PKP4 regulates actin- and adhesion-associated pathway activities

The dynamic connection of AJs to the actin cytoskeleton is tightly regulated by Rho-signaling. When misregulated, abnormal junctional actin polymerization can drive pathological conditions including cancer, vascular and neurodegenerative diseases [39]. As a novel model system to study how PKP4 affects the cortical actin cytoskeleton via Rho signaling, we used murine WT keratinocytes, PKP4-KO keratinocytes, and PKP4-KO keratinocytes overexpressing GFP-tagged PKP4 (Rescue cell line).

Using these cells, we have performed gene expression analyses to identify molecular processes that are affected by PKP4. RNA sequencing of WT, PKP4-KO, and Rescue cells grown for 24 h or 72 h in LCM or HCM was performed followed by assessment of GSEA of protein coding genes to identify differentially regulated biological pathways and molecular functions. Signaling pathways associated with PKP4 were identified using the gene ontology databases [40]. The downregulated genes in PKP4-KO cells and the upregulated genes in rescue cells were primarily associated with actin- and adhesion-associated pathways. (Figs. 1A, B, S1A, B). In addition, GSEA enrichment plots show that PKP4 regulated genes involved in both, actin filament based processes and in cell adhesion (Figs. 1C, D, S1C, D). Taken together, these data indicate that the loss of PKP4 altered actin- and adhesion-associated pathway activities.

Fig. 1

PKP4 regulates actin- and adhesion-associated pathway activities. A, B Normalized enrichment scores of selected gene sets among protein coding genes in PKP4-KO or Rescue (PKP4-KO + PKP4) versus WT cells after (A) 24 h HCM or (B) 72 h HCM. Positive values represent up regulation, negative values represent down regulation. C, D Enrichment plots for the two most significantly enriched Gene Ontology (GO) Biological Processes gene sets for C PKP4-KO relative to WT cells or (D) Rescue (PKP4-KO + PKP4) relative to WT cells after 24 h HCM incubation, respectively. The plots show the profile of the running enrichment scores and positions of gene set members on the rank-ordered list. Genes on the far left (red) correspond to the most upregulated actin- or adhesion-associated genes, whereas genes on the far right (blue) correspond to the most downregulated actin- or adhesion-associated genes. *P < 0.05; **P < 0.01; ***P < 0.001; ns, not significant. See also Fig. S1

PKP4 affects actin organization

Since AJs and the associated actin cytoskeleton are major determinants of tissue tension and dynamics [41], we investigated the role of PKP4 in generating tension. Keratinocyte monolayers detached from the culture dish by dispase treatment typically contract due to intrinsic forces [42]. This contraction was essentially lost in PKP4-KO cells compared to their WT counterparts as determined by quantifying epithelial sheet areas (Fig. 2A). The effect was reversed in Rescue cells confirming a role of PKP4 in generating tissue tension.

Fig. 2

PKP4 affects actin organization. A Dispase-based tension assay of WT, PKP4-KO, and Rescue (PKP4-KO + PKP4) cells grown for 24 h or 72 h in HCM. Left: Representative images showing the monolayers detached from the culture flask. Scale bar = 5 mm. Right: Quantification of monolayer size relative to WT cells. Box plots show the monolayer area from eighteen independent experiments. The whiskers extend to the minimum and the maximum values. B Immunofluorescence analysis of F-actin organization. Left: Representative immunofluorescence images showing PKP4 and F-actin localization in WT, PKP4-KO, and Rescue (PKP4-KO + PKP4) cells grown for 24 h in HCM. Scale bar = 50 µm, detail 10 µm. Right: Number of cells with stress fibers. Averages + SD from three independent experiments are plotted. n ≥ 100 cells per condition. C G-actin/F-actin ratio in WT, PKP4-KO, and Rescue (PKP4-KO + PKP4) cells. Top: Representative western blot of actin. Ponceau S staining was used as a loading control. Bottom: Quantification of G-actin/F-actin ratio. Box plots show the fold change from five independent experiments. The whiskers extend to the minimum and the maximum values. D Immunofluorescence analysis of Vinculin and F-actin localization. Top: Representative immunofluorescence images showing F-actin and Vinculin localization in WT and PKP4-KO cells and a histogram of the relative average intensities of lateral Vinculin. Scale bar = 10 µm, detail 5 µm. Averages ± SD from n ≥ 100 cells per condition from two independent experiments are plotted. Bottom: Bicellular/cytoplasm ratio of Vinculin fluorescence intensity. n ≥ 100 cells per condition from two independent experiments are shown. Representative histograms of the relative average intensities of lateral F-actin and Vinculin in WT cells and PKP4-KO cells. E Immunofluorescence analysis of FAK and Vinculin localization in WT and PKP4-KO cells. Left: Representative immunofluorescence images showing FAK and Vinculin localization in WT and PKP4-KO cells. Scale bar = 10 µm, detail 5 µm. Right: Quantification of co-localization of FAK and Vinculin. Mander’s coefficient describes the amount of overlap in fluorescence intensity between two channels, here, the FAK and vinculin channel. It ranges from 0 for no co-localization to 1 for complete co-localization. Box plots show the Mander’s coefficient from n ≥ 30 images per condition from two independent experiments. The whiskers extend to the minimum and the maximum values. F Schematic depicting the difference in actin organization between WT and PKP4-KO cells. Created with biorender.com. *P < 0.05; **P < 0.01; ***P < 0.001; ns, not significant. Significance was determined by one-way ANOVA with Tukey’s multiple comparisons test (A–C) or by student’s unpaired two tailed t-test (D, E). See also Fig. S2

Actin filaments and AJs function as tension sensors [9, 43]. Therefore, we asked if actin organization might be altered in PKP4-KO cells. WT cells and Rescue cells grown in HCM for 24 h to induce junction formation revealed a prominent cortical actin ring (Figs. 2B, S2A), whereas PKP4-KO cells displayed a striking increase in stress fibers (Fig. 2B). In order to find out if cortical ring formation was only delayed or completely inhibited, cells were grown for 120 h in HCM to allow junction maturation. Even after this prolonged incubation in differentiation medium, WT and Rescue keratinocytes showed low numbers of cells with stress fibers whereas PKP4-KO cells were characterized by a loss of the cortical actin ring and increased stress fibers (Fig. S2B), suggesting a general failure rather than a delay of cortical actin assembly.

Dynamic alterations of cellular organization depend on actin remodeling. During polymerization, actin undergoes a rapid transition from its globular, monomeric state (G-actin) to its filamentous (F-actin) form [44], revealing that the actin cytoskeleton is a highly dynamic structure. To analyze if the altered actin organization correlates with a change in the G-Actin to F-actin ratio, we determined the G-actin/F-actin ratios in WT, PKP4-KO, and Rescue cells (Fig. 2C). The loss of PKP4 resulted in an increased G-actin/F-actin ratio, suggesting that PKP4 promotes actin polymerization, which might facilitate cortical ring formation.

Stress fibers are often associated with focal adhesion contacts linking them to the underlying extracellular matrix [45]. Vinculin is a marker of focal adhesions but in addition has been recognized to localize at mature AJs that are under tension. Vinculin recruitment to AJs relies on force-dependent changes in α-catenin conformation [8, 46]. To analyze whether cytoskeletal changes affected vinculin distribution, WT and PKP4-KO cells were maintained in HCM for 24 h and processed for immunofluorescence (Fig. 2D). In WT keratinocytes, F-actin was particularly apparent at the cell periphery and vinculin localized predominantly at lateral AJs. In contrast, PKP4-KO cells lacked vinculin staining at the cell periphery. Here, vinculin localized at the tips of actin stress fibers suggesting that the loss of PKP4 led to vinculin localization at focal contacts. Signals from focal adhesions are transduced by the focal adhesion kinase (FAK) to regulate mechanosensing [47]. Co-staining of FAK and vinculin revealed increased co-localization of both proteins in PKP4-KO compared to WT cells (Fig. 2E). Since protein levels of FAK and vinculin were not affected by PKP4 (Fig. S2C), we assume that PKP4 primarily affected the localization of vinculin via generation of tension.

Taken together, we show that loss of PKP4 leads to extensive changes of the actin cytoskeleton, with an increase in cytoplasmic stress fibers at the expense of the cortical actin ring resulting in reduced tension (Fig. 2F).

PKP4 modulates the morphology of junctions and stabilizes keratinocyte cell–cell adhesion

Actomyosin promotes cadherin clustering and thus AJ morphology. Based on the changes in actin organization and the localization of PKP4 in AJs we asked if AJ morphology and/or composition were altered in PKP4-KO keratinocytes. Cells were cultured in HCM for either 24 h to allow for junction formation or for 72 h to enable junction maturation and keratinocyte differentiation. Western Blot analysis revealed essentially unaltered expression of the AJ proteins analyzed (Fig. 3A). Unlike predicted, no compensatory upregulation of p120 was found. A transient and minor reduction of p120 was observed at 24 h after Ca2+ switch but the level was adjusted after 3 days in HCM.

Fig. 3

PKP4 modulates the morphology of junctions and stabilizes keratinocyte cell–cell adhesion. A Levels of AJ proteins in WT and PKP4-KO cells grown for 24 h or 72 h in medium with or without Ca2+. Left: Representative western blots of AJ proteins. Ponceau S staining was used as a loading control. Right: Quantification of protein amounts normalized to Ponceau S staining and relative to WT cells grown for 24 h in medium without Ca2+. Averages + SD from five independent experiments are plotted. B Immunofluorescence analysis of E-cadherin localization. Left: Representative immunofluorescence images showing PKP4 and E-cadherin localization in WT, PKP4-KO, and Rescue (PKP4-KO + PKP4) cells. Scale bar = 50 µm, detail 10 µm. Right: Bicellular/cytoplasm ratio of E-cadherin fluorescence intensity. n ≥ 100 cells per condition from two independent experiments. C Dispase-based dissociation assay of WT and PKP4-KO cells grown for 24 h or 72 h in HCM. Top: Representative images showing the results of dispase assays before and after mechanical stress. Scale bar = 5 mm. Bottom: Quantification of fragment numbers. Box plots show the fragment numbers from six independent experiments. The whiskers extend to the minimum and the maximum values. D Immunofluorescence analysis of E-cadherin localization in WT and PKP4-KO cells treated with EGTA (3.3 mM) for 3 h. Top left: Representative immunofluorescence images showing E-cadherin localization in WT and PKP4-KO cells before (0 h) or 3 h after EGTA treatment. Scale bar = 10 µm. Top right: Overlaying mask (yellow marked polygons) indicating exposed areas. Scale bar = 10 µm. Bottom: Quantification of the exposed areas. Box plots show the exposed area in percent per image from 20 images from two independent experiments. The whiskers extend to the minimum and the maximum values. E Schematic of junction morphology in WT compared to PKP4-KO cells. Created with biorender.com. *P < 0.05; **P < 0.01; ***P < 0.001; ns, not significant. Significance was determined by one-way ANOVA with Tukey’s multiple comparisons test (A, B) or by student’s unpaired two tailed t-test (C, D). See also Fig. S3

To analyze whether the loss of PKP4 affects the morphology of AJs, WT, PKP4-KO, and Rescue keratinocytes were processed for immunofluorescence. AJ morphology was dramatically altered as revealed by E-cadherin staining (Fig. 3B). The loss of PKP4 resulted in an increased cytoplasmic pool of E-cadherin and decreased lateral E-cadherin localization with a punctate pattern instead of more continuous lines observed in WT and Rescue cells. Moreover, tricellular regions were not sealed. These findings indicate that loss of PKP4 interferes with AJ maturation and cortical actin organization.

Since AJs and desmosomes are interconnected [48], we asked whether PKP4 might alter the composition and morphology of desmosomes as well. Western blot analysis revealed a slight increase of differentiation-associated desmosomal proteins desmoglein 1 and PKP1 in PKP4-KO cells after 72 h HCM (Fig. S3A). The most profound effect was on desmoplakin 1/2, which was increased in PKP4-KO cells. Since desmoplakin 1/2 levels were considerably affected, WT, PKP4-KO, and Rescue keratinocytes were immunostained for desmoplakin (Fig. S3B). Similar to AJs, desmoplakin localization was altered with punctate instead of linear staining patterns in PKP4-KO cells. This suggests that the increased amounts of desmoplakin accumulate in the cytoplasm of PKP4-KO but did not incorporate into desmosomes. Again, the tricellular regions remained unsealed. To validate the dramatic loss of junctional proteins at tricellular regions in PKP4-KO cells, WT, PKP4-KO, and Rescue keratinocytes were immunostained for tricellulin, which accumulates primarily at tricellular junctions [49]. Tricellulin was decreased at tricellular regions in PKP4-KO cells (Fig. S3C), suggesting that PKP4 was required to close tricellular regions.

To analyze whether PKP4 interacts with desmosomal components, immunoprecipitation of endogenous PKP4 was performed. PKP1, PKP3, and desmoplakin did co-purify with PKP4 (Fig. S3D), suggesting that an association with PKP4 might facilitate desmosome assembly and/or alter desmosome composition.

To test whether PKP4 also affects formation of desmosomes, we performed a time course analysis of desmosome assembly in WT and PKP4-KO cells (Fig. S3E). HCM induced rapidly E-cadherin recruitment to lateral cell membranes in both, WT and PKP4-KO cells. The desmosomal markers desmoplakin and PKP3 also localized at lateral membranes after HCM treatment, but the loss of PKP4 delayed incorporation of desmoplakin and PKP3 into lateral junctions. These data suggest that PKP4 promotes desmosome formation.

Taken together, these data indicate that the loss of PKP4 results in a reduction of junctional proteins at lateral and tricellular membranes, suggesting that PKP4-KO cells fail to form mature junctions.

Since AJs and desmosomes maintain the mechanical integrity of cell–cell adhesion [50,51,52,53], we asked whether the PKP4-depdendent changes in morphology of AJs and desmosomes would correlate with changes in intercellular adhesion. Epithelial sheet assays were performed to test the strength of cell–cell adhesion (Fig. 3C). In line with the reduced association of junctional proteins with lateral and tricellular membranes in PKP4-KO cells, the loss of PKP4 reduced cell–cell adhesion as revealed by an increased number of fragments generated by mechanical stress. This confirms that PKP4 strengthens cell–cell adhesion.

Upon maturation, desmosomes become Ca2+-independent which marks a state of strong intercellular cohesion. To test if cell junctions in PKP4-KO cells would reach Ca2+-independence, WT and PKP4-KO cells were treated with EGTA and processed for immunofluorescence (Fig. 3D). In WT cells, EGTA treatment for 3 h induced small tricellular openings whereas lateral contacts remained intact as revealed by E-cadherin staining. In contrast, the loss of PKP4 resulted in a reduction of E-cadherin at tricellular contacts already in the absence of EGTA. Incubation with EGTA slightly increased tricellular gaps but in addition strongly reduced lateral E-cadherin staining. These data strongly support a role of PKP4 in junction maturation and stabilization.

Taken together, we propose that PKP4 promotes an association of AJ and desmosomal proteins with lateral and tricellular membranes to stabilize cell–cell adhesion. The loss of PKP4 resulted in decreased lateral localization of junctional proteins and increased tricellular gaps, which weakens cell–cell adhesion (Fig. 3E).

PKP4 promotes ROCK-signaling

Given that the expression of PKP4 is low compared to E-cadherin it seemed unlikely that PKP4 strengthens intercellular adhesion by directly recruiting E-cadherin and catenins. At the same time, intercellular adhesion increases with the application of force [54] supporting a role of actin organization in junction stability. Therefore, we hypothesized that the effects of PKP4 depletion on junctions may be mediated by the loss of cortical actin organization and thus focused on regulators of actin polymerization and actomyosin tension. The dynamic organization of the actin cytoskeleton is regulated by Rho GTPases [55]. GTP-bound RhoA activates ROCK1/2 which leads to activation of myosin light chain kinase (MLCK) and inhibition of MLC phosphatase (MYPT1), resulting in an increase of MLC phosphorylation (P-MLC) to induce actomyosin-based contractility [56]. ROCK also directly phosphorylates LIMK, which results in downstream phosphorylation and inactivation of the actin depolymerizing factor cofilin leading to filament stabilization [57] (Fig. 4A).

Fig. 4

PKP4 promotes ROCK-signaling. A Schematic of ROCK-signaling. P – phosphorylation. Inactive proteins are shown in white. Created with biorender.com. B Amount and phosphorylation of proteins involved in ROCK-signaling. Left: Representative western blots of proteins in WT, PKP4-KO, and Rescue (PKP4-KO + PKP4) cells grown for 24 h in medium with or without Ca2+. GAPDH was used as a loading control. Right: Quantification of protein amounts normalized to GAPDH and relative to WT cells grown in medium without Ca2+. Averages + SD from three independent experiments are plotted. C Immunofluorescence analysis showing the localization of proteins involved in ROCK-signaling. Left: Representative immunofluorescence images showing ROCK2, MLCK, MLC2, and P-MLC2-Ser19 localization in WT, PKP4-KO, and Rescue (PKP4-KO + PKP4) cells. Scale bar = 50 µm, detail 10 µm. Right: Bicellular/cytoplasm ratio of fluorescence intensities. n ≥ 100 cells per condition from two independent experiments. D GFP-PKP4 or GFP was affinity-purified from WT + GFP or PKP4-KO + PKP4 cells. Left: Representative western blots of co-purified proteins. Right: Enrichment of ROCK2, MLCK, and MYPT1 normalized to precipitated GFP and relative to values of GFP cells (second lane in immunoblot, which was set to 1). Average + SD from three independent experiments was plotted. E Endogenous PKP4 was affinity-purified from WT cells. PKP4-KO cells were treated in parallel as control. Top: Representative western blots of input and co-purifying proteins. Bottom: Enrichment of ROCK2, MLCK, and MYPT1 normalized to Ig and relative to values of PKP4-KO cells (second lane in immunoblot, which was set to 1). Average + SD from three independent experiments was plotted. *P < 0.05; **P < 0.01; ***P < 0.001; ns, not significant. Significance was determined by one-way ANOVA with Tukey’s multiple comparisons test (B, C) or by student’s unpaired two tailed t-test (D, E). See also Fig. S4

To analyze the putative effect of PKP4 on ROCK-signaling, we quantified the amounts and activation of several ROCK1/2 effectors in WT, PKP4-KO, and Rescue cells maintained in LCM or HCM for 24 h by western blotting (Fig. 4B). Protein levels of ROCK1, ROCK2, MLC2, cofilin, and myosin IIA were unaltered by PKP4. More importantly, the loss of PKP4 led to increased protein levels of the opponents MLCK and MYPT1 suggesting increased actin dynamics. Strongly decreased phosphorylation of MLC2 supports the reduced contractility observed in the PKP4-KO keratinocyte sheets (Fig. 2A). Furthermore, decreased Cofilin phosphorylation in PKP4-KO cells fosters the data showing increased G-actin levels in these cells (Fig. 2C).

Since cortical actin was essentially lost but stress fibers were increased in PKP4-KO cells, we wondered if this correlates with altered localization of ROCK1/2 and their effectors. WT, PKP4-KO and Rescue keratinocytes were maintained for 24 h in HCM and processed for immunofluorescence. Whereas localization of ROCK1, MYPT1, cofilin, phospho-cofilin, and myosin IIA was unaltered in PKP4-KO cells (Fig. S4), lateral localization of ROCK2 and MLCK was reduced in PKP4-KO cells compared to WT and Rescue cells (Fig. 4C). Moreover, PKP4 dramatically promoted the lateral localization of total MLC2 and phospho-MLC2.

To analyze whether PKP4 interacts with effectors of ROCK1/2-signaling to modulate actin dynamics, GFP-PKP4 was affinity purified from Rescue cells (Fig. 4D). RhoA, ROCK1, MLC2, phospho-MLC2, as well as cofilin, phospho-cofilin, and myosin IIA, did not co-purify with PKP4. In contrast, ROCK2, MLCK, and MYPT1 co-precipitated. This was further validated in an immunoprecipitation of endogenous PKP4 which revealed again a co-precipitation of ROCK2, MLCK, and MYPT1 in WT cells (Fig. 4E), suggesting that an association with PKP4 might alter their activity to control actin dynamics at the cortical ring.

Taken together we show that PKP4 promotes MLC and cofilin phosphorylation to regulate actomyosin-dependent tension and cortical ring formation. PKP4 dramatically increased the lateral localization of active phospho-MLC2, which facilitates actomyosin contraction at the cortical ring. These data suggest disturbed mechanosignaling in PKP4-KO keratinocytes. We conclude that PKP4 might function as a scaffold to locally regulate a ROCK2-MLCK-MLC2 axis to promote cortical actin tension.

PKP4 promotes RhoA activation at the cell cortex and cortical actin ring formation

The balance of the small GTPases RhoA, Rac1, and Cdc42 controls actin cytoskeleton dynamics through ROCK signaling [58]. The disruption of RhoA activity reduced phosphorylation of MLC2 [59] as observed in PKP4-KO cells. To analyze if RhoA activity would be affected, a RhoA activation assay was performed to measure total RhoA activities (Fig. S5A). In accordance with decreased phosphorylation of MLC2, the loss of PKP4 reduced total RhoA activity.

To analyze the roles of the Rho-GTPases Rho, Rac, and Cdc42 in stress fiber and cortical actin organization in more detail, WT, PKP4-KO, and Rescue cells were treated with Rho or Rac/Cdc42 activators and processed for immunofluorescence (Fig. 5A). Rho activation induced stress fiber formation in WT cells whereas Rac/Cdc42 activation suppressed stress fibers in PKP4-KO cells. To further validate the PKP4-dependent regulation of Rho activity, constitutively active or constitutively negative mutants of RhoA GTPases were ectopically expressed in WT and PKP4-KO cells, followed by immunofluorescence (Fig. 5B). In agreement with the Rho activator studies, constitutively active RhoA increased stress fiber formation in WT cells whereas constitutively negative RhoA suppressed stress fibers in PKP4-KO cells suggesting that PKP4 affects the activity of Rho-GTPases.

Fig. 5

PKP4 promotes RhoA activation at the cell cortex and cortical actin ring formation. A Immunofluorescence analysis of F-actin organization after Rho or Rac/Cdc42 activation. Left: Representative immunofluorescence images showing PKP4 and F-actin localization in WT and PKP4-KO cells treated with PBS (Mock), Rho activator II (5 µg/ml), or Rac/Cdc42 activator II (5 units/ml). Scale bar = 50 µm, detail 10 µm. Right: Number of cells with stress fibers. Averages + SD from three independent experiments are plotted. n = 50 cells per condition. B Immunofluorescence analysis of F-actin organization in constitutive active or negative RhoA expressing cells. Left: Representative immunofluorescence images showing F-actin localization in WT and PKP4-KO cells after ectopic expression of constitutive active RhoA (RhoA-CA(Q63L)) or constitutive negative RhoA (RhoA-CN(T19N)). Scale bar = 50 µm, detail 10 µm. Right: Number of cells with stress fibers. Averages + SD from three independent experiments are plotted. n = 10 cells per condition. C Immunofluorescence analysis showing total RhoA, Rac1, and Cdc42 localization. Left: Representative immunofluorescence images showing total RhoA, Rac1, or Cdc42 localization in WT and PKP4-KO cells. Scale bar = 100 µm, detail 10 µm. Right: Bicellular/cytoplasm ratio of fluorescence intensity. n ≥ 100 cells per condition from two independent experiments. D Immunofluorescence analysis of the localization of active RhoA, Rac1, and Cdc42. Left: Representative immunofluorescence images showing active RhoA, Rac1, or Cdc42 localization in WT and PKP4-KO cells. Scale bar = 100 µm, detail 10 µm. Right: Bicellular/cytoplasm ratio of fluorescence intensity. n ≥ 100 cells per condition from two independent experiments. E Immunofluorescence analysis showing the localization of a RhoA-GTP biosensor. Top: Representative immunofluorescence images showing EGFP-RhoA biosensor localization in WT and PKP4-KO cells. Scale bar = 50 µm, detail 10 µm. Bottom: Bicellular/cytoplasm ratio of fluorescence intensity. n ≥ 50 cells per condition from two independent experiments. *P < 0.05; **P < 0.01; ***P < 0.001; ns, not significant. Significance was determined by student’s unpaired two tailed t-test (A, B, C, D, E). See also Fig. S5

To resolve the apparent contradiction between the reduced total RhoA activity found in PKP4-KO cells and the fact that hyperactive RhoA mimicked the PKP4-phenotype, we focused on the local regulation of Rho-GTPases. Active RhoA was shown to localize in the cytoplasm and at the membrane but to activate effector proteins primarily when localized at the membrane where RhoA needs to be stabilized to engage downstream pathways of contractility [60]. Therefore, we asked if PKP4 would contribute to the stabilization of active RhoA at the membrane. To evaluate if the cortical localization and/or activity of RhoA, Rac1, and/or Cdc42 was indeed dependent on PKP4, WT and PKP4-KO cells were processed for immunofluorescence (Fig. 5C, D). The loss of PKP4 correlated with decreased lateral localization of both total and active RhoA. Rac and Cdc42 localization were essentially unaffected but lateral localization of active Rac1 was increased in PKP4-KO cells compared to WT cells whereas Cdc42 activity was unaltered. These data suggest that PKP4 promotes the localization of active RhoA at the lateral membrane but suppresses lateral Rac1 activity.

To further validate a PKP4-dependent localization of active RhoA, a RhoA biosensor was ectopically expressed in WT and PKP4-KO cells and the cells were processed for immunofluorescence (Figs. 5E, S5B). This sensor visualizes endogenous Rho-GTP. Like E-cadherin, the sensor accumulated strongly at the lateral membranes in WT cells, indicating that RhoA was locally active at the plasma membrane. In contrast, the loss of PKP4 resulted in the loss of lateral RhoA-GTP which instead showed a cytoplasmic localization although E-cadherin membrane association was unaltered. This indicates reduced RhoA activity at the lateral membrane of PKP4-KO cells.

Taken together, we show that PKP4 suppresses cytoplasmic RhoA activation to restrain stress fiber formation but promotes RhoA activity at lateral membranes to facilitate cortical actin ring formation and the generation of tension through a RhoA-ROCK-MLCK-MLC2 axis.

PKP4-dependent actin dynamics and tension are regulated by ARHGAP23 and ARHGEF2

Local changes in Rho-GTPase activity in the PKP4-KO cells raised the question how PKP4 modulates Rho-signaling. Spatio-temporal control of Rho-GTPases depends on the local balance of GEF and GAP activities. Therefore, we hypothesized that PKP4 might affect Rho-signaling by modulating GEF and/or GAP localization and/or activities at keratinocyte AJs.

GEFs promote the release of GDP in exchange for GTP to activate the GTPase. In contrast, GAPs increase the intrinsic hydrolytic activity that converts GTP into GDP thereby inactivating the GTPase [10, 11]. In order to find out how PKP4 modulates the activity of Rho-GTPases we investigated interactions with these upstream regulators. For this purpose, specific GEFs and GAPs were selected for a detailed characterization based on several criteria (Fig. 6A): First, we analyzed those candidates that were already known to associate with PKP4 in a different context. A direct interaction between PKP4, the GEF ECT2, and RhoA has been previously identified during cytokinesis [20]. Moreover, RACGAP1 [61] and ARHGAP21 had been described as a PKP4 interacting protein [12]. Since ARHGAP23 is similar to ARHGAP21 and both contain a PDZ domain, which might mediate PKP4 binding via its PDZ binding motif, ARHGAP23 was also included. Secondly, GEFs and GAPs that were known to modulate cell junctions were examined. ARHGEF2 regulates the assembly of AJs [62], and ARHGAP24 promotes the formation of AJs by accumulating E-cadherin [63].

Fig. 6

PKP4-dependent actin dynamics and tension are regulated by ARHGAP23 and ARHGEF2. A Schematic of selected GEFs and GAPs and their role in promoting or inhibiting RhoA-activity, respectively. Created with biorender.com. B GFP-PKP4 or GFP was affinity-purified from WT + GFP or PKP4-KO + PKP4 cells. Representative western blots of co-purified proteins. C Protein level of selected GEFs and GAPs. Left: Representative western blots of GEF and GAP proteins in WT, PKP4-KO, and Rescue (PKP4-KO + PKP4) cells grown for 24 h in medium with or without Ca2+. GAPDH was used as a loading control. D Immunofluorescence analysis of the F-actin organization in siRNA treated WT and PKP4-KO cells. Left: Representative immunofluorescence images showing F-actin localization in WT and PKP4-KO cells after knockdown of the indicated GEFs and GAPs. Scale bar = 50 µm, detail 10 µm. Right: Number of cells with stress fibers. Averages + SD from three independent experiments are plotted. n ≥ 100 cells per condition. E Schematic of the effects of ARHGEF2 and ARHGAP23 in WT and PKP4-KO cells. Created with biorender.com. F Immunofluorescence analysis of F-actin organization in WT cells ectopically expressing ARHGAP23 and in PKP4-KO cells ectopically expressing ARHGEF2. Left: Representative immunofluorescence images showing F-actin localization in WT and PKP4-KO cells after ectopic expression of ARHGAP23 or ARHGEF2, respectively. Scale bar = 100 µm, detail 10 µm. Right: Number of cells with stress fibers. Averages + SD from three independent experiments are plotted. n ≥ 30 cells per condition. G Dispase-based tension assay of WT and PKP4-KO cells treated with ARHGEF2- or ARHGAP23-directed siRNAs. Left: Representative images showing the monolayers. Scale bar = 5 mm. Right: Quantification of monolayer size relative to non-targeting siRNA (siCtrl) treated cells. Averages + SD from five independent experiments are plotted. *P < 0.05; **P < 0.01; ***P < 0.001; ns, not significant. Significance was determined by student’s unpaired two tailed t-test (D, F, G). See also Fig. S6

To analyze whether PKP4 interacts with selected GEFs and GAPs, PKP4-GFP was affinity-purified from Rescue cells (Fig. 6B, for quantification see Fig. S6A). ARHGAP21 did not co-purify with PKP4. In contrast, the GAPs ARHGAP23, ARHGAP24, and RACGAP1 co-precipitated. In addition, the GEFs ECT2 and ARHGEF2 (also called GEFH1) were both co-precipitated. The co-precipitation indicated an association with PKP4 and emphasized a role of PKP4 as a scaffold in the spatio-temporal control of Rho signaling.

The function of GEFs and GAPs is regulated by expression as well as localization. To analyze a putative effect of PKP4 on the protein level of the selected GEFs and GAPs, we quantified their amounts in WT, PKP4-KO, and Rescue cells maintained for 24 h in LCM where PKP4 is cytoplasmic or in HCM where PKP4 localizes at cell junctions (Fig. 6C, for quantification see Fig. S6B). Protein levels of ARHGAP21, ARHGAP24, RACGAP1, and ECT2 were unaltered by PKP4 as determined by western blotting. More importantly, the loss of PKP4 slightly increased ARHGAP23 protein expression but dramatically decreased the ARHGEF2 protein level. These effects were reversed in Rescue cells, supporting a role of PKP4 in regulating ARHGAP23 and ARHGEF2 levels.

To directly link the selected GEFs and GAPs with actin dynamics, the impact of siRNA-mediated repression of these GEFs and GAPs was studied with respect to actin organization. WT and PKP4-KO cells were treated with control (siCtrl) or GEF/GAP-directed siRNAs (siARHGAP21, siARHGAP23, siARHGAP24, siRACGAP1, siECT2, siARHGEF2), maintained for 24 h in HCM, and processed for immunofluorescence (Fig. 6D, for quantification of knockdown efficiencies see Fig. S6C). ARHGEF2 depletion increased stress fibers in WT cells whereas control-treated WT cells revealed a cortical actin ring, suggesting that ARHGEF2 depletion reduced cortical RhoA activity and thus mimicked the effect of PKP4 loss. In PKP4-KO cells, ARHGAP23 depletion correlated with a loss of stress fibers, suggesting that ARHGAP23 is active in the PKP4-KO cell cytoplasm to reduce cortical RhoA activity and suppress cortical ring formation. Thus, ARHGEF2 and ARHGAP23 might be the key regulators for actin organization in a PKP4-dependent manner (Fig. 6E).

Based on these findings, we predicted that overexpression of ARHGAP23 should promote stress fibers in WT cells. Transfected cells were maintained for 24 h in HCM and processed for immunofluorescence (Fig. 6F). In accordance with our assumption, overexpressed ARHGAP23 partially localized at the cell cortex where it could inactivate RhoA to prevent cortical actin ring formation and led to increased stress fibers in WT cells. Moreover, we wondered if overexpression of ARHGEF2 in PKP4-KO cells might partially rescue cortical actin ring formation because of a general massive increase in its activity although PKP4 would not mediate its cortical localization. Indeed, overexpression of ARHGEF2 reduced stress fibers in PKP4-KO and improved cortical actin to some extent.

To reveal the functional consequences of the PKP4-dependent actin regulation by ARHGAP23 and ARHGEF2 we investigated if this would also affect cell tension. WT and PKP4-KO cells maintained for 24 h in HCM were treated with control (siCtrl), ARHGEF2-directed (siARHGEF2), or ARHGAP23-directed (siARHGAP23) siRNAs and processed for the dispase assay (Fig. 6G). AR