Aberrations in the Fragile Histidine Triad (FHIT) Gene in Idiopathic Pulmonary Fibrosis¹

The development of primary lung cancer is the result of a stepwise accumulation of multiple acquired genetic events including mutations of TSGs³ (1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11). Several TSGs or candidates, including p53 (2), p16 (12), and FHIT (13, 14, 15, 16), have been reported to be mutated, deleted, or hypermethylated in lung cancers and preneoplastic lesions including metaplasia. To elucidate the genetic events that accumulate in the early stage of the development of lung cancer, genetic analyses of various genes in preneoplastic lesions have been conducted. These studies showed that 3p allelic loss may be the earliest change, followed by 9p allelic loss, 17p allelic loss (including p53 mutation), 5q allelic loss, and ras mutations, which suggests that one or more 3p TSGs may act as gatekeepers of lung cancer pathogenesis (8). The human FHIT (fragile histidine triad) gene was identified at 3p14.2, which included the FRA3B fragile site, and was found to be abnormal in an epithelial cancer cell line (13, 17). Accumulating evidence suggests that the FHIT gene is a TSG (18, 19, 20, 21). Aberration of the FHIT gene has been reported in various kinds of cancers including lung (13, 14, 15, 16), breast (22), head and neck (23), esophageal (24, 25), gastric (26), pancreatic (27), renal (28), cervical (29), and colon cancers (30, 31). The FHIT gene was also reported to be altered in preneoplastic lesions of lung cancer (14, 15).

Lung cancers are divided into two groups, the central type and the peripheral type, according to the location from which the tumor is derived. Most central-type tumors are derived from the trachea and central bronchi and consist of squamous cell carcinomas and small cell carcinomas. Peripheral-type tumors consist of all histological types of tumors including adenocarcinoma and large-cell carcinoma. The frequency of metaplasia or dysplasia in the central lesion as observed by bronchoscopy has been included in most genetic studies of preneoplastic lesions of the lung, and the results of these studies showed the genetic events involved in the carcinogenesis of a central lesion in the lung.

The incidence of lung cancer among patients with IPF is much higher than that among the general population (32, 33, 34, 35). Most lung cancers in IPF patients develop in the peripheral field of the lung. When we examined 49 lung cancer patients with IPF in our hospital in a preliminary study, 41 (84%) of the 49 patients showed the peripheral arising type of lung cancer. Therefore, we suspected that IPF includes precancerous lesions that develop into peripheral type tumors in which TSGs are damaged. The chronic inflammation in IPF could lead to repeated cycles of damage and repair of the epithelial cells in the bronchi and alveoli, and IPF patients have many lesions of metaplasia. The lesions of metaplasia in IPF are a good target for the research on human carcinogenesis in the peripheral lung.

To determine the mechanism of carcinogenesis of IPF, we performed genetic analysis of chromosome 3p, alteration of which is thought to be one of the early events in the development of central-type lung cancer, in the metaplasias and bronchiolar epithelia in patients with IPF. We analyzed the FHIT locus, which is thought to be a crucial gene in the pathogenesis of lung carcinomas at chromosome 3p. We also performed FISH and immunohistochemical analysis of FHIT expression to confirm the results of the LOH analysis.

This study included 146 tissues samples of metaplasia surrounding a honeycomb formation and bronchiolar epithelia in IPF lesions obtained from eight patients with IPF with primary lung cancer (cases 1–8) and three patients with IPF without cancer (cases 9–11). All of the patients were diagnosed pathologically as having usual interstitial pneumonia of Liebow (UIP). The patients were seen at the Fourth Department of Internal Medicine at Nippon Medical School Main Hospital in Tokyo, Japan. From each lung cancer patient, we obtained tissue samples from the primary lung cancer site, normal lung, and IPF lesions, and from each IPF patient without cancer, we obtained tissues from normal lung and IPF lesions at the time of autopsy or surgical resection. The histological type of the lung cancers included three squamous cell carcinomas, four adenocarcinomas and one small cell carcinoma. The patients consisted of nine males and two females, ranging in age from 47 to 82 years.

Serial 8-μm-thick sections were prepared from formalin-fixed, paraffin-embedded blocks. We microdissected a total of 146 lesions (119 metaplasias, 27 bronchiolar epithelia) that surrounded honeycombing and normal areas (histologically normal-appearing respiratory epithelial cell areas and vessel wall cells). The slides of normal areas did not include cancer lesions and were distant from cancer lesions. The microdissected cells were digested in 15 μl of buffer containing proteinase K as described previously (36), and then, DNA extraction from the microdissected tissues was performed.

A PCR-based approach of microsatellite polymorphism analysis using microsatellite markers within and flanking the gene locus at 3p14.2 was used. The microsatellite sequences were obtained from the gdb-Genome Database Site⁴ (hosted by Johns Hopkins University School of Medicine). The loci examined were D3S1300, D3S1234, D3S1312, and D3S1295. D3S1300 is located within intron 5 of the FHIT gene locus, close to exon 5, which is the first coding exon of the gene. This location coincides with the region of common loss in a variety of tumor cell lines. D3S1234 is located in intron 7; D3S1295 is located telomeric to the gene and is positioned at 3p14.3. Centromeric at 3p14.2 is D3S1312. A PCR reaction volume of 25 μl was used for the PCRs of all of the microsatellites. The PCR reaction mixture contained 25 pmol of each primer, 2.5 mm dNTPs, 1.25 units of Ex Taq DNA polymerase, and 1× Ex Taq Buffer (Takara, Tokyo, Japan). The 25-μl reaction mixture was placed in a DNA thermal cycler (Perkin-Elmer Cetus, Norwalk, CT) when the heating block reached 94°C. The PCR conditions for each set of primers were optimized. Standard denaturing and extension temperatures of 94°C and 72°C, respectively, were used. The annealing temperatures were typically between 52°C and 60°C, and the number of cycles ranged between 35 and 45. The PCR product from each reaction was resolved on a 5% metaphor agarose gel (FMC Bioproducts, Rockland, ME) using ethidium bromide staining and UV illumination. An informative sample was considered to have allelic loss if the intensity of one or more bands was reduced by 50% or more in the lesion; DNA was compared with that in the normal DNA sample of the same patient. Allelic loss was confirmed at least twice in separate PCRs.

To perform FISH of the FHIT gene in IPF lesions, we had to preserve the shape of the IPF lesions. We examined and diagnosed IPF lesions including metaplasias and bronchiolar epithelia around honeycomb formations in the paraffin-embedded tissue sections with H&E staining. Then, we cut 5-μm-thick serial sections, mounted each section on a positively charged, deparaffinized slide in xylene, and washed them in 100% ethanol. After hydration in graded alcohols, the slides were incubated in 1 m NaSCN at 80°C for 15 min; washed in distilled water; incubated in 4 mg/ml pepsin (Sigma Chemical Co., St. Lois, MO) at 37°C for 15 to 20 min; immersed in 3:1 methanol/acetic acid solution at room temperature for 15 min; and dried. After dehydration in graded alcohols, they were immersed in 70% formamide/2× SSC solution at 74°C for 5 min, followed by dehydration in graded alcohols. They were then dried for hybridization. The BAC, 1-O-12 probe (kindly given to us by Dr. Kay Huebner, Thomas Jefferson University Kimmel Cancer Center, Philadelphia, PA), spanning exon 5 of the FHIT gene, and the chromosome 3 centromere probe (kindly given to us by Dr. Joe W. Gray, University of California—San Francisco Comprehensive Cancer Center, San Francisco, CA) were labeled by nick translation with digoxigenin-11-dUTP (Roche, Mannheim, Germany) and fluorolink Cy3-dUTP (Amersham Pharmacia Biotech, Piscataway, NJ), respectively. Two hundred ng of the 1-O-12 probe was mixed with 10 μg of Cot-1; this was treated by ethanol precipitation, dissolved to hybridization mixture, denatured at 74°C for 5 min, reannealed at 37°C for 1 h, and mixed with 40 ng of the chromosome 3 centromere probe that had been denatured in the hybridization mixture at 74°C for 5 min. This mixture was pipetted on a 10-mm × 7-mm area of the IPF lesion that was surrounded by rubber cement on the slide that had been placed on a slide warmer. The slides were agitated in a humidified case on a rocking shaker in an incubator at 37°C, over 2 nights. Posthybridization washes were carried out three times for 10 min each in 50% formamide/2× SSC at 45°C; two times at 7 min 30 s each in PN buffer (37); and 5 min in 2× SSC/0.1% Triton X at room temperature. They were then rinsed in 2× SSC at room temperature. Detection was accomplished using 2 μg/ml of FITC-anti-digoxigenin (Roche, Mannheim, Germany). After 10-μl 4′,6-diamidino-2-phenylindole dihydrochloride n-hydrate (DAPI) counterstaining, ∼100 nuclei were counted in each sample. The same procedure was carried out on a control slide with normal peripheral blood.

Immunohistochemical staining was performed on paraffin sections with rabbit polyclonal anti-GST-Fhit antiserum (Zymed Laboratories Inc., South San Francisco, CA) as described previously (15). Briefly, formalin-fixed paraffin-embedded tumors and IPF lesions were serially cut and mounted on poly-L lysine (Sigma Chemical Co.)-coated slides, deparaffinized in xylene, and rehydrated in graded alcohols. Endogenous peroxidase activity was blocked by treatment with 0.3% hydrogen peroxidase in methanol for 30 min. Antigen enhancement was performed by steaming the sections in 5 mmol/liter sodium citrate buffer in distilled water (pH 6.0) at 100°C for 2 min. The slides were cooled under tap water, washed three times in 0.05 m PBS-0.1% Triton and then incubated with blocking serum (normal goat serum at a 1:10 dilution in PBS) for 10 min at room temperature. The slides were incubated overnight with primary rabbit polyclonal anti-GST-Fhit serum at a 1:200 dilution in PBS. After the incubation, the slides were washed three times in PBS-0.1% Triton and then incubated with secondary antibody (biotinylated goat antirabbit diluted at 1:500 in PBS; DAKO, Glostrup, Denmark) for 30 min at room temperature. A final incubation with an avidin-biotin peroxidase complex (DAKO) for 30 min at room temperature was performed. After subsequent PBS washings, the tissue sections were reacted with a solution of 0.2 ng/ml 3,3′-diaminobenzidine, 0.005% H₂O₂ and 50 mmol/liter Tris-HCl. The slides were counterstained with Mayer’s hematoxylin.

The incidence of LOH of the FHIT gene among the samples from the IPF patients with lung cancer and that among the samples from the IPF patients without lung cancer were compared using Fisher’s exact test and the χ² test. To determine whether the variability between persons was greater than that between diagnoses, the response variables (allelic deletions) were analyzed by fitting an ANOVA model (SAS system) with source of variation extracted for persons and diagnosis (tumor, squamous metaplasia, metaplasia, and bronchiole).

To examine the abnormalities of the FHIT gene in IPF lesions, we studied a total of 146 metaplastic lesions or bronchiolar epithelia around honeycomb formations in IPF lesions that had been obtained from eight IPF patients with lung cancer and three IPF patients without lung cancer. The results of PCR-based LOH analysis of four microsatellite markers within or surrounding the FHIT gene showed that LOH of the FHIT gene region was frequently observed among the metaplastic lesions and bronchiolar epithelia of IPF lesions (Fig. 1; Tables 1 and 2). None of the normal tissue samples (normal-appearing respiratory epithelial cell areas and vessel wall cells) showed LOH of the FHIT gene (data not shown). LOH of the FHIT gene region in the metaplastic lesions and bronchiolar epithelia was found in all of the informative samples obtained from six of the eight patients with lung cancer (cases 1–5 and 7) and from the three patients without lung cancer (cases 9–11). Fifty-four (73%) of the 74 informative metaplastic lesions and bronchiolar epithelia from the IPF patients with lung cancer, and 8 (17%) of the 46 informative samples from the IPF patients without lung cancer had LOH of the FHIT gene, and this difference was statistically significant (P < 0.0001). The pattern of allelic loss in the samples of metaplastic lesions and bronchiolar epithelia from an individual IPF patient were similar but not identical. The metaplastic lesions and bronchiolar epithelium samples from an individual patient tended to have allelic loss of the same microsatellite markers. To determine whether the variability between persons was greater than the variability between diagnoses, the allelic deletions were analyzed as response variables in an ANOVA model (SAS system) with the source of variation extracted for person and diagnosis (tumor, squamous metaplasia, metaplasia, and bronchiole). The variability between persons was greater than that between diagnoses (Table 3). In each of the six IPF patients with lung cancer who showed allelic loss in all of the informative samples, the allelic loss in the tumor sample was similar to that in the metaplasias and bronchiolar epithelia of the respective patient.

To confirm LOH of the FHIT gene in the IPF lesions, we performed FISH of the FHIT gene in the IPF lesions of seven patients with lung cancer (cases 1–4 and 6–8; four adenocarcinomas and three squamous cell carcinomas). We conducted FISH analysis of the FHIT gene in IPF lesions using the BAC, 1-O-12 probe (kindly given to us by Dr. Kay Huebner) that spanned exon 5 of the FHIT gene, and the chromosome 3 centromere probe (kindly given to us by Dr. Joe W. Gray). Because it was difficult to achieve hybridization to the cells in each IPF lesion using the formalin-fixed, paraffin-embedded sections, we evaluated LOH in the limited lesions that had hybridized with the probe. In all seven of the patients, the nuclei of cells in metaplasias and bronchiolar epithelia around honeycomb formations showed LOH of the FHIT gene by FISH (Fig. 2), whereas most of the normal cells did not show this (Table 4).

Using immunohistochemistry, we also analyzed Fhit protein expression in the IPF lesions of six patients with lung cancer (cases 1–3 and 6–8; Table 4). The samples of normal bronchial epithelium and bronchiolar epithelium showed strong staining for Fhit protein in all of the patients and were used as an internal control for immunostaining. In the areas of honeycomb formation, the bronchiolar metaplastic cells showed uniform, strong staining for Fhit protein (Fig. 3, A and B). In contrast, the cuboidal, flattened metaplastic epithelial cells showed reduced Fhit protein staining (Fig. 3, A and B). The squamous metaplastic epithelial cells showed different patterns of Fhit immunostaining in some lesions. Some squamous metaplastic epithelial cells showed strong Fhit protein staining, whereas other cells showed reduced Fhit protein staining (Fig. 3, A and B). The metaplastic epithelial cells of 5 of the 6 cases had either reduced or no staining for Fhit protein (Table 4).

This is the first study that demonstrates a high incidence of aberration of the FHIT gene, a candidate TSG, among IPF lesions. Because our preliminary examination showed that lung cancer develops in the peripheral field of the lung in ∼84% of patients with IPF, we hypothesized that the metaplasias of IPF are involved in the carcinogenesis of lung cancer and seem to be good tissue samples for research on human carcinogenesis in the peripheral lung. The high incidence of FHIT gene allelic loss among the metaplasias of IPF lesions in this study supports this hypothesis. In addition, FHIT gene allelic loss was seen more frequently among the metaplasias and bronchiolar epithelia samples obtained from IPF patients with lung cancer than among the samples obtained from those without lung cancer. Sozzi et al. (15) showed that metaplasias in the bronchus had genetic and functional aberration of Fhit, and their results implied that Fhit may be related to carcinogenesis derived from the bronchus. Our immunohistochemical analyses also demonstrated reduced FHIT expression in metaplastic lesions surrounding honeycomb formations. Case 3 showed no allelic loss of the FHIT gene in the lung cancer samples; however, this patient had a heterogeneous pattern of elevated and reduced Fhit protein expression. As to the lung cancer sample from Case 3, there may have been an inactivation mechanism of the FHIT gene different from allelic loss (38, 39); it may have been contaminated with normal cells; or other noninformative regions may have been lost.

It is often difficult to evaluate microdissected samples that are fixed in formalin and embedded in paraffin, by PCR-based analysis. There is a possibility of artificial modification by the fixation and embedding process, which may influence the results. After we extracted DNA from a microdissected sample, we first conducted PCR to amplify D3S1339, which exists at 3p, to screen the quality of the DNA for PCR assay. We performed PCR-based analysis several times and confirmed the results of the PCR-based analysis. In samples with weak signals, i.e., the samples from cases 2 and 3, the results were confirmed by PCR-based assay using microdissected samples of corresponding sites in serially cut sections. In each individual IPF patient with lung cancer in the present study, the pattern of allelic loss in the metaplasias and bronchiolar epithelia was similar (Table 3); therefore, the influence of artificial modification seems to be small. Our data from a FISH analysis of the FHIT gene supported the results of the PCR-based analysis. Additionally, the results of the immunohistochemical analysis of Fhit protein implied functional aberration of the Fhit protein in the metaplasias and bronchiolar epithelia of patients with IPF.

The allelic loss of FHIT was similar in the samples of metaplasia and bronchiolar epithelia from an individual patient. These results raise important issues. Preneoplastic changes are often extensive and multifocal, occurring throughout the respiratory tree, a phenomenon referred to as field cancerization (40). If this phenomenon is applicable to IPF lesions, the following questions are raised: (a) whether specimens exhibiting allelic loss are derived from the same clone; and (b) whether exposure of the IPF patient to a mutagen led to carcinogenesis. Additionally, Park et al. (10) determined the size of clonal patches of bronchial epithelium (both normal and abnormal) by studying the molecular changes (allelic losses and microsatellite alterations) at several locations. They showed that multiple small patches or subclonal patches containing molecular abnormalities were present in the normal or slightly abnormal bronchial epithelia of patients with lung cancer, and that each patch usually consisted of 200–400 cells. Because our microdissected foci consisted of ∼200 cells, their findings may be in agreement with our results. However, to clarify clonality of these regions, fine analysis of LOH at several loci using contiguous foci containing a minimal number of cells adjacent to multiple metaplasia exhibiting allelic loss, is required. Furthermore, IPF is a chronic inflammatory disorder and is associated with repeated cycles of damage and repair. Under these circumstances, cells of metaplasias that had developed to repair the lung structure, might incur chromosomal and genetic aberrations. The metaplastic lesions in patients with non-small cell lung cancer have been reported to have a K-ras gene point mutation (41) and oncogene expression including c-myc mRNA and c-jun and p53 proteins (42). These aberrations may occur not only de novo but also on exposure to mutagens. Occupational or environmental exposure may increase the risk for both apparent IPF and lung cancer. Several studies have shown that metal dust, cigarette smoking, and asbestos may be risk factors for IPF (43, 44, 45, 46). Cigarette smoking and asbestos were reported to be associated with FHIT gene alterations and abnormal Fhit expression (40, 46, 47). All of the eight patients with lung cancer in the present study had a smoking history, and silica was detected in three of the lung cancer patients. The similarity in the pattern of allelic loss in the samples of metaplasias from an individual may represent the individual’s specific susceptibility to mutagens or a specific characteristic of the FHIT gene to mutagens.

In summary, we found aberration of the FHIT gene and reduced Fhit expression in IPF lesions. These results suggest that LOH of the FHIT gene is involved in the carcinogenesis of the peripheral type of lung cancer in patients with IPF. FHIT gene alteration is thought to be one of the early events in human carcinogenesis of the peripheral type of lung cancer. Preventing FHIT gene alteration in IPF lesions may prevent lung carcinogenesis in IPF patients, and may be a useful tool for the prolongation of survival of IPF patients.

We gratefully acknowledge Drs. Gabriella Sozzi and Masayuki Noguchi for helpful advice.