A polarized Ca2+, diacylglycerol and STIM1 signalling system regulates directed cell migration

Abstract

Ca(2+) signals control cell migration by regulating forward movement and cell adhesion. However, it is not well understood how Ca(2+)-regulatory proteins and second messengers are spatially organized in migrating cells. Here we show that receptor tyrosine kinase and phospholipase C signalling are restricted to the front of migrating endothelial leader cells, triggering local Ca(2+) pulses, local depletion of Ca(2+) in the endoplasmic reticulum and local activation of STIM1, supporting pulsatile front retraction and adhesion. At the same time, the mediator of store-operated Ca(2+) influx, STIM1, is transported by microtubule plus ends to the front. Furthermore, higher Ca(2+) pump rates in the front relative to the back of the plasma membrane enable effective local Ca(2+) signalling by locally decreasing basal Ca(2+). Finally, polarized phospholipase C signalling generates a diacylglycerol gradient towards the front that promotes persistent forward migration. Thus, cells employ an integrated Ca(2+) control system with polarized Ca(2+) signalling proteins and second messengers to synergistically promote directed cell migration.

Figure 1

Receptor tyrosine kinase (RTK) signaling is restricted to the front of migrating leader cells. (a,b) bFGF-induced tyrosine phosphorylation was higher in the front of migrating cells (white arrows). Addition of the pan-RTK inhibitor Ponatinib blocked tyrosine kinase signaling in the front, but not in the back of leader cells. Follower cells did not respond to bFGF. HUVECs were fixed and stained with pY20 anti-phospho-tyrosine antibody (n = 107, 105, 115, 110 and 107 cells for SFM, follower cells, and Ponatinib 0 nM, 25 nM and 100 nM, respectively). SFM: serum-free medium. (c,e) Fluorescence ratio images of leader cells co-expressing YFP-Akt-PH (PIP₃ sensor) or YFP-C1AC1A (DAG sensor) and a plasma membrane marker (CFP-mCD4). PIP₃ (c) and DAG (e) were enriched in the front of migrating cells. (d,f) Front-to-back gradients of PIP₃ and DAG were present in leader, but not follower cells (24 leader and 42 follower cells in (d), 28 and 62 cells in (f)). (g) Ca²⁺ pulses in migrating HUVECs were measured as relative increases in local PM targeted GCaMP6s fluorescence intensity. Higher activities were observed in the front (#1) compared to the middle (#2) or back (#3) of migrating cells. (h) Relative mean amplitudes of local Ca²⁺ fluctuations measured over 3 minutes in the front of migrating cells in response to serum or serum plus Ponatinib (see also Supplementary Fig. 1e,f). Amplitudes of Ca²⁺ fluctuations were normalized to basal cytosolic levels (0.3 R.U. means the fluctuation is 30% of the average cytosolic [Ca²⁺] level; n = 24 cells). (i,j) Migrating HUVECs expressing GCaMP6s-CAAX and the reference membrane marker mCD4 were used to measure Ca²⁺ gradients in leader and follower cells (n = 83 leader and n = 86 follower cells). Bars denote mean ± SEM in Fig. 1b,h. Student t test was used for Fig. 1b,d,f,h,j. In Fig. 1d,f,j, p values were calculated by comparing the ratio of the sensor / PM intensity ratios in the front and back (both regions were 10% of cell length).

Figure 2

Store-operated Ca²⁺ (SOC) influx controls cell migration by regulating cell-matrix adhesion in the front of migrating cells. (a) HUVEC migration into open space was monitored by staining cells with CellMask (see Methods). Accelerated sheet migration was observed in STIM1-depleted compared to control cells. (b) Comparing changes in the rate of sheet migration and cytosolic Ca²⁺ levels in HUVECs treated with siRNAs targeting different Ca²⁺ signaling regulators. Average cytosolic [Ca²⁺] was normalized to the level of cells treated with siCntrl (n = 4 experiments for each siRNA). (c) Reduced single cell migration speed in cells over-expressing YFP-STIM1. Cells expressing YFP-ER were used as control (n ~ 10,000 cells per condition). (d,e) Effects of the ER Ca²⁺ pump blocker thapsigargin (d) and the SOC inhibitor BTP2 (e) on cytosolic Ca²⁺ levels and on sheet migration speed. Notice that increasing cytosolic Ca²⁺ levels by thapsigargin decreased migration speed, and lowering Ca²⁺ levels by BTP2 increased migration speed (n = 4 experiments per condition). (f,g) Migrating HUVECs were treated with different concentrations of BTP2 or thapsigargin to reduce or elevate cytosolic Ca²⁺ levels. Cells were then fixed and stained with anti-phospho-myosin light chain (pMLC) antibody. (f) pMLC signals were lower when SOC was blocked by BTP2 but higher when ER Ca²⁺ pumps were blocked by thapsigargin. CAAX: plasma membrane marker. (g) pMLC levels increased with increasing cytosolic [Ca²⁺] (n = 123, 134, 127, 126, 123, 117 and 142 cells per condition from left to right). (h,i) Effect of BTP2 treatment on cell-matrix adhesion. Focal adhesion formation was monitored by expressing GFP-paxillin. BTP2 treatment rapidly decreased the intensities of GFP-paxillin puncta in the front of migrating cells, consistent with SOC influx promoting cell-matrix adhesion. Bars are mean ± SEM in Fig. 2b,d,e,g.

Figure 3

SOC increases migration speed when cell-matrix adhesion is weak, but slows down migration when adhesion is strong. (a) Knocking down STIM isoforms increased sheet migration speed in HUVEC (left) but caused a small and significant reduction in speed in H1299 cells (right) (n = 4 experiments per condition). (b) H1299 cells migrated faster on high fibronectin. BTP2 treatment decreased migration speed of H1299 cells on low, but increased migration speed on high fibronectin. Each box shows the median (horizontal line) and the 25th and 75th percentiles. Lower and upper whiskers are minima and maxima, respectively (n = 3 experiments per condition). (c–f) Fibronectin and SOC jointly regulate focal adhesions. H1299 cells were plated on low or high fibronectin and treated with BTP2 or DMSO before fixation and staining with anti-paxillin antibody to label endogenous focal adhesions. (c) Paxillin puncta (white arrows) were prominent in cells on high fibronectin but were decreased in low fibronectin or when treated with BTP2. (d) Integrated punctate paxillin signals for different treatments in (c) (n = 115, 121 and 104 cells from left to right per condition). (e) Overexpression of YFP-STIM1 in H1299 cells (#1, 2) on low fibronectin increased cell substrate adhesion (paxillin staining) compared to control cells (#3). (f) Average paxillin puncta intensity for control cells and for cells expressing high levels of YFP-STIM1 (n = 42 and 23 cells from left to right per condition). (g) Single cell speed in sheet migration as a function of YFP-STIM1, mCitrine-paxillin or YFP-ER overexpression in H1299 cells. Overexpression of YFP-STIM1 and mCitrine-paxillin accelerated sheet migration on low fibronectin but decreased migration speed on high fibronectin. Solid lines and dashed lines are mean ± SEM; n = 4714 (red) & 4538 (blue) cells with STIM1, 3873 (red) & 4327 (blue) cells with paxillin, and 5698 (red) & 5368 (blue) with ER marker. p value compares the blue and red group for cells with log_eYFP > 6.5. Student’s t test was used in Fig. 3a,b,d,f,g. Bars are mean ± SEM in Fig. 3a,d,f.

Figure 4

STIM1 is enriched in the front of migrating cells. (a,b) Migrating HUVEC expressed YFP-STIM1 (STIM1) and a CFP-tagged ER marker (ER). Merged and ratio images are shown here and in Supplementary Fig. 4a to show the relative enrichment of STIM1 compared to an ER marker towards the front. White arrow indicates the direction of cell migration. (b) Quantification of the ratio of YFP-STIM1 / CFP-ER marker from front to back (see Methods). YFP-STIM1 was enriched in the front, whereas (c) the control YFP-ER was not (n = 36 cells per condition). (d–f) Enrichment of STIM1 in the front of migrating cells is mediated by binding to the microtubule plus-end binding protein EB1. (d) Domain structure of STIM1 and mutations preventing binding to EB1. (e) Unlike wild-type STIM1 (S1wt) protein (Fig. 4a,b and Supplementary Fig. 4a), the S1NN mutant was not enriched in the front of migrating cells. (n = 27 cells) (f) Over-expression of the S1NN mutant suppressed cell migration less than overexpression of S1wt. Bars are mean ± SEM (n ~ 10,000 cells per condition).

Figure 5

STIM1 is locally activated in the front of migrating cells. (a) HUVEC cells were co-transfected with YFP-STIM1 and the ER-PM junction marker CFP-ER-PM. Confocal images show focal planes at the bottom of the cell. The white arrow marks the direction of migration.YFP-STIM1 was enriched at front ER-PM junctions in migrating cells. (b) Quantification of the ratio of YFP-STIM1 / CFP-ER-PM from front to back in migrating cells. (n =14 cells) (c,d) Similar analysis as in (b) but for cells coexpressing YFP-S1NN and CFP-ER-PM (c) or coexpressing YFP-ER-PM and CFP-ER-PM. A smaller increase in relative S1NN activity was observed in the front (n = 10 cells for S1NN and n = 12 cells for the control group). (e,f) Decreasing luminal ER Ca²⁺ levels towards the front of migrating leader cells. (e) Ratio-imaging of a modified luminal ER Ca²⁺ FRET probe T1ER (see Methods) in migrating HUVECs. Adding the SERCA inhibitor thapsigargin (2 μM) and EGTA (3 mM) decreased ER Ca²⁺ levels (lower panel). (f) Gradient in luminal ER Ca²⁺ measured using the T1ER probe. Note that the lower Ca²⁺ levels in the front are still sensitive to EGTA+thapsigargin treatment (n = 79 cells for the control group; n = 49 cells for the thapsigargin + EGTA group).

Figure 6

Polarized plasma membrane Ca²⁺ pump activity keeps cytosolic Ca²⁺ low in the front. (a) Basal cytosolic Ca²⁺ levels (measured using Fura-2) in the front were about 50% of the levels in the back of migrating cells (n = 14 cells). Local [Ca²⁺] was normalized to the average cytosolic level. Cytosolic measurements using GCaMP6s-CAAX are shown in Fig. 1 i,j. (b) High Fura-2 increased the diffusion speed of Ca²⁺, so as to decrease the Ca²⁺ gradient between the front and back of migrating cells. Bars are mean ± SEM (n = 160, 160, 80, 160 and 80 cells from left to right for each group). (c) Addition of the PMCA inhibitors Caloxin (200 μM) and La³⁺ (200 mM) significantly decreased cytosolic Ca²⁺ gradients in migrating cells. Bars are mean ± SEM (n = 47 cells per condition). (d–g) Cytosolic Ca²⁺ is transported out of the cell faster in the front than in the back. (d) Addition of thapsigargin, SOC inhibitor BTP2, and Ca²⁺ chelator EGTA to migrating HUVECs caused a transient increase in cytosolic Ca²⁺. Local extrusion pump rates were measured as a function of the Ca²⁺ level in the front or back over time. (e) Graph showing the relative pump rates (derivative of Ca²⁺ change) in the front and the back as a function of local [Ca²⁺]. The slopes reflect relative pump activity differences in the front and the back for each cell. (f) Statistical analysis of the relative Ca²⁺ pump activities in the front versus the back of migrating cells. Bars are mean ± SEM (n = 25 cells). (g) Cells pretreated with inhibitors of PMCA lost their differential Ca²⁺ pump activities. Bars are mean ± SEM (n = 47 cells for each group). Student t test was used for Fig. 6a,c,f,g. One-way ANOVA was used for Fig. 6b.

Figure 7

Higher Ca²⁺ pump activity in the front compared to the back generates a gradient of basal [Ca²⁺] in migrating cells. (a–d) UV flash photolysis experiments confirmed differential Ca²⁺ pump activities in migrating HUVECs as shown in Fig. 6c–e. (a) Migrating HUVECs were pre-loaded with Fluo-3/AM and NP-EGTA. Thapsigargin and EGTA were added 10 minutes before imaging to block the activity of SERCA and influx Ca²⁺ channels. A UV pulse was used to induce a Ca²⁺ spike 1 minute after recording began. (b,c) Ca²⁺ pump activities (k) in the front and in the back were calculated based on Fluo-3 measurements following UV photolysis, similar to Fig. 6d,e. (d) Quantification of relative Ca²⁺ pump activities in the front and in the back of migrating cells. (n = 41 cells.) (e) Inhibitors of Na⁺-Ca²⁺ exchangers 2,4-DCB or 3,4-DCB both caused a small increase cytosolic Ca²⁺ levels as measured by Fura-2. EGTA, Thapsigargin (Th) and BTP2 were used as positive controls. [Ca²⁺] was normalized using average cytosolic levels in the DMSO group (n = 4 wells for each group). (f) Addition of 10 μM 2,4-DCB or 3,4-DCB did not affect the differential Ca²⁺ pump activities. LaCl₃ was used as a positive control (n = 41, 40, 35 & 42 cells in DMSO, 2,4-DCB, 3,4-DCB and LaCl₃). (g) PMCA4 is enriched in the front of migrating cells. HUVEC co-expressing GFP-PMCA4 and the plasma membrane marker tdimer2-lyn were imaged by confocal microscopy. The merged (left) and the ratio-image (right) indicated enrichment of GFP-PMCA4 in the front (see also Supplementary Fig. 6d). The white arrow depicts the direction of cell migration. (h) Statistical analysis of the relative spatial distribution of GFP-PMCA4/tdimer2-lyn from the front to the back (n = 13 cells). A Student t test was used for Fig. 7d,f,h. In Fig. 7h, p values were calculated based on the ratio of the sensor / PM ratio in the front 10% region to that in the back 10% region of migrating cells. In Fig. 7d,e,f, Bars are mean ± SEM.

Figure 8

A phospholipase C (PLC)-induced gradient of diacylgycerol (DAG) controls cell motility and directionality. (a) Addition of the PLC inhibitor U73122 (1μM) suppressed the DAG sensor accumulation observed in the front of control cells (n = 64 cells for control; n = 22 cells for U73122). (b) The PLC inhibitor U73122 slowed down single cell speed of HUVEC as well as sheet migration speed in a dose-dependent manner (Supplementary Fig. 7a). U73122 was added to the cell sheets prior to time-lapse imaging (n = 2 experiments for each group). (c–e) Overexpression of YFP-C1AC1A reduced the measured migration parameters directionality, single cell speed and directional persistence. (c) Cell migration traces of 50 randomly chosen cells not expressing C1AC1A (left) and 50 cells overexpressing C1AC1A (right) are shown. The traces were aligned to start at the origin (0 μm, 0 μm) with the direction of the wound at the top. Cells expressing C1AC1A had relatively shorter traces and partially lost their orientation towards the open space. (d) Schematic of how the migration parameters speed, persistence and directionality shown in (e) were determined. (e) Statistical analysis of the change in directionality, speed, and persistence in response to YFP-C1AC1A overexpression were calculated for each migrating cell and correlated with binned levels of YFP-C1AC1A expression. Increasing YFP-C1AC1A expression resulted in decreasing directionality, speed, and persistence. Bars are mean ± SEM. (n = 1200 cells) (f) Partial inhibition of PKC by Ruboxistaurin decreased the rate of sheet migration. The reduction could be rescued by inhibiting DAG kinase using two types of inhibitors. Bars are mean ± SEM (n = 3 experiments per condition). (g) Schematic representation of the identified gradients in the Ca²⁺ and diacylglycerol signaling system. In Fig. 8a, p values were calculated by Student t test based on the ratio of the sensor / PM ratio in the front 10% region to that in the back 10% region of migrating cells. One-way ANOVA was used for Fig. 8b,e,f to determine the significance of difference between multiple groups.