Protein kinase A is a functional component of focal adhesions

Abstract

Focal adhesions (FAs) form the junction between extracellular matrix (ECM)-bound integrins and the actin cytoskeleton and also transmit signals that regulate cell adhesion, cytoskeletal dynamics, and cell migration. While many of these signals are rooted in reversible tyrosine phosphorylation, phosphorylation of FA proteins on Ser/Thr residues is far more abundant yet its mechanisms and consequences are far less understood. The cAMP-dependent protein kinase (protein kinase A; PKA) has important roles in cell adhesion and cell migration and is both an effector and regulator of integrin-mediated adhesion to the ECM. Importantly, subcellular localization plays a critically important role in specifying PKA function. Here, we show that PKA is present in isolated FA-cytoskeleton complexes and active within FAs in live cells. Furthermore, using kinase-catalyzed biotinylation of isolated FA-cytoskeleton complexes, we identify 53 high-stringency candidate PKA substrates within FAs. From this list, we validate tensin-3 (Tns3)-a well-established molecular scaffold, regulator of cell migration, and a component of focal and fibrillar adhesions-as a novel direct substrate for PKA. These observations identify a new pathway for phospho-regulation of Tns3 and, importantly, establish a new and important niche for localized PKA signaling and thus provide a foundation for further investigation of the role of PKA in the regulation of FA dynamics and signaling.

Keywords: biosensor; focal adhesions; phosphorylation; protein kinase A (PKA); signal transduction.

Figure 1

Inhibition of PKA alters FA morphology, distribution, and dynamics in spreading cells.A, anti-vinculin immunofluorescence of REF52 cells plated onto fibronectin (FN)-coated coverslips for 1 h in the absence (Ctrl) or presence of an inhibitor of PKA activity (Rp-8-CPT-cAMPS (+Rp-cAMPS); 50 μM). Images are shown using an inverted, grayscale lookup table for ease of visualization. To facilitate analysis of FA distance from the cell center, images were ‘unwrapped’ using a clockwise polar transformation (Polar Xfm; bottom panels). B, histograms of relative FA distance from cell centroid (0 = cell center; 1 = cell periphery) in control- and PKA inhibitor-treated cells (600 and 601 FAs for Ctrl and +Rp-cAMPS samples, respectively). Distributions were analyzed using a Kolmogorov-Smirnov test (K-S distance = 0.5058). C, anti-paxillin immunofluorescence of REF52 cells 1 h after plating on FN in the absence (Ctrl) or presence of Rp-8-CPT-cAMPS. Enlargements of the dash-outlined areas in each panel are shown in the middle column (inset scale bar = 5 μm). D, FA morphometrics from control- and PKA inhibitor-treated cells. Violin plots summarize data from three experiments (2784 and 2155 total FAs for Ctrl and PKA-inhibited samples, respectively), with the width of the plot proportional to the number of points, top and bottom depicting the maximum and minimum values, and black lines depicting the median and colored lines depicting the 25th and 75th quartiles (∗∗p < 0.05; ∗∗∗∗p < 0.005). E, anti-talin immunofluorescence of REF52 cells plated onto FN for 4 h in the absence or presence of PKA inhibitor. The middle column shows enlargements of the dash-outlined areas in each panel (inset scale bar = 5 μm). F, temporal color-coded live-cell images of mCherry-paxillin-expressing REF52 cells spreading on FN for 1 h min in the absence or presence of Rp-8-CPT-cAMPS. Image acquisition began 30 min after plating. G–I, immunofluorescence images were analyzed at 90, 120, and 180 min to determine FA size (G and H) and aspect ratio (I). G, all measured adhesion values at each time point. H, all measured adhesion values greater than 15 μm² to highlight differences at larger adhesion area values. I, mean aspect ratio for all groups, analyzed by unpaired t-tests (n = >170 cells from two experiments; ∗p < 0.05, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001).

Figure 2

PKA type-II regulatory (RII) subunits are present in isolated focal adhesion/cytoskeleton (FACS) fractions.A, REF52 cells plated on FN-coated coverslips were fixed with DSP without (for whole cells, top) or with (for FACS preparation, bottom) subsequent fluid shear, then stained for talin (green), F-actin (red), and nuclei (blue). Scale bar = 30 μm. B, REF52 cells were FACS-prepped then stained to visualize vinculin (green), F-actin (red), and PKA RIIα subunits (blue). Scale bar = 5 μm. C and D, ten micrograms of whole-cell extract (wce) or isolated FACS proteins (FACS) were separated by SDS-PAGE and immunoblotted with antibodies against talin, vinculin (Vinc), actin, PKA RIIα (RIIα) or catalytic (PKAc) subunits, tubulin (Tub), GAPDH (GDH), and peroxiredoxin-3 (Prx3). Molecular weight markers (in kDa) are shown on the right.

Figure 3

Phosphorylated RII subunits PKA localize to FAs.A, FN-adherent REF52 cells were fixed, unroofed, left untreated or incubated with lambda phosphatase (+P’tase), then immunostained with antibodies against vinculin (Vinc) and against autophosphorylated PKA RII subunits (pSer99-RII). B and C, colocalization of pSer99-RII and vinculin with and without phosphatase treatment was analyzed using Pearson’s correlation coefficient (CC) and Li’s intensity correlation quotient (ICQ). Data represent 300 100 × 100 μm microscopic fields from two different experiments and were analyzed by unpaired t test (∗∗∗∗p < 0.001). D, FN-adherent REF52 cells were fixed and immunostained (without unroofing) with antibodies against vinculin (Vinc) and against autophosphorylated PKA RII subunits (pSer99-RII). E, Manders’ overlap coefficients were calculated to determine relative amount of vinculin signal that overlaps with pSer99-RII (M1) and of pSer99-RII that overlaps with vinculin (M2). Data represent 120 cells from three experiments. F, REF52 cells were fixed and stained as described for panel (D). Vinculin images were used to generate binary mask images that were multiplied by input vinculin and pSer99-RII images to eliminate non-FA signal. Ratiometric images were generated by dividing pixel-by-pixel pSer99-RII intensities by corresponding vinculin intensities and pseudocolored using the S-Pet lookup table in Fiji.

Figure 4

An FA-targeted PKA biosensor, PaxRAKS, reveals spatially and temporally dynamic PKA signaling events within individual FAs.A, schematic of PaxRAKS (Paxillin-fused Ratiometric A-Kinase Sensor). A PKA activity-responsive domain (ARD) coupled to a conformation-sensitive EGFP (csEGFP) is fused to the C-terminus of paxillin for localization to FAs. The monomeric red fluorescent protein mCherry is fused to the paxillin N-terminus as a denominator, allowing normalization of phosphorylation-dependent changes in EGFP fluorescence to the local abundance of the biosensor. B, images of a PaxRAKS-expressing cell, depicting simultaneous capture and spatial coincidence of the PKA-responsive csEGFP and reference mCherry signals. C, ratios of csEGFP:mCherry intensity over time in selected, individual FAs containing either wild-type PaxRAKS (WT; green) or a phospho-resistant Thr/Ala point mutant (TA; red). Black and blue arrows indicate the addition and wash-out, respectively, of Rp-8-CPT-cAMPS. Light and dark lines depict raw ratio values (acquired at three frames/min) and boxcar-averaged (−/+1 time point) values for smoothing, respectively. D, color-coded Ratiometric (csEGFP:mCherry) PaxRAKS signal in individual FAs, color-coded using a rainbow lookup table, such that low ratios, indicating low PKA activity, are depicted in cooler colors and high ratios (i.e., high activity) in warmer colors (left; see Movie S1). A set of ImageJ macros was used to isolate, vertically align, and spatially register individual adhesions (middle) to aid in visualization (right) and subsequent analysis (see below; Movie S2). Time-stamped panels show the localization and dynamics PKA activity within an individual FA at the indicated time points (min:s). E, linescan analysis of PaxRAKS dynamics over 90 s under the region of interest depicted by the dotted line in the FA depicted in panel (C). Signal is plotted along the y-axis as a function of time, along the x-axis. F, analysis of the location of peak PKA activity (i.e., maximum PaxRAKS ratio value). In 300 FA from 26 cells in three experiments. For each cell, 8 to 12 FA were imaged every 3 s for 1 min, kymographs were produced as described for panel (E), and the peak signal was scored as being located in the distal, central, or medial (relative to the center of the cell) third of the trace. Locations were tallied on a per-cell basis. Data are presented for each cell (scatter graph, left; ∗∗∗p < 0.005; ∗∗∗∗p < 0.001). And summarized for all cells (pie chart, right). G, minimum, mean, and maximum PaxRAKS ratios in growing, stable, and shrinking adhesions. Peripheral FAs with lengths that changed at rates above +0.02 μm/min for at least 5 min were considered Growing (n = 30), those with rates between +0.02 and −0.02 were considered Stable (n = 30), and those with rates below −0.02 μm/min considered Shrinking (n = 25). Data were analyzed using ordinary one-way ANOVA with multiple comparisons (∗p < 0.05; ∗∗p < 0.01; ns, not significant). H, the same data depicted in (G) was re-plotted as XY scatter plots of minimum, maximum, and mean PaxRAKS ratio values versus growth rate magnitude in individual FAs. The only significant ratio/rate correlation was between the maximum ratio and growth rate in growing FAs (∗r² = 0.174, p = 0.0217).

Figure 5

Candidate PKA substrates from kinase-catalyzed biotinylation of isolated focal adhesion/cytoskeleton fractions. Focal adhesion/cytoskeletal (FACS) proteins were isolated from REF52 cells stably adhered to FN-coated plates and incubated with ATP-biotin with or without exogenous PKA catalytic subunit. Biotinylated proteins were captured on immobilized avidin resin and processed for proteomic analysis by LC-MS/MS. A, analysis of PKA-catalyzed biotinylated FACS proteins by GOBP (Gene Ontology Biological Process). The 53 high stringency hit list of proteins (Table S1C) was analyzed for inclusion in the indicated GOBP classes. The size of the class circles is proportional to the number of hits in that class, while the shading represents the false discovery rate (FDR) as indicated in the scale. B, the 53 high stringency hits (Table S1C) were analyzed using the GeneMANIA plugin in Cytoscape to identify and visualize known physical interactions between and among PKA and potential FACS substrates. The PKA catalytic subunit (top, yellow) is shown with direct interactors (first row) and indirect interactors separated by one degree (first neighbors, second row), two degrees (second neighbors, third row), or three or more degrees of freedom (third and fourth+ neighbors, bottom two rows). Proteins with no prior physical interactions with PKA are shown (non-interactors, two rows in the top right). The two proteins that did not contain PKA substrate sites (see below) are indicated as diamonds, while Tns3, which was chosen for further validation, is shaded in dark red. C, scoring the presence of putative PKA motifs in candidate substrates. The amino acid sequences of the high stringency hits were analyzed using NETPhos 3.1 to identify predicted PKA-specific phosphorylation sites. For each phosphoacceptor residue, this analysis generates a score representing the probability (from 0 to 1) that the residue is a PKA phosphosite based on the homology the flanking amino acids to known PKA consensus sequences. The bars indicate that the given protein contains at least one site with a score greater than the indicated cutoff of 0.5 (the NETPhos minimum to include only positive predicted sites), 0.7, 0.75, or 0.8. The red asterisk indicates the high cutoff score for Tns3.

Figure 6

Tensin-3 (Tns3) is a direct physiological substrate for PKA.A, HEK293T cells were transfected with GFP or GFP-Tensin3 (GFP-TNS3), lysed at 48 h post-transfection, and GFP-Trap pulldowns were incubated in vitro with ATP in the absence or presence of PKA catalytic subunits for 30 min. Reaction products were separated by SDS-PAGE and immunoblotted with anti-GFP or anti-phospho-PKA substrate (P-PKA) antibodies. B, HEK293T cells were transfected with GFP-Tns3, serum-starved at 32 h post-transfection, and 16 h later treated with Fsk and IBMX (50 μM and 100 μM) or DMSO for 10 min. Cell lysates and GFP-Trap pulldowns were analyzed as described in (A). C, HEK293T cells were transfected with GFP-Tns3 and treated 48 h post-transfection with DMSO, Fsk + IBMX (25 μM and 50 μM) or H89 (20 μM) in complete media for 10 min. Lysates and GFP-trap pulldowns were immunoblotted as described in (A). The positions of molecular weight markers, in kDa, are shown. D, U2OS were serum-starved for 16 h then treated for 10 min with either DMSO (1 μl/ml) or Fsk/IBMX (25 μM/50 μM) in serum-free media. Lysates were immunoprecipitated with either rabbit IgG or anti-Tns3 antibody then immunoblotted as indicated. E, Tns3 immunoprecipitates from 8 μg of whole cell extract or purified FACS proteins, along with 2 μg of each lysate, were analyzed by immunoblotting with the indicated antibodies. Positions of molecular weight markers (in kDa) are indicated throughout.

Figure S1

PKA inhibition changes FA morphology in spreading REF52 cells.A, REF52 cells plated on fibronectin coated glass in the absence (control) and presence of the selective PKA inhibitor (Rp-8-CPT-cAMPS (+Rp-cAMPS); 50 μM) were fixed 30, 60, 90, 120 and 180 min after plating and stained for paxillin. B, insets from cells highlighting morphological differences between groups upon treatment with Rp-cAMPs.

Figure S2

Addition of forskolin + IBMX inhibits cell spreading and FA formation. REF52 cells were plated on fibronectin-coated coverslips for 45 min in the presence of solvent (DMSO; 0.1% vol:vol) or a combination of forskolin and IBMX (Fsk/IBMX; 25 & 50 μM, respectively) to activate adenylyl cyclase and inhibit phosphodiesterase and fixed and stained for vinculin.

Figure S3

PKA-catalyzed biotinylation of isolated FACS proteins. Focal adhesion/cytoskeleton (FACS) fractions were prepared from FN-adherent REF52 cells and processed as described in Experimental procedures. Kinase-dependent biotinylation was performed on FACS protein (5 μg) with either ATP (2 mM, lane 1) or ATP-biotin (2 mM, lanes 2–4) in the absence (lane 2) or presence (lanes 1, 3, and 4) of exogenously added PKA (50 or 500 ng, as indicated) for 2 h with 250 rpm mixing at 31 °C. The biotinylated proteins were separated by SDS-PAGE, transferred to PVDF membrane, and visualized by streptavidin-HRP. PKA is labeled with an arrow, potential substrates are indicated with brackets, and molecular weight markers (in kDa) are shown at left.

Figure S4

GFP-Tns3 localizes to focal and fibrillar adhesions. REF52 cells transfected with a plasmid encoding GFP-tagged Tns3 were plated onto FN-coated glass-bottom dishes and imaged by live-cell microscopy (A) or FN-coated coverslips and fixed, stained with antibodies against vinculin, and imaged by two-color fluorescence microscopy (B). Individual GFP-Tns3 and vinculin images are shown in inverted grayscale (top left & right, respectively) or pseudo-colored and overlaid (bottom).