Overexpression of Rab1B and MMP9 predicts poor survival and good response to chemotherapy in patients with colorectal cancer

The expression of Rab1B and MMP9 is up-regulated in CRC cell lines

To determine the expression of Rab1B and MMP9 in CRC, we first performed immunoblot and RT-PCR in 11 CRC cell lines and one human normal colon cell line CCD112CoN. The result showed that the expression levels of Rab1B and MMP9 proteins were significantly increased in 81.8% (9/11) and 63.6% (7/11) of CRC cell lines, respectively, compared with those in normal cell line (Fig. 1A). As expected, relative expressions of Rab1B and MMP9 mRNA in CRC cell lines were also higher than those in CCD112CoN cells (Fig. 1B). Furthermore, we found that there is a positive correlation trend between Rab1B and MMP9 expression in both mRNA and protein (for mRNA: r = 0.473, P=0.121; for protein: r = 0.537, P = 0.072, Fig. 1C and 1D), although the correlations were not statistically significant. These results indicate that both Rab1B and MMP9 are up-regulated in CRC cells and may have a positive correlation.

The protein expression of Rab1B and MMP9 is increased and has a positive correlation with each other in CRC tissues

To further investigate the expression and clinical significance of Rab1B and MMP9 protein in CRC, we collected 179 pairs of cancer and corresponding adjacent non-tumor tissues from CRC patients. Their demographic and clinicopathological data are shown in Table 1. The protein expression of Rab1B and MMP9 in the paired CRC and non-tumor tissues was examined by immunohistochemistry (IHC). Both Rab1B and MMP9 proteins are mainly distributed in the cytoplasm and membrane of CRC cells (Fig. 2A) with 81% (145/179) and 71.5% (128/179) of CRC samples displaying higher expression of Rab1B and MMP9 proteins than the matched adjacent non-tumorous tissues, respectively (P <0.0001, Fig. 2B and 2C). To verify the IHC results, we performed immunoblot analysis on another 8 paired CRC and non-tumorous tissues. As expected, high expression of Rab1B and MMP9 was found in 75% (6/8) and 100% CRC tissues (Fig. 2D), respectively, which is consistent with the results of IHC. Furthermore, we analyzed the relationship between Rab1B and MMP9 expressions. In the same cohort of CRC tissue samples, 71.1% (64/90) samples with MMP9 overexpression had high expression of Rab1B protein, while 69.7% (62/89) samples with MMP9 down-regulation also had low expression of Rab1B (P < 0.001, Table 2 and Fig. 2E). Pearson’s correlation analysis showed that there is a positive correlation between Rab1B and MMP9 protein expressions in 179 CRC samples (Fig. 2F, r =0.388, P< 0.001), which is similar to that in CRC cell lines. These data show that Rab1B and MMP9 proteins are significantly co-overexpressed in CRC tissues.

Rab1B and MMP9 overexpression is correlated with tumor progression and metastasis in CRC patients

To evaluate the clinical relevance of Rab1B and MMP9 protein expression in CRC patients, the median IHC scores of 180 and 156 were defined as the cutoff value for high- and low-expression of Rab1B and MMP9, respectively, which divided CRC patients into high- or low-expression groups. As showed in Table 2, high Rab1B and MMP9 protein expressions in CRC are significantly associated with deep invasion, lymph node metastasis, and advanced TNM stage, suggesting that both Rab1B and MMP9 are involved in the progression and metastasis of CRC.

Rab1B and MMP9 protein expressions are associated with poor prognosis

To explore the prognostic value of Rab1B and MMP9 proteins in this disease, we analyzed overall survival (OS) and progression-free survival (PFS) of CRC patients with Rab1B and/or MMP9 expressions. Patients with Rab1B high-expression have significantly shorter 5-year OS rate (63.5% vs. 92.6%) and 5-year PFS rate (56.3% vs. 88.1%) than those with Rab1B low-expression (All P < 0.001, Fig. 3A). Similarly, patients with MMP9 high-expression have much poorer OS and PFS than those with MMP9 low-expression (Fig. 3B), which is consistent with the conclusion from a meta-analysis of 13 cohort studies on MMP9 expression and prognosis in CRC [25]. These results indicate that the increased Rab1B or MMP9 protein is significantly correlated with poor prognosis of CRC patients.

We next analyzed the effect of their expressions and clinicopathological parameters on patient survival by using univariate and multivariate Cox model. Univariate analysis indicates that TNM stage, adjuvant chemotherapy, Rab1B and MMP9 proteins are significant predictors for OS and PFS of CRC patients (All P < 0.05, Table 3), and Tumor depth is a marginally significant predictor for OS and PFS (P =0.065 and P =0.059, respectively, Table 3). Multivariate Cox regression analysis further demonstrates that Rab1B protein is an independent risk predictor for OS (HR: 3.605, 95% CI: 1.481-8.775, P = 0.005) and PFS (HR: 3.394, 95% CI: 1.579-7.297, P = 0.002) of CRC patients (Table 3), and MMP9 protein is a marginally significant independent unfavorable predictor for OS and PFS in CRC patients (Table 3). In addition, TNM stage and adjuvant chemotherapy also are independent prognostic factors for OS and PFS in CRC patients.

Moreover, we found when patients were stratified by TNM stage, stage I-II patients with high level of Rab1B protein has marginally significantly poorer OS and PFS than those with low level (Fig. 3C), and stage III patients with high Rab1B level has significantly poorer survival than those with low level (Fig. 3D). This result reveals that Rab1B can predict survival of CRC patients independent of clinical stage. Therefore, Rab1B may provide additional prognostic information to the current clinical staging system.

Combination of Rab1B and MMP9 expression significantly improves predictive efficiency for the outcome of CRC patients

We found that in patients with low MMP9 expression, those with high Rab1B expression had significantly worse survival than those with low Rab1B expression (Fig. 4A). Similar result was observed in patients with MMP9 high expression (Fig. 4B). These results suggest that Rab1B is a prognostic predictor independent of MMP9. Therefore, we hypothesize that combination of both proteins will improve their predictive efficiency for survival. To this end, a new combined risk score was calculated as the sum of Rab1B score (0 or 1) and MMP9 score (0 or 1) for each case. Survival analysis shows that patients with low- (score 0), intermediate- (score 1) or high-risk (score 2) have significantly different 5-year overall survival rates, 96.7%, 80.2% and 57.5%, respectively. Multivariate Cox regression analysis reveals that the new combined risk score is an independent prognostic factor for OS and PFS in CRC patients (Table S1). Compared with TNM staging system, Rab1B/MMP9 combined risk score has some better predictive accuracy for OS (ROC area: 0.76 vs. 0.71, Fig. S1A) and PFS (ROC area: 0.73 vs. 0.69, Fig. S1B) of CRC patients though there is no statistically significant difference between the two ROC areas. Thus, the new combined risk score is a potentially useful biomarker that provides additional prognostic information for physician to evaluate survival and make decision on adjuvant chemotherapy in post-operative CRC patients.

Rab1B and MMP9 protein expression predicts outcome of adjuvant chemotherapy in CRC patients

In clinical practice, physicians determine the use of adjuvant chemotherapy based on the clinical stages and high-risk features of patients with CRC. However, clinical stages and high-risk features do not always accurately predict the outcome of chemotherapy. To investigate the utility of Rab1B and MMP9 proteins in CRC management, we explored the use of Rab1B and/or MMP9 protein expression to predict the response to adjuvant chemotherapy in CRC patients. First, we divided patients into high- and low-Rab1B groups. Survival analysis revealed that patients with low Rab1B protein had a similar outcome (including OS and PFS) regardless whether they had adjuvant chemotherapy (Fig. 5A), indicating that these patients do not benefit from adjuvant chemotherapy. Thus these patients should avoid the unnecessary toxicity and economic burden associated with chemotherapy. We next conducted the same analysis on patients with high Rab1B protein. Surprisingly, patients received adjuvant chemotherapy had markedly better OS and PFS than those not received the chemotherapy (Fig. 5B), clearly indicating that patients with high Rab1B protein level may obtain benefit from adjuvant chemotherapy. We also performed the same analysis on MMP9 and observed similar results (Fig. 5C and 5D). Finally, we combined Rab1B and MMP9 into the aforementioned new risk score model and conducted the same analysis. The results show that in patients with low or intermediate risk, there is no difference in OS and PFS between patients with and without chemotherapy (Fig. 6A and 6B), but in patients with high risk, those with chemotherapy had significantly better survival than those without chemotherapy (Fig. 6C).

According to the current criteria of adjuvant chemotherapy, 109 patients were not administered adjuvant chemotherapy in this study. When these patients were divided into low- and high-expression groups based on Rab1B or MMP9 expression, those with high-expression of Rab1B or MMP9 had significantly worse survival (including OS and PSF) than those with low-expression (Fig. S2A and S2B), indicating that patients with high-expression of Rab1B or MMP9 should be administered adjuvant chemotherapy to improve survival. Next, we combined both proteins to predict outcome of these 109 patients, and the result showed that patients with low combined risk had the best survival and those with high combined risk had the worst survival among the three risk subgroups (Fig. S2C). Finally, we compared single proteins with the combined model using ROC analysis. The result demonstrates that the combination risk score model is remarkably better than single proteins (Fig. S2D), suggesting that the combined risk model can provide more detailed and accurate information to help physicians make chemotherapy decisions.

Patients and tissue samples

The tumor and matched colorectal tissues were obtained from 179 consecutive patients with stage I-III CRC who underwent radical resection at Sun Yat-Sen University Cancer Center (Guangzhou, China) between January 2009 and December 2010. None of patients received preoperative chemotherapy and/or radiotherapy. After radical resection, patients with stage III and high-risk stage II CRC further received postoperative adjuvant chemotherapy. Fluorouracil-containing regimens are standard, including FOLFOX, XELOX and Xeloda. All the samples were pathologically diagnosed by two experienced pathologists. Histological classification and tumor differentiation were determined according to the criteria of the World Health Organization. Clinical stage was defined according to the 7^th edition of tumor-node metastasis (TNM) classification of the American Joint Committee on Cancer Staging (AJCC) [31]. This cohort of CRC patients included 95 males (52.8%) and 85 females (47.2%), with a mean age of 59.5 years old. During the follow-up, tumor assessment was done by colonoscopy, abdominal ultrasound or computed tomography scanning at 3-month intervals for the first 2 years, then at 6-month intervals for 3 years for a total of 5 years. The average follow-up time was 54.2 months, ranged from 1 to 80 months. Overall survival (OS) is defined as the time from the date of surgery to the date of death from any cause or last date of follow-up; Progression-free survival (PFS) is defined as the time from the date of surgery to the date of relapse or metastasis of CRC or death from CRC or last date of follow-up.

Another randomly selected eight pairs of fresh CRC tissues and matched adjacent non-tumorous colorectal tissues from patients undergoing surgical resection in 2011 were collected for immunoblot analysis. This study was reviewed and approved by the Ethical Committees of Sun Yat-Sen University Cancer Center. Written informed consent was obtained from all patients before surgery.

Cell lines and cell culture

Eleven human CRC cell lines (SW480, SW620, HT29, CACO2, HCT8, HCT116, CW2, DLD-1, RKO, LS174T and LoVo) and human immortalized colon cell lines (CCD-112CoN) were obtained from the American Type Culture Collection (ATCC). All cell lines were cultured in the conditions specified by the manufacturer, and authenticated by short tandem repeat DNA fingerprinting and tested for mycoplasma before use at Medicine Laboratory of the Department of Forensic Medicine, Sun Yat-Sen University (Guangzhou, China).

Immunohistochemistry (IHC)

IHC staining was performed using standard streptavidin-peroxidase complex method as described previously [32]. Briefly, fresh surgical tissue specimens were fixed in 10% formaldehyde and routinely processed for paraffin embedding. Then these blocks were cut into 4μm thick sections. Endogenous peroxidase activity was blocked by 3% hydrogen peroxide in methanol. After antigen retrieval, the sections were incubated with rabbit polyclonal Rab1B antibody (1:200, 17824-1-AP,Proteintech Group, USA) or mouse monoclonal MMP9 antibody (1:50, sc-21733,Santa Cruz, USA) overnight at 4°C and a negative control was set up by replacing the primary antibody with phosphate buffered solution (PBS). Subsequently, the sections were incubated with horseradish peroxidase (HRP)-conjugated anti-rabbit or anti-mouse IgG secondary antibody (Dako, Denmark) for 30 min at room temperature, followed by developing using 3, 5-diaminobenzidine (DAB, Dako, Denmark) substrate and counterstaining with Mayer’s hematoxylin for the nuclei.

Scoring of immunostaining

The scoring of immunostaining was independently performed by two experienced pathologists who were blinded to the clinical information. The immunostaining intensity was scored as 0 - 3 (0, no staining; 1, weak staining; 2, moderate staining; 3, strong staining), and percentage of immunostained tumor cells was scored as 0 - 100. The immunostaining intensity was multiplied by the percentage of stained tumor cells, which resulted in a semi-quantitative immunostaining score (ISS) between 0 and 300.

Immunoblotting

The fresh CRC tissue samples and CRC cell lines were directly homogenized in RIPA Lysis buffer. 30μg of lysate protein were run on a 10% SDS-PAGE gel and then transferred onto polyvinylidene fluoridemembranes (PVDF, Millipore, USA). After blocking nonspecific binding site with TBST containing 5% non-fat milk, the membranes were incubated with a rabbit polyclonal anit-Rab1B antibody (1:300, sc-599, Santa Cruz, USA), a mouse monoclonal anti-MMP9 antibody (1:200, sc-21733, Santa Cruz, USA) or GAPDH antibody (1:6000, #5174, Cell Signaling Technology, USA) at 4°C overnight. Then the membranes were incubated with HRP-conjugated secondary antibody (1:5000, Jackson Immunoresearch Inc, USA) for about 60 min at room temperature. The blots were scanned and the intensities of protein bands were quantitated by the Bio-Rad software Quantity One (Bio-Rad Laboratories Inc., USA).

Isolation of total RNA and Real-time quantitative PCR (RT-PCR)

Total RNA from cancer cells was isolated using TRIzol reagent (Invitrogen) as per the manufacturer’s instruction. RNA samples were purified and extracted with phenol and chloroform. The quantity of RNA samples was measured by a NanoDrop^TM 2000 spectrophotometer (ThermoFisher Scientific, Waltham, MA, USA). One microgram of total RNA was used to synthesize cDNA via reverse transcription reaction according to the protocol of GoScript^TM Reverse Transcription System kit (Promega, A5001, Madison, WI, USA). After 1:20 dilution, 2μl of cDNA products was employed for PCR, which was performed on a Roche Lightcycler 96 real time PCR machine (Roche Diagnostics, Indianapolis, IN, USA) according to a standard method as described previously [33]. All samples were amplified in triplicate and GAPDH was detected as an internal control. The relative quantification of target genes was calculated using the comparative 2^-ΔΔCT method [34]. Primer sequences for this experiment are as follows:Rab1B forward primer: 5’-GGACTTCAAGATCCGAACCAT-3’, reverse primer: 5’-ATACACCACGATGATGCCA-3’, and the amplicon length: 135bp; MMP9 forward primer: 5’-GGGACGCAGACATCGTCATC-3’, reverse primer: 5’-TCGTCATCGTCGAAATGGGC-3’, and the amplicon length: 139bp; GAPDH forward primer: 5’-CTCCTCCTGTTCGACAGTCAGC-3’, reverse primer: 5’-CCCAATACGACCAAATCCGTT-3’, and the amplicon length: 204bp.

Statistical analysis

SPSS 17.0 for windows (SPSS, Inc., Chicago, IL, USA) and GraphPad PrismV6 (GraphPad Prism, Inc., USA) were used for statistical analyses. The results were expressed as mean ± SD or SEM. The correlation between Rab1B expression and clinicopathological parameters was analyzed by Chi-square test or Fisher’s exact test. The Student’s t-test was used for the analysis of comparisons. The relationship between Rab1B and MMP9 expressions was tested by Pearson’s correlation analysis. The prognostic variables were evaluated by univariate and multivariate Cox proportional hazards regression model. Kaplan-Meier plots and log-rank test were used to estimate cancer specific survival curves.