Background
Despite the modern advances in cancer therapy, PDAC remains a devastating disease, owing to its late and difficult diagnosis on the one hand and the aggressiveness of the PDAC cells on the other hand [
1]. Even with the low incidence of PDAC, notoriously, it ranks fourth among the cancer related deaths in the United States and Europe [
2,
3]. Although most of cancer related deaths are predicted to be declining by 2020, the death rate from pancreatic cancer will increase and be the second cause of cancer-related deaths within the next decades [
4,
5]. Therefore, a better understanding of PDAC development is still urgently needed.
PDAC is an epithelial tumour that arises from the cells of the pancreatic duct [
6] or from acinar cells undergoing acinar to ductal metaplasia (ADM) thus, exhibiting a ductal phenotype [
7]. In healthy exocrine pancreas, epithelial cells align with neighbouring cells and adhere to the basement membrane to create a well organised epithelial sheet and give rise to three-dimensional tubuloacinar glands [
8]. Moreover, the continuous layer of pancreatic ductal epithelial cells possess clear epithelial features including specialized cell-to-cell contacts of tight junction and a polarized morphology, by which the cells exhibit three types of surfaces. The basal surface interacts with the extracellular matrix, the subsequent lateral surfaces communicate with other cells, and the luminal surface faces the lumen [
8]. Therefore, a correct three-dimensional (3D) organization and tissue architecture are core requirements for tissue homeostasis, i.e. control of cellular proliferation, survival, regulating cell adherence and differentiation [
9]. Particularly in exocrine glands, the polarity of the epithelial cells is also essential for the control of the cellular absorption and secretion [
8,
10].
The pancreatic ductal system secretes an enormously bicarbonate-rich fluid, which is required to neutralize the acidic chime entering the duodenum and to provide an optimal pH microenvironment for the activity of digestive enzymes [
11]. From a physiological perspective, the ion/fluid transport causes a transepithelial osmotic gradient that directly influences the intracellular volume [
12,
13]. The secretory cells counteract the fluctuation of cellular volume, by coordinating the net transport of potassium, sodium and chloride ions across the luminal and basolateral plasma membranes [
14]. That implies that cell volume homeostasis is an essential part of the secretory function of the pancreatic ductal cells [
15]. Disruption of this homeostatic state was reported in pathophysiological conditions of renal diseases or brain ischemia, causing dysregulation of cell volume regulatory transporters (imbalance of sodium and potassium intake) and an impaired acid/base transport (reviewed in Hoffmann, Lambert, & Pedersen, 2009). Moreover, the huge acid-base fluxes across the ductal epithelium require a very efficient control of the intracellular pH homeostasis [
16]. Possibly, the ability of pancreatic cells to cope with such enormous acid-base fluxes also contributes to the aggressiveness of PDAC [
17].
The switching of cell polarity alters the localization of the transport proteins [
18]. As a result, some apical ion channels and transport proteins move to the rear end, whereas some basolateral transporter re-localize at the leading edge of the migrating tumour cells. On the one hand, that causes the dysregulation of cellular volume homeostasis, as observed in many secretory epithelia afflicted by cancers i.e. colorectal, gastric, mammary gland and pancreatic [
19‐
22]. On the other hand, it may contribute to cell migration [
23]. Therefore, a focused analysis of the transportome in differentiated/undifferentiated cells will help to define the role of ion channels and transporters in PDAC.
Hitherto, 3-dimensional (3D) culturing is not intensively investigated [
24] partly because the assessment and scalability of the biological behaviour of the cells under 3D culturing condition is still problematic. However, it is generally accepted that the in vitro organotypic 3D cell culture system better resembles the physiological condition of the tissues in vivo, in regards of i) architectural organization and the processes of glandular lumen formation, ii) cell-to-cell interaction, and iii) the role of cancer genes in cell polarity, therefore, allowing the investigation of the different aspects of tumour biology and pathophysiology [
25]. Among the 3D culture methods, the spontaneous cell aggregation is a widely used technique [
25,
26]. Hereby, malignant cells spontaneously aggregate on the substrate preventing cell adherence and promoting the formation of spheroids that grow in suspension [
25]. Beforehand, it has been described that other human pancreatic cancer cell lines (HPAF-II, HPAC and PL45) derived from PDAC [
26] can develop spheroids with a compact structure similar to avascular tumours.
The human PDAC cell line A818–6 bears an activating mutation in codon 12 of the KRAS gene [(G12R); personal communication Franziska Wilhelm, Institute of Pathology, CAU Kiel], which is the most common alteration in PDAC. The A818–6 cells can be grown in two different physical forms, 2D or 3D model, and both are significantly different. When the cells are grown under 3D culturing conditions, A818–6 cells form hollow sphere (HS) structures and when grown under 2D conditions, the cells grow as a monolayer (ML). The 3D HS structure is formed when the A818–6 cells are not allowed to adhere to the bottom of the culture flask/plate. Under these conditions, of one-layer cell spheres with a hollow centre are build, hence the name hollow spheres. In contrast, when the A818–6 cells are while the ML cells were allowed to adhere, and they grow as a monolayer ML on the bottom of the culture flask/plate. It was formerly reported that the cells in HS proliferate slower than ML and display morphological and functional polarity. Furthermore, in contrast to ML cells, they are not able to form tumours when orthotopically inoculated into SCID-mice. Importantly, A818–6 cells possess a high degree of cellular plasticity. When HS are mechanically disrupted they regrow as a ML regaining all the founder ML attributes and vice versa [
27‐
29]. This cellular plasticity enables the A818–6 cell to transform from a rather benign/differentiated cell (HS) state a fully malignant/mesenchymal (ML) state. Cellular plasticity is essential to enable cancer cells to migrate to other organs and form metastases [
30]. This cell line provides an opportunity to study the role of proteins critically involved in the process of epithelial-mesenchymal transition (EMT) and the reversal of this process, MET. In the current study, a whole genome-wide analysis of the two forms of A818–6 was performed to predict how cellular plasticity governs the malignant transformation in this cell line. Specifically, we aimed to explore the possible utility of this model in studying the role of transportome in PDAC.
Discussion
In the current study, we aimed to investigate the differences between the two A818–6 cell forms, ML and HS and to test whether the HS/ML (de)differentiation model could be useful in studying the role of transportome in PDAC development. Therefore, a whole-genome cell microarray in addition to the nCounter analyses was performed to investigate the transportome genes that may play a role in the process of malignant transformation. The expression data showed that the two forms of A818–6 are strictly distinctive. On the one hand, the 2D ML cells are more aggressive and exhibiting mesenchymal features. On the other hand, the HS structures show less aggressiveness with more epithelial characteristics. This degree of plasticity provided these cells with cancer stem cell properties that drove them in a series of transformations between the epithelial and mesenchymal cell types [
51]. These properties were confirmed via detection of protein levels of respective markers and bioinformatic analyses of the data from the two arrays. Known epithelial marker like E-cadherin (CDH1) and alpha catenin (CTNNA1) and keratin 15 (KRT15) were restored in HS. However, markers like HMGA2, CD44, Caveolin 1 and the mesenchymal marker vimentin were boosted in ML. HMGA2 is a known transcriptional regulator that facilitates the transcription of many other pro-tumoural genes and it was previously linked with shortened survival in PDAC patients [
55]. CD44 plays not only a role in EMT but also is a dedifferentiation marker that has been formerly reported to be highly expressed in anaplastic lesions and is correlated with cancer stem cells in PDAC [
56,
57] and Caveolin 1 which was formerly suggested to be considered as an aggressiveness marker in PDAC [
58]. Intriguingly, T-box transcription factor 2 (TBX2) gene, which is a key player in the development of the embryo and its overexpression has been associated with several malignancies including PDAC, was also upregulated in HS [
59]. In other words, the mesenchymal characteristics of ML and the possible involvement of EMT is not entirely correlated with metastasis, as was previously confirmed in lung cancer and PDAC [
60,
61]. Otherwise these two cell forms could represent two intermediate cell types on the scale of EMT transition.
Interestingly, we found that 22 upregulated genes in HS are targets of
miR-9. This may denote that either the level of this microRNA is boosted in cells grown as ML or that miR-9 is actively deregulated in HS and thus implicating
miR-9 – oncomir – in PDAC malignant development. The role of
miR-9 in malignant transformation is not clear so far. On the one hand,
miR-9 level is upregulated and reported to be involved in malignant progression of both hepatocellular [
62] and prostatic carcinoma [
63]. On the other hand,
miR-9 level is regarded as a tumour suppressor microRNA in breast cancer, where its induction leads to anti-proliferative, anti-invasive and pro-apoptotic effects [
64]. However, it was also implicated in the promotion of neovascularization [
65]. Here, our system suggests a role of microRNA
miR-9 in inducing the mesenchymal features in A818–6 ML cells.
Furthermore, we investigated the involvement of the transportome genes in the malignant transformation in our HS/ML model. The overall analysis of the three arrays denoted the implication of some key transportome genes in PDAC malignant transformation. Among those genes that were overexpressed in PDAC tissues/ML was GJB2 that was previously found to be barely expressed in normal pancreatic ductal epithelium, while highly expressed [
66] and correlating with poor prognosis in PDAC patients [
66,
67]. GJB2 was also suggested as a prognostic marker in pancreatic cancer [
68]. Moreover, the current analyses showed that the expression of TRPV6, SLC4A4 and KCNQ1 were down-regulated in PDAC/ML. This transcriptomic profiling of these transportome genes points out to the possible loss of differentiated secretory epithelial cells functions. Therefore, it can be concluded that the control of the resting membrane potential via KCNQ1, the vectorial bicarbonate transport via SLC4A4, as well as the epithelial fluid secretion via KCNQ1 and TRPV6 were inhibited in PDAC (tumour epithelium and ML), while maintained in the normal epithelium (normal epithelium and HS). Another function of, the calcium channel, TRPV6 in the normal pancreatic epithelium was the activation of programmed cell death [
69], and the inhibition of TRPV6 resulted in cell survival in gastric cancer cells [
70]. Under normal conditions, this calcium - permeable channel leads to a cytosolic calcium increase that results in apoptosis thus restoring the capability to control the elimination of the cells from the circulation. In other words, the down-regulation of TRPV6 in PDAC could aid the cancer cells to evade apoptosis. However, it was also recently found that TRPV6 gene was upregulated in some PDAC cells and its inhibition was correlated with decreased invasiveness and metastasis [
71,
72]. In other tumour entities, TRPV6 was also correlated with high proliferation rate of the prostate and breast cancer cells, as the calcium conductance activates the calcium/ calmodulin/calcineurin dependent transcription factor NFAT affecting the expression of the cell-cycle regulators [
73].
Another two transportome members (KCNQ1 and SLC4A4) were previously predicted to enhance pancreatic secretion. Generally, potassium channels were previously implicated in the development of cancer [
74]. In the current study, only one down-regulated potassium channel gene, KCNQ1 (K
V7.1), was detected in PDAC. Since, K
V7.1 channel was annotated as a protein engaged in the control of cell volume, stabilization of membrane potential and maintenance of electrogenic epithelial electrolyte transport. Additionally, K
V7.1 channel has been detected and characterized in many absorbing and secretory epithelia including the pancreatic ducts [
75,
76]. In these tissues, the K
V channels enhance the potassium conductance in response to cell swelling or purinergic stimulation in the epithelial transport [
76,
77]. The resulting potassium efflux stabilizes the resting membrane potential and supports chloride exit, which finally coordinates the electroneutral potassium, chloride secretion at the basolateral membrane [
76]. Therefore, KCNQ1 has been previously proposed as a tumour suppressor gene [
78], Rapetti-Mauss et al. also demonstrated a positive correlation between high KCNQ1 expression and well-differentiated epithelial cell lines and consequently with patient survival in primary stage of colorectal carcinoma [
79]. In the same study, the authors described a progressive loss of KCNQ1 with increasing mesenchymal phenotype in poorly differentiated cells, as a consequence of repression of the KCNQ1 promoter.
The lower mRNA level of the cotransporter SLC4A4 in PDAC suggests that the pancreatic bicarbonate secretion and intracellular pH were dysregulated during PDAC development. SLC4A4 is an important protein in healthy pancreatic ducts, as the acid-base homeostasis is a key mechanism to control bicarbonate-rich fluid secretion [
11]. As a bicarbonate transporter, SLC4A4 drives the transport of sodium and bicarbonate ions across the basolateral membrane of the pancreatic duct. The widely accepted secretory model describes the role of SLC4A4 in the vectorial transport of bicarbonate from the interstitium to the duct [
11,
80]. SLC4A4 expression was formerly found to be down regulated in PDAC [
81]. It should be noted that the cystic fibrosis transmembrane conductance regulator (CFTR) gene, the principal channel controlling fluid transport in secretory cells [
13,
74,
82], did not show any significant change between HS and ML cells. Consequently, the current study could show a clear difference in some key transportome genes’ expression that could be implicated in differentiation/dedifferentiation process in PDAC.
To date, there are several other 3D models that involve PDAC cell lines but each serves a distinct aim. The first 3D PDAC model was the A818–6 HS/ML system and it was introduced in 1999 [
27‐
29] to investigate the process of differentiation in the pancreatic epithelium. Another 3D model was developed using normal pancreatic ductal epithelial cell with a KRAS mutation (HPDE-E6E7) [
83]. This model aimed to characterize the early PDAC stages via growing the HPDE-E6E7 in a 3D condition by using Matrigel® as a stimulus. In the currently studied HS/ML system, it was previously reported that A818–6 HS form duct-like tube when grown in Matrigel® [
27,
29]. A third model was aimed as a drug screening system for pancreatic cancer. In this model, a crowding agent (20% methyl cellulose) was used to incite the 3D aggregates in several PDAC cell lines [
84]. Another PDAC in vitro 3D model focused on establishing a model that closely mimics PDAC stromal microenvironment [
85]. Here, methyl cellulose was used to generate compact spheroids that entailed both PDAC cells and pancreatic stellate cells that in turn produced extensive stroma (Collagen I and III, fibronectin and smooth muscle actin). Another model that was also dedicated to create a 3D PDAC model with regard to PDAC desmoplasia, widened the scope to include a ECM submerged and air-liquid model to investigate the PDAC-stroma crosstalk during the invasion process [
86]. In contrast to all these systems, our HS/ML model depends on the prevention of adhesion to elicit the 3D HS formation and it is a suitable model to investigate the involvement of the PDAC-relevant transportome in the malignant transformation.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.