Legumain Expression in Relation to Clinicopathologic and Biological Variables in Colorectal Cancer

Legumain is a cysteine endopeptidase, displaying high specificity for hydrolysis of asparaginyl bonds; hence, it has also been called as a novel asparaginyl endopeptidase. This lysosomal protease belongs to the peptidase family C13. The gene is located on the chromosome 14q32.1 and encodes 433 amino acids (1, 2). Legumain was found to be highly expressed in several types of tumors, such as colon, prostate and breast cancers. In addition, it is found highly expressed in vivo in the murine colon carcinoma model. Its expression was observed to be up-regulated during tumor development in vivo, suggesting an environmental response, and seems to be a stress-responsive gene, being markedly elevated in cells subjected to environmental stress (2). Legumain is expressed both intracellular and on the cell surface by tumor cells and tumor associated endothelial cells, where it is colocalized with integrins. It is found especially in membrane-associated vesicles concentrated at the invadopodia of tumor cells. Notably, cell surface proteases are often associated with invasive and metastatic tumor cells. Cells that highly express legumain exhibit enhanced migratory and invasive properties. These properties may be mediated by increased extracellular matrix degradation, resulting from activation of zymogens such as progelatinase A. Legumain activates the gelatinase A zymogen, an important mediator of extracellular matrix degradation, and thus may be important for tumor cells to adapt a more invasive and metastatic phenotype. Animal tumor models generated with cells overexpressing legumain showed an in vivo behavior that is vigorous with more increased invasive growth and metastasis (2). This phenotype is proposed to result from the proteolytic function of legumain that results in activation of other protease zymogens. Some proteases are linked to other properties of tumors such as angiogenesis and growth signaling and may be activated by legumain. Protease cascades are characteristic of many biological pathways such as the coagulation, apoptosis, and complement cascades (2). Legumain may represent a target for inhibition of tumor growth and metastasis based on its enhancement of tumor growth and its unique restricted specificity.

These studies show that increased expression of legumain may play a role in promoting tumor progression and suggest that tumors that express high levels of legumain should be expected to display a more aggressive behavior and have a poor prognosis. The aim of the present study was to test this hypothesis by examining legumian expression in primary colorectal cancers, distant, adjacent normal mucosa, and metastases, and determine the relationships between legumain expression and various clinicopathologic and biological variables.

Materials. The tissue sections used for immunohistochemical staining were obtained from 164 patients with primary colorectal adenocarcinoma who underwent surgical resection at Linköping Hospital, Linköping and Vrinnevi Hospital, Norrköping, Sweden. Our study also included 34 corresponding normal mucosa specimens taken from the margin of distant resection (distant normal mucosa), which were histologically free from pretumor and tumor, 89 adjacent normal mucosa samples (the normal mucosa adjacent to primary tumor) and 33 corresponding metastases in the regional lymph nodes. The patient's gender, age, tumor location, differentiation, and Dukes' stage were obtained from surgical and pathologic records from the hospitals. The mean age was 71 years (range, 34-94). The growth pattern was based on the patterns of growth and invasiveness. Differentiation was graded as good, moderate, poor, or mucinous carcinoma including signet-ring cell carcinoma. The extent of necrosis was classified as <10% or >10%. All the patients were followed up until the end of 2003, the mean follow-up time was 75 months and the median was 50 months. By that time, 77 patients died of colorectal cancer. The data on proliferating cell nuclear antigen (PCNA) (3) and p53 expression (4), stromelysin-3 (5), PINCH expression (6) determined by immunohistochemistry, and the information about apoptosis (7) obtained by terminal deoxynucleotidyl transferase–mediated nick end labeling assay were taken from previous studies carried out in our laboratory.

Western blotting analysis was carried out on three additional colorectal adenocarcinomas and three colonic cell lines, KM12C, KM12SM, and KM12L4a (kindly provided by Prof. I.J. Fidler, M.D. Anderson Cancer Center, Houston, TX).

Immunohistochemical Staining. Immunohistochemical staining was done on 5-μm-thick formalin-fixed, paraffin-embedded sections. The sections were incubated at 60°C for 12 hours, deparaffinized in xylene, and rehydrated with a series of ethanols. The sections were transferred to 0.01 mol/L Tris-EDTA buffer (pH 9.0) and subjected to high pressure cooker, for 8 minutes, and the sections incubated at room temperature for 30 minutes, for antigen retrieval. The sections were then rinsed in PBS (pH 7.4) and incubated with 3% H₂O₂ in methanol, which blocks endogenous peroxidase activity, for 20 minutes. Nonspecific binding of antibody was prevented by preincubating the sections with serum-free DAKO Protein Block (DAKO Co., Carpinteria, CA) for 10 minutes. After removing the blocking solution, the sections were incubated with rabbit polyclonal anti-legumain antibody (obtained kindly from Prof. G. David Roodman, University of Pittsburgh Cancer Institute, Pittsburgh, PA) in 1:500 dilution, diluted in antibody diluent (DAKO) at room temperature for 1 hour. Subsequently, the sections were incubated with DAKOPATTS swine anti-rabbit immunoglobulins (DakoCytomation, Glostrup, Denmark) for 25 minutes. The sections were washed with PBS between each incubation step. The peroxidase reaction was done for 8 minutes, using DakoCytomation ChemMate EnVision Detection Kit (DakoCytomation) by mixing 50 parts of substrate buffer (containing hydrogen peroxide and preservative) with one part of 3,3′-diaminobenzidine chromogen (containing 0.05% 3,3′-diaminobenzidine tetrahydrochloride in organic solvent). Then the sections were washed by water, followed by counterstaining with hematoxylin.

The sections known to stain positively were included as positive and negative controls. For the negative controls, the sections incubated with PBS instead of the primary antibody were not stained, whereas the positive controls were stained.

The stained sections were microscopically examined and scored independently by two of the authors without any clinicopathologic information. To avoid artifacts, tissue in the areas with poor morphology, necrosis and in the margins of the sections were not considered. Among the 164 cases evaluated, there was disconsonance in 32 sections in the first round of evaluation. These sections were reread by the two authors individually and matched. The final five discrepant sections were reexamined by dual-microscope and concurrent score was achieved. The staining intensity was classified as negative, weak, moderate, and strong, based on the staining intensity of the cytoplasm of normal epithelial, tumor, and stromal cells, irrespective of the percentage of positive cells. Considering the similarities of the clinicopathologic features, the cases with negative and weakly stained were grouped as weakly stained group, and the cases with moderate and strong staining as strongly stained group. The staining percentage of positive cells was classified as 0, 0 to <5%, 5 to <25%, 25 to <50%, 50 to <75%, and >75%, irrespective of the staining intensity. Although most of the cases (80%) had >75% staining, we cannot further subgroup these cases since they usually showed all the staining (i.e., nearly 100%). Importantly, it seemed that the cases with >75% expression had different clinicopathologic features from those cases with <75% expression. Therefore, we used 75% as a cutoff point in our statistical analyses. We did not examine the staining percentage in distant normal mucosa, adjacent normal mucosa and metastasis, due to the small stained areas.

Western Blotting. The protein samples used for blotting were from cell lines and colorectal cancers. The samples were thawed on ice and homogenized with a sonicator in the presence of radioimmunoprecipitation assay buffer lysis buffer [150 mmol/L NaCl, 1% Triton X-100, 0.5% NaDOD, 0.1% SDS, and 50 mmol/L Tris (pH 8.0)] maintaining the temperature at 4°C throughout the procedure. Then the samples were centrifuged at 4°C, at 12,000 × g for 5 minutes and the supernatant containing protein was collected. The protein concentrations of the lysates were determined using bicinchoninic acid protein assay reagent (Pierce, Rockford, IL). The lysate samples were mixed thoroughly in Eppendorf tubes with an equal volume of 2× electrophoresis sample buffer, containing β-mercaptoethanol and boiled in thermomixer at 99°C for 5 minutes at 300 rpm. Denatured protein lysates were centrifuged and resolved using gradient (4-20%) Tris-Hcl gels (Bio-Rad, Hercules, CA) placed in an electrophoresis unit containing electrophoresis buffer (resolving gel buffer) at 200 V, 350 mA, for 1 hour, with suitable marker. The separated proteins were immobilized onto polyvinylidene fluoride membranes (Hybond-P polyvinylidene fluoride membrane, Amersham Biosciences, Buckinghamshire, England) assembled in an electroblotting cassette placed between the electrodes in the blotting unit containing transfer buffer, at 100 V, 350 mA, for 1 hour, with cooling. The polyvinylidene fluoride membranes were washed with PBS-Tween 20 for 5 minutes with five changes and blocked with blocking buffer (5% nonfat dry milk solution) and incubated for 2 hours on an orbital shaker, at room temperature. And then the membranes were reacted with the rabbit polyclonal anti-legumain antibody, in 1:1,000 dilution over night at 4°C. This was followed by incubation with peroxidase conjugated goat anti-rabbit immunoglobulins (1: 5,000, DakoCytomation) in 3% nonfat dry milk, for 2 hours on an orbital shaker at room temperature. The membranes were washed with PBS-Tween 20 between each incubation step, for 5 minutes with five changes on an orbital shaker at room temperature. Then the membranes were incubated with the Enhanced Chemiluminescence Plus detection reagent for 5 minutes, protected from light. The membranes were developed, using a developer, in dark by exposing it onto autoradiography film: Hyperfilm Enhanced Chemiluminescence (Amersham Biosciences) for suitable durations of time.

Legumain protein migrates in sodium dodecyl polyacrylamide gel as a 50-kDa monomer. Reaction of the legumain antibody with the protein resulted in the bands, which could be detected against the standard marker used (Fig. 1).

Statistical Analysis. The significance of the difference in intensity of legumain expression among normal mucosa samples, primary tumors, and metastases was tested by McNemar's or χ² statistical methods. The relationship between legumain expression and clinicopathologic/biological factors was examined by the χ² method. The relationship between legumain expression and survival, including both univariate and multivariate analyses, was tested using Cox's proportional hazard model by using STATISTICA 6.0 program. The odds ratio, an estimate of the relative risk, with 95% confidence intervals was computed to assess the relationship of the legumain expression to the risk of colorectal cancer. Survival curves were calculated using the Kaplan-Meier method. Two-sided Ps < 5% were considered as statistically significant.

Legumain Expression in Distant Normal Mucosa, Adjacent Normal Mucosa, Primary Tumor, and Metastasis. The preparation, specificity, and reliability of the rabbit polyclonal anti-legumain antibody used in the study were as described previously (2, 8, 9). The specificity of the antibody was further confirmed by performing Western blotting in the present study (Fig. 1).

Among 34 distant normal mucosa samples, 22 (65%) cases showed weak staining and 12 (35%) showed strong staining. Among 89 cases with adjacent normal mucosa, 64 (72%) were weak and 25 (28%) were strong staining. Among164 primary tumors, 69 (42%) presented weak and 95 (58%) presented strong staining. Of 33 metastasis, 11 (33%) had weak and 22 (67%) had strong staining. Regarding staining percentage of the primary tumors, 32 (20%) cases were in <75% and 132 (80%) in >75% staining. Looking at the stromal percentage of the staining, 42 (26%) cases in <75%, and 122 (74%) in >75% staining. Figure 2 presents legumain expression in normal mucosa, primary tumor, and metastasis in the lymph node in a case.

We compared the legumain expression in distant normal mucosa, adjacent normal mucosa, primary tumor, and metastasis. In the unmatched cases, legumain expression increased from distant normal mucosa (P = 0.02) or adjacent normal mucosa (P < 0.0001) to primary tumors, but there was no significant change of the staining between primary tumors and metastases (P = 0.35). Furthermore, considering the matched cases, legumain expression increased from distant (P = 0.04) or adjacent normal mucosa (P < 0.0001, Table 1) to primary tumor. Again there was no statistically significant change of the staining between primary tumors and metastases (P = 0.29). There was no significant difference in the staining intensity between distant and adjacent normal mucosa (P = 0.44 for the unmatched cases and P = 0.62 for the matched cases).

Relationship of Legumain Expression in Primary Tumor to Clinicopathologic and Biological Variables.Table 2 presents legumain expression in primary tumor cells in relation to clinicopathologic and biological variables. Tumors exhibiting strong expression of legumain had PCNA expression (P < 0.0001) and a high apoptotic rate (P < 0.0001) and had a tendency towards more necrosis (P = 0.07) and strong PINCH expression (P = 0.05). Considering the staining percentage, the high percentage of legumain expression was more frequent in more poorly differentiated tumors/mucinous carcinoma (P = 0.04) and tumors exhibiting positive p53 expression (P = 0.049) and tended to have PCNA positive expression (P = 0.08).

Table 3 presents legumain expression in stroma in relation to clinicopathologic and biological variables. Strong staining for legumain was related to more necrosis (P = 0.03), positive PCNA expression (P < 0.0001), and a high apoptotic rate (P < 0.0001) and showed a trend towards strong PINCH expression (P = 0.06). The staining percentage showed a positive trend towards stromelysin 3 expression (P = 0.058).

In terms of the relationship of legumain expression with patients' survival, neither the staining intensity (P = 0.93) nor the staining percentage (P = 0.08) in tumors was of significance. The same was found with stromal staining intensity (P = 0.50) and percentage (P = 0.26). However, when the staining intensity was combined with the percentage, the results showed that patients with tumor cells having both weak staining and <75% of legumain expression had a significantly more favorable prognosis than the other patients (P = 0.01, Fig. 3A). This significance difference remained even in a multivariate analysis after adjustment for gender, age, tumor location, Dukes' stage, growth pattern, and differentiation (P = 0.01, Table 4). The same results were seen with legumain expression in the stroma (i.e., the weak and <75% expression of legumain predicted a better prognosis; P = 0.04, Fig. 3B), although the significance was lost in a multivarite analysis (P = 0.11, data not shown).

In the present study, we observed that legumain expression either in tumor cells or in stroma was increased in primary tumors compared with distant/adjacent normal mucosa, but there was no significant increase in the expression from primary tumors to metastases. An earlier study has shown that legumain is highly expressed in several types of human solid tumors including colorectal cancer, comparing with the corresponding normal tissues, but they did not include metastasis in their study (2). Our present findings not only had confirmed that legumain may be involved in the development of colorectal cancers but also further suggested that legumain may play a greater role in the early development of colorectal cancers.

We furthermore investigated the relationship of legumain expression with clinicopathologic and biological variables and found that abundant expression of the legumain was related or tended to be related to poorer differentiation/mucinous carcinoma, more necrosis and apoptosis, positive expression of PCNA, p53, PINCH, and stromelysin 3. It has been determined that poorer differentiation, abundant necrosis, and increased PCNA and p53 expression represent a higher malignant potential of cancers (3, 4, 10–13). Regarding legumain positively related to apoptosis, we previously found a positive relationship between PCNA and apoptosis (7). It seems that PCNA functions as a communication point between proteins involved in DNA metabolism. One of these proteins, MyD118, has been shown to interact with both PCNA and p21 and has been shown to play a part in one apoptotic pathway (14). Regulation of the apoptotic cascade is a complex phenomenon and many factors may be involved. Overexpression of oncoproteins such as c-Myc, c-Fos, or c-Jun, not only leads to proliferation but also simultaneously induces apoptosis (15). PINCH and stromelysin 3 are factors present in stroma related to tumor development and aggressiveness. PINCH is the LIM protein that functions as an adapter protein at a key convergence point for integrin and growth factor signal transduction and is up-regulated in the stroma, associated with tumors and is particularly intense in stromal cells at invasive edges. These data suggest that the up-regulation of PINCH is a marker for stroma that can facilitate cancer invasion (6, 16). Stromelysin 3 is a matrix metalloproteinase which is specifically expressed in fibroblastic cells of most invasive carcinomas including colorectal cancer and represents a potential new prognostic indicator (17–20). Taken together, our results showed that legumain expression was not only related to factors present in tumor cells but also related to factors present in stroma. Whether legumain had a potential interaction with those factors in tumor development needs to be further studied.

We did not find that either the intensity or the percentage of legumain expression was related to prognosis, but we did find that the combination of the intensity with the percentage was a prognostic factor. The patients with both weak staining and lower percentage of legumain expression had a more favorable outcome. This may suggest that not only intensity but also covered areas of legumain expression should be considered for evaluating its prognostic importance.

In conclusion, legumain expression may be involved in early development of colorectal cancer, and further can be considered as a prognostic factor in the patients.

We thank Helen Richard, Gertrud Stride, and Gunnel Lindell for kindly preparing tissue sections.