Background
Tumor cells that reside or metastasize to the bone and BM can develop pro-survival interactions with stromal cells, including osteoblasts (OSB) at the endosteum, through adhesion molecules. Evidence suggests that recurrent and resistant malignant stem cells can remain relatively protected within the bone microenvironment during treatment and later re-initiate growth [
1‐
3]. In particular, we have shown that N-cad is a necessary mediator of CD138
+ patient-derived multiple myeloma (MM) cells adhesion to the endosteum, and that down-regulation of N-cad in OSB decreased MM-OSB adhesive interactions, restricting the ex vivo survival of these tumor cells [
4]. These adhesive interactions are considered to be major factors by which cancer cells remain “dormant” and escape the cytotoxic effects of therapeutic agents. The mechanism of drug resistance has been described in MM, as well as in disseminated/metastatic prostate cancer (PCa) cancer [
5‐
7]. Bone is a preferred site for PCa metastases, and currently no curative treatments exist once the tumor is established within this niche [
8‐
10].
The 78 kDa glucose-regulated protein (GRP78) is a chaperone protein that serves as the main sensor for misfolded proteins in the endoplasmic reticulum (ER) and triggers the unfolded protein response (UPR) [
11]. Additionally, GRP78 regulates intracellular signaling events associated with embryonic development, aging, Ca
2+ homeostasis and insulin/IGF-1 signal transduction [
11]. GRP78 is expressed ubiquitously in all cell types, and is located primarily in the endoplasmic reticulum where it chaperones protein folding activity, in the mitochondria where it interacts with pro-survival and apoptotic executors and at the cell surface where it directs cell signaling [
12]. In cancer, GRP78 overexpression leads to a variety of signaling pathways associated with tumor initiation, proliferation, adhesion and invasion [
12,
13]. Importantly, GRP78: 1) is highly active in osteoblastic, androgen-independent prostate cancer [
14], suggesting that it might play a pivotal role in the interaction of PCa cells with OSB, 2) plays a critical role in the adhesion and invasion of hepatocellular carcinoma [
15] and MM [
16], 3) can mediate resistance against cytotoxic chemotherapy in PCa cells [
17], and 4) is overexpressed in a quiescent MM cell sub-population resistant to treatment [
18,
19]. Of note, GRP78 has been correlated with the expression of N-cad, E-cad and β-catenin, respectively in colon cancer and hepatocellular carcinoma [
15,
20,
21]. While high levels of N-cad have been linked to poor prognosis of MM patients [
22,
23] and to PCa metastasis and castration resistance [
24], no studies have examined the potential interplay between GRP78 and N-cad in MM and PCa for modulating tumor-bone adhesion.
GRP78 overexpression can affect the EMT signaling pathways related to Snail/Slug and TGF-β/smad. These pathways are closely associated with metastasis of epithelial tumors [
15,
21]. Classically, an EMT is associated with the upregulation of mesenchymal markers such as N-cad and vimentin along with parallel downregulation of the epithelial marker E-cad [
25,
26]. In this study, we described a novel correlation between GRP78 and N-cad in MM and PCa cells where GRP78 KD induced the concomitant downregulation of both N-cad and E-cad, which decreased the adhesive interactions of metastatic PC3 cells with OSB. These results may contribute to a better understanding of the underlying survival properties conferred by cell-cell adhesions, aiding in the development of more effective therapeutic strategies against cancers that interact with the bone niche [
9,
27‐
30].
Methods
ONCOMINE data mining
The ONCOMINE repository (
https://www.oncomine.org/) is a repository of cDNA microarrays [
31]. We searched ONCOMINE using the following filters: Gene: HSPA5 (GRP78) or CDH2 (N-cad), Analysis Type: Cancer vs. Normal Analysis, Cancer Type: prostate cancer or multiple myeloma. A summary table containing fold change and significance (
P < 0.05) for each comparison, are presented as log2 median-centered intensity as reported by ONCOMINE. References from independent studies presented can be found in the Additional file
1.
Cell culture
MM.1S (MM cell line, ATCC®CRL-2974), MM.1R (MM cell line, ATCC® CRL-2975), RPMI 8226 (ATCC® CCL-155), PC3 (bone metastatic PCa cell line, ATCC®-CRL-1435), and hFOB 1.19 (OSB cell line, ATCC® CRL-11372) were purchased from ATCC. MM cell lines were cultured in high glucose RPMI-1640 medium supplemented with 15% fetal bovine serum (FBS), 2.5 mM of L-glutamine and 1% penicillin/streptomycin. PC3 cells were cultured in RPMI-1640 medium containing 10% FBS, 2.5 mM of L-glutamine and 1% penicillin/streptomycin. Both cell lines were cultured at 37 °C in a humidified tissue culture incubator containing 5% CO2. hFOB 1.19 cell medium consisted of 1:1 mixture of Ham’s F12 Medium Dulbecco’s Modified Eagle’s Medium, with 2.5 mM L-glutamine, 10% FBS and 0.3 mg/ml G418 (Sigma-Aldrich). These cells were maintained and propagated at 33 °C and 5% CO2 except during co-culture experiments which were conducted at 37 °C. All cell lines were periodically checked for mycoplasma using MycoAlert™ Mycoplasma Detection Kit (Lonza). Authentication of cell lines was performed by STR DNA profiling analysis conducted by the Molecular Resources Facility at Rutgers University. Cell populations were frozen after 3 passages from the time of initial receipt and growth and were discarded after 30 passages.
ER stress induction
For ER stress induction, MM.1S cells were seeded at a density of 7.5 × 105–1.0 × 106 cells/well in a 24-well plate. Cells were treated for 18 h with 10 nM bortezomib (BTZ) (Caymen Chemicals), 1 nM thapsigargin (Tg) (Sigma-Aldrich) or dimethyl sulfoxide (DMSO, Sigmal-Aldrich) as vehicle controls. Total RNA was isolated and subsequently analyzed via qRT-PCR.
siRNA transfection
HSPA5 (GRP78) targeting siRNAs were purchased from Ambion (Carlsbad, USA). Two different siRNAs;
Silencer® Select Pre-designed siRNA s6979 (5’ UUC UGG ACG GGC UUC AUA Gtt 3′) and s6980 (5’ UCU AGU AUC AAU GCG CUC Ctt 3′) targeting exons 6 and 8, respectively, were tested. For control, the
Silencer® select negative control No. 2 siRNA was used (Ambion). siRNA transfections were performed using a modified reverse transfection technique [
32] using a cocktail containing equimolar quantities of each GRP78 siRNA to maximize silencing potential. The GRP78 siRNA cocktail (or siRNA control) was diluted in Opti-MEM reduced serum medium and incubated with the TransIT-X2 dynamic delivery system (Mirus Bio) according to the manufacturer’s protocol. The siRNA-TransIT-X2 complexes were added to wells of either a 6- or 24- well plate upon which either MM or PC3 cells seeded in complete growth medium at a cell density of 7.5-9 × 10
5 cells/well (6 well plate) or 0.75-1 × 10
5 cells/well (24 well plate). GRP78 siRNA cocktail or control siRNA were used at a final concentration of 50 nM for PC3 and 100 nM for MM cell lines.
RNA isolation and qRT-PCR
Total RNA was isolated following transfections (48 h) from TriZol (Ambion) preserved cells using a TriRNA Pure Kit (Geneaid), following the manufacture’s instructions. The collected RNA was quantitated on a Qubit 3.0 fluorimeter using the Qubit Broad Range (BR) assay kit (Thermo Fisher Scientific). RNA (200 ng) was reverse transcribed into cDNA using a high capacity cDNA kit (Applied Biosystems). RT-PCR was performed using pre-developed TaqMan™ gene expression primer-probes for GRP78 (assay ID Hs99999174_m1), N-cad (assay ID Hs00983056_m1), GRP94 (assay ID Hs00437665_g1), GRP75 (Hs00269818_m1), and GAPDH (Hs99999905_m1) and TaqMan™ fast advanced master mix. qPCR fast assay was carried out on a StepOnePlus RT-PCR system (Applied Biosystems). Fold changes were calculated with the ΔΔCt method using GAPDH as endogenous control and the negative siRNA as the control sample.
Western blot
Total protein was isolated from the cell cultures following transfection (78 h). Protein lysates were prepared by lysing the cells in ice-cold RIPA buffer (G-Biosciences) supplemented with protease and phosphatase inhibitors (Millipore Sigma) which were diluted 1:10 as per the manufacturer’s recommendations. Cell debris was removed by centrifugation at 16,000×g at 4 °C and protein concentrations were determined using a Pierce™ BCA kit (Thermo Fisher Scientific). A sample (20–35 μg) of the supernatant protein was mixed with LDS buffer and DTT, incubated at 70 °C for 10 min and resolved on a 4–12% Bis-Tris PAGE gradient gel before being transferred to a PVDF membrane. Following transfer, the membrane was blocked in 5% skim milk for 1 h, washed and incubated at 4 °C overnight with a rabbit 1° mAb against human GRP78, GRP94, GRP75, N-cad, E-cad, TGF-β1, Slug, B-catenin or GAPDH (all purchased from Cell Signaling Technology) at a 1:1000 dilution. The membrane was subsequently washed and incubated with an anti-rabbit HRP-conjugated 2° Ab (Cell Signaling Technology) for 1 h at room temperature at 1:2000 dilution. The bands were visualized using a SignalFire™ ECL reagent (Cell signaling Technology) on a ProteinSimple FluorChem E imager. No changes in GAPDH band intensity between control siRNA and GRP78 siRNA were detected, therefore target protein bands were normalized against the loading control GAPDH.
Flow cytometry and cell viability
Cell viability was assessed using an Annexin V/PI kit (Biolegend). Annexin V and PI were added to the cell samples post-transfection at 24, 48, and 72 h according the manufacturer’s recommendation, incubated for 15 min at room temperature in the dark, and followed by immediate analysis by flow cytometry (FC500 flow cytometer, Beckman Coulter). Data was processed with Kaluza® (Beckman Coulter) flow analysis software.
Cell morphology assay
For cell morphology analysis, PC3 cells were transfected with GRP78 siRNA and harvested 48 h later or treated overnight with 20μg/mL of a N-cad neutralizing monoclonal antibody, clone CG-4 (N-cad NAb, Sigma-Aldrich) prior to analysis. The cells were counted on Z2 Particle Counter and Size analyzer (Beckman Coulter) and re-seeded at 20,000 cells/well in a standard 96 well plate and left to culture for an additional 18 h. The wells were subsequently washed and imaged under bright field on a Cell Insight CX5 high content screening instrument (Thermo Fisher Scientific). Images were analyzed using ImageJ software package (NIH). 15 fields of view (5 each from 3 independent experiments) of control cells or cells treated with GRP78 siRNA were analyzed using the ImageJ software package (NIH). Cells with a clearly defined spherical and darker border (under bright field) were considered as rounded. Morphology was calculated as rounded: Nrounded cells/Ntotal cells or elongated: [Ntotal cells-Nrounded cells]/Ntotal cells and represented as a mean percentage ± SD.
Adhesion assay
To determine the effects of GRP78 KD on PC3 cells adhesion to bone, PC3 cells were transfected with GRP78 siRNA and harvested 48 h later and cultured with hFOB 1.19 cells plated in a 96 -well plate until confluent. To track the PC3 cells in co-culture, transfected cells were labeled with the Vybrant CFDA SE (green) Cell Tracer Kit (5 μM) (Thermo Fisher Scientific) counted on a Z2 particle counter and size analyzer (Beckman Coulter), and then seeded at 10,000 cells/well onto confluent OSB. For comparison, parallel co-cultures were treated overnight with 20μg/mL of a N-cad NAb, clone CG-4 (Sigma-Aldrich). Number of fluorescently labeled cells [Ncoculture] were counted by high content screening (Cell Insight CX5, Thermo Fisher Scientific) following an 18 h co-incubation period. Supernatants were then transferred from co-culture wells into empty ones to determine the number of fluorescently labeled floating cells [Nsupernatant]. Adherence was calculated as %Nsupernatant/ Nco-culture.
Statistical analysis
All data was plotted and analyzed using the GraphPad Prism software, V 7.0d (La Jolla, CA). Each experiment was performed in triplicates (N = 3). Data is represented as the mean ± SD. Comparisons between two groups were analyzed using unpaired student’s t-tests. A probability (P) value of less than 0.05 was considered statistically significant.
Discussion
In this study, we describe an important correlation between GRP78, a master regulator of the UPR, and N-cad, an adhesion molecule associated with MM and PCa progression [
43‐
45] and cell-cell adhesion implicated in drug resistance [
24,
46]. N-cad expression has been shown to be directly proportional to GRP78 levels in hepatocellular carcinoma and colorectal cancer [
15,
21]. Furthermore, circulating levels of N-cad have been linked to poor prognosis of MM patients [
22,
23] while upregulation of this molecule was linked to PCa metastasis and castration resistance [
24]. Together these findings served as basis for our hypothesis, implicating GRP78 and N-cad as important markers involved in to tumor adhesion and metastasis of PCa with bone tissue. To test this hypothesis, we investigated the effect of GRP78 KD in MM cell lines (MM.1S, MM.1R, RPMI 8226) known to reside in the BM niche and a metastatic PC3 cell line, derived from metastatic bone. Our results suggest that the pro-survival advantages conferred by GRP78 [
47,
48] may also be linked to its role in modulating markers associated with cancer cell adhesion to the bone niche.
An initial survey of basal GRP78 and N-cad levels revealed comparable expressions of these proteins in MM.1S and PC3 cancer cell lines (Additional file
3, A). While GRP78 upregulation is well documented in many cancers [
14,
18,
19], MM and PCa cells often display aberrant N-cad expressions [
23,
24,
45,
49]. We found that GRP78 KD induced a concomitant suppression of N-cad protein levels, suggesting a regulatory relationship between these two biomarkers in MM.1S and PC3 cells (Fig.
1). Interestingly, the gene transcript levels of N-cad were not affected upon GRP78 downregulation, suggesting preferential downregulation of N-cad translation over transcriptional modulation. Importantly, only one of three MM cell lines (MM.1S) assayed showed appreciable GRP78 KD at high dose concentrations of siRNA demonstrating not only the difficulty of siRNA based gene silencing in MM, but also the dependence of MM on this important chaperone of the ER/UPR systems [
37]. Alternatively, the MM.1S cells were found to be susceptible to apoptosis with the highly concentrated and long-lasting siRNA transfections (100 nM, 24-72 h) rendering further follow-up experiments unfeasible. The cancer cell lines studied (i.e., MM.1S and PC3 lines) did not manifest cell membrane bound GRP78 (data not shown) and have been shown in other reports to lack surface GRP78 under specified culture conditions [
50]. Thus, our results likely indicate a role for cytosolic rather surface GRP78 on N-cad expression. However, surface expression of GRP78 has been observed in varying tumor cell lines and with ER stressors known to upregulate surface GRP78 expression [
39] allowing for potential future studies aimed at investigating the effects of exogenous GRP78 blocking antibodies on N-cad activity. Likewise, additional methods [
34] which fall beyond the scope of the current study may be needed to fully characterize the functional relationship in between GRP78 and N-cad among multiple MM and PCa cell lines.
We queried the Oncomine microarray repository to find any correlations in the expression profiles of HSPA5 (GRP78) and CDH2 (N-cad) in patient tissue samples. The ONCOMINE analyses revealed that both HSPA5 and CDH2 genes are overexpressed in various MM and PCa compared to normal tissue samples. In particular, HSPA5 was upregulated in 40% (6/15) and in 66% (2/3) of the clinical studies available for prostate carcinoma and MM tissues, respectively. For MM, the fold change was small likely because plasma cells and leukocytes (used as comparison samples) are secretory cells that also have high basal levels of the UPR markers [
51]. CDH2 on the other hand was only significantly upregulated in one PCa study. That notwithstanding, a study describing the de novo expression of N-cad using two parameter immunofluorescence shown that this protein is expressed in high-grade human PCa, whereas no expression was found in normal prostatic tissue [
52]. Likewise, circulating N-cad has also been associated with poor MM prognosis [
22], supporting an underlying relationship between N-cad and GRP78 in the progression of both tumor types.
The lack of a strong transcriptional downregulation in the MM.1S cells prompted us to evaluate how the expression profiles of GRP78 and N-cad would be affected in the presence of acute drug challenge using the ER stress inducers BTZ and Tg; both of which are known to stimulate GRP78 expression and are used as current clinical treatments. We also wanted to examine whether ER stress would sensitize MM.1S cells to siRNA transfection, leading to a more robust KD effect. This strategy may be a potential clinically viable approach to facilitate the transfection of UPR dependent cancer cells. Acute drug challenge was found to induce GRP78 mRNA expression in MM.1S cells, with a concomitant decrease in N-cad mRNA levels (Fig.
2a). BTZ has been found to suppress the expression of N-cad [
53] but little is known about the effects of Tg on this molecule. N-cad expression has been associated with intracellular Ca
2+ flux, so inhibition of the ATPase-dependent Ca
2+ flux by Tg may be correlated with its effects on N-cad transcription levels [
54]. Nevertheless, when MM.1S cells were pre-treated with BTZ followed by GRP78 siRNA transfection, GRP78 KD caused an additional decrease in N-cad (Fig.
2b) gene expression. Of note, siRNA treatment did not affect genetic levels of N-cad (Fig.
1c) but the combination of ER stressors (i.e., drug treatments) and siRNA transfection did induce significant downregulation in N-cad mRNA levels, suggesting that ER stressors (such as that induced by BTZ) may sensitize MM cells to siRNA treatment.
Unlike MM, which is an exclusively BM-localized malignancy, PCa is a solid epithelial tumor that seeds and invades the bone/ BM niche. GRP78 expression has been linked to cancer progression and metastasis in part through its effects on EMT markers in PCa [
55]. Induction of GRP78 has been shown to trigger EMT in colorectal cancer cells, while GRP78 KD using shRNA reversed the EMT by suppressing N-cad and upregulating E-cad expression levels, referred to as a “cadherin switch.” [
21]. However, using our GRP78-silencing approach, GRP78 KD in PC3 cells resulted in significant decreases of both N-cad and E-cad (Fig.
1 & Fig.
4) protein levels. The EMT process is controlled, in part, by the transcription factor Snail-2 which acts as a strong repressor of E-cad [
56,
57]. GRP78 KD induced Snail-2 expression (Fig.
4) which may account for the downregulation of E-cad.
GRP78 overexpression, its localization to the cell surface, and its association with Cripto [
58,
59] have been correlated with the activation of the TGF-β pathway [
60]. TGF-β is a multifunctional cytokine which regulates prostate cell growth and epithelial cell proliferation [
61,
62]. However, active TGF-β exists mainly as an extracellular matrix protein which can function both as a tumor suppressor or as a key player in promoting tumorigenesis in advanced cancers [
61‐
64]. We showed that in PC3 cells, GRP78 KD induced a strong and significant increase in TGF-β1 protein expression; consistent with our findings, induction of Snail-2 expression has been credited to TGF-β associated pathways [
65]. While we observed TGF-β1 upregulation and the expected downstream effects that this molecule has on Snail-2 and E-cad, we found that GRP78 silencing decreased N-cad expression in PC3 cells. TGF-β1 upregulation is typically reported to be accompanied by N-cad upregulation [
21], hence our results suggest that N-cad expression is highly dependent on GRP78 in this cell line and its regulation via GRP78 KD may supersede the effects of TGF-β1; i.e., N-cad was downregulated in spite of the fact the TGF-β1 expression was significantly increased upon GRP78 KD. This may indicate a unique mechanism in which GRP78 KD can directly modulate the expression of certain adhesion and EMT molecules. We also showed that downregulation of GRP78 did not change the protein expression levels of other chaperone GRPs nor cause cytotoxicity in PCa. These results confirm that the observed effects on the EMT markers were GRP78-dependent and not the result of global changes in closely related chaperones that also maintain ER homeostasis or the result of apoptosis.
TGF-β has also been implicated in EMT signaling through alternative non-smad pathways. For example, Ras homolog gene family member A (RhoA)-dependent signaling is activated by TGF-β and induces mesenchymal characteristics in epithelial cells [
66]. Signaling by RhoA and its effector proteins Rho kinase-ROCKI and ROCKII promote amoeboid movement of tumor cells and the adoption of a more rounded shape [
67]. Consistent with these morphological changes, we observed that following GRP78 KD, PC3 cells underwent a shift from a flatter, elongated shape to a more rounded configuration (Fig.
5) Importantly, incubation with a N-cad NAb did not significantly change the morphology of the cells (Additional file
3, B) implying that the observed changes were directly related to the diminished GRP78 levels rather than a function of N-cad activity.
We suspected that N-cad downregulation upon GRP78 KD could lead to reduced adhesion of PCa to the bone microenvironment. We confirmed this hypothesis by co-culturing PC3 cells transfected with GRP78 silencing siRNAs with a monolayer of OSB cells. The cells were found to be less adherent to OSB, relative to untreated control cells, supporting the hypothesis that GRP78 KD and suppression of N-cad could significantly inhibit PCa cell-based adhesion to bone. In a comparable manner N-cad NAb treatment also diminished PCa adhesion to OSB (Fig.
6). That notwithstanding, other adhesion molecules have been reported to play a role in PCa-bone interaction [
68], potentially accounting for the only moderate decrease in PC3 adhesion to OSB in our assays.