Telomere Shortening and Telomerase Reverse Transcriptase Expression in Preinvasive Bronchial Lesions

Lung cancer is the leading cause of cancer deaths in the Western world, killing about one million people worldwide each year. Most patients with primary lung cancer present advanced disease at the time of initial diagnosis, with a 5-year survival rate estimated to be <15%; therefore, a way to improve the outcome of this disease is an earlier detection of lung cancer precursors, such as preneoplastic bronchial lesions, and the assessment of their potential to progress to invasive lesions. Bronchial carcinogenesis is a multicentric and a multistep process known as “field cancerization,” which associates multifocal morphologic transformation of the bronchial epithelium and multistep accumulation of molecular, genetic, and epigenetic abnormalities, both initiated by tobacco smoke carcinogens, such as benzo(a)pyrene and nitrosamines (1). Histologically, according to the criteria provided by the recent WHO classification of lung cancers and precursors (2), preinvasive lesions of the bronchial epithelium are represented by squamous metaplasia, dysplasia of various grade of severity (mild, moderate, and severe), and carcinoma in situ (CIS). Concomitant molecular and genetic abnormalities consist in mutations of oncogenes and deletions, mutations, and loss of heterozygosity of tumor suppressor genes implicated in cell cycle regulation and apoptosis. The main tumor suppressor barriers to uncontrolled proliferation are represented by p53-ARF and p16^INK4a-Rb pathways, which function in cell cycle checkpoints and receive high pressure for invalidation during the carcinogenic process (3, 4). As a result, p53 mutation occurs in >60% of high-grade preneoplastic lesions and correlates with protein stabilization and immunohistochemical detection in 36% of moderate dysplasia, 59% of severe dysplasia, and 69% of CIS (5). Furthermore, P53 accumulation in preneoplastic bronchial lesions seems to be predictive for their progression to invasion (5, 6). Deregulation of p53 transcription pathway as well as an imbalance of Bax and Bcl2 expression in preinvasive bronchial lesions endow resistance to apoptosis as these proteins exhibit proapoptotic and antiapoptotic properties, respectively (6, 7). Alternatively, Rb pathway alterations participate to G₁ arrest evasion, with loss of p16^INK4 occurring at the level of squamous dysplasia (8). Cyclin D1, which allows Rb phosphorylation and inactivation by releasing the critical transcriptional factor for G₁-S transition E2F1, is overexpressed in 30% of dysplasia (8, 9). Overall, these abnormalities are more frequently observed in high-grade than in low-grade dysplasia (5, 9) and accumulation of molecular abnormalities is predictive of risk of lung cancer in smokers. Indeed, more than two aberrant expressions of P53, cyclin D1, cyclin E, Bax, or Bcl2 in dysplasia are correlated with progression to CIS and/or lung cancer (10).

However, although p53 and Rb pathway inactivation is specifically required for cell transformation, cellular immortality and unlimited proliferation need the cooperation of a ribonucleoprotein, telomerase, which prevents telomere shortening and allows tumor cells to escape from lethal crisis (11). Telomeres, composed of hexameric DNA repeat sequences (TTAGGG) prevent chromosomal rearrangements and fusion and shorten at each cell division due to incomplete telomeric DNA replication by DNA polymerases at the 3′ end. In the absence of adequately sized telomeres, normal somatic cells undergo cellular senescence as short telomeres are perceived as damaged DNA leading to p53/ATM pathway activation (12), which is called cellular mortality stage 1 (13). In contrast, tumor cells, lacking of P53- and Rb-mediated checkpoints, can escape mortality stage 1 and proliferate until the mortality stage 2 or “crisis,” where they suffer huge genetic instability. At this stage, telomeric dysfunction, leading to breakage-fusion-bridge cycles and formation of dicentric chromosomes, allows in turn new genetic abnormalities and promotes tumor progression in surviving cells (14). Clones surviving crisis either reactivate telomerase, which stabilizes telomere length by adding telomeric sequences onto chromosome ends, or lengthen telomeres through an alternative mechanism (15). Telomerase ribonucleoprotein complex is composed of two main subunits, human telomerase reverse transcriptase (hTERT), representing the catalytic subunit and the limiting factor for enzyme activity, and human telomerase RNA component (hTERC), the RNA component serving as template for telomeric repeat synthesis (16). Telomerase expression has been widely reported in most human cancers and in up to 85% of lung carcinomas (17–21). In preneoplastic lesions, elevated hTERT mRNA was detected in dysplastic cells during oral carcinogenesis, in CIS of the breast, and in cervical intraepithelial neoplasia grade III lesions (22–24). hTERT mRNA and telomerase activity were also reported to increase proportionally to the severity of histologic changes in short series of preneoplastic bronchial lesions, concomitantly with elevated hTERC levels, which were detected as early as squamous metaplasia (22–27). Mechanisms of telomerase reexpression during carcinogenesis remain incompletely understood, involving mainly hTERT promoter activation by c-myc, Mad/Max, and SP1 at the transcriptional level (28, 29), posttranslational modifications, such as hTERT phosphorylation (30–33), and subcellular delocalization at the protein level (34–36).

Telomere shortening likely represents an earlier event in carcinogenesis than telomerase activation. Indeed, telomere shortening occurs in prostate and pancreatic carcinogenesis at the level of intraepithelial neoplasia (37, 38). Recently, a new method based on fluorescence in situ hybridization (FISH) has been developed to assess telomere length at the level of individual cells and in tissue sections from archival material. This technique provides morphologic information at the level of preinvasive lesions and enables independent analysis of epithelial and stromal cell telomeres without requiring large unfixed tissue samples as Southern blot does (39).

With the aim to determine the sequential timing of telomerase activation during bronchial carcinogenesis along with p53 and Rb pathway alterations and its relationship with telomere shortening, we analyzed hTERT expression by both in situ hybridization and immunohistochemistry, in comparison with that of cell cycle and apoptotic regulators P53, P16^INK4, cyclin D1, Bax, Bcl2, and Ki67 expression in preinvasive lesions and their normal and invasive counterparts. Concomitantly, we did a qualitative assessment of telomeric signal intensity using FISH.

Patients and Tissue Samples. Formalin-fixed bronchial specimen were collected from lung resections in 27 patients for lung cancer including 19 invasive squamous cell carcinoma, 4 basaloid carcinoma, 2 adenocarcinoma, and 2 large cell carcinoma according to the WHO classification (2). The 27 patients, 26 men and 1 woman, ranging in age from 41 to 79 years (mean, 60.9 years), were staged according to the International Union Against Cancer classification in 11 stage I, 5 stage II, and 11 stage III. All tissue samples were obtained within 1 hour after surgical removal.

Preinvasive lesions were classified according to the WHO classification criteria (2) by two pathologists experienced in lung pathology (E. Brambilla and S. Lantuejoul). Preinvasive lesions including 17 squamous metaplasia, 18 mild dysplasia, 18 moderate dysplasia, 20 severe dysplasia, and 33 CIS were compared with 21 normal or hyperplastic bronchial mucosae and 27 concomitant carcinoma, 23 of them being located in the vicinity but distinct from the preinvasive lesions. Squamous metaplasia and mild dysplasia were considered as low-grade dysplasia, whereas moderate dysplasia, severe dysplasia, and CIS were considered as high-grade dysplasia. Cases were considered as positive for hTERT and P53 when >20% of the cells were stained and positive for cyclin D1 when >5% of the cells were stained. Loss of P16^INK4 expression was defined by a negative staining observed in dysplastic cells contrasting with a positive nuclear staining expressed by cells serving as internal controls (fibroblasts, endothelial cells). Expression of P16^INK4 in internal controls was considered as relevant when at least 10% of the cells show a nuclear expression (8).

Immunohistochemical Analysis. Three-micrometer-thick serial formalin or Bouin fixed serial sections were deparaffinized and incubated at room temperature with primary monoclonal antibodies anti-hTERT (clone 44F12, Novocastra, Newcastle upon Tyne, United Kingdom) at the dilution 1:20 (35 μg/mL), anti-P53 (DO7, Dakopatts, Glostrup, Denmark) at the dilution 1:75 (5 μg/mL), anti-cyclin D1 (AM29, Calbiochem, Cambridge, United Kingdom) at the dilution 1:50 (5 μg/mL), anti-Bax (N19, Santa Cruz Biotechnology, Inc., Santa Cruz, CA) at the dilution 1:50 (4 μg/mL), Bcl2 (124, Dakopatts) at the dilution 1:50 (3 μg/mL), and anti-Ki67 (KiS5, Immunotech, Marseille, France) at the dilution 1:50 (1 μg/mL). Polyclonal antibody against P16^INK4 (C20, Santa Cruz) was used at the dilution 1:50 (2 μg/mL). Microwave antigen retrieving was done in citrate buffer (pH 6) for hTERT, Bax, Bcl2, and Ki67 immunostaining, in 1 mmol/L EDTA (pH 9) for cyclin D1, and in Tris citrate (pH 9.15) for P53. A three-stage indirect immunoperoxidase technique was done on Ventana NexES staining module (Ventana Medical Systems, Tucson, AZ). Negative control consisted in omission of the primary antibody and incubation with immunoglobulins of the same species. Levels of protein expression were evaluated by two pathologists (S. Lantuejoul and E. Brambilla) and scored using the product of the percentage of tumor cell positive nuclei by the staining intensity (0, null; 1, faint; 2, moderate; 3, strong); scores range from 0 to 300. Nuclear and/or nucleolar staining was considered as a specific pattern of hTERT expression. Lymphocytes in the bronchial mucosae served as positive internal control for hTERT, Bcl2, and Bax, whereas endothelial cells and fibroblasts were considered as negative internal controls for hTERT and positive for P16^INK4. In the absence of these positive internal controls, immunostaining were considered as nonevaluable. P16^INK4 loss of expression was considered as nonevaluable in 6 lesions among 154 (4%). Positive external controls for P53 and cyclin D1 were tissues known to express these antigens, and normal cells represent negative internal control. Because the number of sections representative of the lesion was getting short in 14 cases, one marker among seven was missing.

In situ Hybridization. For riboprobe generation and RNA in situ hybridization, a TOPO TA cloning vector (PCR II-TOPO, Invitrogen, Carlsbad, CA) was used containing a 430 bp EcoRV-BamH1 fragment of the hTERT cDNA as previously described (40) to generate a digoxigenin-labeled RNA probe (riboprobe) specific for the antisense strand of the hTERT cDNA, which hybridizes to the full-length transcript complementary to the conserved region from exon 7 to exon 12 corresponding to the catalytic domain of the enzyme. The plasmid was linearized with EcoRV and then transcribed in vitro with SP6 RNA polymerase (Promega, Madison, WI) using a digoxigenin-UTP labeling mixture (DIG RNA labeling kit; Roche Diagnostics Inc., Indianapolis, IN). The resulting digoxigenin-labeled RNA probe was mixed with RNase inhibitor (Roche Diagnostics) and stored in aliquots at −80°C.

In situ hybridization was done in 67 samples of formalin-fixed paraffin-embedded tissue sections in RNase-free conditions including 7 normal/hyperplastic mucosae, 8 mild dysplasia, 10 moderate dysplasia, 12 severe dysplasia, 16 CIS, and 14 invasive carcinoma. The slides were deparaffinized and then transferred on the heating blocks of a Discovery module (Ventana Medical Systems, Strasbourg, France) for an automated in situ hybridization procedure. Briefly, the sections were treated with 2.5μg/mL proteinase K (Roche Diagnostics, Meylan, France) for 14 minutes at 37°C, washed in TBS buffer (reaction buffer, Ventana Medical Systems), and postfixed in 4% paraformaldehyde for 8 minutes at room temperature. Before hybridization, the antisense riboprobe diluted at 800 ng/mL in hybridization buffer [10% 20× sodium saline citrate (3 mol/L sodium chloride and 0.3 mol/L sodium citrate), 50% deionized formamide, 250 μg/mL predenatured salmon sperm DNA, 100 mg/mL dextran sulfate, 2% 100× Denhardt's solution (2% Ficoll 400, 2% polyvinylpyrrolidone, 2% bovine serum albumin), 2% DTT, and 400 μg/mL yeast tRNA] was denatured in boiling water for 15 minutes. Each section was incubated with 100 μL hTERT antisense probe for 8 hours at 42°C. Sense and antisense actin riboprobes have been used as negative and positive controls, respectively, to assess the good mRNA preservation in the same conditions on duplicate slides. After hybridization, the sections were washed twice in 2× sodium saline citrate (Ribowash buffer, Ventana Medical Systems) at 37°C. For immunodetection of hybrids, the slides were then incubated overnight at room temperature with 100 μL of an alkaline phosphatase–conjugated antidigoxigenin antibody (Roche Diagnostics) diluted 1:200 in 0.9% NaCl, 2% normal sheep serum, and 0.3% Triton X-100. The slides were washed twice in TBS, and then briefly rinsed with 100 mmol/L Tris-HCl, 100 mmol/L NaCl, and 50 mmol/L MgCl₂ (pH 9.5). Alkaline phosphatase was detected using 5-bromo-4-chloro-3-indolyl phosphate and nitroblue tetrazolium chloride (Roche Diagnostics) as chromogens. Levels of hTERT mRNA expression were evaluated by two pathologists (S. Lantuejoul and E. Brambilla) and scored using the product of the positive cells by the intensity of staining (1, mild; 2, moderate; 3, strong), total scores ranging from 0 to 300.

Telomere Fluorescence In situ Hybridization. Telomere FISH was done with the DAKO telomere peptide nucleic acid FISH kit (DAKO, Glostrup, Denmark) only in formalin-fixed sections including 7 normal or hyperplastic bronchial mucosae, 35 preneoplastic lesions including 6 squamous metaplasia, 6 mild dysplasia, 8 moderate dysplasia, 6 severe dysplasia, and 9 CIS, and 9 invasive carcinoma. Briefly, deparaffinized sections underwent a microwave heat–induced antigen retrieval 15 minutes in citrate buffer (pH 6). They were then placed in 0.1% PBS Tween for 2 minutes and in TBS twice for 5 minutes. Slides were then placed in Proteinase K Pretreatment Solution (diluted 1:20) for 20 minutes at room temperature. They were thoroughly placed in 70% ethanol for 2 minutes, 85% ethanol for 2 minutes, and ethanol 96% for 2 minutes, and then air-dried. Ten microliters of a Cy3-labeled specific peptide nucleic acid were applied to each sample, which was then coverslipped, and denaturation was done for 5 minutes at 85°C. Slides were then moved in the dark for hybridization 2 days at 37°C. They were then placed in rinse solution (diluted 1:50) for 1 minute and in wash solution (diluted 1:50) for 5 minutes at 37°C, and were immersed in cold ethanol series (70%, 85%, 96%). They were then air dried and counterstained with Hoescht stain and mounted with aqueous mounting medium (Mowiol 4-88, Calbiochem-Merck, Darmstadt, Germany). Nuclear telomeric spots were considered as specific signals and were scored as 0, no staining; 1, spots observed in <10% of the nuclei; 2, spots in 20% to 60% of the nuclei; and 3, spots in 70% and more of the nuclei.

Statistical Analysis. The staining scores were compared in different categories using Mann-Whitney U test and Kruskal-Wallis H test. The χ² test was used to test the association between two categorical variables. P < 0.05 was considered statistically significant. All the tests were done with the StatView program (Abacus Concepts, Berkeley, CA).

Immunohistochemical Analysis of Human Telomerase Reverse Transcriptase. Mild to moderate nuclear immunostaining with nucleolar reinforcement (intensity 1 or 2) was observed on ∼20% of basal cells of normal or hyperplastic bronchial epithelium. Some serous cells of bronchial glands exhibited a faint nuclear staining, whereas fibroblasts and endothelial cells remained negative. A lymphocyte subpopulation, present in the bronchial mucosae, presented a moderate nuclear staining and served as internal positive control with a staining intensity of 2 for evaluating the intensity of staining in neoplastic cells (from 0 to 3; Fig. 1A).

hTERT was expressed in 14 of 20 (70%) normal/hyperplastic epithelium restricted to the basal layer. In 12 of 16 (75%) squamous metaplasia, in 15 of 17 (88%) mild dysplasia, in 16 of 16 (100%) moderate dysplasia, in 20 of 20 (100%) severe dysplasia, in 31 of 32 (96%) CIS, and in 27 of 27 (100%) invasive carcinoma, immunostaining was more diffuse and intense (Fig. 1C). Five lesions (one squamous metaplasia, two moderate dysplasia, one CIS, and one normal or hyperplastic bronchial mucosa) were lost on serial sections. Mean scores and standard deviations of hTERT expression according to histology are given in Table 1. Telomerase mean levels of expression were increasingly expressed from normal/hyperplastic mucosae, squamous metaplasia, and mild dysplasia to moderate dysplasia, severe dysplasia, and CIS. Mean scores of hTERT expression decreased in corresponding invasive squamous cell carcinoma. No case of normal or hyperplastic bronchial mucosa presented levels of hTERT ≥ 100, in contrast with 4 of 17 (23%) of squamous metaplasia, 2 of 17 (11%) of mild dysplasia, 12 of 16 (75%) of moderate dysplasia, 16 of 20 (80%) of severe dysplasia, 29 of 32 (90%) of CIS, and 23 of 27 (85%) of invasive carcinoma. Levels of protein expression assessed by scores were significantly different according to histologic grade of preinvasive lesions (Kruskal-Wallis test, P < 0.0001). Such statistical difference was maintained between low-grade dysplasia (squamous metaplasia and mild dysplasia) and high-grade dysplasia (including moderate dysplasia, severe dysplasia, and CIS) levels of expression (Mann-Whitney test, P < 0.0001).

Nucleolar staining was predominantly observed in basal cell nuclei of normal/hyperplastic bronchial epithelium, whereas all preinvasive and invasive lesions regardless of grade exhibited a diffuse nuclear staining with nucleolar reinforcement.

In situ Hybridization with Human Telomerase Reverse Transcriptase Riboprobe.In situ hybridization staining with hTERT riboprobe was located in the cytoplasm and had the same cell distribution than that obtained by immunohistochemistry. A positive staining was observed in basal bronchial cells of normal and hyperplastic bronchial epithelium, some serous cells of bronchial glands, and a lymphocyte subpopulation (Fig. 1B). Hybridization with sense actin riboprobe gave the expected negative signal in all cases.

hTERT mRNA were detected at variable levels from normal or hyperplastic bronchial mucosae to preinvasive lesions and invasive carcinoma, increasing in proportion with the severity of the bronchial epithelium changes, hyperplasia mucosae, squamous metaplasia, and mild dysplasia exhibiting low mean scores. In contrast, high scores were increasingly observed in moderate dysplasia, severe dysplasia, and CIS with a slight decrease in corresponding squamous cell carcinoma (Fig. 1D). Mean scores and standard deviations of hTERT mRNA expression according to histology are given in Table 1.

Comparatively, a statistical difference between the levels of hTERT mRNA according to histologic grade of lesions could be shown (P < 0.0001) as well as between low-grade dysplasia (mild dysplasia) and high-grade dysplasia including moderate dysplasia (P < 0.0001). Levels of protein and mRNA expression were highly correlated in preinvasive bronchial lesion, as well as in normal/hyperplastic mucosae and invasive carcinoma (P < 0.0001).

Immunohistochemical Analysis of P53 Accumulation. P53 nuclear immunostaining was considered as positive when >20% of the cells were positive, arranged in suprabasal clusters in bronchial epithelium. Normal/hyperplastic mucosae was always negative (0 of 20), whereas increased percentage of P53 reactive cells were seen from squamous metaplasia to CIS to invasive carcinoma (Fig. 1E; Table 2). P53 accumulation was significantly more frequent in high-grade preinvasive lesions than in low-grade lesions (P < 0.0001). A direct correlation was found between hTERT and p53 expression (P = 0.01) at the level of each histologic type of lesion.

Loss of P16^INK4 Expression. P16^INK4 immunostaining was considered as specific when nuclei of normal or dysplastic epithelial cells were labeled as well as stromal endothelial and fibroblasts considered as positive internal controls. Loss of P16 expression started at the level of normal or hyperplastic bronchial epithelium (21%), to reach a plateau around 50% from squamous metaplasia to CIS, and increased to 63% in invasive carcinoma (Fig. 1G; Table 2). Four lesions (two squamous metaplasia, one moderate dysplasia, and one severe dysplasia) were lost on serial sections and six lesions (two normal or hyperplastic bronchial mucosae, one squamous metaplasia, two moderate dysplasia, and one CIS) were nonevaluable for P16^INK4 analysis in absence of positive internal controls. No correlation was observed between hTERT expression and P16^INK4 loss of expression when all histologic grades were considered together, or when low- and high-grade dysplasia were compared together.

Cyclin D1 Overexpression. Cyclin D1 overexpression was considered as positive when >5% of the cell nuclei were labeled. This was observed in 9% of normal/hyperplastic epithelia and progressively increased form squamous metaplasia to CIS to invasive carcinoma (Fig. 1H; Table 2). Two moderate dysplasia and three severe dysplasia were lost on serial sections. A statistical difference concerning cyclin D1 overexpression according to histologic grade was observed (P = 0.002). No correlation was found between hTERT and cyclin D1 expression whatever the histologic group considered.

Bax and Bcl2 Expressions. In normal bronchial epithelia, Bax staining was cytoplasmic and granular in cytoplasm of epithelial cells, whereas Bcl2 staining was restricted to the basal bronchial cell cytoplasm. Bax-to-Bcl2 ratio (BBR) was calculated as previously described by Jeanmart et al. (10). Briefly, BBR ratio was calculated compared with the normal epithelium where the score of Bax was consistently higher than that of Bcl2 (BBR > 1). In contrast, BBR < 1 was considered when Bax staining was lower than Bcl2 staining, corresponding to an inversion of the normal BBR. Whereas normal epithelium, metaplasia, and mild dysplasia showed an inversion of BBR, an increase of frequency of BBR < 1 was observed from moderate dysplasia to CIS to invasive carcinoma (Fig. 1F; Table 2). A statistical difference was seen between the frequency of BBR inversion according to histology (P = 0.002). A correlation was observed between hTERT expression and inversion of BBR (P = 0.0003) at the level of each histologic lesions.

Mean Ki67 Expression. Ki67 staining was considered as positive when nuclei of normal basal and dysplastic epithelial cells were labeled, stromal cells representing negative internal controls. Mean scores of Ki67 staining were 7% in normal/hyperplastic epithelium, 18% in squamous metaplasia, 26% in mild dysplasia, 37% in moderate dysplasia, 34% in severe dysplasia, 36% in CIS, and 39% in invasive carcinoma. This increase from normal/hyperplastic epithelia to low-grade to high-grade dysplasia to invasive carcinoma was statistically significant (P < 0.0001). No correlation was found between Ki67 mean scores and hTERT levels of expression.

Qualitative Telomere Assessment by Telomere Fluorescence In Situ Hybridization. Histologic analysis of telomeric signals showed a high number of intense fluorescent spots (∼10-15) in the nuclei of normal basal epithelial cells from specimen of all normal and hyperplastic mucosae (7 of 7); this staining was given a score of 2 for half of the cells, especially the basal cells, giving a signal consistent with a maintained telomere length. In the same way, high number of fluorescent spots was observed on the nuclei of stromal inflammatory cells, fibroblasts, endothelial cells, and seromucous glandular cells, which served as internal positive controls. Stromal lymphocyte nuclei frequently exhibited a higher staining frequency of 3 but also a higher intensity of staining (Fig. 2A). In contrast, only 16% (1 of 6) of squamous metaplasia, 28% (2 of 7) of mild dysplasia, 37% (3 of 8) of moderate dysplasia, 66% (4 of 6) of severe dysplasia, 55% (5 of 9) of CIS, and 55% (5 of 9) of invasive carcinoma exhibited telomeric signals scored as 2 or 3 (Fig. 2B-D). Although for scores 2 and 3, the number of spots detected per nuclei was rather constant, a strong reduction in number (<10) was frequently noted in lesions exhibiting score 1 signals. The lowest telomeric signals were observed in squamous metaplasia (Fig. 3; Table 1). A statistical difference of telomeric signals was observed between low-grade dysplasia (squamous metaplasia, mild dysplasia) and high-grade dysplasia (P = 0,03). A direct correlation was found between telomerase expression and telomere length from squamous metaplasia to invasive carcinoma (P = 0.02).

Comparison of Human Telomerase Reverse Transcriptase Expression with Telomere Length and P53/Rb Status in Preinvasive and Invasive Lesions. In normal/hyperplastic bronchial epithelium, all cases showed a P53-negative staining (normal immunophenotype) and low hTERT scores (<100), whereas Rb was inactivated in only 3 of 20 cases as a consequence of a p16 loss of expression. In squamous metaplasia, the telomeric signals were at the lowest level and hTERT scores were low in 76% of the cases, of which P53 was predominantly negative (84%), P16 positive, and cyclin D1 negative (61%). In mild dysplasia, hTERT scores were low in 88% of the cases, which exhibited a P53-negative staining and a Rb inactivation (due to a loss of P16 and/or a cyclin D1 overexpression) in 60% and 53% of the cases, respectively. In moderate dysplasia, hTERT scores were predominantly high (i.e., ≥100) in 75% of the cases, of which P53 was positive (mutant immunophenotype) and Rb inactivated in 66% of the cases. In severe dysplasia, telomeric length was at the highest level and hTERT scores were high in 80% of the cases, which exhibit a P53-positive staining and a Rb inactivation in 56% and 50% of these, respectively. In CIS, telomere length was stabilized and hTERT scores were high in 90% of the cases, which showed P53 accumulation and Rb inactivation in 51% and 58% of the cases, respectively. In invasive carcinoma, hTERT scores were high in 85% of the cases, which presented P53 accumulation and Rb inactivation in 39% and 63% of the cases, respectively. Overall, the switch for low to high rate and grade of these molecular alterations occurs apparently at the level of moderate dysplasia.

Using a monoclonal specific antibody for the catalytic subunit of telomerase hTERT, we have examined here the expression of telomerase in preinvasive bronchial lesions and their invasive and normal counterparts. The direct comparison of hTERT immunoreactivity between normal, dysplastic, and tumoral bronchial epithelium showed that dysplasia of progressive severity displays parallel increase of hTERT expression. In our study, telomerase expression was not restricted to invasive lesions because a weak staining was observed in bronchial basal cell consistent with the low levels of hTERT mRNA levels and telomerase activity detected in normal and hyperplastic epithelium, in contrast with activated lymphocytes of the bronchial mucosae, which exhibited a strong immunohistochemical staining and served as positive internal controls. Determining hTERT expression by immunohistochemistry on routine formalin-fixed specimens was facilitated by the use of a new commercially available monoclonal antibody 44F12, whose reliability and specificity is already shown in lung tumors providing Western blot and immunohistochemical results in good concordance with PCR-based telomere repeat amplification protocol assay and in situ hybridization with an hTERT riboprobe (36). The advantage of immunohistochemistry is the ability to visualize at the cellular level and topographically telomerase expression, activated lymphocytes serving as a convenient internal control to detect the possible heterogeneity in tissue immunoreactivity related to differences in fixation procedures. In contrast, telomere repeat amplification protocol assay, which should require microdissection from fresh samples for telomerase activity assessment at the level of dysplastic clones, is susceptible for contamination of tumor cells by the same activated lymphocytes in tissue, and in situ hybridization still remains a laborious and time-consuming method.

In bronchial carcinogenesis, telomerase expression has been first studied through hTERC assessment by in situ hybridization, leading to the demonstration of a mild up-regulation of hTERC levels as early as squamous metaplasia (25). Moreover, a strong increase of hTERC expression was focally observed in CIS in close vicinity with invasive process implicating telomerase complex in malignant transformation and tumor invasion (24). Expression of hTERT subunit, which represents a limiting factor for telomerase activity, was also evaluated through telomerase activity measure and hTERT mRNA detection in preneoplastic lesions in short series of preinvasive bronchial lesions, including squamous metaplasia and dysplasia, levels increasing along with the severity of their grade (25–27).

Telomerase represents a critical marker of cellular immortality in malignancies by maintaining the telomere length in tumor cells. Telomerase is expressed early during the cervical, oral, breast, and colorectal carcinogenic process, predominating in high-grade invasive lesions (21, 24, 41). For instance, telomerase activity has been detected in nearly 70% of breast CIS (24), and elevated hTERT mRNA levels were detected by in situ hybridization in 40% of cervical intraneoplastic grade III lesions, where they correlated with the presence of high-risk human papillomavirus DNA (22). hTERT overexpression was also observed in 77% of high-grade atypical alveolar hyperplasia representing the lung adenocarcinoma precursor lesion, 97% of nonmucinous bronchioloalveolar carcinoma (21), but in only 27% of low-grade atypical alveolar hyperplasia.

Interestingly, a nucleolar hTERT staining was predominantly observed in basal cells of bronchial epithelium but rarely in preneoplastic lesions, suggesting that hTERT nucleolar localization corresponds to a physiologic compartmentalization process in normal cells with moderate levels of telomerase expression, whereas high amounts of protein are accumulated diffusely in nucleoplasms in activated lymphocytes and dysplastic cell nuclei. We have previously shown that telomerase nucleolar localization correlated with a shorter survival in stage I non–small cell lung carcinoma, suggesting that in tumors, telomerase could be sequestrated away from its telomeric targets during DNA repair or DNA replication (36). In contrast, to preneoplastic lesions, tumor cells in invasive cancer suffer numerous chromosomal abnormalities and nucleolar relocalization of telomerase would enable them to control their genetic instability.

Timing of telomerase activation in preneoplastic bronchial lesions was still not completely defined with regard to with P53 and Rb/P16^INK4 pathway inactivation. In this respect, we observed a slight correlation between telomerase expression and P53 accumulation and BBR inversion (P = 0.01), and our data concerning P53, Ki67, and cyclin D1 expression in preinvasive bronchial lesions remain consistent with previous reports (10), their frequency of expression being significantly higher in high-grade dysplasia including moderate dysplasia than in low-grade preneoplastic lesions. The statistically significant correlation between hTERT expression and P53 accumulation is in agreement with the in vitro models suggesting tumor progression through the bypass of M₁ and M₂ mortality stages (42). Interestingly, resistance to apoptosis as reflected by inversion of BBR ratio also parallels telomerase expression (43). In contrast, we failed to show a correlation between P16^INK4 loss or cyclin D1 overexpression and telomerase reactivation at each level of preneoplastic lesions. Concerning P16^INK4, this lack of correlation is possibly due to a low sensitivity of our immunohistochemical technique as P16^INK4 loss is technically not easy to assess, the frequent lack of positive internal controls precluding interpretation in several cases. This bias was responsible for a limited number of valid cases for statistical analysis and probably for a higher frequency of P16^INK4 loss of expression than previously reported. Nevertheless, we still observed an increased frequency of P16^INK4 loss and inversion of BBR from low-grade to high-grade preneoplasia. As already mentioned for telomerase, moderate dysplasia behaved as regards to these markers as a high-grade lesion (5, 8, 10) and the time of switch from low-grade to high-grade molecular alterations.

Using an in situ hybridization technique, we have indirectly evaluated the length and/or the shortening of telomere ends. Not surprisingly, telomerase reactivation and telomere shortening were correlated only in preinvasive and invasive lesions, telomere attrition representing an earlier abnormality in bronchial carcinogenesis preceding hTERT expression. Telomere dysfunction promotes chromosomal instability and favors cytogenetic abnormalities. Microsatellite instability and loss of heterozygosity occurring as early as low-grade colorectal and breast dysplastic lesions precedes telomerase reactivation predominantly detected in the corresponding high-grade dysplasia (44, 45). Accordingly, we found a dramatic reduction in telomeric signals as early as squamous metaplasia, which seems as a stage of crisis, whereas 3p and 9p allelic loss are readily detectable because they are present already in normal and hyperplastic bronchial epithelium of heavy smokers, respectively (1). The early occurrence of loss of heterozygosity and methylation (46) at loci of tumor suppressor genes might increase the cell cycle rate and cell turnover, which in turn is the starter of telomerase activation. Indeed, whereas telomerase reactivation occurs in low-grade dysplasia, it becomes predominant in high-grade preneoplasia where the progressive addition of telomeric sequences into chromosome ends results in an increasing detection of telomeric signals by FISH. Similar findings were observed in prostatic and pancreatic intraepithelial neoplasia, which exhibit 4-fold shorter telomeres than normal basal epithelial cells, whereas stromal cell telomeres are longer (37, 38). As shown in RNA(Terc)-null, P53-mutant mice by Rudolph et al. (47), a progressive telomere shortening leads to cancer initiation by generating anaphase bridges, but telomerase reactivation is required secondarily for tumor progression. Therefore, telomerase participates in telomere structure stability by capping telomeres (48), stabilizing in cancer cells chromosome rearrangements and karyotypic abnormalities acquired during crisis and promoting proliferation and resistance to apoptosis (49). Although the predictive value of telomere shortening for occurrence of lung cancer in high-risk patients with preneoplastic lesions remains to be assessed using logistic regression analysis of larger series, we postulate that squamous metaplasia, exhibiting the lowest telomere length may represent a M₁ stage preceding P53/Rb inactivation and telomerase reactivation, whereas these alterations raise at the level of moderate dysplasia to culminate in severe dysplasia/CIS, thus featuring an immortal post-M₂ stage.

As telomere maintenance is essential to the proliferation of tumor cells, several approaches have being developed to target either telomerase or telomeric complex. More recently, the use of specific ligands leading to G quadruplex telomeric structure stabilization and, therefore, limitation of telomerase accessibility to its target seems as a promising area of development (50). Such promising drugs might become potential chemopreventive agents in high-risk patients, such as heavy smokers presenting preinvasive bronchial lesions with short telomeres.