14-3-3σ Expression Is an Independent Prognostic Parameter for Poor Survival in Colorectal Carcinoma Patients

14-3-3σ (also called stratifin) belongs to a family of abundant, widely expressed 28 to 33 kDa acidic polypeptides that spontaneously self-assemble as homodimers or heterodimers. There are seven closely related genes, encoding β, ε, η, γ, τ, ζ, and σ isoforms, that are conserved across mammalian species. They can bind to >100 functionally diverse cellular proteins and thereby play an important role in various cellular processes like signal transduction, cell cycle regulation, apoptosis, cytoskeleton organization, and malignant transformation (for a recent review, see refs. 1–3). 14-3-3 proteins contain no catalytic domain or function and a number of studies showed that binding of 14-3-3 requires target protein phosphorylation. Each monomer within the dimeric 14-3-3 molecule binds substrates through a binding cleft that preferentially recognizes proteins containing the optimal phosphoserine/phosphothreonine sequence motifs RSx[pS/pT]xP and RxΦx[pS/pT]xP (where uppercase bold characters denote the most highly conserved positions, the prefix p denotes a phosphorylated amino acid, Φ denotes aromatic or aliphatic amino acids and x represents any amino acid; refs. 4, 5). In addition, 14-3-3 proteins regulate enzyme activity and may act as localization anchors, controlling the subcellular localization of proteins. Furthermore, 14-3-3 proteins can function as adaptors or scaffolds stimulating protein-protein interactions (6).

14-3-3σ was originally identified as a p53 inducible gene that is responsive to DNA-damaging agents (7). 14-3-3σ sequesters the mitotic initiation complex, cdc2-cyclin B1, from entering the nucleus, thus preventing initiation of mitosis (8). In this manner, 14-3-3σ induces G₂ arrest and allows the repair of damaged DNA. Furthermore, 14-3-3σ can bind cyclin-dependent kinase-2 and cyclin-dependent kinase-4 (CDK2, CDK4) and thereby block the transition of the eukaryotic cell cycle (9). These findings define 14-3-3σ as a negative regulator of cell cycle progression.

14-3-3σ expression is lost in ∼95% of breast carcinomas by methylation-mediated scilencing (10). Additionally, it has been shown that hypermethylation of 14-3-3σ occurs at an early stage in the progression of invasive breast cancer (11). Hypermethylation of CpG islands and loss of 14-3-3σ expression was also frequently found in human hepatocellular carcinoma (12), oral carcinoma (13), lung cancer (14), and prostate cancer (15). In colorectal carcinoma, however, Suzuki et al. (16) found that the CpG island of 14-3-3σ was unmethylated. Very recently, it was shown that hypermethylation of 14-3-3σ promoter is a rare event in colorectal carcinoma (17). The colorectal carcinoma cell lines HCT116, Lovo, and Lim2405 were found to express 14-3-3σ RNA (7). However, proteomic analysis revealed down-regulation of 14-3-3σ in human colon adenoma (18). Given the importance of 14-3-3σ silencing in cancer progression, only few data are available on 14-3-3σ expression in colorectal carcinoma.

Here, we analyzed the prognostic significance of 14-3-3σ expression in colorectal carcinoma compared with common clinicopathologic parameters in 121 patients with a long follow-up period, including univariate and multivariate analysis.

Reagents and chemicals. Unless otherwise indicated, all reagents were purchased from Sigma (Vienna, Austria). Cell culture reagents (media, serum, and antibiotics) were obtained from PAA Laboratories (Linz, Austria). Murine anti–14-3-3σ monoclonal antibody (mAb) was obtained from Neomarkers (Fremont, CA). Mouse anti–α-tubulin mAb was obtained from Sigma and the sheep anti-mouse horseradish peroxidase (HRP)-IgG was purchased from Amersham Pharmacia (Freiburg, Germany).

Cell culture, cell proliferation. Experiments were conducted with established human colon cancer cell lines (HRT-18, Caco-2, Lovo, DLD-1, SW480, CX-1, HCT-15, and HT-29) obtained from DSMZ (Darmstadt, Germany) and cultivated as suggested by the supplier with 10% FCS and 1% penicillin/streptomycin. New cultures were reestablished from frozen stocks every 3 months. The HT-29F cell line, which is a subclone of HT29 resistant to high concentrations of 5-fluorouracil (5-FU; ref. 19), was obtained from Dr. H. Hendriks (Faculty of Veterinary Medicine, Department of Pathology, Utrecht University, Utrecht, the Netherlands). Determination of IC₅₀ of 5-FU exponentially growing cells of cell line HT29, and HT-29F were seeded at 7 × 10⁵ cells per flask in the presence of increasing concentrations of the drug. Cell growth was assessed after 3 days by cell counting in a CASY (Schärfe Systems, Reutlingen, Germany) and IC₅₀ values were determined by nonlinear curve fitting of data according to the general dose-response curve.

Reverse transcription-PCR analysis. Genomic DNA free RNA was isolated with the SV total RNA isolation system (Promega, Madison, WI) according to the manufacturer's protocol. Reverse transcription-PCR (RT-PCR) was done as described earlier (20). The following 14-3-3σ–specific primers was used: forward 5′-AGAGCGAAACCTGCTCTCAG-3′ and reverse 5′-GGCATCTCCTTCTTGCTGAT-3′. For control, glyceraldehyde 3-phosphodehydrogenase cDNA was amplified. RT-PCR without reverse transcriptase was run in parallel to verify no genomic DNA contamination. All PCR products were visualized by electrophoresis and ethidium bromide staining in 2% agarose gels.

Immunoblot analysis. Cells were detached by trypsinization and washed in PBS. Lysates were prepared in ice-cold cell lysis buffer (21). Tissue samples were homogenized, centrifuged, and the supernatant was taken for immunoblots. Equivalent protein concentrations of 30 μg were resolved in 12% SDS polyacrylamide gels on a Minigel apparatus (Novex, Groningen, the Netherlands) and transferred to a nitrocellulose membrane (Schleicher&Schuell, Einbeck, Germany). Membranes were incubated with the primary antibodies: anti–14-3-3σ (1:200) and mouse anti–α-tubulin (1:2,000). After the washing steps, membranes were incubated with IgG-HRP sheep anti-mouse (1:1,000). Protein-antibody complexes were detected by enhanced chemiluminescence (Amersham Pharmacia).

Patients and tissues. Formalin-fixed and paraffin-embedded samples from colorectal cancer (n = 121) were obtained from Department of Pathology, University of Innsbruck, Austria, from 1994 to 2003. Clinical and pathologic data were documented prospectively and entered into a specific tumor registry after surgery and after follow-up (Table 1). Each case was classified according to pT stage and nodal status (pTNM system), histologic tumor type, tumor differentiation (grades 1-3), tumor localization (right colon, left colon, rectum), and tumor size (<2, 2-5, and >5 cm).

Immunohistochemistry. 14-3-3σ expression was determined by immunohistochemistry using a murine monoclonal antibody (Neomarkers). Four-micrometer sections were cut from paraffin block, deparaffinized, rehydrated, and put into endogenous peroxidase-blocking solution. The sections were then boiled (86°C) for 15 minutes in 10 mmol/L citrate-buffer (pH 6.0) in a water bath. Thereafter, slides were incubated with the primary antibody (1:100) at 4°C overnight. After washing in TBS, the slides were incubated in biotinylated rabbit anti-mouse IgG (DAKO, Copenhagen, Denmark) at a dilution of 1:500, and detected with an ABC-peroxidase kit (Vector Laboratories, Bulingame, CA) and diaminobezidine as substrate. The tissue sections were washed in water, counterstained with Mayer's hematoxylin (Merck, Darmstadt, Germany), and coverslipped. Negative controls were run in parallel.

Evaluation of 14-3-3σ expression. Slides of cancer specimens were analyzed in parallel by two independent pathologists (P. Obrist and S. Stadlmann) who had no prior knowledge of the clinical data. A 14-3-3σ immunostaining score was calculated on the basis of a proportion score and an intensity score as described previously (22). The proportion score reflects the estimated percentage of positively stained tumor cells (score 1, 0-10%; score 2, 10-50%; score 3, 51-80%; score 4, >80%). The intensity score represents the estimated staining intensity (score 0, no staining; score 1, weak; score 2, moderate; score 3, strong). 14-3-3σ overexpression was defined for tumors that show both a proportion score ≥3 and an intensity score ≥2.

Statistical analysis. All statistical analyses were done using SPSS 11.0 for Windows (SPSS, Inc., Chicago, IL). To trace correlations between 14-3-3σ expression and several clinicopathologic parameters, data were cross-tabulated and Fisher exact test was done. The correlation of 14-3-3σ staining with patient survival was evaluated using life tables constructed from survival data with Kaplan-Meier plots. Comparison of the different groups were done with the log-rank test. The end point in the present study was overall survival ranging from date of surgery until date of death or, if no information was documented, until date of last follow up information (=censored). Multivariate survival analysis was carried out on samples where all clinical parameters were available (n = 107) using the Cox proportional hazard model to evaluate the independent power of each variable. Results were considered significant when P < 0.05.

Expression of 14-3-3σ in colorectal carcinoma cell lines and tissues. To evaluate the expression of 14-3-3σ in colorectal carcinoma RT-PCR and Western blots of nine human colorectal carcinoma cell lines and Western blots of eight matched pairs of carcinoma and normal mucosa were done. RT-PCR revealed expression of 14-3-3σ in all nine colorectal carcinoma cell lines (Fig. 1A). Western blots determined a strong 14-3-3σ expression in CX-1, DLD-1, Caco-2, HT-29F, and Lovo cells; however, weak expression in HRT-18 and SW480 cells and no expression in HCT-15 and HT-29 cells were observed (Fig. 1B). In six tissue samples, as shown in Fig. 1B, 14-3-3σ expression was higher in carcinoma than in matched normal mucosa. In two samples (samples 5 and 6), 14-3-3σ expression was nearly similar in mucosa and carcinoma tissue (Fig. 1C). These results show differential 14-3-3σ expression between colorectal carcinoma cell lines on protein level but not on mRNA level. Furthermore, a significantly higher 14-3-3σ expression in carcinomas compared with normal mucosa was shown.

Expression of 14-3-3σ in colorectal carcinoma tissue. To distinguish which cell type expresses 14-3-3σ in tissue samples, we did immunohistochemical analysis of 121 colorectal carcinoma tissues. One hundred sixteen (95.9%) of 121 patients with colorectal cancer showed a positive staining reaction for 14-3-3σ, whereas 5 (4.1%) specimens were completely 14-3-3σ negative (Fig. 2A). The atypical gland cxells showed membranous and cytoplasmic staining, which is most pronounced at the luminal surface (Fig. 2C′ and D′). The staining intensity was weak in 40 (34.5%), moderate in 43 (35.5%), and strong in 33 patients (27.3%; Fig. 2B-D). The mean percentage of stained tumor cells was 38.5 ± 17.8% (range 0-90%). Thirty-one specimens (25.6%) showed an area ≥50% with 14-3-3σ–positive tumor cells. The distribution of the percentage score (1-4; see Materials and Methods) was 8.3%, 71.9%, 18.2%, and 1.7%, respectively. The combination of staining intensity and percentage of stained tumor cells revealed 14-3-3σ overexpression in 47 cases (38.8%; for definition, see Materials and Methods).

Clinicopathologic parameters and expression of 14-3-3σ are illustrated in Table 1. Univariate analysis revealed a significant correlation of 14-3-3σ overexpression with tumor differentiation (P = 0.001) and tumor stage (P = 0.003). A borderline association was found with increasing tumor size (P = 0.084). Colorectal carcinomas with small tumor size (<2 cm) had a lower percentage of 14-3-3σ overexpression (15.4%) when compared with carcinomas >5 cm (51.6%). The percentage of 14-3-3σ overexpression in colorectal carcinoma increases with tumor grade: Overexpression was observed in 13.8% of specimens with grade 1, 43.3% with grade 2, and 77.8% with grade 3. There was no significant correlation between the level of 14-3-3σ expression and other clinical and pathologic features, including sex, age, tumor localization, nodal status, and liver metastasis (Table 1). Survival analysis according to the Kaplan-Meier method revealed a significantly lower overall survival for patients with 14-3-3σ–overexpressing carcinomas (P = 0.0096; Fig. 3). Within this group, the 5-year survival rate was 56.2% compared with 73.6% in 14-3-3σ negative or low positive tumors. The median overall survival was 65 months (14-3-3σ overexpression) and not reached (14-3-3σ low expression), respectively. By multivariate analysis, 14-3-3σ overexpression was an independent prognostic marker for overall survival (P = 0.037). Of all clinicopathologic features, the presence of liver metastases was the most independent parameter predicting prognosis (P = 0.006; Table 2).

Cell proliferation. 5-FU is still one of the most widely used agents in the first line therapy of colorectal carcinoma; therefore, we examined the effects of 5-FU on the growth of two colorectal carcinoma cell lines, which express 14-3-3σ (HT-29F) or not (HT-29). HT-29F is a subclone of HT29 cell line resistant to high concentration of 5-FU. The mean calculated antiproliferative IC₅₀ values for 5-FU were 10.2 ± 0.31 μmol/L for HT-29F and 2.20 ± 0.18 μmol/L for HT-29, respectively. The results showed a significantly lower IC₅₀ in the 14-3-3σ–negative than in the 14-3-3σ–positive cell line.

Among the seven distinct 14-3-3 genes, 14-3-3σ has been directly implicated in the etiology of human cancer (6). Recent studies have shown that 14-3-3σ is selectively lost by epigenetic hypermethylation of the gene promotor in breast cancer. In fact, loss of 14-3-3σ by gene methylation is reported to be the most consistent molecular alteration thus far discovered in breast cancer (10, 11). In contrast, methylation of 14-3-3σ was never detected in breast epithelial tissue derived from cancer-free patients. A second mechanism of 14-3-3σ loss in breast cancer is its targeting for ubiquitin-mediated proteolysis by Efp, an estrogen-dependent E3 ubiquitin ligase (23). 14-3-3σ inactivation by epigenetic silencing was also detected in a number of other cancer types, such as small cell lung cancer (14), ovarian cancer (24), hepatocellular carcinoma (12), prostate cancer (15), and gastric cancer (16). Furthermore, silencing of 14-3-3σ by methylation was detected in vulval and oral squamous cell carcinoma (13, 25) and in basal cell carcinomas (26). These data suggest that 14-3-3σ acts as a tumor suppressor and that its inactivation contributes to tumor progression.

In contrast to these findings, our results clearly show that 14-3-3σ is overexpressed in a subset of colorectal cancers and predicts poor overall survival in this patient group. Furthermore, 14-3-3σ overexpression was mainly found in dedifferentiated carcinomas and in more advanced tumor stages, indicating that 14-3-3σ is involved in colorectal carcinoma progression. In human colorectal carcinoma cells, 14-3-3σ is a p53 regulated gene and can induce cell cycle arrest at G₂-M phase by a mechanism that involves its binding to and sequestering CDK1-cyclin B1 complexes in the cytoplasm (7). Experimentally induced loss of both 14-3-3σ alleles resulted in failure to maintain the G₂ checkpoint after DNA damage, causing mitotic catastrophe as the cell entered mitosis (8). Our data suggest that 14-3-3σ expression is not solely dependent on intact p53 activity because most of the cell lines that express 14-3-3σ are p53 mutated (DLD-1, CX-1, Caco-2, SW480), whereas other cell lines are 14-3-3σ negative and also p53 mutated (HCT-15, HT-29, HT-29F). The p53 status of the Lovo cell line was wild type, whereas the status of HRT-18 (identical to HCT-8) cells is controversially reported (p53 status of different cell lines was obtained from recent literature; refs. 27, 28).

According to present and previous studies, epigenetic silencing of 14-3-3σ seems to be not involved in colorectal cancer because studies from Suzuki et al. (16) and Ide et al. (17) showed no CpG methylation of the 14-3-3σ promoter in colorectal carcinoma cell lines and carcinomas. Using RT-PCR, we found 14-3-3σ mRNA expression in all nine colorectal carcinoma cell lines. Similar results were also found in previous studies (7, 17). However, 14-3-3σ protein was differentially expressed in the analyzed colorectal carcinoma cell lines, suggesting the involvement of posttranscriptional regulatory mechanisms. Modifications of translational efficiency are increasingly reported in cancer and are controlled through a complex network of RNA/protein interactions involving recognition of mRNAs by RNA-binding proteins (29). It is tempting to suggest that a different composition of RNA/protein complexes in colorectal carcinoma cell lines might control translational efficiency of 14-3-3σ mRNA and account for the observed differences of 14-3-3σ expression in colorectal carcinoma cell lines. However, it is not known whether 14-3-3σ expression in colorectal carcinoma results from increased translational efficiency.

The specific inhibition of 14-3-3σ expression in various cancer types, but not in colorectal carcinoma, suggests a different role of 14-3-3σ in the transformation process in colorectal carcinoma than in other cancers. Cancer most commonly arises via two major abnormalities, which are uncontrolled cellular proliferation and lack of adequate cell death. 14-3-3σ overexpression might be involved in establishing of both abnormalities. Very recently, Zhang et al. showed that insulin-like growth factor-I increased 14-3-3σ expression and cell cycle progression in MCF7 breast cancer cells (30). These effects of insulin-like growth factor-I were p53 independent. Using small interfering RNA to reduce 14-3-3σ mRNA levels resulted in decrease of insulin-like growth factor-I–induced cell proliferation, demonstrating, at least in MCF7 cells, a positive link between 14-3-3σ expression and growth factor–induced cell cycle progression (30). In colorectal carcinoma, Ide et al. (17) provided evidence for a higher proliferative activity at the invasive front in 14-3-3σ–positive carcinomas than in 14-3-3σ–negative cases. These data suggest a possible growth advantage of 14-3-3σ–expressing cancer cells at least under specific growth conditions.

However, 14-3-3σ might also be involved in cell death regulation because 14-3-3σ was shown to prevent apoptosis in colorectal carcinoma cell lines through sequestration of BAX, a member of the bcl-2 family of apoptosis-regulating proteins, to the cytosol (31). Unbound BAX associates with mitochondria, mediates mitochondrial permeability transition, and subsequently results in cytochrome c release and apoptosis. Lack of 14-3-3σ in colorectal cancer cells sensitizes them to chemotherapy-induced apoptosis (31). By analysis of cell growth, we showed a positive correlation between 14-3-3σ expression and 5-FU sensitivity of colorectal carcinoma cells. Studies have correlated BAX and thymidylate synthase expression to 5-FU resistance of cancer cells (32–34). By Western blot analysis, both proteins were equally expressed in HT-29 and HT-29F cell lines, suggesting that these proteins are not involved in the different sensitivity to 5-FU (data not shown). However, other proteins might be also involved in chemoresistance and, therefore, the suggested 14-3-3σ–mediated resistance to 5-FU chemotherapy remains to be proven.

Other functions of 14-3-3σ cannot be excluded because the different 14-3-3 monomers bind tightly to each other to form homodimers and mixed heterodimers and, therefore, display significant functional redundancy in vivo and in vitro (2). Except 14-3-3η and 14-3-3γ, all other members of the 14-3-3 protein family are expressed in all nine colorectal carcinoma cell lines (data not shown). Therefore, 14-3-3σ monomers may interact with other monomers of the 14-3-3 protein family to form functional heterodimers and activate tumor-promoting pathways. Finally, our results show overexpression of 14-3-3σ in ∼40% of colorectal cancer patients and predicts poor prognosis for these patients. Therefore, 14-3-3σ might be a potentially new target in colorectal carcinoma therapy.

We thank Martin Heitz for excellent technical assistance.