What conclusions can be drawn from the enhanced expression of growth factors, cell adhesive molecules and oncogenes in the process of metaplasia?
G. Triadafilopoulos (Palo Alto)
Esophageal columnar metaplasia, is a premalignant lesion for adenocarcinoma of the esophagus and the gastric cardia, in which columnar epithelium containing goblet cells (intestinal metaplasia) replaces the squamous esophageal epithelium . Although most frequently occurring in the context of gastroesophageal acid reflux, esophageal columnar metaplasia is of uncertain pathogenesis. Adenocarcinoma in the setting of esophageal columnar metaplasia does not arise de novo, but follows a multistep process in which the metaplastic epithelium develops low-grade dysplasia, high-rade dysplasia, and eventually invasive adenocarcinoma . Several recent studies have focused on the identification of biological markers that may characterize the metaplastic process at the molecular level or herald the development of dysplasia and neoplasia. The following discussion specifically outlines the potential role of growth factors, cell adhesion molecules, and oncogenes in the process of metaplasia, dysplasia, and adenocarcinoma.
Both gastroesophageal reflux disease (GERD) and Barrett's esophagus (BE) may be complicated by esophageal fibrosis and stricture formation. Transforming growth factor beta-one (TGFb1) is a multifunctional cytokine which may play a central role in the pathogenesis of chronic inflammation and fibrosis . It is the first member of a large family of secreted signaling molecules, known as the TGFb superfamily, that appear to mediate many key events in normal growth and development as well as tissue repair after injury . TGFb1 is a 25-kD homodimeric or heterodimeric protein that can mediate a broad spectrum of biological activities . Although TGFb was originally recognized by its ability to transform normal rat fibroblasts in vitro, it has been subsequently shown that it is chemoattractant for fibroblasts, stimulates fibroblast proliferation, and enhances fibroblast collagen and fibronectin synthesis, while inhibiting collagenase gene expression in vitro [6, 7].
Figure 1. Western blot analysis of TGFb1 in three representative patients studied. Biopsies from normal esophagus, esophagitis, Barrett's intestinal metaplasia, and Barrett's adenocarcinoma were solubilized, then analyzed by immunoblotting. Lanes N represent normal esophagus, lane E esophagitis, lane B Barrett's intestinal metaplasia, and lane C esophageal adenocarcinoma.
Figure 2. Cummulative data of densitometric analysis of immunoblots in the three groups studied. Compared to normal esophageal epithelium (6.3 ± 1.4 densitometric units), the esophagitis group revealed a two-fold increase in TGFb1 expression (12.4 ± 3.09, p < 0.03). Such increase was even more pronounced in the Barrett's group which revealed a four-fold increase in expression
(23.3 ± 6.1, p < 0.0 1). The adenocarcinoma group showed enhanced expression of ten-fold magnitude (40.0 ± 5.8) when compared to the normal esophageal epithelium (p < 0.001). No differences could be detected between the two groups of Barrett's esophagus (i.e. with or without dysplasia).
In order to explore a potential relation between TGFb1 expression and metaplasia or carcinogenesis, we compared endoscopic mucosal biopsy specimens of esophagitis (n = 8), BE of the intestinal metaplasia (IM) type (n = 8) and adenocarcinoma (n = 5) to normal esophagus (n = 21) and duodenum (n = 13). The samples were analyzed for the expression of TGFb1 by both western analysis and immunoperoxidase staining .
Densitometric analysis of western blots revealed the expression of TGFb1 to be significantly higher in patients with erosive esophagitis (p < 0.03) when compared to normal squamous esophageal epithelium. When studied repeatedly over time in a selected group of patients with esophagitis, TGFb1 expression remained increased (sixfold over control). As compared to both squamous esophageal epithelium and columnar epithelium of duodenal mucosa, TGFb1 expression was even more pronounced in Barrett's esophagus (p < 0.01) and in the esophageal adenocarcinoma group (p < 0.001) (Figures 1 and 2).
Immunoperoxidase staining, to examine the localization of TGFb1 in these tissues, complemented the western analysis results. All normal esophageal epithelia demonstrated homogeneous TGFb1 expression around the vascular papillae. TGFb1 expression in esophagitis was significantly enhanced around the vascular papillae with more positive staining in the surrounding epithelial layer and the thickened basal layer. Out of 8 cases in the Barrett's esophagus group, 6 showed maximal staining (2+), while the remaining, expressed TGFb1 in less than 75% of the cells. The immunolocalization of TGFb1 was not only in the glands, but also in the lamina propria. There were no significant differences between the Barrett's esophagus patients with or without dysplasia. In the adenocarcinoma group, out of a total of 5 cases, the two well differentiated adenocarcinomas expressed TGFb1 levels that were near normal (i.e. < 75% of the cells were positively stained), while the other three showed maximal (2+) staining throughout.
These results suggest that TGFb1 is constitutively expressed in normal esophageal epithelia and that its expression is enhanced in the inflammed epithelium of GERD, the metaplastic epithelium of Barrett's esophagus, and in esophageal adenocarcinoma in a progressively increasing fashion. Therefore, TGFb1 might play a role in the progression of GERD and Barrett's metaplasia to esophageal stricture and adenocarcinoma formation.
Cell adhesion molecules
Cell-cell interaction and cell-matrix interaction mediated by cell adhesion molecules play an essential role in the induction and maintenance of a differentiated epithelial cell phenotype and in structural and functional tissue differentiation . Cadherins are a family of transmembrane glycoproteins that are thought to play a vital role in cell-cell adhesion; loss or down-regulation of their expression has been implicated in neoplasia. Unstable or reduced expression of E-cadherin has been shown in lung, gastric, hepatocellular and breast carcinomas, and in a human esophageal cancer cell line .
We recently evaluated the expression of E-cadherin, a 120 kD protein, in esophageal columnar metaplasia and Barrett's adenocarcinoma compared with normal and inflamed esophageal epithelia and the relationship between E-cadherin expression and the histologic grading of metaplasia and differentiation . Endoscopically obtained mucosal biopsy samples of esophagitis (n = 6), esophageal columnar metaplasia with or without dysplasia (n = 16), and esophageal adenocarcinoma (n = 6) were analyzed for E-cadherin by both Western analysis and immunoperoxidase staining.
Figure 3. Relative E-cadherin expression in esophagitis, Barrett's esophagus, and esophageal adenocarcinoma as percentage compared with expression in the normal esophagus for each individual patient. E-cadherin expression was evaluated by immuno-blotting of detergent lysate of mucosal biopsies from 28 patients.
Densitometric analysis of immunoblots revealed the expression of E-cadherin to be significantly lower in patients with Barrett's esophagus compared to normal esophageal epithelium, regardless of the presence or absence of dysplasia (p < 0.03); there were no differences between the normal esophagus and the esophagitis groups. In the adenocarcinoma group, one patient showed complete loss of E-cadherin expression and the other five showed significantly reduced expression that was even lower than in Barrett's esophagus (p < 0.01) (Figure 3). These immunoblot results were confirmed by immunohistochemistry which revealed strong E-cadherin expression on the cell-cell boundaries in normal and inflamed esophageal epithelium, less than maximal staining in all cells of Barrett's mucosa, and near absent or absent expression in adenocarcinoma cells.
Carcinogenesis can result from genetic alterations resulting in activation of cellular transforming genes. Several recent studies have identified genes whose structure, expression, or activity is altered in different stages of esophageal cancer [12, 13].
The c-erb-2 protooncogene encodes a growth factor receptor similar in sequence to the epidermal growth factor (EGF) receptor . The protein has tyrosine kinase activity, and its expression in a variety of fetal tissues suggests a widespread developmental role in cell growth and differentiation. Although expression in adult tissues is generally at lower levels and appears to be largely confined to epithelial cells, a recent study has shown that c-erb-2 gene is amplified in Barrett's adenocarcinoma .
The ras family of oncogenes (H, K, and N) encode specific proteins designated p21 which may be essential components of normal cell division and differentiation. In addition, these protooncogenes may modulate cell growth processes by their ability to abrogate specific growth requirements of various target cells . The expression of protooncogenes may differ in various stages of development and in different regions of the gastrointestinal tract. Although aberrant expression of cellular protooncogenes has been demonstrated in premalignant and malignant transformation in the gastrointestinal tract, the potential role of ras protooncogenes in the metaplastic condition of BE has not been studied.
We therefore studied ras protooncogene expression in upper gastrointestinal epithelia using three monoclonal antibodies in order to determine the relationship between aberrant differentiation and ras protooncogene expression . Specimens from 30 patients of normal esophagus, stomach, and duodenum, were examined and compared with specimens of esophageal columnar metaplasia.
In esophageal mucosal biopsies obtained proximal to the squamocolumnar junction, ras protooncogene reactivity was totally absent. In specimens obtained from squamous epithelium adjacent to Barrett's specialized columnar epithelium, N-ras labeling was enhanced and similarly expressed in the cell membrane and cytoplasm. N-ras protooncogene expression in these biopsies was limited to the specimens in which associated N-ras expression was observed on the adjacent Barrett's specialized columnar epithelium. There was no reactivity of H-ras and K-ras detected in any of the squamous epithelial biopsies studied.
In Barrett's metaplasia of the gastric fundic-type, there were no appreciable differences in the reactivity to any of the three antibodies and that of the gastric fundus in every component studied. Staining was faint and similar for all three protooncogenes studied. Specimens with junctional-type Barrett's epithelium were categorized into four distinct groups. Group 1, comprising 18% of the biopsies, revealed expression of all three protooncogenes localized to the basal cell membrane. In these samples, N-ras expression was augmented as compared to H- and K-ras. Group 2 consisted of 10% of tissues studied and revealed enhanced N-ras labeling in relation to H-ras, and absent K-ras immunoreactivity. Antigen localization was predominantly in the basal membrane. Group 3 consisted of 20% of specimens revealing marked N-ras expression in the basal membrane with absent H-ras and K-ras immunoreactivity. Finally, group 4, consisted of 52% of specimens in which none of the three protooncogenes revealed any labeling (Table I).
Table II summarizes the expression of ras protooncogenes in Barrett's specialized columnar epithelial metaplasia. On the basis of the differential expression of the
3 protooncogenes studied, six distinct groups were identified. Group I consisted of 24% of specimens that revealed immunolabeling for all three protooncogenes, localized mainly in the basal membrane and with N-ras predominating. There was no significant difference between ras immunoreactivity in goblet or mucus cells. Group 2 comprised 22% of specimens in which N-ras was markedly expressed, H-ras was present, and K-ras was absent. Antigen localization was most pronounced and similarly present in the basal membranes of goblet and mucus cells. Group 3 consisted of 8% of specimens, in which
H-ras reactivity was absent, and both N-ras and K-ras were expressed. Group 4 consisted of 5% of specimens studied in which N-ras labeling was absent and both H-ras and K-ras were expressed, mainly along the basal membrane. Group 5 consisted of 10% of samples in which N-ras was the only protooncogene expressed. Finally, group 6 consisted of 31% of specimens that revealed no ras immunolabeling.
In the stomach, all three protooncogenes revealed similar pattern of expression, consisting of faint to absent staining in biopsies of fundic gastric mucosa. Immunoreactivity of ras protooncogenes was limited in most specimens to the parietal cell cytoplasm with absent staining in the other components of the mucosa. Biopsies of antral mucosa revealed no immunolabeling for any of the three protooncogenes studied. Similarly, in the duodenum, ras immunoreactivity was absent in the great majority of the specimens studied, and only faint labeling was observed in some samples without predominance of any antigen.
These data suggest that there was no expression of ras protooncogenes in the great majority of specimens obtained from the duodenum, stomach, esophagus, and Barrett's fundic-type epithelium. In Barrett's junctional and specialized columnar (intestinal) epithelium variable augmentation of expression of each of the ras proteins was observed, with predominance of N-ras in most specimens. Enhanced expression of N-ras was also encountered in normal esophageal mucosa adjacent to the Barrett's specialized columnar epithelium. The observed aberrant expression of N-ras antigens may be due to an altered differentiation program in Barrett's metaplastic epithelium and could play a role in the process of metaplasia and carcinogenesis in Barrett's esophagus. Prospective serial sampling of Barrett's epithelium over years of patients' follow-up may allow correlation of changes in protooncogene expression with increasing trend towards neoplasia and elucidation of the mechanisms involved in aberrant differentiation and carcinogenesis.
In a more recent study, we examined whether activation of the Src tyrosine kinase is involved in Barrett's metaplastic and neoplastic transformation. The cellular oncogene
c-src and its viral homologue v-src encode a 60 kDa, cytoplasmic, membrane-associated, protein kinase [18, 19]. For the viral protein (v-src), or the cellular protein (c-src), a close correlation exists between elevated specific kinase activity and cell transformation . Src activity is elevated in three human cancers: colon, breast, and neuroblastoma. In terms of preneoplastic intestinal epithelia, Src activity is elevated in colonic adenomas that are at greatest risk for developing colon cancer, such as those that contain villous architecture and/or severe dysplasia . Src activity is also elevated in dysplastic epithelia of ulcerative colitis, a chronic inflammatory disease of the colon with an increased risk for colon cancer .
Figure 4. In vitro Src tyrosine kinase activity (left) and Src protein levels (right) in normal esophagus (E), Barrett's esophagus (B), and duodenum (D). Proteins were precipitated from tissue lysates with excess anti-Src Mab327. Immunoprecipitates were incubated with [gamma-32P] ATP and enolase and resolved on a 7% SDS-polyacrylamide gel. Equivalent immunoprecipitates to those used for the kinase assay were also analyzed for Src content by immunoblotting using Mab327. Representative blots from one patient (patient #10).
Figure 5. In vitro Src tyrosine kinase activity (left) and Src protein levels (right) in normal esophagus (E), Barrett's adenocarcinoma (Ca), and duodenum (D). Proteins were precipitated from tissue lysates with excess anti-Src Mab327. Immunoprecipitates were incubated with [gamma-32P] ATP and enolase and resolved on a 7% SDS-polyacrylamide gel. Equivalent immunoprecipitates to those used for the kinase assay were also analyzed for Src content by immunoblotting using Mab327. Representative blots from one patient (patient #5).
Since activation of the Src tyrosine kinase occurs in malignant and premalignant lesions of the colon, we measured Src expression and tyrosine kinase activity in 11 patients with BE and 5 patients with esophageal adenocarcinoma . Src specific activity was 3 to 4-fold higher in Barrett's epithelia than in normal esophagus (squamous control) or duodenum (columnar epithehal control) (Figure 4), and 6-fold higher in esophageal adenocarcinoma than in adjacent normal mucosa (Figure 5). Different regions of BE from the same patient revealed significant heterogeneity in Src specific activity in contrast to the uniform Src activity observed in different regions of normal esophagus and duodenum. In all tissues examined, Src kinase activity and protein were preferentially associated with the Triton X100-soluble rather than the remaining insoluble fraction. Immunoblotting with antiphosphotyrosine antibodies and tryptic phosphopeptide mapping of ex vivo phosphorylated Src suggested relative dephos-phorylation of Src at Tyr 527 in BE, and a mechanism for the observed increase in specific activity. Therefore, the activation and microheterogeneous distribution of Src in the premalignant and malignant Barrett's epithelia, together with the strong historical correlate of elevated Src activity in cell transformation, suggest that activation of this kinase may play a part in the early events of Barrett's metaplasia and adenocarcinoma.
Many forms of cancer, particularly those that occur on surfaces exposed to the environment, such the gastrointestinal tract, are preceeded by preneoplastic events. These events are characterized by dysplastic changes, particularly in the cell nucleus, that are frequently associated with metaplasia and carcinoma in situ. The underlying basal membrane is not interrupted, there is no invasion, metastasis or other features of cancer behavior. Through multiple steps, the preneoplastic lesions may progress to neoplasia but they may also remain stable for long periods and may even regress, particularly if the inciting environmental agent(s) is removed. In the case of Barrett's esophagus, the preneoplastic change is typically recognized endoscopically in about 10% of patients with gastroesophageal reflux disease but no interception of the preneoplastic process by specific chemoprevention or other treatment strategies has thus far been feasible.
The molecular markers such as those described above will not only allow a better understanding of the early and late molecular events taking place in the course of metaplasia, dysplasia and neoplasia but they will undoubtedly serve as tools in enhancing our understanding of the role of environmental factors that may be involved in Barrett's malignant transformation.
6. Ignotz RA, Massague J. Transforming growth factor-b stimulates the expression of fibronectin and collagen and their incorporation into the extracellular matrix. J Biol Chem 1986;261:4337-4345.
7. Ignotz RA, Endo T, Massague J. Regulation of fibronectin and type I collagen mRNA levels by transforming growth factor-b . J Biol Chem 1987;262:6443-6446.
8. Triadafilopoulos G, Kumble S. Transforming growth factor beta one (TGFbeta1) expression is enhanced in gastroesophageal reflux disease, Barrett's esophagus, and esophageal adenocarcinoma. (Poster presentation at the American Gastroenterological Association Meeting, San Francisco, CA 1996).
11. Swami S, Kumble S, Triadafilopoulos G. E-cadherin expression in gastroesophageal reflux disease, Barrett's esophagus, and esophageal adenocarcinoma: an immunohistochemical and immunoblot study. Am J Gastroenterol 1995;90:1808-1813.
23. Kumble S, Omary ME, Cartwright CA, Triadafilopoulos G. Src activation in malignant and premalignant epithelia of Barrett's esophagus. Poster presentation at the American Gastroenterological Association Meeting, San Francisco, CA 1996.