Relationship of the K-ras/c-mos expression patterns with angiogenesis in non-small cell lung carcinomas.

BACKGROUND
Neo-angiogenesis is an acquired capability vital for a tumor to grow and metastasize. Evidence has shown that the mitogen-activated protein (MAP) kinase pathway is involved in this process. Alterations of K-ras and c-mos, two pivotal components of this pathway, have been implicated in non-small cell lung carcinogenesis. In the present report, we examine, in a series of non-small cell lung carcinomas (NSCLCs), the status of K-ras and c-mos oncoproteins in correlation with the tumor neo-angiogenesis state and the major angiogenic factor, vascular endothelial growth factor (VEGF).


MATERIALS AND METHODS
c-mos and p-ERK1/2 status was evaluated immunohistochemically in a total of 65 NSCLCs, whereas the presence of K-ras mutations was examined by reverse transcriptase-polymerase chain reaction (RT-PCR) restriction fragment length polymorphism (RFLP) in available matched normal tumor material from 56 cases. Microvessel density (MVD) was estimated by immunodetection of CD3, endothelial marker, and VEGF expression was assessed by immunohistochemistry. All possible associations were examined by a series of statistical methods.


RESULTS
Expression of oncogenic activated K-ras and c-mos overexpression was observed in 12 of 49 (25%) and in 16 of 61 (26%) informative cases, respectively. Only 1 of the 25 deregulated for K-ras or c-mos cases exhibited both alterations, suggesting a mutually exclusive relationship between activated K-ras and c-mos overexpression (p = 0.074) in a subset of NSCLCs. In these cases, the MAPK kinase kinase/MEK/ERK pathway was found to be activated. MVD and VEGF expression were 36.9 +/- 10.6 mv/mm2 and 73.1 +/- 20.0%, respectively. The most intriguing finding was that the [K-ras(No)/c-mos(P)] profile was significantly associated with low MVD levels compared to normal cases (p = 0.004); by contrast, no correlation was found between the other K-ras/c-mos patterns and MVD. Furthermore, the former group exhibited the lowest VEGF levels.


CONCLUSIONS
The mutually exclusive relationship between mutated K-ras and c-mos overexpression in a subset of NSCLCs implies a common signal transduction pathway in lung carcinogenesis. The effect of this pathway on NSCLC neo-angiogenesis seems to depend upon the status of c-mos, which acts as a molecular "switch," possibly exerting a negative selective pressure on tumor progression.


Introduction
A vascular network is vital for a tumor to grow beyond a critical size and metastasize (1). Angiogenesis is a precisely regulated process, whereby many factors are implicated in a fine-tuned mechanism for the tissues to gain the appropriate blood supply for their survival (2). According to this scenario there are two major categories of modulators of angiogenesis: activators (e.g., VEGF-A, [angiopoietin 1] Ang1, basic fibroblast growth factor, transforming growth factor-␣[TGF-␣], hepatocyte growth factor, and tumor necrosis factor-␣[TNF-␣]), and inhibitors (e.g., thrombospondin-1, Ang2, angiostatin, and endostatin). The balance between these groups of factors determines the angiogenic potential (2,3). Under normal conditions, the angiogenic switch is in the "off" state Original Articles and leads to a vascular quiescence. Otherwise, when the angiogenic switch is "on," the balance is tipped in favor of angiogenesis (3).
Among these factors, the members of the VEGF family are specific for the vascular endothelium with crucial roles in the angiogenic process (4). VEGF acts on vascular epithelial cells as a mitogenic and motogenic factor, promoting the sprouting and formation of the new vessel-like structures (3,4). Its expression is induced in tumor cells by hypoxia, cytokines, and growth factors such as platelet-derived growth factor (PDGF) and transforming growth factor-␤(TGF-␤) (5). Recently, it has been found that VEGF expression in tumor cells is stimulated by the Raf/MEK/ERK MAP kinase cascade components src (6), v-raf and v-H-ras (7), K-ras, and MEK-1 (8, and references therein).
C-mos is an upstream activator of the MAPK pathway with serine-threonine kinase activity that activates MAPK kinase (MAPKK) (21). The role of c-mos is well established in oocyte maturation, where it is involved in oocyte arrest at metaphase II by stabilizing the maturation-promoting factor, and in the asymmetric division of the oocyte and the production of the first polar body by modulating the formation and orientation of the meiotic spindle (22). On the other hand, very little is known about its expression and functions in human somatic cells (23), even though its expression induces oncogenic transformation (22). C-mos causes meiotic-like modifications in M phase that lead to the production of binucleated cells and may indicate a novel mechanism of chromosomal instability (24). However, c-mos tumorigenic ability is mainly attributed to its mitogenic function as a stimulator of the MAPK pathway and the consequent activation of many downstream effectors including c-fos, c-jun, c-myc, S6 kinase II, and Elk-1 (25 and references therein). Among them, c-fos has been suggested to be the key molecule for the c-mos-mediated cellular transformation (26). However, the observed deregulation of certain cyclins (D, E, and A), cyclin-dependent kinases (p33 cdk2 and p43 cdk2 ), and S phase-specific E2F complexes in serum-starved v-mos-transformed cells adds a further complexity to the situation (27,28).
Mounting evidence suggests that the MAPKK/ MEK/ERK pathway may be involved in tumor neoangiogenesis (6)(7)(8). However, the data concerning the implication of defects of this pathway in nonsmall cell lung carcinoma (NSCLC) angiogenesis are limited and conflicting (29,30). In the present study, we investigated, in a series of NSCLCs, the relationship between aberrations of the MAPK components, K-ras and c-mos, and tumor neo-angiogenesis. To the best of our knowledge, this is the first report that addresses this issue.

Tissue Samples
A panel of 65 NSCLCs, consisting of 29 adenocarcinomas, 31 squamous cell carcinomas, and 5 undifferentiated large carcinomas, was analyzed immunohistochemically for c-mos, p-ERK1/2, VEGF, and neo-angiogenesis status. In 56 of these cases, frozen material and adjacent normal lung tissue was available for K-ras mutation analysis. These tumors were classified according to World Health Organization (WHO) criteria and Tumor-Nodes-Metastases (TNM) system, and a part of them has been previously employed in the study of c-mos status in NSCLCs (31) ( Table 1).

Controls
The human cervical cell line ME180 and the appropriate control peptide (Santa Cruz Bioanalytica) were used as positive controls for c-mos expression, while the DB lymphoma c-mos negative cell line was used as a negative control (31 and references therein). The specificity of E-4 anti p-ERK antibody was tested by preincubating the latter with the appropriate control peptide (Santa Cruz Bioanalytica). Elimination of the staining verified p-ERK specificity. For CD31 analysis, staining of vessel endothelial cells in the surrounding normal tissue was considered as positive control. As positive controls for VEGF analysis, normal kidney tissue sections recommended by the manufacturer were used. Furthermore, in each set of immunoreactions, antibody of the corresponding IgG fraction, but of unrelated specificity, was used as a negative control.
Evaluation c-mos Cytoplasmic and membranous immunoreactivity was considered to be evidence of c-mos expression. As we have previously described, nuclear staining in the presence of cytoplasmic and/or membranous reactivity was also evaluated as specific, whereas sole nuclear staining was disregarded (31,32). Immunohistochemistry was evaluated by examining all discreet areas of each tumor specimen. Tumors were considered c-mos positive when more than 30% of the tumor cells were stained; otherwise, they were scored as negative. The criteria were assessed in a recent study showing that carcinomas in which Ͼ30% cells exhibit immunoreactivity, are associated with increased c-mos mRNA levels (31).
p-ERK1/2 Tumor cells were evaluated as positive when staining was mainly nuclear because activated ERK translocates to the nucleus (31 and references therein). CD31 Hematoxylin-eosin staining was used to detect invasive areas of the tumor. In immunostained tissues, we defined microvessels as any single brownstaining endothelial cell or aligned clusters of endothelial cells, with or without a formed lumen, separated from adjacent microvessels, tumor cells, and other connective tissue material; plasma cells and T-cell subsets, which similarly stain positive for CD31, were clearly identified by their typical morphology. We note here that other groups exclude single immunohistochemically stained cells claiming, on one hand, that a lumen is necessary for them to be classified as vessels and on the other that single immunostained cells may not be of vascular origin.
VEGF Tumor cells were evaluated as positive when cytoplasmic staining was observed. The percentage of positive cells was counted. Evaluations were performed by three independent observers (P.Z., C.K., V.G.), and interobserver variability was minimal (p Ͻ 0.01).

RNA Extraction and cDNA Preparation
Cancerous material with more than 90% tumor cells was used for RNA extraction, because microdissection is not suitable for RNA handling methods. RNA was extracted with Trizol reagent (Life Technologies, AntiSel, Athens, Greece) according to the manufacturer's instructions. Two micrograms of total RNA were reverse-transcribed with oligo-dT (NEB, Bio-Line, Athens, Greece) and 200 U Moloney murine leukemia virus reverse transcriptase (Life Technologies, Anti-Sel), according to the supplier's instructions.

K-ras PCR-RFLP Mutation Detection
Codons 12 and 13 of K-ras were screened for mutations in our database. Briefly, BstNI artificial restriction fragment length polymorphic (RFLP) sites for detection of codon 12 mutations were introduced by nucleotide substitution after polymerase chain reaction (PCR) amplification of cDNA transcripts (33). The most common mutation at codon 13 (Gly to Asp) was detected by employing a naturally occurring RFLP (33). The cell lines SW480 (human colon carcinoma) and MDA-MB231 (human breast cancer) carry K-ras mutations at codons 12 and 13, respectively, and were used as positive controls for K-ras mutation analysis (33).

Statistical Analysis
The possible associations between K-ras and c-mos status independently, and (K-ras/c-mos) patterns with MVD, VEGF status, and clinicopathologic parameters were assessed with t-test, the nonparametric Kruskal-Wallis, and Pearson's chi-square tests. Moreover, analysis of variance (ANOVA) was used to evaluate more specifically the possible association between MVD and (K-ras/c-mos) patterns. Finally, Spearman bivariate analysis was performed to

K-ras, c-mos, and p-ERK1/2 Status: Relationship With Tumor Clinicopathologic Features
Oncogenic activated K-ras [K-ras(A)] (Fig. 1) and c-mos overexpression [c-mos(P)] (Fig. 2) was found in 12 of 49 (25%) and in 16 of 61 (26%) informative cases, respectively (Table 1). Among the 25 samples informative for both K-ras and c-mos status, alterations in K-ras or c-mos were observed in 11 and 13 cases, respectively, and only one case harbored both K-ras mutation and c-mos overexpression. In the latter cases, the MAPK/MEK/ERK pathway was found activated as evaluated by the p-ERK nuclear immunostaining. These observations indicate a mutually exclusive relationship between the presence of K-ras(A) and c-mos(P) in a subset of NSCLCs, which was further supported by a statistical trend (p ϭ 0.074, by Pearson's 2 test; group categorization in Table 2). No correlation was found between K-ras and c-mos status and clinicopathologic features ( Table 3).

Estimation of MVD in Tumor Areas: Relationship With VEGF Status and Clinicopathologic Parameters of the Patients
MVD ranged from 14-62 microvessels per square millimeter (mv/mm 2 ), with a mean value of 36.9 Ϯ 10.6 mv/mm 2 (Table 1 and Fig. 3). No correlation was found between MVD and any of the clinicopathologic features examined (e.g., histology, lymph node invasion, and tumor stage) ( Table 3).
VEGF expression varied from 30-100% (mean 73.1 Ϯ 20.0%) ( Table 1 and Fig. 4). No statistically significant association was found between VEGF estimation and any of the clinicopathologic parameters was found (Table 3). MVD positively correlates with VEGF expression (r ϭ .204), but this association does not reach significance (p ϭ 0.106, by Spearman test).

Relationship of K-ras and c-mos Status With MVD and VEGF Expression
K-ras mutational activation was not statistically associated with MVD or VEGF expression (Table 2). However, on closer inspection, samples without K-ras mutations [K-ras(No)] exhibited a larger variance in MVD (variance ϭ 124.1) compared to those with activated K-ras (variance ϭ 22.7), suggesting a more heterogeneous group (Fig. 5A) (p ϭ 0.03, by F-test).
An interesting finding was that cases with c-mos overexpression appeared to have lower MVD than those with normal c-mos status [c-mos(N)] (p ϭ 0.003, by t-test) ( Table 2 and Fig. 5B). Moreover, in the former group with c-mos overexpression, VEGF expression was found to be lower than that observed in the latter group (p ϭ 0.045 by Kruskal-Wallis test) ( Table 2).

Relationship of K-ras/c-mos Patterns With MVD, VEGF Expression, and Clinicopathologic Features
Because both K-ras and c-mos are activators of the MAPK pathway, samples were divided into two groups based on the complete absence (No) or presence of at least one aberration (Ab) in the aforementioned molecules. Statistical analysis revealed that the (Ab) group [K-ras-c-mos(Ab)] had lower MVD than that of the (No) one [K-ras-c-mos(No)] (p ϭ 0.012, by t-test). However, the levels of significance are lower than that of the association between c-mos status and MVD, thus excluding a putative synergistic effect of K-ras and c-mos on MVD (Table 2).
Cases were further categorized into three groups according to K-ras and c-mos status:  No correlation was found between K-ras-c-mos or K-ras/c-mos patterns and clinicopathologic features (Table 3).

Discussion
A discrete critical step for tumor development is to set the angiogenic switch to the "on" state (34). Although deregulation of oncogenes is implicated in the release of cells from controlled proliferation, programmed cell death, migration, and adhesion, little pattern was statistically associated with lower MVD, as compared to the normal status K-ras(No)/c-mos (N) (p ϭ 0.003) ( Table 2 and Fig. 6A). Moreover, Kruskal-Wallis analysis did not reveal any association between the examined patterns and VEGF expression (Table 2 and Fig. 6B).   is known about their involvement in tumor angiogenesis (8,34,35). Aberrant expression of oncogenic components of the MAPKK/MEK/ERK pathway, is known to be implicated in NSCL carcinogenesis (11,20,31,32). Limited data are available regarding the involvement of molecules participating in this signaling cascade in tumor neo-angiogenesis, as well as their role in the regulation of the pivotal angiogenic factor VEGF (7,8,29,30). In the present study, we investigated the relationship between alterations of the MAPK pathway components, K-ras and c-mos, with the angiogenic status in a series of NSCLCs. Examining the status of K-ras and c-mos, we observed that in almost half of the cases, analyzed for both molecules, a mutually exclusive relationship exists between mutationally activated K-ras and overexpressed c-mos. This finding suggests that K-ras and c-mos participate in the same MEK/ERK cascade in a subset of NSCLCs. This notion is further strengthened by the fact that all these cases exhibited nuclear immunoreactivity for activated ERK1/2, as previously demonstrated in our database (31).

MVD (microvessels/mm 2 ) VEGF (%)
Our analysis concerning the relationship between K-ras/c-mos patterns and MVD status revealed several intriguing findings. The carcinomas with the K-ras(No)/c-mos(P) profile exhibited a significantly lower MVD compared to cases with normal K-ras/ c-mos pattern (p ϭ 0.004, Table 2), whereas the latter profile demonstrated similar MVD values with the K-ras(A)/c-mos(N) one. A question arising from these observations is how do they fit in the currently proposed angiogenesis model (2,36). According to this model, the following steps are required for angiogenesis: (1) proteolytic degradation of the parental vessel membrane and surrounding matrix, (2) migration of endothelial cells, (3) proliferation of endothelial cells, (4) formation of capillary tubes, and (5) stabilization of vessels by recruiting periendothelial cells (3,36). The role of VEGF in these steps is vital for the maturation and stabilization of the newly formed vessels, because its absence leads to vessel regression (36). Furthermore, an overwhelming body of evidence implicates members of the MAP kinase kinase/MEK/ERK module in these processes (6)(7)(8)(37)(38)(39). In light of these considerations, our results concerning the relationship between the K-ras/c-mos patterns and MVD have complicated the issue.
One possible explanation is that the K-ras(No)/ c-mos(P) pattern is related to vessel regression, which acts in a particular stage of tumor development as a negative selective-pressure mechanism, by creating a hypoxic environment. In accordance with this hypothesis, the K-ras(No)/c-mos(P) carcinomas exhibited the lowest VEGF levels ( Table 2, Fig. 6B). Additionally, we have recently observed at the ultrastructural level disruption of the endothelium integrity and absence of the tunica externa in vessels next to c-mos-positive tumor cells (data not shown). However, how can the higher MVD and VEGF levels in the other K-ras/c-mos patterns be explained, in as much as K-ras and c-mos act in the same pathway?
The most likely answer lies on the effect K-ras and c-mos exert on VEGF expression or other angiogenic factors. It should be mentioned to this end that VEGF is triggered by a variety of growth factor-regulated    Cases with mutated K-ras and/or stimulated growth factor (GF) autocrine loops (A and B, respectively) activate signaling cascades, such as MEK/ERK and PI3K pathways (8, and references therein), that lead to high VEGF levels and increased neo-angiogenesis. On the other hand, overexpression of c-mos regulates negatively the growth factor-dependent signaling cascades diminishing VEGF expression and tumor vascularization (42) (C).