Osteoglycin (OGN) reverses epithelial to mesenchymal transition and invasiveness in colorectal cancer via EGFR/Akt pathway

The following antibodies and reagents were used: anti-OGN (Western blot) from R&D Systems anti-human OGN (IHC) from Sigma-Aldrich; anti-Akt, anti-phospho-Akt (Ser473), anti-phospho-EGFR (Y1068), anti-EGFR, anti-Slug, anti-Erk1/2, anti-Zeb-1, anti-Snail, anti-Twist, anti-Snail from Cell Signaling Technology; anti-CD31, anti-β-actin from prteintech; cross-linking reagent BS3 from Thermo Scientific Pierce; Eps15, epsin1 from Santa Cruz Biotechnology; AKT activator: sc79 from Selleck; OGN human enzyme-linked immunosorbent assay (ELISA) kit from USCN. All chemicals, unless otherwise specified, were purchased from Sigma-Aldrich.

The protocol of current study was reviewed and approved by the specialty committee on ethics of biomedicine research, Fudan University Shanghai Cancer Center (FUSCC). The tissue acquisition and utilization followed the National Regulations on the Use of clinical samples in China. All patients from FUSCC dataset have provided written informed consent. The enrollment criteria included: (1) pathologically confirmed with primary colorectal adenocarcinoma and no history of other cancers; (2) undergoing curative surgical resection, but without any preoperative anticancer treatment; and (3) with complete clinical and follow-up data.

The TMA used for this study includes unselected, non-consecutive, primary, and sporadic colorectal cancers enrolled between January 2007 and November 2009 in FUSCC. Construction of this TMA has been previously described in detail [14]. Every section was scored independently by two pathologists utilizing a semiquantitative scoring system [15, 16]. The staining intensity was scored as 0 (negative), 1 (weak), 2 (medium) or 3 (strong). Extent of staining was scored as 0 (< 5%), 1(5–25%), 2 (26–50%), 3 (51–75%) and 4 (> 75%) according to the percentages of the positive staining areas in relation to the whole carcinoma area. We multiplied the percentage score by the staining intensity score to generate the immunoreactivity score (IRS). High expression of OGN was defined as detectable immunoreactions in cytoplasm and stoma with IRS > 4.

Human OGN, Zeb-1, Slug, Snail cDNA was chemically synthesized and cloned into a lentiviral expression vector, pCDH-CMV-MCSEF1-Puro generating pCDH-OGN. The human colon cancer cell lines (SW620, HT29, RKO and HCT116) were transfected with the pCDH-OGN expression vector or the control vector.

Cells were implanted onto a chamber slide for 24 h reaching 80% confluent, then fixed with paraformaldehyde for 30 min, and followed by permeabilized with 0.5% Triton X-100 for 10 min at room temperature, thereafter the primary antibodies for EGFR were added incubation for overnight at 4 °C, and then the alexa flours 488 Goat anti-rabbit (1:200, Invitrogen) was used as the secondary antibody for an hour at room temperature. At last, nuclei were stained with DAPI for 5 min when necessary. Fluorescence images were photographed with a fluorescence microscope.

RIPA lysis buffer was used for protein extraction. Protein from either cell lysates were separated by SDS-PAGE (6–20% gel) and then transferred to PVDF membranes. Membranes were probed with primary antibodies. Following incubation with horseradish peroxidase-conjugated secondary species-specific antibodies, immunoreactive proteins were detected using ECL (Pierce, Thermo Scientific) and detected using a BioImaging System.

Cell viability: A count of 5 × 10³ cells/well was plated into 96-well plates in triplicate and were allowed to adhere overnight. After adherence, 10 μl/100 μl of CCK8 solution was added to each well to co-incubate for 2 h at 37 °C. Then, cell viability was measured by spectrophotometrically at 450 nm, and this was recorded as day 1. After 2, 3, 4, 5 and 6 days, cell viability and proliferation were re-assessed.

SW620 cells (10⁶ cells/mouse) which transfected with pCDH-OGN or empty vector control were suspended in 100 μL Matrigel, injected subcutaneously into the right flank of nude mice (n = 8, male; 4 weeks old Balb/C athymic nude mouse), and allowed to grow for 4 weeks. Tumor growth was monitored using calipers every 5 days and animals were euthanized when tumors reached 10% of body weight. Primary tumors and organs were harvested and fixed in 10% formalin and paraffin embedded for pathological analysis. Tumor volumes were calculated using the following equation: Volume = (d² × D)/2, where D is the long side and d is the short side. For the metastasis assay, stable OGN-SW620 cells (5 × 10⁵ cells) were orthotopically implanted into the spleens of nude mice.

Statistical analysis was performed with SPSS 22.0 (SPSS Inc., Chicago, IL, USA) and GraphPad Prism v.6 (La Jolla, CA, USA). The optimal cut-off values for serum OGN levels were identified by X-tile 3.6.1 software [17] (Yale University, New Haven, CT, USA). Chi-square test was used to analyze the relationship between clinicopathological parameters and OGN expressions. Survival analysis was performed using the Kaplan-Meier method and Cox regression model. For normal or approximately normal variables, Student’s t-test (for two group comparisons) or analysis of variance (three or more groups) was utilized. P < 0.05 was considered statistically significant. All confidence intervals (CIs) were stated at the 95% confidence level.

OGN is expressed within the epithelial cancer cell and stromal compartments of human colorectal tumors. OGN staining in the epithelium displayed a cytoplasm-accentuated expression, often combined with a clear membrane staining. Staining in the epithelial cells was frequently accompanied with staining in the stroma, while stromal staining was usually diffuse. To evaluate the expression patterns and clinical importance of OGN, we divided patients into two groups with low expression (Low, IRS ≤ 4) and high expression (High, IRS > 4). In 109 CRC patients (40.1%), a high expression of OGN was observed and OGN expression was designated as low in 163 cases (59.9%). No demographic or baseline clinical data were statistically associated with any pattern of OGN expression. Gender, age, histological type, TNM stage, pathological grading, venous/perineural invasion, MS/MMR status, and adjuvant therapy in FUSCC cohorts are shown in Table 1.

Cox regression analysis was performed to test the associations between OGN expression and oncologic outcomes, and the univariate Cox regression model implied that OGN expressions, TNM stages, venous/ perineural invasion and adjuvant therapy were associated with prognosis of CRC patients in terms of cancer specific survival (P < 0.05, Table 2). Consistently, multivariate analysis after adjustment revealed that OGN expression was also independent prognostic factor for cancer specific survival in CRC patients, besides T stages and adjuvant therapy (P < 0.05). Hence, Kaplan-Meier analysis showed the OGN expression was markedly associated with longer time post-surgical resection, with prolonged cancer specific survival as 75.7 months in the High OGN expression group versus 61.6 months in the Low OGN expression group. Moreover, this survival benefit can translate into prolonged disease-free survival (DFS) evidenced by univariate Cox proportional hazards model (Table 3). What is more, in the multivariate Cox proportional hazards model after adjusting for other effects, OGN expression (p = 0.035), TNM stages (p < 0.001), perineural invasion (p = 0.042), and lymph node examined (p = 0.032) were all independently associated with improved DFS. Meanwhile, Kaplan-Meier survival analysis for OGN expression also revealed that high OGN expression correlated with a longer disease-free survival (p = 0.037).

The potential prognostic effect of OGN expression in CRC patients was further examined by the level of serum OGN. In the FUSCC cohort, the cut-off values of serum OGN levels were determined by X-tile program, which were 12.651 ng/ml, (Fig. 1e). Patients were divided into 2 groups for further analysis (High serum level > 12.651 ng/ml, Low serum level ≤ 12.651 ng/ml). Although OGN level was not correlated with overall survival, but with regard to DFS, high OGN level was indicative of longer survivals (P = 0.032 Fig. 1f).

The expression of OGN in CRC cell lines were all at lower levels compared to the normal human colon mucosal epithelial cells (NCM460), see Fig. 2A1, so we created OGN over-expressing colorectal cell lines. The colon cancer cell lines were transduced with lentiviral constructs encoding human OGN cDNA to generate stable OGN over-expression cell lines (HCT116-OGN⁺, RKO-OGN⁺, SW620-OGN⁺, HT29-OGN⁺). Over-expression efficiency was verified by western blotting (Fig. 2A1). The CCK-8 assays were applied to examine whether OGN expression affected colorectal cancer cells proliferation and viability in vitro. As present in Fig. 2A2, there were significant growth suppression rates when OGN over-expressed in cell lines compared with the controls. What is more, cell lines SW620, HT29 showed better response to OGN challenging in cell CCK-8 assay. To determine the effect of altered OGN expression on migration and invasion, OGN transfected cells were used for wound-healing and transwell assays. Over-expression of OGN attenuated the migration and invasion ability of HT29 and SW620 cells (p < 0.05). In detail, OGN over-expression resulted in a 31.13% reduction in migration in the HT29 cells, and decreased the invasion by 37.01% in the SW620 cells (Fig. 2b, c). In line with the results obtained in the wound-healing assays, OGN over-expression prevented the average invasive cells of HT29 cells by 77.32%, as well as in SW620 cells, the average cells transwelled were reduced by 89.19% (Fig. 2d, e).

Pairwise gene correlation analysis for READ sets of TCGA was performed finding OGN correlated with EGFR closely (p = 0.0015). The growth of colorectal cancer cells is dependent on or facilitated by EGFR, hence, we explored whether EGFR and related signaling may be involved in OGN-induced cell inhibition. Firstly, we examined whether activation of the EGFR signaling pathway was modulated by OGN over-expression. HT29-OGN⁺ and SW620-OGN⁺ cells, which exposed to extracellular EGF (100 ng/ml), exhibited suppressed activation of EGFR protein phosphorylation level compared to Control cells. After 15 min of EGF exposure, the increase in p-EGFR/EGFR levels were reduced by 30.14% in HT29-OGN⁺ and 33.22% in SW620-OGN⁺ cells compared to the control cells (Fig. 3a). To address the mechanism for EGFR activation reduction, immunofluorescent staining of EGFR demonstrated a reduction in membranous, with a concomitant increase in cytoplasmic, after OGN over-expression (Fig. 3b). The signaling properties of EGFR are closely regulated by its sub-cellular localization, exactly, suppression of EGFR signaling may initiate from receptor endocytosis induction, eventually proteasome targeting and degradation of internalized receptors were followed. Hence, it was indicated that OGN enhanced the internalization of EGFR from the cell membrane into the cytoplasm, and then inhibited the activation of EGFR protein phosphorylation levels. Meanwhile we assessed whether OGN controlled EGFR activity by protein degradation via the proteasome. We observed the previously observed p-EGFR/EGFR inhibition was blocked, when HT29/SW620 cells with OGN over-expressed were pretreated with MG132 (the proteasome inhibitor) before exposing to EGF (Fig. 3c). The degradation was always resulted from the formation of EGFR dimmers, so we investigated whether OGN affected EGFR dimerization. Colon cancer cells were exposed to EGF for three-time intervals and then pretreated with the membrane-impermeable chemical cross-linker BS3 allowing resolution of dimeric components of EGFR. As a result, OGN over-expressed cells produced a greater increase in EGFR dimmers compared to the control cells (Fig. 3d). Besides the dimerization triggering EGFR endocytosis, the recruitment of EGFR internalization machinery, such as early endocytic adaptors (EGFR pathway substrate, Eps15 and epsin1) also played a role in the protein degradation. Therefore, we investigated whether the over-expression of OGN influenced the recruitment of Eps15 and epsin1 to EGFR. It was shown that the recruitment of Eps15 and epsin1 following EGF challenge was prominently increased in cells over-expressed for OGN (Fig. 3e). Collectively, these data implied a model in which over-expression of OGN detained proliferative signaling by inhibiting EGFR activity.

There are three major downstream pathways of EGFR, including PI3K/Akt, Stat3 and ERK1/2-mediated signaling pathways. Regarding to the observations above that OGN suppressed EGFR activity in HT29 and SW620 colorectal cancer cells, we aimed to identify which downstream pathways of EGFR will be inhibited after OGN over-expression. Firstly, the response of Akt, Stat3 and ERK1/2 to EGF exposure in OGN over-expressed cells were tested and only a decline in Akt phosphorylation was observed, while there was not any significant difference in Stat3 and ERK1/2 (Fig. 4a). As OGN abolished Akt activity, and then whether it was via EGFR that OGN reduced Akt activity? So, the proteasome inhibitor MG132, which referred to block EGFR phosphorylation, was used to pretreated cells with OGN over-expression, consequently the above observed Akt phosphorylation reduction was reversed as was the EGFR degradation blocked (Fig. 4b).

EGFR activation is frequently induced to mediate carcinoma invasion and metastasis through promoting EMT. Immunoblotting of protein lysates from OGN over-expressed HT29/SW620 cells abolished a significant reduction in expression of E-cadherin induced by EGF (Fig. 4c). Furthermore, the expressions of EMT-associated molecular biomarkers and EMT transcription factors, such as Slug, Zeb-1, and Snail decreased simultaneously, but had no effect of Twist (Fig. 4c). To assess whether suppression of p-Akt modulated the previously observed inhibition of EMT, the effect of OGN pretreated with SC79 (constitutive Akt activator) was explored again. Re-expression of p-Akt leaded to reverse the decrease in E-cadherin protein levels after OGN over-expression (Fig. 4d). Levels of EMT markers, such as Vimentin, Slug, Zeb-1, Snail and the protein levels of E-cadherin (Fig. 4d) were similarly affected, hence a critical role of Akt inhibition was indicated in the observed OGN-mediated decrease in EMT. The above results demonstrated the fact that OGN could reduce EMT via inhibition of the Akt pathway.

EMT plays a vital role in evolution of cancer cells gaining migratory and invasive properties. In order to confirm the ability of OGN to inhibit malignant potential and to determine the molecular mechanism by which OGN regulated EMT, we evaluated the altered EMT transcriptional markers above (Zeb-1, Slug and Snail) in OGN over-expressed cells and reassessed the cell invasion ability in rescue experiments. In rescue experiments, cancer cells with stable ectopic OGN expression were transiently transfected with Zeb-1, Slug or Snail. It has been confirmed that only rescue of Zeb-1 can reverse the inhibition of EMT in OGN over-expressed HT29/SW620 cells (Fig. 5a). wound-healing and transwell invasion assays were performed after treatment with ectopic OGN and EMT transcriptional markers expression. As shown in Fig. 5b–c, ectopic OGN expression markedly impeded the migration and invasion of HT29/SW620 cells, however, Zeb-1-over-expression abolished OGN-inhibition migration and invasion. These results indicated that Zeb-1 was a key molecular for OGN induced EMT inhibition and EMT signaling was essential for OGN-induced CRC cell migration and invasion.

Understanding the effect of OGN in tumorigenicity requires consideration of the cancer cell per se and the microenvironment established by complex interactions between the host and the cancer cell. To validate our data in vitro models of colorectal cancer, we injected male nude mice subcutaneously with equal numbers of SW620-OGN⁺, or SW620 control cells and monitored both tumor growth and final tumor mass. Primary tumor growth and the emergence were assayed over time. OGN over-expression in cancer cells led to a decrease in growth rate and final mean tumor volume compared with control cells (Fig. 6a). In consistent with the above findings, as shown in Fig. 6c, xenograft tumors derived from OGN over-expressing cells had consistently less Zeb-1 staining compared to their control counterpart cells, at last, led to E-cadherin (marker of EMT) upregulation. In addition, the liver metastases were observed with fewer and smaller lesions in the mice injected with SW620-OGN⁺ (Fig. 6D1, D2). Collectively, these results suggested that OGN-reduced tumorigenicity and metastasis in vivo.