Cross-talk between Rac and Rap GTPases in migrating cells

Abstract

To enable effective cell migration, local cell protrusion has to be coordinated with local cell attachment. Here, we investigate spatiotemporal activity patterns of key regulators of cell protrusion and adhesion, the small GTPases Rac and Rap, in migrating cells. These analyses show that Rac activity correlates very tightly with instantaneous cell protrusion events, while the Rap activity stays elevated for prolonged time periods after protrusion and is also detectable before cell protrusion. Direct analysis of activity cross-talk in living cells via light-based perturbation methods revealed that Rap can efficiently activate Rac; however, reciprocal cross-talk from Rac to Rap was not detectable. These findings suggest that Rap plays an instructive role in the generation of cell protrusions by its ability to activate Rac. Furthermore, prolonged Rap activity suggests that this molecule also plays a role in maintenance or stabilization of cell protrusions. Indeed, analysis of Rap1-depleted A431 cells revealed a significant reduction of cell attachment, suggesting that Rap-stimulated cell adhesion can stabilize newly formed protrusions. Taken together, our study suggests a mechanism, by which cell protrusion is coupled to cell adhesion via unidirectional cross-talk that connects the activity of the small GTPases Rap and Rac.

• During cell migration, protrusion and adhesion have to be coordinated in space and time. How the regulators of these processes are connected is not known.

• To directly investigate these connections, we combined activity measurements of the small GTPases of the Ras/Rap and Rho families, with light-based perturbations in living cells. We observed partially overlapping activity dynamics and unidirectional cross-talk from Rap to Rac.

• Our results offer a deeper understanding of cell migration, by suggesting that Rap has an instructive role in stimulating both initial cell protrusion and subsequent adhesion, and that activity cross-talk coordinates these processes in space and time.

INTRODUCTION

Directed cell migration plays a fundamental role in many biological processes, including embryonic development, the immune response and regeneration of injured tissues (Trepat et al., 2012). In addition to these physiological roles, cell migration can also drive the progression of diseases, such as the aberrant migratory behavior of malignant cancer cells that drives metastasis and cancer invasion (Trepat et al., 2012; Scarpa and Mayor, 2016).

The process of cell migration involves four steps. 1) First, cell protrusions are formed at the front of migrating cells. 2) Subsequently, new adhesions are generated at the leading edge of the cell. 3) Contractile structures, called stress fibers are formed that are linked to these adhesions, and that generate forces between the front and back of the cell. 4) Finally, the trailing end of the cell detaches and retracts after adhesions at the back of the cell are released. This last step generates excess dorsal surface to enable newly generated protrusions at the cell front (Ridley, 2001).

To facilitate migration, the processes of cell protrusion, adhesion, and retraction must be precisely coordinated in space and time. Consequently, the signaling network components that regulate these processes have to be closely interconnected. Interestingly, recent studies have shown that the network components that control various aspects of cell migration are controlled by positive and negative feedback loops that generate excitable system dynamics (Bement et al., 2015; Yang et al., 2016; Devreotes et al., 2017; Graessl et al., 2017; Miao et al., 2017). The causal links that connect protrusion and retraction signals can either enable highly dynamic cell shape changes, resulting in a more random, exploratory migration mode with frequent changes in direction (Arrieumerlou and Meyer, 2005; Nanda et al., 2023), or result in a more persistent directional migration mode via a more stable spatial segregation of protrusion and retraction (Svitkina et al., 1997; Patwardhan et al., 2024). For the spatiotemporal coordination of the many processes that occur during cell migration, regulators of the Rho and Ras families were shown to play important roles (Devreotes et al., 2017).

Rac1, a member of the Rho family of GTPases, is a master regulator for the formation of flat cell protrusions, which are called lamellipodia (Ridley et al., 1992). Within these structures, Rac1 activates the WAVE regulatory complex, a molecule that subsequently stimulates the actin nucleator Arp2/3. The newly formed actin filaments generate a polymerization-driven force that pushes the cell edge forward (Ridley, 2001; Steffen et al., 2013). After new adhesions are formed, another Rho family GTPase called RhoA stimulates contractile forces, which are generated by the activity of the molecular motor Myosin-II on actin filaments (Ridley, 2001).

The Ras/Rap family of small GTPases is also thought to play important roles in cell migration (Wittchen et al., 2005; Devreotes et al., 2017). Similar to other small GTPases, Ras/Rap proteins are activated by particular guanine nucleotide exchange factors (GEF), inhibited by specific GTPase-activating proteins (GAP) and they relay their activity state to cellular processes via specific effectors. Interestingly, previous studies suggested that the activity of Ras/Rap proteins might be linked to other regulators of cell migration, in particular, to the cell protrusion master regulator Rac1. Based on previous studies, Ras/Rap proteins can act upstream of Rac1, for example via the GEF Tiam1 (Lambert et al., 2002) and also downstream of Rac1 via ROS (Diekmann et al., 1994; Ferro et al., 2014) or growth factor signaling (Joshi et al., 2023). Ras GTPases are well-studied oncogenes that mediate growth factor signaling (Hobbs et al., 2016). Rap GTPases are key regulators of integrin-mediated cell matrix adhesion and cadherin-mediated cell–cell junction formation. In particular, the Rap1 isoforms are known to activate integrins via the Rap1-GTP–interacting adapter molecule (RIAM) (Lafuente et al., 2004).

Here, we developed and applied new, improved live cell activity sensors to investigate the function of Ras/Rap proteins in the keratinocyte-derived cancer cell line A431. We particularly focused on activity patterns of Ras/Rap and the related Rho family GTPase Rac1 during cell migration, and how these activity patterns are associated with dynamic cell shape changes. To investigate potential causal links between these key regulators, we combined direct, light-controlled activity perturbations with simultaneous monitoring of the signal network response. Our findings reveal a clear hierarchy between Rap and Rac activities and support a model, in which Rap signals have an instructive role to stimulate the Rac-dependent initiation of cell protrusion events and the subsequent processes that consolidate these newly formed cell protrusions.

RESULTS

Ras/Rap activity in migrating cells

In previous studies of cell migration in the keratinocyte-derived A431 cell line, we found that highly dynamic protrusion–retraction cycles are stimulated by an unexpected cross-talk that activates the cell contraction regulator Rho downstream of the cell protrusion regulator Rac1 (Nanda et al., 2023). These studies raised the question, which signals act upstream of Rac to initiate the dynamic protrusion–retraction cycle. Previous studies, in particular in Dictyostelium, suggested that members of the Ras/Rap family of small GTPases act upstream of Rac1; however, little information was available on the spatiotemporal activity state of these molecules in mammalian cells. To fill this gap in our knowledge, we used a similar strategy as in our previous study (Nanda et al., 2023) to generate sensitive activity sensors for Ras and Rap-family GTPases (Figure 1A). Briefly, we fused two tandem copies of effector domains that either preferentially bind the active form of Rap-subfamily (RalGDS-GBD) or Ras subfamily (Raf-GBD) GTPases to a fluorescent protein, expressed the sensor proteins at very low levels using the delCMV promotor (Watanabe and Mitchison, 2002) to avoid competition with effector proteins, and monitored the plasma membrane translocation of these sensors via total internal reflection fluorescence microscopy (TIRF-M).

FIGURE 1:

Spatiotemporal Ras and Rap activity dynamics. (A) Schematic for Ras and Rap activity sensors used in this study. Similar to other GTPases, inactive GDP-bound Ras/Rap is largely cytosolic, where it binds to a solubilization factor. GEFs catalyze the transition into the active, GTP-bound Ras/Rap state, which is preferentially localized at the plasma membrane, where it interacts with effectors via GTPase-binding domains (GBD). The sensors used in this study contain tandem repeats of GBDs that are fused to a fluorescent protein, and their translocation from the cytosol to the active GTPase at the plasma membrane is monitored via TIRF-M. (B) Representative TIRF image of an A431 cell coexpressing the Rap (mCitrine-2xRalGDS-GBD) and Ras (2xRaf-GBD-mCherry) activity sensors (see also Supplemental Movie S1). Cyan arrows point at cell protrusions and yellow arrows point at Rap activity pulses in central cell attachment areas. (C) Kymograph corresponding to white arrow in B. (D) Quantification of signal intensity changes in the black square region in B. (E) Sequential TIRF images of a single migrating A431 cell that expresses the Rap activity sensor and a volume marker (see also Supplemental Movie S2). Cyan arrows point to newly formed cell protrusions.

Although translocation sensors based on Raf-RBD are frequently used to investigate Ras, RalGDS-based sensors are less commonly used. We therefore performed a more detailed characterization of our 2xRalGDS-based sensors by combining sensor readouts with local perturbations of Rap activity at the plasma membrane via photochemically-induced dimerization (Supplemental Figure S1) (Chen et al., 2017). Of the five known human Rap isoforms, we concentrated our investigations on the well-studied Rap1 subfamily and specifically on Rap1a, as this isoform is expressed at a higher level in A431 cells compared with Rap1b (Klijn et al., 2015). Briefly, we fused Rap1a variants that lack their plasma membrane targeting CAAX-Box to a first heterodimerization domain: E. coli dihydrofolate reductase (eDHFR). In addition, we expressed a second heterodimerization domain, Halo-Tag, fused to the CAAX-Box plasma membrane targeting sequence. We treated cells with the photocaged dimerizer NvocTMP-Cl and induced plasma membrane targeting of Rap1a variants by local photouncaging of the dimerizer within a diffraction limited spot. As expected, we found that plasma membrane targeting of constitutively active Rap1a G12V lead to efficient cotargeting of the sensor (Supplemental Figure S1B). On average, a ≈20% change in the Rap1a G12V perturbation signal lead to a ≈10% change in the sensor signal. In contrast, a corresponding control sensor that lacks the RalGDS domain (Supplemental Figure S1A), or a perturbation based on the dominant negative Rap1a mutant S17N (Supplemental Figure S1C) did not result in plasma membrane cotargeting. A perturbation based on Rap1a wt resulted in efficient plasma membrane targeting of the sensor, which was, however, slightly slower compared with perturbations via Rap1a G12V. This suggests that plasma membrane-bound Rap-GEFs might be able to activate Rap1a wt after its plasma membrane targeting.

Using this sensor, we detected highly dynamic patterns of Rap activity in A431 cells, including transient pulses in central cell attachment areas that were robustly observed in nonmigratory cells >40 h after plating (Figure 1, B–D; Supplemental Movie S1). In migratory cells shortly after replating on fresh fibronectin (<6 h), Rap activity was strongly enriched at the cell edge, where it closely correlated with active cell protrusion (Figure 1E; Supplemental Movie S2). Active Ras on the other hand was more homogenously distributed and did not show strong local enrichment (Figure 1, B–D; Supplemental Movie S1), which is in agreement with its major role in controlling cell proliferation and cell growth.

To quantify Rap and Ras activity dynamics, we analyzed how their signals change with protrusion and retraction of the cell edge using a modification of the ADAPT ImageJ plugin (Figure 2) (Barry et al., 2015; Nanda et al., 2023). These analyses show that Rap activity signals strongly and positively correlate with cell edge velocity (Figure 2, D, E, and M; Supplemental Movie S3). On average, Rap activity is slightly delayed with a maximal correlation at ≈1min (65s) after cell protrusion.

FIGURE 2:

Ras/Rap activity dynamics in migrating cells. (A–L) Analysis of control (A–C), Rap (D–F), Ras (G–I), and Rac (J–L) activity dynamics at the cell periphery during protrusion–retraction cycles (see also Supplemental Movie S3). (A, D, G, and J) Representative TIRF images of the activity sensor and a volume marker (left), and kymographs (right) corresponding to the white arrows in left panels. (B, E, H, and K) Cross-correlation functions for sensor signals and cell edge velocity plotted against the time shift between these measurements. Black arrow heads point to time shift with maximal signal after cell protrusion. (C, F, I, and L) Enrichment of sensor signals in protrusions (>0.5 µm/min) and retractions (≤0.5 µm/min). Values are normalized to mean control sensor measurements. (M) Measurements of the signal-cell edge velocity correlation coefficient at a time shift of 0 min of the correlation functions shown in B, E, H, and K. (N) Measurements of the signal enrichment at a time shift of 0 min of the enrichment plots shown in C, F, I, and L. (O) Difference between the signal enrichment before and after protrusion (defined as time shift of 0 min), extracted from data shown in C, F, I, and L. Data corresponding to Rac activity shown in J–L and in the last column of M–O were reproduced from previously acquired data (Nanda et al., 2023) (data are protected by CC BY 4.0 license). (*P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001; ns, not significant; One-sample, two-sided t-test; n > 18 cells from at least three independent experiments, exact numbers of cells are indicated in panels C, F, I, and L. Error bars represent the SEM.

The cell edge velocity correlation cannot distinguish between activity signal increase with positive velocity during cell protrusion or signal decrease with negative velocity during cell retraction. Furthermore, the cell edge velocity correlation only quantifies how similar signal changes are compared with cell edge velocity changes and does not indicate how strongly the sensor signals are locally enriched within cells. We therefore also performed a protrusion–retraction enrichment analysis that we recently developed (Nanda et al., 2023), to quantify how much the sensor signals are increased at the cell edge compared with the entire cell attachment area during cell protrusion and cell retraction phases. These analyses showed that Rap activity is strongly enriched by over 80% with a maximum at ≈2min (112s) after cell protrusion (Figure 2, F and N). Conversely, Rap activity is depleted by ∼15% at ≈2min (112s) after cell retraction. Correlation and enrichment measurements of signals of the Ras activity sensor were much weaker compared with the Rap activity sensor (Figure 2, G–I, M, and N; Supplemental Movie S3), but were nevertheless clearly above levels obtained using a cell volume marker that served as control (Figure 2, A–C, M, and N; Supplemental Movie S3).

Interestingly, the Rap correlation and enrichment measurements showed a strong asymmetry for time periods before and after cell protrusion. This was particularly clear for the signal enrichment during protrusion. We did not observe such an asymmetry using an identical assay for Rac activity in our previous studies (Nanda et al., 2023). For comparison, the measurements that were obtained in this previous study are reproduced in Figure 2, J–L (see also Supplemental Movie S3). To quantify this asymmetry, we subtracted the mean enrichment in the time period before protrusion from the mean enrichment after protrusion (Figure 2O). This quantification clearly shows that Rap activity is more strongly enriched at the edge of the cell after protrusion compared with before protrusion. No asymmetry was measurable for Rac activity, and Ras activity showed a weak, intermediate trend (Figure 2O).

CROSS-TALK BETWEEN RAP AND RAC GTPases

The asymmetric enrichment of Rap activity relative to cell protrusion events suggests a delayed activation of this regulator. In contrast, the more symmetric enrichment of Rac activity suggests a tight spatiotemporal link to cell protrusion. Indeed, the maximal enrichment of Rap activity is substantially shifted after cell protrusion (τ = 112s), while the maximal enrichment of Rac is not shifted at all (τ = 0s). Intuitively, this delay could indicate that Rap acts downstream of Rac. Simultaneous imaging of both Rac and Rap activity supports this idea: although both Rap and Rac are highly active during cell protrusion, Rac activity quickly diminishes after protrusion, while Rap activity often remains high within regions that corresponded to the leading edge shortly before (Figure 3A; Supplemental Movie S4). However, these experiments also revealed a more complex pattern. In addition, Rap activity was often observed before cell protrusion in the absence of active Rac (Figure 3A; Supplemental Movie S4). Thus, we both observed events, in which Rap is active before Rac, and events in which Rac is active before Rap. As previous studies could explain both a causal link from Rac to Rap (Lambert et al., 2002), as well as from Rap to Rac (Diekmann et al., 1994; Ferro et al., 2014; Joshi et al., 2023), we directly investigated potentially bidirectional cross-talk between these molecules by combining rapid activity perturbations with readouts of the signal network response.

FIGURE 3:

Investigation of GTPase cross-talk revealed unidirectional activation of Rac downstream of Rap. (A) Representative TIRF images (left) and corresponding kymographs (right) of A431 cells coexpressing Rap (mCitrine-2xRalGDS-GBD) and Rac (mCherry-3xp67phox-GBD) activity sensors (see also Supplemental Movie S4). Yellow arrows point to weak transient Rap signals that can be detected before Rac activation. Blue arrows point to prolonged Rap activity after initial protrusion. (B) Schematic for the photochemical dimerizer-based perturbation method (top) and for PA-Rac1–mediated perturbations (bottom), and their combination with activity measurements. (C) Representative TIRF images of an A431 cell that coexpresses the dimerization fusion proteins mTurquoise2-NES-eDHFR-Rap1a G12V (“Rap1a perturbation”), mTagBFP-Halotag-CAAX (not shown) as well as the Rac activity sensor mCherry-3xp67^phoxGBD (“Rac sensor”) and the volume marker delCMV-mCitrine (not shown), immediately before (0 s) and after (10, 120, and 360 s) photouncaging at the entire cell attachment area via a single TIRF illumination pulse at 405 nm. (D) Kymographs that correspond to the white arrows in left panels (see also Supplemental Movie S5). Magenta arrows represent the timepoint of perturbation. Blue arrows point to transient activity of Rac1. (E) Quantification of Rap1A perturbation and parallel measurement of Rac activity sensor recruitment dynamics and volume marker intensity in the entire cell attachment area or at the cell periphery. (F) Quantification of the early Rac activity response at the cell periphery. (G) Quantification of changes in the cell attachment area after photouncaging (n= 16 for control and n= 19 cells for the experimental condition, from 3 independent experiments). (H) Left, Representative TIRF images of an A431 cell that coexpresses the dimerization fusion proteins mTurquoise2-NES-eDHFR-Rap1a G12V (“Rap1a perturbation”), mTagBFP-Halotag-CAAX (not shown) as well as the Rac activity sensor mCherry-3xp67^phoxGBD (“Rac sensor”). Right, Kymographs that correspond to the white arrows in left panels (see also Supplemental Movie S6). (I) Quantification of local Rap1A perturbation and parallel measurement of Rac activity sensor recruitment dynamics. (J) Quantification of the early Rac activity response (n= 29 for control and n= 37 cells for the experimental condition, from three independent experiments). (K) Left, Representative TIRF image of an A431 cell that coexpresses mCerulean-PA-Rac1 and the mCitrine-2xRalGDS-GBD Rap activity sensor. The small inset shows a TIRF image of the mCerulean-PA-Rac1 distribution that was acquired at the end of the imaging sequence. Right, Kymograph that corresponds to the white arrow in the left panel (see also Supplemental Movie S7). Photoactivation was performed in the whole-cell attachment area via TIRF illumination at 445 nm. (L) Quantification of Rap1A recruitment dynamics to the entire cell attachment area. (M) Quantification of the early Rap activity response (n= 38 for control and n= 42 cells for the experimental condition, from three independent experiments). (N) Proposed signal network that links Rap and Rac activity in cells. (**P<0.01; ns, not significant; two-sided Student's ttest). Error bars represent the SEM. Quantifications in F, J, and M considered average signals in the time ranges before (−50 to 0 s) and after (0 to 50 s) illumination.

To test, whether Rap can activate Rac, we again used our photochemically induced dimerization method to introduce rapid perturbations of constitutively active Rap1a G12V at the plasma membrane (Chen et al., 2017), (Figure 3B). We combined these perturbations with the Rac activity sensor that we also used in Figure 2, J–L. We first introduced perturbations at the entire cell-attachment area via a single TIRF illumination pulse at 405 nm. Using this method, we measured a very rapid and transient Rac activity response (Figure 3, C–E; Supplemental Movie S5). Interestingly, although the levels of active Rap1a remained elevated for more than 5 min, the activity of Rac decreased much faster and reached baseline levels already after 1.5 min. This suggests that the Rac activity is effectively and rapidly inhibited after the initial activation, either via a constant inhibition or via a negative feedback. Importantly, Rac activity was clearly detectable in peripheral cell regions, where this molecule is thought to play important roles in stimulating cell protrusions (Figure 3, E and F). Indeed, we were able to detect an increase in the cell attachment area that was associated with Rap1a targeting (Figure 3G) or the Rac1 activity response (Figure 3, C and D), however, as expected for a more complex morphological phenotype, this process was significantly slower (Figure 3, E and F). Interestingly, we were also able to measure cross-talk in central cell attachment areas that did not generate cell protrusions, suggesting that the Rap/Rac activity cross-talk is independent of cell shape changes and therefore not an indirect result of a morphological phenotype. We confirmed this cross-talk in central cell attachment areas by introducing highly local perturbations within diffraction limited spots in central cell areas (Figure 3, H–J; Supplemental Movie S6). These observations clearly show that Rap1a acts upstream of Rac in the A431 cell line.

To investigate the reciprocal causality, if Rac can activate Rap, we used the previously established, PA-Rac1 construct (Wu et al., 2009), which enables light-controlled activation of Rac1 in living cells (Figure 3B). PA-Rac1 was previously shown to be enriched in the plasma membrane (Wu et al., 2009), and we activated Rac1 activity in the entire cell attachment area by continuous TIRF illumination at 445 nm and combined this rapid perturbation with Rap activity measurements via the activity sensor used in Figure 2, D–F. As expected, rapid activation of Rac1 with light lead to the generation of new lamellipodial cell protrusions. However, in contrast with lamellipodia that spontaneously form in A431 cells, which always enriched high levels of active Rap (Figures 1E and 2D), the PA-Rac1–triggered lamellipodia did not display an enrichment in Rap activity (Figure 3K; Supplemental Movie S7). Furthermore, quantification of the Rap sensor signal in the entire cell attachment area did not show any Rap activation by Rac, and instead even suggested a small reduction of its activity (Figure 3, L–M).

Taken together, these two observations suggest a unidirectional causality between Rap and Rac activities (Figure 3N): Rap can very quickly and efficiently activate Rac within seconds. Subsequently, Rac activity is inhibited, either by constant inhibition or via negative feedback regulation, which acts within 1 to 2 min (Figure 3, E and I). Conversely, Rac does not strongly influence Rap activity. The weak reduction of Rap activity downstream of Rac over the time course of ∼5 min is too slow to explain the adaptive downregulation of Rac1 by itself, but it might play a role in a slower process that might act in parallel.

Efficient cell attachment is dependent on the Rap1a isoform

Parallel imaging of Rac and Rap activity showed that these regulators are both activated very strongly during active protrusion (Figure 3A). However, as shown in Figures 3A and 2, D–F and O, Rap activity remained elevated much longer, even after the instantaneous protrusion events. As new cell attachment sites have to be established during and shortly after cell protrusion, this observation fits to the well-known role of Rap1 in stimulating new cell adhesions (Boettner and Van Aelst, 2009). To investigate, whether these processes are indeed controlled by Rap1 GTPases in A431 cells, we combined knockdown experiments with phenotypic readouts. We first identified two siRNAs, which were highly effective and reduced the levels of Rap1a mRNA by more than 90% within 48 h (Figure 4A). Using these tools, we found that the knockdown of Rap1a leads to a significant reduction of the cell adhesion area, suggesting that this molecule is indeed required for normal spreading of A431 cells on extracellular substrates (Figure 4, B and C).

FIGURE 4:

Efficient cell attachment is dependent on the Rap1a isoform. (A) Quantification of Rap1a mRNA levels via qPCR, 48 h after siRNA-mediated knockdown. Knockdown efficiency was highly efficient (91 ± 3% for siRNA#7; n = 3 and 93.4 ± 1% for siRNA#5; n = 4; mean±SEM). (B and C) Effect of Rap1a knockdown on the cell attachment area 8 h after seeding. Cells were stained for DNA and filamentous actin; area quantification was performed using Cell Profiler (****P < 0.0001; two-sided Student's ttest; left: n = 10,949 for control and n = 6712 cells for siRNA #7, from three independent experiments; right: n = 17,746 for control and n = 10,006 cells for siRNA #5, from four independent experiments). (D and E) Effect of Rap1a knockdown on FA formation. (D) Cells were fixed 3 h after replating on fresh fibronectin, and stained for the FA marker paxillin, as well as for DNA and filamentous actin. FAs were identified and analyzed via Cell Profiler. Top, Representative wide-field fluorescence images. Bottom, Objects identified via Cell profiler (cells, nuclei, FAs), are shown as colored outlines. (E) Quantification of cell adhesion area (****P < 0.0001; ***P < 0.001; ns, P > 0.05; Ordinary one-way ANOVA with Dunnet's posttest), paxillin maximal signal per cell and the fraction of the cell adhesion area that is covered by FAs (****P < 0.0001; ***P < 0.001; Kruskal–Wallis test with Dunnet's posttest; n > 377 cells from three independent experiments). (F) Schematic for the activity state of Rap and Rac in relation to cell protrusion. (G) Schematic for the proposed role of Rap in cell protrusion. An upstream signal, for example from growth factor receptor activity, stimulates Rap, which in turn activates Rac, leading to cell protrusion. Rap activity can persist over longer periods of time to stabilize the newly formed protrusion by the formation of new cell adhesions. Error bars represent the SEM.

In the context of cell adhesion, Rap1a was previously linked to integrin inside-out signaling (Shimonaka et al., 2003; Lafuente et al., 2004). As focal adhesions (FAs) are critically dependent on normal integrin function, we further investigated the impact of Rap1a knockdown on the formation of these critical adhesion structures. We fixed cells either 3 or 8 h after replating on fresh fibronectin and used the FA marker paxillin to quantify the formation of FAs (Figure 4E). These studies first confirmed our previous observation on cell spreading, in particular at the earlier 3 h timepoint. Furthermore, both the paxillin staining intensity and the coverage of the cell adhesion area with FAs were strongly reduced after Rap1a knockdown, clearly showing that this signal molecule is essential for the normal function of FAs during initial spreading of A431 cells on fibronectin.

Taken together, our studies show that Rap and Rac activities are coordinated in space and time during migration of A431 cells (Figure 4, F–G). We reveal a clear hierarchy between these signal molecules and find that the isoform Rap1a can efficiently activate Rac. We also find that Rac is downregulated in the continued presence of active Rap (Figure 3, E and I). This suggests strong constant inhibition or negative regulation of Rac.

These direct investigations of activity cross-talk can fully explain the observed coordination between Rap and Rac activities during cell protrusion–retraction cycles (Figure 4F). Initial Rap activity is already observed before Rac activity, presumably due to upstream signals, for example, downstream of growth factor receptor signaling. If Rap is sufficiently active, it activates Rac, leading to efficient cell protrusion (Figure 4G). Rac activity is inhibited via constant inhibition or via negative feedback (Figure 3, E and I), for example, via activation of Rho (Nanda et al., 2023) and subsequent activation of Rac GAPs (Guilluy et al., 2011), while Rap activity is still high, supporting subsequent steps, including cell adhesion (Figure 4G). Thus, our investigations suggest a mechanism that ensures effective cell protrusion by coupling this process to increased cell adhesion in space and time via unidirectional cross-talk from Rap1 to Rac1.

MATERIALS AND METHODS

Request a protocol through Bio-protocol

Cell culture

A431 (CRL-1555, ATCC) cells were maintained using standard cell culture techniques at 37°C and 5% CO₂using DMEM medium (2 mM l-Glutamine, PAN Biotech and 10% FBS, PAN Biotech or Sigma-Aldrich Chemie GmbH/Merck). For imaging, cells were either freshly replated (<6 h) onto glass-bottom dishes (MatTek) for conditions requiring highly migratory cells with dynamic cell edge movements (Figures 1E, 2, and 3A); or allowed to fully attach for >40 h on LabTek glass surface slides (Thermo Fischer Scientific), which lead to a less dynamic phenotype characterized by reduced cell migration (Figures 1, B and C and 3C and D, H and K). For knockdown experiments, cells were freshly replated onto LabTek slides to stimulate cell spreading and fixed after specified timepoints of 3 or 8 h (Figure 4). All glass-bottom dish types were coated with 10 µg/ml fibronectin for 45 min at room temperature (RT). Transfection of plasmid DNA was performed using Lipofectamine 3000 (Invitrogen, Thermo Fisher Scientific).

siRNA-mediated knockdown

Knockdown experiments were performed using ON-TARGETplus siRNAs (Dharmacon, Horizon Discovery): (siControl: #2 5′-UGGUUUACAUGUUGUGUGA-3′, Rap 1A: #5 5′-GAACAGAUUUUACGGGUUA-3′, #7 5′-GCGAGUAGUUGGCAAAGAG-3′). For knockdown, A431 cells were transfected with 30 nM siRNA using Lipofectamine RNAiMAX (Invitrogen, Thermo Fisher Scientific) and incubated for 24 h, followed by cell passaging to remove excess siRNAs. Knockdown efficiency was assessed 48 h after siRNA treatment via qPCR. Briefly, RNA samples were prepared using the RNeasy Mini Kit (Qiagen) and QIAshredder (Qiagen) and processed by subsequent cDNA synthesis (GoScript Reverse Transcription System, Promega), followed by qPCR. 96 h after siRNA transfection, phenotypic analysis was performed on fixed cells.

Plasmid constructs

The Rac activity sensor construct, delCMV-mCherry-3x-p67phox-GBD, and the control constructs delCMV-mCitrine and delCMV-mCherry were described previously (Nanda et al., 2023). The PA-Rac construct used for optogenetic perturbations, mCerulean-PA-Rac1Q61L, was a kind gift from Klaus Hahn (University of North Carolina) (Wu et al., 2009). The plasma membrane associated HaloTag construct used for photochemically induced dimerization, mTagBFP-HaloTag-CAAX, and the photo-caged, small-molecule dimerizer, NvocTMP-Cl, were described previously (Chen et al., 2017). The Rap activity sensor construct delCMV-mCitrine-2xRalGDS-GBD was generated by restriction digestion and ligation of 3xGFP-RalGDSRBD (kind gift from Philippe Bastiaens, MPI Dortmund) and delCMV-mCitrine-Rhotekin-GBD (Kamps et al., 2020), using AccIII and MfeI, and subsequent doubling of the RalGDS-GBD insert by Gibson assembly using AccIII and the primers 5′-CATGGACGAGCTGTACAAGTCCGGAGGTTCCGGAAGTGGATCCGCGCTGCCGCTCTACAAC-3′ and 5′-GCGCAGCTCGAGATCTGAGTCCGGATCCACTTCCGGAAC CGGTCCGCTTCTTGAGGAC-3′. The Ras sensor delCMV-2xRaf-GBD-mCherry was cloned by restriction digestion and ligation of RafRBD-mCherry (kind gift from Philippe Bastiaens, MPI Dortmund) and delCMV-mCherry using NheI and AgeI, followed by doubling of the Raf-GBD via Gibson assembly using AgeI and the primers 5′-ACGGGTTCTGGAAGTGGATCGGTTCTCATGTCCCTGGTGGAGGC-3′ and 5′-GGCCTCCACCAGGGACATGGAATTCGATCCACTTCCAGAACCCGTCGCCTTGCCTAGGTAATC-3′. The exchange of the fluorophore in delCMV-2xRaf-GBD-mCherry to delCMV-2xRaf-GBD-mCitrine was performed by restriction digest and ligation using AgeI and BsrGI. The perturbation construct mTurquoise2-NES-eDHFR-Rap1a G12V and related control constructs mTurquoise2-NES-eDHFR-Rap1a wt and mTurquoise2-NES-eDHFR-Rap1a S17N, which contain the eDHFR dimerization domain and a nuclear export sequence (NES) were generated as follows: First, delCMV-MCP-mCitrine-Rap1a(wt) was generated by restriction digestion and ligation of EYFP-C1 Rap1a(wt) (kind gift from Philippe Bastiaens, MPI Dortmund) and delCMV-MCP-mCitrine (based on MCP-YFP, from Addgene #101160 and delCMV-mCitrine; described in Master thesis Olga Just, TU Dortmund) using BsrGI and MfeI. The dominant positive mutation G12V was then introduced into this construct via site-directed mutagenesis using the primers 5′- GGTCCTTGGTTCAGTAG GCGTTGGGAAG-3′and 5′-CTTCCCAACGCCTACTG AACCAAGGACC-3′ to obtain delCMV-MCP-mCitrine-Rap1aG12V. Finally, mTurquoise2-NES-eDHFR-Rap1a G12V was generated by restriction-free Gibson assembly after PCR amplification of delCMV-MCP-mCitrine-Rap1aG12V using primers 5′-CTCTAGATCCATGCGTGAGT ACAAGCTAG-3′ and 5′-AATTGAGTTATTCCACTGGTGTTTTCCTATTTATC-3′, and of mTurquoise2-NES-eDHFR-RhoAQ63L (Kamps et al., 2020), using primers 5′-ACCAGTGGAATAACTCAATTGTTGTTGTTAAC-3′ and 5′-ACTCACGCATGGATCTAGAGGTGGATCC-3′). mTurquoise2-NES-eDHFR-Rap1a(wt) was generated from mTurquoise2-NES-eDHFR-Rap1aG12V via QuickChange, using the primers 5′-GGTTCAGGAGGCGTTGGGAAGTC-3′and 5′-CAACGCCTCCTGAACCAAGGACC AC-3′. Subsequently, mTurquoise2-NES-eDHFR-Rap1aS17N was generated from mTurquoise2-NES-eDHFR-mCitrine-Rap1a(wt) via QuickChange, using the primers 5′-GGCGTTGGGAAGAATGCTCTGACA-3′ and 5′-CTGAACTGTCAGAGCATTCT TCCCAACG-3′.

qPCR

Evaluation of siRNA-mediated mRNA knockdown was performed using qPCR. PPIB was used as a housekeeping gene to normalize measurements for individual mRNAs and scrambled siRNA was used as a control treatment to calculate knockdown efficiency. qPCR reactions were performed using the “GoTaq qPCR System” Kit (Promega), using 300 nM of each primer. Primer sequences were selected based on previously published studies: Rap1A (Lin et al., 2015): TGTCTCACTGCACCTTCAATGGCAT (fw), ACGCCTCCTGAACCAAGGACCA (rv); PPIB (Nazet et al., 2019): TTCCATCGTGTAATCAAGGACTTC (fw), GCTCACCGTAGATGCTCTTTC (rv).

Microscopy

TIRF microscopy was performed using an Olympus IX-81 microscope, equipped with a TIRF-MITICO motorized TIRF illumination combiner, an Apo TIRF 60x/1.45 NA oil immersion objective and a ZDC autofocus device. All TIRF measurements used a single dichroic mirror (ZT405-440/514/561) that was used in combination with a matched emission filter set (HC 435/40, HC 472/30, HC 542/27 and HC 629/53), a 514 nm OBIS diode laser (150 mW) (Coherent, Santa Clara), and the Cell R diode lasers (Olympus) with wavelength 405 nm (50 mW), 445 nm (50 mW), and 561 nm (100 mW). In some experiments, TIRF measurements were combined with wide-field illumination via the Spectra X light engine (Lumencor). For detection, an EMCCD camera (C9100-13; Hamamatsu, Herrsching am Ammersee, Germany) was used at medium gain without binning. In addition, the microscope was equipped with a temperature-controlled incubation chamber. Time-lapse live-cell microscopy experiments were carried out at 37°C in CO₂-independent HEPES-stabilized imaging medium (PAN Biotech or Sigma-Aldrich Chemie GmbH/Merck) supplemented with 10% FBS. Automated scanning for cell morphometric analysis was performed using the same microscope using wide-field illumination and an UPlanSApo 10x/0.4 NA air objective and software-based automated focusing. The only criterion for including cells in live cell measurements were that they expressed all constructs and that they were viable during the entire video acquisition time.

Analysis of cell morphodynamics and local fluorescence signals at the cell edge

Cell morphodynamics were measured and analyzed essentially as described previously (Nanda et al., 2023). Briefly, the local enrichment of the Ras and Rap GTPase activity was investigated in A431 cells by transfecting the corresponding activity sensors together with soluble fluorescent proteins that act as cell volume markers. Movements of the cell edge were analyzed automatically on the basis of the cell volume markers independent of varying GTPase activity signals. The subsequent quantitative analysis procedure was also described previously (Nanda et al., 2023). Briefly, a modified version of the ADAPT plugin (Barry et al., 2015) was combined with custom ImageJ analysis scripts to generate the signal/velocity cross-correlation functions and protrusion/retraction enrichment measurements. Cell edge movements >0.5 µm/min were considered to be protrusions, and movements ≤0.5 µm/min were considered to be retractions.

Analysis of Rac/Rap GTPase activity cross-talk via optogenetic perturbations

Cross-talk between Rac and Rap activity was performed essentially as described before (Nanda et al., 2023). Briefly, photoactivatable Rac1 (mCerulean-PA-Rac1) was cotransfected with the Rap activity sensor delCMV-mCitrine-2xRalGDS-GBD. Light-based activation of Rac1 was performed by TIRF illumination using the 445-nm Cell R diode laser. To prevent excess PA-Rac1 activation, a 10,000x neutral-density filter was added into the 445-nm TIRF illumination light path. The built-in neutral-density filter wheel of the microscopy setup was additionally set to 30%. Within photoactivation time intervals, 445 nm TIRF illumination was constantly on, except for the exposure times during image acquisition. Detection of mCerulean was always performed after the experiment. The subsequent analysis was also described previously (Nanda et al., 2023). Briefly, the activity sensor measurement A_Rap was calculated by first measuring the fluorescence intensity of the Rap activity sensor I_Rap via TIRF microscopy in the entire cell adhesion area. These raw intensity measurements were normalized by subtracting the background signal outside the cell area I_{Rap, BG}, and dividing by the initial, background-corrected intensity value I_{Rap, 0} − I_{Rap, 0, BG} before the perturbation:

Analysis of Rap/Rac GTPase activity cross-talk via photochemically induced dimerization

Cross-talk between Rap and Rac activity was performed using photochemically induced dimerization, essentially as described previously (Chen et al., 2017). Briefly, local or whole attachment area photouncaging of the NvocTMP-Cl dimerizer was performed by a single illumination pulse (200 ms, 405 nm, 180 nW) at a diffraction-limited spot via the FRAP mode of the TIR-MITICO motorized TIRF illumination combiner, or via TIRF illumination (200 ms, 405 nm, 540 nW). Laser power was measured at the aperture of the 60x objective. Stabilized and background-corrected time series of the perturbation and response signals were measured at the perturbation site. To exclude cells in subsequent analyses that did not generate a sufficiently strong perturbation, only cells with ≥5% of signal increase in the perturbation signal within 30 to 50 s after illumination were considered.

Analysis of cell morphology

96 h after siRNA transfection, 15,000 A431 cells per condition were reseeded onto LabTek wells that were previously coated with fibronectin (10 mg/ml). For quantification of the cell attachment area, the cells are washed 3x with 37°C PBS and fixed with prewarmed formaldehyde (3.7% in PBS) for 20 min at 37°C, 8 h after seeding. After 3x washing with PBS, the cells were permeabilized with Triton X-100 (0.1% in PBS) for 15 min at RT. After 3x washing with PBS, cells were stained for F-actin and nuclei using Rhodamin Phalloidin (1:1000 in PBS) and Hoechst 33342 (1:1000 in PBS) for 30 min at RT in the dark. Cells were washed 3x with PBS and stored at 4°C until imaging within 24 h of fixing. A total of 64 images were obtained for each condition by automated scanning of individual wells. Subsequently, images were analyzed using Cell Profiler (Carpenter et al., 2006) to measure the cell adhesion area. Briefly, nuclei were detected as primary objects using Hoechst signals and adaptive Otsu thresholding with two classes and a typical object diameter of 5 to 25 pixels. Then, cells were detected using phalloidin signals as secondary objects using nuclei as input objects, the propagation method and global three class Otsu thresholding. The cell adhesion area was then obtained as the area of the secondary cell objects.

For quantifying the formation of FAs, the cells are washed 3x with 37°C PBS and fixed with prewarmed formaldehyde (3.7% in PBS) for 20 min at 37°C, 3 and 8 h after seeding. After 3x washing with PBS, the cells were permeabilized with Triton X-100 (0.1% in PBS) for 15 min at RT. After 3x washing with PBS, cells were blocked with 2% BSA in PBS for 1 h at RT and subsequently incubated with the primary antibody (1:400 in 2% BSA/PBS, purified mouse anti-paxillin, BD Transduction Laboratories, catalogue no. 610051) for 1 h at RT. After 3x washing with PBS, cells were simultaneously stained for F-actin and nuclei using Rhodamin Phalloidin (1:1000) and Hoechst 33342 (1:1000) and incubated with the secondary antibody (1:1000 in 2% BSA/PBS, goat anti-mouse IgG, Alexa Fluor 514, Invitrogen, catalogue no. A-31555) for 1 h at RT in the dark. Cells were washed 3x with PBS and stored at 4°C until imaging within 24 h of fixing. Of note, >20 images that had 40 to 60% cell coverage were obtained for each condition. Subsequently, images were analyzed using Cell Profiler (Carpenter et al., 2006) to measure the cell adhesion area, maximal paxillin intensity, and the fraction of FA area with respect to the entire cell attachment area. Briefly, the contrast of nuclei images was increased by applying a median filter with a window size of three pixels, followed by a morphological tophat filter und a disk structuring element (size:25 pixels) and suppression of background with a feature size of 25 pixels. Nuclei were detected in contrast enhance images as primary objects using a manual threshold of 0.005, and a typical object diameter of 10 to 500 pixels. Then, cells were detected using widefield paxillin images, which were dominated by cytosolic signals, as secondary objects, using nuclei as input objects, and the propagation method with Adaptive Sauvola thresholding. The cell adhesion area was again obtained as the area of the secondary cell objects. FA structures of typical size were first enhanced in TIRF paxillin images using the speckle method and a feature size of six pixels, and subsequently identified as another set of primary objects, using a typical diameter of 5 to 100, and a manual threshold of 0.05. FA primary objects were then defined as child objects of the corresponding, overlapping secondary cell objects. Signal intensity of FA objects was measured in the original TIRF paxillin images and FA size was determined based on the FA primary objects.

Subsequent verification of the analyses showed that in up to 5% of the automatically acquired images, the automated thresholding procedure resulted in inaccurate cell segmentation, and these images were therefore excluded from the analyses.

Software for image and video analysis

Image and video analysis was performed using FIJI (https://imagej.net/software/fiji/), the FIJI Kymograph plugin, the image stabilizer plugin (K. Li, “The image stabilizer plugin for ImageJ,” https://www.cs.cmu.edu/∼kangli/code/Image_Stabilizer.html, February, 2008), and a modified version of the ADAPT plugin (Barry et al., 2015; Nanda et al., 2023). Cell Profiler (Carpenter et al., 2006) was used for cell adhesion area calculation, and data were further processed using MatLab. Data plotting and statistical analyses were performed using Prism (GraphPad). One sample ttest was used to compare values of a single condition with the expected value “0” based on the null hypothesis (Figure 2, M–O). All other statistical tests either compared two conditions via Student's ttest or more than two conditions via one-way ANOVA analysis and Dunnet's posttest.

Supporting information

Supporting Video 1. Movie S1.

Spatio-temporal activity patterns of the small GTPases Rap, and Ras in sessile A431 cells (related to Fig. 1B and C). Time-lapse TIRF video of the Rap (mCitrine-2xRalGDS-GBD) and Ras (2xRaf-GBD-mCherry) activity sensors. Images were collected with a frame rate of 12/min.

Supporting Video 2. Movie S2.

Spatio-temporal activity patterns of the small GTPases Rap in migrating A431 cells (related to Fig. 1E). Time-lapse TIRF video of the Rap activity sensor and a volume marker. Images were collected with a frame rate of 6/min.

Supporting Video 3. Movie S3.

Spatio-temporal activity patterns of the small GTPases Rap, Ras and Rac in migrating A431 cells (related to Fig. 2A,D,G,J). Time-lapse TIRF videos of the Rap, Ras, Rac or control activity sensors together with corresponding volume markers. Images were collected with a frame rate of 6/min.

Supporting Video 4. Movie S4.

Spatio-temporal activity patterns of the small GTPases Rap and Rac in migrating A431 cells (related to Fig. 3A). Time-lapse TIRF video of the Rap (mCitrine-2xRalGDS-GBD) and Rac (mCherry-3xp67phox-GBD) activity sensors in a A431 cell. Images were collected with a frame rate of 30/min.

Supporting Video 5. Movie S5.

Measurement of Rap/Rac crosstalk in A431 cells in the entire cell attachment area (related to Fig. 3C). Time-lapse TIRF videos of the Rap1a perturbation and the Rac activity response in A431 cells. Photouncaging occurred in the entire cell attachment area via TIRF illumination at 405nm. Images were collected with a frame rate of 6/min.

Supporting Video 6. Movie S6.

Measurement of Rap/Rac crosstalk in A431 cells within local spot regions in central cell attachment areas (related to Fig. 3H). Top: Time-lapse TIRF videos of the Rap1a perturbation and the Rac activity response in A431 cells. Photouncaging occurred in a diffraction-limited spot via illumination at 405nm. Images were collected with a frame rate of 6/min. Bottom: Intensity plots corresponding to the signals inside the yellow circle in videos above. The y axis corresponds to signals (arbitrary units) and the x axis corresponds to frames in the video. A vertical blue line in the intensity plots indicates the current frame shown above.

Supporting Video 7. Movie S7.

Measurement of Rac/Rap crosstalk in A431 cells in the entire cell attachment area (related to Fig. 3K). Time-lapse TIRF video of the Rac activity response in A431 cells expressing photoactivatable Rac (PA-Rac1). Images were collected with a frame rate of 12/min.

Abbreviations used:

eDHFR

E. coli dihydrofolate reductase

FAs

Focal adhesions

GAP

GTPase activating protein

GBD

GTPase binding domain

GEF

Guanine nucleotide exchange factor

RIAM

Rap1-GTP-interacting adapter molecule

TIRF-M

Total internal reflection fluorescence microscopy.