Polymorphisms in the Promoter Region of the Neutrophil Elastase Gene Are Associated with Lung Cancer Development¹ Free

It is not known whether inherited factors are influencing predisposition to rapid progression in lung cancer development. During the invasion of cancer cells, the proteases are involved in cell-extracellular matrix interaction. Collagens, proteoglycans, and elastins were known to be components of extracellular matrix including basement membrane and interstitial stroma (1, 2). NE³ is a member of the trypsin superfamily and is a powerful serine protease capable of degrading most protein components of the extracellular matrix, various proteins of the coagulation and complement cascades, or cell wall of bacteria (3). The NE protein is stored in granules of neutrophil along with other proteases such as cathepsin G, proteinase 3, or azurocidin, and it is released when the neutrophil is activated or destroyed (4). The crucial role of NE protein could be a double-edged sword in normal host defense or tissue turnover, and in abnormal self-destruction. NE protein has been studied in regard to the aggressiveness of the tumor. It was reported that NSCLC cell lines were secreting NE protein, and the local production of the NE was involved in the invasion of the tumor and associated with a poor prognosis (2). NE is known to be the only protein that is able to cleave insoluble elastin, and the expression level of the NE protein was positively associated with tumor progression, especially in the aorta that is an abundant source of insoluble elastin (5). In this report, we have investigated an etiological role of NE in lung cancer development under the design of a clinical epidemiology study and using the tools of molecular genetics.

The Institutional Review Board has approved the study, and all of the participants have signed appropriate informed consent forms. Included in this study were 348 primary lung cancer cases which were consecutively enrolled in an ongoing comprehensive lung cancer study at the Mayo Clinic (Table 1). A more detailed description of the main study was published previously (6), and a brief account is provided as follows. Since March of 1997, all of the patients who were newly diagnosed with primary lung cancer have been identified and evaluated for research studies. Patients were ascertained daily from our computerized pathology reporting system, which identifies ∼95% of all lung cancer cases seen at our institution. The remaining 5% of the patients were identified directly from the patient-care physicians. Information abstracted from the medical records of the patients includes pathology, clinical staging, treatment, history of previous diseases, lifestyle (tobacco, alcohol, and coffee use), education, occupation, a brief family history, demographics, and follow-up data.

Two hundred and ninety-nine control subjects, from two sources, were recruited from our institution during the same time period as the case enrollment (Table 1). The first group (control-1) was 108 Olmsted County residents who have no current or previously diagnosed lung cancer, or other malignancies as of the date of their clinic visit. The second group (control-2) was comprised of 191 cancer-free subjects who were heavy smokers and participated in an ongoing spiral computed tomography study. Any individual who was found having lung cancer was excluded from the control groups.

A peripheral blood sample, as the source for DNA, was obtained from each study subject. The PCR primers used in this study are shown in Table 2. Samples used for mutation screening and sequencing were amplified in 25-μl reaction volumes containing 50 ng of genomic DNA, 25 pmol of F (forward) and R (reverse) primers for each exon, 200 μm deoxynucleotide triphosphates (Perkin-Elmer, Foster City, CA), 0.2 μl of Taq polymerase (AmpliTaq Gold; Perkin-Elmer), and 2 mm of MgCl₂. The condition for PCR amplification was for 35 cycles; 94°C for 30 s, the annealing temperature optimized for each primer set (shown in Table 2) for 45 s, and 72°C for 45 s (final extension was 72°C for 10 min) after denaturation at 95°C for 9 min.

The PCR products were denatured at 95°C for 3 min and gradual reannealing from 95°C to 65°C over a period of 30 min before analysis. For the REP-b product, part of the PCR product was aliquoted and mixed 1:1 molar ratio with the PCR product from the DNA with known genotype before the denaturation-reannealing step was performed. PCR products were confirmed by 2% agarose gel electrophoresis and were examined for heteroduplex content by subjecting ∼50 ng of the amplified product to DHPLC (WAVE System; Transgenomic, San Jose, CA) under conditions of partial denaturation (7). The DNA separation column (DNASep Cartridge) was provided by Transgenomic. The column mobile phase consisted of a mixture of 0.1 m triethylamine acetate (buffer A) and buffer A with 25% acetonitrile (buffer B). The flow rate was 0.9 ml/min. To ensure the precision and reproducibility of the assay, each series of the samples were tested along with a set of known controls confirmed by sequencing analysis.

The PCR product that showed heteroduplex profiles by DHPLC analysis was directly sequenced. Briefly, the PCR product was treated with exonuclease and shrimp alkaline phosphatase based on the protocol provided by United States Biochemical. They were mixed with 3.2 pmol of each forward or reverse primer of each exon and sequenced at the Mayo Molecular Core Facilities. Sequencing reactions were performed in a DNA thermal cycler 9600 (Perkin-Elmer) with fluorescent terminations, and the sequencing products were analyzed on an ABI377 sequencer.

5′Flanking region of the NE was amplified with the Expand High Fidelity PCR System (Roche Diagnostics, Indianapolis, IN) from the DNA with known genotype. The primers used were: 5′-CGCAGTGAGTGCCCGACACA-3′ and 5′-GGTCTCTGCCCCTCCGTGCCC-3′ (8). The PCR product was gel-purified and ligated into SmaI site of pGL3 vector (Promega, Madison, WI). The clones were sequenced, and the genotype was confirmed for each SNP with the absence of PCR artifact.

Human NSCLC cell line A549 was obtained from American Type Culture Collection (Manassas, VA) and was maintained at 37°C in F-12K medium (American Type Culture Collection) supplemented with 10% fetal bovine serum. Transient transfections were performed in triplicate with FuGENE 6 (Roche Diagnostics) using 2 μg of the pGL3-NE promoter reporter plasmid and 0.1 μg of pRL-TK (Promega) as an internal control when the cells were 40–50% confluent. After a 48-h incubation, cell lysates were prepared, and luciferase activity was assayed with the Dual-Luciferase Reporter Assay System (Promega) following the instructions from the manufacturer. Luciferase activities were corrected for differences in transfection efficiency among each experiment as estimated by Renilla luciferase activity from pRL-TK. All of the experiments were performed three to four times for each plasmid.

The association between the binary outcome of genotypes and lung cancer was examined using standard two-way contingency-table methods. Differences in genotype frequencies were tested using χ² and Fisher’s exact statistics. Ps of 0.05 and 95% confidence intervals were calculated for assessing the significance of the association. ORs are used to estimate the relative risks of lung cancer of individuals carrying a high-risk allele or combination of allele types.

We analyzed 30 normal blood DNA samples by DHPLC to search for polymorphisms in the entire coding region and the core expression control region of the NE gene, which contains six repetitive tandem motif of a 53-bp sequence (REP53, the fourth motif has 52 bp). Only one SNP was identified in the coding region, which is in exon 5, codon 173 TCC to TCA, with heterozygosity of 13.3%. Three kinds of polymorphisms were identified in the six repetitive tandem motif of the promoter region: an extra 52-bp repetitive sequence between the fourth and fifth repetitive motifs, T to G SNP at −903 bp (named REP-a), and A to G at −741 bp (named REP-b) from transcription initiation site (Fig. 1).

Because of the difficulty to distinguish each genotype according to the shape of DHPLC elution profile, two primer pairs, REP-1 and -2, were designed to split two SNP sites into each fragment to be analyzed separately. REP-a, REP-b, and a repetitive number of REP53 were screened in control-1. For REP-a, TG and TT genotypes were observed in 15 (13.3%) and 93 subjects (86.1%), respectively, whereas no GG genotype was found (Fig. 2, A and B; Table 3). To rule out the possibility that TT and GG could not be distinguished, we mixed all of the samples that showed homoduplex profile with known TT product and screened again, and no sample showed as heteroduplex. For REP-b, AA, AG, and GG was 6 (5.5%), 45 (45.7%), and 57 subjects (53.7%), respectively (Table 3). AA and GG homozygous samples were distinguishable according to the difference of the elution shape and peak retardation time (Fig. 2, C and D). Heterozygous samples for REP53 repeat number were observed in only 6 of 143 (4.2%), and it was not included in additional analysis because of small sample size.

Control-2, heavy smokers without lung cancer, was also screened. For REP-a, TG and TT genotypes were observed in 25 (13.1%) and 166 subjects (86.9%), respectively, and no GG genotype was found. For REP-b, AA, AG, and G/G was 12 (6.4%), 76 (40.2%), and 101 subjects (53.4%), respectively. Among cases, −903TG and −903TT genotypes were observed in 22 (6.5%) and 318 cases (93.5%); −741AA, −741AG, and −741G/G were 10 (2.9%), 121 (34.8%), and 217 (63.4%) cases, respectively. There was no −903GG and very few −741AA in our entire study population. Therefore, TT and TG were contrasting allele types at REP-a, and GG and AA+AG were contrasting allele types at REP-b.

Allele types were compared between cases and controls singly and in combination. Measured by OR, individuals with TT type at REP-a or GG type at REP-b had a 2.3- and 1.4-times higher risk of developing lung cancer when compared with those with TG and AA at REP-a or AG at REP-b, respectively (Table 3). When we assess the combined effects of the high-risk allele at both −903 and −741 polymorphic sites (Table 4), four combinational types were formed: TT-GG, TT-AG, TT-AA or TG-AG, and TG-AA or TG-AG. When TT-GG type was compared with the group with TG-AA or TG-AG combinations, the OR was 24.8 between cases and control-1. This comparison could not be made using control group 2 because of the lack of individuals with TG-AA or TG-AG combination among cases who were under the age of 50 years (therefore, the OR reached infinity because of a zero in the denominator). The OR was 23.2 when comparing cases with all controls. Likewise, comparisons were made between TT-AG type and TG-AA or TG-AG; the ORs were 19 and 17.6 using control-1 or all controls, respectively. The third combinational group, TT-AA or TG-AG, also showed a significant OR of 12–13 for lung cancer over the fourth group TG-AA or TG-AG.

It has been reported that the 53-bp element in the promoter region contains a potential binding site for a basic helix-loop-helix protein and has an enhancer activity (9). Thus, we evaluated the role of the two SNPs in the regulation of NE transcription by transient transfection and luciferase reporter assay. We generated four constructs with different genotypes of the NE promoter: G and A, T and A, G and G, and T and G for REP-a and -b site, respectively. They were subcloned into luciferase reporter plasmid and transfected into A549 NSCLC cells, which was reported to produce NE protein (2) and measured the basal transcription activity (Fig. 3). There was a trend toward a higher activity of luciferase if the REP-a site being T was compared with G, and if the REP-b site being G was compared with A allele. The construct with −903T and −741G (T-G) showed 1.9-fold activity, which was significantly higher than −903G and −741A (G-A) indicating that the significant increase in transcription might be dependent on the combination of the two SNPs.

Our results from the molecular genetic epidemiology studies suggested that individuals with a combination of the two SNPs (the −903 T and −741G allele) in NE promoter had a higher risk of developing lung cancer than the other allele types. This observation was supported by the functional analysis of the transcriptional activity of the NE promoter constructs containing these different SNPs in vitro. This evidence indicates that genotype with higher risk correlates with a higher promoter activity and its SNPs. To our knowledge, this is the first report about lung cancer risk and molecular variants of the NE gene. It has been reported that SNPs within transcription control region of a gene can affect disease susceptibility. For example, SNPs in the 5′-flanking region of the tumor necrosis factor gene were reported to be correlated with inflammatory bowel disease (10), narcolepsy (11), or severe infection (12). SNP caused the helix-turn-helix transcription factor OCT-1 to bind to a novel region and altered the tumor necrosis factor expression in human monocytes, and this OCT-1-binding genotype is associated with 4-fold increased susceptibility to severe malaria (12). The matrix metalloproteinases family, MMP-1, has G to GG polymorphism in the 5′-flanking region, which creates the ETS binding site, thus showed a higher transcription activity. It was shown that ETS-binding genotype was observed frequently in tumor cell lines (13). The NE expression is mainly regulated in transcription level (14, 15). In the NE promoter, T genotype of REP-a corresponded to the third nucleotide of inverted GATA, whereas it was abolished with G genotype. The REP-b located four bp 3′ side of inverted GATA, thus it is within the matrix sequence, and it could also be able to affect the stability of protein binding. Because REP53 has an E-box consensus motif and an enhancer activity, REP-a and -b could also alter the binding affinity through changing the second structure of the gene. Additional analysis is required to identify which transcription factor(s) binding is affected by the presence of REP-a and/or REP-b.

α-1AT, a glycoprotein known to inactivate a wide variety of proteinases including NE, has been widely reported regarding the relationship between its genotype and disease predisposition. COPD, including emphysema and chronic bronchitis, develops in α-1AT deficiency individuals (16). The low circulating level of α-1AT is unable to inhibit proteinases and leads to panlobular emphysema. Individuals with α-1AT deficiency are also predisposed to juvenile hepatitis, cirrhosis, and hepatocellular carcinoma (16). We have reported previously the possibility that imbalance between α-1AT and NE may be a predisposed condition for lung cancer development, i.e., individuals who carried the deficient allele of α-1AT had a significantly higher risk for developing squamous cell or bronchoalveolar carcinoma of the lung (6). It was also reported that strong α-1AT expression in lung adenocarcinoma correlates with poor prognosis (17); however, it is not known whether the α-1AT protein was functionally normal and whether there is a strong NE expression among the same patients.

Our work demonstrated a unique paradigm to show the role of NE in lung cancer etiology. On the basis of the findings of α-1AT on lung cancer, we logically hypothesized a potential role of NE in lung cancer development. We then identified the SNPs on the NE gene promoter, demonstrating an epidemiological association, and performed functional analysis showing higher activity of the NE promoter construct T-G for REP-a and REP-b site compared with the G-A. This result supports the epidemiological findings that the combination of TT genotype for REP-a and GG for REP-b showed markedly higher OR for lung cancer. Because the database for SNPs has been filled up, it will be possible to apply this methodology for various candidate genes. It should be noted that measuring the relative abundance of NE levels in lung cancer and adjacent normal tissue would demonstrate a functional consequence of the different SNPs in the regulatory regions of the NE gene. However, none of NE antibodies commercially available to date is good enough for immunostaining and quantitating the NE level in lung tissues. In addition, this effort could only be applied to study subjects who received surgical resection and had adequately procured tissue specimens.

Another strength of our study is the use of two contrasting control groups to corroborate new marker-disease association in the same study. This is a step ahead in precluding inconsistent results from different studies using different controls, a typical controversial phenomenon when establishing marker-disease associations in epidemiological studies. We at the same time acknowledge the sample size limitation for certain comparisons when examining combined risk markers, which is commonly encountered in patient-oriented epidemiological studies.

As a future direction, it is of great interest whether an association between polymorphism in NE and COPD, especially emphysema, is there or not. Because a deficiency of α-1AT, an inactivation factor of NE protein, is strongly correlated to emphysema, it is expected that polymorphisms in the control region of the NE could also affect the risk for emphysema. To prevent an onset and a progression of COPD, it is very important to control risk factors in a high-risk group, and we would like to evaluate REP-a and REP-b for elucidating the potential high-risk group of COPD. Consequently, it is plausible that the imbalance between α-1AT and NE could be a predisposed condition for lung cancer development.

In conclusion, using a combined approach of clinical epidemiology and molecular genetics, we identified a potentially important genetic marker for risk of lung cancer and for assisting clinical decision in treatment modalities. The α-1AT/NE system could be a useful genetic marker for screening the high-risk group. Additional analysis for the correlation between NE genotypes and clinical feature of the tumors, especially disease stage and histology, is warranted.

We thank the following physicians at Mayo Clinic Rochester for supporting this study: R. S. Marks, E. S. Edell, M. S. Alen, D. L. Miller, G. A. Croghan, J. L. Myers, D. E. Midthun, H. D. Tazelaar, S. N. Thibodeau, A. M. Patel, V. F. Trastek, and A. Jatoi. We also thank Susan Ernst for her secretarial support for the manuscript preparation.