform loading of the samples, blots were TRF1 (Ad.DTRF1) were created using the two-plasmid FLP stripped and reprobed with a 1.5-kb DNA fragment specific recombinase system (Fan et al., 2000; Ng et al., 2000). for 18S rRNA (ATCC clone HHCSA65). Deletions of 21 nucleotides (2250–2270) in the RT motif of hTERT gene and Myb domain in TERF1 were introduced by RT-PCR PCR and confirmed by Sequencing. The cloning strategy RNA was treated with RNase-free DNase (Boehringer placed the hTERT or TRF1 gene under the control of the Mannheim) (1 mg/ml) for 2 h at 371C, followed by heat mouse CMV early promoter and replaced the E1 region of inactivation at 651C for 10 min. RT reaction contained 1 mg adenovirus. This derivative was then cotransfected, into DNase-treated RNA, 0.25 1 mg/ml pdN6 random primers subconfluent 293 cells, with the large adenovirus genome- (Pharmacia), 1 Â first strand buffer (GIBCO-BRL), 0.5 mm containing plasmid pBHGfrtdelE13FLP, which encodes the dNTP (Pharmacia), and 200 units MMLV-RT (GIBCO-BRL), FLP recombinase but lacks a packaging signal necessary and were incubated for 1 h at 371C. PCR was performed by for the production of viable virus. Site-specific recombination using gene-specific primers along with the primers for alpha- between FRT sites present on both plasmids resulted in unit actin. The PCR samples were resolved by electrophoresis and length genomes containing both the packaging signal and the the products were quantitated by ImageQuant software. hTERT or DhTERT or TRF1orDTRF1 gene that replicated to give a packageable viral yield. Following initial character- ization of the viral DNA to confirm the presence of the Microarray analysis hTERT or TRF1 coding sequences, virus was plaque-purified A complete description of the composition, performance of the and small stocks were prepared for use in subsequent microarray system and methodologies used in this study have experiments. been described recently (Vafa et al., 2002) and can be viewed at (http://sequence.aecom.yu.edu/bioinf/funcgenomic.html). Genes Western analysis, immunoprecipitation and telomerase assay whose expression levels changed by a mean average of three- fold or more are presented in the Results section because of Immunoblots for hTERT proteins were done according to large experimental variations for genes with intensity values procedures described previously (Fu et al., 2000). Immuno- o100 and expression ratio o3. The identities of all genes precipitation of ectopically expressed hTERT or telomerase discussed in this report were confirmed by DNA sequence activity from cell extracts was performed by the procedure analysis of the corresponding cDNAs. described recently (Ford et al., 2000; Wood et al., 2001). Briefly, cells were broken open in lysis buffer (0.01% NP-40, Determination of telomere–nuclear matrix association 10 mm Tris pH7.5, 50 mm KCl, 5 mm MgCl2,2mm dithio- threitol, 20% glycerol plus protease inhibitors). Lysates were Nuclear matrix halos were isolated by removing histones and then centrifuged for 30 min at 16 000g to remove insoluble other loosely bound proteins as described (Smilenov et al., material and were used for immunoprecipitation. hTERT 1999). The nuclear halos were washed with a restriction proteins were immunoprecipitated with anti-hTERT antibo- enzyme buffer, 6 Â 106 halos were cleaved in a volume of 0.5 ml dies and protein A/G Sepharose. Immunoprecipitates were containing 1000 Units of restriction enzyme Sty1 for 3 h at washed three times with lysis buffer. After washing, the beads 371C and the nuclear matrices were pelleted by centrifugation.

Oncogene Associating telomeres GG Sharma et al 144 To purify released and attached DNA fragments to the nuclear this technique for the determination of DNA crosslinks is that matrix, both fractions were treated with proteinase K in a under alkaline conditions, crosslinked DNA behaves as a solution containing 10 mm EDTA, 0.5% sodium dodecyl duplex and this DNA can be separated from totally denatured sulfate and 10 mM Tris-HCl (pH 7.4) and incubated overnight (noncrosslinked) DNA by alkaline elution (Pommier et al., at 371C. DNA was purified as described previously (Smilenov 1985). The details of estimating DNA crosslinks by the filter et al., 1999). Agarose gel electrophoresis was performed for alkaline elution method have been described previously fractionation of DNA. For Southern blot analysis, equal (Nishikawa et al., 1992). To allow detection of DNA–DNA volumes from about 106 halos were fractionated on 0.8% crosslinks, harvested cells were treated with 3 Gy of g-rays at a agarose gels. Prior to DNA loading, RNase was added to a dose rate of 1 Gy/min. final concentration of 10 mg/ml. Fractionation of DNA, DNA crosslinking was estimated as transfer to a Hybond-N membrane, or slot blotting of DNA, hybridization with a 32P-labeled (TTAGGG) probe, and 1 À r0 5 Crosslinkratio ðcGy equivalentÞ¼ À 1Â300 detection were done as described previously (Smilenov et al., 1 À r 1999). Quantitation and comparison of the telomeric DNA was performed on a phosphorimager using total (T), released in the supernatant (S) or in the pellet (P). where r and r0 are the fractions of DNA remaining on the filter for CDDP-treated and control cells, respectively. Determination of terminal restriction fragment (TRF) analysis and detection of telomeres In vitro repair assay For the determination of telomere length, DNA was isolated HeLa cells were grown in suspension in a spinner flaskin and digested with the restriction enzymes Rsa1 and Hinf1, S-MEM supplemented with 5% BCS. Whole cell extracts which do not cut TTAGGG sequences, processed for were prepared by a procedure described previously (Iliakis 32 et al., 1999) except that the salt concentration after the dounce fractionation, and hybridized with a P-labeled (TTAGGG)5 m probe. Detection and measurement of TRF lengths were homogenization step was adjusted to 500 m KCl. In all, 50 ml anti-hTERT serum or 50 ml of normal rabbit serum was performed as described earlier using ImageQuant version 1.2., 1 Molecular Dynamics (Smilenov et al., 1999). Detection of incubated overnight at 4 C with 25 ml of rProtein A (RepliGen) telomeres on metaphases was done by FISH using telomere- and 25 ml of rProtein G (GiBCO-BRL) sepharose beads in 1 ml specific probe (Dhar et al., 2000). of dialysis buffer containing 50 mg/ml of BSA under contin- uous slow rotation. Subsequently, the beads were washed and distributed into three tubes. A volume of 200 ml of HeLa G1 Chromosome studies whole-cell extract (12.5 mg/ml) was added to one tube. After Johnson and Rao (1970) discovered the premature chromo- rotating at 41C for 2 h, the supernatant was transferred to the some condensation (PCC) technique that allows chromatin of second tube, and after a further 2 h to the third tube. The G1 or G2 phase cells to become visibly condensed as interphase supernatant was removed from the last tube, tested for the chromosomes, when such cells are fused with mitotic cells. level of immunodepletion by Western blotting, telomerase Factors from mitotic cells cause the interphase chromatin to activity and ability to support plasmid end joining. undergo a prophase-like reaction whereby the nuclear mem- Plasmid pSP65 (3.0 kb) was prepared using CsCl2–EtBr brane dissolves and the interphase chromatin is induced to gradients, and then digested with SalI and used for reactions. condense into chromosomes prematurely. Thus, this technique Plasmid end-joining reactions were performed in 20 mm m m m allows direct visualization of chromosomes in interphase cells. HEPES–KOH (pH 7.5), 10 m MgCl2,80m KCl, 1 m Cells from G1 phase exhibit a single chromatid per chromo- DTT, 0.25 mg of DNA, and different amounts of cell extract 1 some as compared to two chromatids at G2 phase or in in a final volume of 20 mlat25C for 1 h. Reactions were metaphase cells. The premature chromosome condensation stopped by addition of 2 ml of 5% SDS, 2 ml of 0.5 M EDTA technique in combination with in situ hybridization was used and 1 ml of proteinase K (10 mg/ml) and were then incubated to visualize telomeric associations and chromosome damage for 1 h at 371C. One-half of the reaction was loaded on 0.7% directly in G1 phase cells (Pandita et al., 1994, 1995). A total of agarose gel and run at 45 V (2 V/cm) for 5 h. Gels were stained 200 PCC spreads were microscopically examined for frequen- in SYBR Gold and scanned in a Fluoroimager (Molecular cies of chromosome end associations and chromosome Dynamics). Quantitation was carried out using the Image- damage. Quant software. Plasmid end joining was also performed in the presence of Determination of DNA DSBs anti-hTERT serum. Anti-hTERT serum and normal rabbit serum were diluted from 1 : 1 to 1 : 32 in PBS, 1 ml was added to The level of DNA double-strand breaks (DSBs) in cells was the end-joining reaction prior to the addition of ATP and estimated using a neutral DNA filter elution method described plasmid DNA. After incubation at 251C for 10 min, ATP and previously (Pandita and Hittelman, 1992). The relative elution plasmid DNA were added to start the rejoining reaction. (RE) was calculated as RE ¼Àlog (Fi/Fc), where Fi and Fc are the fractions of DNA left on the filter for the irradiated (i) and Meiotic chromosome preparations, immunocytochemistry and 7 unirradiated (c) cells. In all experiments, Fc was 97 2%. fluorescence in situ hybridization Mice deficient for TERT were obtained from Dr L Harrington DNA interstrand crosslinking repair (Ontario Cancer Institute/Amgen Institute, Toronto, Canada). The covalent interactions between the two strands constituted The alleles are carried on mixed genetic background mice by crosslinks in duplex DNA molecules can be detected by gel- (129SvEv  BlackSwiss). Generally, mice of about 4 months electrophoretic mobility, filter alkali elution and special of age, were killed and testes resected for further processing or density-labeling procedure (Roberts, 1978). Here, we employed instant snap freezing in liquid nitrogen. Structurally preserved the filter alkaline elution method to determine cis-platin- suspension nuclei were prepared by crosslinking fixation with induced DNA interstrand crosslinking. The principle behind PBS-buffered formaldehyde (Scherthan et al., 2000). A

Oncogene Associating telomeres GG Sharma et al 145 polyclonal rabbit anti-SCP3 antiserum was utilized to detect the radioactivity on the disks was determined using liquid axial cores and complete SCs (Pandita et al., 1999). Antibodies scintillation counting and compared with that in the standard to TRF1 were kindly provided by Titia de Lange (Rockefeller dNTP samples. University, New York, USA). Immunofluorescence staining of Ribonucleotide (NTP): For ribonucleotide analyses, NTPs the SC lateral element SCP3 protein was done as described were obtained from Sigma and were used as standard on high- previously (Pandita et al., 1999). A directly labeled pressure liquid chromatography (HPLC). Exponentially grow- (TTAGGG)3 PNA probe was used for FISH to detect ing cells were used to determine normal ribonucleotide pools. telomere repeats (telo-FISH). Telo-FISH was combined with After being washed with phosphate-buffered saline, cells were IF. This was achieved by first performing immunostaining and processed using standard extraction procedures with perchlo- then preparations were subjected to simultaneous denaturation ric acid (Plunkett et al., 1980). Ribonucleotides were separated in the presence of the telomere probe (Scherthan et al., 2000). on an anion-exchange Partisil-10 SAX column (Waters Labeling, FISH and detection of oligonucleotides were Corporation, Milford, MA, USA) using HPLC as described performed as described (Scherthan et al., 2000). in detail previously (Rodriguez et al., 2000). The intracellular concentration was calculated and expressed as the quantity of Determination of intracellular nucleotide pools nucleotides contained in the extract from a given number of cells of a determined mean volume. This calculation assumes Deoxynucleotides (dNTP): For measurement of cellular dNTP that nucleotides are uniformly distributed in total cell water. In pools, dNTP standards were obtained from Pharmacia general, the lower limit of detection of this assay was about (Piscataway, NJ, USA). Nucleotides were extracted using 10 pmol in an extract of 2 Â 107 cells, corresponding to a 60% methanol for the determination of dNTP pools. The cellular concentration of about 2 mM. DNA polymerase assay as modified by Sherman and Fyfe (1989) was used to quantitate dNTPs in the cell extracts. A Klenow fragment of DNA polymerase I lacking exonuclease Acknowledgements activity (US Biochemical Corporation, Cleveland, OH, USA) We thankDrs R Kucheralapati and Y Heyer for microarray was used to start a reaction in a mixture that contained 100 mm facility, and Drs D Thanos and S Mitra for the suggestions on HEPES buffer, pH 7.3, 10 mm MgCl2, 7.5 g BSA, and synthetic ChIP. We thankDr Titia de Lange for providing us reagents oligonucleotides of defined sequences as templates annealed to for this study. Thanks are due to T. Rounsville for making a primer, [3H]dATP or [3H]dTTP and either standard (dNTP) adenovirus constructs. This workwas supported by NIH or the extract from 4 or 6 Â 105 cells. Reactants were incubated Grant NS34746 and A-T Children’s project to TKP, NCI CN- for 1 h and applied to filter disks. After the disks were washed, 15015 to JWS and CA13696 to CSHY.

References

Allsopp RC, Vaziri H, Patterson C, Goldstein S, Younglai EV, Fu W, Killen M, Culmsee C, Dhar S, Pandita TK and Futcher AB, Greider CW and Harley CB. (1992). Proc. Natl. Mattson MP. (2000). J. Mol. Neurosci., 14, 3–15. Acad. Sci. USA, 89, 10 114–10 118. Gottschling DE, Aparicio OM, Billington BL and Zakian VA. Baur JA, Zou Y, Shay JW and Wright WE. (2001). Science, (1990). Cell, 63, 751–762. 292, 2075–2077. Goytisolo FA, Samper E, Martin-Caballero J., Finnon P, Blackburn EH. (2001). Cell, 106, 661–673. Herrera E., Flores JM, Bouffler SD and Blasco MA. (2000). Bodnar AG, Ouellette M, Frolkis M, Holt SE, Chiu CP, J. Exp. Med., 192,1625–1636. Morin GB, Harley CB, Shay JW, Lichtsteiner S and Wright Hastie ND, Dempster M, Dunlop MG, Thompson AM, Green WE. (1998). Science, 279, 349–352. DK and Allshire RC. (1990). Nature, 346, 866–868. Braunstein M, Rose AB, Holmes SG, Allis CD and Broach Hemann MT, Strong MA, Hao LY and Greider CW. (2001). JR. (1993). Genes Dev., 7, 592–604. Cell, 107, 67–77. Chan SW and Blackburn EH. (2002). Oncogene, 21, 553–563. Hsu HL, Gilley D, Blackburn EH and Chen DJ. (1999). Proc. Choi D, Whittier PS, Oshima J and FunkWD. (1997). FASEB Natl. Acad. Sci. USA, 96, 12 454–12 458. J., 15, 1014–1020. Iliakis G, Wang Y and Wang HY. (1999). Methods Mol. Biol., Constantinou A, Davies AA and West SC. (2001). Cell. 104, 113, 543–553. 259–268. Johnson RT and Rao PN. (1970). Nature, 226, 717–722. Counter CM, Avilion AA, LeFeuvre CE, Stewart NG, Greider Kawakaim Y, Kitamoto M, Nakanishi T, Yasui W, Tahara E, CW, Harley CB and Bacchetti S. (1992). EMBO J., 11, Nakayama J, Ishikawa F, Tahara H, Ide T and Kajiyama G. 1921–1929. (2000). Oncogene, 19, 3888–3893. de Lange T. (2002). Oncogene, 21, 532–540. Kharbanda S, Kumar V, Dhar S, Pandey P, Chen C, Dhar S, Squire JA, Hande MP, Wellinger RJ and Pandita TK. Majumder P, Yuan ZM, Whang Y, Strauss W, Pandita (2000). Mol. Cell. Biol., 20, 7764–7772. TK, Weaver D and Kufe D. (2000). Curr. Biol., 10, 568–575. Etheridge KT, BanikSSR, Armbruster BN, Zhy Y, Terns Kondo S, Kondo Y, Li G, Silverman RH and Cowell JK. RM, Terns MP and Counter CM. (2002). J. Biol. Chem., (1998). Oncogene, 16, 3323–3330. 277, 24764–24770. Krude T and Keller C. (2001). Cell Mol. Life Sci., 58, Fan Z, Chakravarty P, Alfieri A, Pandita TK, Vikram B and 665–672. Guha C. (2000). Cancer Gene Ther., 7, 1307–1314. Kyrion G, Liu K, Liu C and Lustig AJ. (1993). Genes Dev. 7, Feng J, FunkWD, Wang SS, Weinrich SL, Avilion AA, Chiu 1146–1159. CP, Adams RR, Chang E, Allsopp RC, Yu J et al. (1995). Levy MZ, Allsopp RC, Futcher AB, Greider CW and Harley Science, 269,1236–1241. CB. (1992). J. Mol. Biol., 225, 951–960. Ford PP, Suh JM, Wright WE and Shay JW. (2000). Mol. Cell Liu Y, Snow BE, Hande MP, Yeung D, Erdmann NJ, Biol., 20, 9084–9091. Wakeham A, Itie A, Siderovski DP, Lansdorp PM,

Oncogene Associating telomeres GG Sharma et al 146 Robinson MO and Harrington L. (2000). Curr. Biol., 10, Sherman PA and Fyfe JA. (1989). Anal. Biochem., 180, 1459–1462. 222–226. Luderus ME, van Steensel B, Chong L, Sibon OC, Cremers FF Smilenov LB, Dhar S and Pandita TK. (1999). Mol. Cell Biol., and de Lange T. (1996). J. Cell Biol., 135, 867–881. 19, 6963–6971. Martin-Rivera L, Herrera E, Albar JP and Blasco MA. (1998). Smith S and de Lange T. (1997). Trends Genet., 13, Proc. Natl. Acad. Sci. USA, 95, 10 471–10 476. 21–26. Ng P, Cummings DT, Evelegh CM and Graham FL. (2000). Steinert S, Shay JW and Wright WE. (2000). Biochem. Biotechniques, 29, 524–526. Biophys. Res. Commun., 273, 1095–1098. Nishikawa K, Rosenblum MG, Newman RA, Pandita TK, Struhl K. (1998). Genes Dev., 12, 599–606. Hittelman WN and Donato NJ. (1992). Cancer Res., 52, Tahara H, Yasui W, Tahara E, Fujimoto J, Ito K, Tamai K, 4758–4765. Nakayama J, Ishikawa F, Tahara E and Ide T. (1999). Olovnikov AM. (1992). Izv. Akad. Nauk Ser. Biol., 4, 641–643. Oncogene, 18, 1561–1567. Ouellette MM, McDaniel LD, Wright WE, Shay JW and Vafa O, Wade M, Kern S, Beeche M, Pandita TK, Hampton Schultz RA. (2000). Hum. Mol. Genet., 9, 403–411. GM and Wahl GM. (2002). Mol. Cell, 9, 1031–1044. Pandita TK. (2002). Oncogene, 21, 611–618. Van Komen S, Petukhova G, Sigurdsson S, Stratton S and Pandita TK and Hittelman WN. (1992). Radiat. Res., 130, 94– Sung P. (2000). Mol. Cell, 6, 563–572. 103. Vaziri H and Benchimol S. (1998). Curr. Biol., 8, 279–282. Pandita TK, Gregoire V, Dhingra K and Hittelman WN. Vaziri H, Squire JA, Pandita TK, Bradley G, Kuba RM, (1994). Cytogenet. Cell. Genet., 67, 94–101. Zhang H, Gulyas S, Hill RP, Nolan GP and Benchimol S. Pandita TK, PathakS and Geard CR. (1995). Cytogenet. Cell (1999). Mol. Cell Biol., 19, 2373–2379. Genet., 71, 86–93. Vaziri H, West MD, Allsopp RC, Davison TS, Wu YS, Pandita TK, Westphal CH, Anger M, Sawant SG, Geard CR, Arrowsmith CH, Poirier GG and Benchimol S. (1997). Pandita RK and Scherthan H. (1999). Mol. Cell Biol., 19, EMBO J., 16, 6018–6033. 5096–5105. Vaziri H, Dragowska W, Allsopp RC, Thomas TE, Harley CB Plunkett W, Hug V, Keating MJ and Chubb S. (1980). Cancer and Lansdorp PM. (1994). Proc. Natl. Acad. Sci. USA, 91, Res., 40, 588–591. 9857–9860. Pommier Y, Schwartz RE, Zwelling LA and Kohn KW. Weinrich SL, Pruzan R, Ma L, Ouellette M, Tesmer VM, (1985). Biochemistry, 24, 6406–6410. Holt SE, Bodnar AG, Lichtsteiner S, Kim NW, Trager JB, Renauld H, Aparicio OM, Zierath PD, Billington BL, Taylor RD, Carlos R, Andrews WH, Wright WE, Shay JW, Chhablani SK and Gottschling DE. (1993). Genes Dev., 7, Harley CB and Morin GB. (1997). Nat. Genet., 17, 1133–1145. 498–502. Roberts JJ. (1978). Adv. Radiat. Biol., 7, 211–435. Wong KK, Chang S, Weiler SR, Ganesan S, Chaudhuri J, Rodriguez Jr CO, Plunkett W, Paff MT, Du M, Nowak B, Zhu C, Artandi SE, Rudolph KL, Gottlieb GJ, Chin L, Ramakrishna P, Keating MJ and Gandhi V. (2000). J. Alt FW and DePinho RA. (2000). Nat. Genet., 26, Chromatogr. B. Biomed. Sci. Appl., 745, 421–430. 85–88. Rudd GN, Hartley JA and Souhami RL. (1995). Cancer Wood LD, Halvorsen TL, Dhar S, Baur JA, Pandita RK, Chemother. Pharmacol., 35, 323–326. Wright WE, Hande MP, Calaf G, Hei TK, Levine F, Shay Sawant SG, Gregoire V, Dhar S, Umbricht CB, Cvilic S, JW, Wang JJ and Pandita TK. (2001). Oncogene, 20, Sukumar S and Pandita TK. (1999). FASEB J., 13, 1047–1054. 278–288. Scherthan H. (2001). Nat. Rev. Mol. Cell Biol., 2, 621–627. Wright WE and Shay JW. (1992). Trends Genet. 8, 193–197. Scherthan H, Jerratsch M, Dhar S, Wang YA, Goff SP and Zhang X, Mar V, Zhou W, Harrington L and Robinson MO. Pandita TK. (2000). Mol. Cell Biol., 20, 7773–7783. (1999). Genes Dev. 13, 2388–2399.

Oncogene