Regulation of hemidesmosome dynamics and cell signaling by integrin α6β4

Hemidesmosomes (HDs) play a crucial role in maintaining tissue integrity by means of integrin α6β4-mediated adhesion to laminin-332 in the underlying basement membrane. The importance of integrin α6β4 (referred to hereafter as α6β4) for stable cell–matrix adhesion is apparent from mice with targeted deletion of the genes encoding either integrin α6 (Itga6) (Georges-Labouesse et al., 1996) or β4 (Itgb4) (Dowling et al., 1996; van der Neut et al., 1996) and from patients, in which the absence of this integrin results in a devastating congenital skin disease called junctional epidermolysis bullosa (JEB) (Bardhan et al., 2020). JEB patients exhibit not only skin blistering, but also defects in the respiratory and gastrointestinal tract (Bardhan et al., 2020). In addition to mediating stable cell adhesion, there is evidence that HDs also contribute to cell proliferation and migration (Lipscomb and Mercurio, 2005), and play a role in skin stem cell homeostasis and ageing (Liu et al., 2019). In this Review, we summarize our current understanding of the posttranslational modifications of α6β4 and the roles these modifications play in the assembly and disassembly of HDs. We also discuss the role of tyrosine phosphorylation of integrin β4 in signal transduction processes and highlight the role of α6β4 in HDs in controlling cell mechanics and signaling pathways associated with focal adhesions (FAs).

Stratified and pseudostratified epithelia, including the epidermis and the airway epithelium, contain type I HDs, which consist of the transmembrane proteins α6β4, bullous pemphigoid antigen (BP) 180 (also known as BPAG2, COL17A1 or collagen type XVII) and the tetraspanin CD151 (Litjens et al., 2006; Walko et al., 2015). In addition, they contain the two cytoplasmic proteins plectin and BP230 (BPAG1e, an epidermal isoform of BPAG1, which is also known as dystonin or DST) (Fig. 1A). The latter two belong to the family of plakins, which are large modular proteins that connect cytoskeletal elements to each other and to junctional complexes (Sonnenberg and Liem, 2007). In type I HDs, plectin and BP230 mediate linkage of α6β4 and BP180 to the keratin intermediate filament (IF) system (Fig. 1B) (Bouameur et al., 2014; Geerts et al., 1999; Rezniczek et al., 1998). Notably, simple epithelia, such as that of the small intestine, contain alternative complexes, termed type II HDs, which consist of α6β4 and plectin but lack BP180 and BP230 (Stutzmann et al., 2000).

A distinguishing feature of type I HDs in electron micrographs is a tripartite structure with inner and outer electron-dense plaques (Fig. 1C). The inner plaque provides attachment sites for keratin filaments, while the outer plaque associates with anchoring filaments that traverse the lamina lucida to connect the HD to the lamina densa of the epidermal basement membrane. The ultrastructural features of HDs are largely unaffected by the absence of plectin or CD151 (Andra et al., 1997; Karamatic Crew et al., 2004; Wright et al., 2004). However, plectin deficiency results in a reduction of the number of keratin filaments associated with inner plaques (Andra et al., 1997; Nakamura et al., 2005). In the absence of either BP230 or BP180, the inner plaques are poorly developed or even completely absent (Groves et al., 2010; Guo et al., 1995; Jonkman et al., 1995).

Type II HDs have reduced association with keratin filaments, which renders these structures less stable and more dynamic than type I HDs, at least in cultured cells (Geuijen and Sonnenberg, 2002; Tsuruta et al., 2003). However, despite this reduced stability, the type II HD-mediated adhesion of intestinal epithelial cells to the underlying basement membrane is sufficiently strong enough to prevent their detachment when exposed to external mechanical forces from the movement of intraluminal contents, or during their migration up the crypt–villus axis (De Arcangelis et al., 2017; Krausova et al., 2021). Indeed, unlike the more structurally robust type I HDs, the more dynamic quality of type II HDs may enable cell migration. Nonetheless, whereas type I and type II HDs have clear differences in composition and dynamics, both play essential roles in cell adhesion and tissue integrity.

In the epidermis, the expression of α6β4 is transcriptionally downregulated when basal cells commit to terminal differentiation and move upward to form the suprabasal cell layers (Tennenbaum et al., 1996). During wound healing and in psoriatic lesions, the differentiation program of keratinocytes is dysregulated, and expression of α6β4 is no longer restricted to the basal cells and is found in the suprabasal layers (Hertle et al., 1992). Furthermore, suprabasal and enhanced expression of α6β4 have been observed in squamous cell carcinomas and are linked to malignant progression (Mercurio and Rabinovitz, 2001). Multiple processes, including epigenetic mechanisms and microRNA-mediated gene regulation have been reported to regulate the expression of genes encoding the integrin subunits α6 (ITGA6) and β4 (ITGB4) (Ferraro et al., 2014; Yang et al., 2009). Indeed, an array of transcriptional activators and repressors are involved in HD formation, either through the direct or indirect regulation of ITGA6 and ITGB4 (Carroll et al., 2006; Gaudreault et al., 2008; Niu et al., 2015; Oommen et al., 2012).

In addition to the normal regulation of transcription and translation of mRNAs, expression of the α6 and β4 subunits is regulated by alternative splicing in different tissues and during development. Alternative splicing of ITGA6 produces mRNAs that encode two cytoplasmic variants of the α6 subunit: α6A and α6B (Cooper et al., 1991; Hogervorst et al., 1991). When cells gain specialized functions during embryonic and fetal development, α6B often becomes replaced by α6A (Hogervorst et al., 1993; Tamura et al., 1991). However, studies with genetically modified mice have revealed no essential function for α6A during early development or in the adult (Gimond et al., 1998). Furthermore, it has been shown that the formation of HDs in adult skin is unaffected by the substitution of α6A for α6B. In contrast, changes in the expression of α6A and α6B have been linked to altered signaling, proliferation and cell adhesion to laminin in several tumors (Goel et al., 2014; Tennenbaum et al., 1995).

Five cytoplasmic variants have been reported for the β4 subunit. Two of these variants differ from β4A, the major variant expressed by keratinocytes, in that they contain insertions of 53 (β4B) or 70 (β4C) amino acids in the connecting segment (CS) that separates the two pairs of fibronectin type III (FnIII) domains in the cytoplasmic tail (Fig. 1A) (Hogervorst et al., 1990; Suzuki and Naitoh, 1990; Tamura et al., 1990). A fourth variant (β4D) lacks a region of seven amino acids in the FnIII-4 domain, whereas the C-terminal 902 amino acids are replaced by a stretch of 114 unique amino acids in the β4E variant (van Leusden et al., 1997). The latter variant is produced by partial retention of an intron during RNA splicing. As yet, the functions of all of these variants are not clear.

The assembly of HDs has been extensively studied in cultured human keratinocytes. HD formation is initiated by the binding of the β4 subunit to plectin, which involves two distinct binding sites (Fig. 2A) (Rezniczek et al., 1998; Schaapveld et al., 1998). A primary binding site comprising the first pair of FnIII domains and the N-terminal region of the CS of β4 specifically interacts with the N-terminal actin-binding domain (ABD) of plectin (Fig. 2B) (de Pereda et al., 2009; Geerts et al., 1999). This interaction involves electrostatic interactions between two positively charged residues (R1225 and R1281) in the FnIII-2 domain of β4 and two negatively charged amino acids (D151 and E95) in the plectin ABD. This association prevents plectin from interacting with the actin cytoskeleton when it is associated with HDs (de Pereda et al., 2009; Geerts et al., 1999). The importance of the interaction between the two acidic residues in the ABD and the basic residues in the second FnIII domain of β4 for the localization of plectin at HDs has been demonstrated in keratinocytes and JEB patients carrying missense mutations that affect R1225 and R1281 (Chung and Uitto, 2010; Koster et al., 2001). The C-terminal region of the CS and a region of β4 near its C terminus form a secondary binding site for the plakin domain of plectin adjacent to the ABD (Fig. 2A) (Koster et al., 2004; Rezniczek et al., 1998). CD151 associates with α6β4–plectin complexes through interaction with the α6 subunit, and this may facilitate the clustering of the resulting complexes (Sterk et al., 2000). These complexes are further stabilized by keratin filaments that bind to the C terminus of plectin (Bouameur et al., 2014; Seltmann et al., 2013). BP180 is subsequently recruited by interactions with both α6β4 and plectin (Koster et al., 2003). Finally, BP230 is incorporated into HDs through interactions with β4 and BP180, after which bundles of keratin filaments become abundantly associated with the newly established HDs through interaction with both plectin and BP230. The interaction of β4 with BP230 involves the second pair of FnIII domains (FnIII-3 and FnIII-4), which interact with the epidermal-isoform-specific N-terminal region of BP230, and is associated with a conformational change in β4 that stabilizes the interaction (Fig. 2C) (Koster et al., 2003; Manso et al., 2019).

HDs are dynamic structures that disassemble when epithelial cells migrate, for example during wound repair and cancer invasion, and reassemble when migration has been terminated. HDs are also disassembled during mitotic cell rounding, when the overall cell–substrate contact area is decreased and cells are only attached to the substratum through a fine network of retraction fibers (Geuijen and Sonnenberg, 2002). Retraction fibers that contain α6β4 adhesions are also formed at the rear of migrating carcinoma cells (Rabinovitz and Mercurio, 1997) and keratinocytes (Geuijen and Sonnenberg, 2002). Like the mitotic retraction fibers, these retraction fibers contain laminin-332-ligated α6β4 but are devoid of HD components. Furthermore, they contain tetraspanin (CD151), as well as β1 and β5 integrins, but no FA proteins (Geuijen and Sonnenberg, 2002; Yamada et al., 2013). Exactly how the functional state of HDs, and hence the dynamics of cell adhesion, is controlled is not known, but phosphorylation and Ca²⁺ have been recognized as key regulators.

The integrin β4 subunit contains several serine and threonine residues that are phosphorylated upon activation of receptor tyrosine kinases (RTKs) and reduce the association of β4 with plectin (Margadant et al., 2008; Rabinovitz et al., 2004). Phosphorylation of T1736 in the C-terminal region of the β4 cytoplasmic domain by ribosomal S6 kinase 1 (RSK1, also known as RPS6KA1) or protein kinase D 2 (PKD2, also known as PRKD2) causes dissociation of the plakin domain of plectin from β4 (Fig. 3A) (Frijns et al., 2012; te Molder and Sonnenberg, 2015). S1356 and S1364, both located in the CS, are phosphorylated by extracellular signal-regulated kinases 1 and 2 (ERK1/2; also known as MAPK3 and MAPK1, respectively) and by RSK1 and RSK2 (also known as RPS6KA3), respectively (Frijns et al., 2010). This phosphorylation occurs downstream of epidermal growth factor receptor (EGFR) or protein kinase C (PKC) activation and prevents the interaction of the ABD of plectin with β4 (Fig. 3A). Intriguingly, these serine residues are located outside the ABD-binding site on the β4 cytoplasmic domain, and the mechanism by which their phosphorylation inhibits the binding of plectin to β4 is unknown. One potential mechanism is that the two negatively charged phosphoserines compete with the two acidic amino acids in the plectin ABD for binding to the positively charged residues (R1225 and R1281) in the FnIII-2 domain of β4 (Fig. 3B) (Frijns et al., 2012; Litjens et al., 2006). Notably, during mitotic cell rounding when the activity of the ERK1/2 is diminished, phosphorylation of β4 at S1356 is then mediated by another proline-directed kinase, most likely Cdk1, which is activated at the G2-M phase transition (Frijns et al., 2010). In line with a role for phosphorylation of β4 in the disassembly of HDs, phosphorylated β4 (S1356) has also been detected in retraction fibers that are formed during rear-end contraction (Kashyap et al., 2011).

In addition to the phosphorylation-dependent regulation of β4–plectin interactions, other phosphorylation events downstream of RTK activation can alter HD dynamics (Bouameur et al., 2013; Germain et al., 2009). Phosphorylation of β4 at S1424 may cause the disassembly of type I HDs by preventing the interaction of β4 with BP180 and BP230 (Germain et al., 2009). Furthermore, plectin is phosphorylated at S4642 downstream of the ERK/MAP kinase-interacting serine/threonine-protein kinase 2 (MNK2) pathway, which reduces plectin binding to keratin filaments (Bouameur et al., 2013), and BP180 is phosphorylated by PKC, leading to its translocation away from HDs (Kitajima et al., 1999). PKC has also been implicated in the phosphorylation of the α6A subunit; however, there is presently no evidence that the phosphorylation of this subunit has an effect on HD dynamics (Hogervorst et al., 1993). Finally, the interaction between BP230 and β4 might be inhibited directly by phosphorylation: phosphomimetic substitutions of T39 or S46 in the unique N-terminal region of BP230 severely reduce its affinity for β4 (Manso et al., 2019). Yet, whether BP230 is phosphorylated during HD disassembly is unknown.

It has long been established that the formation of HDs requires Ca²⁺ and that there is a direct relationship between the number of mature HDs and the concentration of Ca²⁺ (Trinkaus-Randall and Gipson, 1984). Similarly, immortalized human keratinocytes assemble more robust HDs upon culture in Ca²⁺-containing medium (Schaapveld et al., 1998). Because the calmodulin antagonist W7 inhibits HD formation, it has been suggested that Ca²⁺ mediates this effect by a calmodulin-regulated mechanism (Trinkaus-Randall and Gipson, 1984). However, the effects of Ca²⁺ on HD formation are complex, and both positive and negative effects have been reported. There is also evidence that calmodulin activated by Ca²⁺ causes the disassembly of HDs in primary mouse keratinocytes (Kostan et al., 2009). However, calmodulin does not mediate this effect by controlling enzyme activity, but rather by direct binding to the alternative sequences preceding the ABD in the plectin 1a isoform and subsequent disruption of the plectin–integrin β4 complex by steric repulsion (Song et al., 2015) (Fig. 3). Additionally, Ca²⁺ may induce HD disassembly by triggering calpain- and caspase-dependent proteolysis of β4 and/or plectin (Potts et al., 1994; Walko et al., 2011; Werner et al., 2007). However, Ca²⁺ has also been shown to promote HD stability through calcineurin-dependent inhibition of β4 phosphorylation (Kashyap and Rabinovitz, 2012; te Molder and Sonnenberg, 2015). The precise cause for the differential effects of Ca²⁺ on the formation of HDs is not known but may be related to differences in the responsiveness and sensitivity of human and murine keratinocytes to Ca²⁺-induced differentiation.

In addition to growth factor receptor-induced serine and threonine phosphorylation, β4 is also phosphorylated on several tyrosine residues. Y1257 in the FnIII-2 domain, Y1422 and Y1440 in the CS, as well as Y1494 and Y1526 in the FnIII-3 domain and Y1642 in the FnIII-4 domain of the β4 cytoplasmic region have been identified as phosphorylation-dependent docking sites for a variety of proteins (Fig. 3C) (Giancotti, 2007; Margadant et al., 2008; Mercurio and Rabinovitz, 2001). These include Src homology 2 (SH2) domain-containing protein tyrosine phosphatase (Shp2, also known as PTPN11), as well as the adaptor proteins Src homology and collagen transforming protein 1 (Shc1) and insulin receptor substrate 1 (IRS1), which might link α6β4 to the ERK/mitogen-activated protein kinase (MAPK) and phosphoinositide 3-kinase (PI3K) signaling pathways that stimulate cell proliferation and survival, respectively. The evidence for phosphorylation-dependent binding of adaptor proteins to tyrosine residues in the FnIII domains primarily comes from overexpression experiments and/or far-western analysis, and therefore the biological significance of these interactions is unclear. In fact, these sites might only be recognized when the FnIII domains are partially or completely unfolded. Structural analysis has confirmed that Y1494, Y1526 and Y1642 are not accessible for protein kinases due to steric hindrance (Alonso-García et al., 2015). However, there are no structural constraints for kinases to phosphorylate or adapter proteins to bind to Y1422 and Y1440 in the CS. Recently, peptide pulldown assays combined with mass spectrometry have identified phospholipase C γ1 (PLCγ1, also known as PLCG1), but not Shc1, as a candidate binding partner for Y1422 and Y1440 (te Molder et al., 2021). Because this interaction could not be confirmed in cells, its significance for downstream signaling remains uncertain. Furthermore, the Src family kinases (SFKs) have been implicated in the phosphorylation of Y1422 and Y1440 downstream of EGFR activation (Gagnoux-Palacios et al., 2003; Mariotti et al., 2001), but siRNA-depletion of individual SFKs or combinations of these kinases does not abrogate β4 tyrosine phosphorylation (te Molder et al., 2021). It is not clear which kinase phosphorylates the tyrosines in β4, but given the fact that tyrosine phosphorylation of β4 is primarily found in cells that overexpress RTKs and express α6β4 at high levels, it is plausible that EGFR can directly phosphorylate the β4 subunit due to their close proximity.

Several other RTKs, including Met, Ron (also known as MST1R) and ErbB2, have been shown to associate with β4 and phosphorylate β4 tyrosine residues (Gambaletta et al., 2000; Santoro et al., 2003; Trusolino et al., 2001). Association of the Ron receptor with β4 has been suggested to be mediated by 14-3-3 proteins, which simultaneously bind to Ron and β4 when they are phosphorylated at specific serine residues (Santoro et al., 2003). 14-3-3 binding involves the phosphorylation of S1356, S1360 and S1364 in β4 and may mediate the disassembly of HDs and relocation of α6β4 to other parts of the cell (Santoro et al., 2003). Alternatively, β4 may contribute to pro-tumorigenic signaling via focal adhesion kinase 1 (FAK1, also known as PTK2) (Abdel-Ghany et al., 2002; Lakshmanan et al., 2016; Ni et al., 2018; Seo et al., 2017; Tai et al., 2015). FAK1 activation is observed when α6β4 is ligated to the Ca²⁺-activated chloride channel regulator 1 (mCLCA1) and is believed to involve a direct interaction between a phosphotyrosine residue in β4 (pY1494) and FAK1 (Ni et al., 2018). Another region of β4 that contains the Calx-β domain and is proximal to the plasma membrane has been found to interact with an 11-amino-acid sequence in the linker region between the FERM and kinase domains of FAK1 (Tai et al., 2015). Furthermore, β4 has been implicated in the activation of FAK1 through its binding to tensin-4 (TNS4) (Seo et al., 2017). Thus, β4 might contribute to FAK1 activation through both direct and indirect interactions with the kinase. However, recent findings suggest that the activity of FAK1 is regulated by an effect of β4 on the mechanical properties of cells (Wang et al., 2020). Further studies will be required to determine the importance of β4–FAK1 interactions in controlling FAK1 activation.

Endocytic trafficking of integrins is an important mechanism through which the availability of integrins at the cell surface can be regulated. Additionally, it allows cells to spatiotemporally control the formation and turnover of adhesion complexes such as FAs, which are important for cell migration (Paul et al., 2015). Integrin trafficking might also play a role in regulating HD dynamics. β4 lacks the typical tyrosine-based internalization signals that mediate clathrin-dependent sorting of β1 and β3 integrins in the endocytotic and secretory pathways (Paul et al., 2015). However, clathrin-dependent endocytosis of α6β4 might be facilitated by complex formation with CD151. This protein harbors a YXXɸ motif (where X indicates any amino acid and ɸ indicates a hydrophobic residue), which is recognized by the µ2 subunit (AP2M1) of the clathrin adaptor complex AP-2 (Sincock et al., 1999). Although CD151 plays a role in α6β1 endocytosis, there is currently no evidence that it participates in α6β4 internalization (Liu et al., 2007).

The evidence suggests that α6β4 follows an Arf6- and caveolae-dependent pathway of internalization (Osmani et al., 2018). In addition, caveolae have a critical role in HD remodeling independent of their role in endocytosis by acting as membrane reservoirs that flatten out in response to mechanical stress (Osmani et al., 2018). The flattened membrane is then populated by α6β4 either by lateral diffusion through the plasma membrane or through Arf6-dependent vesicular trafficking. Additionally, it has been shown that endocytic β4 trafficking in mouse keratinocytes is triggered by phosphorylation at S1354 and S1362 (corresponding to S1356 and S1360 of human β4) and depends on Rab5- and Rab11-mediated transport along microtubules (Seltmann et al., 2015; Yoon et al., 2005). Microtubules are targeted and captured at the plasma membrane by the cortical microtubule stabilizing complex (CMSC), which is localized in close proximity to FAs and plays a role in their dynamics (Chen et al., 2018). Although CMSCs are also found in the vicinity of HDs, these complexes do not appear to contribute to the formation and/or disassembly of HDs (Hotta et al., 2010; te Molder et al., 2020).

In breast cancer cells, the surface levels of β4 are regulated by an endocytic pathway that involves the adaptor protein arrestin domain-containing protein 3 (ARRDC3) (Draheim et al., 2010). This protein interacts with S1424 of β4, one of the phosphorylated residues involved in the disassembly of HDs, and targets the integrin for ubiquitylation and subsequent degradation by the proteasome (Draheim et al., 2010). Finally, specific factors along the biosynthetic pathway play a role in the spatiotemporal regulation of HDs. For example, Lgl2 (also known as Llgl2), which controls the basolateral targeting of de novo synthesized α6β4, is required for the formation of HDs in zebrafish (Sonawane et al., 2005). It is likely that many different trafficking pathways have vital roles in HD regulation by regulating the cell-surface levels and/or the subcellular distribution of α6β4.

CD151 not only forms stable complexes with α6β4, but can also associate with other laminin-binding integrins, including α3β1 (Sterk et al., 2002; Yauch et al., 1998). α6β4 clusters together with α3β1–CD151 complexes at the basal surface of keratinocytes in pre-HD adhesions (Litjens et al., 2006; Sterk et al., 2000). Recent data indicate that these pre-HD adhesions also contain plectin and that neither α3β1 nor CD151 are required for mature HD formation in cultured keratinocytes (te Molder et al., 2019). However, the absence of either CD151 or the α3 subunit has been associated with skin fragility in humans, and loss of α3 induces mild skin abnormalities in neonatal mice (DiPersio et al., 1997; Has et al., 2012; Karamatic Crew et al., 2004; Wright et al., 2004).

A defining feature of tetraspanins is their ability to establish associations with a variety of partner molecules and to cluster into specialized regions of the plasma membrane called tetraspanin-enriched microdomains (TEMs) (Hemler, 2005). In addition, many tetraspanins, including CD151, are palmitoylated at membrane-proximal cysteines. The integrin subunits α3, α6 and β4 are also palmitoylated (see Box 1), which contributes to their incorporation and clustering into CD151 TEMs (Fig. 4). CD151 strengthens α3β1-, but not α6β4-mediated adhesion (te Molder et al., 2019). It is noteworthy that in the CD151 TEMs, α3β1 is not associated with actin stress fibers, probably because the lipid composition of these TEMs hinders a stable interaction between α3β1 and the cytoskeletal protein talin.

Although CD151 does not strengthen α6β4-mediated cell adhesion and has no clear role in the formation of HDs, it might be important in tissue remodeling during processes such as wound healing, when cells have to dismantle their type I HDs in order to migrate (Wright et al., 2004). CD151-mediated co-clustering of α6β4 and α3β1 in the pre-HD adhesions might enable cells to withstand external forces during migration, thereby facilitating an efficient and effective formation of type I HDs when migration has been completed. CD151 might also contribute to the regulation of α6β4 function by stimulating its recycling and association with RTKs. Furthermore, there is evidence that CD151 associates with PKC, and thus might play a role in PKC-dependent phosphorylation of β4 (Charrin et al., 2014).

Laminin-332 is an important regulator of dermal–epidermal adhesion and plays a critical role in the formation and turnover of HDs (Bardhan et al., 2020; Geuijen and Sonnenberg, 2002; Litjens et al., 2006; Ryan et al., 1999). Laminin-332 is deposited by keratinocytes in its unprocessed form and is then proteolytically processed to yield mature laminin-332 (Marinkovich et al., 1992). Although in carcinoma cells binding of α6β4 to mature laminin-332 may be favored over unprocessed laminin-332, in cultured keratinocytes, both ligands can promote the clustering of α6β4 and induce the formation of HDs (Fig. 4) (Baudoin et al., 2005; Goldfinger et al., 1998). However, in three-dimensional organotypic cell cultures, the formation of HDs that resemble those formed in vivo (i.e. HDs that exhibit electron-dense plaques and anchoring filaments) depends on processed laminin-332 (Baudoin et al., 2005). Thus, while both processed and unprocessed laminin-332 can facilitate the clustering of HD components, the formation of mature HDs relies on the processing of laminin-332.

A role for plectin in the clustering of α6β4 has been revealed by studies utilizing a chimeric protein consisting of the extracellular and transmembrane domains of interleukin 2 receptor fused to the β4 cytoplasmic domain (Geuijen and Sonnenberg, 2002; Nievers et al., 2000, 1998). This receptor, which cannot associate with either laminin-332 or CD151, still assembles into large clusters at the basal surface of cells, which resemble type II HDs. When this chimeric receptor can no longer associate with plectin through the introduction of a mutation in the β4 cytoplasmic domain or by expressing the chimeric receptor in keratinocytes that lack plectin, clustering of the chimeric protein is abrogated (Nievers et al., 2000). Plectin may promote α6β4 clustering through the formation of tetramers and oligomers of anti-parallel dimers via a lateral association of its central rod domains (Fig. 4) (Walko et al., 2015). Interestingly, yeast two-hybrid studies also suggest that the β4 cytoplasmic domain may dimerize at a region in the CS (Nievers et al., 2000). Interaction of these β4 dimers with dimeric plectin may then lead to the formation of large protein clusters. In conclusion, cells have multiple mechanisms in place to cluster α6β4 and thereby initiate HD assembly and reinforce adhesion strength.

In addition to mediating firm adhesion, α6β4 has been implicated in cell migration, which is driven by dynamic changes in the actin cytoskeleton under the control of Rho family GTPases (Lipscomb and Mercurio, 2005; Wilhelmsen et al., 2006). RhoA contributes to the deposition of laminin-332 by facilitating integrin α2β1-mediated cell spreading on collagen (Nguyen et al., 2000), while Rac1 contributes to migration on the deposited laminin-332 by promoting the formation of lamellipodia and by directing integrin α3β1 to these structures (Choma et al., 2004; Nguyen et al., 2000). Many studies have demonstrated an important role of α6β4 in stimulating Rac1 activation. For example, α6β4 ligation to laminin-332 supports growth factor-induced keratinocyte migration and tumor survival through activation of Rac1 (Zahir et al., 2003). α6β4 also stimulates PI3K-dependent activation of Rac1 in the absence of growth factors by allowing cells to adhere to either substrate-bound laminin or antibodies against the integrin (Mercurio and Rabinovitz, 2001; Shaw, 2001). α6β4 ligation may promote the activation of PI3K through inducing the phosphorylation of the signaling adaptors IRS1 and IRS2 (referred to collectively as IRS1/2; Fig. 3C); however, this may still depend on RTKs (Shaw, 2001). Indeed, integrin-mediated adhesion to the extracellular matrix (ECM) can lead to the activation of RTKs in the absence of soluble growth factor ligands (Marcoux and Vuori, 2003; Moro et al., 1998). Integrins may facilitate ligand-independent activation of RTKs by promoting the clustering of RTKs through lateral associations (Yamada and Even-Ram, 2002). In addition to its ability to support Rac1 activation, α6β4 has been implicated in regulation of RhoA activity (Mercurio and Rabinovitz, 2001). MDA-MB-435 carcinoma cells grown on laminin-111 are unable to form large lamellae in response to lysophosphatidic acid (LPA) but form these structures after α6β4 expression; α6β4 mediates this effect by suppressing the concentration of cAMP in these cells (Mercurio and Rabinovitz, 2001).

Recently, it has been shown that the β4 subunit in type I HDs decreases cell contractility through the inhibition of RhoA and RhoA-associated kinase (ROCK)-dependent myosin light chain (MLC) phosphorylation and FAK–PI3K mechanosignaling pathways downstream of FAs (Fig. 5) (Wang et al., 2020). HDs mediate this inhibitory effect because of their proximity to FAs and the ability of plectin to mechanically couple filaments of keratin attached at HDs to F-actin-anchored at FAs. Consistent with an important role of plectin in the suppression of RhoA activity, plectin-deficient keratinocytes exhibit increased RhoA-mediated cell contractility (Andra et al., 1998; Wang et al., 2020). It is suspected that the inhibitory effect on contractile responses is stronger as more keratin filaments are associated with HDs, and therefore type II HDs are less efficient in decreasing RhoA activity than type I HDs. Notably, HD-mediated inhibition of RhoA activity has been observed in keratinocytes cultured in medium containing high Ca²⁺ levels (Wang et al., 2020). In this medium, undifferentiated keratinocytes that are attached to the substratum assemble numerous, larger HDs and display reduced proliferation and migration, while descendants of these cells, which have lost contact with the substratum, undergo terminal differentiation (Wang et al., 2020). Although these observations indicate that α6β4 clustering can regulate the actin cytoskeleton by modulating Rho GTPase activity downstream of RTKs, there is also evidence that α6β4 can dynamically associate with F-actin and influence carcinoma cell migration through the stabilization of a polarized lamellipodium (Gipson et al., 1993; Lipscomb and Mercurio, 2005; Santoro et al., 2003). It is not known whether the association of α6β4 with these actin-rich sites requires plectin, and thus the formation of type II HDs, but in light of present knowledge we suspect that this is indeed the case.

In addition to a role for HDs in regulating RhoA-induced contractility, a recent study suggests that the formation of type II HDs in MCF10A cells also depends on RhoA-mediated mechanical force generation (Fujiwara et al., 2018). In these cells, the knockdown of Solo (also known as ARHGEF40), a RhoA-specific guanine-nucleotide-exchange factor that associates with keratin filaments and β4, suppresses HD formation. This suggests that RhoA activity may increase HD assembly in MCF10A cells (Fujiwara et al., 2018). Notably, RhoA is not necessary for the formation of type I HDs in keratinocytes, because no abnormalities in the structure of HDs are observed in mice with an epidermal-targeted deletion of RhoA (Jackson et al., 2011). It is not clear why type I and II HD formation would differ in their requirement for RhoA; however, one possible explanation is that BP180, which is present in type I HDs but absent from type II HDs, renders the requirement for RhoA superfluous. In support of this idea, BP180 has been shown to facilitate traction forces by stabilizing actin bundles through its binding to α-actinin-4 (Hiroyasu et al., 2016). Furthermore, because keratins are important for the assembly and stability of HDs and Solo is required for proper organization of the keratin network, the effects of Solo knockdown on the formation of HDs might also be mediated by a diminished and/or altered association of HDs with keratin (Seltmann et al., 2013). Another protein that associates with actin structures and regulates HD formation is flightless (FLII), a conserved member of the gelsolin family of actin regulators (Kopecki et al., 2009). Although targeted deletion of this protein in mice does not affect the structure of HDs, its overexpression is associated with an altered appearance and reduced number of HDs. The mechanism by which flightless regulates HD formation is not known, but may be related to the effects of altered actin remodeling on the keratin network (Kopecki et al., 2009).

In contrast to the polarity of actin filaments and microtubules, keratin filaments assemble into non-polar structures. Keratin filaments are also highly elastic and do not associate with myosin motors that can produce cell contraction (Yoon and Leube, 2019). Notable exceptions are keratin 6a (KRT6A) and keratin 6b (KRT6B), which interact with myosin IIA and may play a role in stabilizing myosin IIA protein levels (Wang et al., 2018). Some evidence suggests that myosin IIA can also bind to filaments formed by keratin 8 (KRT8) and keratin 18 (KRT18) (Kwan et al., 2015); however, a role of the associated myosin IIA molecules in generating tension in the keratin network is unlikely, because of the biomechanical properties of IFs. Since HDs connect to keratin filaments, their role in mechanosensing and mechanotransduction is not immediately apparent. However, because the keratin IFs and the actin cytoskeleton form an intricate network of proteins interconnected through plectin, keratin filaments that are anchored at HDs can be pre-stressed by actomyosin-generated tension (Fig. 5). The high elasticity of keratin filaments necessitates that HDs and FAs are in close proximity to each other, in order for the actomyosin contractile machinery to generate sufficient tension on HD-anchored keratin filaments and to enable HDs to sense and respond to matrix rigidity. However, as discussed above, a further increase in force generation by the actomyosin cytoskeleton can be counterbalanced by pre-stressed keratin filaments, leading to the inhibition of FA signaling (Wang et al., 2020).

In the Caenorhabditis elegans hypodermis, HD-like junctions respond to external tension by recruiting and activating a signaling pathway involving G-protein-coupled receptor kinase-interactor-1 (GIT-1), PAK-interacting exchange factor-1 (PIX-1), Rac1 (CED-10) and p21-Rac-activated kinase-1 (PAK-1) that promotes junction maturation (Zhang et al., 2011). The maturation of HD-like junctions is important because it enables the hypodermal cells to resist mechanical stress during muscle activity. In human keratinocytes, a homologous signaling pathway functions in controlling FA function and dynamics but has not been found to be associated with HDs (Hiroyasu et al., 2017). Additionally, it has been suggested that VAB-10A, the sole BP230 and plectin homolog in C. elegans (Zhang and Labouesse, 2010), can serve as a force sensor in HDs and play a role in mechanotransduction, similar to the role of talin at FAs (Goult et al., 2018). Plectin contains nine spectrin repeats (SR1–SR9) and a Src homology 3 (SH3) domain that forms an intramolecular autoinhibitory interaction with SR4 (Fig. 1A) (Ortega et al., 2011, 2016). Mechanical deformation of SR4 could relieve this autoinhibitory interaction and enable plectin to serve as a mechanosensitive hub (Daday et al., 2017; Suman et al., 2019). Most SH3 domains bind to poly-Pro motifs; yet, the SH3 domain of plectin lacks the canonical poly-Pro-binding site and may therefore recognize different, still unknown, ligands (Ortega et al., 2011). Furthermore, force-induced unfolding of the β4 cytoplasmic domain may expose phosphorylation sites that promote biochemical signaling events. In this manner, it is feasible that the tyrosine residues in the FnIII-3 and FNIII-4 repeats of the β4 cytoplasmic tail, which are normally not accessible to kinases, as described above, become phosphorylated. Nonetheless, it is unclear whether tension in keratin filaments induced by the actomyosin cytoskeleton is sufficiently high to result in unfolding of plectin and β4.

In the initial decades following the discovery that α6β4 is a component of HDs, research primarily focused on the interaction of this protein with other HD components and the regulation of HD dynamics by growth factor receptors. Subsequently, the role of α6β4 in inherited skin diseases and cancer has been the focal point of interest and the subject of intense research. These studies have established an important role for this integrin in the formation of HDs and maintenance of a stable attachment of the epidermis to the dermis. They have also revealed that signaling by α6β4 cooperates with RTKs to regulate cell proliferation and migration. This cooperation, which is believed to occur outside HDs, depends on tyrosine phosphorylation of β4 by RTKs and subsequent binding of adaptor proteins, thus linking the integrin to signaling pathways that regulate cell proliferation, migration and survival. However, in the majority of the related studies, significant tyrosine phosphorylation of β4 has only been observed in cells that overexpress the respective RTK and not in cells expressing basal levels of RTKs. Therefore, its contribution to the signaling output of RTKs is likely negligible. In an alternative way, α6β4 might impact signaling by reducing the tension exerted by the actomyosin cytoskeleton through FAs. Two signaling pathways downstream of FAs that are regulated by α6β4 are the RhoA–ROCK–MLC and FAK–PI3K signaling pathways.

Finally, it should be noted that an understanding of α6β4 regulation needs to be placed in the context of multiple RTKs and adhesion complexes formed by other integrins, which can crosstalk extensively and thereby positively or negatively affect each other's function. A better understanding of how the balance between these complexes is established and how they influence oncogenic signaling pathways will be necessary to fully understand the role of laminin-binding integrins in cell adhesion, and in tumor development and progression. Furthermore, most of our knowledge regarding the laminin-binding integrin-containing adhesion complexes comes from cell culture assays, and still very little is known about their dynamics in vivo.

We apologize to the authors of papers that could not be cited owing to space restrictions. We thank Coert Margadant, Katelyn Richards, Kevin Wilhelmsen and Roy Zent for discussion and critical reading of the manuscript. We also would like to thank Hans Janssen (The Netherlands Cancer Institute) for preparing the electron micrograph image for Fig. 1C. J.M.d.P. acknowledges institutional support funded by the Programa de Apoyo a Planes Estratégicos de Investigación de Estructuras de Investigación de Excelencia of the Junta de Castilla y León (CLC-2017-01) and the European Regional Development Fund.