Cph1-PM-mediated transcellular adhesion alters collective cell migration
To explore the contribution of signals that originate at the plasma membrane in collective cell migration, we engineered artificial photoswitchable cell-cell adhesion molecules, which don’t link to the actin cytoskeleton. Specifically, we used as the extracellular domain of the Cph1 protein from Synechocystis sp., which forms homodimers under red light (660 nm) and reversibly dissociates into monomers under far-red light (720 nm). To anchor Cph1 in the plasma membrane, without connecting to cytoskeletal components, Cph1 was solely fused to the transmembrane domain of the platelet-derived growth factor receptor (Fig. 1a). We named the membrane-anchored artificial and photoswitchable adhesion molecule ‘Cph1-PM’. In general, when Cph1-PM is expressed on the surfaces of cells, the adhesions between them will form under red light and reverse under far-red light as previously characterized 25.
As a starting point, we employed MDA-MB-231 (MDA) cells, which lack type-1 cadherins, display weak cell-cell adhesions and migrate as single cells 27. To explore if these artificial cell-cell adhesions can induce collective cell migration, we transfected MDA cells with Cph1-PM and generated monoclonal stable cell lines with Cph1-PM expression from them (Cph1-PM-MDA). In addition, we used MCF-7 cells as a positive control for a cell type that displays high expression levels of type-1 cadherins, strong cell-cell adhesions, and collective cell migration 28. To probe for collective and single-cell migration, we used wound healing assays and set a baseline for the migration behavior using wild-type MDA and MCF-7 cells. As expected, we find significantly faster the wound closure rates in MCF-7 cells compared to MDA cells (Fig. 1b). At the same time, MCF-7 cells also display a higher correlation length than MDA cells, a measure of persistence distance and indicative of a higher degree of coordinated cell movement (Fig. 1c). We further observed individual MDA cells at the wound edge depart from the confluent cell sheet, indicative of the weak intercellular adhesions between them, while the leading edge of the MCF-7 wound remained intact (Supplementary Fig. 1a and Supplementary Movie 1).
Next, we probed the effect of increased intercellular adhesions on MDA cell migration. For this, we compared the cell migration behavior of Cph1-PM-MDA cells under red light, where the artificial cell-cell adhesions are active, to cells under far-red light, where the artificial cell-cell adhesions are inactive. Exposure to red light caused Cph1-PM- MDA cells to migrate collectively and with a unified front (Fig. 1d and Supplementary Movie 2). Directional rate vectors for the migration showed that under red light the majority of motion vectors were directed toward the wound space. On the other hand, the orientation of the rate vectors of Cph1-PM-MDA cells under far-red light and wild-type MDA cells were random (Supplementary Fig. 1a). Moreover, we found a significantly higher correlation length for Cph1-PM-MDA cells under red light than under far-red light, which had a comparable correlation length to the parent MDA cells (Fig. 1c). Actually, the correlation length pf Cph1-PM-MDA cells under red light and MCF-7 cells were comparable.
Counterintuitively, the wound closed faster for Cph1-PM-MDA cells under red light than either those under far-red light or MDA cells lacking the construct (Fig. 1b), despite increased friction between the cells and the cells being more restricted in their freedom to move independently 26. These differences in wound closure rate were consistent over the entire 12-hour duration of the experiment (Supplementary Fig. 1b). To further analyze the origin of the faster wound closure rate under red light, we performed single-cell motion analysis (Fig. 1e and Supplementary Movie 3). The traces of individual Cph1-PM-MDA cells under red light exhibited a directional migration towards the wound during the analyzed 16-hour time window, whereas Cph1-PM-MDA cells under far-red light showed frequent random trajectories. For analysis, cells were grouped based on localization into “front cells”, (i.e. 3 cell diameters from the wound edge) and “rear cells” (Supplementary Fig. 2a). Under red light, Cph1-PM-MDA cells at the “front” moved more directly towards the wound compared cells under far-red light (Figs. 1f, g, Supplementary Figs. 2b, c). Moreover, also at the single cell level Cph1-PM-MDA cells at the front migrated faster under red light than under far-red light (Fig. 1h).
Cph1-PM mediated transcellular adhesion augments plasma membrane tension
Increasing the friction between individual cells was previously reported to slow down the median speed while increasing the collectivity of individual agents 26. In stark contrast, we found that increased intercellular adhesion in Cph1-PM-MDA cells augments cell speed and directionality at the same time (Fig. 1b, h). These findings suggest that Cph1-PM activation does not merely increase the adhesion between adjacent cells, but that it also elicits downstream cell signaling. Since Cph1-PM lacks a cytosolic tail that could directly interact with intracellular proteins, we hypothesized that this signaling must originate from within the plasma membrane. To test this assumption, we probed for changes in membrane properties.
First, we measured whether Cph1-PM activation influenced lipid ordering. For that, we took advantage of the hydrophobic fluorescence membrane dye Pro12A, which exhibits a blue-shifted emission with increased lipid ordering. In this ratiometric imaging, a rise in generalized polarization (GP) is associated with stiffer gel-like lipid conditions and a decrease is associated with increased fluidity 29. Focusing on a single cell-cell interface, we observed an increase in GP when activating the Cph1-PM with red light for 15 min (Fig. 2a). Opposingly, when the cells were exposed first to red light for 15 min and then for 15 min to far-red light illumination, to inactivate the Cph1-PM based adhesions, the GP value decreased (Fig. 2b). The results indicated that the direct physical cell-cell membrane adhesions locally alter membrane structure and make the interphase membranes stiffer at the cell-cell contact sites.
To further analyze the alternations in membrane properties on induction of the artificial cell-cell adhesions, we performed static tether pulling using atomic force microscopy (AFM) to examine changes in apparent plasma membrane tension at the cellular scale. Specifically, a membrane tether on cells positioned close to the wound edge was maintained at a constant distance until it ruptures, and the corresponding force difference was quantified (Fig. 2c). We found that Cph1-PM-MDA cells had a significantly higher static tether force under red light than these cells under far-red light (Fig. 2d). Moreover, arguing against profound tension gradients between different cells, we did not observe significant differences in membrane tension in the front vs. rear of the wound for either condition (Supplementary Figs. 2a, b).
Cph1-PM mediated changes in membrane tension affect membrane-to-cortex attachment and formation of focal adhesions
The static tether force measured in the previous experiments is proportional to the apparent membrane tension, which integrates the in-plane tension of the lipid bilayer as well as the membrane-to-cortex attachment 19, 30, 31. To investigate the origin of the observed difference in apparent membrane tension, we next probed the membrane-cortex interface. Previous studies showed that phosphorylation of Ezrin/Radixin/Moesin proteins (pERM) increases the connection between F-actin and the plasma membrane, thereby increasing the apparent membrane tension (Fig. 2e) 32. To probe for changes in pERM phosphorylation upon Cph1-PM activation, we exposed Cph1-PM-MDA cells to either red or far-red light, fixed the cells under the respective illumination, and then stained for F-actin, pERM, and DAPI (Fig. 2f). Here, we find quantitative increase of the pERM signal in Cph1-PM-MDA cells under red compared to far-red light (Fig. 2g) as another indicator of increased membrane tension on formation of the artificial cell-cell adhesions.
Considering that membrane tension has been reported to impact focal adhesion in single cells 33, we investigated how cell membrane adhesion affected cell spreading and the number of focal adhesions 34, 35. We found an increased number of focal adhesions in Cph1-PM-MDA cells under red light compared to those under far-red light by staining the focal adhesion associated protein vinculin, which is only associated with focal adhesions in the Cph1-PM-MDA system (Figs. 2h, l). While the total area of the focal adhesions per cell increased, the cell spreading area and the area of individual focal adhesions between Cph1-PM-MDA cells under red and far-red light did not show significant differences (Figs. 2i, j, k). Previous reports showing that not the total number but the size of the focal adhesions are the determining factor of cell speed would argue that the increased number of cell adhesions under red light compared to far-red light would not slow down cell migration 36.
Cph1-PM mediated effects are independent of E-cadherin signaling
Having observed collective migration, a feature generally associated with epithelial cell types, in the Cph1-PM-MDA originating from a mesenchymal cell type (MDA), we investigated changes in the expression of epithelial and mesenchymal markers. First of all, we found no increase in the expression of the cell-cell adhesion molecule E-cadherin (marker of the epithelial phenotype) in Cph1-PM-MDA cells independent of light treatment, while it was clearly expressed in control MCF7 cells (Fig. 3a).
The observed collective migration in Cph1-PM-MDA cells under red light and the faster movement of individual cells raises the question of what signaling pathways are triggered downstream. A logical starting point was to investigate whether signaling pathways that are associated with cadherin signaling are indirectly activated due to these artificial cell-cell adhesions. First, to determine the role of actin polymerization in Cph1-PM dependent signal propagation, we added cytochalasin D (CytoD), an inhibitor of actin polymerization (Fig. 3b) and jasplakinolide (Jasp), an actin nucleation promoter (Fig. 3c) to the wound healing assays. Treatment with CytoD reduced the wound closure rate both in Cph1-PM-MDA cells under red and far-red light, only confirming the central role of actin in cell migration (Figs. 3b, d). Moreover, the addition of Jasp yielded no significant changes compared to the untreated cells (Figs. 3c, d), except for a slight increased correlation length for Cph1-PM-MDA cells both under red and far-red light when treated with Jasp (Fig. 3e).
Next, we evaluated Cph1-PM-dependent changes in the mechanosignaling associated with E-cadherin. E-Cadherin has been described to activate YAP-dependent gene expression with changes in mechanical strain 37. When evaluating the localization of YAP under different illumination, we found no difference in the cytoplasmic versus nuclear localization (Figs. 3f, g). Consistently, expressions of two YAP down-streaming markers, TEA domain transcription factor 1 (TEAD1) and connective tissue growth factor (CTGF), showed no significant differences between red and far-red (Fig. 3h).
Finally, we evaluated EMT markers that would be expected to be down regulated with increasing epithelial characteristic. Firstly, vimentin, a mesenchymal cell marker, showed no change in the Cph1-PM-MDA compared to the parent cell type in the western blot (Fig. 3a). Likewise, the gene expression levels of two representative mesenchymal markers, zinc finger E-box binding homeobox transcription factor 1 (ZEB1) and fibronectin 1 (FN1), were comparable with red and far-red light exposure as observed with RT-PCR (Fig. 3i). Collectively, we did not find any evidence that increased Cph1-PM based cell-cell adhesions with red light exposure alters the expression of genes that are associated with the transition from a mesenchymal to an epithelial cell type. Hence, these findings argue that Cph1-PM signaling and the observed collectivity in the wound healing did not rely on any cadherin-dependent circuits.
Cph1-PM mediated transcellular adhesion promotes PLD2 signaling
The findings to this point establish that activation of the artificial cell-cell adhesions results in substantial changes in membrane tension as well as in faster and collective cell migration. At the same time, we find no changes in signaling cascades originating from classical E-cadherin signaling. Unlike E-cadherin with truncated cytosolic tail, which cannot mediate cell-cell adhesions due to the missing interactions with the cytoskeleton, Cph1-PM is able to maintain stable cell-cell adhesions under red light without connecting to the cytoskeleton. Therefore, we reasoned that the origin of the signal resulting in the altered phenotype may reside in the plasma membrane. Indeed, membrane tension has been implicated promoting biochemical signals at the single-cell level 20. More precisely, increased membrane tension was reported to augment phospholipase D2 (PLD2) activity, which breaks down phosphatidylcholine (PC) into phosphatidic acid (PA) and choline 20. Downstream, PA modulates the properties of cells through activation of various pathways, chief among others, is the mammalian target of rapamycin (mTOR) pathway 38. Considering its mechanosensitive properties, we thus probed for changes in PLD2 activity, through the quantification of its product PA 21. Consistent with increased PLD activity, we observed significantly elevated PA levels in Cph1-PM-MDA cells under red light compared to cells under far-red light and the MDA cells lacking Cph1-PM (Fig. 4a). This increase in PA in Cph1-PM-MDA cells under red light was clearly due to the increased cell-cell adhesions and not the expression of Cph1-PM in MDA cells itself as no significant differences in PA content was observed between Cph1-PM-MDA cells under far-red light and the parent MDA cells.
To further consolidate these findings, we modulated the PLD2 pathway by adding either propranolol (Pro), which prevents the degradation of PA through phosphatidic acid phosphatase (PAP)39, or by adding VU-0285655-1 (VU), which inhibits PLD2 directly (Fig. 4b) 40. Following treatment with Pro, PA levels in Cph1-PM-MDA cells under red light remained as high as for untreated cells (Fig. 4a). In contrast, the addition of VU decreased PA levels in Cph1-PM-MDA cells exposed to red light to the level of cells under far-red light. Other than this, no significant changes in PA levels were observed when Cph1-PM-MDA cells were treated with either Pro or VU under far-red light or in parent MDA cells. These findings are relevant, as it argues that no additional PA is produced in Cph1-PM-MDA cells under red light illumination without the activation of PLD2 (Fig. 4a).
To connect the increases in PA with the increased migratory capacity as has been demonstrated for single cells, 20, 31, 41 we evaluated the effect of altered PLD2 activity on wound closure rates and coordination of the cellular movement (Figs. 4c, d). In the presence of Pro, Cph1-PM-MDA cells under red light showed similar wound closure rates as Cph1-PM-MDA cells under far-red light and untreated MDA control cells, respectively. In contrast, the treatment with VU resulted in a significant decrease in wound closure rate under red light, while it did not alter the wound closure rate under far-red light (Fig. 4e). This result directly indicates that PLD2 activity and PA levels contribute to the improved wound closure rate and faster movement of individual cells observed for Cph1-PM-MDA cells under red light. This relationship is further supported by the decreased correlation length with VU treatment under red, which comparable to the correlation length of Cph1-PM-MDA cells under far-red light (Fig. 4f). Collectively, these results suggest not only that the PLD2 activation modulates the migration potential of MDA cells, but also that Cph1-PM activates the PLD2 pathway.
Finally, we increased the membrane tension in cells through the addition of oleic acid (OA), which inserts itself into the plasma membrane resulting in a uniform increase in membrane tension (Fig. 4g). Exogenous increase of membrane tension with OA led to Cph1-PM-MDA cells under far-red light and parent MDA cells to migrate at the same pace as Cph1-PM-MDA cells under red light (Figs. 4h, i). However, it should be noted that under far-red light in the presence of OA, the Cph1-PM-MDA cell migration was not as coordinated as for the cells under red light due to the lack of cell-cell connections (Fig. 4j, Supplementary Video 4).
Cph1-PM mediated membrane signaling impacts the PI3K/mTOR pathway
The results to this point suggest that the direct connection between cells through Cph1-PM increased membrane tension, activating PLD2, which in turn increased PA levels. To understand how this may affect cell dynamics, we investigated the role of PA-associated pathways. In particular, we considered the PA-dependent PI3K/mTOR activity shown to impact cell migration 42. To this end, we perturbed the PI3K/mTOR pathway with wortmannin, an inhibitor of PI3K 43; SF1670, an inhibitor of PTEN 44; and rapamycin, an inhibitor of the mTOR complex 45. After wortmannin treatment, the wound closure rate of Cph1-PM-MDA cells under red light significantly decreased and was comparable to cells under far-red light (Fig. 5a). Similarly, the addition of rapamycin led to a significant decrease in the wound closure rate of Cph1-PM-MDA cells under red light, but had not affect under far-red light. In contrast, SF1670 treatment resulted in a significant increase in the wound healing rate even under far-red light, but did not alter Cph1-PM-MDA cells under red light (Fig. 5b). Complementarily, we evaluated single-cell tracks during the wound healing assay under red and far-red light in conjunction with SF1670 and rapamycin treatment (Fig. 5c, Supplementary Figs. 4a, b). In the absence of any inhibitor, as already observed, the cell trajectories were straighter and the individual cells moved faster under red light than under far-red light. Yet, the SF1670 treatment resulted in an increased coordination of the cells even under far-red light. These findings are consistent with previous studies that reported an increase in the wound healing rate, and a tendency for collective cell movement, upon inhibition of PTEN 46. Importantly, the wound closure rate decreased for Cph1-PM-MDA cells treated with rapamycin under red, but not under far-red light, which argues that inhibition of the mTOR pathway selectively suppressed the collective behavior of the Cph1-PM-MDA cells induced by the artificial cell-cell adhesions.