PPM1F controls integrin activity via a conserved phospho-switch

Abstract

Control of integrin activity is vital during development and tissue homeostasis, while derailment of integrin function contributes to pathophysiological processes. Phosphorylation of a conserved threonine motif (T788/T789) in the integrin β cytoplasmic domain increases integrin activity. Here, we report that T788/T789 functions as a phospho-switch, which determines the association with either talin and kindlin-2, the major integrin activators, or filaminA, an integrin activity suppressor. A genetic screen identifies the phosphatase PPM1F as the critical enzyme, which selectively and directly dephosphorylates the T788/T789 motif. PPM1F-deficient cell lines show constitutive integrin phosphorylation, exaggerated talin binding, increased integrin activity, and enhanced cell adhesion. These gain-of-function phenotypes are reverted by reexpression of active PPM1F, but not a phosphatase-dead mutant. Disruption of the ppm1f gene in mice results in early embryonic death at day E10.5. Together, PPM1F controls the T788/T789 phospho-switch in the integrin β1 cytoplasmic tail and constitutes a novel target to modulate integrin activity.

Figure 1.

The integrin β1 T788/T789 motif constitutes a conserved phospho-switch to regulate integrin activity. (A) Alignment of cytoplasmic amino acid residues of human integrin β subunits. The conserved threonine motif (red), the proximal NPxY motif (blue), the distal NPxY motif (green), and the binding sites of talin, kindlin-2, and filaminA are marked. (B) Strep-tag-integrin β1 (Strep-ITGB1) cytoplasmic domains in the WT form, with modifications of the T788/T789 motif (T/D, TT/DD, TT/AA), alanine mutation of Tyr-795 (Y795A), or of Tyr-783 (Y783A) were incubated with His-tagged enolase, FLN19-21, or talin-F3. Upon streptactin pulldown, bound His-tagged proteins were detected by WB with α-His antibody or Coomassie staining. 50% of His-tagged proteins were directly loaded (input) for comparison. (C) His-tagged proteins as in B were immobilized in triplicate wells and incubated with the indicated Strep-ITGB1 variants at 4°C. After washing, binding was detected by incubation with streptactin-HRP. Bars represent mean ± SEM of triplicates from a representative experiment. (D) Biotinylated integrin β1 peptides in the nonphosphorylated form (β1-762-798) or phosphorylated at T788/T789 (β1-762-798 pTpT) were bound to streptavidin-agarose (0.5 mg/ml beads) before being incubated with His-tagged enolase, His-talin-F3, GST, or GST-tagged FLN19-21. Beads without peptide loading were used as negative control. Upon pulldown, bound His- and GST-tagged proteins were detected by WB with α-His or α-GST antibody and Coomassie staining. 50% protein amount was directly loaded as input. (E and F) Structural models of filaminA-Ig21 domain (E) and talin-F3 domain (F) with phosphorylated integrin β1A based on the known structures of filaminA Ig21/integrin β7 complex (PDB2BRQ) or talin/integrin β1D complex (PDB3G9W). The phosphorylated β1A T788/T789 motif was generated in Coot. The electrostatic surfaces of filaminA-Ig21 and talin-F3 are depicted in red (negative charge, value −5) and blue (positive charge, value 5), β1 is shown in yellow; the phosphorylated threonine residues are shown in ball and stick representation. See also Fig. S1. Coot, Crystallographic object-oriented toolkit; H. sapiens, Homo sapiens; w/o, without.

Figure S1.

Related to Fig. 1, Fig. 2, Fig. 7, Fig. 8, and Fig. 9: The T788/T789 motif in the integrin β1 cytoplasmic tail is evolutionary conserved, and its phosphorylation regulates association with filaminA and talin. (A) Alignment of cytoplasmic amino acid residues of integrin β subunits derived from different species. The conserved threonine residues (red), the proximal NPxY motif (blue), and the distal NPxY motif (green) are marked. (B) Alignment of amino acid residues of strep-tagged integrin β1 cytoplasmic tails with indicated binding sites of talin, filaminA, and kindlin-2. The threonine residues (red) and the proximal (blue) and distal (green) NPxY motifs are marked. Point mutations are marked by black boxes. (C) The indicated strep-tag-integrin β1 (Strep-ITGB1) cytoplasmic domains and His-tagged enolase, talinF3 domain, FLN19-21, or kindlin-2 were expressed as soluble proteins in BL21 DE3 bacteria and equal amounts detected by WB and Coomassie staining. (D) Schematic model of the OPTIC principle. OPTIC fusion constructs consist of integrin β1 cytoplasmic domains (ITGB-C) and the transmembrane and extracellular domain of CEACAM3 (CEA3). Receptor clustering is triggered by binding of Opa₅₂ protein expressing Neisseria gonorrhoeae (N. gonorrhoeae) to CEACAM3, thereby potentially recruiting an intracellular protein of interest (POI). (E) 293T cells were transiently transfected with the indicated expression constructs and WCL subjected to WB; CEACAM1 protein served as positive control for the CEACAM antibody, and nontransfected 293T cell lysate served as negative control. (F) GST-FLN19-21 and Strep-ITGB1 WT were incubated with increasing amounts of His-talinF3. Upper panels show the input proteins. Upon streptactin pulldown, proteins bound to ITGB1 WT were visualized by Coomassie staining (bottom part; top panel), WB with α-GST antibody to detect FLNIg19-21 (second panel), or with α-His to detect talinF3 (third panel). Coomassie staining also verified similar amounts of precipitated Strep-ITGB1 WT (lowest panel). (G) Strep-ITGB1 WT or the T788D/T789D variant was immobilized in triplicate wells and incubated with His-tagged talinF3 and increasing amounts of GST-FLN19-21 at 4°C. After washing, talinF3 binding was detected by incubation with α-6xHis antibody and secondary HRP-coupled antibody. Bars represent mean ± SEM of triplicates from a representative experiment. Coomassie staining verified similar amounts of input FLN19-21 (upper panel), talinF3 (middle panel), and Strep-ITGB1 WT and TT/DD (lowest panel). (H) GST-FLN19-21 and Strep-ITGB1 WT were incubated with increasing amounts of His-kindlin-2. Left panels show the input proteins. Upon streptactin pulldown, proteins bound to ITGB1 WT were visualized by Coomassie staining (right; top panel), WB with α-His to detect kindlin-2 (second panel), or α-GST antibody to detect FLNIg19-21 (third panel). Coomassie staining also verified similar amounts of precipitated Strep-ITGB1 WT (lowest panel). C. owczarzaki, Capsaspora owczarzaki; D. melanogaster, Drosophila melanogaster; D. rerio, Danio rerio; G. gallus, Gallus gallus; H. sapiens, Homo sapiens; M. musculus, Mus musculus; X. laevis, Xenopus laevis; mCh, mCherry.

Figure 2.

The integrin β1 T788/T789 phospho-switch regulates talin versus filaminA binding in intact cells. (A) 293T cells were cotransfected with indicated CEA3-integrin β1 cytoplasmic tail fusion proteins together with GFP or GFP-talin-1. After 48 h, cells were infected with CEACAM-binding bacteria (Ngo) for 1 h, fixed, and stained for CEACAM3. Micrographs (left panel) show representative infection sites of bacterial attachment (blue) and CEA3-integrin β1 clustering (red). Recruitment of GFP-talin (green) to clustered integrin cytoplasmic tails is indicated (white arrowhead). Scale bar, 1 µm. Talin recruitment was quantified with the enrichment ratio (ER) indicating the -fold enrichment of GFP-intensity at bacterial attachment sites versus the overall cellular GFP level (middle panel). Dots represent individual ERs of 30–60 recruitment sites from n ≥ 2 independent experiments. Horizontal lines indicate mean values and 95% confidence intervals (whiskers). The red line indicates the threshold of positive recruitment. The bar graph (right panel) depicts the percentage of cells showing an ER of talin-1 recruitment ≥ 2. Statistically significant differences were evaluated using one-way ANOVA, followed by Bonferroni post hoc test (***, P < 0.001; **, P < 0.01; *, P < 0.05). (B) Cells transfected with the indicated CEA3-integrin β1 cytoplasmic tail fusion proteins together with GFP-FLN19-21 were evaluated as in A. (C) His-talinF3 and Strep-ITGB1 WT or Strep-ITGB1 TT/DD were incubated with increasing amounts of GST-FLN19-21. Left panels show the input proteins. Upon streptactin pulldown, proteins bound to ITGB1 WT or ITGB1 TT/DD were visualized by Coomassie staining (right; top panel), WB with α-GST antibody to detect FLN19-21 (second panel), or with α-His to detect talinF3 (third panel). Coomassie staining also verified similar amounts of precipitated Strep-ITGB1 WT and TT/DD (lowest panel). (D) Schematic view of talin versus filaminA association with the integrin β1 cytoplasmic tail depending on T788/T789 (pseudo-)phosphorylation. See also Fig. S1. D, aspartate; T, threonine; P, phospho-.

Figure S2.

Related to Fig. 3: knock-down of PPM1F in 293T cells or NHDFs does not alter expression of core focal adhesion proteins and does not affect integrin surface levels. (A) Overview of protein phosphatases present in the human integrin adhesome (Zaidel-Bar et al., 2007). These enzymes have been targeted individually by specific shRNA-encoding lentiviral particles. Control cells were treated with lentiviral particles lacking shRNA. (B) Immunoblotting of WCL from PPM1F knock-down (shPPM1F) and control 293T cells probed with antibodies against the indicated focal adhesion proteins. Monoclonal α-tubulin antibody was used as loading control. (C) Immunoblotting of WCLs from control and PPM1F knock-down (shPPM1F) NHDF probed with antibodies against core focal adhesion proteins. Probing with monoclonal α-tubulin antibody confirmed equal loading of WCLs. (D) Integrin surface expression levels of control and shPPM1F NHDF were analyzed by flow cytometry. Cells were stained with the indicated monoclonal integrin-specific antibodies. As a comparison, cells were stained the fluorescent labeled second antibody only (second Ab), or remained unstained; count ≥10,000 cells. (E) FilaminA was depleted in NHDF with a shRNA encoding lentivirus. WCLs were analyzed by WB with α-filaminA (upper panel) and α-tubulin (lower panel) antibodies. (F) FilaminA-depleted NHDFs were seeded for 1.5 h onto 1 µg/ml FN_III9-11-coated coverslips. After fixation, cells were stained with indicated antibodies and analyzed by confocal microscopy; scale bar, 20 µm. Insets: Higher magnification of boxed areas; arrowheads point to enrichment of active integrin β1 and filaminA; scale bar, 10 µm. (G) Integrin β1 was depleted in NHDF with a shRNA encoding lentivirus. WCLs were analyzed by WB with α-integrin β1 (upper panel) and α-tubulin (lower panel) antibodies. Control and integrin β1-depleted NHDF were seeded for 1.5 h onto 1 µg/ml FN_III9-11-coated coverslips. After fixation, cells were stained with a monoclonal antibody against active integrin β1 (9EG7) and phalloidin. Additional samples were stained with the second antibody only; scale bar, 20 µm.

Figure 3.

The phosphatase PPM1F regulates integrin activity and integrin-dependent cell adhesion. (A) 293T cells were transduced with lentiviral particles encoding shRNA directed against the indicated protein phosphatases. Stable knock-down cells were plated for 40 min on 1 µg/ml collagen I, 0.8 µg/ml FN_III9-11, or 10 µg/ml poly-L-lysine and washed, and adherent cells were fixed and stained with crystal violet. In parallel, cells were plated for 3 h, fixed, and stained without washing (total seeded cells). Crystal violet staining was quantified and normalized to total seeded cells; control = lentivirus without shRNA; bars represent mean ± SD of three wells. Shown is a representative result repeated twice with comparable results. (B) Representative micrographs from A showing shPPM1F-transduced and control cells on collagen I and poly-L-lysine; scale bar, 150 µm. (C) NHDFs were transduced with lentiviral particles encoding shRNA targeting PPM1F or without shRNA (control). Knock-down was confirmed by WB of WCL with α-hPPM1F and α-tubulin antibody as loading control. (D) Control and PPM1F knock-down NHDFs were subjected to adhesion assays (1 µg/ml collagen I, 0.8 µg/ml FN_III9-12). Values were normalized to total seeded cells. Bars represent mean ± SEM of three independent experiments, each done in triplicate; unpaired t test, ***, P < 0.001. (E) NHDFs as in D were seeded onto fibronectin-coated dishes for 40 min, fixed, and stained for active integrin β1 (9EG7) or total integrin β1 (AIIB2). Graphs show the ratio of active integrin β1 to total integrin β1. Bars represent mean ± SEM of triplicates from one representative experiment. (F) NHDFs as in D were seeded for 2 h onto 1 µg/ml FN_III9-11-coated coverslips. Fixed cells were stained as indicated. Arrowheads point to active integrin β1/talin enrichment; scale bar, 20 µm. Insets: Higher magnification of boxed area; scale bar, 10 µm. See also Fig. S2.

Figure 4.

PPM1F KO results in constitutive integrin activity and an exaggerated cell adhesion phenotype. (A) A172 cells expressing Cerulean (WT) were treated with sgRNA against Cerulean (control) or sgRNA against Cerulean and PPM1F (PPM1F KO) combined with Cas9. Clonal cell lines were derived and analyzed by WB with polyclonal α-PPM1F (upper panel) or α-tubulin (lower panel) antibodies. (B) Cells from A were seeded onto 10 µg/ml FN_III9-11 for 30 min in triplicate. Cell adhesion was measured as in A. Adhesion of A172 WT cells was used as reference. Bars represent mean ± SEM from three independent experiments. Statistics was calculated using one-way ANOVA, followed by Bonferroni post hoc test (*, P < 0.05). (C) Cells from A were replated onto 0.1 µg/ml FN_III9-11-coated 96-well plates for 40 min. Samples were stained for active integrin β1 (9EG7), total integrin β1 (AIIB2), or with secondary antibody only as a control. The graph shows the ratio of active integrin β1 versus total integrin β1 after background subtraction. Bars represent mean ± SEM from a representative experiment done in triplicate. (D) Cells transfected with GFP-talin were seeded onto FN_III9-11 for 30 min. Cells were fixed, stained with antibodies against active integrin β1, and analyzed by confocal microscopy; scale bar, 10 µm. Arrowheads point to active integrin β1/talin enrichment. Insets: Higher magnification of boxed area; scale bar, 5 µm. (E) Quantification of peripheral belt formation by active integrin (upper graph) or talin-1 (lower graph) in A172 WT and PPM1F KO cells. Data are shown in percentages of all cells analyzed; n ≥ 30 derived from ≥ 2 independent experiments. (F) A172 WT, control, and PPM1F KO cells were seeded onto 2 µg/ml FN_III9-11-coated coverslips for 30 min or 1.5 h before fixation and staining with DAPI and phalloidin-Cy5. Representative pictures are shown; scale bar, 25 µm. (G) Quantification of cell area from cells in F; n ≥ 100 cells from two or more independent experiments; mean values and 95% confidence intervals are shown, outliers are represented in dots, and statistics was performed using one-way ANOVA with post hoc Bonferroni test (***, P < 0.001). (H) A172 WT and PPM1F KO cells were transduced with lentiviral particles harboring shRNA against human filaminA or empty pLKO.1 vector as a control. After puromycin selection, lysates were prepared and subjected to WB analysis using indicated antibodies. (I) Cell adhesion assays were performed with starved A172 cells from H for 20 or 60 min using 0.4 or 10 µg/ml FN_III9-11 as integrin-dependent matrix. After a washing step, adherent cells were fixed and stained with crystal violet. Staining was quantified and normalized to the total number of seeded cells. Bar graphs show mean ± SEM of four independent experiments done in triplicate referenced to WT cell adhesion (= 1); one-way ANOVA and Bonferroni post hoc test (***, P < 0.001; **, P < 0.01; *, P < 0.05). (J) Cell lines from H were kept in suspension for 45 min and incubated for 15 min with 10 µg/ml FN_III9-11 before staining for total (AIIB2) or active integrin β1 (9EG7). Samples were analyzed by flow cytometry, 10,000 counts; unstained cells and an IgG-matched irrelevant antibody served as negative controls. The mean fluorescence intensity (MFI) ratio of active to total integrin β1 was calculated and normalized to the WT sample (= 1). Bars represent mean MFI ± SEM of three independent experiments; one-way ANOVA and Bonferroni post hoc test (**, P < 0.01; *, P < 0.05). See also Fig. S3 and Fig. S4. FN, FN_III9-11.

Figure S3.

Related to Fig. 4: PPM1F KO in A172 cells does not increase integrin surface levels or alter expression of core focal adhesion proteins, but strongly affects integrin-dependent processes. (A) A172 cells expressing Cerulean (WT) were treated with sgRNA against Cerulean combined with Cas9 or with sgRNAs against Cerulean and PPM1F combined with Cas9. Clonal Cerulean-negative (Control) and clonal Cerulean/PPM1F-negative cell lines (hPPM1F KO) were derived. The hPPM1F KO cells were stably transduced with mKate2 encoding lentivirus (hPPM1F KO plus mKate2), or lentivirus encoding hPPM1F WT (PPM1F KO plus hPPM1F-mKate2), or lentivirus encoding PPM1F D360A (PPM1F KO plus hPPM1F DA-mKate2). WCL from the different cell lines were analyzed by WB with antibodies against indicated core focal adhesion proteins. Monoclonal α-tubulin antibody was used as loading control. (B) A172 cell lines as in A were analyzed by flow cytometry for surface expression levels of indicated integrins by staining with integrin-specific antibodies. IgG control: WT cells receiving an isotype-matched control antibody. In the case of integrin β3, NIH3T3 cells served as positive controls. Count ≥10,000 cells. WT, hPPM1F-mKate2; DA, hPPM1F D360A-mKate2. (C) A172 WT, control, or hPPM1F KO cells were seeded onto 0.4 or 2 µg/ml FN_III9-12 for 30 min in triplicate. Cells were washed, fixed, and stained with crystal violet. Crystal violet staining was quantified and normalized to A172 WT cells. Bars represent mean ± SEM from three independent experiments. Significance was calculated using one-way ANOVA, followed by Bonferroni post hoc test (*, P < 0.05). For cell adhesion on 10 µg/ml FN, see Fig. 2 H. (D) A172 WT, control, or PPM1F KO cells were seeded onto 10 µg/ml FN for 15 and 60 min and processed as in C. Representative pictures of crystal violet–stained wells; scale bar, 150 µm. (E) A172 WT or hPPM1F KO cells were transiently transfected with GFP-talin-1 before seeding onto 2 µg/ml FN_III9-11 for 1.5 h. Fixed cells were stained with antibodies against active integrin β1 (9EG7) and analyzed by confocal microscopy. Scale bars, 10 µm. Arrowheads point to clusters of active integrin β1/talin. Insets: Higher magnification of boxed area; scale bar, 5 µm. (F) Quantification of cells from E showing a peripheral active integrin belt (upper graph) or peripheral talin-1 clustering (lower graph). Data are shown in percentages of all cells analyzed by confocal microscopy; n ≥ 30 derived from two or more independent experiments. (G) A172 WT, control, or PPM1F KO cells were seeded onto 0.4 µg/ml FN_III9-12-coated coverslips (left side) or onto 10 µg/ml FN_III9-11-coated coverslips (right side) for 30 min (upper panels) or 1.5 h (lower panels) before fixation and staining with DAPI and phalloidin. Cells were analyzed by confocal microscopy. Representative pictures are shown; scale bar, 25 µm. Quantification of cell areas was done in ImageJ using a custom plugin. Cells not recognized by the plugin were manually analyzed with Leica LAS AF Lite software; n ≥ 100 cells from two or more independent experiments; box plots depict means with 95% CIs (whiskers) and outliers (dots). Significance was determined using one-way ANOVA with post hoc Bonferroni test (**, P < 0.01; ***, P < 0.001). (H) A172 hPPM1F KO and derived reconstituted cell lines were lysed and subjected to WB with α-PPM1F (upper panel) and α-tubulin (lower panel) antibodies to test for PPM1F expression levels. The two bands correspond to the hPPM1F-mKate2 fusion protein and the hPPM1F protein, which results from proteolytic separation of the mKate tag. FN, FN_III9-11.

Figure S4.

Related to Fig. 4: filaminA knock-down in A172 cells pheno-copies integrin-dependent effects of PPM1F depleted cells. (A) Cell adhesion assays were performed with indicated A172 cell lines using 10 µg/ml FN_III9-11. Shown are representative pictures of crystal violet stained wells after 20 min adhesion; scale bar, 150 µm. (B) Indicated cell lines were seeded onto 5 µg/ml FN_III9-11 for 2 h, fixed, and stained against active integrin β1 before analysis by confocal microscopy; scale bar, 10 µm. Insets show higher magnification of boxed areas. Arrowheads point to active integrin β1, which accumulates at the cell periphery in PPM1F KO and filaminA knock-down cells; scale bar, 5 µm. (C and D) Indicated cell lines were seeded onto 10 µg/ml FN_III9-11 for 30 min or 2 h, fixed, and stained with DAPI and phalloidin-Cy5. Samples were imaged using confocal microscopy. (C) Representative images from cells at indicated time points; scale bar, 25 µm. Cell spreading was quantified in D. Bars show mean with 95% CIs from two independent experiments; n ≥ 80 cells. Statistics was performed using one-way ANOVA, followed by Bonferroni post hoc test (***, P < 0.001). (E) Model summarizing effects of filaminA knock-down on integrin activity. In filaminA knock-down cells, the balance between active and inactive integrins is shifted toward the active conformation. Talin-integrin association is increased due to the lack of the counterregulator filaminA, thereby promoting cell adhesion, reducing cell spreading and pheno-copying effects of PPM1F KO cells. FilaminA KD, filaminA knock-down; P, phospho-; shFilA, short-hairpin RNA targeting filaminA; T, threonine.

Figure 5.

The phosphatase PPM1F regulates the phosphorylation state of the integrin T788/T789 motif in intact cells. (A) Starved A172 WT and PPM1F KO cells were kept in DMEM plus 0.25% BSA for the indicated time periods before WCLs were prepared and subjected to WB analysis with indicated antibodies. (B and C) Starved A172 WT (B) or PPM1F KO cells (C) were seeded onto 2 µg/ml FN_III9-11 for the indicated time periods before being lysed and analyzed by immunoblotting as in A; PPM1F KO cells were used as a control in B to directly compare integrin β1 phosphorylation levels. (D) PPM1F KO cells were reconstituted with either WT human PPM1F (PPM1F KO + hPPM1F-mKate2) or the phosphatase-dead mutant PPM1F D360A (PPM1F KO + hPPM1FD360A-mKate2). Furthermore, PPM1F KO cells were stably transduced with mKate2 (PPM1F KO + mKate2). As a comparison, A172 WT cells and A172 cells transduced with a vector lacking the PPM1F sgRNA (control) were used. The cells were seeded onto 2 µg/ml FN_III9-11 for 45 min and WCLs subjected to WB with indicated antibodies (left panels). Bar graphs show densitometric quantification of band intensities from pT788/pT789-β1 versus total β1 integrin antibody signal for the indicated samples from three independent experiments; WT was set to 1 (right graph). Statistics was performed using one-way ANOVA, followed by Bonferroni post hoc test (*, P < 0.05; ***, P < 0.001). (E) Indicated A172 cell lines were kept in suspension for 45 min and incubated for 15 min with 10 µg/ml FN_III9-11 before being stained for total (AIIB2) or active integrin β1 (9EG7). Samples were analyzed by flow cytometry, 10,000 counts. The mean fluorescence intensity ratio of active to total integrin β1 was calculated and normalized to the WT sample (= 1). Bars represent mean fluorescence intensity ± SEM of four independent experiments; statistics was performed using one-way ANOVA and Bonferroni post hoc test (***, P < 0.001). (F and G) Cell adhesion assays were performed with starved A172 cell lines for 20 or 60 min using 0.4 or 10 µg/ml FN_III9-11; 2% BSA-coated wells were used as a negative control. After a washing step, adherent cells were fixed and stained with crystal violet. Staining was quantified and referenced to WT cell adhesion (= 1). (F) Representative pictures after 60 min adhesion on 10 µg/ml FN_III9-11; scale bar, 150 µm. (G) Bar graphs show mean ± SEM of four independent experiments pipetted in triplicate; statistics was performed using one-way ANOVA and Bonferroni post hoc test (***, P < 0.001; **, P < 0.01). (H) Model summarizing effects of PPM1F KO on integrin activity. In PPM1F KO cells, the balance between active and inactive integrins is shifted toward the active conformation by constitutive phosphorylation of integrin β1 at T788/T789, thereby prohibiting filaminA binding, while promoting increased talin association and cell adhesion. DA, phosphatase-dead mutant of PPM1F D360A; FN, FN_III9-112; P, phospho-; T, threonine.

Figure 6.

Recombinant PPM1F dephosphorylates the conserved T788/T789 motif in the integrin β1 cytoplasmic domain. (A) Recombinant GST-PPM1F and GST-PPM1F D360A (DA) were expressed in E. coli, purified, and analyzed via SDS-PAGE and Coomassie staining. A BSA standard was used to evaluate the protein content. (B) The activity of purified enzymes from A was determined using the fluorogenic substrate 4-MUP. Curves were obtained by a direct nonlinear fit of the data to the Michaelis–Menten equation, and V_max and K_m values for PPM1F were determined from one representative experiment. (C) Sequences of double or single phosphorylated integrin β1 peptides and the single phosphorylated control peptide MLC2. (D) 500 ng GST-PPM1F or GST-PPM1F DA were incubated at 30°C with synthetic integrin β1 peptides from C. Release of phosphate was measured after 60 min by malachite green. Bars depict mean ± SD of three independent experiments performed in duplicate; unpaired t test (***, P < 0.001). (E) CaMKIIβ (120 ng) was incubated with GST-integrin β1 cytoplasmic domain in the presence or absence of ATP/calmodulin/CaCl₂. 2 µg GST-PPM1F or GST-PPM1F DA were added as indicated, and samples were incubated for 60 min at 30°C. Reactions were stopped via addition of SDS sample buffer and subjected to WB with indicated antibodies; *GST is free GST cleaved off from PPM1F proteins. (F) 200 ng of the indicated recombinant phosphatases were separated by SDS-PAGE and probed with α-GST (hPTP1B; PPM1F) and α-His (hPP5; hILKAP) antibodies (left panels); 200 ng of the purified phosphatases were incubated at 30°C with β1-22pT788pT789 peptide. Phosphate release was measured after 60 min with malachite green. Bars depict mean ± SEM of triplicates from a representative experiment. Blank (= no enzyme) values were subtracted from each sample. (G) Phosphatase assay as in F using purified GST-integrin β1 cytoplasmic domain phosphorylated in vitro by CaMKIIβ as a substrate. (H) Assays conducted as in G were analyzed by WB using indicated antibodies; *PPM1F is free PPM1F released from GST. See also Fig. S5. CaMKII recomb., recombinant human Ca²⁺/calmodulin-dependent protein kinase II; w/o, without.

Figure S5.

Related to Fig. 6: PPM1F purified from 293T cells dephosphorylates the conserved T788/T789 motif in the integrin β1 cytoplasmic domain. (A) GST-tagged PPM1F or PPM1FD360A was expressed in 293 cells and affinity-purified from lysates with glutathione-coupled beads. Equal amounts of beads were subjected to WB analysis using α-GST antibody. (B) The integrin β1 derived synthetic peptide β1-pT788/pT789 was incubated with increasing amounts of cell-purified GST-PPM1F, with 200 ng GST-PPM1F D360A (PPM1F DA), or 100 ng calf intestine alkaline phosphatase (CIAP) for 1 h at 30°C. Released phosphate was detected by malachite green solution. Background values (buffer plus malachite green) were subtracted; w/o, peptide without phosphatase. Shown are mean ± SD of three independent experiments; unpaired t test, *, P < 0.05; **, P < 0.01; ***, P < 0.001. (C) β1-pT788/pT789 was incubated with 100 ng of cell-purified GST-PPM1F or GST-PPM1F DA for indicated time periods at 30°C. Released phosphate was detected as in B. (D) 50 ng PPM1F or PPM1F DA were incubated with the indicated phospho-peptides for 60 min and released phosphate measured as in B. (E) Dephosphorylation of increasing β1-pT788/pT789 peptide concentrations by 150 ng cell-purified PPM1F was analyzed after 60 min incubation at 30°C as in B. The indicated curve was obtained by direct fit of the data to the Michaelis–Menten equation, and V_max and K_m values were determined. (F) 120 ng His-tagged CaMKIIβ was incubated with recombinant GST-tagged WT β1 integrin cytoplasmic domain, the nonphosphorylatable integrin β1 TT/AA mutant, or without any substrate. As indicated, samples received calmodulin (CaM), Ca²⁺, and ATP. After 60 min, reactions were stopped via addition of SDS sample buffer and subjected to WB using the indicated antibodies. Samples without kinase or ATP/Ca²⁺/calmodulin served as additional controls. (G) 200 ng of the indicated recombinant phosphatases were employed in enzyme assays using the fluorogenic substrate 4-MUP. The increase in fluorescence of 4-MU was recorded over 30 min for each enzyme (green squares), while samples without phosphatase (black dots) or including the phosphatase together with a phosphatase inhibitor (orange triangles) served as controls. Depicted is a representative experiment.

Figure 7.

Kindlin-2 association with the phosphorylated integrin β1 cytoplasmic tail requires the presence of talin. (A) Strep-tag-integrin β1 (Strep-ITGB1) cytoplasmic domains in the WT form, with modifications of the T788/T789 motif (T/D, TT/DD, TT/AA), alanine mutation of Tyr-795 (Y795A), or of Tyr-783 (Y783A) were incubated with His-kindlin-2. Upon streptactin pulldown, bound kindlin-2 was detected by WB with α-His antibody or Coomassie staining. 50% of His-kindlin-2 was directly loaded (input) for comparison. (B) His-tagged kindlin-2 was immobilized and incubated with the indicated Strep-ITGB1 variants at 4°C in triplicate wells. After washing, binding was detected by incubation with streptactin-HRP. Bars represent mean ± SEM of triplicates from one representative experiment. (C) Biotinylated integrin β1 peptides with or without phosphorylated T788/T789 were bound to streptavidin-agarose (0.5 mg/ml beads), before incubation with His-kindlin-2. Upon streptavidin pulldown, bound kindlin-2 was detected as in A. *Kindlin-2 is the N-terminal kindlin-2 fragment, which is also detected by α-His-antibody and interacts with integrin tails. (D) His-kindlin-2 and Strep-ITGB1 WT were incubated with increasing amounts of His-FLN19-21. Left panels show the input proteins, right panels protein visualization upon pulldown of ITGB1 by streptactin. Proteins were visualized by Coomassie staining (upper panel) and with α-His to detect bound kindlin-2 (lower panel). Coomassie staining also verified similar amounts of precipitated Strep-ITGB1 WT (lowest bands in the upper panel). (E and F) Strep-tag-integrin β1 (Strep-ITGB1) cytoplasmic domains in the WT form, with modifications of the T788/T789 motif (T/D, TT/DD, TT/AA), alanine mutation of Tyr-795 (Y795A), or of Tyr-783 (Y783A) together with or without T788D/T789D mutation were incubated with kindlin-2 and talin-F3 alone or together in a 1:1 ratio. Upon streptactin pulldown, bound His-tagged proteins were detected by WB with α-His antibody or Coomassie staining. Input is shown in E, pulldown in F; *Kindlin-2 as in C. (G) Indicated biotinylated integrin β1 peptides were bound to streptavidin-agarose (0.25 mg/ml beads) before incubation with His-tagged talin-F3 (1×) and kindlin-2 (1×) with or without GST-tagged FLN19-21 (5× molar amount). Upon streptavidin pulldown, bound His- and GST-tagged proteins were detected by WB using α-His or α-GST antibody and Coomassie staining (right panels); left panels show protein input (100%); *Kindlin-2 as in C. (H) Schematic view of talin and kindlin-2 versus filaminA association with the integrin β1 cytoplasmic tail depending on T788/T789 (pseudo-)phosphorylation. See also Fig. S1. P, phospho-; T, threonine; w/o, without; wt, wild type.

Figure 8.

Kindlin2 and talin versus filaminA recruitment are dictated by integrin β1 phosphorylation and PPM1F activity in intact cells. (A) 293T cells were cotransfected with indicated CEA3-integrin β1 cytoplasmic tail fusion proteins together with GFP-kindlin-2. After 48 h, cells were infected with CEACAM-binding bacteria (Ngo) for 1 h, fixed, and stained for CEACAM3. Micrographs (left panel) show representative infection sites of bacterial attachment (blue) and CEA3-integrin β1 clustering (red). Recruitment of GFP-kindlin-2 (green) to clustered integrin cytoplasmic tails is indicated (white arrowhead). Scale bar, 1 µm. Kindlin-2 recruitment was quantified with the enrichment ratio (ER) indicating the -fold enrichment of GFP intensity at bacterial attachment sites versus the overall cellular GFP level (middle panel). Data from GFP-transfected cells (Fig. 2 A) were used as negative control for statistical calculations. Dots represent individual ER values of 30–60 recruitment sites from n ≥ 2 independent experiments. Horizontal lines indicate mean values and 95% confidence intervals (whiskers). The red line indicates the threshold of positive recruitment. The bar graph (right panel) depicts the percentage of cells showing a ratio of kindlin-2 recruitment ER ³ 2. Statistically significant differences were evaluated using one-way ANOVA, followed by Bonferroni post hoc test (***, P < 0.001; **, P < 0.01). (B) A172 WT, PPM1F KO, or PPM1F KO cells reexpressing mKate-tagged PPM1F or PPM1F D360A were seeded directly or 48 h after transfection with GFP-talin-1 onto 2 µg/ml FN_III9-11 for 1.5 h. Cells were fixed, stained with antibodies against active integrin β1 (9EG7) and optionally against kindlin-2, and analyzed by confocal microscopy; scale bar, 10 µm. Arrowheads point to active integrin β1/talin or kindlin-2 enrichment. Insets: Higher magnification of boxed area; scale bar, 5 µm. (C and D) 293T cells were cotransfected with indicated CEA3-integrin β1 cytoplasmic tail fusion protein in the WT (CEA3-WT) or T788D/T789D (CEA3-TT/DD) form together with GFP, GFP-talin-1 (C), or GFP-FLN19-21 (D). After 48 h, cells were infected with CEACAM-binding bacteria (Ngo) for 1 h, fixed, and stained for CEACAM3. Micrographs (left panel) show representative infection sites of bacterial attachment (blue) and CEA3-integrin β1 clustering (red). Recruitment of GFP proteins (green) to integrin cytoplasmic tails is indicated (white arrowhead). Scale bar, 1 µm. Protein recruitment was quantified with the ER indicating the fold enrichment of GFP intensity at bacterial attachment sites versus the overall cellular level (middle panel). Dots represent individual ER values of 60 recruitment sites from three independent experiments. Horizontal lines indicate mean values and 95% CIs (whiskers). The red line indicates the threshold of positive recruitment. The bar graph (right panel) depicts the percentage of cells showing a ratio of protein recruitment ER ≥ 2. Statistically significant differences were evaluated using one-way ANOVA, followed by Bonferroni post hoc test (***, P < 0.001; *, P < 0.05). See also Fig. S1.

Figure 9.

Homozygous disruption of the ppm1f gene results in embryonic lethality and ppm1f^−/− fibroblasts show constitutive T788/T789 phosphorylation, elevated integrin activity, and increased cell adhesion. (A) Schematic representation of the WT and targeted ppm1f locus. Insertion of a lacZ-neomycin-resistance cassette into exon 4 resulted in gene disruption and expression of β-galactosidase under the control of the ppm1f gene promoter. The primers used for genotyping (blue) and the resulting PCR fragments (red) are shown. E, exon number; P1, gene specific primer forward; P2, gene specific primer reverse; P3, targeted primer forward. (B) DNA extracts from tail biopsies were genotyped by PCR using primers indicated in A to result in a 250-bps product (WT allele, P1, P2) or a 450-bps product (targeted allel, P2, P3). (C) ppm1^+/+ and ppm1f^+/− mice were mated as depicted, and the offspring was genotyped after weaning by PCR using DNA extracts from tail biopsies. (D) Close-up view of head morphology at E10.5 from WT, ppm1f^+/−, and ppm1f^−/− embryos. Ppm1f^−/− embryos are smaller in size and exhibit a stunted telencephalon (white arrow) and reduced development of branchial arches (black arrowhead). Scale bars, 1 mm. (E) Two ppm1f^+/− mice were mated, and genomic DNA was extracted from fibroblasts isolated from ppm1f^+/+ (WT), ppm1f^+/− (heterozygous), and ppm1f^−/− mouse embryos at E10.5 (MEFs). Genotyping PCR identified WT, heterozygous, and homozygous ppm1f KO embryos. (F) WCLs from MEFs isolated from WT (PPM1F^+/+) or from ppm1f^−/− embryos were probed with polyclonal α-mPPM1F antiserum (upper panel) or monoclonal α-tubulin (lower panel). PPM1F^−/− cells do not express truncated versions of PPM1F. (G) MEFs as in F were seeded onto 1 µg/ml FN_III9-12 for 2 h. Samples were fixed and stained for talin (left row) or the active integrin β1 (right row) before analysis by confocal microscopy; scale bar, 20 µm. Insets show higher magnification of boxed areas; scale bar, 5 µm. Arrowheads point to active integrin β1 or talin enrichment. (H) MEFs as in F were seeded onto 2 µg/ml FN_III9-12 for 45 min, and WCLs were subjected to WB with indicated antibodies (left panels). Bar graphs (right panels) show densitometric quantification of band intensities from pT788/pT789-β1 versus total β1 integrin antibody signal from four independent experiments; WT was set to 1. Statistics was performed using one-way ANOVA, followed by Bonferroni post hoc test (***, P < 0.001). (I) MEFs as in F were kept in suspension for 45 min and incubated for 15 min with 10 µg/ml FN_III9-12 (FN). Samples were stained for total (Hmb1-1) or active β1 integrin (9EG7) and analyzed by flow cytometry, ≥10,000 counts. The mean fluorescence intensity (MFI) ratio of active to total β1 integrin was calculated and normalized to the WT sample (= 1). Bars represent mean MFI ± SEM of three independent experiments; statistics was performed using unpaired Student’s t test (**, P < 0.01). (J) MEFs as in F were seeded in triplicate onto fibronectin-coated wells for 20 min, and cell adhesion was quantified. Bars represent mean ± SEM of five independent experiments performed in triplicate. Values were normalized to MEF WT cells (= 1). Statistics was performed using unpaired Student’s t test (***, P < 0.001; *, P < 0.05). (K) Serum-starved MEFs were stimulated by addition of 10% FCS, and cell migration was monitored every 30 min for 12 h using time-lapse microscopy. Cell tracks were evaluated for velocity and covered distance. Bars show mean ± SEM of three independent experiments. Samples were done in duplicate, each n = 15; unpaired t test (***, P < 0.001).

Figure 10.

Phosphorylation and dephosphorylation of the conserved threonine motif in the integrin β tail orchestrate the regulation of integrin activity. (A) FilaminA stabilizes the inactive, clasped low-affinity integrin conformation by outcompeting talin and kindlin-2 at the integrin β tail. (B and C) Upon phosphorylation of integrin β1 at T788/T789 by integrin-targeted kinase(s), filaminA is displaced, and integrins are primed for (C) inside-out activation by talin, resulting in integrin tail separation and a conformational change to the active, unclasped high-affinity integrin state. (D) Talin-initiated tail reorientation promotes cooperative kindlin association (sequentially or simultaneously), integrin clustering (with multivalent ligands), and further integrin downstream signaling. Counteracting these processes, PPM1F-mediated dephosphorylation of integrin β1 T788/T789 allows (re)association of filaminA, which displaces talin and kindlin from the cytoplasmic tail and favors the closed integrin conformation. P, phospho-; T, threonine; w/o, without.