Epithelial–mesenchymal transition, IP3 receptors and ER–PM junctions: translocation of Ca2+ signalling complexes and regulation of migration

During epithelial–mesenchymal transition IP3Rs relocate from tight junctions to the leading edge of migrating pancreatic cancer cells and regulate dynamics of focal adhesions. STIM1-competent ER–PM junctions position closely behind IP3Rs and, together with IP3Rs, regulate cell migration.

Disconnection of a cell from its epithelial neighbours and the formation of a mesenchymal phenotype are associated with profound changes in the distribution of cellular components and the formation of new cellular polarity. We observed a dramatic redistribution of inositol trisphosphate receptors (IP 3 Rs) and stromal interaction molecule 1 (STIM1)-competent endoplasmic reticulum-plasma membrane junctions (ER-PM junctions) when pancreatic ductal adenocarcinoma (PDAC) cells disconnect from their neighbours and undergo individual migration. In cellular monolayers IP 3 Rs are juxtaposed with tight junctions. When individual cells migrate away from their neighbours IP 3 Rs preferentially accumulate at the leading edge where they surround focal adhesions. Uncaging of inositol trisphosphate (IP 3 ) resulted in prominent accumulation of paxillin in focal adhesions, highlighting important functional implications of the observed novel structural relationships. ER-PM junctions and STIM1 proteins also migrate to the leading edge and position closely behind the IP 3 Rs, creating a stratified distribution of Ca 2 + signalling complexes in this region. Importantly, migration of PDAC cells was strongly suppressed by selective inhibition of IP 3 Rs and store-operated Ca 2 + entry (SOCE), indicating that these mechanisms are functionally required for migration.

INTRODUCTION
Pancreatic ductal adenocarcinoma (PDAC) is a leading cause of cancer-related death [1]. Epithelial-mesenchymal transition (EMT), migration and invasion are cellular processes that are crucially important for the formation of lethal metastases in this and other types of cancer (reviewed in [2][3][4]). Currently, understanding the fundamental contributions that Ca 2 + signalling makes to cell migration is an important research avenue of potential clinical relevance [5][6][7]. It is particularly relevant for identifying putative therapeutic targets that could delay or prevent the formation of metastases (reviewed in [8]). Two prominent components of the Ca 2 + signalling cascade are inositol trisphosphate receptors (IP 3 Rs) and the store-operated Ca 2 + entry (SOCE) mechanism. IP 3 Rs are intracellular Ca 2 + -releasing channels [9] which mediate responses to numerous hormones and neurotransmitters (reviewed in [10,11]). SOCE restores the Ca 2 + concentration in the ER ([Ca 2 + ] ER ) following its depletion due to the Ca 2 + -releasing activity of IP 3 Rs or other intracellular channels (reviewed in [12,13]). The endoplasmic reticulumplasma membrane junctions (ER-PM junctions) are regions of close contact between the two organelles [14][15][16] which serve as hubs for cAMP [17,18], phospholipid [19] and Ca 2 + [16,20,21] signalling. SOCE is a Ca 2 + influx mechanism triggered by Ca 2 + store depletion, which involves the oligomerization of stromal interaction molecule 1 (STIM1, an EF-hand-containing protein that serves as the ER Ca 2 + sensor), translocation of STIM1 oligomers to ER-PM junctions and opening of PM Ca 2 + channels [16,[22][23][24][25]. STIM1-competent ER-PM junctions can thus be classed as platforms for SOCE. Orai proteins are poreforming components of SOCE channels that are opened by STIM1 [22,24,26], although some types of transient receptor potential (TRP) channels are probably also involved [27]. Following [Ca 2 + ] ER depletion, Orai proteins translocate to the PM component of the ER-PM junctions where they interact with STIM and form Ca 2 + -selective channels. Ca 2 + is heavily buffered in the cytosol of most cell types (e.g. [28,29]) and therefore proximity of Ca 2 + channels to their downstream targets is frequently crucial for the efficiency and specificity of signalling. In the case of migrating cells and the regulation of migration, the relative positioning of Ca 2 + signalling complexes, proteins that define the leading edge and focal adhesions (structures responsible for interaction between the cell and extracellular matrix), is of particular interest.
occurring when PDAC cells disconnect from their neighbours and develop a migratory phenotype. Furthermore, we have characterized novel stratified localization of the Ca 2 + signalling complexes in the leading edge and their structural relationships with the components of migratory apparatus. Finally, we have revealed the functional importance of IP 3 Rs and of SOCE for cell migration.

Reagents
Xestospongin-B was provided by Dr J. Molgó (Institut de Biologie  et Technologies  DNA constructs coding for LL-FKBP-mRFP (where LL denotes a long linker, FKBP is FK506-binding protein and mRFP is monomeric RFP), CFP-FRB-LL (FRB is FKBP12-rapamycinbinding) [21] and for YFP-STIM1 with a TK (thymidine kinase) promoter [21] were gifts from Dr T. Balla (National Institute of Child Health and Human Development, Bethesda, MD, U.S.A.).
To express the proteins of interest, cells were transfected at approximately 60-70 % confluence with 1-2 μg of DNA per plasmid construct for 24 h using PromoFectin reagent (PromoKine) according to the manufacturer's instructions. For the knockdown of cellular proteins of interest, siRNA oligomers directed against human IP 3 R1, IP 3 R2 and IP 3 R3 isoforms were used. Cells were transfected at approximately 30-40 % confluence with 50 nM per siRNA oligomer for 72 h using Lipofectamine 2000 (Life Technologies) according to the manufacturer's instructions.

Immunofluorescence and visualising ER-PM junctions
Cells were seeded on to 35-mm-diameter glass-bottom dishes from Mattek or ibidi.
For visualising STIM1 puncta, cells were transfected with YFP-TK-STIM1, and 24 h later, were treated with 30 μM CPA for 1 h at 37 • C/5 % CO 2 and imaged using a confocal microscope.
Additional three PBS washes were carried out prior to imaging in PBS. In the specified experiments, Alexa Fluor 647 phalloidin was used (at 1:50 dilution).
Two different confocal microscopes were used to visualize the distribution of specific proteins and ER-PM junctions in fixed cells: Leica TCS SP2 (AOBS) confocal microscope with ×63 oilimmersion objective [NA (numerical aperture) − 1.4] and Zeiss LSM 710 confocal microscope with ×63 oil-immersion objective (NA − 1.4). The pinhole was set between 1 and 2 airy units.

Live-cell Ca 2 + imaging and uncaging
To investigate inositol trisphosphate (IP 3 )-induced Ca 2 + responses, cells were loaded with caged IP 3 and with Ca 2 + indicator Fluo-4 by incubation in the solution containing 1 μM caged IP 3 /PM and 5 μM Fluo-4-AM. A Zeiss LSM 510 confocal microscope with a ×63 water-immersion objective (NA 1.2) was utilized in these experiments; the 488 nm laser line was used to excite Fluo-4 (emission recorded at LP 505 nm), 351 nm and 364 nm laser lines were used for uncaging.
To investigate SOCE, PANC-1 cells were loaded with Fura-2 by incubation for 1 h in Fura-2-AM-containing solution. Cells were washed for 30 min to allow de-esterification of the probe. A Till Photonics Imaging system was used in these experiments. Fluorescence of cells loaded with Fura-2 was excited at 340 and 380 nm, and emission collected using a 510 nm bandpass filter. Data recorded in these experiments are expressed as the ratio of fluorescence excited by 340 nm (F340) and 380 nm (F380), after corresponding background subtraction. Changes of extracellular solution were made using gravity-fed perfusion system.

Migration assay
Boyden chambers were purchased from Corning. The pore size was 8 μm. Migration was measured in conditions of symmetrical FBS (1 % FBS in both the upper and lower chambers) and asymmetrical FBS (0 % FBS in the upper chamber and 5 % FBS in the lower chamber). After seeding, PANC-1 cells were allowed to migrate for 6 h in a humidified environment at 37 • C/5 % CO 2 in the presence or absence of the inhibitors of various components of the Ca 2 + signalling cascade; Boyden chamber inserts were then fixed using 100 % methanol and non-migrated cells were removed from the top-side of the inserts using cotton buds. The inserts were then rinsed two times in PBS, prior to the staining of migrated cells on the underside of the chamber inserts using 100 μg/ml propidium iodide. The inserts containing fixed and stained cells were imaged on a Leica AOBS TCS SP2 confocal microscope using a ×10 air objective with an NA of 0.3 as described in [34]. Five representative regions of interest were imaged per insert. Fluorescently stained migrated cells were counted using CellProfiler software cell counting algorithm.

Super-resolution imaging
For super-resolution imaging, PANC-1 cells were seeded into Lab-Tek chambered coverglass eight-well #1.0 with low thickness variation (Thermo Scientific). To visualize IP 3 R1, PANC-1 cells were fixed with 4 % PFA for 10-15 min at room temperature and subsequently immunostained with anti-IP 3 R1 antibody, followed by the use of an appropriate species-specific Alexa Fluor 647conjugated secondary antibody. To visualize ER-PM junctions at super-resolution level, PANC-1 cells co-transfected with PM-targeted LL-FKBP-mRFP and ER-targeted CFP-FRB-LL constructs for 24 h were fixed using 4 % PFA for 10-15 min at room temperature after treatment with 100 nM rapamycin for 4-5 min at 37 • C/5 % CO 2 to highlight the pre-existing ER-PM junctions without ER Ca 2 + store depletion. Briefly, the heterodimerization of both ER-and PM-targeted constructs revealed the ER-PM junctions as punctate structures in both CFP and RFP fluorescence channels. To highlight the ER-targeted FRB-LL-CFP counterpart of the ER-PM junctions' puncta, cells were immunostained using anti-GFP antibody (which also recognizes CFP), and followed by appropriate species-specific Alexa Fluor 647-conjugated secondary antibody.
After immunostaining of IP 3 R1 or ER-PM junctions, samples were immersed and imaged in a dSTORM buffer containing 100 units/ml glucose oxidase, 2000 units/ml catalase, 50 mM mercaptoethylamine/HCl and 50 mg/ml glucose in PBS [35]. Wells were filled and sealed with a coverslip to exclude oxygen. Super-resolution microscopy was performed using a custom-built instrument, as described previously [35,36]. An Olympus IX71 microscope formed the basis of an inverted objective total internal reflection fluorescence (TIRF) instrument with a UAPON ×100 TIRF, NA − 1.49 objective. Laser illumination was provided by a 640 nm 150 mW diode laser (Toptica Photonic AG) and a 561 nm 200 mW optically pumped semiconductor laser (Coherent Europe). Additionally, a 405 nm laser diode (Mitsubishi Electronics) was available for re-activation of fluorophores if required. Laser power on the diode lasers was controlled directly, whereas a rotating quarter-wave plate was used to alter the power of the 561 nm laser. Low powers (1-10 %) were used for field of view selection, context and conventional fluorescence images. Higher powers (50-100 %) were used for dSTORM imaging. Fluorescence and excitation light were spectrally separated by the dichroic mirror and emission filter from a multi-edge filter set (LF405/488/561/635-A-000, Semrock), and an additional bandpass filter was used to remove cross-talk in each channel (FF01-676/37 with 640 nm excitation and FF01-600/37 with 561 nm excitation, both from Semrock).
Images were acquired using an EMCCD camera (Andor iXon 897) using software written in LabVIEW (National Instruments). Typically dSTORM image 'stacks' were composed of 10 000 frames, with 10 ms exposure time per frame. Super-resolution images were reconstructed from these image stacks using the open-source rainSTORM package [35,37,38].

IP 3 Rs knockdown and cell migration
Migration of PANC-1 cells after 72 h of siRNA knockdown of IP 3 R1 or IP 3 R2 or IP 3 R3 isoforms was measured using a Boyden chamber assay. siRNA sequences targeting each of the IP 3 R isoforms are: IP 3 R1 silencer select siRNA sense 5 -GCACGACAGUGAAAACGCAtt-3 , antisense 5 -UGCGUUUUCACUGUCGUGCct-3 ; IP 3 R2 silencer select siRNA sense 5 -GGUGUCUAAUCAAGACGUAtt-3 , antisense 5 -UACGUCUUGAUUAGACACCag-3 ; and IP 3 R3 silencer select siRNA sense 5 -GCAUGGAGCAGAUCGUGUUtt-3 , antisense 5 -AACACGAUCUGCUCCAUGCtg-3 . Following 72 h of treatment with the specified siRNA constructs, PANC-1 cells were allowed to migrate for 6 h in a humidified environment at 37 • C/5 % CO 2 . Boyden chamber inserts were then fixed using 100 % methanol. Non-migrated cells were removed from the topside of the chamber inserts using cotton buds and rinsed two times in PBS, prior to the staining and counting of migrated cells on the underside of the chamber inserts (see the Migration Assay section above for the description of staining and counting procedures).

Immunoblotting
Cells were treated with trypsin, removed from flasks and then collected by centrifugation. Cells were then lysed using RIPA lysis extraction buffer supplemented with Halt protease inhibitor cocktail and EDTA (Pierce-Thermo Scientific). Lysed samples were separated on a 4-12 % NuPAGE Bis-Tris gradient gels and protein transferred on to nitrocellulose membranes by transverse electrophoresis. Nitrocellulose membranes were blocked in 3 % (w/v) non-fat dried skimmed milk powder dissolved in PBS for 1 h at room temperature, and probed with primary antibodies including anti-IP 3 R1 (1:500 dilution), anti-IP 3 R2 (1:1000 dilution), anti-IP 3 R3 (1:500 dilution) and anti-β-actin (1:1000 dilution) at 4 • C overnight. After overnight incubation, nitrocellulose membranes were incubated with appropriate species-specific HRP-conjugated secondary antibodies (1:400 dilution) for 1 h at room temperature. Bands were visualized using enhanced chemiluminescence (ECL) Western blotting substrate and a Bio-Rad Quantity One imaging system. Band intensities were quantified and analysed using ImageJ software (NIH). Blotting for β-actin was used as a loading control for siRNA knockdown experiments.

Image, data and statistical analyses
Image acquisition and preliminary analysis of confocal images was performed using either Leica LAS or Zeiss LSM 510 or Zeiss Zen software. Further analysis was performed using ImageJ software. Linear adjustments of contrast and brightness were applied if necessary using ImageJ. The 'mask' images used for illustrating the co-localization of the rapamycin-inducible linker components (images labelled ER-PM linkers) were created using the Co-localize RGB ImageJ plugin as described in [33].
In data presentation (for all components of the study) the error margins represent the S.E.M. The results were analysed using a Student's t test; P < 0.05 was considered statistically significant and is indicated by the symbol * in the Figures.

IP 3 Rs translocate from cell-cell contacts in cellular clusters to the leading edge of individual migrating cancer cells
In monolayers of PANC-1 cells, IP 3 R1 was observed primarily in the areas of cell-cell contacts (see Figure 1A and Supplementary Figure S1). Similar distribution was observed in smaller cellular clusters ( Figure 1B). Fluorescence profiles measured along the lines selected to cross junctional regions demonstrate approximately 4-5-fold increased density of IP 3 R1 in these regions in comparison with the neighbouring regions of the cytoplasm ( Figures 1A and 1B, and Supplementary Figures S1A and S1C, upper panel). In cell-cell contact regions IP 3 R1 was closely co-positioned with occludin, which we used as a marker of tight junctions ( Figure 1C). The white colour in the 'Merge' panel of this figure indicates that in a part of this region the IP 3 R1 and occludin are in such close proximity that the distance between these proteins is below the resolution of a confocal microscope (i.e. less than 300 nm). Notably, in the monolayers of PANC-1 cells, E-cadherin (a component of adherence junctions and an important marker of the epithelial phenotype) was also preferentially found in the cell-cell contact regions (Supplementary Figure S2), confirming cellular connectivity.
The observed proximity of IP 3 Rs to tight junctions was described before in non-transformed Madin-Darby canine kidney (MDCK) cells, where it was associated with the developed epithelial phenotype [30,39]. Interestingly, in our experiments this distribution was observed in the monolayers of cancer cells (PANC-1 cells) and even in relatively small clusters of these cells. Importantly, cells disconnected from the clusters or monolayers also clearly display polarized distribution of IP 3 R1, as these cells preferentially positioned their IP 3 R1s at the leading edge ( Figure 1D, and Supplementary Figures S1A, right panels, and S1C, central panel). Notably, similar increased density of IP 3 R1 at the leading edge was observed in migrating PANC-1 cells after treatment with transforming growth factor (TGF)-β1, an agonist known to induce a migratory mesenchymal phenotype in this cell type [40,41] (see Supplementary Figure S3). Preferential positioning at the leading edge was also observed for IP 3 R2s (Supplementary Figure  S4). Pan-IP 3 R antibodies, which were raised against a conserved C-terminal region common to all IP 3 R subtypes, also revealed a similarly polarized distribution of IP 3 Rs with preferential localization at the leading edge of individual migrating cells ( Figure 1E and Supplementary Figure S1C, lower panel). The leading edge of migrating cells is characterized by the region of polymerized actin (reviewed in [4]). We therefore examined the relative positioning of the F-actin-enriched region and IP 3 R1; Figure 1(F) illustrates close apposition of the two proteins at the leading edge of isolated migrating PANC-1 cells. In addition to actin polymerization, cellular migration requires the formation of focal adhesions at the front of migrating cells; we therefore extended our analyses to examine the relative positioning of IP 3 Rs and focal adhesions.

Focal adhesions are closely surrounded and regulated by IP 3 Rs at the leading edge of migrating cancer cells
Dual immunostaining of focal adhesions (using antibodies against vinculin) and IP 3 R1s revealed remarkable relative localization of the adhesions and the receptors. As expected, focal adhesions were preferentially localized close to the leading edge of the migrating cells; the leading edge was also enriched with IP 3 R1s (Figure 2A). Importantly, focal adhesions were not co-localized but were instead closely surrounded by the receptors forming 'potholes' on the background of IP 3 R1 immunostaining (see the fragment shown in the bottom panels in Figure 2(A) and the associated fluorescence profile). The preferential positioning of focal adhesions and IP 3 Rs as well as 'potholes' were observed in the smooth-shaped (lamellipodia-like) regions of the leading edge ( Figure 2A and the central row of images in Figure 2B with the associated fluorescence profile) as well as in spiky, filopodialike protrusions at the leading edge ( Figure 2B and specifically the lower row of panels and the associated fluorescence profile). The intimate spatial relationship between focal adhesions and IP 3 Rs concentrated in this region should make these structures particularly sensitive to Ca 2 + signals generated by IP 3 Rs at the leading edge of migrating cells and could explain the prominent effect of the inhibition of IP 3 Rs on cell migration (see the last subsection of the Results section). Using intracellular photorelease of IP 3 from its caged precursor (uncaging) we directly probed the influence of this second messenger on the dynamics of focal adhesions. In these experiments we used cells expressing simultaneously Pax-mCh and Pax-GCaMP5. Paxillin was used in these experiments because it is an important regulatory component of focal adhesions [42].
Uncaging of IP 3 -induced rapid accumulation of paxillin in focal adhesions and the loss of paxillin from the cytosol (Figures 3A  and 3B). These findings highlight the importance of IP 3 Rs for focal adhesion remodelling -the process immediately related to migration ( [43], reviewed in [42]). Uncaging of IP 3 also induced Ca 2 + rises resolvable both in the cytosol and localized at focal adhesions ( Figure 3A). Interestingly, inhibition of IP 3 Rs with xestospongin-B reduced the paxillin content of focal adhesions in unstimulated PANC-1 cells (Figure 3C), suggesting that IP 3 Rs are involved in the regulation of focal adhesion even when cells are not stimulated with IP 3 -producing agonists. These results suggest that focal adhesions are modulated by low 'background' activity of IP 3 Rs.
Non-excitable cells rely on SOCE as the main [Ca 2 + ] ERreloading mechanism for efficient calcium signalling mediated by IP 3 Rs. To activate SOCE, STIM1 has to translocate to the ER-PM junctions [44]. Indeed the presence of STIM1competent ER-PM junctions in the proximity of the leading edge has recently been reported [7,33]. We therefore next investigated the relative positioning of IP 3 R1, STIM1 and ER-PM junctions.

Repositioning of IP 3 Rs from cell-cell contacts to the leading edge of migrating cells is accompanied by the accumulation of ER-PM junctions and STIM1 puncta in the adjacent cytoplasmic region
In migrating PANC-1 cells, the region with the increased density of IP 3 R1s at the leading edge was closely followed by the region with a high density of STIM1 puncta ( Figure 4A). Similar relative positioning was seen for IP 3 R1s and co-localized ER and PM linkers (white dots highlighting the ER-PM junctions in Figure 4B), suggesting that the ER-PM junctions migrating just behind the IP 3 Rs are STIM/SOCE-competent. The similarity of distributions of STIM1 puncta and ER-PM junctions (identified by the linkers) is consistent with other studies that observed the co-localization of STIM1 with ER-PM junctions [21,33]. The preferential localization of IP 3 R1s and ER-PM junctions at the front of migrating cells was also confirmed using super-resolution microscopy. In these experiments, employing the dSTORM technique, we observed that the leading edge of migrating cells indeed had increased density of IP 3 R1s ( Figure 4C) and higher concentration of ER-PM junctions ( Figure 4D). Note the increased resolution of dSTORM images (in the x-y plane) in comparison with diffraction-limited (in the x-y plane) TIRF images taken from the same cellular regions (insets in Figures 4C  and 4D). The actual size of both the ER-PM junctions and clusters of IP 3 Rs is significantly smaller than the limit of resolution of diffraction-limited microscopy but the preferential localization at the leading edge was observed using all types of microscopy. dSTORM imaging, which has considerably improved axial and lateral resolution in comparison with conventional microscopy, confirmed that both IP 3 R1s and ER-PM junctions can be observed close to the leading edge and in the immediate proximity to the ventral membrane of the migrating cells (i.e. portion of the membrane that is involved in forming contacts with the substratum and that is sliding along the substratum). A number of recent studies reported the importance of Ca 2 + signalling for cell migration and invasion [5][6][7][45][46][47]. The Ca 2 + responses have been shown to both potentiate [7,46] and suppress [7] migration, depending on cell type and extracellular environment. Considering the observed prominent stratified localization of IP 3 Rs and STIM1/ER-PM junctions near the leading edge of migrating PANC-1 cells and the proximity of these structures to the components of migratory apparatus (e.g. focal adhesions and actin fibres) we next decided to test the importance of IP 3 Rs and SOCE for the migration of this cell type.

Inhibition of IP 3 Rs and STIM-Orai channels suppresses migration of PANC-1 cells
The selective inhibitor of IP 3 Rs -xestospongin-B [48]effectively suppressed cytosolic Ca 2 + responses induced by IP 3 uncaging in PANC-1 cells ( Figure 5A). SOCE in this cell type was significantly (by 61+ − 1 %, n = 151) inhibited by 30 μM GSK-7975A ( Figure 5B), a selective inhibitor of SOCE mediated by STIM-Orai interaction [49]. Note that 10 μM GSK-7975A produced only a slightly weaker inhibition than 30 μM (inhibited by 53 + − 1 %, n = 162; results not shown) and 100 μM was not more effective than 30 μM (n = 145; results not shown). Migration in our experiments was tested using Boyden chambers. In the absence of FBS, PANC-1 cells migrate very inefficiently (leftmost bars in Figures 5C and 5D). We therefore investigated the effect of the inhibitors on migration of these cells in the presence of FBS using symmetrical FBS distribution (1 % FBS in both the upper and lower chambers, Figure 5C) and asymmetrical FBS distribution (0 % FBS in the upper chamber and 5 % FBS in the lower chamber; this configuration can be considered as a model of chemotactic migration, see Figure 5D). Xestospongin-B significantly inhibited migration of PANC-1 cells in conditions of symmetrical FBS ( Figure 5C). The effects of this IP 3 R inhibitor on migration were even stronger for cells migrating along the gradient of FBS ( Figure 5D); in this condition xestospongin-B inhibited migration by 74 + − 9 %. These findings are consistent with the results of IP 3 R-knockdown experiments that suggested the involvement of IP 3 Rs (particularly IP 3 R1 and possibly IP 3 R2) in migration ( Figure 6). GSK-7975A supressed migration of PANC-1 cells (Figures 5C and 5D) and, as for xestospongin-B, the effect was particularly prominent in the experiments with asymmetrical FBS ( Figure 5D, in these experiments GSK-7975A inhibited migration by 84 + − 3 %). Both xestospongin-B and GSK-7975A also inhibited cell migration as measured by wound-healing assay (Supplementary Figure S7). Neither xestospongin-B nor GSK-7975A induced substantial cellular toxicity (Supplementary Figure S8). Strong inhibition of migration by xestospongin-B and GSK-7975A suggest that the striking accumulation of IP 3 Rs and STIM/SOCE-competent ER-PM junctions in the leading edge of PANC-1 cells has a clear function, which is to provide signals important for migration of this type of cancer cells. There was a difference between the effect of xestospongin-B and GSK-7975A on the paxillin content of focal adhesions. Incubation for 1 h with xestospongin-B produced a statistically significant decrease in paxillin content in the focal adhesions of unstimulated PANC-1 cells (see Figure 3C); we did not, however, observe changes in paxillin content in focal adhesions following 1 h of application of 30 μM GSK-7975A (results not shown; n = 44 for GSK-7975A treated and n = 39 for the control group). It is therefore conceivable that the two inhibitors utilize different mechanisms for supressing the cell migration and that Ca 2 + release and influx regulate different processes contributing to the cell migration.
It is interesting to note that in primary normal pancreatic acinar cells IP 3 Rs are preferentially positioned in close proximity to the tight junctions in the cell-cell contact regions near the apical part of the cells [14,[50][51][52][53]. Indeed, we verified and confirmed this previously reported distribution using the same antibodies as employed for immunostaining of IP 3 Rs in PANC-1 cells (see Supplementary Figure S9). PDAC cells probably originate from pancreatic acinar cells (evidence for this has recently been reviewed in [54]). Therefore the changes in the distribution of IP 3 Rs from cell-cell contacts to the leading edge, that were characterized using a cellular model in our study, are likely to reflect the pathophysiological process associated with EMT in vivo. In other words the present study suggests a novel switch of polarity and function of Ca 2 + signalling complexes during EMT associated with cancerogenesis. Ca 2 + signalling complexes move from the intercellular contact regions (proximal to the apical part of the cell) to the leading edge and change function from regulating vital physiological processes (exocytosis and fluid secretion) to regulating migration of cancer cells. It is important to note that the leading edge attracts not only IP 3 Rs but also ER-PM junctions and STIM1 and that both Ca 2 + release and Ca 2 + influx are important for migration.
The present study describes prominent changes in the distribution of Ca 2 + signalling complexes that develop when cancer cells disconnect from their neighbours and form a migratory 'mesenchymal' phenotype. The observed preferential distribution of IP 3 Rs and STIM1/ER-PM junctions signifies the formation of novel structural and functional (signalling) polarity in migrating PDAC cells. Disrupting this polarized distribution could present a mechanism for inhibiting migration and invasion of PDAC cells and ultimately suppressing the formation of metastasis of this type of cancer.