Plakophilin 3 phosphorylation by ribosomal S6 kinases supports desmosome assembly Free

Desmosomes mediate strong cohesion between cells in tissues exposed to mechanical strain, such as the skin and the heart (Johnson et al., 2014; Broussard et al., 2015; Hatzfeld et al., 2017). They are composed of desmosomal cadherins, desmogleins (DSGs) and desmocollins (DSCs), which connect the adjacent cells. Cadherin function depends on the desmosomal plaque proteins junctional plakoglobin (JUP) and plakophilin (PKP) 1–3, which act as scaffolds to link the desmosomal cadherins to desmoplakin (DSP). DSP in turn anchors keratin filaments (Thomason et al., 2010; Kowalczyk and Green, 2013; Johnson et al., 2014; Spindler and Waschke, 2014). In the epidermis, desmosome adhesive strength is controlled by PKP isoform expression. While PKP1 promotes the hyper-adhesive state to stabilize intercellular adhesion, PKP3 renders desmosomes more dynamic (Garrod and Tabernero, 2014; Tucker et al., 2014; Keil et al., 2016). This facilitates tissue remodeling as required during regeneration and wound healing.

PKPs are also involved in transducing signals to regulate cellular processes including protein synthesis, cell growth, proliferation and migration, and have been implicated in tumor development (Wolf and Hatzfeld, 2010; Wolf et al., 2010; Fischer-Keso et al., 2014; Hatzfeld et al., 2014; Munoz et al., 2014; Rietscher et al., 2018). PKP1 and PKP3 regulate mRNA stability of specific targets including desmosomal mRNAs (Fischer-Keso et al., 2014). Moreover, PKP1 controls cell proliferation and size by regulating protein synthesis via the translation initiation factor eukaryotic initiation factor 4A (eIF4A; also known as EIF4A1) (Wolf et al., 2010).

A preponderance of adhesive or signaling functions either attenuates or promotes cell growth and proliferation, respectively. Therefore, the balance between these functions needs to be tightly controlled to allow for tissue remodeling and repair, and to prevent over-proliferation and cancer. Thus, it is important to understand how PKPs and their differential contributions to intercellular adhesion are regulated in the epidermis. We have recently shown that PKP1 function is regulated by insulin/insulin-like growth factor 1 (IGF1) signaling via AKT2-mediated phosphorylation. Whereas non-phosphorylated PKP1 localizes in desmosomes and stabilizes cell adhesion, its phosphorylated form is translocated into the cytoplasm, where it promotes proliferation and induces anchorage-independent growth, a hallmark of cancer (Wolf et al., 2013). So far, the regulation of PKP3 is incompletely understood. A highly transient tyrosine phosphorylation was shown to promote PKP3 cytoplasmic localization under oxidative stress conditions (Neuber et al., 2015). PKP3 phosphorylation at serine residues including serine 285 induced an interaction of cytoplasmic PKP3 with 14-3-3σ/stratifin (Roberts et al., 2013) and supported the segregation of PKP1 to lateral, and of PKP3 to tricellular, contacts in keratinocytes (Rietscher et al., 2018). However, it is not known which signaling pathways or kinases are involved in these processes.

Here, we provide evidence that PKP3 is regulated by epidermal growth factor (EGF) signaling via ribosomal S6 kinases (RSKs). RSK1 and RSK2 promote PKP3 localization at the plasma membrane and increase intercellular cohesion in proliferating keratinocytes. This emphasizes that PKP1 and PKP3 not only display differential functions in regulating keratinocyte adhesion, but also are independently regulated by distinct signaling pathways. The coordinated action of insulin/IGF-1 signaling on PKP1 and EGF signaling on PKP3 prevents hyper-adhesion, but allows for formation of more dynamic PKP3-dependent desmosomes to ensure intercellular cohesion in proliferating keratinocytes.

PKP1 and PKP3 fulfill distinct and partially non-overlapping functions in desmosome adhesion. Whereas PKP3 facilitates desmosome assembly, PKP1 promotes desmosome maturation to a Ca²⁺-independent state referred to as ‘hyper-adhesion’ (Garrod and Tabernero, 2014). Therefore, we set out to investigate whether these distinct functions are regulated by distinct signals and are mediated by specific post-translational modifications. Keratinocyte proliferation, differentiation and migration depend on growth factors such as insulin/IGF1 and EGF. Therefore, we tested the effects of insulin and EGF on PKP3 phosphorylation in mouse keratinocytes and compared them to untreated starved mouse keratinocytes. The activity of both growth factors was validated by testing downstream kinase activation of the MEK–ERK and AKT–mTOR signaling pathways using phospho-specific antibodies (Fig. 1A,B; Fig. S2A). MEK (also known as MAP2K), ERK (also known as MAPK), RSK and AKT proteins, S6 kinase (S6K; also known as RPS6KB1) and the ribosomal protein S6 were all phosphorylated after insulin and/or EGF treatment. Immunoprecipitated PKP3 was then tested for its phosphorylation using an antibody that recognizes the cAMP-dependent protein kinase, cGMP-dependent protein kinase and protein kinase C (AGC-kinase) consensus motif RXXpS/pT. PKP3 was phosphorylated at such motifs after EGF, but not after insulin treatment (Fig. 1A; see Fig. S1A for putative antibody recognition sites). In contrast, PKP1 was phosphorylated in response to insulin or IGF1 treatment but not after EGF stimulation (Wolf et al., 2013). To further validate these data, we checked a human cell line, A431, for PKP3 modification by EGF (Fig. 1A; Fig. S2B). Again, activation of the AKT or MEK–ERK pathways was confirmed by phospho-specific antibodies against AKT, S6K, MEK, ERK and RSK proteins. Probing immunoprecipitated PKP3 for phosphorylation at RXXS/T motifs confirmed a strong effect of EGF and a lack of phosphorylation after insulin treatment. We also tested a phospho-protein kinase A (PKA) substrate antibody that recognizes a RRXpS motif found at amino acid position 131–134 of PKP3 (Fig. S1A). This confirmed a phosphorylation of PKP3 in response to EGF, but not to insulin, and suggested a role of serine 134 in EGF-dependent regulation of PKP3 (Fig. S1B,C). Since EGF activates MEK and ERK kinases upstream of RSKs we also checked whether a serine/proline motif as detected by the phospho-MAPK/CDK substrate antibody (PXpSP or pSPXR/K) would become phosphorylated in response to EGF stimulation. However, we did not detect differential signals upon stimulation with EGF or insulin, suggesting that serine 238 (SPSR motif) was not phosphorylated by MEK or ERK kinases (Fig. S1B,C). These data indicate that EGF stimulates PKP3 phosphorylation at AGC-kinase consensus sites, whereas insulin, which promoted PKP1 phosphorylation and its accumulation in the cytoplasm, had no effect on PKP3.

We next asked whether phosphorylation of PKP3 by EGF signaling induces altered PKP3 localization. Mouse keratinocytes were starved and then treated for 15 min with EGF, insulin or complete medium and processed for immunofluorescence. In starved cells, PKP3 (Fig. 2A) and DSP (Fig. S3A) were hardly found at cell contact sites. Insulin treatment did not improve PKP3 or DSP recruitment to the plasma membrane. In contrast, a 15 min EGF treatment considerably improved PKP3 association with lateral membranes, although tricellular contacts were not formed. Incubation in complete medium for 15 min was sufficient to induce PKP3 lateral and tricellular localization (Fig. 2A,B; for method of quantification see Fig. S4).

Since clustering of PKP3 at tricellular contacts is supported by its displacement from lateral membranes by PKP1 (Keil et al., 2016), we wondered whether a longer time period of EGF stimulation would allow PKP3 distribution to cell corners even in the absence of complete medium, and, if so, whether a brief stimulation would be sufficient to induce PKP3 recruitment to the membrane. Therefore, we analyzed localization changes after either a prolonged incubation with EGF for 6 h or a short EGF stimulation for 15 min followed by 6 h incubation without growth factors (Fig. 2C,D; Fig. S3B). Immunofluorescence analysis showed that PKP3 localization at tricellular regions increased and gaps in the cell monolayer were closed after 6 h of EGF stimulation, whereas incubation without growth factors following a brief stimulation did not improve tricellular contact formation. However, EGF alone was clearly less efficient than complete medium. This suggests that EGF supports desmosome formation but that their full maturation requires additional signals.

Similar effects were observed in A431 cells, although due to the low expression levels of PKP1, PKP3 does not accumulate at tricellular contacts in these cells. As observed in keratinocytes, insulin had essentially no effect on PKP3 membrane association, whereas EGF induced a strong increase in PKP3 and DSP membrane association (Fig. 2E,F; Fig. S3C). Medium containing FCS also improved junctional localization of both desmosomal proteins.

Taken together, these data indicate that a 15 min incubation with EGF was sufficient to promote desmosome assembly by recruiting PKP3 to lateral membranes. However, a brief EGF treatment was not sufficient for PKP3 sorting into tricellular contacts.

To test whether EGF-mediated desmosome assembly depended indeed on PKP3, we compared desmosome formation in wild-type (WT) and PKP3 knockout (PKP3-KO) keratinocytes (Sklyarova et al., 2008; Keil et al., 2016). Ca²⁺ was added to induce desmosome formation and cells were either kept without growth factors or treated with insulin, EGF or complete medium. After 2 h incubation in high Ca²⁺ medium (HCM), DSP membrane association was significantly improved in the presence of EGF or complete medium compared to insulin, supporting the notion that EGF improves desmosome initiation in WT keratinocytes. In contrast, PKP3-KO keratinocytes revealed hardly any desmosomes after 2 h of Ca²⁺ addition, either in the presence or in the absence of EGF, indicating that the effect of EGF signaling in facilitating early desmosome formation depended on PKP3 (Fig. 3A,B). Therefore, we conclude that PKP3 specifically mediates EGF-facilitated desmosome assembly.

EGF activates the MEK–ERK kinase cascade as well as AKT-mediated signaling. In order to identify the kinase(s) responsible for PKP3 phosphorylation, starved mouse keratinocytes were treated with EGF in the absence or presence of a series of kinase inhibitors, namely Gefitinib [epidermal growth factor receptor (EGFR) inhibitor], Rapamycin (mTOR inhibitor), PF-4708671 (S6K inhibitor), Osu-03012 (PDK1 inhibitor), U0126 (MEK inhibitor), FR180204 (ERK inhibitor) or BI-D1870 (RSK inhibitor). Pathway activation by EGF and impact of inhibitors was validated by probing cell lysates with phospho-specific antibodies as indicated (Fig. 4). PKP3 was immunoprecipitated and its phosphorylation determined using the RXXpS/pT motif antibody. PKP3 phosphorylation was reduced upon inhibition of EGFR, MEK, ERK and RSK kinases, whereas inhibition of mTOR, S6K or PDK1 did not interfere with PKP3 phosphorylation (Fig. 4). This confirms the importance of EGF signaling and strongly suggests a role of a member of the RSK family in the regulation of PKP3.

The RSK family has four members: RSK1 (gene name RPS6KA1), RSK2 (RPS6KA3), RSK3 (RPS6KA2) and RSK4 (RPS6KA6). To identify the kinase responsible for PKP3 phosphorylation in mouse keratinocytes or A431 cells, we analyzed the expression pattern of all four RSKs at mRNA (quantitative RT-PCR, Fig. S5A) and protein (western blotting, Fig. S5B) levels. RSK1 and RSK2 were the predominant kinases in both cell types, whereas RSK3 and RSK4 were not detectable at the protein level and rare or undetectable at the mRNA level. Therefore, we focused on RSK1 and RSK2.

To distinguish between overlapping or unique roles of RSK1 and RSK2, we performed knockdown studies in EGF-stimulated mouse keratinocytes (Fig. 5A, left panel) and A431 cells (Fig. 5A, right panel), and analyzed PKP3 phosphorylation after immunoprecipitation (IP) as described before. Whereas the knockdown of RSK1 was efficient and specific, RSK2 knockdown led not only to a strong downregulation of RSK2 but also to a moderate downregulation of RSK1 in both keratinocytes and A431 cells. PKP3 phosphorylation was strongly reduced after RSK1 knockdown, whereas RSK2 knockdown did not lead to a significant reduction in PKP3 phosphorylation. Treatment of EGF-stimulated mouse keratinocytes with the LJH685 RSK inhibitor (Aronchik et al., 2014) confirmed reduced PKP3 phosphorylation at RXXpS motifs (Fig. 5A). Next, we analyzed PKP3 phosphorylation after overexpression of RSK1 and RSK2 (Fig. 5B). RSK1 strongly increased PKP3 phosphorylation at RXXS/T motifs, whereas RSK2 induced a much milder increase in PKP3 phosphorylation. Moreover, RSK1, and to a lesser extent RSK2, were co-precipitated with PKP3 (Fig. 5B). To further validate this association, we performed bimolecular fluorescence complementation (BiFC) analyses for RSK1 and RSK2 together with PKP3 (Fig. 5C,D). This experiment confirmed a strong association between PKP3 and RSK1 and a much weaker signal for RSK2 in line with the results from phosphorylation studies.

Since RSKs are known as cytosolic and nuclear proteins, we wondered whether RSKs and PKP3 might colocalize during cell contact formation. Therefore, we analyzed RSK and PKP3 localization in low Ca²⁺ medium (LCM) before cell contact formation and after 1 h and 24 h in HCM (Fig. 5E). A considerable portion of RSKs was found to colocalize with PKP3 at the plasma membrane at 1 h after induction of cell contacts. At 24 h post-Ca²⁺ addition, RSKs remained at lateral contacts, where they colocalized with PKP3. However, this colocalization did not extend to the tricellular regions, suggesting that PKP3 phosphorylation by RSK occurs primarily at lateral membrane regions. Taken together, these data uncover a role of RSKs in the regulation of PKP3.

Based on the observed association between PKP3 and RSK1 and PKP3 phosphorylation following RSK1 overexpression, we asked if this is reflected by changes in PKP3 localization and desmosome structure. Overexpression of RSK1 was compatible with strong tricellular localization of PKP3. Lateral desmosomes appeared as straight lines, as typically observed after desmosome maturation (Fig. S5C–E). Overexpression of RSK2 supported strong lateral PKP3 localization but did not promote its tricellular accumulation (Fig. S5C–E).

We next analyzed if inhibition or knockdown of RSKs would modulate PKP3 localization. RSK1 and RSK2 knockdown was confirmed by western blotting (Fig. S6A). The knockdown of either RSK1 or RSK2 prevented tricellular localization of PKP3 and led to punctate desmosomes characteristic of native immature desmosomes (Fig. 6A,B). A similar effect was observed after application of RSK inhibitors BI-D1870 and LJH685 (Fig. 6C). The impact of the inhibitors on RSK phosphorylation, S6K activation and ribosomal protein S6 phosphorylation (S6 is a target of RSKs and S6K) was validated by western blotting (Fig. S6B). This revealed that the LJH685 inhibitor was specific for RSKs, whereas BI-D1870 additionally reduced S6K phosphorylation and thus activity (Fig. S6B).

Since sorting of PKP1 and PKP3 to lateral and tricellular contacts, respectively, correlated with an increase in intercellular cohesion, we asked whether the changes in PKP3 distribution to punctate lateral desmosomes would correlate with changes in intercellular adhesion. Epithelial sheet assays were performed to test the strength of cohesion. In line with the altered localization of PKP3, the knockdown of RSK1 or RSK2 reduced intercellular cohesion as revealed by an increased number of fragments generated by mechanical stress (Fig. 6D; Fig. S7A). This impact on intercellular adhesion was completely lost in PKP3-KO cells (Fig. 6E; Fig. S7B), confirming that the effect of RSKs on desmosomes depended on PKP3. Since the BI-D1870 inhibitor turned out to be less specific than the LJH685 inhibitor, and to inhibit S6K in addition to RSKs, we used the LJH685 inhibitor to assess the role of RSKs in intercellular adhesion. Although fragment number increased in LJH685-treated WT keratinocytes, the effect was lost in PKP3-KO cells (Fig. 6F; Fig. S7C), in agreement with the knockdown studies. Taken together, these data substantiate a specific role of RSKs in the regulation of desmosomes involving PKP3.

We also tested the effect of ethylene glycol-bis-(2-aminoethylether)-N,N,N′,N′-tetraacetic acid (EGTA) treatment on the stability of intercellular cohesion. In WT cells treated with non-targeting small interfering RNA (siRNA), EGTA induced small gaps at tricellular regions as described previously (Keil et al., 2016; Loschke et al., 2016), indicating that these regions are more dynamic than lateral contacts. Prolonged EGTA treatment for up to 3 h led to a widening of tricellular openings but lateral contacts remained intact (Fig. S7D). In contrast, the knockdown of RSK1 resulted in a strong reduction of PKP3 and DSP at tricellular contacts after 1 h of EGTA incubation. Prolonged incubation with EGTA led to an extensive loss of PKP3 and DSP from lateral membranes, indicating that RSK1 stabilizes lateral contacts. In the absence of PKP3 (PKP3-KO cells), RSK1 knockdown had no effect on tricellular localization of either PKP1 or DSP, and tricellular gaps remained small even in the absence of RSK1 (Fig. S7D), strongly suggesting a PKP3-dependent stabilization of cell cohesion by RSK1.

To identify the phosphorylation site(s) essential for PKP3 regulation by RSKs we compared kinase-specific predictions for the RXXS motifs in PKP3 (Table S4; GPS5.0, http://gps.biocuckoo.org/; NetPhos3.1, http://www.cbs.dtu.dk/services/NetPhos/; Kinexus, http://www.phosphonet.ca/). Moreover, we took into consideration if a specific site had been detected in high-throughput screens for phosphorylated proteins, as summarized at PhosphoSitePlus (https://www.phosphosite.org/proteinAction?id=3262&showAllSites=true), and compared these data to the results of an in vitro phosphorylation assay of PKP3 peptides using purified RSK1. Serine 134/135 became phosphorylated in our in vitro assay, was predicted as an RSK target by all three programs and had been detected as being phosphorylated in several high-throughput screens (Table S4). Moreover, the RRXpS (PKA substrate) antibody suggested that this residue became phosphorylated in response to EGF stimulation (Fig. S1B,C). Thus, we determined the importance of this motif in detail by analyzing the localization of the corresponding non-phosphorylatable (S to A) and phospho-mimetic PKP3 mutants (S to E). We used PKP3-KO keratinocytes for overexpression in order to avoid interference with endogenous WT protein. Expression levels of PKP3-WT and both mutants were similar (Fig. 7C). The PKP3-S134/135A mutant revealed a punctate or frayed appearance of lateral desmosomes without tricellular PKP3 (Fig. 7A,D). Frequently, this was accompanied by small gaps at tricellular regions in the cell monolayers. In contrast, the phospho-mimetic mutant PKP3-S134/135E assembled into straight lateral desmosomes and localized at tricellular junctions (Fig. 7A,D). Line scans across the lateral contacts highlight the significant broadening of these contacts in PKP3-S134/135A mutant-expressing cells (Fig. 7B). Quantification supports a slight increase of the phospho-mimetic mutant at lateral membranes and a reduction of the PKP3-S134/135A mutant at tricellular contacts accompanied with a significant broadening of lateral contacts (Fig. 7D,E). We also mutated serines 256/257 to alanine and glutamic acid although only one program predicted this to be a putative RSK motif (Table S4). However, a peptide containing these residues became phosphorylated by RSK1 in vitro. Comparison of the unphosphorylated S256/257A and S256/257E mutants revealed a similar localization pattern (Fig. S8), strongly suggesting that S134/135 is more important in PKP3 regulation by RSKs.

Taken together, these data support our notion that phosphorylation of PKP3 at S134/135 by RSKs supports its localization at cell contacts and that a lack of phosphorylation interferes with its accumulation at tricellular regions.

Although the overall importance of PKPs in desmosome function is well established and cell type-specific expression of plakophilins has been known for years, the response of PKP isoforms to context-dependent signals and its consequence for desmosome adhesion and signaling remain incompletely understood. Recent data from several groups, including our own, have established that PKP1 is vital for stable cohesion (McGrath et al., 1997; Tucker et al., 2014; Rietscher et al., 2016; Fuchs et al., 2019), whereas PKP3 is more dynamic and facilitates desmosome assembly and remodeling (Keil et al., 2016). These complementary functions suggest a coordinate and differentiation-specific regulation in keratinocytes that express both PKP isoforms. Here, we establish a role of EGF signaling through RSKs in supporting plasma membrane recruitment of PKP3 to facilitate desmosome assembly in basal keratinocytes. Our study unravels the mechanism that links EGFR signaling to PKP3 function, providing new insight into the role of EGF signaling in modulating intercellular adhesion via PKP3.

So far, RSKs have not been implicated in the control of desmosomal adhesion. Our data show that RSK phosphorylates PKP3 in response to EGFR activation but not in response to insulin/IGF1 receptor activation. Moreover, EGF, but not insulin, accelerated PKP3 recruitment to the plasma membrane to facilitate desmosome formation. Inhibition or depletion of RSK1 resulted in the loss of PKP3 from tricellular contacts along with the formation of small gaps in these regions, whereas lateral contacts resembled immature desmosomes, as observed early after cell contact initiation. These changes in localization correlated with reduced intercellular cohesion. The importance of PKP3 as a transmitter of EGF signaling was convincingly shown by the complete loss of the effects on desmosome assembly and intercellular cohesion in PKP3-KO keratinocytes.

There are several possibilities as to how RSK-mediated PKP3 phosphorylation might contribute to the regulation of its localization. It appears likely that phosphorylation at S134 alters the affinity for specific interacting proteins. This has been shown for AKT2-mediated phosphorylation of PKP1, which reduced its interaction with DSG1 and DSP, but not eIF4A, resulting in its loss from desmosomes and accumulation in the cytoplasm (Wolf et al., 2013). In the case of PKP3, phosphorylation by RSKs might modulate its delivery to the plasma membrane and/or its interaction with junctional proteins including adherens junction proteins and desmosomes. As a consequence, desmosome dynamics might be altered to facilitate its distribution into tricellular contacts.

EGF and insulin/IGF1 are both essential for keratinocyte proliferation. Their receptors are most prominently expressed in the basal layer of the epidermis, where stem cells reside to allow skin regeneration, whereas expression decreases in suprabasal keratinocytes to enable terminal differentiation (Nanney et al., 1984; Hodak et al., 1996). The balance between proliferation and differentiation relies not only on mitogenic signals but also on the tight control of cell adhesion. Cell proliferation in the basal layer requires dynamic cell–cell interactions, whereas an intact barrier that protects the body from environmental insults depends on strong and stable cell cohesion in the suprabasal layer. This observation implicates a change in cell contacts from basal to suprabasal layers that can be achieved by altered composition and post-translational modifications. PKP1 levels are low in basal keratinocytes (Neuber et al., 2010; Wolf et al., 2013), with sparse desmosomal PKP1 but a considerable cytoplasmic pool. We have shown previously that insulin/IGF1 signaling activates the PI3K–AKT signaling cascade and promotes PKP1 phosphorylation by AKT2. Phosphorylated PKP1 localized in the cytoplasm, where it stimulated translation and proliferation, while at the same time intercellular adhesion was reduced (Wolf et al., 2010, 2013). Since IGF1 receptor expression decreases from basal to suprabasal cells, signaling is restricted to the basal cell compartment, where it keeps PKP1-mediated adhesion low and supports proliferation by activating its function in translation. As a consequence, PKP1-dependent hyper-adhesion is prevented (Tucker et al., 2014; Keil et al., 2016; Rietscher et al., 2016). Whereas PKP1 desmosome association is negatively regulated by insulin/IGF1, PKP3 phosphorylation at AGC-kinase consensus sites was not affected by insulin/IGF1 signaling (Fig. 1A), strongly suggesting that the IGF1 pathway does not regulate PKP3. In contrast, EGF signaling promoted PKP3 recruitment to the lateral plasma membranes. PKP3-dependent cell contacts can ensure tissue cohesion in the basal layer but at the same time allow for dynamic rearrangement of cells, as required during epidermal regeneration. Dynamic remodeling may be additionally facilitated by the lack of tricellular junctions, which are hotspots of epithelial tension (Higashi and Miller, 2017).

Here, we propose a model in which PKP3 in response to EGFR signaling initiates the formation of dynamic desmosomes in the basal layer of the epidermis (Fig. 8). At the same time, IGF1-mediated modification of PKP1 prevents the formation of PKP1-dependent hyper-adhesive desmosomes (Tucker et al., 2014; Keil et al., 2016). This results in stable, but dynamic, cell cohesion in the basal cells. When cells commit to differentiate and move to the spinous layer, attenuation of insulin/IGF1 signaling allows PKP1 membrane recruitment. Increased membrane associated PKP1 displaces PKP3 from lateral membranes and, together with 14-3-3σ/stratifin, facilitates PKP3 localization at tricellular contacts to promote barrier formation (Rietscher et al., 2018). Finally, we hypothesize that EGFR-dependent signaling in the stratum granulosum not only induces tight junction formation as shown by others (Rübsam et al., 2017), but also promotes epidermal sealing by directing PKP3 into the tricellular junctions.

Our data show how signaling pathways balance PKP isoform functions to regulate adhesive strength and allow desmosomes to adapt in a context-dependent manner to their environment. This implicates that desmosomes are active signaling hubs that receive signals (phosphorylation of PKP1 mediated by IGF1 and of PKP3 mediated by EGF) and transduce this information to modulate adhesive strength and control proliferation. Moreover, it indicates that PKP1 and PKP3 sense signals from different growth factors and interpret these signals in distinct ways that either promote (PKP3) or prevent (PKP1) their desmosomal localization. Taken together, our work demonstrates how insulin/IGF1 and EGF signaling cooperate to regulate the opposing functions of PKP1 and PKP3 in the basal layer of the epidermis to promote PKP3-dependent formation of dynamic desmosomes while at the same time preventing PKP1-dependent formation of Ca²⁺-independent stable desmosomes.

Human complementary DNAs (cDNAs) of RSK1 [pKH3-human RSK1, Addgene plasmid #13841, deposited by John Blenis (Richards et al., 2001)] and RSK2 [pWZL Neo Myr Flag RPSK6A3, Addgene plasmid #20627, deposited by William Hahn and Jean Zhao (Boehm et al., 2007)] were subcloned into pEGFP-C2 (Takara Bio, Nojihigashi, Japan), pVen2-HA-C2 and myr-HA. Human cDNAs of PKP3 were subcloned into pLVX-IRES-puro (Takara Bio) containing a C-terminal EGFP tag. Vectors for production of lentiviral particles pMD2.G and psPAX2 were Addgene plasmids #12259 and #12260, deposited by Didier Trono. To create site-specific non-phosphorylatable (S to A) as well as phospho-mimetic (S to E) mutations of PKP3, PCR-based site-directed mutagenesis was performed using oligonucleotides with the corresponding codons. FLAG-PKP3 and pVen1-FLAG-C2 constructs have been described previously (Wolf et al., 2006, 2010).

Primary antibodies and dilutions used for western blotting and immunofluorescence analysis are listed in Table S1. Secondary antibodies were donkey anti-mouse and anti-rabbit IgG conjugated to horseradish peroxidase and donkey anti-mouse, anti-guinea pig and anti-rabbit IgG conjugated to Alexa Fluor 488 or Cy3 (Jackson ImmunoResearch, West Grove, PA).

Immortalized keratinocytes from WT and PKP3-KO mice have been described (Keil et al., 2016). HEK293 and A431 cells were grown in Dulbecco's modified Eagle medium (DMEM) supplemented with 10% (v/v) fetal calf serum (FCS). To generate PKP3-KO keratinocytes expressing PKP3-WT-EGFP, PKP3-S134/135A-EGFP, PKP3-S134/135E-EGFP, PKP3-S256/257A-EGFP and PKP3-S256/257E-EGFP, HEK293 cells were co-transfected by CaPO₄ precipitation with pMD2.G, psPAX2 and pLVX-IRES-puro containing human PKP3-WT-EGFP, PKP3-S134/135A-EGFP, PKP3-S134/135E-EGFP, PKP3-S256/257A-EGFP or PKP3-S256/257E-EGFP. Lentiviral particles were purified 48 h after transfection using Lenti-X concentrator (Takara Bio) according to the manufacturer's protocol. PKP3-KO cells were incubated with the lentiviral particles for 24 h and subsequently used for the indicated experiments.

Mouse keratinocytes were grown on collagen I in LCM [DMEM/Ham's F12 medium containing 50 μM CaCl₂, 10% (v/v) Ca²⁺-free FCS, 1 mM sodium pyruvate, 1 mM glutamate, 0.18 mM adenine, 0.5 μg/ml hydrocortisone, 5 μg/ml insulin, 10 ng/ml EGF, 100 pM cholera toxin] at 32°C, 5% CO₂ and 90% humidity. To induce differentiation of keratinocytes, LCM was supplemented with 1.2 mM Ca²⁺ (HCM) for the indicated time points.

Transfection of plasmid DNA was performed using Xfect™ (Takara Bio) according to the manufacturer's protocol. Keratinocytes were grown in LCM for 24 h, incubated with the plasmid-Xfect mixture for 4 h and maintained in HCM for an additional 24 h.

siRNA pools (defined pools of 30 selected siRNAs) were generated by siTools Biotech (Martinsried, Germany) and siRNAs were transfected with Lipofectamine^® RNAiMax (Thermo Fisher Scientific, Waltham, MA) according to the manufacturer's protocol. Keratinocytes were transfected in suspension with non-targeting (nt), RSK1 or RSK2 directed siRNAs (2 pmol), were switched to HCM at 48 h after transfection and kept in HCM for another 24 h.

Keratinocytes grown for 24 h in HCM were starved in DMEM/Ham's F12 medium without any supplements (−GF) for 5 h and stimulated with growth factors (50 µg/ml insulin, 100 ng/ml EGF) for the indicated time points. A431 cells were serum starved for 24 h before stimulation with growth factors (50 µg/ml insulin, 100 ng/ml EGF). For application of inhibitors, cells were treated with the indicated inhibitor (Table S2) for 1 h before stimulation with EGF for 15 min.

For knockdown of RSKs, A431 cells were transfected with siRNA pools in suspension. The next day, cells were serum starved for 24 h and stimulated with 100 ng/ml EGF for 15 min.

To separate proteins under denaturing conditions sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) was performed. Proteins were transferred to nitrocellulose membranes using a semi-dry blotter. Membranes were blocked using 3% (w/v) skimmed milk/Tris-buffered saline with Tween 20 (TBST) or 3% (w/v) bovine serum albumin (BSA)/TBST and subsequently probed with the appropriate antibodies (Table S1). For analyzing phosphorylation, phospho-specific signals were detected. Then, the same membrane was washed in TBST (three times, each for 10 min), incubated in stripping buffer [0.2 M glycine, 0.05% Tween-20 (pH 2.5)] for 1 h, washed again in TBST (three times, each for 10 min), blocked using 3% (w/v) skimmed milk/TBST or 3% (w/v) BSA/TBST and probed with the non-phospho antibody. Chemiluminescence was detected via the Fusion-SL 3500.WL imaging system (Peqlab, Erlangen, Germany).

Total RNA was isolated from mouse keratinocytes and A431 cells grown for 3 days in DMEM/Ham's F12 or DMEM, respectively. Cells were homogenized in Trizol, and RNA was isolated by phenol/chloroform extraction and isopropanol precipitation. Reverse transcription was carried out using SuperScript^® II Reverse Transcriptase (Thermo Fisher Scientific). Real-time PCR was performed using the Maxima SYBR Green/ROX qPCR Master Mix (Thermo Fisher Scientific) and a Roche LightCycler 480 II Real Time PCR system, with the following PCR conditions: 95°C for 15 min followed by 40 cycles of 95°C for 15 s, 62°C for 15 s and 72°C for 20 s. Primer pair sequences are listed in Table S3.

For analysis of protein expression, keratinocytes were lysed in SDS buffer [20 mM Tris-HCl (pH 7.5), 2% SDS (supplemented with 1 μg/ml aprotinin, 1 μg/ml leupeptin, 1 mM Pefabloc, 5 mM NaF, 1 mM NaVO₃)] and centrifuged for 15 min at 13,000 g. Total protein concentration was determined using a BCA Protein Assay Kit (Thermo Fisher Scientific). Equal amounts of protein were separated by SDS-PAGE, transferred to nitrocellulose, probed with the appropriate antibodies and chemiluminescence was detected. α-Tubulin was used as loading control.

Keratinocytes and A431 cells were lysed in IP buffer [20 mM Tris-HCl (pH 7.5), 137 mM NaCl, 2 mM EDTA, 10% (v/v) glycerol, 1% (v/v) NP-40, 10 μg/ml aprotinin, 10 μg/ml leupeptin, 1 mM Pefabloc, 1 mM NaF and 1 mM NaVO₃]. Lysates were cleared by centrifugation for 15 min at 4°C and 13,000 g, and incubated with protein A agarose beads (Thermo Fisher Scientific) and anti-PKP3 antibody for 4 h at 4°C on an overhead rotator. Bound proteins were eluted in SDS-PAGE loading buffer, separated by SDS-PAGE and analyzed by western blotting.

For FLAG-IP, HEK293 cells were transfected with FLAG-PKP3-pcDNA3 in the presence or absence of RSK-myr-HA-pcDNA3. Cells were serum starved for 24 h and lysed in IP buffer containing 1 mM dithiothreitol (DTT). Lysates were incubated with anti-FLAG M2 affinity gel (Sigma-Aldrich, Taufkirchen, Germany). Bound proteins were solubilized in SDS-PAGE loading buffer, separated by SDS-PAGE and analyzed by western blotting.

Mouse keratinocytes grown on collagen I-coated coverslips and A431 cells grown on uncoated coverslips were fixed for 10 min in methanol at −20°C or for 20 min in 3.7% (w/v) formaldehyde in PBS at room temperature (RT), permeabilized in detergent buffer [100 mM PIPES (pH 6.9), 4 M glycerol, 2 mM EDTA, 1 mM EGTA, 0.5% (v/v) Triton X-100] for 15 min at RT and blocked in 1% (w/v) skimmed milk/PBS for 30 min at RT. Primary antibodies were diluted in blocking solution and incubated overnight at 4°C in a humid chamber. The next day, coverslips were briefly blocked in blocking solution and incubated for 1 h at RT with the fluorophore-conjugated secondary antibody. DNA was stained with Hoechst 33342 (Thermo Fisher Scientific). Coverslips were mounted in Mowiol. Images were taken on a Nikon Eclipse E600 microscope, with a charge-coupled device camera and a Plan APO 60×/1.40 NA oil objective, controlled with the NIS-Elements AR 4.12.00 software. ImageJ was used for image processing.

To determine the enrichment factors for PKP3 and DSP at lateral, and PKP3 at tricellular, contacts, fluorescence intensities were measured in segments of equal length [∼100 pixels (pxl)] and width (bicellular, 40 pxl; tricellular, 10 pxl) covering the cytoplasm as well as bicellular and tricellular contacts as illustrated in Fig. S4 (Rietscher et al., 2018). Each line scan results in a histogram depicting the fluorescence intensities along the bar [cytoplasm (green), bicellular (blue), tricellular (purple)]. The enrichment factors for PKP3 and DSP at bicellular contacts were calculated by dividing the mean junctional value (10 pxl length of the 100 pxl scan line) by the mean cytoplasmic value (both ends of a scan line, each 10 pxl length) for each line scan. To quantify PKP3 fluorescence intensities at tricellular contacts, the mean tricellular value (10 pxl length at the middle of the 100 pxl scan line) was divided by the mean bicellular value (one end of a scan line, 10 pxl length). All calculated enrichment factors are shown as boxplots displaying the first to third quartile; whiskers extend to the 5th and 95th percentile. The statistical significance of differences between the mean bicellular and tricellular values was determined using a one-way analysis of variance (ANOVA) followed by a Tukey's multiple comparison test.

For BiFC analysis, keratinocytes were co-transfected with the indicated pVen1 and pVen2 constructs. At 4 h after transfection, LCM was switched to HCM, and cells were incubated for 24 h before fixation in 3.7% (w/v) formaldehyde in PBS and immunostaining with FLAG and human influenza hemagglutinin (HA) tag-directed antibodies. BiFC images of cells expressing both FLAG- and HA-tagged fusion proteins were taken with identical exposure times (2 s) to enable a comparison of BiFC efficiencies. Image processing and the mean BiFC fluorescence intensities of individual transfected cells were determined using ImageJ.

For analysis of intercellular cohesion in combination with knockdown, WT and PKP3-KO keratinocytes transfected with non-targeting (nt), RSK1- or RSK2-directed siRNA pools were switched to HCM at 24 h after transfection and kept in HCM for another 24 h before incubation with 2.4 U/ml dispase II (Roche Diagnostics, Indianapolis, IN) in DMEM/Ham's F12 medium supplemented with 1.2 mM Ca²⁺ and 25 mM HEPES for 30 min at 37°C. After detachment, monolayers were kept for an additional 30 min in DMEM/Ham's F12 medium containing 1.2 mM Ca²⁺, 25 mM HEPES and 3.3 mM EGTA before submitting to mechanical stress on an orbital shaker at 750 rpm. For analyzing the effect of RSK inhibition, the RSK inhibitor LJH685 (50 µM) was applied together with dispase II for 1 h at 37°C. Floating monolayers were treated as above. Images were taken using a Sony DSC-H300 camera. For image processing and counting of fragments, the ImageJ tool ‘Cell Counter’ was used.

To identify the phosphorylation sites essential for PKP3 regulation by RSK we compared kinase-specific predictions for the RXXS motifs in PKP3 (Table S4) using GPS5.0 (http://gps.biocuckoo.org/, December 2019), NetPhos3.1 (http://www.cbs.dtu.dk/services/NetPhos/, December 2019) and Kinexus (http://www.phosphonet.ca/, December 2019). We took into consideration if a specific site had been detected in high-throughput screens for phosphorylated proteins as summarized at PhosphoSitePlus (https://www.phosphosite.org/proteinAction?id=3262&showAllSites=true, December 2019) and compared these data to the results of an in vitro phosphorylation assay on PKP3 peptides using purified RSK1.

Peptide chip experiments were performed essentially as described (Wolf et al., 2013). Peptide arrays (JPT Peptide Technologies, Berlin, Germany) containing overlapping (ten amino acids) 15 amino acid-long peptides of PKP3 were blocked overnight in 0.5 mg/ml PEG 20,000 and 200 µM ATP. Then, 1 U (1 nmol/min) of full-length RSK1 (Lot 880119J, Thermo Fisher Scientific) was incubated for 10 min at RT in 500 µl reaction buffer (5 mM MOPS, 2.5 mM β-glycerophosphate, 10 mM MgCl₂, 1 mM EGTA, 0.4 mM EDTA, 50 µM DTT, pH 7.2) containing 10 µM ATP to facilitate auto-phosphorylation with non-radioactive ATP. Subsequently, 100 µCi γ-³²P-ATP (6000 Ci/mmol, Perkin Elmer) was added, and the reaction mix was applied onto the peptide chip and incubated at 30°C for 2 h. To stop the reaction, the array was washed three times with aqueous solution containing 1% SDS, once with PBS containing 1% Triton X-100, three times with Millipore water and once with ethanol for 10 min each. Phosphorylated peptides were detected by autoradiography.

All statistical analyses were performed using GraphPad Prism 8.3. For two independent data sets, statistically significant differences were determined with a two-tailed Student's t-test. To compare more than two independent data sets with normal distribution, one-way ANOVA followed by a Tukey's multiple comparison test was used. Asterisks indicate statistical significance (*P<0.05, **P<0.01, ***P<0.001).

We thank Andrej Mun for excellent technical assistance, and Annemarie Jordan for her contribution in an initial phase of the project. We thank Prof. Mike Schutkowski for his support with the peptide chip experiments.

The peer review history is available online at https://jcs.biologists.org/lookup/doi/10.1242/jcs.238295.reviewer-comments.pdf