• Open Access

The making of hemidesmosome structures in vivo



Hemidesmosomes are evolutionarily conserved attachment complexes linked to intermediate filaments that connect epithelial cells to the extracellular matrix. They provide tissue integrity and resistance to mechanical forces. Alterations in hemidesmosome structures are responsible for skin blistering, carcinoma invasion, and wound-healing defects. Valuable information about hemidesmosome assembly and disassembly has been obtained from in vitro cell culture studies. However, how these processes take place in vivo still remains elusive. Here, we discuss recent data about the formation and reorganization of hemidesmosomes in several in vivo model systems, particularly zebrafish and Caenorhabditis elegans, focusing on various factors affecting their dynamics. Mechanisms found in different organisms reveal that hemidesmosome formation and maintenance in vivo are carefully controlled by ECM protein folding, ECM-receptor expression and trafficking, and by post-translational modification of hemidesmosome components. These findings validate and extend the in vitro studies, and shed light on our understanding about hemidesmosomes across species. Developmental Dynamics 239:1465–1476, 2010. © 2010 Wiley-Liss, Inc.


Epithelial attachment structures are crucial for maintaining tissue integrity and resisting tension in live organisms. One type of such adhesive junctions, the hemidesmosome, connects intermediate filaments to the extracellular matrix through trans-membrane attachment complexes (Green and Jones,1996). Hemidesmosomes are found in epithelial tissues, in particular those exposed to mechanical stress such as the skin, oesophagus, and intestine. Genetic and autoimmune diseases altering their composition affect tissue stability, and cause, for instance, the human skin blistering disease termed Epidermolysis Bullosa (Pulkkinen and Uitto,1999). Hemidesmosomes are also involved in cell migration processes, such as wound healing and carcinoma invasion (Mercurio et al.,2001; Lipscomb and Mercurio,2005).

Most of our present knowledge about hemidesmosomes comes from studies using mouse and human tissues. As tracking this junction in real-time in mammalian tissues is difficult, the analysis of hemidesmosome dynamics has mainly come from in vitro keratinocyte cell-culture systems. Indeed, very little is known about how hemidesmosomes disassemble and reassemble in vivo (Litjens et al.,2006). As a consequence, the molecular mechanisms maintaining hemidesmosomes' stability in the normal epidermis remain poorly understood.

Progress in understanding the in vivo formation of hemidesmosomes has come in recent years from zebrafish and Caenorhabditis elegans, which are powerful genetic models (Sonawane et al.,2005,2009; Dodd et al.,2009; Labouesse,2006). Their transparent body makes it possible to monitor changes in hemidesmosome structures in vivo under different environmental cues. Furthermore, the availability of mutations in major hemidesmosome components in C. elegans allows deciphering the roles of each component during hemidesmosome assembly, remodeling, and renewal.

In this review, we will briefly summarize our past knowledge on hemidesmosomes in different organisms in vivo. Recent advances gained from the in vitro and in vivo analysis of mammalian hemidesmosomes have been nicely reviewed in several reports (Litjens et al.,2006; Margadant et al.,2008). Therefore, we will specially emphasize findings, implications, and perspectives originating from C. elegans and zebrafish hemidesmosome studies, and discuss how they can help understand the making of hemidesmosomes in mammalian tissues.


Mammalian hemidesmosomes include type I hemidesmosomes in the skin, mouth, and oesophagus, and type II hemidesmosomes mainly found in intestinal epithelial cells (Margadant et al.,2008). Mammalian type I hemidesmosomes contain a set of unique proteins: the membrane receptor α6β4-integrin, which connects to the basement membrane through laminin-332, the cytolinker plakins HD1/plectin and BP230/BPAG1e, and another membrane receptor BP180/collagen XVII. The intermediate filaments, keratin-5 and keratin-14, are associated with α6β4-integrin indirectly through the plakin family proteins plectin and BP230. The integrin α6 subunit also interacts with the tetraspanin CD151, and BP180 weakly associates with laminin-332 (Fig. 1A) (de Pereda et al.,2009). In contrast, type II hemidesmosomes consist of only α6β4-integrin and plectin, and are linked to keratins 8 and 18.

Figure 1.

The structures of mammalian and C. elegans hemidesmosomes and the reorganization of C. elegans hemidesmosome components. A: Components and structures of mammalian type I hemidesmosomes in the skin. B: Components and structures of C. elegans fibrous organelles in the epidermis adjacent to body-wall muscles. Color codes in A and B represent homologous components (identical colors: intermediate filaments; plectin and BP230 vs. VAB-10A. Note that VAB-10A is homologous to both plectin and BP230) or analogous components (similar colors: β4-integrin and BP180 vs. LET-805; laminin vs. UNC-52). Overlapping components, direct interaction. Touching components, predicted binding. C: Immunostaining against muscles (red), LET-805 (orange), IFA2/3 (yellow), VAB-10A (green), and UNC-52 (purple), and reporters mcherry::EPS-8 (cyan), VAB-19::GFP (blue) of wild-type C. elegans embryos at 1.7- or 3-fold stages. The pattern of most fibrous organelle components changes from a narrow band in 1.7-fold to two rows of parallel stripes (corresponding to 2 rows of muscle cells) in 3-fold embryos, except for EPS-8, which does not localize to fibrous organelles until striped patterns form. The ECM ligand UNC-52 co-localizes and expands together with fibrous organelles, though does not form clear parallel stripes. Scale bar = 5 μm.

In zebrafish larvae, hemidesmosomes are only found in the basal domains of epidermal cells covering head and flanks, connecting the cells to the extracellular basal lamina (Sonawane et al.,2005). The structure and components of zebrafish hemidesmosomes are less well studied. Much like in mammals, the zebrafish genome encodes all components necessary to form type I hemidesmosomes (Table 1). However, with the exception of the integrin α6 homolog (Sonawane et al.,2009), their localization and functions await further investigation.

Table 1. Hemidesmosome Components During Evolutiona
SpeciesbITGA6ITGB4PlectinBP230MACFBP180LET-805MUA-3Cytoplasmic IFs
  • a

    Y, clear homologue present; N, no homologue found by Blast searches.

  • b

    Survey for homologues was done using the following Blast servers: D. reriohttp://zfin.org/action/blast/blast; C. intestinalishttp://crfb.univ-mrs.fr/aniseed/; B. floridaehttp://www.ncbi.nlm.nih.gov//genomes/geblast.cgi?bact=off&gi=6106; S. mansonihttp://www.ncbi.nlm.nih.gov//genomes/geblast.cgi?bact=off&gi=24951; D. melanogasterhttp://flybase.org/blast/; Potential homologues were cross-examined to identify the closest human or C. elegans homologue through NCBI-Blast and http://www.wormbase.org/db/searches/blast_blat searches, respectively.

  • c

    An earlier genome-wide search had predicted the existence of ITGA6/ITGB4 and a spectraplakin locus (Sasakura et al.,2003).

  • d

    Genome annotations for these species are still incomplete.

  • e

    An ultrastructural study reported the existence of hemidesmosomes in Amphioxus (Bocina and Saraga-Babic,2006).

  • f

    The genome contains a predicted LET-805 homologue, which aligns well up to the last Fibronectin type III repeat, but has no apparent TM domain and shows no homology through LET-805 cytoplasmic domain.

  • A survey of the genomes of some representatives of major metazoan branches indicates the following trends. With the notable exception of insects that lack cytoplasmic intermediate filaments, other protostomes for which the genome sequence is available feature several cytoplasmic intermediate filaments related to nuclear lamins, whereas deuterostomes have cytoplasmic intermediate filaments related to vertebrate ones (Erber et al.,1998). Vertebrates contain plectin and two giant spectraplakins (Macf1 and Dst); cephalochordates, urochordates, nematodes, and insects contain a single predicted spectraplakin locus (C. elegans vab-10 and D. melanogaster shot) (Roper et al.,2002). Alternative splicing of the spectraplakin locus generates an isoform with the structure of vertebrate Plectin and BPAG1e in C. elegans, but not in D. melanogaster (no information currently available for C. intestinalis, B. floridae, or S. mansoni) (Roper et al.,2002; Bosher et al.,2003). The zebrafish genome encodes homologues of all mammalian hemidesmosome components, suggesting that it can form type I hemidesmosomes. In contrast, the urochordate C. intestinalis may lack a BP180 homolog, but has a probable α6β4-integrin homologue (http://aniseed-ibdm.univ-mrs.fr/) (Sasakura et al.,2003), suggesting that it could form a structure related to type II hemidesmosomes. For the cephalochordate B. floridae, the situation is more ambiguous since it may not contain a true α6β4-integrin (Putnam et al.,2008). The lophotrochozoa S. mansoni seems to lack both α6β4-integrin and nematode (LET-805, MUP-4, MUA-3) ECM receptor homologues (Berriman et al.,2009). Note, however, that in the latter two cases genome annotations are still incomplete. Finally, C. elegans contains the membrane receptors LET-805/myotactin basally and MUA-3/MUP-4 apically. One possible evolutionary scenario is that plakin and intermediate filaments form the core components of hemidesmosomes, and that they recruited either α6β4-integrin in vertebrates, or myotactin/MUP-4/MUA-3 in nematodes, and yet possibly other ECM receptors in other species during evolution.

M. musculusYYYYNYNNY
C. intestinaliscYYSingle spectraplakin locusNNNY
B. floridaed,eITGA6/7NSingle spectraplakin locusNNfNY
S. mansonidITGA7NPartial homology to spectraplakinNNNY
C. elegansNNSingle spectraplakin locusNYYY
D. melanogasterNNSingle spectraplakin locusNNNN

C. elegans embryos and larvae contain hemidesmosome-like structures, which anchor dorsal and ventral epidermal cells to distinct extracellular matrices on both apical and basal surfaces. The two junctional complexes are connected by intermediate filaments, and together called fibrous organelles (FOs), sometimes also referred to as trans-epithelial attachments (TEAs) (Cox and Hardin,2004; Ding et al.,2004; Labouesse,2006). These structures mainly serve two functions. One is to attach the epidermis to the ECM and maintain epidermis integrity; the other is to act as tendon-like structures that transmit muscle tension to the cuticle exoskeleton. Similar structures are present in the pharynx marginal epithelial cells, rectum, vulva, and in areas of the epidermis that contact the mechanosensory touch neurons (Cox and Hardin,2004). As detailed below, there are some differences between the hemidesmosome-like junctions found apically and basally.

C. elegans hemidesmosomes have the morphological hallmarks of vertebrate hemidesmosomes. They appear as electron-dense membrane plaques and are associated with C. elegans intermediate filaments IFA-2/MUA-6, IFA-3, and IFB-1 (Francis and Waterston,1991; Bosher et al.,2003; Woo et al.,2004). C. elegans intermediate filaments are more closely related to nuclear lamins (Dodemont et al.,1994), yet can form IFB-1/IFA-2 and IFB-1/IFA-3 heterodimers (Karabinos et al.,2003). Fibrous organelles contain the sole plectin and BPAG1 homologue in the C. elegans genome, VAB-10A, which is generated by alternative splicing from the unique spectraplakin locus vab-10 and should thus play the combined role of both plakins in vertebrate hemidesmosomes (Table 1) (Bosher et al.,2003). VAB-10A has a putative intermediate filament binding-domain at its C-terminal domain, and is required to anchor intermediate filaments at fibrous organelles. (Fig. 1B) (Bosher et al.,2003; Hapiak et al.,2003; Woo et al.,2004).

Fibrous organelle membrane receptors are distinct from integrins in vertebrate hemidesmosomes. At the basal side, they contain LET-805/myotactin, a single-pass trans-membrane protein with extracellular fibronectin type III repeats; it is large enough to potentially reach the muscle membrane across the ECM and to physically interact with muscle proteins. By analogy to mammalian plakins and integrins, it is speculated that myotactin binds to VAB-10A in fibrous organelles, although no direct evidence is yet available to support this notion (Labouesse,2006). At the apical side, VAB-10A is predicted to associate with two single trans-membrane receptors, MUP-4 and MUA-3, which both harbor EGF repeats in their extracellular domains, and are weakly homologous to vertebrate matrilins (Bercher et al.,2001; Hahn and Labouesse,2001; Hong et al.,2001).

The ankyrin repeat–containing protein VAB-19, homologous to the mammalian tumor suppressor Kank, co-localizes with known fibrous organelle components (Ding et al.,2003). In addition, yeast two-hybrid screening identified the signaling adaptor EPS-8 as a potential physical interaction partner of VAB-19 (Ding et al.,2008). The central domain of EPS-8 is both required and sufficient for VAB-19 binding. Based on EPS-8 localization in the pharynx marginal cells (which are bigger and offer greater spatial resolution), EPS-8 is predicted to associate only with the apical moiety of fibrous organelles. Unlike VAB-19, which is recruited to fibrous organelles right after they begin to form, EPS-8 remains diffuse and does not reorganize until very late during embryogenesis (Fig. 1C) (Ding et al.,2008). It is still unclear how VAB-19 and EPS-8 are attached to fibrous organelles.

While C. elegans hemidesmosomes anchor epidermal cells to the underlying muscle quadrants, they are distinct from the vertebrate plectin-containing muscle dystroglycan complex. Firstly, the C. elegans dystroglycan homologue DGN-1 is not expressed in the epidermis. Second, in contrast to vab-10 mutants, dgn-1 null animals are viable and have no muscle or epidermis defects (Johnson et al.,2006). Hence, it is reasonable to assume that many of the mechanistic principles derived from myotactin-coupled hemidesmosomes will likely be applicable to integrin-coupled type I hemidesmosomes rather than muscle dystroglycan complex in vertebrates.


In mammals, the hemidesmosome assembly process begins at the embryonic stage. In human skin, for example, hemidesmosome-like plaques first appear at about 9 weeks' gestational age, and their number increases over the next 6 weeks. Association of hemidesmosomes with intermediate filaments is also taking place over this period. Hemidesmosome structures appear fully developed by 15 weeks' gestation (McMillan and Eady,1996). Mature hemidesmosomal plaques have an average diameter of 200 nm (Jonkman et al.,2002). In vitro data indicate that the interaction between β4-integrin and plectin is a key step in hemidesmosome assembly (Koster et al.,2004).

Targeted inactivation of hemidesmosome components in mouse shows that laminin-332 and α6β4-integrin are critical to construct stable hemidesmosomes (Table 2A) (Georges-Labouesse et al.,1996; van der Neut et al.,1996; Ryan et al.,1999; Meng et al.,2003). Their absence leads to severe skin blistering and near absence of hemidesmosomes, although reduced amounts of BP230, BP180, and plectin can still associate with the basal membrane. Loss of the plakins plectin and BP230 affects hemidesmosome formation and induces mechanical fragility of the skin, but to a lesser extent than integrin loss, since plaques could still be seen. In BP230 KO mice, keratins fail to attach to defective hemidesmosomes, whereas in plectin KO mice, they are absent only in areas undergoing blistering and histolysis (Guo et al.,1995; Andra et al.,1997). In contrast, human patients with mutations affecting BP180 display milder hemidesmosome defects (McGrath et al.,1995). Therefore, BP180 might be recruited only to strengthen the whole hemidesmosome structure (Table 2A) (Margadant et al.,2008).

Box 2

C. elegans embryos epidermal cells start to differentiate approximately 4 hr post-zygote formation, first by assembling adherens junctions. They undergo three morphogenetic processes (Chisholm and Hardin,2005): (1) dorsal epidermal cells intercalate, which creates a slight ventral bend; (2) subsequently, ventral epidermal cells extend ventrally by a process of epiboly to enclose the embryo; (3) finally the embryo elongates within 2 hr by a process involving cell-shape changes but not cell division nor migration. Myoblasts migrate to reach their final positions in contact with dorsal and ventral epidermal cells during the second step and at the beginning of the third step. C. elegans hemidesmosome-like junctions, or fibrous organelles, also start to form during this time period. Genetic analysis has revealed that embryonic elongation depends on proper establishment of epithelial polarity, adherens junction assembly, hemidesmosome assembly, and cytoskeleton remodeling (Chisholm and Hardin,2005; Labouesse,2006). The initial phase of elongation depends mainly on the lateral epidermal cells (Diogon et al.,2007; Gally et al.,2009). The second part of elongation also requires muscle activity, although it is not known what kind of input muscles provide to the epidermis (Waterston,1989; Williams and Waterston,1994). Mutants with defective fibrous organelles arrest before the twofold stage (see Fig. 2A) generally with strong epidermis integrity defects. Mutants with defective muscles also arrest at the twofold stage and fail to move within the egg-shell due to paralysis, hence their common phenotype name known as Pat (Paralyzed at twofold).

Table 2. Interdependence of Hemidesmosome Components for Their Assemblya
inline image

Zebrafish hemidesmosomes firstly become visible in the basal epidermal cells at about 4.5 days post-fertilization, and mature into electron-dense plaques of ±100 nm in length one day later (Sonawane et al.,2005). However, the membrane receptor of zebrafish hemidesmosomes, integrin α6 (Itga6), localizes to the basal domain of the epidermis at as early as 2.5 days post-fertilization. It then forms clusters with intermediate filaments prior to hemidesmosome formation. The assembly of mature hemidesmosomes appears to depend on the recruitment of laterally localized Itga6 at a later stage. An insufficient amount of Itga6 protein results in non-functional hemidesmosomes (Sonawane et al.,2009).

Although hemidesmosomes have long been considered as stable entities, several studies have challenged this view. In vitro cell culture studies have suggested that hemidesmosomes undergo active disassembly and reassembly in response to environmental cues. Time-lapse analysis of the hemidesmosome markers BP180 and β4-integrin has revealed the dynamic properties of these proteins, especially during keratinocyte migration and wound healing in scratch assays (Geuijen and Sonnenberg,2002; Tsuruta et al.,2003). Moreover, it has been observed that hemidesmosomes get rapidly disassembled during carcinoma invasion (Rabinovitz and Mercurio,1997; Santoro et al.,2003; Rabinovitz et al.,2004). However, the dynamic nature of hemidesmosomes has not yet been confirmed in vivo.

C. elegans provides a nice system in which to investigate hemidesmosome assembly and remodeling in vivo, as indeed its hemidesmosome-like junctions undergo dynamic changes during development. They are assembled prior to embryonic elongation, a process during which the embryo transforms its shape from a ball of cells to a vermiform larva in the absence of cell division (Box 2). They initially form puncta aligned along the anterior-posterior axis and progressively form parallel stripes oriented along the circumference when muscles start to contract (Figs. 1C, 2A,B) (Labouesse,2006). The average size of C. elegans fibrous organelles is about 150 nm in diameter, which is comparable to that of vertebrate hemidesmosomes (Francis and Waterston,1991; Bosher et al.,2003).

Figure 2.

The assembly process of C. elegans hemidesmosomes. A: Immunostaining against VAB-10A of wild-type C. elegans embryos at comma, 1.7- or 3-fold stages, and larvae at L1 stage. Scale bar = 10 μm. B: Schematic representation showing fibrous organelle reorganization during embryonic elongation (for details see Box 2), corresponding to the images shown in A. After recruitment to the muscle-adjacent areas, fibrous organelle components first form punctate aggregates and progressively align into parallel stripes along the epidermis. Pink, dorsal hypodermis; red, ventral hypodermis; yellow, seam cells; orange, muscle cells; blue, fibrous organelles. C: Schematic representation showing the distribution of fibrous organelle components at the cellular level during the stages illustrated in B. They are first randomly distributed (left). Then, presumably upon input from adjacent tissues (possibly from muscles), the core component VAB-10A/plakin binds to the membrane receptors MUA-3/MUP-4 apically and LET-805/myotactin basally, and begins the first step of assembly (middle). As elongation progresses, newly formed hemidesmosome complexes are concentrated at the ridge of annuli and reorganized into short stripes (right).

VAB-10A/plakin and LET-805/myotactin, like their mammalian functional counterparts plectin, BPAG1 andα6β4-integrin, are also crucial for hemidesmosome assembly. In vab-10A or let-805 loss-of-function mutants, fibrous organelle structures are strongly affected. Neither VAB-10A nor LET-805/myotactin is required for the initial localization of VAB-19 and intermediate filaments to the muscle contact area, but both are essential for their assembly into mature fibrous organelles. In let-805 mutants, VAB-19 becomes diffuse in the epidermis and intermediate filaments form stripes that occupy the entire epidermis, whereas in vab-10A mutants VAB-19 never form stripes and intermediate filaments form abnormal aggregates or bundles. As a result, strong mutations in vab-10A or let-805 affect epidermis integrity and cause detachment of muscles from the cuticle (Table 2B) (Hresko et al.,1999; Bosher et al.,2003).

VAB-19 is not required for fibrous organelle assembly at an early stage; however, it becomes crucial during reorganization. In vab-19 cryo-sensitive mutants, LET-805/myotactin fails to form parallel stripes beyond the twofold stage, whereas VAB-10A and intermediate filaments form stripes that occupy the entire dorsal and ventral epidermis (instead of just the muscle contact areas) (Table 2B) (Ding et al.,2003). It suggests that VAB-19 functions during fibrous organelle remodeling and helps them mature into stable circumferential stripes along the epidermis. Consistent with this notion, the putative VAB-19 interaction partner EPS-8 co-localizes with fibrous organelles only during late elongation, corresponding to the time point when the remodeling process takes place (Figs. 1C, 2C). In addition, loss of EPS-8 function also results in defective remodeling of fibrous organelles, since parallel stripes never form in eps-8 mutants (Table 2B) (Ding et al.,2003). Therefore, it is reasonable to speculate that EPS-8, together with VAB-19, promotes fibrous organelle reorganization during late embryogenesis.

Hence, fibrous organelle components appear to be recruited to the muscle contact area independently from each other, and to form physical interactions later during fibrous organelle maturation, presumably to withstand increasingly stronger muscle pulling force (Fig. 2C). The unique intermediate filament-bundling phenotype observed in vab-10A mutants is consistent with a role of VAB-10A in anchoring and organizing intermediate filament distribution, which is conserved in mammalian hemidesmosomes (de Pereda et al.,2009). However, since fibrous organelle formation involves coordination between apical and basal hemidesmosomes, the underlying mechanisms might be more complicated than the formation of vertebrate hemidesmosomes, which only happens on the basal side of the epidermis. The initial nucleation of hemidesmosomes might take place independently on either side of the epidermis. Newly formed hemidesmosomes on both sides could later reorganize and align with each other during development to form mature fibrous organelles. Indeed, as discussed in the next section, recent studies have found a mechanism that specifically regulates basal hemidesmosome formation in C. elegans (Zahreddine et al.,2009). However, the detailed process of hemidesmosome assembly into fibrous organelles still awaits further investigation.


Extracellular Matrix

The components of the extracellular matrix not only serve as anchorage for hemidesmosome structures, but their changes also provide guidance for hemidesmosome formation and reorganization (Ryan et al.,1999; Meng et al.,2003). In vitro, a cleaved form of laminin-332, the high-affinity ligand for α6β4-integrin, stabilizes hemidesmosomes, while unprocessed laminin-332 triggers disassembly and turnover of hemidesmosomes (Goldfinger et al.,1999; Hintermann and Quaranta,2004). This process is yet to be confirmed in vivo.

In C. elegans, the main ECM protein necessary for fibrous organelle formation does not appear to be laminin. Four laminin family members exist in C. elegans, laminin αA (EPI-1), αB (LAM-3), β (LAM-1), and γ (LAM-2). Together with other myoblast-produced ECM proteins, including collagen IV, C. elegans, laminins affect muscle polarization and their adhesion to the epidermis but it is not clear whether they also affect fibrous organelle formation (Graham et al.,1997; Gupta et al.,1997; Huang et al.,2003). The ECM protein HIM-4 is required for the formation of a subset of fibrous organelles, in regions where the epidermis contacts the uterus and mechanosensory axons (Vogel and Hedgecock,2001). Further investigations are still required to understand whether these ECM components are directly or indirectly linked to fibrous organelles. Molecular mechanisms underlying their roles in fibrous organelle formation also remain to be examined.

The main C. elegans ECM protein produced by the epidermis is the perlecan homologue UNC-52 (Spike et al.,2002). Its localization coincides with that of fibrous organelles during development (Fig. 1C). In unc-52 loss-of-function mutants, the distribution of both LET-805 and VAB-10A is highly abnormal (Hresko et al.,1994). Recent data have suggested that the folding of UNC-52 and its subsequent secretion to the ECM depends on the function of the molecular chaperone CRT-1/calreticulin. In crt-1 mutants, UNC-52 protein level is reduced by 50% (Zahreddine et al.,2009). This minor reduction is detrimental to fibrous organelle formation in a background with compromised VAB-10A function, resulting in detachment of muscles from the cuticle, and disorganized fibrous organelle patterns in areas with detached muscles (Fig. 3A,B) (Zahreddine et al.,2009). This finding, together with UNC-52 distribution and unc-52 mutant phenotype, suggests that UNC-52 function is essential for fibrous organelle formation, and that UNC-52 directly interacts with the membrane receptor LET-805/myotactin (Fig. 1).

Figure 3.

Regulation of hemidesmosome assembly. A: The level of the membrane receptor LET-805 is controlled indirectly by the E3-ubiquitin ligase EEL-1. EEL-1 likely targets the degradation of a factor activating let-805 expression or controlling let-805 mRNA stability. The folding of UNC-52/perlecan, the presumptive ECM ligand, is assisted by the molecular chaperone CRT-1/calreticulin, which when properly folded, is secreted by epidermal cells into the ECM. The assembly of IFB-1 filaments is regulated by sumoylation. Sumoylated IFB-1 remains unpolymerized to ensure normal turnover of IFB-1 filaments. B: In eel-1 mutants, over-produced LET-805 accumulates at the basement membrane, which affects the timely reorganization of hemidesmosomes. In crt-1 mutants, UNC-52 mis-folding reduces its export to the ECM, which weakens hemidesmosome connection to the basement membrane. In smo-1 mutants lacking the sumoylation process, IFB-1 proteins aggregate ectopically, disrupt normal filament formation, and reduce IFB-1 exchange rate within the hemidesmosomes. In all three cases hemidesmosome structure can be compromised. C: In zebrafish, the interplay between E-cadherin on the lateral membrane, the polarity protein Lgl2, and the vesicular protein Clint controls Itga6 delivery to the basal membrane beyond day 4 of larval development. Too much Itga6 (when E-cadherin is absent) or too little (when Lgl2, or Clint are absent) affects hemidesmosome structure.

Post-Translational Modification

Biochemical studies in culture systems have outlined the importance of post-translational modifications of proteins in controlling hemidesmosome stability in mammalian system. For example, the phosphorylation state of the β4-integrin cytoplasmic domain determines the strength of its interaction with plectin, and may act as a regulation point for hemidesmosome assembly and disassembly. Several protein kinases can phosphorylate β4-integrin, including PKC, PKA, the Epidermal Growth Factor Receptor, and the Macrophage Stimulating Factor Receptor (Santoro et al.,2003; Rabinovitz et al.,2004; Wilhelmsen et al.,2007). However, controversy remains in explaining their roles in hemidesmosome formation due to unphysiological experimental conditions (Margadant et al.,2008).

Although signaling events have yet to be discovered in C. elegans fibrous organelles, there have been suggestions that fibrous organelles may respond to signaling pathways. For instance, ablation of muscle precursors affects fibrous organelle formation, raising the possibility that muscles send a signal to promote epidermis differentiation and/or fibrous organelle formation (Hresko et al.,1999). Recently, a genetic screen for genes acting together with vab-10A to promote fibrous organelle formation identified two protein kinases, the p21-activated kinase PAK-1, and the Leucine zipper kinase PIG-1. Their loss of function combined with a weak vab-10A mutation affects fibrous organelle stability (Zahreddine et al.,2009). It would be interesting to explore the functions of these kinases in fibrous organelle assembly, and identify their phosphorylation targets. It might confirm the findings achieved by in vitro studies, and gain further insight into the roles of phosphorylation during hemidesmosome formation.

Limor Broday and collaborators recently found through a proteomic screening approach that sumoylation contributes to the maintenance and turn-over of fibrous organelles (Kaminsky et al.,2009). Specifically, they showed that the intermediate filament protein IFB-1 is sumoylated at lysine-460. FRAP experiments suggest that this modification ensures a stable pool of non-polymerized IFB-1, which can exchange with fibrous organelle-linked IFB-1. Loss of sumoylation results in ectopic aggregation of IFB-1 in the epidermis, which as a consequence lowers the amount of IFB-1 available for fibrous organelle and disrupts their formation, with indirect consequences on LET-805/myotactin and VAB-19 distribution. It suggests that IFB-1 sumoylation serves as a sequestering factor that regulates intermediate filament polymerization (Fig. 3) (Kaminsky et al.,2009). Sumoylation is a conserved process also found in mammalian cells; therefore, its function in mammalian hemidesmosomes should also be explored.

Intracellular Trafficking

Hemidesmosomes are polarized structures in both invertebrates and vertebrates. In C. elegans epidermis, although hemidesmosomes are present at both sides of the cells, their compositions are slightly different apically and basally. They consist of different membrane receptors, with MUA-3/MUP-4 at the apical side and LET-805 at the basal side. In vertebrates, hemidesmosomes are found only at the basal side of epidermal cells in contact with the basement membrane. Therefore, a carefully controlled intracellular trafficking machinery must exist to ensure the correct targeting of these membrane receptors to either the apical or basal sides of the polarized epidermis.

The dependence of hemidesmosome on trafficking mechanism is nicely demonstrated in zebrafish (Sonawane et al.,2005,2009; Dodd et al.,2009). Before zebrafish hemidesmosome formation, the membrane receptor α6-integrin (Itga6) is initially localized at both the lateral and basal domains of epidermal cells. However, when hemidesmosome clusters appear, the laterally-localized Itga6 molecules are transported to the basal side of the epidermis. This process depends on the cell polarity protein called Lgl2, which regulates polarized exocytosis, and the ENTH-domain protein Clint1, which is involved in vesicle trafficking (Fig. 3C) (Dodd et al.,2009; Sonawane et al.,2009). The re-localization of Itga6 is further antagonized by E-cadherin. Either insufficiently-transported or over-transported Itga6 to the basement membrane results in defects of hemidesmosome assembly (Sonawane et al.,2009). It will be important to reinvestigate the phenotypes of mouse E-cadherin and Lgl2 knockouts to see whether they also modulate hemidesmosome formation. Interestingly, β4-integrin and BPAG1 interact with yet another cell polarity protein, Erbin (Favre et al.,2001). The biological role of this interaction has yet to be established.

Control of Hemidesmosome Component Levels

Absence or severe reduction of key components in the hemidesmosomes can have great impact on its formation. It is unexpected, however, that a minor increase or decrease of certain components also affects hemidesmosome structures, as outlined in recent studies using both vertebrate and invertebrate model systems.

The recent C. elegans RNAi screen for enhancers of vab-10A revealed several genes regulating protein turnover (Zahreddine et al.,2009). Among them, the E3-ubiquitin ligase EEL-1 negatively regulates LET-805 expression through an unknown factor. In eel-1 mutants, LET-805 protein levels increase by 50%, which proves harmful for fibrous organelle formation when combined with a VAB-10A/plakin that is partially functional (Fig. 3A,B) (Zahreddine et al.,2009).

One explanation for why an increase in membrane receptors affects hemidesmosome biogenesis is that it increases the adhesion between the epidermis and the ECM while the epidermis still undergoes shape changes during morphogenesis. If hemidesmosome reorganization does not occur properly, then, in the tug-of-war between increased adhesion against the forces coming from within the epidermis and from outside, the whole hemidesmosome might rupture if not remodeled in time. Consistent with this hypothesis, reducing elongation forces in C. elegans by affecting the amount of the βH-spectrin SMA-1, which is required for elongation efficiency (McKeown et al.,1998; Praitis et al.,2005), indeed partially rescues the hemidesmosome defects caused by an increased membrane receptor LET-805 level (Zahreddine et al.,2009).

Strikingly, in zebrafish too slight increase of Itga6 can affect hemidesmosome formation. The loss of E-cadherin function mentioned above increases the α6-integrin targeting to the basal domain by less than twofold. This increase in α6-integrin level, however minor, results in abnormal hemidesmosomes that exhibit mat-like patterns instead of discrete electron-dense plaques (Sonawane et al.,2009).

Taken together, these findings further demonstrate that the formation of hemidesmosomes is a delicate process fine-tuned by different mechanisms, and that minor fluctuation of hemidesmosome protein abundance can result in great effect. Such effect could be particularly strong in hemidesmosomes undergoing constant disassembly and reassembly, for example in wounded tissue. The hemidesmosomes in healing epithelia resemble C. elegans fibrous organelles in many ways. They are connected to the ECM underlying contracting myofibroblasts, and they undergo a dynamic reorganization process during epidermal cell migration and wound healing (Goldfinger et al.,1999; Werner et al.,2007). If the membrane receptor integrin level is aberrantly elevated, it will create unnecessary adhesion to the ECM and hamper cell migration. Therefore, studying mechanisms affecting minor fluctuation of hemidesmosome components might greatly improve our understanding of the wound healing process.


Remarkable progress has been made in the study of vertebrate and invertebrate hemidesmosome formation in recent years. Simple model organisms such as zebrafish and C. elegans, due to their optical clarity and powerful genetic tools, help validate findings from in vitro systems and improve our understanding of hemidesmosome formation in a physiological context. Indeed, several novel C. elegans hemidesmosome components have been discovered in recent years, such as VAB-19 and EPS-8, which possess mammalian homologues (Ding et al.,2003,2008). Moreover, genetic screens in zebrafish and C. elegans (Sonawane et al.,2005,2009; Dodd et al.,2009; Zahreddine et al.,2009) identified novel players involved in diverse cellular processes, including signaling, trafficking, transcription, RNA splicing, protein folding, and ubiquitin pathway, all of which have mammalian homologs (Zahreddine et al.,2009). It would be worthwhile to explore their functions in vertebrate hemidesmosomes, especially during wound healing and carcinoma invasion.

C. elegans can also be a useful system to study the influence of environmental cues on hemidesmosomes in vivo. Recent studies have indicated that focal adhesions, another cell-matrix attachment structure, mature under the influence of mechanical force (Geiger and Bershadsky,2002). Whether mechanical tension also promotes the maturation of hemidesmosomes, which are meant to resist mechanical forces, is an open question that has never been investigated. For instance, C. elegans hemidesmosomes become submitted to tension from the underlying muscles, and it would be interesting to see to what extent muscles might stimulate their reorganization during embryonic morphogenesis. Exploring the mechanisms underlying hemidesmosome remodelling in C. elegans should greatly facilitate our understanding of hemidesmosome formation in general.

Several studies, mainly performed in vitro, suggested that hemidesmosomes might act as signal transducers. For example, biochemical studies of plectin showed that it can interact with the receptor for activated protein-kinase C (RACK1), the γ1 subunit of AMP-activated protein kinase, the Fer kinase, and that it modulates the activity of stress-activated MAP-kinases such as p38 (Lunter and Wiche,2002; Osmanagic-Myers and Wiche,2004; Gregor et al.,2006; Osmanagic-Myers et al.,2006). Plakins also influence Rho GTPase signaling in vivo (Andra et al.,1998; Lee and Kolodziej,2002). Future analysis in zebrafish and C. elegans should be instrumental in defining whether hemidesmosomes serve, not only as attachment structures, but also as active signaling machineries in specific physiological circumstances.


We thank Dr. Andrew D. Chisholm for providing mcherry::EPS-8(central). We apologize to the authors whose work we did not include in this review due to space constraints.