Clara cell 10 kDa protein mRNA in normal and atypical regions of human respiratory epithelium

We used RNA‐RNA in situ hybridization to study expression of the human CC10 gene in morphologically normal and atypical areas of 32 non‐neoplastic lung specimens resected from 26 non‐small cell lung cancer patients. We scored strong, moderate or weak levels of CC10 mRNA expression in 3 distinct lung compartments. In morphologically normal lungs, strong and moderate levels of CCIO mRNA were observed in bronchioli and bronchi, respectively, but the expression was rarely observed in the alveolar region. Distinct alterations in CC10 mRNA expression were noted in specific histologic abnormalities within bronchi and the alveolar region. CC10 hybridization signal decreased markedly in bronchi containing diffuse goblet cell hyperplasia or squamous metaplasia, while CC10 mRNA expression remained unchanged in bronchi with basal cell hyperplasia or focal goblet cell hyperplasia. Bronchiolar CC10 mRNA levels remained unchanged in sections containing abnormalities elsewhere. Interestingly, in alveoli with bronchiolization of the alveoli, high levels of CC10 mRNA were observed. These regions also contained strongly stained keratin 14‐positive cells, which may indicate a concurrent metaplastic process. In lungs with morphologic atypias, no correlation was found between abnormalities detected in bronchi and alveoli from the same lung. A comparison of mRNA expression and clinicopathologic features demonstrated that the amount of histologic abnormalities increased with smoking history (pack years); however, no correlation between CC10 mRNA expression and sex, age or smoking history was found.

o 1994 Wiley-Liss, Inc.* We and others have noticed an increase in the proportion of pulmonary adenocarcinomas, with up to 50% containing features suggestive of peripheral airway cell differentiation (Linnoila, 1990). In addition, these tumors are the most common tumor type in women, in younger age groups and in patients with no smoking history, suggesting a distinct biology for cancers arising in the peripheral lung. Pulmonary adenocarcinomas are generally resistant to radiation and chemotherapy, and the majority of patients develop distant metastases long before the diagnosis of lung cancer is made, calling for more effective early detection. However, unlike centrally located squamous cell carcinomas of the lung, for which the sequence of preneoplastic changes has been well described (Nasiell et al., 1987), events preceding the more peripherally located pulmonary adenocarcinomas are not well understood.
One of the progenitor cells for the peripheral airway epithelium and adenocarcinomas arising in this region is the Clara cell, a morphologically defined subtype of non-ciliated secretory cells in bronchiolar epithelium (Massaro et al., 1994). Prior to the availability of sequence information, the molecular weight of the major Clara cell secretory protein was estimated from SDS/PAGE to be about 10 kDa (Singh et al., 1985) and for this reason has been referred to as CC10. The expression of CClO was originally thought to be restricted to Clara cells (Singh et al.,198S), often serving as a marker for these cells. We have previously demonstrated that non-ciliated secretory cells containing CClO can be detected throughout the human tracheobronchial tree (Broers et al., 1992;Linnoila et al., 1992). While CClO expression was abundant in non-neoplastic human lung, it was detectable in tumors and corresponding cell lines at markedly lower levels (Broers et al., 1992;Linnoila et al., 1992). Interestingly, CClO levels were significantly lower in serum and bronchoalveolar lavage specimens obtained from smokers and lung cancer patients compared with specimens from healthy non-smokers (Bernard et al., 1992).
It has been well documented that the entire lung exposed to pulmonary carcinogens such as tobacco smoke may show histologic abnormalities reflecting field cancerization (Auerbach et al., 1961). The previously reported findings of alterations in CClO expression during carcinogenesis prompted us to examine the morphological basis of CClO dysregulation. Difficulty in obtaining normal lung tissue has led us to examine surrounding non-neoplastic lung obtained from patients undergoing resection for cancer.
In this study, we used RNA-RNA in situ hybridization to examine patterns of CClO mRNA expression in morphologically normal and atypical areas of 32 non-neoplastic lung specimens obtained from 26 patients with non-small cell lung carcinomas (NSCLC) or mesotheliomas. Our findings suggest that the expression of CClO mRNA becomes altered in distinct lung compartments and may implicate a role for CClO in the development of pulmonary carcinomas.

Tissues
Thirty-two non-neoplastic lung tissue specimens from 26 patients were reviewed in this study. All specimens were resected from patients who had NSCLC or mesotheliomas and were prepared as previously described (Broers et al., 1992). Tumor histologies of the 26 cases are summarized in Table VII.

In situ hybridization
Preparation of the human CClO kDa mRNA probe and tissue in situ hybridization were performed as previously described (Broers et al., 1992) with the following modification. The 366 bp insert was re-oriented to avoid any incorporation of vector-derived material in the generated RNA probes. Results of the in situ hybridization were scored independently by 2 authors (R.I.L. and S.M.J.) for both the number of positive cells (distribution score: 0 = no positive cells; 1 for 1-10%; 2 for 11-50%; 3 for 51-100% of epithelial cells positive) and the labeling of grains per cell (0 = negative; 1 = weak; 2 = moderate; 3 = strong). Using the sum of these values, a hybridization signal index (HSI = distribution score + intensity score, possible values: 0,2-6) was established for each lung compartment (Table I). All slides were hybridized in duplicate and exposed for 1 or 2 week intervals. Sections from one block were evaluated for 23 of the 26 patients studied, while 2, 3 and 4 separate blocks were reviewed for the remaining 3 patients.

Immunohistochemisty
The mouse monoclonal antibody (MAb) anti-keratin 14 (RCK 107) (Wetzels et al., 1989)  basal cell hyperplasia (a gift from Dr. J.L.V. Broers, University of Limburg, Maastricht, The Netherlands). Frozen sections were cut and thawed onto glass slides pre-coated with 0.1% poly-L-lysine and fixed by dipping once in -20°C methanol, and 3 times in 4°C acetone. Slides were air dried and stored at -20°C until use.
Immunohistochemical staining was performed using the Vectastain ABC staining kit (Vector, Burlingame, CA) following the vendor's instructions with modifications as described (Linnoila et al., 1988).

Histopathology
Changes in 3 distinct lung compartments (bronchi, bronchioli, alveolar region) were recorded. Hematoxylin and eosinstained sections of all tissues were screened for the following histologic abnormalities: basal cell hyperplasia, focal and diffuse goblet cell hyperplasia, squamous metaplasia, dysplasia, type I1 cell hyperplasia, fibrosis, and bronchiolization of the alveoli. The identification of histologic abnormalities was based on histologic characteristics described in previous reports (Nasiell, 1963;Nettesheim and Szakal, 1972;Adekunle et al., 1991). Basal cell hyperplasia was scored when three or more basal cell layers were seen. Focal goblet cell hyperplasia was represented by normal appearing respiratory epithelium interrupted by clusters of 3 or more goblet cells with no more than 50% of bronchial lining cells composed of goblet cells, while diffuse goblet cell hyperplasia was scored in areas of more than 50% goblet cells. Squamous metaplasia was characterized by areas containing layers of flattened epithelial cells with cilia absent. Dysplasia was used to describe 2 types of lesions: 1) atypical squamous metaplasia (borderline carcinoma in situ) (Nasiell, 1966); and 2) atypia in epithelium that retained the general character of columnar epithelium with preserved cilia (Nasiell, 1963).

Clinicopathologic analyses
Data were obtained from 26 NSCLC patients. Because multiple slides were obtained in 3 of the cases and one slide was available on each of the other 23 patients, data from each of the multiple slides were averaged to form a single value per patient. Multiple values obtained per case were averaged to form one value per compartment resulting in one HSI value for each of the 3 lung compartments per patient (Tables 111-VII). Comparisons between 2 groups of patients' HSI values were made using the Wilcoxon rank sum test while the Kruskal-Wallis test was used to test for differences when multiple histologies were compared. Spearman rank correlation analysis was used to demonstrate the association between CClO HSI values and pack years and age. All p values reported are two-sided.
'Distribution score equals the percent of positive epithelial cells in high-power field.-'Hybridization signal index (HSI) equals the sum of distribution and intensity scores.-)For example, moderate score could otentially be distribution score 2 plus intensity score 1 = 3, as we1 as distribution score 2 plus intensity score 2 = 4.

CClO mRNA in histologically normal human lung
Mean CClO HSI was calculated for each of 3 lung compartments (bronchi, bronchioli, alveolar region) ( Table 11) 'BCH, basal cell hyperplasia; FGCH, focal goblet cell hyperplasia; DGCH, diffuse goblet cell hyperplasia; SQM, squamous metaplasia; DYS, dysplasia; FIB, fibrosis; T2H, type I1 cell hyperplasia; BOA, bronchiolization of the alveoli; AD, adenocarcinoma; ADSQ, adenosquamous; CRC, carcinoid; ME, mesothelioma; OT, other.-2HSI, hybridization signal index, the sum of distribution and intensity.-'Control, specimens containing no histologic abnormalities. Dysplasia was used to describe two ypes,of, lesions: 4atypical squamous metaplasia (borderline carcinoma in situ), (Nasiell, 1966), and atypia in epithelium that has retained the general character of columnar epithelium and preserved cilia (Nasiell, 1963).     'Includes focal and diffuse goblet cell hyperplasia.-2Mean number of pack years was calculated for all patients whose specimens contained one or more histologic abnormalities in each lung compartment. of 8 of 21 bronchi and 62 of 63 bronchioles examined were classified as controls, with an average HSI of 3.9 and 5.0, respectively. A total of 11 of 29 alveolar regions examined were classified as controls, with an average HSI of 0.5.
Large bronchi contained numerous scattered non-ciliated columnar cells demonstrating variable CClO mRNA expression ( Fig. la,b). We also detected focal hybridization signal in the epithelium of bronchial glands (Fig. lc,d). In terminal bronchioli, both the distribution and intensity of CClO mRNA 'One adenosquamous.-21ncludes two sarcoma, one small cell t~mor.-~Tissue specimens of the 26 patients we studied contained different combinations of bronchi, bronchioli and alveoli. expression was higher than that seen in large bronchi. It appeared that most cells did express CClO mRNA (Fig. le). While grain density in epithelial cells was similar to that seen in terminal bronchioli, the total number of cells expressing CClO mRNA in respiratory bronchioli was lower than that seen in bronchi and terminal bronchioli (Fig. If). CClO mRNA expression was rarely observed in histologically normal alveoli (Fig. lg).

Keratin 14 staining in histologically normal human lung
To aid in the detection of basal cell hyperplasia, we stained all tissue sections with the basal cell-specific keratin 14 MAb. As expected, keratin 14 immunoreactivity was strong in bronchial basal cells and glands (Fig. 2a), and only weakly expressed in bronchiolar basal cells (Fig. %). No immunoreactivity was observed in the alveolar region (Fig. 2).

Distribution of histological abnormalities
When we examined the distribution of histologic abnormalities in a given lung we found no association between findings of atypia in either bronchial versus alveolar compartments. For example, 3 of the 6 (50%) control specimens that lacked histologic abnormalities in the bronchial compartment also lacked abnormalities in the alveolar compartment, while the other 50% contained alveolar abnormalities. Furthermore, of the 9 specimens containing bronchial abnormalities, 519 (55%) demonstrated normal alveolar regions and 4/9 (45%) showed atypia.
Despite numerous abnormalities in bronchi and alveoli, histologic abnormalities in bronchioli were rare. This disparity, and the fact that many changes occurred simultaneously in various compartments, prompted us to focus on defining CClO mRNA expression patterns in individual compartments when-  (Table VI).

CClO mRNA in histologically abnormal human lung
Histologic abnormalities in conducting airways. CClO mRNA expression was examined in bronchi and bronchioli containing histologic abnormalities associated with the conducting airways (basal cell hyperplasia, focal and diffuse goblet cell hyperplasia, squamous metaplasia, dysplasia) (Table 11). CClO mRNA expression was detected in bronchial glands adjacent to abnormal bronchi, similar to that seen in normal controls (Fig. 2c,d). When compared with controls, CClO mRNA expression increased in one bronchus containing basal cell hyperplasia only, while levels of expression remained unchanged in bronchi containing both basal cell hyperplasia and focal goblet cell hyperplasia. However, in bronchi containing diffuse goblet cell hyperplasia, the average CClO HSI decreased from 3.9 to 2.8 (Fig. 3a,b). Squamous metaplasia was detected in one bronchus with an HSI of 3.0. We obtained discordant results for bronchi containing dysplasia with high CClO mRNA levels in one (not shown) and decreased expression in the other (Fig. 3c,d). Keratin 14-positive basal cells did not express CClO mRNA (Fig. 3e,f).
While histologic abnormalities were rarely detected in bronchioli, minor variation occurred in CClO HSI. Only 1/63 demonstrated dysplasia in which CClO mRNA expression remained unchanged (Table 11).
Histologic abnormalities in the alveolar region. In sections containing histological abnormalities associated with the alveolar region (fibrosis, type I1 cell hyperplasia, bronchiolization of the alveoli), 8 alveolar regions contained fibrosis only and demonstrated no CClO mRNA expression (Table 11). In alveolar regions containing both fibrosis and focal type I1 cell hyperplasia (Fig. 4a,b), the average CClO mRNA HSI re-mained unchanged from that of control alveoli. However, a marked increase in the average CClO HSI (3.7) was observed in alveoli containing fibrosis, type I1 cell hyperplasia, and bronchiolization of the alveoli (Fig. 4c,d). These findings were most remarkable since little or no CClO mRNA expression was detected in histologically normal alveoli (Fig. lg).

Keratin 14 staining in histologically abnormal human lung
Histologic abnormalities in conducting airways. In histologically abnormal conducting airway epithelium, strong keratin 14 immunoreactivity was detected in bronchial basal cells containing areas of basal cell hyperplasia and squamous metaplasia, while immunoreactivity remained unchanged in bronchioli.
Histologic abnormalities in the alveolar region. While normal alveolar cells lacked immunoreactivity for the basal cell marker keratin 14, focal areas of strong keratin 14 immunoreactivity were observed in alveoli containing bronchiolization of the alveoli (Fig. 4eJ). The increase in CClO mRNA expression, along with the appearance of keratin 14 positive cells, suggests that 2 distinct cell types may be associated with regions undergoing hyperplastic growth and squamous metaplasia in the alveolar region of human lung.

Clinicopathologic correlation
Histologic abnormalities and CClO mRNA expression in various lung compartments were correlated with clinicopathologic features such as smoking history (pack years), sex, age and histologic diagnosis (Tables 111-VII). Smoking history was available for 24 of the 26 patients.
To illustrate the correlation of CClO and smoking history (pack years), we grouped these patients into those who smoked 0-49 pack years, or 50+ pack years. We found no statistically significant correlation between male or female smokers and mean CClO mRNA levels in any of the 3 lung compartments (Table 111). We also compared sex and age of each patient with the expression of CClO mRNA in each lung compartment. As expected, CClO mRNA expression was highest in bronchioli. No significant correlation was found with age in either gender (Table IV). Spearman rank correlation similarly revealed no associations with CClO expression and smoking history or age of patients (Table v).
In addition, we compared the expression of each histologic abnormality with smoking history. A greater than 50 pack year smoking history was associated with basal cell hyperplasia, goblet cell hyperplasia, squamous metaplasia and dysplasia in the conducting airways, and type I1 cell hyperplasia and bronchiolization of the alveoli in the alveolar region (Table   VI) .
Lastly, we investigated CClO mRNA expression and its possible correlation with histologic subtype of the resected tumor. To simplify the analysis, we combined squamous and adenosquamous histologies into one group (squamous, n = 9). Once again, CClO mRNA levels were higher in bronchioli than bronchi in all groups; however, we observed no statistically significant correlation between CClO mRNA expression and histologic subtype (Table VII). DISCUSSION Our study has demonstrated the wide range of changes associated with field cancerization in the human lung. We were able to show that smoking history correlated with an increase in histologic abnormalities that were unevenly distributed in the 3 main lung compartments. The least number of changes were observed in bronchioli. Histologic abnormalities were associated with marked changes in CClO mRNA expression, which is a well-defined product of non-ciliated secretory epithelial cells. In bronchi, CClO mRNA expression levels decreased in the presence of diffuse goblet cell hyperplasia and squamous metaplasia, while CClO mRNA levels increased in the alveoli with bronchiolization of the alveoli.
The spectrum of histologic changes that we observed in the conducting airways was in accordance with previously published reports outlining events leading to squamous cell carcinoma (Nasiell, 1963). We also detected some of these changes in lungs from patients with adenocarcinoma. These findings are in agreement with those of Solomon et al. (1990), who showed focal areas of basal cell hyperplasia, squamous metaplasia and dysplasia in grossly unremarkable bronchi and bronchioles taken from lobectomy specimens containing primary adenocarcinoma. Histologic abnormalities that we observed in the alveolar region included bronchiolization of the alveoli and squamous metaplasia. These lesions are known to occur after a variety of insults to the lung (respiratory infection, exposure to chemical irritants and carcinogens) (Nettesheim and Szakal, 1972;Fukuda et al., 1989); however, their significance in the development of lung cancer is unknown. The discordant manifestation of histological changes in various lung compartments may be due to the possibility that individual lung compartments respond to carcinogens at different rates. Many of the changes we observed were focal in distribution; therefore without evaluating an extensive number of tissue specimens, many of these lesions are likely to be missed (Slaughter et al., 1953). Consequently, further studies are needed to establish the relationship between histologic abnormalities and the development of non-squamous cell carcinomas of the lung.
We have previously demonstrated that while CClO mRNA was abundant throughout the conducting airways, expression was detected infrequentIy and at low levels in lung tumors (Broers et al., 1992). Our current findings demonstrated that alterations in CClO mRNA levels were associated with regions undergoing a change of cell type. While no change was detected in non-ciliated secretory epithelium with underlying basal cell hyperplasia, altered CClO mRNA expression was observed in bronchi with diffuse goblet cell hyperplasia and squamous metaplasia and alveolar regions containing bronchiolization of the alveoli. Like the histologic abnormalities, the changes in CClO may be focal and dependent on lung compartment. It is not clear whether CClO mRNA levels are changing due to alterations in the pattern of gene expression in a given cell. Other approaches are necessary to elucidate this point further. Our findings of decreased CClO levels in bronchi are in accordance with the studies that showed decreased CClO protein expression in serum and bronchoalveolar lavage specimens of smokers and lung cancer patients (Bernard et al., 1992). While a significant decrease in Clara cell number in distal airways of smokers has been described (Lumsden et al., 1984), our results on CClO expression in bronchioli were inconclusive. The decrease in Clara cell number may be of interest because Clara cells are active in the metabolism of xenobiotics (Boyd and Schuller, 1984).
CClO protein is one of the major respiratory tract-derived proteins, amounting to 7% of the total protein content of lung lavages obtained from healthy non-smokers (Bernard et al., 1992). While the function of CClO protein has not been fully determined, it has been shown to inhibit phospholipase A2 (Singh et al., 1990) and bind polychlorinated biphenyls, a major component of industrial pollution (Anderson et al., 1991). Phospholipase A2 inhibition may modulate inflammation and chemotaxis in the human respiratory tract. Previous studies have demonstrated an imbalance of T-cell subsets (Ginns etal., 1982) and antitumor activity of peripheral blood monocytes in sera of lung cancer patients (Mantovani et al., 1979). Antiinflammatory agents such as aspirin and piroxicam have been implicated in the suppression of neoplastic transformation in the gastrointestinal tract (Marnett, 1992). However, more studies are needed to determine the nature of pulmonary defenses against neoplastic disease. If CClO does in fact provide a protective effect in the human lung, it is reasonable to assume that reduced CClO levels may contribute to impaired respiratory tract function and subsequent neoplastic cell growth. The ability to bind polychlorinated biphenyls suggests that CClO may be important in the clearance of harmful substances deposited within the respiratory tract.
We conclude that: 1) resected lung cancer specimens provide excellent material to study field effects; 2) the process is complex, and changes may be focal or proceed into different directions; 3) altered CClO mRNA expression may indicate a change in cell type; and 4) more studies are needed to evaluate cancer-specific lesions.
In the future, it will be important to perform the following studies: 1) extensive tissue sampling to study alterations in bronchiolar CClO expression; 2) determination of normal CClO expression in various lung compartments from nonsmoking subjects; and 3) assessment of CClO levels as a potential marker for the early detection of lung cancer.