Expression of Telomere-Associated Genes as Prognostic Markers for Overall Survival in Patients with Non–Small Cell Lung Cancer

Telomeres are specialized nucleic acid-protein complexes that cap the ends of eukaryotic chromosomes and are composed of TTAGGG repetitive sequences bound by a collection of proteins (1). TRF1, TRF2, POT1, and RAP1 are telomere-associated proteins that are part of the telomere structure and have essential roles in controlling telomere length. TRF1 was identified as a suppressor of telomere elongation (2, 3). Long-term overexpression of TRF1 in a telomerase-positive tumor cell line resulted in gradual and progressive telomere shortening, whereas telomere elongation was induced by expression of a dominant-negative form of TRF1 (2). Like TRF1, TRF2 binds specifically to double-stranded TTAGGG repeats at the telomere (3). TRF2 is also a negative regulator of telomere length. Overexpression of TRF2 results in progressive shortening to telomere length, similar to TRF1 (4). More importantly, TRF2 protects human telomeres from end-to-end fusions; thus, TRF2 plays a key role in maintaining telomere integrity (5, 6). POT1 binds to the 3′ single-stranded TTAGGG overhang at the end of the chromosome, in agreement with a protective role of chromosome end stability (7, 8). POT1 controls telomerase-dependent telomere elongation by first recruiting telomerase to the single-stranded 3′ telomeric overhang, allowing extension of the G strand. Subsequently, POT1 limits further elongation of the telomere by telomerase (9, 10). Human RAP1 is recruited to the telomere by TRF2; it does not bind to DNA directly (11). RAP1 negatively regulates telomere length in vivo (11–13). Depletion of endogenous human RAP1 by small interference RNA leads to longer telomeres; in addition, overexpression of dominant-negative forms of human RAP1 extends telomeres (13).

Telomeres play key roles in protecting linear chromosomes from end-to-end fusion, recombination, and degradation; thus, telomeres are critical for maintaining genome integrity (14, 15). Changes in telomere-associated proteins can lead to telomere dysfunction, which in turn could cause genomic instability and cancer development. Without telomere protection, fusion of unprotected chromosome ends can generate dicentric chromosomes or chromosome circulation. Expression of a dominant-negative form of TRF2 disrupted telomere function, resulting in chromosomal end-to-end fusions and anaphase bridges, leading to induction of an ATM- or p53-dependent apoptotic response (5, 16). Similarly, POT1 knockdown cells displayed chromatin bridges between interphase cells and improper chromosomal segregation (17). It was shown that RAP1 loss genotype in yeast causes telomere fusions, and that this telomere protection role of RAP1 could be evolutionarily conserved (18). In addition to its primary role as a negative regulator of telomere length, TRF1 was indicated to be important in chromosome end protection, as end-to-end fusions at telomeres were found accumulated in TRF1-deficient mouse embryonic stem cells (19).

Given the importance of these telomere-associated proteins in controlling telomere length, telomere stability, and genomic integrity, we were interested in investigating the association between the expression levels of these telomeric genes in normal and tumor lung tissues and their association with lung cancer prognosis. Gene expression of TRF1, TRF2, POT1, and RAP1 were analyzed by reverse transcription-PCR and quantified by real-time PCR. Three housekeeping genes [glyceraldehyde-3-phosphate dehydrogenase (GAPDH), 18S rRNA, and β-actin] were used to normalize the expression of each target gene. It has been shown that to measure expression level accurately, normalization by multiple housekeeping genes instead of one is required (20). The geometric mean of three housekeeping genes was validated as an accurate normalization factor by analyzing publicly available microarray data (20). In this study, we used the geometric mean expression of GAPDH, 18S rRNA, and β-actin, three housekeeping genes, as internal control for normalization of gene expression.

We analyzed samples from 148 non–small cell lung cancer patients who were enrolled in our ongoing lung cancer case-control study and underwent surgical resection at the M.D. Anderson Cancer Center from 1993 to 1997. These samples were fresh lung tumor tissues, and their paired adjacent normal tissues were collected at thoracotomy. The specimens were stored immediately after excision in a liquid nitrogen freezer (−130°C) until subjected to total RNA extraction. The current study did not restrict patient recruitment on the basis of age, gender, ethnicity, or tumor stage. As of June 2005, 99 (66.9%) of these patients were deceased; 93.9% of our study subjects were Caucasians. No statistically significant differences in sex, ethnicity, smoking status, age, and pack-years were observed when comparing alive patients to deceased patients, although there were more never smokers in alive patients (12.2%) than in deceased patients (6.0%; Table 1).

RNA extraction. Total RNA samples were isolated from surgically removed tumor or normal tissues from the lung using an E.Z.N.A. total RNA kit (Omega Bio-tek, Doraville, GA). The kit used the reversible binding properties of HiBind matrix, a new silica-based material. Briefly, tissues were first lysed under denaturing conditions that inactivate RNases. Samples were then applied to the HiBind spin columns to which total RNA was bound while debris and contaminants were washed away. Finally, high-quality RNA was eluted in DEPC-treated sterile water. The quality of the RNA samples was determined by electrophoresis through agarose gels and stained with ethidium bromide. The 18S and 28S RNA bands were visualized under UV light.

cDNA synthesis. Reverse transcription of total RNA was done with a Taqman reverse transcription reagents kit (Applied Biosystems, Branchburg, NJ) in a final volume of 20 μL containing 1× RT buffer, 5 mmol/L MgCl₂, 250 μmol/L each deoxynucleotide triphosphate, 20 units of RNase inhibitor, 50 units of multiscribe reverse transcriptase, 2.5 μmol/L random hexamers, and 0.5 μg total RNA. The samples were incubated at room temperature for 10 minutes and 42°C for 30 minutes; reverse transcriptase was inactivated by heating the mixture at 99°C for 5 minutes.

Primers and probes. Primers and probes for TRF1, TRF2, POT1, RAP1, GAPDH, 18S rRNA, and β-actin genes were designed using the Primer Express software (version 2.0, Applied Biosystems, Branchburg, NJ). We conducted searches in National Center for Biotechnology Information databases to confirm the total gene specificity of the nucleotide sequences chosen for the primers and probes as well as the absence of single nucleotide polymorphisms. In particular, the primer pairs were selected to be unique when compared with the sequences of the closely related family member genes. To avoid amplification of contaminating genomic DNA, one of the two primers or the probe was placed at the junction between two exons. Agarose gel electrophoresis allowed us to verify the specificity of PCR amplicons. The sequences of the primers and probes of these five genes are as followed: TRF1, forward primer 5′- CCACATGATGGAGAAAATTAAGAGTTAT-3′, reverse primer 5′-TGCCGCTGCCTTCATTAGA-3′, probe 5′-FAM-TTATGTGCTAAGTGAAAAATCATCAACCT-TAMRA-3′; TRF2, forward primer 5′- ACCAGGGCCTGTGGAAAAG-3′, reverse primer 5′- GGTGGTTGGAGGATTCCGTA-3′, probe 5′- FAM-CACCCAGAGAACCCGCAAGGCAG-TAMRA-3′; POT1, forward primer 5′-TCAGATGTTATCTGTCAATCAGAACCT-3′, reverse primer 5′-TGTTGACATCTTTCTACCTCGTATAATGA-3′, probe 5′-FAM- ACGACAGCTTTCCAAGCTCTGGA-TAMRA-3′; RAP1, forward primer 5′-GCCACCCGGGAGTTTGA-3′, reverse primer 5′-GGGTGGATCATCATCACACATAGT-3′, probe 5′-FAM-AGGTTGTGGTGGATGAGAGCCCTCC-TAMRA-3′; GAPDH, forward primer 5′-AAGGCTGAGAACGGGAAGC-3′, reverse primer 5′-GAGGGATCTCGCTCCTGGA-3′, probe 5′-FAM-TGTCATCAATGGAAATCCCATCACCATC-TAMRA-3′; 18S rRNA, forward primer 5′-AGTCCCTGCCCTTTGTACACA-3′, reverse primer 5′-GATCCGAGGGCCTCACTAAAC-3′, probe 5′-FAM-CGCCCGTCGCTACTACCGATTGG-TAMRA-3′; β-actin, forward primer 5′-CGAGCGCGGCTACAGCTT-3′, reverse primer 5′-TCCTTAATGTCACGCACGATTT-3′, probe 5′-FAM-ACCACCACGGCCGAGCGG-TAMRA-3′.

PCR amplification. Quantitative values were obtained from the cycle number (C_t). At this specific cycle number, the increase in fluorescent signal associated with exponential growth of PCR products started to be detected by the laser detector of the ABI Prism 7900 Sequence Detection System (Applied Biosystems, Foster City, CA). The build in analysis software was used according to manufacturer's manuals. For each tested gene, a standard curve was plotted to quantify the amount of each transcript.

All PCR reactions were done using an ABI Prism 7900 Sequence Detection System (Applied Biosystems, Foster City, CA) with the Taqman 1000 Rxn Gold/Buffer A Pack (Applied Biosystems, Foster City, CA). Each PCR reaction was done in a final volume of 10 μL containing 1× buffer A, 3.4 mmol/L MgCl₂, 100 μmol/L each deoxynucleotide triphosphate, 0.2 μmol/L each primer, 0.1 μmol/L probe, 0.02 unit of AmpliTaq Gold DNA polymerase, and 1 μL of each synthesized cDNA sample. The thermal cycling conditions comprised an initial denaturing step at 95°C for 10 minutes and 40 cycles at 95°C for 15 seconds and 60°C for 1 minute.

The precise amount of total RNA added to each reaction, RNA quality, and reverse transcription efficiency were difficult to assess. We, therefore, also quantified transcripts of GAPDH, 18S rRNA, and β-actin as endogenous RNA controls and normalized each sample based on its geometric mean expression of all three housekeeping RNAs. Results were expressed as N-fold differences in target gene expression relative to the geometric mean of three internal control genes. The PCR reaction for each gene was duplicated for each sample, and the mean was used in the analysis.

The smoking status of our studied patients was obtained from personal interview and defined as: ever smoker, ≥100 cigarettes in patient's lifetime; former smoker, smoking cessation for ≥1 year before diagnosis. Pack-years were defined as the number of cigarettes per day divided by 20 and then multiplied by the number of years smoked. Survival data were obtained from patient's chart review. The χ² test and Fisher's exact test were used to test the association between the distribution of characteristics and survival status. Student's t test and Wilcoxon-Mann-Whitney test were used for continuous variables. The difference of protein expression in normal and tumor lung tissues was assessed by Sign rank test. Kaplan-Meier plots and log-rank test were applied by using the time of event-free survival, which was calculated from the date of lung cancer diagnosis to the date of end point event or the date of last patient follow-up. The end point in this study was overall mortality. The Cox proportional hazards model was applied to multivariate analysis to assess the effect of expression of individual protein on the risk of the end point event. Gene expression levels in tumor tissues were dichotomized by CART software for the hazard ratio (HR) analysis. The cutoff point for expression of RAP1/geometric mean expression of GAPDH, 18S rRNA, and β-actin was 0.0153. Patients' age, gender, ethnicity, smoking status, tumor grade, and clinical stage were added into the model to control for confounding factors. The reference HR of 1 was set for low expression. The Kruskal-Wallis test was used to assess the correlation of expression of individual protein and tumor grade. All tests were two sided with a significant level of 0.05. STATA 8.0 software (STATA Corp., College Station, TX) was used to perform statistical analysis.

Clinical characteristic of study population. A total of 148 confirmed lung cancer patients diagnosed from 1993 to 1997 were included in this analysis. Clinical characteristics of these patients were summarized in Table 2. According to the tumor-node-metastasis staging system, most of the patients were at stage IA (20.27%), stage IB (27.7%), and stage IIIA (18.92%) when diagnosed. Tumor grades are defined as grade 1, well-differentiated (low grade) tumor; grade 2, moderately differentiated (intermediate grade) tumor; and grade 3, poorly differentiated (high grade) tumor. Most of the patients were at grade 2 (38.51%) and grade 3 (39.19%). The majority of the patients had adenocarcinoma (47.3%) or squamous cell carcinoma (35.81%).

TRF1 expression level is down-regulated in the tumor lung samples. Total RNA from normal and tumor lung tissues of the sample individuals were analyzed by reverse transcription-PCR and quantified by real-time PCR to compare the expression levels of TRF1, TRF2, POT1, and RAP1 in normal and tumor tissues. We observed significantly lower expression of the TRF1 gene in tumor tissues than in adjacent normal tissues (P < 0.0001; Table 3). However, no significant differences in expression between tumor and normal tissues were observed for TRF2 (P = 0.8111), POT1 (P = 0.1007), and RAP1 (P = 0.3525).

RAP1 expression in lung tumor tissue is correlated with overall survival of the lung cancer patients.RAP1 expression levels in tumor samples were significantly correlated with overall survival of the patients (Table 4, Fig. 1). High level of RAP1 expression in tumor tissues was associated with a 53% reduced risk of cancer death (HR, 0.47; 95% confidence interval, 0.24-0.91; Table 4). In Kaplan-Meier survival estimates, the median survival time for patients with low RAP1 expression was 15.34 months, whereas the median survival time for patients with high RAP1 expression was 51.21 months (P < 0.0009; Fig. 1). This protective effect of higher RAP1 expression was especially evident in men and in ever smokers (Table 4). In men, the adjusted HR was 0.45 (95% confidence interval, 0.22-0.996). Ever smokers with higher levels of RAP1 had a 50% reduced risk of death with borderline significance (HR, 0.50; 95% confidence interval, 0.24-1.02). These data suggest that RAP1 expression level could be a useful prognostic marker for lung cancer survival.

We also studied the association between TRF1, TRF2, and POT1 expression in tumor tissues and overall survival of these patients, but we did not find statistically significant associations (data not shown).

TRF2 expression in lung tumor tissues is correlated with tumor grade. We next analyzed the association between TRF1, TRF2, POT1, and RAP1 expression levels in the tumor tissue and tumor grade. We observed that TRF2 expression was significantly correlated with tumor grade (P = 0.0114; Table 5). Higher level of expression of TRF2 was correlated with lower tumor grade, suggesting that TRF2 has a protective role in lung cancer progression.

We studied the expression of four telomere-associated proteins (TRF1, TRF2, POT1, and RAP1 genes) in normal and tumor tissues of lung cancer patients and observed significant down-regulation of TRF1 in tumor samples and no significant difference in expression between tumor and normal tissues for TRF2, POT1, and RAP1 (Table 3). The down-regulation of TRF1 in tumor tissue was consistent when we used GAPDH alone, 18S rRNA alone, β-actin alone (data not shown) or the geometric mean of three genes as internal control. In previous published studies, controversial data were presented for these telomeric genes in cancer. Some studies suggested that TRF1 and TRF2 were down-regulated in tumor tissues (21–26), whereas others showed that TRF1 or TRF2 was up-regulated (27–30). Different tumor type and tumor stage may account for some of these differences. One study showed that tumor stage and telomere length might also influence POT1 expression in cancer (31). Their data indicated that in stage I/II gastric cancer, POT1 is mostly down-regulated; and in stage III/IV gastric cancer, POT1 is frequently up-regulated. In addition, POT1 expression decreased as telomere length shortened (31).

Normal adjacent tissues are widely used as baseline (or comparison) specimen class in gene expression profiling and telomerase activity study between tumor and normal tissues (32–34). However, some studies have raised the question that whether normal adjacent tissues can truly represent normal tissues in these type of studies due to a “field” effect (35, 36). They suggested that, in some cases, tissues adjacent to cancer, although appearing morphologically normal, may contain genetic changes associated with cancer (35, 36), which might even influence gene expression (35). Therefore, difference in baseline or normal tissue collection might also contribute to controversial reports on telomere-associated protein expression in cancer.

More importantly, we observed significant association between RAP1 expression and overall survival of the lung cancer patients (Table 4; Fig. 1). When we used GAPDH alone, 18S rRNA alone, and β-actin alone as internal control, we also observed significant association in all three cases (data not shown). This result is consistent with the telomere protection role of RAP1 in yeast (18), indicating that this protective role is evolutionarily conserved. We observed that in men and ever smokers, the influence of RAP1 expression is prominent (Table 4). It is well documented that females live longer than males; the different rate of telomere erosion in males and females may lead to differential requirements of telomeric protein levels. This gender disparity in telomeric protein requirement might explain the differences exhibited between males and females in our research. Looking at different risk factors, smoking produces a state of heightened oxidative stress. Increased smoking is associated with accelerated telomere shortening, which may sensitize the tumor to the protective role of RAP1.

We also found TRF2 expression in tumor tissues was inversely correlated with tumor grade: the higher the gene expression, the lower the tumor grade (Table 5). Tumor grade refers to how much the tumor cells resemble normal cells of the same tissue type. The cells of grade 1 tumors resemble normal cells and tend to grow and multiply slowly, whereas grade 3 tumors grow rapidly. Tumor grade is an indicator of the severity of the cancer. Our data suggested that TRF2 might have a protective role in lung cancer progression.

Studies have shown that shortened, dysfunctional telomere might activate a DNA damage response pathway involving ATM, p53, CHK2, and H2AX proteins and lead to cell cycle arrest and senescent phenotype (37). This phenomenon might be a tumor suppressor mechanism. Disruption of DNA damage response mechanism might result in accelerated telomere shortening and elevated frequencies of end-to-end chromosome fusions. Previously, we studied the association among telomere activity, p53 protein overexpression, and genetic instability in lung cancer using samples overlapping with this study (38). We found that p53 overexpression carried a 6.7-fold (95% confidence interval, 1.7-27.7) increased risk for positive telomerase activity, and it is common in individuals genetically susceptible to exposure (38). However, p53 overexpression and telomerase activity were not associated with telomere-associated protein expression in lung cancer tissues (data not shown). We also measured telomere length of tumor samples of these lung cancer patients using real-time PCR; however, no association between telomere length and overall survival was observed (data not shown). Neither was any association between telomere length and telomere-associated gene expression (data not shown). These data suggested the predictive role of RAP1 expression in lung cancer survival might be independent of telomerase activity and telomere length.

In this study, we used the geometric mean of the expression of three housekeeping gene as internal control to normalize target gene expression. This method can eliminate the variability of expression of a single gene control in response to various factors and present a more accurate measure of reverse transcription-PCR expression profiling (20). When we compared the data using the geometric mean of three genes as internal control and using individual gene as internal control, we found that the relative expression level between tumor and normal tissues varied. However, RAP1 expression in tumor consistently showed significant association with overall survival disregard which internal control was used. Because the expression of telomere maintenance genes from tumor tissues could be affected by tumor heterogeneity and lymphocyte infiltration, studies at the cellular level either by immunohistochemistry or by mRNA hybridization are warranted to confirm real-time PCR results. Nevertheless, our data strongly suggested that expression of RAP1 is a promising prognosis marker for lung cancer.