ABSTRACT
F-actin binding and bundling are crucial to a plethora of cell processes, including morphogenesis, migration, adhesion and many others. SWAP-70 was recently described as an in vitro F-actin-binding and -bundling protein. Fluorescence cross-correlation spectroscopy measurements with purified recombinant SWAP-70 confirmed that it forms stable oligomers that facilitate F-actin bundling. However, it remained unclear how SWAP-70 oligomerization and F-actin binding are controlled in living cells. We addressed this by biophysical approaches, including seFRET, FACS-FRET and FLIM-FRET. PIP3-mediated association with the cytoplasmic membrane and non-phosphorylated Y426 are required for SWAP-70 to dimerize and to bind F-actin. The dimerization region was identified near the C terminus where R546 is required for dimerization and, thus, F-actin bundling. The in vitro and in vivo data presented here reveal the functional relationship between the cytoplasm-to-membrane translocation and dimerization of SWAP-70, and F-actin binding and bundling, and demonstrate that SWAP-70 is a finely controlled modulator of membrane-proximal F-actin dynamics.
This article has an associated First Person interview with the first author of the paper.
INTRODUCTION
Proteins that govern the dynamics of cytoskeletal rearrangements are key to a wide range of biological processes. Among the multiple families of proteins that modulate the F-actin cytoskeleton in various ways, SWAP-70 is unique. This ∼70 kDa protein features an unusual arrangement of protein domains with a pleckstrin homology (PH) domain and a dbl homology (DH) domain in the N- to C-terminal order, followed by an F-actin-binding site at the C terminus. Purified recombinant SWAP-70 was shown to bind and bundle F-actin (Chacón-Martínez et al., 2013). Because only one F-actin-binding site was identified in SWAP-70, it was hypothesized that SWAP-70 bundles F-actin through dimerization or oligomerization.
Many phenotypes of SWAP-70-deficient cells reflect the role of SWAP-70 in modulating the F-actin cytoskeleton. SWAP-70, which is mainly expressed in hematopoietic cells, is required for proper formation of cell shape of activated B lymphocytes and for their directed migration in vitro and in mice (Pearce et al., 2006, 2011), for B lymphocyte adhesion and homing into tissues (Chopin et al., 2010), for mast cell and dendritic cell migration (Ocaña-Morgner et al., 2011; Sivalenka and Jessberger, 2004), and for mast cell degranulation (Gross et al., 2002; Sivalenka et al., 2008). Through F-actin modulation SWAP-70 is involved in macropinocytosis (Oberbanscheidt et al., 2007), phagocytosis (Baranov et al., 2016), osteoclast F-actin ring formation (Garbe et al., 2012; Roscher et al., 2016), and other F-actin-related processes. In various cell types, SWAP-70 localizes to sites of active F-actin rearrangement such as ruffles, lamellipodia and filopodia (Kriplani et al., 2019; Ocaña-Morgner et al., 2009; Pearce et al., 2011; Shinohara et al., 2002), and analysis of site-specific mutants indicated that a functional PH domain is required for localization to the cytoplasmic membrane (Garbe et al., 2012; Shinohara et al., 2002).
However, it remained unclear if SWAP-70 exists as a dimeric or oligomeric complex in the cytoplasm, and whether it would dimerize or oligomerize only after association with the cytoplasmic membrane or with F-actin. How would translocation to the membrane, and possibly oligomerization, be triggered? Which features define the dimerization domain of SWAP-70? Is PH domain-mediated binding to the cytoplasmic membrane required for SWAP-70 to associate with F-actin? These are among the questions addressed in this communication, in which we report a mechanistic analysis of the behaviour of SWAP-70 in vitro and in cells using a range of biochemical and biophysical methods, and present a model describing the mechanism of action of SWAP-70.
RESULTS
SWAP-70 forms highly stable oligomers in vitro
Purified SWAP-70 is known to bind and bundle F-actin (Chacón-Martínez et al., 2013). SWAP-70 features only one F-actin-binding site, which is located within the C-terminal 22 amino acids (aa) of the protein (aa 564-585) (Chacón-Martínez et al., 2013; Ihara et al., 2006). Therefore, formation of F-actin bundles should be driven by SWAP-70 oligomerization, for only dimers or higher-order oligomers would be able to bind and thus bundle two or more actin filaments. Previous data from size exclusion chromatography using recombinant, purified SWAP-70 indicated that SWAP-70 may form dimers and higher order oligomers, at least in vitro (Chacón-Martínez et al., 2013).
To follow the dynamics of SWAP-70 dimerization in solution, fluorescence cross-correlation spectroscopy (FCCS) was used. FCCS allows quantification of the concentration and diffusion of fluorescently labelled molecules as well as their interaction in a small detection volume (∼1 μm3) (Bacia and Schwille, 2007; Ries et al., 2010). Purified SWAP-70 was labelled on cysteine residues with the thiol-reactive dyes Atto 655 (red) or Alexa 488 (green). These distinctly labelled protein preparations were subsequently mixed to monitor the formation of associations of red and green SWAP-70 proteins. In FCCS experiments three curves are obtained. First, an auto-correlation curve for each, SWAP-70 Atto 655 (red) and SWAP-70 Alexa 488 (green), providing information about the concentration and diffusion of molecules in the corresponding spectral channel. Second, the cross-correlation curve (black) is generated, which reflects the formation of red/green complexes (Fig. 1A,B; Fig. S1A, Eqns 1-5). One would expect an increase in the amplitude of the black cross-correlation curve if mixed green/red complexes of SWAP-70 form, as was shown for the positive control sample containing SWAP-70 labelled simultaneously with both red and green fluorescent dyes (Fig. 1B). However, we did not detect any significant increase in amplitude of the cross-correlation curve compared to the negative control sample containing the mixture of fluorophores only; the black cross-correlation curve remained at the zero level (Fig. 1A). The corresponding calculated cross-correlation percentage (CC%) (Eqn 5) values were similar to the values for the negative control sample, in contrast to the double-labelled positive control sample, which yielded a CC% of 30% (Fig. 1C). Modifying the reaction conditions to increase the probability of dynamic exchange of red and green molecules between oligomers did not change this behaviour of the protein (Fig. S1B). Regardless of whether the temperature was increased to 37°C, whether the salt concentration was increased, or whether MgCl2, DTT, ATP or ATP and CaCl2 were included, no increase of cross-correlation above background levels, defined as <5% cross-correlation of the theoretical maximal cross-correlation, was observed.
Fluorescence cross-correlation (FCCS) measurements reveal highly stable SWAP-70 oligomers in solution. (A,B) Representative FCCS auto- and cross-correlation curves for a mixture of recombinant, purified, single-labelled SWAP-70 green and SWAP-70 red (A) and for double-labelled green/red SWAP-70 (positive control sample) (B). Dots represent experimental data and lines the corresponding fits for green channel auto-correlation (green), red channel auto-correlation (red) and cross-correlation (black). (C) Cross-correlation percentage (CC%) for control [Alexa 488 (green) or Atto 655 (red) fluorophores only], a mixture of single-labelled SWAP-70 green with SWAP-70 red, and double-labelled SWAP-70. (D) Outline of the dissociation experiments. Green-labelled SWAP-70, homodimeric, was mixed with red-labelled SWAP-70, homodimeric, in the presence of various detergents, with subsequent removal of detergents by means of Bio-Beads. Left: The presumptive re-association of red and green complexes dissociated by detergent. This was not observed. Middle: Addition of various detergents failed to dissociate the red and green homodimers. Right: After treatment with acidic pH bigger complexes containing red and green homodimers formed and dissociated when returned to neutral pH. (E) CC% for the mixture of red and green single-labelled SWAP-70 in different experimental conditions: without (wo) treatment; in citrate-phosphate buffer pH 5; in citrate-phosphate buffer pH 3; after increasing the pH of citrate-phosphate buffer to pH 7; in sodium bicarbonate buffer pH 10; in citrate-phosphate buffer pH 3 after adding and removing CHAPS; in citrate-phosphate buffer pH 3 after adding and removing CHAPS and returning the pH to 7. (F) Diffusion time of SWAP-70 in the red spectral channel under the experimental conditions listed for E. Data represent mean±s.d. from one to three independent experiments, with an average of ten cross-correlation curves analysed for each experimental condition.
Fluorescence cross-correlation (FCCS) measurements reveal highly stable SWAP-70 oligomers in solution. (A,B) Representative FCCS auto- and cross-correlation curves for a mixture of recombinant, purified, single-labelled SWAP-70 green and SWAP-70 red (A) and for double-labelled green/red SWAP-70 (positive control sample) (B). Dots represent experimental data and lines the corresponding fits for green channel auto-correlation (green), red channel auto-correlation (red) and cross-correlation (black). (C) Cross-correlation percentage (CC%) for control [Alexa 488 (green) or Atto 655 (red) fluorophores only], a mixture of single-labelled SWAP-70 green with SWAP-70 red, and double-labelled SWAP-70. (D) Outline of the dissociation experiments. Green-labelled SWAP-70, homodimeric, was mixed with red-labelled SWAP-70, homodimeric, in the presence of various detergents, with subsequent removal of detergents by means of Bio-Beads. Left: The presumptive re-association of red and green complexes dissociated by detergent. This was not observed. Middle: Addition of various detergents failed to dissociate the red and green homodimers. Right: After treatment with acidic pH bigger complexes containing red and green homodimers formed and dissociated when returned to neutral pH. (E) CC% for the mixture of red and green single-labelled SWAP-70 in different experimental conditions: without (wo) treatment; in citrate-phosphate buffer pH 5; in citrate-phosphate buffer pH 3; after increasing the pH of citrate-phosphate buffer to pH 7; in sodium bicarbonate buffer pH 10; in citrate-phosphate buffer pH 3 after adding and removing CHAPS; in citrate-phosphate buffer pH 3 after adding and removing CHAPS and returning the pH to 7. (F) Diffusion time of SWAP-70 in the red spectral channel under the experimental conditions listed for E. Data represent mean±s.d. from one to three independent experiments, with an average of ten cross-correlation curves analysed for each experimental condition.
Nevertheless, the diffusion time of SWAP-70 molecules deducted from the auto-correlation curves was higher than would have been expected for a monomeric protein (in the red channel 180 µs is expected for a monomer) (Fig. S1C). The calculated number of molecules per diffusing red or green particles using two different calibration standards (Atto 655- or Alexa 488-labelled monomeric 42 kDa SWAP-70 N-terminal fragment; Fig. S1D) and Atto 655 or Alexa 488 fluorophores was more than two (Table S1, Eqn 6) (Ries et al., 2010). This observation led us to the conclusion that recombinant SWAP-70 formed oligomers prior to mixing of differently labelled protein batches, and that the red and green complexes were very stable, did not dissociate, and thus did not mix with each other. The variations in experimental conditions mentioned above did not alter the diffusion time, supporting the hypothesis that purified SWAP-70 forms highly stable complexes (Fig. S1C). These complexes are functional in that they bind and bundle F-actin.
Further approaches to dissociate the red and green oligomers and to obtain mixed complexes were undertaken by addition of various detergents, which were subsequently removed by means of Bio-Beads. We expected that this would allow re-association and thus assembly of monomers into dimers (Fig. 1D, left). Detergents such as Triton X-100 (0.1%), Tween-20 (0.1%), SDS (2%), cholate (2%) and CHAPS (2%) did not induce any significant mixing of green and red complexes (Fig. S1E). In addition, the diffusion time still pointed to the existence of oligomeric species. This argues against the possibility of incomplete removal of dissociating detergents as that would result in monomers (Fig. S1F). Most likely, these detergents did not dissociate the existing complexes (Fig. 1D, middle). The presence of charged residues in the C-terminal region, suggested to be required for oligomerization (Chacón-Martínez et al., 2013), indicates that electrostatic interactions might be involved in oligomerization. Thus, pH variations may affect the dynamics of oligomerization. Indeed, changing the pH from neutral to acidic (pH 5 or pH 3) slightly increased cross-correlation and shifted the equilibrium towards the formation of red/green associations (Fig. 1E). This effect was reversible as the number of red/green complexes decreased when the pH was brought back to 7 (Fig. 1E), indicating that the newly formed green/red associations were stable only at acidic pH. A change to pH 10 did not have an effect (Fig. 1E). The addition of CHAPS at acidic pH dramatically increased the cross-correlation values, pointing to the formation of mixed red/green oligomers after the removal of the detergent (Fig. 1E). Again, reversal of the pH from 3 to 7 reduced the number of mixed complexes back to control values (Fig. 1E). Interestingly, the diffusion time of the oligomers at pH 3 after detergent treatment also increased (Fig. 1F), suggesting the formation of bigger complexes. A likely interpretation of the increase in diffusion time at low pH is the formation of bigger complexes containing associations of red only and green only homodimers. These complexes fell apart after the pH was returned to neutral (Fig. 1D, right). In this case, the behaviour of SWAP-70 at low pH might be physiologically irrelevant and the mixed complexes were rather artificial.
We concluded that the recombinant, functionally active protein forms highly stable oligomers. Nevertheless, it has to be considered that in these studies a bacterial expression system was used. It is possible that in eukaryotic expression systems, in which relevant post-translational modifications may take place, the oligomeric status of the protein might be different. To reveal even more physiologically relevant SWAP-70 properties, experiments in living cells were performed.
SWAP-70 dimerizes at the membrane in live cells
To detect SWAP-70 dimerization in vivo, fluorescence resonance energy transfer (FRET) measurements in B16 melanoma cells deficient in endogenous SWAP-70, expressing two differently tagged SWAP-70 proteins, were conducted. Compared to FCCS studies, FRET allows analysis of protein oligomerization in fluorescently labelled cellular compartments, including non-homogeneous and dynamic membrane-proximal compartments such as ruffles. First, sensitized emission FRET (seFRET) was chosen as a method to follow the formation of mixed oligomers in cells transfected with SWAP-70-Cerulean (donor) and SWAP-70-Venus (acceptor). In seFRET, the increase in acceptor emission in presence of the donor is detected when the distance between interacting partners is less than 10 nm. FRET efficiency can be calculated from donor and acceptor microscopy images after performing cross-talk corrections, required because of overlap of the Cerulean and Venus emission spectra. In untreated cells, SWAP-70 was almost exclusively present in the cytoplasm (Fig. 2A, left; SWAP-70-GFP) and FRET efficiency was low (Fig. 2B,C), suggesting a monomeric state of the protein. Thus, whereas recombinant, purified protein highly stably oligomerizes, in untreated mammalian cells the protein remains monomeric.
SWAP-70 translocates to the membrane and dimerizes after cell stimulation. (A) Representative confocal fluorescence microscopy images of untreated (left) and vanadate-treated (right) B16 melanoma cells overexpressing wt SWAP-70-GFP. SWAP-70 translocates to the membrane upon stimulation. (B) Representative microscopy-based seFRET image for interaction of SWAP-70-Cerulean and SWAP-70-Venus in the cell. In untreated cells in the cytoplasm (upper image) FRET efficiency is low; in treated cells at the membrane (lower image) FRET efficiency is higher, pointing to SWAP-70 dimerization. (C) FRET efficiency for SWAP-70-Cerulean and SWAP-70-Venus interaction in cells untreated or vanadate-treated calculated from seFRET images. Cells overexpressing wt SWAP-70, the AB mutant (Δ565-585), the PH mutant (K219A, K220A), wt SWAP-70 in cells treated with PIP3 kinase inhibitor wortmannin, and the Y426F mutant. SWAP-70 PH mutant and wt SWAP-70 in cells treated with wortmannin fail to dimerize. Data represent mean±s.d. for three independent experiments, 10-15 cells were analysed in each experiment. (D-H) Representative confocal fluorescence microscopy images of untreated (left) and vanadate-treated (right) cells overexpressing SWAP-70-GFP mutants: (D) AB mutant (Δ565-585), (E) PH mutant (K219A, K220A), (F) Y517F mutant, (G) Y426F mutant, (H) Y426E mutant. Scale bars: 5 µm.
SWAP-70 translocates to the membrane and dimerizes after cell stimulation. (A) Representative confocal fluorescence microscopy images of untreated (left) and vanadate-treated (right) B16 melanoma cells overexpressing wt SWAP-70-GFP. SWAP-70 translocates to the membrane upon stimulation. (B) Representative microscopy-based seFRET image for interaction of SWAP-70-Cerulean and SWAP-70-Venus in the cell. In untreated cells in the cytoplasm (upper image) FRET efficiency is low; in treated cells at the membrane (lower image) FRET efficiency is higher, pointing to SWAP-70 dimerization. (C) FRET efficiency for SWAP-70-Cerulean and SWAP-70-Venus interaction in cells untreated or vanadate-treated calculated from seFRET images. Cells overexpressing wt SWAP-70, the AB mutant (Δ565-585), the PH mutant (K219A, K220A), wt SWAP-70 in cells treated with PIP3 kinase inhibitor wortmannin, and the Y426F mutant. SWAP-70 PH mutant and wt SWAP-70 in cells treated with wortmannin fail to dimerize. Data represent mean±s.d. for three independent experiments, 10-15 cells were analysed in each experiment. (D-H) Representative confocal fluorescence microscopy images of untreated (left) and vanadate-treated (right) cells overexpressing SWAP-70-GFP mutants: (D) AB mutant (Δ565-585), (E) PH mutant (K219A, K220A), (F) Y517F mutant, (G) Y426F mutant, (H) Y426E mutant. Scale bars: 5 µm.
We hypothesized that oligomerization of SWAP-70 in cells might depend on its phosphorylation state, because at least one tyrosine phosphorylation event is known to regulate SWAP-70's function and interaction with the F-actin cytoskeleton (Pearce et al., 2011). Indeed, treatment of the cells with the tyrosine phosphatase inhibitor sodium vanadate dramatically changed SWAP-70 localization. Most of the protein translocated to the membrane by approximately 8 min after addition of vanadate (Fig. 2A, right, Movie 1). The effect of vanadate was reversible. After medium exchange SWAP-70's cytoplasmic localization was restored, confirming a continuous functional state of the protein upon and after vanadate treatment (Movie 2). Vanadate stimulation had a long-lasting effect resulting in the presence of SWAP-70 in the membrane for at least 2 h. This allowed stable seFRET measurements to be conducted. Interestingly, FRET efficiency was significantly increased when the protein translocated close to the plasma membrane, indicating SWAP-70 dimerization or oligomerization at the membrane (Fig. 2B,C). As the number of SWAP-70 molecules in the complex cannot be deduced from these measurements we will refer to this complexes as dimers. Therefore, seFRET proved to be a useful approach to measure SWAP-70 dimerization in vivo. However, seFRET has important limitations. For seFRET, several microscopy images with combinations of excitation and emission filters using samples containing donor only, acceptor only and the FRET sample that contains both fluorophores, need to be acquired. The images have to be further processed to subtract cross-talk components. This significantly increases the noise levels in the final FRET image (Fig. 2B), especially at the membrane compartment.
To distinguish if F-actin or membrane binding of SWAP-70 is responsible for the vanadate-induced protein translocation and dimerization, a mutant lacking the F-actin-binding site (AB mutant, deletion of aa 565-585) and a PH domain mutant lacking residues required for lipid binding (K219A, K220A) were tested. The F-actin-binding domain was not required for SWAP-70-GFP translocation and dimerization (Fig. 2C,D). However, membrane binding via the PH domain proved necessary. The PH mutant of SWAP-70 did not translocate to the membrane after vanadate treatment and the protein remained monomeric in the cytoplasm (Fig. 2C,E). The same result was obtained using a different PH domain mutant (Δ200-225; not shown). In addition, the presence of phosphatidylinositol (3,4,5)-trisphosphate (PIP3) in the membrane appeared to be required for SWAP-70 membrane association and dimerization as the treatment of the cells with PI3 kinase inhibitor wortmannin precluded SWAP-70 membrane binding and dimerization after vanadate stimulation (Fig. 2C; Fig. S2A).
As SWAP-70 membrane translocation induced by the tyrosine phosphatase inhibitor appears to drive its dimerization, we investigated the role of tyrosine residues of SWAP-70. A series of SWAP-70 tyrosine mutants was tested. Deletion of the first 197 aa, which includes tyrosines at positions 69, 72, 130, 141 and 162, did not significantly affect protein translocation to the membrane in vanadate-treated cells (Fig. S2B). This also confirms that there are no protein domains within the N-terminal 197 aa that are required for membrane association. Mutation of individual tyrosine residues 216 or 241/242 located near the PH region of SWAP-70 also did not change the effect of cell stimulation (Fig. S2C,D). Similarly, the deletion mutant Δ353-585 (Fig. S2E), the mutant Y302F (Fig. S2F) and the mutant Y517F, affecting a tyrosine known to be phosphorylated upon B-cell activation (Pearce et al., 2011), did not show a difference compared with the wild-type (wt) protein in response to vanadate treatment (Fig. 2F).
Surprisingly, even in untreated cells the mutant Y426F showed a drastically different cellular distribution compared with wt protein. This mutant protein was largely associated with the cellular membrane in untreated cells and showed an even more pronounced membrane accumulation after vanadate treatment (Fig. 2G). Despite its membrane association, the Y426F protein did not show any significant dimerization without vanadate treatment (Fig. 2C). One explanation is that, as Y426F cannot be phosphorylated, membrane association occurs but another event, triggered by vanadate, drives dimerization. To test this hypothesis, we created the SWAP-70 phosphomimetic mutant Y426E, which we hypothesized would not translocate to the membrane as it resembles a protein carrying a charged phospho-group. However, Y426E did translocate to the cortex after vanadate treatment (Fig. 2H). Thus, phosphorylation at this position is not required for membrane association. Another explanation is that Y426F associates less tightly with the membrane, and instead associates directly with the F-actin cytoskeleton in close proximity to the membrane without dimerization and thus without F-actin bundling.
SWAP-70 Y426F localizes to F-actin patches
Visualization of actin by LifeAct-RFP showed and confirmed localization of the Y426F protein together with F-actin at the cell cortex and cytoplasmic distribution of the wt protein or the AB mutant (Fig. S3A-C). The intensity of Y426F at the membrane in areas of F-actin accumulation was much higher than in the cytoplasm, whereas for the wt protein the intensity was homogeneous within the cell (Fig. S3D). Treatment of cells with cytochalasin D, an inhibitor of actin polymerization, disrupted the actin cytoskeleton leaving SWAP-70 associated with the remaining actin patches (Fig. 3A, Fig. S3E,H,K-M). The AB mutant was almost absent from actin patches (Fig. 3B, Fig. S3F,I,K-M). The Y426F mutant also localized to actin patches, similar to wt protein (Fig. 3C, Fig. S3G,J-M); however, more Y426F-containing patches were present near the membrane (Fig. 3C, Fig. S3J,N). Given the different localizations of the Y426F or AB mutant and wt proteins, we compared their distribution in vanadate-treated cells. As expected, localization of the Y426F mutant in untreated cells was similar to the wt and AB mutant distribution in treated cells. When the equatorial plane of cells was imaged, all proteins localized close to the membrane (Fig. 3D-F). When the bottom of the cell was imaged, the AB mutant showed fewer SWAP-70-associated actin patches or foci compared to wt and Y426F mutant proteins (Fig. 3D-F, Fig. S3O). Wt SWAP-70 in treated cells and Y426F protein in untreated cells localized similarly in actin patches at the cell bottom (Fig. 3D,F, Fig. S3O). To confirm binding of the Y426F protein to F-actin, AB and PH domains were separately deleted in the Y426F mutant and the two double mutants were investigated. Both additional mutations significantly changed Y426F protein localization in untreated cells. Both double-mutant proteins localized to the cytoplasm in a similar manner to SWAP-70 carrying the single AB or PH domain deletion (Fig. 3G,H). It thus appears that despite the presence of non-phosphorylatable Y426F the AB and PH domains both support association of Y426F protein with membrane-proximal F-actin in untreated cells.
SWAP-70 Y426F mutant localizes to F-actin patches. Representative confocal fluorescence microscopy images of untreated cells expressing wt or mutant SWAP-70 GFP (green) and LifeAct-RFP (red), visualizing actin. (A-C) Cells treated with cytochalasin D. Actin patches are indicated with arrows. On the merged GFP-RFP image, intensity profiles along the white lines of red and green channels are presented. (A) wt SWAP-70 GFP, protein localizes to the patches. (B) AB mutant SWAP-70 GFP, protein is homogeneously distributed in the cytoplasm. (C) Y426F mutant of SWAP-70 GFP, protein localizes in the patches. (D-F) Left: Cell equatorial plane. Right: Cell bottom. (D) wt SWAP-70 GFP in vanadate-treated cells. (E) AB mutant SWAP-70 GFP in vanadate-treated cells. (F) Y426F mutant of SWAP-70 GFP in untreated cells. (G) AB mutant of SWAP-70 (left); PH mutant of SWAP-70 (right) (H) AB mutant of SWAP-70 Y426F mutant (left); PH mutant of SWAP-70 Y426F mutant (right). Scale bars: 5 µm.
SWAP-70 Y426F mutant localizes to F-actin patches. Representative confocal fluorescence microscopy images of untreated cells expressing wt or mutant SWAP-70 GFP (green) and LifeAct-RFP (red), visualizing actin. (A-C) Cells treated with cytochalasin D. Actin patches are indicated with arrows. On the merged GFP-RFP image, intensity profiles along the white lines of red and green channels are presented. (A) wt SWAP-70 GFP, protein localizes to the patches. (B) AB mutant SWAP-70 GFP, protein is homogeneously distributed in the cytoplasm. (C) Y426F mutant of SWAP-70 GFP, protein localizes in the patches. (D-F) Left: Cell equatorial plane. Right: Cell bottom. (D) wt SWAP-70 GFP in vanadate-treated cells. (E) AB mutant SWAP-70 GFP in vanadate-treated cells. (F) Y426F mutant of SWAP-70 GFP in untreated cells. (G) AB mutant of SWAP-70 (left); PH mutant of SWAP-70 (right) (H) AB mutant of SWAP-70 Y426F mutant (left); PH mutant of SWAP-70 Y426F mutant (right). Scale bars: 5 µm.
Amino acids required for SWAP-70 dimerization
In vitro studies with recombinant purified SWAP-70 indicated that the dimerization region is located somewhere near the C terminus of the protein (Chacón-Martínez et al., 2013). The dimerization interface of the protein in cells can differ from that observed in vitro. Nevertheless, we tested C-terminal residues for involvement in dimerization of SWAP-70. Mutants covering aa 444-585 were generated and tested by seFRET in living cells treated with vanadate. aa 564-585 comprise the SWAP-70 F-actin-binding site and are not involved in dimerization, for we showed above that the AB mutant of SWAP-70 formed dimers at the membrane, similar to wt protein (Fig. 2C). Among the remaining aa 444-564 only the deletion of aa 500-507 or 526-564 significantly reduced FRET efficiency, indicating the importance of these residues for SWAP-70 dimerization (Fig. 4A). Deletion of aa 545-564 within the 526-564 region also decreased dimerization efficiency.
Amino acids important for SWAP-70 dimerization revealed by microscopy-based seFRET and FACS-seFRET. (A) seFRET efficiency calculated from microscopy images for cells co-expressing SWAP-70 Cerulean and SWAP-70 Venus to follow oligomerization in untreated and treated cells for wt SWAP-70, and its mutants lacking aa 444-478, 478-499, 500-507, 508-525, 526-564, 545-564. Data represent mean±s.d. for three independent experiments, 10-15 cells analysed for each experiment. (B,D) Representative FACS-FRET plots for double-positive (SWAP-70 Cerulean and SWAP-70 Venus) B16 (B) and 293T (D) cells in the FRET channel (donor excitation-acceptor emission) and Cerulean channel (donor excitation-donor emission) at different time points between 0 and 15 min after vanadate stimulation. The percentage of FRET-positive cells falling within this gate is shown in the left corner of each plot. (D). (C) Percentage of FRET-positive cells obtained by the FACS-seFRET approach in vanadate-treated cells transfected with the FRET-positive standard or with wt SWAP-70 Cerulean and wt SWAP-70 Venus. (E) Representative image of SWAP-70-GFP in vanadate-treated 293T cells. (F) seFRET efficiency for SWAP-70 Cerulean and SWAP-70 Venus in untreated and vanadate-treated 293T cells. (G) Representative image of Y426F SWAP-70-GFP in 293T cells. (H) Percentage of FRET-positive cells obtained by FACS-seFRET for vanadate-treated cells co-transfected with wt SWAP-70 Cerulean and wt SWAP-70 Venus or its mutants with the deletions 500-507, 526-564, 545-564. Mutants show decreased dimerization. Data represent mean±s.d. for three to four independent experiments, 300,000-1,000,000 cells per each experiment analysed as depicted in B,D. Scale bars: 5 µm.
Amino acids important for SWAP-70 dimerization revealed by microscopy-based seFRET and FACS-seFRET. (A) seFRET efficiency calculated from microscopy images for cells co-expressing SWAP-70 Cerulean and SWAP-70 Venus to follow oligomerization in untreated and treated cells for wt SWAP-70, and its mutants lacking aa 444-478, 478-499, 500-507, 508-525, 526-564, 545-564. Data represent mean±s.d. for three independent experiments, 10-15 cells analysed for each experiment. (B,D) Representative FACS-FRET plots for double-positive (SWAP-70 Cerulean and SWAP-70 Venus) B16 (B) and 293T (D) cells in the FRET channel (donor excitation-acceptor emission) and Cerulean channel (donor excitation-donor emission) at different time points between 0 and 15 min after vanadate stimulation. The percentage of FRET-positive cells falling within this gate is shown in the left corner of each plot. (D). (C) Percentage of FRET-positive cells obtained by the FACS-seFRET approach in vanadate-treated cells transfected with the FRET-positive standard or with wt SWAP-70 Cerulean and wt SWAP-70 Venus. (E) Representative image of SWAP-70-GFP in vanadate-treated 293T cells. (F) seFRET efficiency for SWAP-70 Cerulean and SWAP-70 Venus in untreated and vanadate-treated 293T cells. (G) Representative image of Y426F SWAP-70-GFP in 293T cells. (H) Percentage of FRET-positive cells obtained by FACS-seFRET for vanadate-treated cells co-transfected with wt SWAP-70 Cerulean and wt SWAP-70 Venus or its mutants with the deletions 500-507, 526-564, 545-564. Mutants show decreased dimerization. Data represent mean±s.d. for three to four independent experiments, 300,000-1,000,000 cells per each experiment analysed as depicted in B,D. Scale bars: 5 µm.
Microscopy-based seFRET measurements are time consuming and limited to low numbers of cells analysed per measurement. To include a much higher number of cells in the analysis, FACS-FRET, a flow cytometry-based FRET assay that is also based on sensitized emission (Banning et al., 2010), was used. Here, FRET efficiency is expressed as percentage of FRET-positive cells. First, intensity of double-positive (Cerulean and Venus) cells in the FRET channel (donor excitation-acceptor emission) is plotted against their intensity in the Cerulean channel (donor excitation-donor emission) (Fig. 4B). A FACS-positive gate (magenta), which excludes FRET-negative cells, is defined (Fig. 4B). Dimerization of SWAP-70 at different time points after vanadate treatment was assessed simultaneously in 300,000 to 1×106 cells by FACS-FRET. As a positive FRET standard, a Cerulean-Venus-SWAP-70 protein showed 100% FRET-positive cells, thus validating our FACS-FRET approach (Fig. 4C). For B16 cells co-expressing SWAP-70-Cerulean and SWAP-70 Venus, the percentage of FRET-positive cells falling within the FRET-positive gate increased with time after vanadate treatment, reaching 81.6% at 15 min (Fig. 4B). However, the transfection efficiency for B16 cells, and therefore the number of events collected, was low. To improve statistics and to validate our findings further, HEK 293T (293T) cells, which could be efficiently transfected, were tested. 293T cells treated with vanadate also showed a gradual increase in the percentage of FRET-positive cells co-expressing SWAP-70 donor and acceptor and in addition higher numbers of collected events (Fig. 4D). B16 and 293T cells showed similar percentages of FRET-positive cells 15 min after vanadate treatment (Fig. 4B-D). SWAP-70 localization in the 293T cell line was similar to that in B16 cells as it translocated to the membrane after vanadate treatment (Fig. 4E). Microscopy-based seFRET measurements confirmed SWAP-70 dimerization at the membrane (Fig. 4F). Mutant Y426F also localized to the cell cortex in untreated 293T cells, similar to B16 cells (Fig. 4G). Thus, 293T cells reproduced the results obtained with B16 and were suitable for performing FACS-FRET experiments (Fig. 4H). Deletion of residues 500-507, 526-564 or 545-564 resulted in reduced FRET-positive cells in agreement with our microscopy-based seFRET data (Fig. 4H). This supports our conclusion that residues 500-507 and 526-564 promote SWAP-70 dimerization.
seFRET allows the analysis of large numbers of cells, as in FACS-FRET, but is an intensity-based method and therefore prone to certain artefacts, caused for example by cross-talk between the fluorescent channels and by the dependence of the results on the intensities of the donor and acceptor read-outs. External calibrations are required prior to all measurements. In addition, as mentioned above, measurements in the membrane compartment had relatively high background resulting in a noisier signal in the final seFRET image (Fig. 2B). Therefore, we employed an additional method, fluorescence lifetime imaging microscopy FRET (FLIM-FRET). Although not suitable for FACS, FLIM-FRET does not rely on intensity measurements and follows changes over the lifetime of a donor in the presence of an acceptor. The lower the lifetime values, the higher the FRET efficiency and thus the association of donor and acceptor molecules. In agreement with our results obtained by seFRET, after treatment of B16 cells with vanadate, the lifetime of SWAP-70-Cerulean (donor) was much lower at the membrane compared to its lifetime in the cytoplasm, suggesting membrane-proximal protein dimerization (Fig. 5A,B, Fig. S4). FLIM images also had a higher signal-to-noise ratio compared to seFRET (Fig. 5A).
Amino acids important for SWAP-70 dimerization revealed by FLIM-FRET. (A) Representative FLIM-FRET images of cells co-transfected with wt SWAP-70 Cerulean and wt SWAP-70 Venus or its mutants. Mean Cerulean fluorescence lifetime (ps) in the cell is presented in colour code according to the scale on the left. From left to right, Cerulean lifetime is shown for: wt SWAP-70 in untreated cells, wt SWAP-70, SWAP-70 Δ526-564 mutant, SWAP-70 R546A mutant and Y426E mutant in treated cells. Cerulean lifetime for wt SWAP-70 in treated cells is lower compared to the indicated mutants. (B,C) Cerulean fluorescence lifetime (ps) calculated from FLIM-FRET images obtained in untreated and treated cells for wt SWAP-70 and its mutants. (B) Cerulean lifetime for wt SWAP-70 and SWAP-70 mutants lacking the residues 526-564, 500-507, 526-534, 535-544. Mutant Δ526-564 showed a higher donor lifetime. (C) Cerulean lifetime for wt SWAP-70 and SWAP-70 mutants R546A, K553A, Y426F and Y426E. Mutant R546A and Y426E showed higher donor lifetimes in treated cells. Data represent mean±s.d. for three to six independent experiments, five cells analysed per each experiment. ***P<0.001; n.s., P>0.05, not significant (unpaired t-test). (D) Percentage of FRET-positive cells obtained by FACS-seFRET for vanadate-treated cells co-transfected with wt SWAP-70 Cerulean and wt SWAP-70 Venus or its mutant R546A. Experiment analysed as depicted in Fig. 4B,D. (E,F) Bundling of actin filaments in a low-speed co-sedimentation assay by purified wt SWAP-70 and R546A mutant. Less bundled actin was present in the pellet when actin was incubated with R546A mutant. (E) Supernatants (S) containing non-bundled F-actin and pellets (P) containing bundled F-actin were analysed by Coomassie-stained gels. Representative gel from one experiment is shown. (F) Densitometry analysis of the gel bands from the gel shown in E for bundled actin (pellets) for the sample containing F-actin only, or the mixture of F-actin with wt SWAP-70 or the R546A mutant. (G) Binding of actin filaments in a high-speed co-sedimentation assay by purified wt SWAP-70 and the R546A mutant. Densitometry analysis of the Coomassie gel bands from for actin-bound SWAP-70 (pellets) for the sample containing the mixture of F-actin with wt SWAP-70 or the R546A mutant is shown. (H) Diffusion time of purified wt SWAP-70 and the R546A mutant in solution determined by FCS. All data represent mean±s.d. from three independent experiments.
Amino acids important for SWAP-70 dimerization revealed by FLIM-FRET. (A) Representative FLIM-FRET images of cells co-transfected with wt SWAP-70 Cerulean and wt SWAP-70 Venus or its mutants. Mean Cerulean fluorescence lifetime (ps) in the cell is presented in colour code according to the scale on the left. From left to right, Cerulean lifetime is shown for: wt SWAP-70 in untreated cells, wt SWAP-70, SWAP-70 Δ526-564 mutant, SWAP-70 R546A mutant and Y426E mutant in treated cells. Cerulean lifetime for wt SWAP-70 in treated cells is lower compared to the indicated mutants. (B,C) Cerulean fluorescence lifetime (ps) calculated from FLIM-FRET images obtained in untreated and treated cells for wt SWAP-70 and its mutants. (B) Cerulean lifetime for wt SWAP-70 and SWAP-70 mutants lacking the residues 526-564, 500-507, 526-534, 535-544. Mutant Δ526-564 showed a higher donor lifetime. (C) Cerulean lifetime for wt SWAP-70 and SWAP-70 mutants R546A, K553A, Y426F and Y426E. Mutant R546A and Y426E showed higher donor lifetimes in treated cells. Data represent mean±s.d. for three to six independent experiments, five cells analysed per each experiment. ***P<0.001; n.s., P>0.05, not significant (unpaired t-test). (D) Percentage of FRET-positive cells obtained by FACS-seFRET for vanadate-treated cells co-transfected with wt SWAP-70 Cerulean and wt SWAP-70 Venus or its mutant R546A. Experiment analysed as depicted in Fig. 4B,D. (E,F) Bundling of actin filaments in a low-speed co-sedimentation assay by purified wt SWAP-70 and R546A mutant. Less bundled actin was present in the pellet when actin was incubated with R546A mutant. (E) Supernatants (S) containing non-bundled F-actin and pellets (P) containing bundled F-actin were analysed by Coomassie-stained gels. Representative gel from one experiment is shown. (F) Densitometry analysis of the gel bands from the gel shown in E for bundled actin (pellets) for the sample containing F-actin only, or the mixture of F-actin with wt SWAP-70 or the R546A mutant. (G) Binding of actin filaments in a high-speed co-sedimentation assay by purified wt SWAP-70 and the R546A mutant. Densitometry analysis of the Coomassie gel bands from for actin-bound SWAP-70 (pellets) for the sample containing the mixture of F-actin with wt SWAP-70 or the R546A mutant is shown. (H) Diffusion time of purified wt SWAP-70 and the R546A mutant in solution determined by FCS. All data represent mean±s.d. from three independent experiments.
The mutant protein lacking residues 526-564 displayed a higher donor lifetime in treated cells compared to the wt protein (Fig. 5A,B). Increased lifetime may indicate some impairment of SWAP-70 dimerization, or could be a result of a conformational change of the mutant protein leading to altered donor-acceptor orientation or localization. Nevertheless, the result was in agreement with seFRET (Fig. 4A) data showing decreased FRET efficiency for this mutant. Unexpectedly, the SWAP-70 Δ500-507 mutant showed a behaviour similar to the wt protein (Fig. 5B). This disagreement with seFRET results might be explained by a possible lower abundance of the mutant at the membrane. This could affect donor intensity and therefore seFRET and seFACS-FRET values.
To specify the residues crucial for SWAP-70 dimerization more precisely, aa 526-564 were further mutated and tested in the FLIM-FRET assay. As was shown in seFRET and FACS-FRET assays, the absence of residues 545-564 was sufficient to reduce the percentage of FRET-positive cells and possibly SWAP-70 dimerization (Fig. 4A,H). However, it was still not clear if the residues between aa 526 and 545 are also important for dimer formation. Deletion of aa 526-534 or 535-544 (Fig. 5B) did not significantly affect SWAP-70 donor lifetime, suggesting that only the region 545-564 contains aa that are involved in dimerization. Our in vitro data hinted at the importance of electrostatic interactions for SWAP-70 dimerization, thus we changed positively charged arginine 546 and lysine 553 to alanine. Replacement of R546, but not of K553, with alanine increased the fluorescence lifetime to values similar to those obtained for the mutant Δ526-564 (Fig. 5A,C), suggesting an important role of R546 in promoting SWAP-70 dimerization. This role of R546 was confirmed in FACS-FRET experiments, in which strongly reduced SWAP-70 dimerization was observed when the R546A mutant was expressed (Fig. 5D).
To obtain more information from FLIM measurements, two-component fit of the data for wt protein and for mutants R546A and K553A was performed (Fig. S5A). The donor lifetimes of the first and the second component were fixed to the average lifetimes obtained in the cytoplasm of unstimulated cells and at the membrane of vanadate-treated cells for wt protein (2544 and 2134 ns, respectively) and the fraction of interacting molecules was estimated. As expected, the 546 mutant, but not the 553 mutant, showed a lower fraction of interacting molecules in vanadate-treated cells (Fig. S5A), suggesting a lower probability of donor-acceptor interaction for this mutant.
To validate further our finding that R546 supports dimerization of SWAP-70 and therefore its F-actin bundling activity, we overexpressed this mutant in Escherichia coli, purified the protein and performed experiments in vitro. The mutant protein was tested in a low-speed co-sedimentation F-actin bundling assay, in which most of the bundled actin is sedimented in the pellet. A much lower amount of bundled actin was obtained with the R546A mutant compared to the wt protein (Fig. 5E,F). This decrease in bundled F-actin could be explained either by reduced binding of SWAP-70 to F-actin or reduced dimerization. To exclude the possibility of reduced F-actin binding of R546A, we tested this mutant in high-speed co-sedimentation assays. The R546A mutant bound F-actin similarly to wt (Fig. 5G). Therefore, reduced bundling activity of R546A should be attributed to its impaired dimerization. To test this further, we labelled purified R546A mutant with the thiol-reactive dye Atto 655 and measured its diffusion time using fluorescence correlation spectroscopy (FCS), which is similar to FCCS approach described above except that only one fluorophore is used and one auto-correlation curve is generated. The diffusion time was reduced for this mutant compared to wt protein (Fig. 5H). The mean diffusion time value was 198.4±30 μs, which would correspond to 1.3±0.6 SWAP-70 molecules per diffusing particle and suggests the presence of SWAP-70 monomers.
These data support our hypothesis that R546A is needed for SWAP-70 dimerization and for F-actin bundling. The FLIM-FRET approach was further used to investigate if phosphorylation of Y426 is important for dimerization. Y426F, which cannot be phosphorylated, and the phosphomimetic mutant Y426E were tested by FLIM. The Y426F mutant showed a slightly decreased lifetime in untreated cells compared to wt protein (Fig. 5C) but not to the same extent as treated cells. Two-component fit of FLIM data for Y426F showed an increase in the fraction of the interacting molecules already in untreated cells, but again the values were different from values of treated cells (Fig. S5B). In vanadate-treated cells, Y426F behaved similarly to wt protein (Fig. 5C, Fig. S5B). In contrast, Y426E behaved similarly to the dimerization mutant R546A (Fig. 5A,C, Fig. S5A,B). This suggests that SWAP-70 needs to be dephosphorylated at Y426 to dimerize at the membrane. Interestingly, F-actin was not required for SWAP-70 dimerization. In cells treated with cytochalasin D, interaction of SWAP-70 with F-actin was disrupted as expected (Fig. S6A). However, both wt and Y426F mutant proteins translocated to the membrane and dimerized (Fig. S6B,C).
SWAP-70 binding to F-actin in live cells
To investigate the dynamic interaction of SWAP-70 with F-actin in cells and its relationship to membrane binding, we performed FLIM-FRET experiments to determine the lifetime of SWAP-70-GFP in the presence of the F-actin reactive probe LifeAct-RFP in treated and untreated cells. SWAP-70-GFP lifetime decreased in the presence of LifeAct-RFP in treated cells, reflecting the enhanced interaction of SWAP-70 with F-actin close to the membrane (Fig. 6). F-actin also showed pronounced colocalization with SWAP-70 in membrane ruffles after vanadate stimulation (Movie 3). As expected, SWAP-70 AB and PH mutant proteins did not significantly interact with F-actin after stimulation (Fig. 6). These data suggest that SWAP-70 membrane translocation is required for its interaction with F-actin. We hypothesize that membrane association and dimerization are prerequisites for SWAP-70 F-actin bundling.
SWAP-70 binds to F-actin in live cells. (A) Representative FLIM-FRET images. Cells were co-transfected with wt SWAP-70 GFP or its mutants and actin LifeAct-RFP. Mean GFP fluorescence lifetime (ps) in the cell is presented in colour code according to the scale on the left. From left to right, GFP lifetime is shown for untreated and treated cells: wt SWAP-70, AB mutant, PH mutant,Y426F mutant, Y426E mutant and R546A mutant. Lifetime for the AB mutant in treated cells is higher than for wt protein, suggesting no association of this mutant with actin. Lifetime for the Y426 mutant in untreated cells is lower than the lifetime of the wt protein, showing the binding of this mutant to actin even in untreated cells. (B) GFP fluorescence lifetime (ps) calculated from FLIM-FRET images obtained in untreated and treated cells for wt SWAP-70, AB mutant, PH mutant, Y426F mutant, Y426E mutant and R546A mutant. The AB and PH mutants did not associate with actin, whereas the Y426F mutant associated with actin in treated and untreated cells. The R546A mutant associated with actin in treated cells. Y426E showed impaired actin binding in treated cells. Data represent the mean of three to six independent experiments, five cells analysed per experiment. *P<0.05 (unpaired t-test).
SWAP-70 binds to F-actin in live cells. (A) Representative FLIM-FRET images. Cells were co-transfected with wt SWAP-70 GFP or its mutants and actin LifeAct-RFP. Mean GFP fluorescence lifetime (ps) in the cell is presented in colour code according to the scale on the left. From left to right, GFP lifetime is shown for untreated and treated cells: wt SWAP-70, AB mutant, PH mutant,Y426F mutant, Y426E mutant and R546A mutant. Lifetime for the AB mutant in treated cells is higher than for wt protein, suggesting no association of this mutant with actin. Lifetime for the Y426 mutant in untreated cells is lower than the lifetime of the wt protein, showing the binding of this mutant to actin even in untreated cells. (B) GFP fluorescence lifetime (ps) calculated from FLIM-FRET images obtained in untreated and treated cells for wt SWAP-70, AB mutant, PH mutant, Y426F mutant, Y426E mutant and R546A mutant. The AB and PH mutants did not associate with actin, whereas the Y426F mutant associated with actin in treated and untreated cells. The R546A mutant associated with actin in treated cells. Y426E showed impaired actin binding in treated cells. Data represent the mean of three to six independent experiments, five cells analysed per experiment. *P<0.05 (unpaired t-test).
The SWAP-70 Y426F mutant, which associated with F-actin without cell treatment, showed donor lifetime values similar to the wt protein upon stimulation. This confirms the direct association of this mutant with the F-actin cytoskeleton in untreated cells (Fig. 6). Interestingly, mutant Y426E showed increased donor lifetime in vanadate-treated cells, suggesting decreased association with F-actin (Fig. 6). This suggests that SWAP-70 dephosphorylation may promote F-actin binding. SWAP-70 dimerization was not important for F-actin binding as dimerization mutant R546A had a donor lifetime similar to the wt protein (Fig. 6).
SWAP-70 dimerizes in B16 cells stimulated with 12(S)-HETE
In the studies described above, vanadate treatment was used to induce SWAP-70 translocation to the membrane. This induced stable and efficient SWAP-70 translocation to the membrane allowing us to study SWAP-70 dimerization and F-actin binding by FRET techniques. However, vanadate affects several cellular pathways and its biological relevance is limited. We stimulated B16 cells with a more natural molecule. 12(S)-hydroxy-(5Z,8Z,10E,14Z)-eicosatetraenoic acid [12(S)-HETE] triggered an obvious effect on SWAP-70 in B16 cells. This arachidonic acid metabolite is known to modulate the metastatic activity of cancer cells (Honn et al., 1994), to increase their adhesion to epithelial cells (Chopra et al., 1991) and to alter cell motility (Nguyen et al., 2016). In metastatic B16 cells, 12(S)-HETE has been shown to affect intracellular localization of integrins responsible for cell adhesion and spreading (Timar et al., 1995). Upon stimulation of B16 cells with 12(S)-HETE, the number of cells with membrane structures such as ruffles and protrusions enriched for SWAP-70 was increased (Fig. S7A-C), and the cells seemed to be more spread (Fig. S7A). There were some prestimulated cells also found in the control sample, but their percentage was much lower (Fig. S7B). As FLIM-FRET experiments allow the measurements in the particular areas of the cell, the donor lifetime of SWAP-70 in the areas of 12(S)-HETE-induced ruffle-like structures was measured. These lifetime values were reduced compared to the lifetime of SWAP-70 in the cytoplasm of unstimulated cells, suggesting dimerization in these particular membrane regions (Fig. S7D,E). The effect of the more physiological stimulus 12(S)-HETE was, as expected, not as dramatic as for the general phosphatase inhibitor vanadate where all the protein was present at the membrane. Still, upon 12(S)-HETE stimulation SWAP-70 translocated to the membrane, mostly to ruffles and dimerized in these particular areas similar to the protein at the membrane of vanadate-treated cells (Fig. S7D,E). We hypothesize that translocation of SWAP-70 to such membrane structures and its dimerization is physiologically relevant as it is the prerequisite for F-actin binding and bundling.
DISCUSSION
A multitude of F-actin regulatory proteins, including SWAP-70, modulate the structure, localization and dynamics of actin filaments (Adams, 2004; George et al., 2013; Hilpelä et al., 2003; Lappalainen, 2016; Lee and Dominguez, 2010; Stevenson et al., 2012). SWAP-70 is known to play an important role in F-actin-dependent processes and the recombinant protein has been shown to bind and bundle F-actin in vitro (Chacón-Martínez et al., 2013). However, SWAP-70-actin interaction in living cells remained unknown.
In this study, we examined the oligomeric state of SWAP-70 within cells, narrowed down the dimerization region and identified an aa residue controlling SWAP-70's association with F-actin via phosphorylation. Our FLIM-FRET and seFRET results show that SWAP-70 remains monomeric in the cytoplasm in untreated cells. PH-domain-dependent binding to PIP3 in the cytoplasmic membrane promotes dimerization of SWAP-70 and its interaction with F-actin as proposed in our model shown in Fig. 7. This is consistent with studies indicating that negatively charged lipids such as PIP3 can induce protein clustering and dimerization at the membrane (Anderluh et al., 2017; Frost et al., 2009; Raja, 2010; Scacioc et al., 2017). Our results show that binding of SWAP-70 to phosphoinositides is also required for its interaction with F-actin. F-actin is enriched at the leading edges of the cells, where high concentrations of phosphoinositides were reported (Meili et al., 1999; Rickert et al., 2000; Servant et al., 2000). Therefore, we propose that PIP3-rich clusters at the leading edges promote association of SWAP-70 with F-actin and thus many of the biological functions of SWAP-70.
Model for SWAP-70 dimerization and interaction with F-actin. (A) SWAP-70 is monomeric in the cytoplasm and does not interact with F-actin. The dimerization region is hidden within the folded protein and is not available for interaction. The phosphorylated Y426 residue prevents the F-actin-binding domain from interaction with F-actin. (B) Upon stimulation of the cells, the PH domain of SWAP-70 binds PIP3 and thus SWAP-70 is recruited to the cytoplasmic membrane. SWAP-70 is dephosphorylated at the membrane at the position Y426 allowing SWAP-70 to dimerize and bind and bundle F-actin. (C) The SWAP-70 Y426F mutant directly binds F-actin but the protein is still monomeric. The dimerization region remains unavailable and no F-actin bundling can be promoted by SWAP-70.
Model for SWAP-70 dimerization and interaction with F-actin. (A) SWAP-70 is monomeric in the cytoplasm and does not interact with F-actin. The dimerization region is hidden within the folded protein and is not available for interaction. The phosphorylated Y426 residue prevents the F-actin-binding domain from interaction with F-actin. (B) Upon stimulation of the cells, the PH domain of SWAP-70 binds PIP3 and thus SWAP-70 is recruited to the cytoplasmic membrane. SWAP-70 is dephosphorylated at the membrane at the position Y426 allowing SWAP-70 to dimerize and bind and bundle F-actin. (C) The SWAP-70 Y426F mutant directly binds F-actin but the protein is still monomeric. The dimerization region remains unavailable and no F-actin bundling can be promoted by SWAP-70.
Binding of SWAP-70 to phosphoinositides at the membrane can occur in presence of different stimuli such as EGF and SCF (Chacón-Martínez et al., 2013; Shinohara et al., 2002). Vanadate used in this study to induce SWAP-70 membrane translocation may also affect the presence of PIP3 in cells through inactivation of PTEN phosphatase (Howes et al., 2003) or activation of PI3 kinases (Okkenhaug, 2013; Vanhaesebroeck et al., 2012). One may speculate that levels of PIP3 became elevated upon vanadate treatment and this, rather than phosphorylation of SWAP-70, attracts it to the membrane, and indeed using a mutational approach we could not identify a tyrosine residue responsible for SWAP-70's membrane association.
Treatment of cells with vanadate was very useful for our FRET experiments, as it causes stable presence of SWAP-70 at the membrane for at least 2 h, allowing FRET measurements at the membrane of the living cells. However, it was interesting to test if a specific stimulant for cancer cells would similarly affect SWAP-70 localization and dimerization in B16 cells. 12(S)-HETE, known to play role in cancer cell adhesion and metastasis, induced SWAP-70 translocation to F-actin-rich ruffle-like structures at the cell edge and protein dimerization. The 12(S)-HETE effect lasted at least 30-40 min, allowing FLIM-FRET experiments. This type of stimulant is also known to activate the PI3K pathway (Zhang et al., 2005), consistent with our hypothesis on SWAP-70 activation and dimerization via PIP3 binding at the cytoplasmic membrane.
Our FCCS studies reported here showed that purified, recombinant and functional SWAP-70 forms very stable oligomers that withstand very harsh conditions such as low pH and treatment with detergents. Is oligomerization an in vitro phenomenon only? We show here dimerization of SWAP-70 in living cells and the importance of particular aa residues for this association. For these studies, for the first time to our knowledge, we also simultaneously compared three different FRET approaches: FACS-FRET, microscopy-based seFRET and FLIM-FRET. FACS-seFRET is the fastest method allowing the analysis of hundreds of thousands of cells within minutes (Banning et al., 2010). The data obtained for SWAP-70 dimerization with FACS-seFRET was in agreement with those gained by microscopy seFRET measurements. Although microscopy seFRET reveals intracellular localization, a relatively high level of noise was present in seFRET images at the membrane. In addition, seFRET is known to be prone to certain artefacts due to donor and acceptor cross-talk and requires extensive image processing, implementing correction factors obtained from the acceptor-only and donor-only standards (Piston and Kremers, 2007; van Rheenen et al., 2004; Broussard et al., 2013). FLIM-FRET is considered to be the more rigid method, and is independent of fluorophore concentrations and donor-acceptor cross-talk. In FLIM-FRET, the fluorescence lifetime of the donor was determined with high precision all over the cell, requiring longer measurement times. Another limitation of FLIM-FRET is that compared to seFRET it cannot be used in simultaneous analysis of large numbers of cells as can be achieved by methods such as FACS.
Nevertheless, for all mutants, except mutant Δ500-507, FLIM-FRET, seFRET and FACS-seFRET data showed similar results, strongly confirming our results regarding SWAP-70 dimerization in cells and confirming the importance of the PH domain and the aa 545-564 region for SWAP-70 dimerization in vivo. In particular, we show that R546 is required for SWAP-70 dimerization in vivo and this observation is in line with our in vitro experiments with this mutant. The different results obtained for the Δ500-507 mutant using seFRET and FLIM-FRET can be explained by the fact that the fluorophores experienced some change in the emission or excitation spectra, quantum yield or orientation as the correction factor δ (Eqn 7), which is particularly sensitive to noise levels (van Rheenen et al., 2004), obtained in seFRET appeared to be different from all other measured samples.
The FLIM-FRET approach was also chosen to study SWAP-70's interactions with F-actin. This method has been implemented previously to investigate protein-actin interactions using LifeAct probes (Jayo et al., 2012). The results obtained for the AB mutant of SWAP-70 validate our FLIM-FRET experiments because, as expected, this mutant did not bind F-actin. SWAP-70 binds F-actin only in treated cells after translocation to the cytoplasmic membrane. Stimulation may induce a conformational change of SWAP-70 to expose the AB site. This hypothesis is supported by previous findings that the C-terminal part of SWAP-70 associates with F-actin independently of cell stimulation, whereas the full-length protein binds F-actin only upon cell stimulation (Ihara et al., 2006). Bundling of F-actin at the membrane would be the consequence of membrane association because it promotes SWAP-70 dimerization. Our in vitro data obtained with the dimerization mutant R546A confirms the requirement of SWAP-70 dimerization for F-actin bundling as this mutation significantly reduced the bundling activity of SWAP-70. However, currently it is not possible to measure F-actin bundling directly in living cells by fluorescence-based techniques. It is worth mentioning, however, that SWAP-70 dimerization is not required for F-actin binding. The Y546R dimerization mutant binds actin in treated cells. On the other hand, F-actin binding is also not required for SWAP-70 dimerization as the AB mutant and wt protein in cytochalasin D-treated cells dimerized. Both F-actin binding and dimerization require translocation of the protein to the membrane.
Using tyrosine mutants of SWAP-70 we revealed an important tyrosine at position 426 that regulates binding of SWAP-70 to F-actin and SWAP-70 dimerization at the membrane. As we show here, the Y426E phosphomimetic mutant fails to efficiently dimerize and bind F-actin at the membrane upon vanadate treatment, whereas the non-phosphorylatable Y426F mutant in untreated cells binds F-actin, but does not efficiently dimerize. In treated cells, Y426F dimerizes, binds and probably bundles F-actin. Binding to the membrane itself is not controlled by Y426 as both mutants associated with the membrane after vanadate treatment of the cells. These data suggests that pY426 keeps SWAP-70 in the cytoplasm in a closed monomeric state and prevents association with F-actin. Y426 becomes non-phosphorylated upon membrane binding of SWAP-70 to allow binding of F-actin, dimerization and, as a consequence, F-actin bundling (Fig. 7).
Together, these data reveal the functional relationship between cytoplasm-to-membrane translocation, dimerization, F-actin binding and bundling of SWAP-70 and reveal that membrane-proximal F-actin dynamics, known to govern many F-actin-dependent processes, are modulated by SWAP-70.
MATERIALS AND METHODS
DNA constructs
pDN-Cerulean and pDN-Venus plasmids for mammalian expression containing wt SWAP-70, SWAP-70 AB mutant (Δ565-585) and SWAP-70 PH mutant (K219A, K220A) were kindly provided by Dr Chacón-Martínez (TU Dresden, Germany). The pENTR-D-TOPO vector containing SWAP-70 Δ444-478 and SWAP-70 Δ478-499 (kindly provided by Dr Chacón-Martínez) were subcloned into pDN-Cerulean and pDN-Venus using the Gateway system (Invitrogen). pDN-Cerulean and pDN-Venus SWAP-70 point mutants Y426F, Y426E, R546A, K553A and the deletion mutants Δ508-525, Δ526-564, Δ545-564, Δ526-534, Δ535-544 were generated by a PCR site-directed mutagenesis using AccuPrime™ Pfx SuperMix (Thermo Fisher Scientific) and the required primers. RV-SWAP-70-GFP plasmid for mammalian expression containing wt SWAP-70 or its mutants Y517F, a PH mutant (Δ220-225), and the Δ1-197 and Δ353-585 mutants were kindly provided by Dr Ocaña-Morgner (TU Dresden, Germany). Point mutants Y302F, Y216F, Y241-242F, Y426F and Y426E in the RV-SWAP-70-GFP vector as well as point mutant Y426F in RV-SWAP-70-GFP AB mutant or RV-SWAP-70-GFP PH mutant plasmids were generated by PCR site-directed mutagenesis. For bacterial expression, SWAP-70 subcloned into the His-tag vector pDEST-17 was used. The point mutant R546A in the SWAP-70 pDEST-17 plasmid was generated by PCR site-directed mutagenesis. The mammalian expression plasmid for LifeAct-RFP staining of F-actin was obtained from Dr A. Roscher.
Protein purification and labelling
His-tagged SWAP-70 was expressed in E. coli (BL21) and purified on a nickel-nitrilotriacetic acid-agarose affinity column (Qiagen). Purification was followed by size-exclusion gel filtration chromatography on Superdex-200 16/60 column (Pharmacia Biotech) in 10 mM phosphate buffer, 150 mM NaCl, 0.1 mM EDTA.
Before labelling, SWAP-70 was incubated for 1 h with 10 molar excess of Tris(2-carboxyethyl) phosphine hydrochloride (TCEP) at 4°C to reduce thiol groups. The protein was labelled on its cysteine residues by the thiol-reactive dyes Alexa 488 or Atto 655, separately for single-labelled protein preparations or mixed for double-labelled protein preparations. SWAP-70 features seven cysteine residues. To ensure that only one of the cysteines was predominantly labelled, the fluorescent dyes were used at only 1.5 molar excess to the protein in the labelling reaction. Labelled protein was separated from free non-reacted dye on Sephadex G-25 gel filtration columns (GE Healthcare). The protein concentration was determined by measuring absorbance at 280 nm using the extinction coefficient of 1.18 Lg−1 cm−1. The degree of labelling was determined by measuring the absorption spectra of the protein and dye, and was found to be 0.7-0.9 dye molecules per protein molecule for SWAP-70 Alexa 488 (SWAP-70 green) and 0.9-1.3 dye molecules for SWAP-70 Atto 655 (SWAP-70 red). Labelled proteins were concentrated using Vivaspin concentrators (Sartorious) and the aliquots were shock-frozen in liquid nitrogen. Activity of the labelled protein was confirmed by low speed sedimentation F-actin bundling assays (see below).
Fluorescence cross-correlation spectroscopy (FCCS)
All FCCS experiments were performed using a LSM780 confocal microscope (Zeiss), which provided FCCS functionality through the ConfoCor3 module. Light was collected using a 40×NA 1.2 UV-VIS-IR C-Apochromat water immersion objective. All experiments were carried out at 22±0.5°C. The sample was excited by the 488 nm line of an Argon-ion laser and the 633 nm line of a He-Ne laser. Fluorescence was detected by Avalanche photo diodes (PerkinElmer) placed behind suitable filters (for Alexa 488: band path filter 505-610; for Atto 655: long path filter 650). Each curve represents the average of 12 measurements. The duration of each measurement was 10 s. Traces containing large fluorescent aggregates disturbing the measurements were removed. Diffusion time, number of particles, CC% and molecular weight were calculated using Eqns 1-6. Data were analysed using open software PyCorrFit (Muller et al., 2014). For the measurements, 50-100 nM SWAP-70 was diluted in PBS buffer (137 mM NaCl, 10 mM phosphate, 2.7 mM KCl, pH 7.4) if the measurements were carried out at pH 7. For the measurements at pH 3 and pH 5, samples were prepared in citrate-phosphate buffer (50 mM sodium-phosphate, 20 mM citric acid, 150 mM NaCl, pH 3 or pH 5). For the measurements at pH 10, a sodium carbonate-bicarbonate buffer was used (100 mM sodium carbonate, 100 mM sodium bicarbonate, 150 mM NaCl). For the detergent-treated samples, labelled SWAP-70 was incubated for 3.5 h with Triton X-100 (0.1%), Tween-20 (0.1%), SDS (2%), cholate (2%) or CHAPS (2%). Detergent was removed overnight by incubation with Bio-Beads SM-2 (Bio-Rad).
Cells, cell transfection, vanadate, wortmannin, cytochalasin D and 12(S)-HETE treatments
B16 mouse melanoma cells in which SWAP-70 was completely removed using the CRISPR/Cas9 system (kindly provided by Dr Pearce, TU Dresden, Germany) were used in microscopy imaging experiments, microscopy seFRET and FLIM-FRET. HEK 293T (293T) cells were used in FACS-FRET experiments. Cells were maintained in DMEM medium supplemented with 10% fetal calf serum and 1% penicillin/streptomycin. For imaging and microscopy-based FRET experiments, cells were seeded on glass-bottom chambers [Lab-Tek 8-well chambered #1.5 glass slides (NUNC)] coated with fibronectin (25 µgml−1). Cells were transfected in medium without antibiotics, using Lipofectamine 2000 reagent (Invitrogen) according to the manufacturer's protocol.
For treatment of the cells with vanadate, 1 ml of 20 mM sodium orthovanadate stock solution was freshly mixed with 330 µl of 30% H2O2 and incubated for 5 min to obtain 6 mM pervanadate. The solution was diluted 1:5 in H2O and added to the cells to obtain a final working solution of 25 µM pervanadate (Wienands et al., 1996). For the treatment of the cells with wortmannin, cells were incubated for 1 h with 1 µM wortmannin (Sigma-Aldrich). For cytochalasin D treatment, cells were incubated for 2 h with 5 µg/ml of the cytochalasin D from Zygosporium mansonii (Calbiochem). For 12(S)-HETE stimulation, a 1 µM solution was added to the cells. For all microscopy experiments, cells were imaged in imaging buffer [20 mM HEPES pH 7.4, 150 mM NaCl, 15 mM glucose, 20 mM trehalose, 5.4 mM KCl, 0.85 mM MgSO4, 0.6 mM CaCl2, 150 µg ml−1 bovine serum albumin (BSA)].
Fluorescence microscopy
Fluorescence images were acquired either on Leica TCS SP5 confocal laser-scanning microscope (Leica) equipped with a 63× NA 1.4 HCX PL APO water-immersion objective or an LSM 780 confocal microscope (Zeiss) using a 40× NA 1.2 UV-VIS-IR C-Apochromat water-immersion objective. Images were analysed in Fiji 2 software.
Microscopy-based seFRET
Here, A, B and C are the intensity signals of donor, FRET and acceptor, respectively. α, β, γ and δ are the calibration factors generated by acceptor-only and donor-only references (van Rheenen et al., 2004).
FACS-FRET
For FACS-FRET experiments, transfected cells were trypsinized and washed twice with FACS buffer (2 mM EDTA+0.1% BSA in PBS). Experiments were performed as described (van Rheenen et al., 2004) on a BD LSRII (BD Biosciences) machine. Donor (Cerulean) and FRET (Cerulean and Venus) samples were excited with a 405 nm laser; donor emission was collected with a 450/40 filter, whereas FRET signal was detected using a 530/30 filter positioned behind a 505 LP filter. Acceptor (Venus)-transfected cells were excited with a 488 nm laser and the emission was collected with the filters used for the FRET channel. Cerulean/Venus double-positive cells were selected for FRET analysis and were used to set the positive gate. The gate (magenta) was adjusted to untreated cells transfected with SWAP-70 Cerulean and with SWAP-70 Venus, which were FRET negative, similar to the control sample containing fluorescent tags only. The percentage of FRET-positive cells falling within this gate after stimulation of the cells was calculated. Data were analysed with FlowJo software 8 (TreeStar Inc.).
FLIM-FRET
seFRET experiments were carried out on living cells at room temperature using an LSM780 confocal microscope (Zeiss), equipped with a FLIM system (Becker & Hickl). The light was collected using a 40× NA 1.2 UV-VIS-IR C-Apochromat water-immersion objective. If Cerulean was used as a donor, the sample was excited with a 440 nm pulsed diode laser (50 MHz). If GFP was used as a donor, excitation with a 473 nm pulsed diode laser (50 MHz) was used. Photons emitted by the donor were collected and counted using SPCM-64 software (Becker & Hickl). Donor lifetimes for each pixel in the field of view (256×256) were calculated in SPC image analysis software (Becker & Hickl) to generate exponential decay curves. The lifetime image of a cell was generated by fitting the lifetime for every pixel in the image with a mono-exponential decay model if not specified otherwise. To calculate the mean lifetime of the donor in one cell, the lifetime of five points within the cell were analysed and averaged.
For two-component fitting of the FLIM data, the donor lifetime (t1) of the first component (a1) was fixed to the average donor lifetime in the cytoplasm in non-treated cells (2544 ns). The lifetime of the second component (t2) was fixed to the donor lifetime in vanadate-treated cells at the membrane where SWAP-70 molecules interacted (2134 ns). The fraction of non-interacting molecules (a1) and interacting molecules (a2) was estimated from the fit.
Low speed co-sedimentation F-actin bundling assay
Non-muscle actin (>99% pure) from human platelets (Cytoskeleton) at a concentration of 0.2 mgml−1 was depolymerized in depolymerization buffer (2 mM Tris/HCl, pH 8.0, 0.5 mM DTT, 0.2 mM ATP, 0.1 mM CaCl2 ) for 1 h on ice. Polymerization was induced by the addition of 5 mM MgCl2 and 20 μM EGTA for 2 min. F-actin (2 µM) was mixed with 2 µM SWAP-70 protein or its R546A mutant and incubated for 30 min at room temperature. Bundled actin was pelleted by centrifugation at 16,000 g for 30 min at 4°C. Supernatants were precipitated with ice-cold acetone. Pellets and supernatants were analysed on Coomassie-stained gels. Protein bands were quantified using Fiji 2 software.
High-speed co-sedimentation F-actin-binding assay
Actin and proteins were prepared in the same way as for the actin-bundling assay, but centrifugation was carried out at 100,000 g for 30 min to pellet all filamentous actin. Pellets and supernatants were analysed as described for the actin-bundling assay. The percentage of SWAP-70 in the pellet was determined and normalized to the percentage of pelleted protein in the absence of actin.
Acknowledgements
We thank Dr Carlos Andrés Chacón-Martínez and Dr Carlos Ocaña-Morgner (Technische Universität Dresden, Germany) for providing DNA constructs and Dr Glen Pearce (Technische Universität Dresden, Germany) for the SWAP-70-deficient B16 melanoma cell line. We also thank the CFCI and CMCB microscopy facilities (Technische Universität Dresden, Germany) for providing the necessary equipment and user support.
Footnotes
Author contributions
Conceptualization: R.J., V.B.; Methodology: R.J., V.B.; Validation: R.J., V.B.; Formal analysis: V.B.; Investigation: R.J., V.B.; Writing - original draft: R.J., V.B.; Writing - review & editing: R.J., V.B.; Supervision: R.J.; Project administration: R.J.; Funding acquisition: R.J.
Funding
Funding was provided by the Deutsche Forschungsgemeinschaft and institutional funds provided by the Faculty of Medicine.
References
Competing interests
The authors declare no competing or financial interests.