Stimulation of resting cells by growth factors leads to an increase in the rate of protein synthesis, which is necessary for cell growth and division. Translation initiation factor eIF-2α is one of the key translation factors mediating the effects of growth factors on protein synthesis. In normal cells, expression of eIF-2α is increased transiently, but its levels are elevated constitutively in oncogene-transformed cells. Overexpression of constitutively active eIF-2α in rodent cells makes them tumorigenic. In this article, the authors report their findings on the increased expression of eIF-2α in human benign and malignant neoplasms originating from melanocytes and colonic epithelium.
Immunohistochemistry was used to analyze the expression of eIF-2α, eIF-4E, and cyclin D1 in melanocytic nevi and melanomas and the expression of eIF-2α in colonic adenomas and carcinomas.
The authors found that the expression of eIF-2α was increased markedly in both benign and malignant neoplasms of melanocytes and colonic epithelium.
Stimulation of resting cells by growth factors leads to elevated levels and activity of translation initiation factors (e.g., eIF-2α and eIF-4E) and an increase in net protein synthesis, which is required for cell growth and division.1–6 Translation initiation factor eIF-2α associates with two other proteins, eIF-2β and eIF-2γ, to form an eIF-2 initiation factor complex responsible for the transfer of initiator methionine tRNA to the 40S ribosomal subunit. Whereas eIF-2β and eIF-2γ mediate methionine tRNA binding and transfer in a GTP dependent manner, the eIF-2α subunit plays a rate-limiting regulatory role, as its phosphorylation by interferon-inducible kinase (PKR) inactivates the eIF-2 complex.6, 7 Translation initiation factor eIF-4E is the least abundant and rate-limiting subunit of the eIF-4F complex.4, 5 The two other subunits are eIF-4A, which is a helicase responsible for unwinding secondary structures, and eIF-4G, which holds the complex together and is responsible for ribosome binding.4, 5 The function of eIF-4E is binding of the 5′ cap structure that is present in virtually all eukaryotic mRNA, thus providing for the association of mRNAs with the eIF-4F complex, eIF-4A-mediated disruption of the secondary structures in 5′-untranslated region, ribosome binding, and initiation of translation.4, 5
Expression of both eIF-2α and eIF-4E is increased by c-Myc, v-Srs, and v-Abl oncoproteins,8–11 Moreover, overexpression of either constitutively active eIF-2α12 or eIF-4E4 leads to malignant transformation of otherwise nontumorigenic rodent cells. Conversely, inactivation of eIF-2 by overexpressing PKR13 or down-regulation of eIF-4E by expression of exogenous antisense encoding vector14 leads to cessation of cell growth. Although experiments with cultured cells suggest that eIF-2α and eIF-4E function as mediators of oncoproteins in neoplastic transformation, their role in human neoplastic diseases has only begun to be elucidated (see reviews4, 11, 15–19). It has been demonstrated that the expression of eIF-4E is elevated in head and neck carcinomas,20 colon neoplasms,21 bronchioloalveolar carcinomas of the lung,22 and thyroid carcinomas.23 It is noteworthy that an increase in eIF-4E expression takes place early, as evidenced by its increase at benign stage (adenoma) in colonic tumor progression.21 In contrast, increases in both eIF-2α and eIF-4E expression have been correlated with the progression of thyroid epithelial tumors to more phenotypically malignant types.23 The expression of eIF-2α and eIF-4E in human melanocytic neoplasms and of eIF-2α in colonic epithelial tumors, however, has not been reported before. In the current article, we describe our results regarding the expression of eIF-2α and eIF-4E in melanocytic neoplasms and the expression of eIF-2α in colonic neoplasms.
MATERIALS AND METHODS
Archival tissues for this study were obtained from paraffin blocks in the files of the Department of Pathology, University of Massachusetts Medical School (Worcester, MA). We analyzed 22 adenomas and 25 adenocarcinomas of the colon and analyzed 9 melanomas (including 3 nodular melanomas, 4 superficial spreading melanomas, 1 melanoma in situ, and 1 spindle cell type melanoma), and 6 melanocytic nevi (including 3 dermal nevi, 2 compound nevi, and 1 dysplastic compound nevus). The tissue was fixed in 10% buffered formalin and routinely processed through a VIP Tissue Tek processor. Sections were cut at 4 microns, heated at 600 °C for 30 minutes, then deparaffinized and hydrated through a series of xylenes and alcohols. The sections for cyclin D1 analysis were antigen retrieved in 1 mM ethylenediamine tetraacetic acid, pH 8.0, for 5 minutes in an 800-watt microwave oven. After the replenishment of this solution, the slides were microwaved again for an additional 5 minutes and then allowed to cool for 20 minutes. The sections for eIF-4E and eIF-2α analysis were antigen retrieved in the same way, except the solution used was 0.01 M citrate buffer, pH 6.0. The histologic sections were stained on a TechMate 1000 automated immunostainer (Ventana Medical Systems, Tucson, AZ) using an avidin/biotin complex staining procedure. After a hydrogen peroxide block of endogenous peroxide and a serum blocking step, the slides were incubated with one of the primary antibodies: anti-eIF-4E monoclonal antibody (1:400 dilution; Transduction Laboratories, Lexington KY), anti-eIF-2α monoclonal antibody (1:2000 dilution; obtained from E. Henshaw, Rochester University, Rochester, NY), and monoclonal anticyclin D1 antibodies (1:40 dilution; Novocastra Laboratories, Newcastle Upon Tyne, United Kingdom) for 45 minutes. This was followed by brief buffer washes and incubation in a cocktail of biotinylated antimouse immunoglobulin G (IgG)/IgM and antirabbit IgG (Ventana Medical Systems) for 30 minutes. The sections were washed, incubated in avidin/biotin complex (Ventana Medical Systems) for 30 minutes, washed again, and reacted with diaminobenzidine and hydrogen peroxide. The sections then were counterstained with hematoxylin. As a technical control, a duplicate set of slides was stained by normal preimmune mouse serum instead of primary mouse monoclonal antibody and showed no reactivity.
We analyzed eIF-4E and eIF-2α expression in nine melanomas, including three nodular melanomas, four superficial spreading melanomas, one melanoma in situ, and one spindle cell type melanoma. The intensity of staining was compared between tumor cells (indicated by M on the figures) and neighboring normal skin (indicated by S on the figures) containing scattered melanocytes as a cell of tumor origin (examples of melanocytes are indicated by arrows in Fig. 1C,D). For all sections, we performed careful evaluation of numerous cells consistent with melanocytes in the epidermis. These cells did not show detectable expression of eIF-4E or eIF-2α. We also evaluated normal skin away from the tumor, and the results were the same. Figure 1 shows the results for a representative sample of nodular melanoma. In all samples of nodular melanoma, we found that eIF-2α expression always was increased markedly compared with the melanocytes in the adjacent basal layer (Fig. 1C). The expression of eIF-4E was weak but appeared stronger than in nonneoplastic melanocytes that displayed no detectable staining (Fig. 1C, arrow). We noted that keratinocytes showed a variable intensity of staining in different samples (compare Figs. 1–3). We believe that this was due largely to unspecific antibody binding to keratin. However, melanocytes, as stated above, did not show detectable staining for either eIF-2α or eIF-4E in all samples. It was found previously that the overexpression of translation initiation factor eIF-4E in NIH 3T3 cells led to a specific increase in the level of cyclin D1.24, 25 Therefore, we wanted to determine whether cyclin D1 expression was correlated with eIF-4E levels. Figure 1 shows that cyclin D1 expression was not detected in the neoplastic cells of nodular melanoma. The results are summarized in Table 1.
Table 1. Expression of eIF-4E, eIF-2α, and Cyclin D1 in Melanomas, Melanocytic Nevi, Colonic Adenomas, and Colonic Adenocarcinomas
The expression of eIF-4E always was increased markedly in colonic neoplasms. The immunohistochemical images of eIF-4E and cyclin D1 expression were published previously (see Rosenwald et al., 199921).
The intensity of staining was graded as weakly increased if the brown signal was only slightly stronger in neoplastic cells than in normal cells of tumor origin. A strong increase was defined as a marked increase in the intensity of the brown immunohistochemical signal, compared with weak or undetectable signal in normal cells of tumor origin.
The expression of eIF-2α always was strong in colonic and melanocytic neoplasms.
Figure 2 shows a representative sample of superficial spreading melanoma. The results on eIF-2α and eIF-4E expression were similar to the results from the samples of nodular melanoma. It can be seen in Figure 2 that eIF-2α was expressed strongly in tumor cells, whereas eIF-4E was expressed weakly or was undetectable in tumor cells. All samples of superficial spreading melanoma that we examined (Fig. 2B, Table 1) displayed strong nuclear expression of cyclin D1 in > 50% of tumor cells despite very little or no increase in eIF-4E expression (Fig. 2D). In contrast, in all other samples of melanoma (including nodular, in situ, and spindle cell melanomas; a representative sample is shown in Fig. 1B; for a summary of results, see Table 1), there was no increase in the proportion of cells that expressed cyclin D1. Therefore, it appears that increased cyclin D1 expression in melanocytic neoplasms does not depend on the up-regulation of eIF-4E (see Discussion).
Because our results indicated that the expression of eIF-2α was increased in melanomas, it was important to determine whether it was increased in benign melanocytic neoplasms. To address this question, we analyzed the expression of eIF-2α in melanocytic nevi (six samples in total, including three dermal nevi, two compound nevi, and one dysplastic compound nevus). Figure 3 demonstrates the results obtained in the dysplastic compound nevus. The results in other types of nevi were similar: The expression of eIF-2α was increased markedly compared with the cells of tumor origin (melanocytes), whereas eIF-4E was expressed weakly in the nevi and was not detectable in normal skin melanocytes (examples of melanocytes are indicated by arrows in Fig. 1C,D). It is noteworthy that there was no change in the expression of cyclin D1, which was undetectable in the overwhelming majority of cells in the nevi and in normal skin samples. The results are summarized in Table 1.
We reported previously that the expression of translation initiation factor eIF-4E was increased in human colonic adenomas and carcinomas.21 In the current report, we present data regarding the expression of eIF-2α in the same types of colonic neoplasms. We analyzed 22 adenomas and 25 adenocarcinomas (Table 1). In all samples, the expression of eIF-2α was increased markedly compared with nonneoplastic epithelium from the same sample. Representative samples are shown in Figure 4. Figure 4A,B (low-magnification and high-magnification views, respectively) shows that the expression of eIF-2α was increased markedly in colonic adenomas compared with normal epithelium seen in the same microscopic field. The transition from normal epithelium to adenoma was accompanied by a dramatic up-regulation of eIF-2α expression (indicated by arrows in Fig. 4B). The expression of eIF-2α in adenocarcinomas also was increased markedly (Fig. 4C). It is noteworthy that we never observed a difference in eIF-2α expression levels in normal epithelium adjacent to adenomas compared with normal epithelium distant from the tumor. However, in the adenocarcinoma samples, we found that nonneoplastic epithelium close to the tumor frequently displayed an adenoma-like appearance (less mucin, tall nuclei) and higher levels of eIF-2α compared with the epithelium farther from the tumor (Fig. 4D). The expression of eIF-4E also was up-regulated notably in the epithelium close to adenocarcinoma compared with the epithelium farther from the tumor; this up-regulation was not observed in the nonneoplastic epithelium adjacent to adenomas (data not shown). All of these changes appear to gradually become more apparent with decreasing distance from the carcinoma foci. These findings suggest that adenocarcinomas, unlike adenomas, secrete autocrine/paracrine growth factors that induce growth response (including increased expression of translation initiation factors) and elicit adenoma-like morphology in the adjacent epithelium.
The results of the current study demonstrate that expression of translation initiation factor eIF-2α is increased in melanomas and melanocytic nevi as well as in colonic adenomas and adenocarcinomas. It has been reported that eIF-2α is increased in gastrointestinal carcinomas.26 Our results indicate that an increase in eIF-2α expression occurs during development of benign melanocytic and colonic neoplasms. At the same time, the results clearly demonstrate that increased expression of eIF-2α alone is not sufficient for malignant transformation, because benign nevi rarely transform into melanoma yet display high expression levels of eIF-2α, which is comparable to the expression observed in melanoma cells (Figs. 1–3). Similarly, increased expression of eIF-2α in colonic epithelium is not sufficient for malignant transformation, because not all adenomas progress to adenocarcinomas, and those that do require many years to achieve a malignant phenotype. However, increased expression of eIF-2α may be a predisposing factor, provided that additional changes occur.
Our observations suggest that an increase in eIF-2α expression in nevi and melanomas is much more pronounced compared with minimal or no increase in eIF-4E expression (Figs. 1–3). In contrast, eIF-4E expression was increased markedly in both benign and malignant colonic epithelial neoplasms.21 It was found previously that eIF-4E increases the expression of cyclin D1 in NIH 3T3 cells.24, 25 Our current results in superficial spreading melanoma demonstrated that despite the lack of an increase in eIF-4E levels, there was a marked increase in nuclear cyclin D1 expression in all four samples examined (Fig. 2). Furthermore, there were low to undetectable levels of eIF-4E expression in both nodular and superficial spreading melanomas, whereas the proportion of cells that expressed cyclin D1 was increased strongly in superficial spreading melanomas (compare Fig. 1B with Fig. 2B). These results indicate that eIF-4E is not the only factor determining cyclin D1 expression in melanocytic tumors. These findings agree with those of our recent report describing a frequent increase of cyclin D1 expression in squamous carcinomas of the lung without a concomitant increase in eIF-4E, which reflects multiple levels of regulation of cyclin D1 expression.22 Although it has been demonstrated that the expression of eIF-4E frequently is elevated strongly in head and neck carcinomas,20 colonic neoplasms,21 bronchioloalveolar carcinomas of the lung,22 and thyroid carcinomas,23 we demonstrated in the current study that there may be no increase or only a small increase in the expression this factor in melanomas. These results are consistent with a report by others on the weak elevation of eIF-4E in melanoma cell lines compared with cultured melanocytes.27 It also was found that eIF-4A expression is elevated in melanoma cell lines compared with cultured normal melanocytes and cells of melanocytic nevi27 and that the suppression of eIF-4A by antisense oligonucleotide treatment decreases proliferation of melanoma cells in culture.28 Thus, a strong increase in eIF-2α probably is followed by additional events (including the up-regulation of eIF-4A) leading to malignant transformation. Regarding colonic neoplasms, it is clear that simultaneous up-regulation of both eIF-2α and eIF-4E is not sufficient to elicit a malignant transformation, because both factors were elevated at a benign stage (Fig. 4; Table 1).21 It is noteworthy that the increased expression of translation elongation factor 1γ also appears to take place at the adenoma stage during neoplastic transformation of colonic epithelium, because both adenomas and carcinomas displayed increased expression of this factor.29, 30 It remains to be determined which other components of translational machinery are up-regulated during the transition to a malignant phenotype in melanocytic and colonic malignant transformation.
Although the role of translation factors in human neoplastic disorders has only begun to be elucidated, the activity of ribosome biogenesis, in the form of the prominence and the number of nucleoli, has been used for a long time as a marker for malignant transformation in various tissues. Therefore, one of the events during malignant transformation may be the acquisition of a constitutively high rate of ribosome biogenesis. It is noteworthy that retinoblastoma protein (pRb) recently was found to negatively regulate the synthesis of ribosomal RNA.19 Functional inactivation of pRb may be due to constitutive phosphorylation by cyclin-dependent kinase (CDK-4) as a consequence of CDK-4 activation by mutation, a loss of CDK inhibitors p16 and p21,31, 32 or cyclin D1 overexpression.31 Increased availability of ribosomes coupled with accompanying elevation of eIF-2α and other translation factors may lead to a markedly increased rate of protein synthesis, thus accelerating cell growth and division. The importance of deregulation of translational machinery as a cause of neoplastic transformation is starting to be recognized,4, 11, 15–19 and targeting components of translational machinery as a strategy for antitumor therapy is discussed in the literature.33, 34 In conclusion, further studies are required to determine whether deregulated expression and function of translation factors plays a key role in human tumorigenesis and whether components of translational machinery can be used as helpful markers for diagnosis and as targets for therapy.