Epiplakin is a large (>725 kDa) cytoskeletal protein exclusively expressed in epithelial tissues. It has a unique structure, consisting entirely of plakin repeat domains (PRDs), one of the hallmarks of spectraplakin protein family members. Previous studies, including the phenotypic analyses of knockout mice, failed to reveal the biological function of epiplakin. Using in vitro binding assays, we show here that all but one of the 16 PRDs of mouse epiplakin bind to keratins of basal keratinocytes. Nevertheless, in primary keratinocyte cell cultures, epiplakin only partially colocalized with keratin intermediate filament networks. However, upon application of cellular stress in the form of keratin hyperphosphorylation, osmotic shock or UV irradiation, the entire cytoplasmic epiplakin pool became associated with keratin. In response to such types of stress, epiplakin initially translocated to the still-intact keratin filament network and remained associated with keratin after its disruption and transformation into granular aggregates. Time-course experiments revealed that serine/threonine (okadaic acid) and tyrosine (orthovanadate) phosphatase inhibitor-induced filament disruption in differentiated keratinocytes proceeded faster in epiplakin-deficient cells compared with wild-type cells. Our data suggest that epiplakin plays a role in keratin filament reorganization in response to stress, probably by protecting keratin filaments against disruption in a chaperone-like fashion.
Epiplakin was originally isolated as an autoantigen with immunoreactivity toward the serum from a patient who suffered from a subepidermal blistering disease (Fujiwara et al., 1994; Fujiwara et al., 1996). The protein is expressed in epithelial tissues, with especially high levels found in skin (Spazierer et al., 2003). Its function is still unknown, because mice deficient in epiplakin did not show any discernible phenotype (Spazierer et al., 2006), except for a slight acceleration of keratinocyte migration, as reported by Goto et al. (Goto et al., 2006). Surprisingly, not even when the skin of knockout mice was challenged by tape stripping or wounding, were differences to wild-type mice observed (Spazierer et al., 2006).
Based on the sequence analysis of human and mouse cDNAs, epiplakin was classified as a member of the spectraplakin protein family of cytolinkers (Fujiwara et al., 2001; Spazierer et al., 2003). Its gene encodes a large protein consisting entirely of plakin-repeat domains (PRDs), of which there are 16 in mouse epiplakin and 13 in the human protein (Fujiwara et al., 2001). The PRD contains a highly conserved ∼20 kDa core region, called the module (Sonnhammer et al., 1997; Janda et al., 2001) and a less conserved linker region of variable length. The lack of other domains typically present in spectraplakins makes epiplakin an unusual member of this protein family, and brings its functional relationship to the other members into question. In general, cytolinker proteins connect cytoskeletal filaments with one another and attach them to plasma-membrane-associated adhesive junctions (for a review, see Röper et al., 2002; Jefferson et al., 2004). The ability to interact with intermediate filaments has been shown for several members of the spectraplakin family. Plectin, for example, binds to intermediate filament proteins of various types via an intermediate-filament-binding site located between modules 5 and 6 of its six PRDs (Nikolic et al., 1996) and an additional vimentin-binding site has been identified in its N-terminal actin-binding domain (Sevcik et al., 2004). For desmoplakin, keratin and vimentin binding have been reported to involve sites within its three C-terminal PRDs (Stappenbeck and Green, 1992; Stappenbeck et al., 1993; Kouklis et al., 1994; Yang et al., 1996; Meng et al., 1997) and an interaction with keratins and neuronal intermediate filaments has been demonstrated for the PRDs of BPAG1n and BPAG1e (Yang et al., 1996). Association with keratin filaments has also been shown for envoplakin and periplakin (Karashima and Watt, 2002; Kazerounian et al., 2002). In the case of epiplakin, binding of the most C-terminal PRD to keratins was shown in dot blot assays (Jang et al., 2005) and, in a recent study, keratin binding of module 8 has been demonstrated using slot blot assays (Wang et al., 2006). It is not known, however, whether all the PRDs in epiplakin can actually bind to keratin. Furthermore, colocalization of epiplakin with keratin filament networks has been reported so far only for human carcinoma (HeLa) and keratinocyte (HaCaT) cell lines, whereas it was not observed in primary keratinocytes (Jang et al., 2005). Thus, it remains an open question whether the interaction of epiplakin with keratins K5 and K14 observed in vitro is of biological significance.
Based on the available data and functions previously analyzed for other spectraplakin protein family members, one might expect epiplakin to play a role in the regulation of keratin filament dynamics and/or stabilization of the filaments. For example, it has been shown for plectin that upon incubation of keratinocytes with okadaic acid (OA), the protein is slowly released from keratin filaments and it is absent from keratin aggregates that eventually form (Strnad et al., 2002; Osmanagic-Myers et al., 2006). Furthermore, plectin deficiency in keratinocytes leads to faster keratin intermediate filament network disassembly upon incubation with OA (Osmanagic-Myers et al., 2006). In contrast to plectin, 14-3-3 proteins associate with OA-induced keratin aggregates and seem to keep the aggregated keratin from quickly re-establishing intermediate filament network systems (Strnad et al., 2002). As the interaction of 14-3-3 proteins with keratin is phosphorylation dependent, they have been proposed to act as solubilization factors for hyperphosphorylated keratins. Binding specifically to keratins hyperphosphorylated on serine/threonine (Liao and Omary, 1996; Strnad et al., 2002), 14-3-3 proteins have also been shown to modulate mitotic progression of cells (Ku et al., 2002). In fact, phosphorylation serves as a common mechanism in modulating the function and accessibility of keratins.
In addition to the large increase observed in mitosis, phosphorylation of keratins is also dramatically increased in response to stress, as shown in both cultured cells and mouse models (Toivola et al., 2002; Fausther et al., 2004; Ridge et al., 2005). Consequently, phosphatase inhibitors, such as okadaic acid (OA) and orthovanadate (OV), which cause hyperphosphorylation of keratins, are increasingly used to disrupt keratin filament networks of cells, thereby mimicking cellular stress (Strnad et al., 2001; Strnad et al., 2002; Osmanagic-Myers et al., 2006). In this study, we have used a similar approach combined with UV irradiation and osmotic stress, to assess whether epiplakin plays any role in the stress response of keratinocytes.
Epiplakin associates with epidermal keratins via multiple binding sites
Epiplakin is expressed in all types of epithelia, including the epidermis where expression is particularly strong (Spazierer et al., 2003). Nevertheless we observed no differences in keratin intermediate filament cytoarchitecture when comparing primary cultures of wild-type and epiplakin-null epidermal keratinocytes (stratified epithelium), nor of corresponding hepatocyte (simple epithelium) cultures (Spazierer et al., 2006). To assess whether epiplakin was associated with keratin filaments in such cells, primary keratinocyte and hepatocyte cell cultures from wild-type mice were subjected to double immunofluorescence microscopy using antibodies to epiplakin combined with anti-pan-keratin (keratinocytes) or anti-keratin K8/18 (hepatocytes) antibodies. In keratinocytes, we observed predominantly dotty epiplakin-positive structures, some of which were clearly overlapping with keratin filaments (Fig. 1A, upper panels). By contrast, in hepatocytes, codistribution of epiplakin with keratin along filaments was clearly evident (Fig. 1A, lower panels).
Using dot blot in vitro binding assays, Jang et al. (Jang et al., 2005) demonstrated binding of one of the human epiplakin repeats (the most C-terminal) to various intermediate filament proteins, including K5/K14, K8/K18 and vimentin. Comparable results were obtained when immobilized module 8, linker 9, and module 8 plus linker 9 were incubated with keratin, vimentin or desmin in slot blot assays (Wang et al., 2006). As mouse epiplakin comprises 16 PRDs (each consisting of a linker region and a core PRD module; Fig. 1B), it was of interest to assess whether other PRDs were also able to bind to keratin. Therefore, we expressed individual (mouse) PRD1 to PRD9 and PRD16 as GST-fusion proteins and overlaid them onto nitrocellulose-immobilized K5 and K14 isolated from immortalized p53–/– keratinocytes. As shown in Fig. 1C, all epiplakin PRDs except PRD5 were found to bind to K5. Interestingly, PRD9, which, because of virtually identical primary structures (Spazierer et al., 2003), is representative of PRD9-15, additionally showed binding to K14. GST alone (negative control) showed no keratin binding. In addition, we investigated which of the PRD9 subdomains, the module and/or the linker, showed K5/K14 binding, and we similarly examined the linker subdomain of PRD8, which accounts for the structural difference between PRD8 and PRD9 (modules 8 and 9 are identical), and the linker-less module of PRD16 (see Fig. 1B), which differs from module 9 in just the last 16 C-terminal amino acid residues. Neither linker 8 nor linker 9 showed binding to keratins in blot overlay assay, whereas modules 9 and 16 both bound to K5, but not to K14 (Fig. 1D). From this it appeared that linker 9 and module 9 were capable of binding to K14 only when combined. In an alternative assay, GST-fusion proteins of PRD8 and PRD9 were covalently linked to Sepharose 4B beads to pull down keratins from keratinocyte cell extracts. As shown in Fig. 1E, K5 as well as K14 co-sedimented with PRD8- and PRD9-coupled beads, whereas beads coupled to GST alone were ineffective. Thus, whereas pull-down and blot overlay results were in full agreement for PRD9, K14 binding of PRD8 was detectable only by pull down assay. This difference can be explained, however, by heterodimer formation of K5 and K14 in cell-lysates, leading to the additional pull down of K14 with PRD8-bound K5.
Epiplakin deficiency has no obvious effect on Ca2+-mediated keratinocyte differentiation
The prominent expression of epiplakin in keratinocytes combined with its partial association with the keratin filament network prompted us to examine whether epiplakin plays any role in keratinocyte differentiation. In fact, Jang et al. (Jang et al., 2005) reported a substantial upregulation of epiplakin during Ca2+-mediated differentiation of normal human epidermal keratinocytes (NHEKs). However, for NHEKs, only 10% of the original (undifferentiated) cell population was found to be epiplakin-positive (Jang et al., 2005), whereas in undifferentiated primary mouse keratinocytes we observed that 100% of the cell population expressed epiplakin (Fig. 1A, and data not shown). Nevertheless, immunoblotting of cell lysates from primary mouse keratinocytes that had been cultivated in medium containing 1.3 mM calcium for up to 5 days revealed a significantly elevated expression of epiplakin. Using tubulin as a loading control (Fig. 2A), a ∼1.2- and ∼1.8-fold upregulation of epiplakin compared with basal levels (day 0) was observed after 3 and 5 days of differentiation, whereas corresponding levels of involucrin, which served as differentiation marker, were up ∼2.4- and ∼3-fold.
To examine whether the increasing epiplakin protein levels had any effect on the time course of differentiation, wild-type and epiplakin-deficient primary keratinocytes grown in 1.3 mM Ca2+-containing medium were subjected to immunofluorescence microscopy using desmoplakin as a differentiation marker. At 3 hours after initiating differentiation in both cell types, desmoplakin showed its typical association with cell-cell junctions, and no differences in the staining pattern became evident thereafter (Fig. 2B). Epiplakin showed few changes in its distribution during differentiation (Fig. 2B). Furthermore, we could not detect any differences between differentiated wild-type and knockout keratinocytes regarding the size or shape of cells, nor did we find any differences in keratin expression or filament organization (Fig. 2B, and data not shown).
Hyperphosphorylation, osmotic stress and UV irradiation stimulate epiplakin-keratin association
To investigate whether the phosphorylation status of keratins influenced their binding to epiplakin, primary keratinocytes were treated with the phosphatase inhibitor okadaic acid (OA) to specifically induce serine/threonine hyperphosphorylation. Immunofluorescence microscopy of cells exposed to OA for 2 hours revealed a predominantly filamentous staining pattern of epiplakin and a complete colocalization of the protein with keratin filaments (Fig. 3A). After 4 hours of OA exposure, keratin filaments were partially disrupted, with epiplakin being present in both the residual thick perinuclear filament bundles and the keratin granules, which appeared mainly at the cell periphery (Fig. 3A). After 6 hours, keratin filaments had disappeared, and keratin as well as epiplakin was found exclusively in form of granules spread all over the cytoplasm (Fig. 3A). Next, we monitored OA-induced epiplakin-keratin association by immunoblotting and protein staining of soluble and insoluble cellular subfractions during OA treatment (Fig. 3B). In untreated keratinocytes (lanes 0h) epiplakin was found in both the insoluble and the soluble fractions (upper two panels), whereas keratins, as expected, were mostly insoluble (lower panel; and data not shown). The proportion of epiplakin in the soluble fraction decreased upon incubation with OA, and after 6 hours, almost all epiplakin was found associated with the insoluble keratin-containing fraction. Increased association of epiplakin with insoluble keratins correlated with increased phosphorylation of keratins, as demonstrated by immunoblotting using phospho-epitope-specific antibodies to K5. The phosphorylation status of epiplakin was found to be unchanged during OA incubation (data not shown).
As OA-induced serine/threonine hyperphosphorylation of keratins led to their association with epiplakin, it was of interest to examine whether a tyrosine phosphatase inhibitor such as orthovanadate (OV) would have similar effects. Strikingly, we observed an entirely filamentous distribution of epiplakin and full colocalization with keratins after a 2 minute incubation of cells with OV (Fig. 4). After a 10 minute incubation with OV, filamentous structures were no longer visible. The entire keratin networks had collapsed, forming numerous aggregates of various sizes, with undistinguishable staining patterns of both keratin and epiplakin (Fig. 4).
Osmotic shock assays, such as treatment with 150 mM urea, serve as a useful tool for monitoring keratin network properties and their alterations in epidermolysis bullosa simplex (EBS) caused by keratin mutations (Morley et al., 1995; D'Alessandro et al., 2002). To examine whether this type of stress, in a similar manner to hyperphosphorylation, leads to an association of epiplakin with keratins, primary mouse keratinocytes incubated with urea for 5 minutes were double labeled for pan-keratin and for epiplakin (Fig. 4). Indeed, compared with untreated cells, the staining pattern of epiplakin appeared more filamentous under these conditions (compare Fig. 1A with Fig. 4), indicating increased epiplakin-keratin association.
UV irradiation of keratinocytes has been reported to cause a redistribution of keratin networks into ring-like structures around the nucleus (Zamansky et al., 1992). As shown in Fig. 4, UV irradiation of primary keratinocytes at 400 J/m2 led to a conspicuous colocalization of epiplakin with such perinuclear structures.
OA-induced keratin filament disruption in differentiated keratinocytes proceeds faster in the absence of epiplakin
The tendency of epiplakin to associate with keratin structures under conditions leading to filament reorganization suggested that the protein might play a role in the cellular stress response. To examine whether epiplakin had any effect on the rate of OA-induced filament disruption, we subjected primary cultures of undifferentiated epiplakin-deficient keratinocytes to a time-course experiment. Immunofluorescence microscopy carried out 2, 4 and 6 hours after OA addition revealed no differences in keratin network disruption patterns in these cells compared with the wild type (Fig. 3; compare keratin staining of +/+ and –/– cells). Similarly, no detectable differences in keratin filament disruption or rearrangement were observed between the two cell types when treated with OV or urea, or with UV irradiation (Fig. 4).
A different situation was observed when epiplakin-deficient and wild-type keratinocytes that had undergone Ca2+-induced differentiation were subjected to a similar experiment. Under these conditions, OA-induced keratin network disruption in wild-type cells proceeded more slowly than in undifferentiated cells (Fig. 5A). In undifferentiated wild-type cells, keratin was found largely in the form of aggregates at the periphery of cells at the 4 hour time point, with just a filamentous ring (eventually also becoming fragmented within the next 2 hours) persisting around the nucleus (Fig. 3A). By contrast, in differentiated cells, keratin aggregates were rarely observed before the 6 hour time point, and even then, the majority of cells displayed just partially disrupted filament networks (Fig. 5A). Interestingly, however, in differentiated epiplakin-deficient cells, the keratin filament system seemed to be more susceptible to hyperphosphorylation. Partially disrupted filament systems became visible in these cells after 2 hours, and the formation of aggregates appeared to progress faster (Fig. 5A). For a statistical analysis, we counted wild-type and epiplakin-deficient cells exhibiting filamentous, partially disrupted (as exemplified by cells indicated with arrows in Fig. 5A) and totally disrupted (Fig. 5A, cells indicated with arrowheads) keratin filaments at 1 hour intervals over a period of 6 hours. As the time courses – including the onsets of the intermediate filament rearrangements – varied between different experiments, we calculated the percentages of cells with changed filament network organization during the first 2 hours after the onset of filament breakdown (Fig. 5B). In almost half (∼46%) of epiplakin-deficient cells, keratin filaments were at least partially disrupted at this time point, whereas only ∼10% of wild-type cells were in this state. Similarly, the percentage of cells with totally disrupted filament networks (displaying exclusively aggregated keratins), calculated over a 2 hour period (starting from the first time point when cells of that type were appearing) was significantly higher for epiplakin-deficient cells than for wild-type cells (Fig. 5B). The faster disruption of filamentous intermediate filament networks, paired with an accelerated appearance of aggregated keratins, was noticeable in four independent experiments. The entire time course for one of these experiments is shown in Fig. 5C. To shed more light on the mechanism underlying increased filament disruption, we next investigated whether it was coupled with accelerated phosphorylation of keratins by western blot analysis using antibodies to phospho-K5. However, epiplakin-deficient and wild-type cells showed no significant differences in the extent or rate of keratin phosphorylation (data not shown). Since MAP-kinases, especially p38, are thought to play a pivotal role in the stress response, we also measured the activation of p38, ERK and JNK MAP kinases using phospho-specific antibodies to each of these proteins. Again, we could not detect any differences in the activation patterns of these kinases over the time course of OA-induced filament breakdown between differentiated wild-type and epiplakin-deficient cells (data not shown).
As the tyrosine phosphatase inhibitor OV led to a similar breakdown of keratin filaments (paired with the recruitment of epiplakin to the resulting aggregates) as serine/threonine phosphatase inhibitors, it was of interest to examine whether epiplakin deficiency had an accelerating effect on keratin filament disruption in this situation. Because of the faster and more distinct disruption pattern observed, only two types of cells could be distinguished: those with an intact keratin filament system and those with a disrupted network. When OV-treated epiplakin–/– and wild-type cells were statistically analyzed at 3-minute intervals over a period of 9 minutes, an acceleration of filament disruption was again found in epiplakin-deficient compared with wild-type cells. After 3 minutes of OV incubation, there was already a significant difference, which then continued over time (Fig. 5D).
Individual PRDs expressed by force mimic endogenous epiplakin without affecting keratin cytoarchitecture
To investigate whether fragments of epiplakin had any influence on keratin filament organization when overexpressed in keratinocytes, module 9, linker 9, PRD9 and PRD5 were expressed in keratinocytes as GFP-fusion proteins. Immunofluorescence microscopy of transfected cells revealed a partially filamentous staining pattern resembling that of endogenous (full length) epiplakin only for PRD9, whereas a diffuse distribution throughout the cells was observed in all other cases (Fig. 6, left column). In no case did the overexpression of epiplakin fragments lead to a change in keratin filament organization (data not shown). Furthermore, we examined whether OA-induced association of cellular epiplakin with keratin structures could be mimicked by overexpressed epiplakin fragments. Both, module 9 and PRD9, showed near-complete colocalization with keratin aggregates after a 5 hour incubation with OA, whereas linker 9 and PRD5 remained diffusely distributed throughout the cell (Fig. 6, +OA), confirming the blot overlay data shown in Fig. 1.
Ever since the unique, highly ordered structure of epiplakin with up to 16 repeat domains became known, speculation has arisen about the function of these domains, especially after it was shown that the most C-terminal repeat domain of human epiplakin (PRD 13) binds to keratin in vitro (Jang et al., 2005). In a recent study, Wang et al. (Wang et al., 2006) reported keratin binding for fragments corresponding to module 8 (with and without the ensuing linker 9) and, at variance with Jang and colleagues, to linker 9 alone. Studying mouse epiplakin, we show here that in addition to PRD8 and PRD16 (the most C-terminal PRD), 15 out of the 16 PRDs of epiplakin have the ability to bind to K5 in blot overlay assays. Binding may in fact also have occurred to K6, which is expressed along with K5 in immortalized keratinocytes, and displays a similar electrophoretic mobility. Additionally, PRD9 to PRD15 were also found to bind to K14. Our data further suggest that the modules, not the linkers, are responsible for K5 binding. Nevertheless, the combination of modules and linkers seems to be required for binding to K14. In general, the blot overlay assays confirm earlier results obtained with human epiplakin, except that strong K14 binding, but no K5 binding was reported for the last PRD (Jang et al., 2005). Whether this difference results from the use of distinct assays, or indeed reflects different properties of the human and mouse protein species, remains to be clarified.
Our data indicate that epiplakin binds to both monomeric and polymeric keratins. Binding to the monomeric form was evident from the blot overlay assays, confirming data obtained for the human protein (Jang et al., 2005). However, co-sedimentation of epiplakin with insoluble keratin filaments, as well as its in vivo association with keratin networks revealed by confocal immunofluorescence microscopy, support the notion that binding occurs to polymeric keratins, as has also been demonstrated by Jang et al. (Jang et al., 2005) and Wang et al. (Wang et al., 2006).
What are multiple keratin-binding sites used for? One plausible answer could be that epiplakin plays a role in filament reorganization during and/or after cellular stress. This explanation is supported by the observation that epiplakin colocalizes with keratin filaments after UV irradiation of keratinocytes and after subjecting them to osmotic stress. Furthermore, it colocalized with hyperphosphorylated keratin. By contrast, plectin, another spectraplakin protein family member and the closest relative to epiplakin, associates with keratin aggregates only after tyrosine phosphatase inhibition with orthovanadate (Strnad et al., 2001), whereas it was no longer detectable in the cytokeratin fraction of keratinocyte lysates after incubation with OA for more than 2 hours (Osmanagic-Myers et al., 2006). The different composition of keratin granules, which is apparently dependent on the type of hyperphosphorylation achieved with different phosphatase inhibitors, seems to determine the properties of the granules. For example, plectin-positive keratin granules, which form rapidly during mitotic filament breakdown, appear to be rather immobile (Strnad et al., 2001) because of their anchorage into the cytoskeleton via plectin. It has been proposed that such anchorage might be a prerequisite for the quick and successful reintegration of keratin into a filamentous network (Strnad et al., 2002). By contrast, the OA-induced filament breakdown into 14-3-3 protein-positive granules and the recovery after drug removal occurred much more slowly, with each taking several hours to be completed (Strnad et al., 2001; Strnad et al., 2002). It is thought that 14-3-3 protein association with keratin granules keeps them soluble (Liao and Omary, 1996; Liao et al., 1997), and might abolish their capacity to quickly re-establish a fully spread filament system (Strnad et al., 2002). As epiplakin was found in both types of granules, it might have function(s) that are independent of the type of aggregate. One model could be that epiplakin, with its 15 keratin-binding modules, serves as a chaperone-like protein during filament remodeling. This would be consistent with the more rapid OA- and OV-induced filament breakdown in differentiated epiplakin-deficient keratinocytes compared with wild-type cells, especially as we could not detect any faster or stronger activation of MAP-kinase-based stress-response pathways in the absence of epiplakin. Furthermore, the phosphorylation status of K5 was not affected in epiplakin–/– cells, making a mechanical role of epiplakin during the stress response more likely. Therefore, we suggest that epiplakin plays a protective function in keratin filament networks. It is conceivable that epiplakin spans a number of keratin subunit proteins within the filaments, serving as a stabilizer for the filament system when cells are subjected to stress, although a more indirect mechanism of stabilization cannot be ruled out. That this protective effect was noticed only in cells that had undergone differentiation, but not under basal conditions prompts us to speculate that epiplakin executes this function mainly in cell layers where filament systems span more cells, such as in the upper epidermis. Nevertheless, the initial analysis of mice deficient in epiplakin did not show any skin abnormalities, and even challenging the skin by tape-stripping or wounding did not reveal any discernible phenotype (Goto et al., 2006; Spazierer et al., 2006). However, this kind of mechanical stress might not have affected the keratin filament network system in the same way that phosphatase inhibitors do. The use of other (organ-specific) injury models with epiplakin-deficient mice might therefore shed more light on the putative filament-stabilizing function of epiplakin. Furthermore, with its 15 keratin-binding repeat domains, epiplakin might not only play a role in conserving the integrity of filaments by stabilizing them against disruption, but also in the recovery of filaments after drug removal. This issue could be addressed in the future by examining the dynamic properties of keratin during filament remodeling in epiplakin-deficient keratinocytes.
The mechanism underlying the rapid transition from a dotted and partially filamentous epiplakin staining pattern to one of complete colocalization of epiplakin with the keratin network, as observed after OA and OV treatment of keratinocytes, is still unknown. Our data suggest that hyperphosphorylation of keratins triggers the relocalization, because in resting keratinocytes, only part of the epiplakin pool showed association with keratin filaments, whereas all of it was recruited to the keratin network and developing aggregates upon stress exposure. Interestingly, in contrast to primary keratinocytes, the epiplakin staining pattern in primary hepatocytes was completely filamentous, showing a full overlap with the K8/18 intermediate filament network. Epiplakin was also found to colocalize with K8/18 in the human colon carcinoma cell line HT29 (our unpublished results) and in HeLa cells (Jang et al., 2005). Furthermore, the distribution of epiplakin in liver matches that of keratins (Spazierer et al., 2003; Spazierer et al., 2006). Therefore, the question arises, whether the subcellular localization of epiplakin is dependent on the type of cell or on the type of keratin involved. As epiplakin has been shown to colocalize with K5/K14 filaments in HaCaT cells almost perfectly (Jang et al., 2005), it seems likely that the degree of colocalization of epiplakin with keratin filaments is dependent on the cell type.
Unlike other spectraplakins, such as plectin and envoplakin (Wiche et al., 1993; DiColandrea et al., 2000), overexpression of epiplakin fragments in immortalized keratinocytes did not affect keratin filament organization. However, like full-length epiplakin, even single epiplakin PRDs showed a stronger association with hyperphosphorylated keratins after OA-induced filament disruption. Only PRD5 did not colocalize with keratins, in full agreement with blot overlay assays, identifying it as the only one of the 16 repeat domains without keratin-binding activity. Regarding the function of PRD5 the question remains as to whether it serves as a docking station for non-keratin proteins or simply acts as a spacer between the other keratin-binding PRDs.
In summary, our studies revealed a relocalization of epiplakin to the keratin intermediate filament network under various types of cellular stress, suggesting a role for epiplakin in the stress response. In line with this, we showed that phosphatase-inhibitor-induced keratin filament breakdown in differentiated keratinocytes occurred more rapidly in the absence of epiplakin. Based on these results, applying organ-specific injury models to epiplakin-deficient mice would appear a feasible and promising approach for future studies to investigate the function of epiplakin in more detail.
Materials and Methods
The following primary antibodies were used for immunoblotting (IB) and immunofluorescence microscopy (IFM): affinity-purified rabbit antibodies to epiplakin (IB and IFM) (Spazierer et al., 2003); antiserum AF 138 to K5 (IB; Covance, Princeton, NJ); monoclonal antibody (mAb) LP 34 to K5/6/18 (pan-keratin) (IFM; Dako-Cytomation, Glostrup, Denmark); mAb LL001 to K14 (IB; kindly provided by J. M. Leigh, Royal London School of Medicine and Dentistry, London, UK); mAB LJ4 to phosphorylated K4/5/6/8 [(Liao et al., 1997; Toivola et al., 2002) kindly provided by M. B. Omary, Palo Alto VA Medical Center, Palo Alto, CA)]; mAb B-5-1-2 to tubulin (IB; Sigma, St Louis, MO), PRB-140C to involucrin (IB; Covance); a mixture of mAbs DP-2.15, DP-2.17 and DP-2.20 to desmoplakin (IFM; Progen, Heidelberg, Germany); a combination of mAbs Ks 8.7 and Ks 18.04 to K8 and K18 (IFM; Progen); and mAb G1160 to GST (IB; Sigma). Secondary antibodies were goat anti-rabbit and goat anti-mouse IgGs conjugated to Alexa Fluor 488 (IFM), Texas red (IFM), AP (IB) or HRP (IB) and were purchased from Jackson Immuno-Research Laboratories (West Grove, PA).
Expression of recombinant proteins
The specified epiplakin fragments PRD1 (encoding amino acids 1-238), PRD2 (239-476), PRD3 (477-806), PRD4 (807-1123), PRD5 (1124-1442), PRD6 (1443-1767), PRD7 (1768-2093), PRD8 (2094-2425), PRD9 (2426-2939), PRD16 (6032-6545), Li8 (2094-2232), M8 (2233-2425), Li9 (2426-2747) and M16 (6353-6545) to be used in overlay and pull-down experiments were obtained by subcloning the corresponding cDNAs into the expression vector pGEX4T-1 (Clontech, Palo Alto, CA), expression as bacterial GST-fusion protein and purification as described (Spazierer et al., 2003). For GST pull-down assays, recombinant GST-fusion proteins were coupled to glutathione Sepharose 4B beads (Amersham Biosciences, Little Chalfont, UK) according to the manufacturer's instructions.
Cell culture, fractionation and immunofluorescence microscopy
Primary keratinocytes were isolated from 2- to 3-day-old mice as described previously (Spazierer et al., 2006), and grown in keratinocyte growth medium (KGM; Cambrex, East Rutherford, NJ) containing 0.1 mM CaCl2. For differentiation of keratinocytes, KGM containing 1.3 mM CaCl2 was used. Immortalized p53–/– basal keratinocytes used for cell transfection experiments were derived from p53–/– mice (Andrä et al., 2003). For the isolation of primary hepatocytes, livers from E12.5 mouse fetuses were mechanically dissociated and plated onto plastic dishes in DMEM containing 10% FCS (Eferl et al., 1999). For fractionation, cells were washed twice with PBS and lysed with 150 mM NaCl, 1.5 M KCl, 1 mM EDTA, 0.5% Triton X-100, 10 mM Tris-HCl, pH 7.5 (high-salt buffer), supplemented with protease inhibitors (Complete Mini; Roche Applied Science, Indianapolis, IN), for 30 minutes at 4°C. After centrifugation at 16,000 g for 10 minutes, the supernatant (soluble fraction) was removed, and the insoluble (keratin) fraction was solubilized in 8 M urea. For immunofluorescence microscopy, cells were fixed with methanol at –20°C and further processed for immunofluorescence microscopy as described (Spazierer et al., 2006).
Immunoblotting and blot overlay assay
For immunoblotting, cells were lysed in 150 mM NaCl, 0.1% SDS, 1% Triton X-100, 1% deoxycholate, 5 mM EDTA, 10 mM Tris-HCl, pH 7.2 (RIPA buffer). Proteins were separated by SDS-PAGE, transferred to nitrocellulose sheets and visualized as previously described (Spazierer et al., 2003). The protocol followed for blot overlay assays was similar to that for immunoblotting, except that prior to immunodetection, membranes were incubated in PBS containing 10 μg/ml of the proteins to be bound.
Keratinocytes were lysed with 1% Empigen BB (Calbiochem) in PBS, supplemented with 10 mM EDTA, 0.5 mg/ml OA (Sigma) and Complete Mini protease inhibitor tablets. The lysates were cleared by centrifugation at 16,000 g for 15 minutes, and incubated with the recombinant protein-coupled glutathione-Sepharose 4B beads over night at 4°C. Beads were then washed extensively with PBS and finally resuspended in SDS-PAGE buffer for immunoblotting.
Exposure of cells to okadaic acid, orthovanadate, urea and UV
OA and OV were used at final concentrations of 100 μg/ml and 40 mM, respectively. Osmotic stress was induced by incubation of cells with 150 mM urea in PBS for 5 minutes at 37°C (Lane and Pekny, 2004). Cells were exposed to UV radiation at a dose of 400 J/m2 (Toivola et al., 2002), while kept in a minimum volume of KGM. After irradiation, cells were incubated in KGM for 24 hours at 37°C before being processed further.
Epiplakin cDNA fragments were cloned into the eukaryotic expression vector pEGFP-N2 (Clontech). For transfection, 5 μg DNA and 8 μl Fugene® reagent (Roche) were added to 200 μl KBM, incubated for 30 minutes and then added to 6 cm culture dishes containing 2.5 ml growth medium. Incubation was carried out for 24 hours at 37°C.
We thank J. M. Leigh and M. B. Omary for generous gifts of antibodies. This work was supported by a grant from DEBRA UK and grant P17862-B09 from the Austrian Science Research Fund.