Introduction

Glucose-regulated protein 78 (GRP78), referred to as BiP, is an immunoglobulin heavy chain-binding protein and stress protein [1]. It shares amino acid sequence similarity with heat-shock protein 70 (HSP70); therefore, it is also regarded as a member of the heat-shock protein 70 (HSP70) family [2]. GRP78 resides primarily in the endoplasmic reticulum (ER). When unfolded proteins accumulate in the lumen of the ER, the expression of GRP78 protein will be up-regulated to restore ER homeostasis. In terms of physiological processes, GRP78 plays essential roles in early embryonic development, and the GRP78 expression level is decreased during aging [3, 4].

During cancer progression, cancer cells are often under ER stress in the tumor microenvironment, which is characterized by glucose deprivation, acidosis, and severe hypoxia [5]. Increased expression of the ER stress protein GRP78 has been found in several types of cancer, including breast cancer, lung cancer, esophageal cancer, gastric cancer, colon cancer, and endometrial cancer [6–11]. However, the role of GRP78 expression in the prognosis of various cancer types is controversial. The GRP78 expression level in esophageal cancer is higher in early-stage, highly differentiated adenocarcinomas, and patients with higher GRP78 levels tend to have better survival rates [8]. In contrast, in gastric cancer, GRP78 expression is up-regulated, and patients with higher GRP78 expression have a poor prognosis [9]. The GRP78 expression level in breast cancer cells also serves as a prognostic marker for responsiveness to hormonal therapy based on estrogen starvation, and combination therapy targeting GRP78 may enhance the efficacy of and reduce the resistance to hormone therapy [6].

In general, GRP78 had an anti-apoptotic function and conferred drug resistance in cancer cells such as gastric cancer, colon cancer, breast cancer, and hepatocellular carcinoma (HCC) cells [6, 12–14]. Suppression of GRP78 enhanced drug efficacy in both proliferative and dormant cancer cells, which implied that GRP78-mediated drug resistance is not limited to proliferating cancer cells but also occurs in dormant cancer cells [15].

A previous study revealed that GRP78 was overexpressed in colon cancers and might be an important biomarker for malignant transformation [10]. GRP78 silencing enhanced epirubicin-induced apoptosis in colon cancer cells and suppressed colon cancer growth through the down-regulation of the VEGF/VEGFR2 signaling pathway [16, 17]. Although overexpression of GRP78 was noted in colon cancers, its expression level was negatively correlated with lymphatic invasion [18]. In addition, colon cancers with higher surface GRP78 expression showed reduced tumor proliferation and growth, but their invasive ability increased with surface GRP78 expression [19, 20]. However, GRP78 localizes mainly to the endoplasmic reticulum, and only a small fraction is found on the cell surface [21]. The role of the overall GRP78 expression level on metastasis in colon cancer remains unclear. In this study, we demonstrate that the down-regulation of GRP78 expression increases the migration and invasion ability of colon cancer cells by inducing vimentin expression, decreasing the E-cadherin level and activating the NRF-2/HO-1 signaling pathway.

Materials and methods

Tissue samples

Two sets of tissue microarrays of colonic adenocarcinoma were purchased from US Biomax (Rockville, MD, USA) and Shanghai Outdo Biotech Co. (Shanghai, China). The first microarray (US Biomax) contained 30 cases of primary colonic adenocarcinoma, five cases of distantly metastatic adenocarcinoma, 25 cases of metastatic adenocarcinoma in lymph nodes, and normal colonic tissue. The second microarray (Shanghai Outdo Biotech Co.) consisted of 63 cases of low-grade (well-differentiated to moderately differentiated) and 29 cases of high-grade (poorly differentiated) colonic adenocarcinomas. The pathologic diagnoses of these cases were reconfirmed using microscopy. Immunohistochemical staining was performed on a BenchMark XT autostainer using an iView DAB detection kit (Ventana, Tucson, AZ, USA). The sections were incubated with GRP78 antibody (1:300; Santa Cruz Biotechnology, Dallas, TX, USA) for 1 h at 37 °C. Appropriate positive and negative controls were included in these assays. Both the intensity and extent of GRP78 expression in carcinoma cells were evaluated and scored by the method previously described [22]. The intensity of cytoplasmic staining was scored semiquantitatively as follows: 0 point, negative; 1 point, weakly positive; 2 points, moderately positive; or 3 points, strongly positive. The percentage of positive tumor cells (0–100 %) was multiplied by the intensity of GRP78 staining; therefore, the overall score ranged from 0 to 300. We divided the GRP78 expression scores (0–300) into three groups: low expression (0–100), intermediate expression (101–200), and high expression (201–300). For statistical analyses, we used Fisher’s exact test to determine whether there were significant differences in GRP78 expression among primary colonic adenocarcinoma, distantly metastatic adenocarcinoma, and metastatic adenocarcinoma in regional lymph nodes. The expression levels of GRP78 in low- and high-grade colonic adenocarcinoma were also compared.

Human colon adenocarcinoma cell culture

Human colon adenocarcinoma HT-29 (HTB-38) and DLD-1 (CCL-221) cell lines were purchased from the American Type Culture Collection (ATCC, Rockville, MD) and were isolated from Dukes’ stage B and C human colon adenocarcinoma, respectively. The cells were cultured in RPMI 1640 with 10 % fetal calf serum (FCS), penicillin (100 U/mL), and streptomycin (100 μg/mL) in a humidified incubator (37 °C, 5 % CO2). The cells were either sub-cultured or used before they reached 80 % confluence.

Small interfering RNA (siRNA) preparation

The expression of GRP78 was knocked down in HT-29 and DLD-1 cells using siRNA, as previously described [23]. The target sequence for the human GRP78 mRNA was 5′-AAGGTTACCCATGCAGTTGTT-3′, and the scrambled siRNA sequence was 5′-AAGGTGGTTGTTTTGTTCACT-3′. The GRP78 siRNA and scrambled siRNA were inserted into pSUPERIOR vectors to generate the pSUPERIOR-GRP78-siRNA and pSUPERIOR-scramble-siRNA plasmids, respectively. The siRNAs were then transfected into cells. Briefly, 1.5 × 105 cells were washed twice with Phosphate buffered saline (PBS) and mixed with 0.5 μg of plasmid. We applied one pulse for 20 ms under a fixed voltage of 1.4 kV on a Neon pipette-type microporator (Invitrogen Life Technologies, Grand Island, NY) [24, 25]. Successfully transfected HT-29 and DLD-1 cells were selected based on their antibiotic resistance.

Transwell migration assay

In vitro cell migration was examined using the BD Falcon cell culture insert (BD Biosciences). Aliquots of 1 × 105 DLD-1 cells expressing scrambled control or GRP78KD were suspended in 500 μL of serum-free RPMI 1640 and were seeded into the upper compartments of each chamber. The lower compartments were filled with 1 mL of RPMI 1640 with 10 % fetal calf serum. After incubation for 24 h at 37 °C in 5 % CO2, the non-migrating cells were removed from the upper surface of the membrane by scrubbing. The cells on the reverse side were stained with 0.1 % crystal violet, and migrating cells were counted under a microscope at ×100 magnification.

Wound-healing assay

Scramble control or GRP78KD HT-29 or DLD-1 cells were seeded into a six-well plate and allowed to grow to 80 % confluence in RPMI 1640. Subsequently, the cell monolayers were wounded using a 10-μL pipette tip. The wounded monolayers were washed four times with phosphate-buffered saline and incubated in RPMI 1640 for the indicated time periods.

Invasion assay

The BD BioCoat™ Matrigel Invasion Chamber (BD Biosciences) was utilized to determine the invasiveness of cells. DLD-1 cells (1 × 105) expressing scrambled control or GRP78KD suspended in 500 μL of serum-free RPMI 1640 were seeded into the upper compartments of each chamber, and the lower compartments were filled with 1 mL of RPMI 1640 containing 10 % fetal calf serum. After incubation for 24 h at 37 °C in 5 % CO2, the non-migrating cells were removed from the upper surface of the membrane by scrubbing. The cells on the reverse side were stained with 0.1 % crystal violet, and invading cells were counted under a microscope at ×100 magnification.

Evaluation of cell migration using the x’CELLigence biosensor system

Experiments were carried out using the RTCA DP instrument (Roche Diagnostics GmbH, Germany), which was placed in a humidified incubator maintained at 5 % CO2 at 37 °C. Cell migration was assessed using specifically designed 16-well plates (CIM-plate 16; Roche Diagnostics GmbH) with 8 μm pores. These plates are similar to conventional Transwell plates but with microelectrodes located on the underside of the membrane of the upper chamber. FCS medium (10 %) was added in the lower chamber, and cells were seeded into the upper chamber at 20,000 cells/well in serum-free medium. The CIM-plate 16 was monitored every 10 s for 40 min and then once every hour. Data analysis was carried out using RTCA v. 1.2 software, which was supplied with the instrument.

Protein extraction and western blot analysis

The method for total protein extraction was described previously [14, 23]. The cells were lysed using cell lysis buffer containing protease inhibitors (Complete Protease Inhibitor Tablets; Boehringer Mannheim, Indianapolis, IN) [14, 23]. The proteins (25 μg) in each sample were separated using 10 % SDS–PAGE under reducing conditions and electrotransferred onto polyvinylidene difluoride membranes (GE Healthcare). The membranes were incubated overnight at 4 °C with a primary antibody against a specific target and subsequently probed with a horseradish peroxidase-conjugated secondary antibody (1:5,000). The products were visualized with an enhanced chemiluminescence reagent (GE Healthcare) and detected using a VersaDoc 5000 (Bio-Rad Laboratories, Hercules, CA) [26].

Statistical analysis

All experiments were repeated a minimum of three times. All data collected from real-time RT–PCR and cell proliferation experiments are expressed as the means ± SD. The data presented in some figures are derived from a representative experiment that was quantitatively similar to the replicate experiments. When two groups of datasets were compared, statistical significance was determined using a two-tailed Student’s t test. Asterisks in the figures indicate significant differences between the indicated experimental groups and the corresponding control conditions (p < 0.05, see figure legends).

Results

Silencing of GRP78 enhanced the migratory ability of DLD-1 cells

To confirm the role of GRP78 in cancer metastasis, we used small interfering RNA (siRNA) to suppress the expression of GRP78 in HT-29 and DLD-1 colon cancer cells. As shown in Fig. 1a, the levels of GRP78 were suppressed by over 80 % in GRP78-siRNA (GRP78KD) DLD-1 cells compared with scrambled control DLD-1 cells. To further evaluate the influence of GRP78 on colon cancer cell migration, Transwell migration, wound-healing, and invasion assays were performed. As shown in Fig. 1, we found that migratory cells were more prevalent among GRP78KD DLD-1 cells than among scrambled control DLD-1 cells (Fig. 1b, c). We further confirmed the migration assay results using the x’CELLigence biosensor system and found that knockdown of GRP78 enhanced the cell migration of DLD-1 cells (Fig. 2a); the difference was statistically significant (Fig. 2b).

Fig. 1
figure 1

Silencing GRP78 expression enhanced cell migratory ability in a Transwell migration assay system. GRP78 expression was suppressed using small interfering RNA (siRNA). GRP78-siRNA and scrambled control RNA were transfected into DLD-1 cells, and transfected cells were selected with antibiotics. a The levels of GRP78 in scrambled control and GRP78-knockdown (GRP78KD) cells were determined by western blotting. b The migratory ability of scrambled control and GRP78-knockdown (GRP78KD) cells was determined using a Transwell migration assay as described in the “Materials and methods” section. The migrated cells were observed and counted under a microscope

Full size image
Fig. 2
figure 2

Silencing GRP78 expression was found to enhance the migratory ability of cells using the x’CELLigence biosensor system. The x’CELLigence biosensor system was used to evaluate cell migration ability in DLD-1 cells transfected with scrambled control and GRP78KD. Cell migration was assessed using specially designed 16-well plates (CIM-plate 16; Roche Diagnostics, GmbH) with 8-μm pores and carried out using the RTCA DP instrument (Roche Diagnostics GmbH, Germany). The CIM-plate 16 was monitored every 10 s for 40 min and then once every hour. Data analysis was carried out using RTCA software 1.2, which was supplied with the instrument. The Y-axis represents the cell index

Full size image

GRP78 silencing enhanced the migratory ability of colon cancer cells in a wound-healing assay

We also used a wound-healing assay to observe the influence of GRP78 on the migration ability of colon cancer cells. The lower migration ability in HT-29 cells compared with DLD-1 cells was also observed (Fig. 3). After 24 h, the gap between the wound edges was smaller in GRP78KD DLD-1 cells than in scrambled control DLD-1 cells (21 vs. 57 %) (Fig. 3b). This phenomenon was also observed in HT-29 colon cancer cells (Fig. 3a). These results indicate that silencing GRP78 may enhance the wound-healing migration ability of colon cancer cells.

Fig. 3
figure 3

GRP78 mediated migration of HT-29 and DLD-1 cells in a wound-healing migration assay. GRP78 expression was suppressed using small interfering RNA (siRNA) in HT-29 and DLD-1 cells. A wound-healing migration assay was performed to compare the migratory ability between scrambled control and GRP78-knockdown (GRP78KD) cells in a HT-29 and b DLD-1 cells. DLD-1 or HT-29 cells transfected with scrambled control or GRP78 KD were seeded into a six-well plate and allowed to grow to 80 % confluence in RPMI 1640. Then, the cell monolayers were wounded using a 10-μL pipette tip. The wounded monolayers were washed four times with phosphate-buffered saline and incubated in RPMI 1640 for the indicated time periods

Full size image

Suppression of GRP78 expression enhanced cells’ invasive ability

The process of metastasis involves several different stages. To further confirm the role of GRP78 in cancer metastasis, an invasion assay was performed. As shown in Fig. 4, the invasive ability was increased dramatically in GRP78KD DLD-1 cells compared with scrambled control cells. These results indicate that suppression of GRP78 may enhance cancer invasiveness. Taken together, our results demonstrate that silencing GRP78 enhances the invasiveness of colon cancer cells.

Fig. 4
figure 4

Knockdown of GRP78 expression enhanced the cells’ invasive ability. The BD BioCoat™ Matrigel Invasion Chamber (BD Biosciences) was applied to determine the invasiveness of cells. Scrambled control or GRP78 KD DLD-1 cells (1 × 105 cells each) were seeded into the upper compartments of each chamber. After incubation for 24 h, the non-migrating cells were removed from the upper surface of the membrane by scrubbing. Cells on the reverse side were stained with 0.1 % crystal violet, and migrated cells were counted under a microscope at ×100 magnification

Full size image

Knockdown of GRP78 resulted in a change in epithelial-to-mesenchymal transition (EMT) marker expression and led to activation of the NRF-2/HO-1 pathway

To dissect the mechanism underlying the phenomena described above, epithelial-to-mesenchymal transition (EMT) markers were evaluated. After the successful suppression of GRP78 expression, E-cadherin expression was down-regulated without a change in N-cadherin levels. The expression of vimentin, a mesenchymal marker, was markedly up-regulated (Fig. 5a). A previous study demonstrated that silencing of GRP78 may cause an increase in NRF-2 expression, which is correlated with sensitivity to epirubicin [16]. It is known that NRF-2 plays a role in mediating cell migration ability [27]. We further evaluated the expression levels of NRF-2 and HO-1 in colon cancer cells. As shown in Fig. 5b, we found that the level of NRF-2 was increased dramatically after silencing GRP78. The expression of the downstream molecule HO-1 was also up-regulated after GRP78 knockdown in colon cancer cells. These results indicate that the knockdown of GRP78 enhanced colon cancer metastasis through alterations in EMT biomarker expression and the activation of the NRF-2/HO-1 pathway.

Fig. 5
figure 5

Silencing of GRP78 altered EMT biomarker expression and activated the NRF-2/HO-1 pathway. a The levels of E-cadherin and vimentin in scrambled control and GRP78-knockdown (GRP78KD) cells were determined by western blotting. GAPDH was used as the control. b The levels of NRF-2 and HO-1 in scrambled control and GRP78-knockdown (GRP78KD) cells were determined by western blotting. GAPDH was used as the control

Full size image

GRP78 expression levels in primary and metastatic colonic adenocarcinomas

To better understand the role of GRP78 in colon cancers, we first compared the expression levels of GRP78 in normal colonic epithelia and primary colonic adenocarcinomas. Normal colonic glandular epithelial cells usually showed low or intermediate GRP78 expression scores (Fig. 6a). Primary colonic adenocarcinoma showed stronger GRP78 immunostaining compared to normal colonic epithelium (Fig. 6b). To clarify the role of GRP78 in colon cancer metastasis, GRP78 expression levels were compared between primary colon adenocarcinomas and lymph node metastases. GRP78 expression in metastatic adenocarcinoma in lymph nodes tended to be weaker than that in primary colonic adenocarcinoma (p < 0.01, Table 1, Fig. 6b, c). We further evaluated whether GRP78 expression levels were associated with differentiation of carcinoma cells. In the microarrays comprising low- and high-grade colonic adenocarcinomas, significantly stronger GRP78 expression was observed in low-grade adenocarcinomas (p < 0.01, Fig. 6d, e).

Fig. 6
figure 6

GRP78 expression in normal colonic mucosa, primary colonic adenocarcinoma, and metastatic colonic adenocarcinoma in lymph nodes. a Intermediate GRP78 expression in normal colonic epithelium. b High GRP78 expression in primary colonic adenocarcinoma. c Intermediate GRP78 expression in metastatic colonic adenocarcinoma in a lymph node. High-grade colonic adenocarcinoma (e) tended to have weaker GRP78 expression than low-grade colonic adenocarcinoma (d). Original magnification ×100

Full size image
Table 1 GRP78 expression of colonic adenocarcinoma in primary site, distantly metastatic site, and lymph node metastasis
Full size table

Discussion

Colorectal cancer is the third most common cancer among both men and women in the United States and is the second leading cause of cancer-related death [28]. According to statistical data from the Department of Health, colorectal cancer has become the most common cancer in Taiwan. In addition, it is also the third most commonly diagnosed malignancy in the world [29]. Colorectal cancer can be surgically cured when it is confined to a primary site, and metastasis is the main cause of death in colorectal cancer patients. In this study, we demonstrated that metastatic colon cancer cells had lower GRP78 expression than colon cancer cells at the primary site. With the down-regulation of GRP78 expression, colon cancer cells gained migration ability through EMT (Fig. 7).

Fig. 7
figure 7

A summary schematic diagram illustrating the role of GRP78 in the migration of colon cancer cells. Suppression of GRP78 expression may influence the EMT program (via a decrease in E-cadherin expression and increase in vimentin expression). In addition, the NRF-2/NO-1 pathway may be activated after GRP78 silencing. The migratory ability increases due to the enhancement of the EMT program and the activation of the NRF-2/NO-1 pathway after GRP78 suppression

Full size image

A previous study found that GRP78 caused resistance to chemotherapy-induced apoptosis in colon cancer [16]. GRP78 down-regulators caused cell death in both glucose-deprived and etoposide-resistant HT-29 colon cancer cells, and it was found that the down-regulation of GRP78 suppressed colon cancer growth through decreased angiogenesis [13, 17, 30]. Although GRP78 typically localizes to the endoplasmic reticulum, cell surface-localized GRP78 is regarded as a possible anticancer treatment target [31]. GRP78 overexpression was noted in colon cancers, but the role of cell surface GRP78 in colon cancer metastasis has remained controversial [19, 20]. Interestingly, GRP78 expression was negatively correlated with lymphatic invasion in a study of human colorectal cancer specimens [18]. In this study, we found that the suppression of GRP78 in colon cancer cells increased their migration ability, increased vimentin expression, and decreased E-cadherin expression.

The epithelial-to-mesenchymal transition is a hallmark of cancer metastasis. E-cadherin is the major epithelial marker and is a key player in cell polarity and organization of the epithelium [32]; it is often decreased during the progression from adenoma to carcinoma and the development of tumor metastasis [33]. Down-regulation of E-cadherin was observed to accompany an increase in migration ability in colon cancers [34–36]. In this study, suppression of GRP78 was associated with the down-regulation of E-cadherin, which may contribute to enhanced migration ability.

Vimentin is an intermediate filament protein. All intermediate filaments share a common tripartite structure consisting of a highly conserved central α-helical rod domain and variable N-terminal head and C-terminal tail domains [37]. Focal contacts participate in the movement of cells on a substratum. The architecture of focal contacts is disturbed in vimentin-deficient fibroblasts, and vimentin regulates focal contacts and helps to stabilize cell–extracellular matrix (ECM) adhesions in endothelial cells; all of these observations support a role for vimentin in regulating focal adhesions [38, 39]. Vimentin is one epithelial-to-mesenchymal transition marker, and it has recently been considered to be a prerequisite for EMT induction [40]. Although the majority of colorectal cancers (53–84 %) carry an aberrantly methylated vimentin gene, vimentin expression may be increased in colon cancers in certain situations [41, 42]. In this study, the suppression of GRP78 was associated with up-regulation of vimentin, which may contribute to enhanced migration ability in colon cancer cells. Calcium homeostasis and influx have been widely reported to be involved in the mesenchymal transition. Calcium signals play a pivotal role in cancer cell migration and invasion [43, 44]. It is reported when increasing the concentration of cytosolic calcium would promote the process of cancer malignant migration through calcium-related proteins, such as Orail and STIM1 [45]. Chelating intracellular calcium would abolish EGF or hypoxia-induced EMT in breast cancer cells [46]. Previous studies have indicated that EGF can induce calcium signaling via store-operated calcium channels along with subsequent regulation of cell proliferation and migration [47–50]. In addition, KCNN4 channels participate in the EMT induced by PRL-3 in colorectal cancer [51]. However, the correlation between GRP78 and calcium influx remains unclear. It will be worthwhile to further dissect this relationship.

Heme oxygenase (HO) regulates the intracellular heme level by cleaving heme into carbon monoxide (CO), biliverdin, and free iron [52], and there are three HO isoforms: HO-1, HO-2, and HO-3. HO-1, as a heat shock protein-32, controls the rate-limiting step in the degradation of heme, and it is an inducible enzyme with relatively low expression in most tissues. HO-1 also serves as a defense mechanism against oxidative stress. Furthermore, HO-1 regulates cell proliferation, modulates the inflammatory response, and facilitates angiogenesis. Previous studies have revealed that HO-1 expression is increased in colon cancer tissue [53, 54]. HO-1 is a target gene of the basic leucine zipper (bZIP) transcription factor NRF-2, which regulates the adaptive response to oxidative stress in multicellular organisms. Under normal conditions, NRF-2 remains transcriptionally inactive in the cytoplasm, and its protein level is kept low through its degradation by the proteasome [55]. Previous reports demonstrated that increased NRF-2 expression is associated with poor prognosis in various cancers [56–58] and that NRF-2 expression increases chemoresistance in colon cancer [16, 59]. A recent study also found that colon cancers with higher levels of NRF-2 expression had an increased incidence of lymph node and distant metastases [60]. In this study, we demonstrate that the suppression of GRP78 increases NRF-2 and HO-1 protein levels in colon cancer cells (Fig. 7), which may contribute to their increased capacity for metastasis.

In conclusion, our work indicates that GRP78 may suppress colon cancer cell migration. GRP78 silencing can enhance the expression of vimentin and decrease the expression of E-cadherin in colon cancers. In addition, the expression levels of NRF-2 and HO-1 expression are elevated after the suppression of GRP78 expression. Our results provide further information about the mechanisms associated with the progression and regulation of colon cancers.