Plakoglobin but Not Desmoplakin Regulates Keratinocyte Cohesion via Modulation of p38MAPK Signaling

ORIGINAL ARTICLE

Plakoglobin but Not Desmoplakin Regulates Keratinocyte Cohesion via Modulation of p38MAPK Signaling Volker Spindler1,3, Carina Dehner1,3, Stefan Hu¨bner2 and Jens Waschke1 Plakoglobin (Pg) and desmoplakin (DP) are adapter proteins within the desmosome, providing a mechanical link between desmosomal cadherins as transmembrane adhesion molecules and the intermediate filament cytoskeleton. As in the severe skin blistering disease pemphigus, autoantibodies against desmosomal adhesion molecules induce loss of keratinocyte cohesion at least in part via p38 mitogen-activated protein kinase (p38MAPK) activation and depletion of desmosomal components, we evaluated the roles of Pg and DP in the p38MAPK-dependent loss of cell adhesion. Silencing of either Pg or DP reduced cohesion of cultured human keratinocytes in dissociation assays. However, Pg but not DP silencing caused activation of p38MAPKdependent keratin filament collapse and cell dissociation. Interestingly, extranuclear but not nuclear Pg rescued loss of cell adhesion and keratin retraction. In line with this, Pg regulated the levels of the desmosomal adhesion molecule desmoglein 3 and tethered p38MAPK to desmosomal complexes. Our data demonstrate a role of extranuclear Pg in controlling cell adhesion via p38MAPK-dependent regulation of keratin filament organization. Journal of Investigative Dermatology (2014) 134, 1655–1664; doi:10.1038/jid.2014.21; published online 6 February 2014

INTRODUCTION Both plakoglobin (Pg) and desmoplakin (DP) are proteins of the desmosomal adhesive complex (Waschke, 2008; Delva et al., 2009; Brooke et al., 2012). Desmosomes are found in epithelial cells and in a variety of cells of non-epithelial origin. They are abundant in tissues that are exposed to high degrees of mechanical stress, such as the epidermis, the oral mucosa, or the heart muscle. The core of desmosomes is built up of desmogleins (Dsgs) and desmocollins, both cadherin-type adhesion molecules (Thomason et al., 2010). They are integral membrane proteins that bind their counterparts of the opposing cell with the extracellular domains and connect indirectly to the intermediate filament cytoskeleton. The latter link is provided by so-called plaque proteins, including the 1

Institute of Anatomy and Cell Biology, Department I, Ludwig-MaximiliansUniversita¨t, Munich, Germany and 2Institute of Anatomy and Cell Biology, Julius-Maximilians-Universita¨t, Wu¨rzburg, Germany

3

These authors contributed equally to this work.

Correspondence: Jens Waschke, Institute of Anatomy and Cell Biology, Ludwig-Maximilians-Universita¨t, Pettenkoferstrasse 11, 80336 Munich, Germany. E-mail: [email protected] Abbreviations: Ab, antibody; DP, desmoplakin; Dsg3, desmoglein 3; GFP, green fluorescent protein; NES, nuclear export sequence; NLS, nuclear localization sequence; pAb, polyclonal antibody; Pg, plakoglobin; PV-IgG, IgG fraction of a pemphigus vulgaris patient; p38MAPK, p38 mitogen-activated protein kinase; siRNA, small interfering RNA Received 15 July 2013; revised 25 November 2013; accepted 10 December 2013; accepted article preview online 17 January 2014; published online 6 February 2014

& 2014 The Society for Investigative Dermatology

Armadillo family proteins Pg and plakophilins, as well as DP. Probably all plaque proteins interact with Dsgs and desmocollins, but it is DP that predominantly mediates the link to the keratins of the intermediate filament cytoskeleton (Smith and Fuchs, 1998). In contrast to DP, other functions of Pg that are not directly related to cell adhesion are well established (Zhurinsky et al., 2000a). Pg is present not only at intercellular junctions but is also detectable in cytosolic pools as well as in the nucleus. Pg has been shown to bind the t-cell factor/lymphoid-enhancer factor (TCF/LEF) transcription factors and to regulate proliferation (Huber et al., 1996; Simcha et al., 1998; Charpentier et al., 2000) and apoptosis (Dusek et al., 2006). DP is well established to primarily serve as an adapter molecule required for cell cohesion, but has also recently been shown to inhibit Wnt/b-catenin signaling and thus proliferation (Yang et al., 2012). Murine whole-body knockout models of Pg or DP are lethal in utero (Bierkamp et al., 1996; Ruiz et al., 1996; Gallicano et al., 1998); epidermis-specific depletions, however, revealed contrasting phenotypes. Whereas depletion of Pg induces mild adhesive defects and a hyperkeratotic, inflammatory phenotype resembling striate palmoplantar keratoderma (Li et al., 2012), reduction of epidermal DP causes severe loss of intercellular adhesion, which is pronounced in the basal layer of the epidermis (Vasioukhin et al., 2001). A similar phenotype with reduced adhesion is seen in patients suffering from the disease pemphigus (Amagai and Stanley, 2012). Here, circulating autoantibodies target desmoglein 3 (Dsg3) www.jidonline.org 1655

V Spindler et al. Plakoglobin and p38MAPK-Dependent Cell Adhesion

Fragments per well

80

*

60

40

*

20

PB S AK 23 PV -Ig G

0

DP

AK23

PBS

Pg

PV-IgG

Pg Dsg3 DP

RESULTS

1656 Journal of Investigative Dermatology (2014), Volume 134

210 kDa

Dsg3

130 kDa

Pg

82 kDa

GAPDH

37 kDa

*

0.8

* * *

0.6 0.4 0.2 0 PBS

DP

1

PV-IgG

PV-IgG

AK23

First, we investigated whether interference with cell adhesion, specifically by loss of Dsg trans-interaction via pemphigus autoantibodies, leads to alterations in Pg and DP distribution and protein levels. Confluent human HaCaT keratinocytes were incubated for 24 hours with AK23, a monoclonal Dsg3 antibody (Ab) derived from a pemphigus mouse model, or with PV-IgG, an IgG fraction of a patient suffering from PV. Under these conditions, both Ab fractions induced profound loss of cell adhesion after 24 hours as detected by dispasebased dissociation assays (Figure 1a). Compared with controls (4.4±1.3), the fragment numbers were increased by AK23 application to 28.7±3.4 and by PV-IgG incubation to 54.1±7.6, respectively. By immunostaining, the linear membrane distribution of Pg in untreated HaCaT keratinocytes appeared fragmented following AK23 incubation. PV-IgG even had a more disturbing effect on membrane localization of Pg. DP localization, which—in line with its distribution exclusively within desmosomes—displayed a punctate pattern at the cell membrane in controls, was only mildly affected by autoantibody incubation. Depletion of a variety of desmosomal proteins, in particular Dsg3, by pemphigus autoantibodies is well established. Similarly, the PV-IgG fraction applied here resulted in decreased Dsg3 levels as demonstrated by western blotting of whole-cell lysates (Figure 1c). Interestingly, PV-IgG also reduced Pg protein

PBS

Pemphigus autoantibodies induced clustering and depletion of Pg

Band density (fold of PBS)

1.2

AK23

and/or Dsg1, which then leads to loss of keratinocyte cohesion. The most common variant, pemphigus vulgaris (PV), is characterized by mucosal or suprabasal epidermal blistering, a phenomenon referred to as acantholysis (Stanley and Amagai, 2006). It is by now well established that both direct inhibition of extracellular Dsg3 trans-interaction as well as altered intracellular signaling are necessary to induce pronounced cell dissociation (Waschke, 2008; Getsios et al., 2010). With respect to the latter, p38 mitogen-activated protein kinase (p38MAPK) activation apparently is important, as it is rapidly phosphorylated following autoantibody binding both in vitro and in vivo, whereas inhibition of p38MAPK signaling blocks loss of adhesion (Berkowitz et al., 2005, 2006; Spindler et al., 2013). Moreover, Pg has been demonstrated to be involved in the pathogenesis of PV, as Pgdeficient keratinocytes were protected against the pathogenic effects of PV autoantibodies (Caldelari et al., 2001). Furthermore, it was shown that PV antibodies induce nuclear c-myc activation through abrogation of the inhibitory effect of Pg on c-myc expression (Williamson et al., 2006). DP also may have a role in PV, as its membrane distribution has been shown to appear fragmented following autoantibody binding (de Bruin et al., 2007; Mao et al., 2009; Saito et al., 2012). Our group has recently demonstrated that a signaling complex consisting of Pg, Dsg3, and p38MAPK exists, and that p38MAPK activation following loss of Dsg3 trans-interaction via PV autoantibodies causes keratin filament retraction and loss of cell cohesion (Spindler et al., 2013). Therefore, in this study, we aimed to investigate the role of DP and Pg in the context of p38MAPK activation and intercellular adhesion.

Figure 1. Pemphigus autoantibodies disrupted the distribution of plakoglobin (Pg). (a) A 24-hour incubation of AK23 or pemphigus vulgaris (PV)-IgG reduced cell cohesion as detected by dispase-based dissociation assays. n ¼ 7, *Po0.05 versus control. (b) AK23 and PV-IgG disrupted the linear staining of Pg at the membrane and, minor pronounced, of desmoplakin (DP). Bar ¼ 20 mm. Panel shows representative images from three independent experiments. (c) Immunoblots of whole-cell lysates demonstrate that AK23 and PV-IgG reduce Pg and desmoglein 3 (Dsg3) protein content, but not that of DP (n ¼ 5, *Po0.05 vs. phosphate-buffered saline (PBS)). GAPDH, glyceraldehyde-3-phosphate dehydrogenase.

content, with AK23 showing less pronounced effects. In contrast, levels of DP remained unchanged. Thus, autoantibody-mediated loss of cell cohesion is accompanied by aberrant Pg distribution and reduced protein content.

V Spindler et al. Plakoglobin and p38MAPK-Dependent Cell Adhesion

# 180 160

#

Fragments per well

140 120 100 80

#

*

60

*

40 20

PB S AK 23 PB S AK 23 PB S AK 23

0

siDP

1.2

Dsg3

130 kDa

Phospho-p38

42 kDa

p38

42 kDa

GAPDH

37 kDa

0.8

*

0.6

*

0.4 0.2

p-p38/p38 band density (fold of siNT)

210 kDa

*

1.4 1.2 1 0.8 0.6 0.4 0.2

0

0

siNT siDP siPg

82 kDa

DP

Band density (fold of siNT)

Pg

1.6 DP Pg

1

siNT siDP siPg

siPg

siPg

siDP

siNT

siNT

42 kDa

P38

42 kDa

β-Actin

43 kDa

1 0.5 0 PV-lgG

Phospho-p38

*

PBS

82 kDa

*

1.5

AK23

210 kDa

Pg

*

2

siNT

DP

p-p38/p38 band density (fold of siNT)

siPg PV-lgG

siNT PV-lgG

siPg AK23

siNT AK23

siPg PBS

siNT PBS

2.5

siPg

Figure 2. Silencing of plakoglobin (Pg) induced loss of cell cohesion and p38 mitogen-activated protein kinase (p38MAPK) activation. (a) Silencing of either Pg (siPg) or desmoplakin (siDP) reduced intercellular adhesion as detected by dispase-based dissociation assays. n ¼ 6, *Po0.05 versus nontarget siRNA (siNT) phosphate-buffered saline (PBS); #Po0.05. (b) Silencing of Pg but not of DP led to increased p38MAPK phosphorylation detected by western blots of whole-cell lysates. n ¼ 4, *Po0.05 versus siNT. (c) A 30-minute AK23 or pemphigus vulgaris (PV)-IgG treatment did not further increase p38MAPK activation induced by Pg silencing as indicated by immunoblots of whole-cell lysates. n ¼ 3, *Po0.05 versus siNT.

Depletion of Pg but not of DP resulted in p38MAPK-mediated cell dissociation and keratin retraction

As PV-IgG reduced Pg protein levels and also affected DP distribution, we silenced either Pg or DP by small interfering RNA (siRNA) in HaCaT cells and evaluated the effects on cell adhesion by dissociation assays. Transfection of nontarget siRNA yielded fragment numbers of 7.4±1.1 (Figure 2a); however, these numbers were significantly increased under conditions of reduced Pg (55.8±6.8) and DP (50.3±7.3)

expression. Interestingly, in western blots of whole-cell lysates, robust p38MAPK phosphorylation was evident under conditions of siRNA-reduced Pg protein content only, which was accompanied by reduced Dsg3 levels (Figure 2b). Knockdown efficacies of Pg and DP were similar (Figure 2b, right panel). Activation of p38MAPK following Pg but not DP silencing was also evident in another human epidermal cell line, SCC9 cells. (Supplementary Figure S1A online). Incubation of HaCaT cells depleted for Pg or DP with AK23 for 24 hours still resulted in higher fragment numbers (166.4±9.7 and 127.2±7.3, respectively, Figure 2a). However, AK23 or PV-IgG did not further increase p38MAPK phosphorylation in cells with siRNAreduced Pg expression as detected by western blotting of whole-cell lysates (Figure 2c). To substantiate that the effect of p38MAPK activation was involved in cell dissociation, we used the p38MAPK-specific inhibitor SB202190. In Pg-depleted cells, fragment numbers were reduced from 89.7±9.8 to 42.8±8.0 after 24 hours of p38MAPK inhibition—i.e., to levels not different from those in experiments using SB202190 alone (Figure 3a). In contrast, in cells silenced for DP, SB202190 did not reduce fragment numbers (58.8±5.0 vs. 59.8±6.8 fragments). Co-incubation of SB202190 with AK23 for 24 hours reduced cell dissociation in the absence or presence of Pg and DP silencing. The same effects were detectable in SCC9 cells (Supplementary Figure S2B online). Under conditions of Pg silencing, AK23induced elevation of fragment numbers (151.2±12.8) was blocked by SB202190 in HaCaT cells (54.3±8.0). Similarly, AK23-mediated cell sheet fragmentation was abrogated by SB202190 under conditions of reduced DP levels (132.2±8.7 vs. 48.0±8.9 fragments). Taken together, p38MAPK inhibition ameliorated reduced cell adhesion induced by Pg silencing but not caused by DP depletion. The collapse and clustering of keratin filaments within keratinocytes is a hallmark of pemphigus, referred to as keratin retraction (Waschke, 2008). Previously, we have shown that the keratin filament collapse induced by PV-IgG and AK23 is mediated via p38MAPK activation, which occurs subsequent to the loss of Dsg3 interaction (Spindler et al., 2013). However, it is unknown whether Pg or DP are involved in this process. In cells depleted for Pg, a marked collapse of keratin filaments was obvious (Figure 3b, middle line). A 24-hour incubation with SB202190 largely abolished this effect, indicating that p38MAPK activation causes keratin retraction following depletion of Pg (Figure 3b, bottom line). Interestingly, DP depletion did not result in a complete retraction of keratin filaments, as filaments appeared weaker underneath the membrane, but still reached cell borders in areas where neighboring cells were in close contact (arrows). SB202190 had no effect on the distribution of keratin filaments after DP silencing (Figure 3c). These effects were also detectable in SCC9 cells (Supplementary Figure S1C and D online). Taken together, silencing of either Pg or DP was sufficient to induce loss of cell cohesion of human keratinocytes. Interestingly, the underlying mechanisms appear to be different, as Pg but not DP depletion induced keratin filament collapse via p38MAPK activation. www.jidonline.org 1657

V Spindler et al. Plakoglobin and p38MAPK-Dependent Cell Adhesion

* 160

*

140

Fragments per well

120

*

100 80

*

60 40 20

AK

SB P 20 BS 21 23 90 + A SB K2 20 3 21 90 SB PB 20 S AK 21 23 90 + A SB K2 20 3 21 90 SB PB 20 S AK 21 23 9 + AK 0 SB 23 20 21 90

0

siNT

panCK

Merge

siDP

siPg siPg + SB202190

DP

siNT

Merge

siDP

siDP + SB202190

panCK

siNT

Pg

siPg

Figure 3. Loss of cell cohesion and keratin filament retraction induced by plakoglobin (Pg) silencing was blocked by p38 mitogen-activated protein kinase (p38MAPK) inhibition. (a) Cell dissociation by Pg depletion (siPg) was blocked by the p38MAPK-specific inhibitor SB202190, whereas p38MAPK inhibition had no effect in desmoplakin (siDP)-depleted cells. SB202190 blocked the effect of AK23 independent of Pg and DP expression state. n ¼ 6, *Po0.05. (b) Pg silencing (siPg) induced keratin filament retraction (arrows), which was prevented by SB202190 (arrowheads). Bar ¼ 20 mm. Panel shows representative images from three independent experiments. (c) Knockdown of DP (siDP) did not induce pronounced keratin filament collapse (arrows) and SB202190 had no effect (arrowheads). Bar ¼ 20 mm. Panel shows representative images from three independent experiments. siNT, nontarget siRNA.

Extranuclear Pg ameliorated cell dissociation, p38MAPK activation, and keratin filament retraction

In addition to its adhesive function, Pg regulates a variety of transcription factors by shuttling to the nucleus (Zhurinsky 1658 Journal of Investigative Dermatology (2014), Volume 134

et al., 2000b) and this has been implicated in loss of cell adhesion in pemphigus (Williamson et al., 2006). Also, MAPKAP kinase 2, the main effector of p38MAPK regarding regulation of cell adhesion (Mao et al., 2013), translocates

V Spindler et al. Plakoglobin and p38MAPK-Dependent Cell Adhesion

from nucleus to the cytosol in a p38MAPK-dependent manner. To dissect whether nuclear or extranuclear localization is important for p38MAPK-mediated cell adhesion, we gene-

3.5

* §

3

* Fold of control

2.5 2

#

*#

1.5 1 0.5

PB S AK 23 PB S AK 23 PB S AK 23 PB S AK 23

0

GFP

Pg-WT

NLS-Pg

NES-Pg

82 kDa

Phospho-p38MAPK

42 kDa

p38MAPK

42 kDa

GAPDH

37 kDa

Pg-WT NLS-Pg NES-Pg

siPg (UTR3′)

GFP

GFP

panCK

Pg

1 0.8

*

*

0.6 0.4 0.2 0

GFP Pg-WT NLS-Pg NES-Pg

Pg-GFP Pg

p-p38/p38 band density (fold of GFP)

NES-Pg

NLS-Pg

Pg-WT

GFP

1.2

Merge

rated Pg-green fluorescent protein (GFP) mutants containing either a nuclear localization sequence (NLS-Pg) or a nuclear export sequence (NES-Pg) (Supplementary Figure S2A online). As expected, NLS-Pg localized primarily to the nucleus, whereas NES-Pg was found predominantly at the plasma membrane (Supplementary Figure S2B online). A wild-type construct (Pg-WT) was detectable in the nucleus, the cytosol, and the membrane, and GFP served as control. In dissociation assays, NLS-Pg (1.7±0.2-fold of controls) but neither NES-Pg (0.8±0.2-fold of controls) nor Pg-WT (1.1±0.2-fold of controls) expression disturbed cell adhesion in HaCaT monolayers (Figure 4a). Furthermore, Pg-WT and NES-Pg (1.7±0.2- and 1.6±0.2-fold of controls, respectively) ameliorated loss of cell adhesion induced by AK23 (2.5±0.2-fold of controls), whereas NLS-Pg had no protective effect. Interestingly, expression of Pg-WT or NES-Pg suppressed basal p38MAPK phosphorylation (Figure 4b). We next induced keratin collapse by reducing the amount of Pg using siRNA directed against the 30 UTR region of the Pg-mRNA (Supplementary Figure S2C online). Under these conditions, keratin retraction was evident in cells expressing either GFP or NLS-Pg (Figure 4c, arrows). In contrast, in cells expressing either Pg-WT or NES-Pg, keratin filament collapse was reduced (Figure 4c, arrowheads). These data indicate that extranuclear localization of Pg, presumably at the plasma membrane, is important for controlling p38MAPK activity and keratin retraction. Furthermore, the observation that NES-Pg ameliorated loss of cell adhesion in response to AK23 suggests that NES-Pg is involved in regulating Dsg3 binding. As Dsg3 and Pg are present both in desmosomes as well as outside of desmosomes in the cell membrane (Yamamoto et al., 2007), we performed triton extraction to investigate both pools separately. Under conditions of Dsg3 depletion following 1-hour PV-IgG exposure, NES-Pg rescued PV-IgG-induced reduction of Dsg3 levels detectable in the soluble pool (Figure 5a). These data may indicate that Pg present at the plasma membrane regulates Dsg3 levels via inhibition of p38MAPK signaling and keratin retraction. Previously we have shown that Dsg3 forms a complex with Pg and p38MAPK (Spindler et al., 2013). Thus, we tested whether Pg exhibits a scaffolding function to locate p38MAPK to the desmosomal

Figure 4. Extranuclear plakoglobin (Pg) ameliorated autoantibody-induced cell dissociation, p38 mitogen-activated protein kinase (p38MAPK) activation, and keratin filament retraction. (a) Expression of a Pg mutant with predominant localization in the nucleus (nuclear localization sequence (NLS)Pg) was sufficient to disrupt HaCaT monolayer integrity. In contrast, extranuclear Pg (nuclear export sequence (NES)-Pg) or wild-type Pg (Pg-WT) reduced AK23-mediated cell dissociation. n ¼ 10, *Po0.05 versus respective phosphate-buffered saline (PBS) condition; #Po0.05 versus green fluorescent protein (GFP) AK23, yPo0.05. (b) Expression of Pg-WT or NES-Pg reduced the amount of active p38MAPK as detected by immunoblots of whole-cell lysates. n ¼ 5, *Po0.05 versus GFP. (c) Under conditions of Pg depletion via small interfering RNA (siRNA) targeting of the 30 UTR, keratin retraction was evident, which was still present in GFP- or NLS-Pg-expressing cells (arrows in first and third line). In contrast, in cells expressing Pg-WT and NES-Pg, keratin filament retraction was reduced (arrowheads in second and fourth line). Bar ¼ 10 mm. Panel shows representative images from three independent experiments.

www.jidonline.org 1659

V Spindler et al.

NES-Pg

GFP

GFP

NES-Pg

Non-cytoskeletal fraction NES-Pg

GFP

GFP

NES-Pg

Cytoskeletal fraction

210 kDa 130 kDa

Pg-GFP Pg β-Actin GAPDH

110 kDa 82 kDa 43 kDa 37 kDa PBS

PBS

#

1.4 1.2 1 0.8

*

0.6 0.4 0.2 0

PV-IgG

N

PV-IgG

Cytoskeletal Non-cytoskeletal

1.6

G F ES P -P g N GF ES P -P g

DP Dsg3

Dsg3 band density (fold of GFP)

Plakoglobin and p38MAPK-Dependent Cell Adhesion

0.4 0.2

sh

N sh

p38

42 kDa

β-Actin

43 kDa

GAPDH

37 kDa

T

1 0.5 0

1h

PB

42 kDa

1.5

G

Phospho-p38

2

PV -Ig

210 kDa

*

2.5

S

DP

IP control

Cytoskeletal Non-cytoskeletal

3

PB

PV-IgG 24 h

PBS 24 h

PV-IgG 1 h

PBS 1 h

Non-cytoskeletal fraction PV-IgG 24 h

PBS 24 h

PV-IgG 1 h

PBS 1 h

Cytoskeletal fraction

p-p38/p38 band density (fold of PBS)

IP DP

Pg

0 N

42 kDa 42 kDa

sh

Phospho-p38MAPK p38MAPK

0.6

PV S -Ig G

82 kDa

** *

0.8

Pg

210 kDa

Pg

p38MAPK p-p38MAPK DP Pg

1

T

DP

1.2

sh

shPg

shNT

IP DP

shPg

shNT

IP control

shPg

shNT

Pure lysate

Band density (fold of shNT IP)

PBS PV-IgG

24 h

Figure 5. Extranuclear plakoglobin (Pg) blocked desmoglein 3 (Dsg3) depletion by pemphigus autoantibodies and tethered p38 mitogen-activated protein kinase (p38MAPK) to the desmosome. (a) Expression of nuclear export sequence (NES)-Pg prevented Dsg3 depletion induced by pemphigus vulgaris (PV)IgG in both the cytoskeleton-bound (Triton X-100 insoluble) and membrane/cytosol (Triton X-100 soluble) pools after 1 hour. n ¼ 5, *Po0.05 versus respective phosphate-buffered saline (PBS) green fluorescent protein (GFP) condition; #Po0.05. (b) Immunoprecipitation (IP) of desmoplakin (DP) from whole-cell lysates revealed that p38MAPK and phospho-p38MAPK form a complex with DP and Pg. Silencing of Pg by shRNA (shPg) reduced the amount of p38MAPK and phosphop38MAPK present in this complex. n ¼ 4, *Po0.05 versus non-target shRNA (shNT). (c) p38MAPK activation was detectable after 1 hour of PV-IgG incubation in the non-cytoskeletal fraction. n ¼ 4, *Po0.05 versus respective phosphate-buffered saline (PBS) condition.

proteins. We immunoprecipitated DP, which is only present in desmosomes, under conditions of Pg silencing via lentiviral expression of Pg shRNA. Indeed, compared with control transductions, the amount of phospho-p38MAPK and p38MAPK attached to DP was substantially reduced after Pg knockdown (Figure 5b). Thus, Pg seems to have a tethering role for p38MAPK in desmosomal complexes. Finally, we investigated the distribution of p38MAPK in the Triton-soluble and -insoluble pools with and without PV-IgG treatment. p38MAPK was mainly localized in the non-cytoskeletal pool; 1660 Journal of Investigative Dermatology (2014), Volume 134

however, a considerable amount was phosphorylated in the cytoskeletal fraction (Figure 5c). Upon treatment with PV-IgG, predominantly soluble p38MAPK was activated after 1 hour. The distribution of p38MAPK did not change after incubation with PV-IgG for 1 hour or 24 hours. DISCUSSION In this study, we provide evidence that Pg regulates cell cohesion via modulation of p38MAPK activity. Our data furthermore indicate that primarily the extranuclear Pg is

V Spindler et al. Plakoglobin and p38MAPK-Dependent Cell Adhesion

responsible for stabilizing cell adhesion via inhibition of p38MAPK and that depletion of Pg is involved in the loss of keratinocyte cohesion in pemphigus. Pg depletion is involved in p38MAPK-mediated loss of cell cohesion induced by pemphigus autoantibodies.

Our data demonstrate that Pg staining in HaCaT cells is disrupted in response to pemphigus autoantibodies and that Pg protein levels are reduced. Similar results have been shown in other studies using primary mouse and human keratinocytes (de Bruin et al., 2007; Mao et al., 2009). Furthermore, some studies suggest fragmentation of DP localization and eventually DP depletion following PV-IgG treatment (Mao et al., 2011; Saito et al., 2012). It is not clear whether depletion of Pg or DP is causally involved in loss of cell cohesion in pemphigus or rather is an epiphenomenon following binding of autoantibodies to Dsg3. It has been shown that Dsg3 and Pg are internalized together after binding of pemphigus autoantibodies (Calkins et al., 2006). In line with this, our data suggest that reduced Pg levels following pemphigus autoantibody binding participate to cause p38MAPK activation and keratin filament collapse. Indeed, loss of cell adhesion by Pg silencing appears to be mediated at least in part by p38MAPK signaling as cell cohesion was efficiently increased by inhibition of this kinase. In contrast to our experiments, a study using embryonic mouse keratinocytes completely devoid of Pg suggested that Pg is necessary to mediate loss of cell adhesion by PV autoantibodies (Caldelari et al., 2001). However, the same group later demonstrated that loss of Pg together with Dsg3 was involved in autoantibody-induced cell dissociation (Williamson et al., 2006). As in our study Pg levels were only transiently diminished, it is possible that a precisely tuned amount of Pg is important to regulate cell cohesion, or that in Pg-null keratinocytes some functions of Pg have been replaced by other proteins such as b-catenin (Bierkamp et al., 1999). Our data further demonstrate a central role of correct desmosomal keratin filament anchorage for intercellular adhesion. In line with this, very recent data demonstrate a function for keratin filaments to limit protein kinase Ca-mediated DP phosphorylation and increase cell adhesion (Kro¨ger et al., 2013). Similarly, it has been shown that mice lacking both keratin 1 and 10 displayed altered desmosomal structure indicative of compromised adhesion (Wallace et al., 2012). Thus, important regulatory functions of keratin filaments for both desmosomal turnover and cell adhesion are emerging and need further clarifications. Extranuclear versus nuclear Pg and their roles in controlling cell adhesion

With regard to pemphigus, it has been shown that nuclear Pg is reduced after autoantibody binding allowing for TCF/LEFmediated increase of proliferation and loss of cell cohesion (Williamson et al., 2006). In our study, extranuclear but not nuclear Pg rescued loss of cell adhesion in response to pemphigus autoantibodies. Thus, our data do not indicate a relevant role of nuclear Pg for p38MAPK activation and keratin retraction. It is therefore possible that both nuclear

and extranuclear Pg signaling is relevant in pemphigus, which is underscored by the notion that NES-Pg expression was only partially effective in limiting AK23-induced loss of cell cohesion. In line with our results indicating extranuclear Pg (which primarily was located at the membrane) to strengthen cell adhesion via limiting p38MAPK-dependent keratin filament retraction, we found that cells transfected with Pg-siRNA demonstrated reduced p38MAPK levels in the desmosome. Thus, Pg may serve as a scaffolding protein necessary to tether p38MAPK to desmosomal complexes. In our previous study (Spindler et al., 2013), we demonstrated that p38MAPK is rapidly phosphorylated by AK23 after 30 minutes in a complex with Dsg3 and Pg. Thus, p38MAPK may initially become phosphorylated in the Dsg3/Pg complex upon autoantibody binding, and then get released from this complex to the cytosol to activate downstream mediators such as MK2 and heat shock protein 27 (Berkowitz et al., 2005; Mao et al., 2013). This is underscored by the finding that p38MAPK is activated specifically in the Triton-soluble pool after 1 hour of PV-IgG treatment. Alternatively, it is also possible that the signal to activate p38MAPK after binding of PV-IgG is relayed by cytoskeleton-unbound Dsg3 in the membrane. Taken together, our data indicate that Pg is required in a Dsg3-based signaling complex to regulate p38MAPK activity and to enhance keratinocyte cohesion by orchestrating keratin cytoskeleton organization. MATERIALS AND METHODS Cell culture and siRNA-mediated mRNA silencing The human immortalized keratinocyte cell line HaCaT was cultured in DMEM containing 1.8 mM Ca2 þ supplemented with 10% fetal calf serum (Biochrom, Berlin, Germany), 50 U ml–1 penicillin G, and 50 mg streptomycin. For culturing the squamous cell carcinoma cell line SCC9, a 1:1 mixture of DMEM and F-12 nutrient mixture (Ham) (Life Technologies, Carlsbad, CA) supplemented with 10% fetal bovine serum, 50 U ml–1 penicillin, and 50 mg ml–1 streptomycin was used. Cells were kept in a humidified atmosphere with 5% CO2 at 37 1C. For transient Pg and DP silencing, siRNA oligos were transfected using Turbofect (Fermentas, Thermofisher, Waltham, MA) or Lipofectamine RNAiMax as a transfection reagent according to the manufacturer’s recommendations. siRNA was purchased from Dharmacon (Dharmacon Thermo-Scientific, Waltham, MA) (Pg: J-01170810, DP: J-019800-07). For some experiments, siRNA against the 30 UTR (50 -CCUCCGUAGGGUCUUUCUU-30 ) was designed and applied using the same procedures. For immunoprecipitation experiments, the following Pg shRNA sequence was ligated into pLKO.1 vector (Addgene, Cambridge, MA, plasmid 8453): 50 -ATCCGTGT GTCCCAGCAATAATTCAAGAGATTATTGCTGGGACACACGGAT-30 . Lentiviral particles were produced and applied as described below.

Production of Pg shRNA lentiviral constructs To facilitate large-scale silencing necessary for immunoprecipitation experiments, we applied a lentiviral system based on the pLKO.1 system (Moffat et al., 2006). shRNA for Pg was designed with appropriate ends and cloned into the EcoRI and AgeI restriction sites of the pLKO.1 vector (Addgene plasmid 10878). pLKO.1-TRC www.jidonline.org 1661

V Spindler et al. Plakoglobin and p38MAPK-Dependent Cell Adhesion

control (Addgene plasmid 10879) served as control vector. Viral particles were assembled by transfecting HEK293T cells (kind gift of Dr Mariya Radeva, LMU Munich) with the pLKO.1 plasmid together with the pCMV delta R8.2 packaging plasmid (Addgene plasmid 12263) and the pCMV-VSV-G envelope plasmid (Addgene plasmid 8454) using the CaCl2 precipitation method. Briefly, HEK293T cells were seeded in 10-cm dishes and transfected with a complex of 15 mg PLKO1-shRNA-Pg-Vector, 3.75 mg VSV-G, and 11.75 mg pCMV delta R8.2-plasmid in 2.5 mol l–1 Ca2 þ solution together with the same amount of 2  N,N-bis(2-hydroxethyl)-2aminoethanesulfonic acid (BBS) (50 mol l–1 N,N-bis(2-hydroxyethyl)2-aminoethanesulfonic acid, 280 mol l–1 NaCl, 1.5 mol l–1 Na2HPO4). Medium was replaced after 24 hours and the supernatant collected 48 and 72 hours after transfection. Both fractions were pooled and stored at –80 1C. For infection of HaCaT cells, 6 ml of virus supernatant was added to 80% confluent cells seeded in T75 flasks, together with 10 mg Polybrene (Sigma Aldrich, St Louis, MO). Medium was exchanged after 24 hours and knockdown was evaluated after another 24 hours.

Test reagents and purification of PV-IgG fractions SB202190, a specific p38MAPK inhibitor (Calbiochem MerckMillipore, Darmstadt, Germany), was applied at 30 mmol l–1 for 6 hours. AK23, a pathogenic Dsg3 mAb derived from a pemphigus mouse model, was purchased from Biozol (Eching, Germany) and used at a concentration of 75 mg ml–1. PV-IgG was collected from a patient suffering from active PV with mucosal and epidermal lesions and applied at 500 mg ml–1. The serum was tested by ELISA (Euroimmun, Lu¨beck, Germany) displaying both Dsg3 (11,550 U ml–1) and Dsg1 (375 U ml–1) reactivity. Total IgG was purified as reported earlier (Waschke et al., 2005). IgG fractions were collected under informed consent and permitted by the ethics committee of the University of Lu¨beck, Germany.

Electrophoresis and western blotting Cells were grown in 24-well plates, incubated for 24 hours with AK23 or with PV-IgG and lysed according to standard procedures (Spindler et al., 2013). Lysates were then subjected to gel electrophoresis and blotted after measurement of protein amount using BCA-kit (Thermo Fisher Scientific, Waltham, MA). Primary antibodies were Pg mAb (Progen, Heidelberg, Germany), DP polyclonal antibody (pAb; Santa Cruz Biotechnology, Dallas, TX), Dsg3 pAb (Santa Cruz), anti-glyceraldehyde-3-phosphate dehydrogenase mAb (Sigma-Aldrich), b-actin mAb (Sigma-Aldrich), phospho-Thr180/182 p38MAPK pAb, and nonphospho-specific p38MAPK pAb (both Cell Signaling). Either a polyclonal, horseradish peroxidase-coupled goat anti-rabbit Ab (Cell Signaling, Danvers, MA), a goat anti-mouse mAb, or a goat anti-rabbit mAb (both Dianova GmbH, Hamburg, Germany) was used as a secondary Ab. All antibodies were diluted in 5% nonfat milk mixed with TBS-Tween. For membrane development, the ECL System (GE Healthcare, Buckinghamshire, UK) was used.

Immunofluorescence HaCaT cells grown to 80% confluency were transiently transfected with siRNA the day after seeding. At 48 hours after transfections, the cells were incubated with 30 mmol l–1 SB202190, a specific inhibitor of p38MAPK for 6 hours. Next, cells were fixed with 2% formalin 1662 Journal of Investigative Dermatology (2014), Volume 134

(freshly prepared from paraformaldehyde) for 10 minutes and subjected to immunostaining procedures as described in detail elsewhere (Spindler et al., 2007). Mouse monoclonal Pg Ab (Progen), rabbit polyclonal DP Ab (NW6, a kind gift of Dr Kathleen Green, Northwestern University, Chicago, IL), FITC-conjugated pancytokeratin mAb (Sigma-Aldrich), or pan-cytokeratin E605 mAb (Ebioscience, San Diego, CA) was used as the primary Ab. Cy3and Cy2-labeled goat anti-mouse or goat anti-rabbit Abs and a Cy5conjugated goat anti-mouse Ab served as secondary Abs. Images of immunostained samples were acquired using a Leica SP5 confocal microscope equipped with a 63  NA 1.4 PL APO objective (both Leica Microsystems, Wetzlar, Germany).

Dispase-based dissociation assays The assay was performed essentially as described earlier (Heupel et al., 2009). Briefly, cells were seeded in 24-well plates, cultured until 80% confluence, and then transiently transfected with Pg or DP siRNA. Again, 48 hours after transfection, fully confluent cells were incubated with AK23 alone or in combination with SB202190 or SB202190 for 24 hours. Following a brief rinse in Hank’s balanced salt solution, dispase II (dissolved in Hank’s balanced salt solution, Sigma-Aldrich) was applied for 30 minutes to release monolayer from the bottom of the well. Hank’s balanced salt solution was added to dispase solution and mechanical stress was applied by 10 pipetting cycles using a 1 ml electric pipette (Eppendorf, Hamburg, Germany). Finally, resulting fragments per well were counted under a binocular microscope (Leica Microsystems).

Immunoprecipitation T75-flasks of HaCaT cells were transfected either with lentiviral shRNA-Pg or with control lentivirus supernatant. After 48 hours, the cells were washed once with ice-cold phosphate-buffered saline and incubated with 1 ml RIPA buffer (0.05 mol l–1 Tris-HCl, 0.15 mol l–1 NaCl, 0.1% SDS, 1% NP-40, 0.1 mmol l–1 EDTA), mixed in a 1:1 ratio with protease inhibitors and phenylmethylsulfonyl fluoride (1:100) for 30 minutes on ice under gentle shaking. After centrifugation at 18,000 g for 5 minutes on 4 1C, the protein content was measured and an amount of 1,000 mg was used. After preclearing with protein A/G-beads (Santa Cruz), the capturing Ab (2 mg of DP pAb, Santa Cruz) was added for 1 hour at 4 1C on a rotator. Next, 40 ml protein A/ G-beads were applied to this mixture and incubated overnight at 4 1C. Beads were washed with RIPA wash-buffer (0.05 mol l–1 Tris-HCl, 0.15 mol l–1 NaCl, 0.1% SDS, 0.1% NP-40, phenylmethylsulfonyl fluoride 1:100). Following centrifugation and washing, beads were mixed with 3x Laemmli sample buffer, boiled for 10 minutes at 95 1C, and subjected to western blot analysis.

Triton extraction HaCaT cells were seeded in 24-well plates, transfected with either GFP-control or Pg-NES as described above and incubated for 24 hours with PV-IgG. After a washing step with ice-cold phosphatebuffered saline, Triton buffer (0.5% Triton X-100, 50 mmol l–1 MES, 25 mmol l–1 EGTA and 5 mmol l–1 MgCl2) was added for 5 minutes on ice. Cells were scraped and lysate was centrifugated at 18,000 g. The pellet represents the cytoskeleton-bound pool, whereas the supernatant is the non-cytoskeleton-attached pool. Both fractions were mixed with 3  Laemmli sample buffer and subjected to western blotting.

V Spindler et al. Plakoglobin and p38MAPK-Dependent Cell Adhesion

Generation of Pg constructs To generate constructs for the production of GFP-tagged fusion proteins of human Pg (AA 2-743) together with nuclear localization or NESs (NLS and NES, respectively), the full-length Pg complementary DNA-containing vector pCMV-GFP-Pg was used as a template. From this template, the complementary DNAs encoding NLS-Pg and NES-Pg, respectively, were amplified applying high fidelity PCR in the presence of the NLS-containing forward primer 50 -CTCGAGCTCCTAAGAAGAAGAGAAAGGTGGAGGTGATGAACCTGATGGAGC-30 or the NES-containing forward primer 50 -CTCGAGCTCTTCAGCTAC CACCGCTTGAGAGACTTACTCTTGATGAGGTGATGAACCTGATG GAGC-30 in conjunction with the reverse primer 50 -GAATTCCT AGGCCAGCATGTGGTCTGC-30 . Both forward primers were equipped with sequences representing the cloning site for XhoI (50 CTCGAG-30 ) and encoding either the simian virus 40 large-T antigen (SV40 T-ag)–derived NLS (PKKKRKV132: 50 -CCTAAGAAGAAGAGAA AGGTG-30 ) (Kalderon et al., 1984) or the NES (LQLPPLERLTLD84: 50 -CTTCAGCTACCACCGCTTGAGAGACTTACTCTTGAT-30 ) derived from the human immunodeficiency virus 1 Rev protein (Fischer et al., 1995). The reverse primer provided the cloning sequence for EcoRI. The yielded amplicons were subcloned into pGEM-T (Promega, Madison, WI) and the respective complementary DNAs subsequently transferred as XhoI- and EcoRI-flanking fragments into the same sites within expression vector pEGFPC1 (Clontech, Mountain View, CA). Both constructs were verified by sequencing. pCMV-GFP-N1 served as control.

Image processing and statistics Images were processed using Photoshop CS5 (Adobe Systems). Data were analyzed using Excel (Microsoft) and are presented as means±SE. Statistical analysis was performed using Prism (Graphpad Software, La Jolla, CA). All values were Gaussian-distributed and oneway analysis of variance was performed followed by Bonferroni correction. Statistical significance was assumed when Po0.05. CONFLICT OF INTEREST The authors state no conflict of interest.

ACKNOWLEDGMENTS We thank Dr Kathleen Green and Dr Mariya Radeva for providing cells and antibodies, and Dr Enno Schmidt for donating PV-IgG. Furthermore, we thank Angelika Antonius, Veronika Heimbach, Martina Hitzenbichler, Sabine Muehlsimer, and Andrea Wehmeyer for their skillfull technical assistance and Dr Katharina Huetz for helpful scientific discussion. SUPPLEMENTARY MATERIAL Supplementary material is linked to the online version of the paper at http:// www.nature.com/jid

REFERENCES Amagai M, Stanley JR (2012) Desmoglein as a target in skin disease and beyond. J Invest Dermatol 132:776–84 Berkowitz P, Hu P, Liu Z et al. (2005) Desmosome signaling. Inhibition of p38MAPK prevents pemphigus vulgaris IgG-induced cytoskeleton reorganization. J Biol Chem 280:23778–84 Berkowitz P, Hu P, Warren S et al. (2006) p38MAPK inhibition prevents disease in pemphigus vulgaris mice. Proc Natl Acad Sci USA 103: 12855–60 Bierkamp C, McLaughlin KJ, Schwarz H et al. (1996) Embryonic heart and skin defects in mice lacking plakoglobin. Dev Biol 180:780–5

Bierkamp C, Schwarz H, Huber O et al. (1999) Desmosomal localization of beta-catenin in the skin of plakoglobin null-mutant mice. Development 126:371–81 Brooke MA, Nitoiu D, Kelsell DP (2012) Cell–cell connectivity: desmosomes and disease. J Pathol 226:158–71 Caldelari R, de Bruin A, Baumann D et al. (2001) A central role for the armadillo protein plakoglobin in the autoimmune disease pemphigus vulgaris. J Cell Biol 153:823–34 Calkins CC, Setzer SV, Jennings JM et al. (2006) Desmoglein endocytosis and desmosome disassembly are coordinated responses to pemphigus autoantibodies. J Biol Chem 281:7623–34 Charpentier E, Lavker RM, Acquista E et al. (2000) Plakoglobin suppresses epithelial proliferation and hair growth in vivo. J Cell Biol 149: 503–20 de Bruin A, Caldelari R, Williamson L et al. (2007) Plakoglobin-dependent disruption of the desmosomal plaque in pemphigus vulgaris. Exp Dermatol 16:468–75 Delva E, Tucker DK, Kowalczyk AP (2009) The desmosome. Cold Spring Harb Perspect Biol 1:a002543 Dusek RL, Godsel LM, Chen F et al. (2006) Plakoglobin deficiency protects keratinocytes from apoptosis. J Invest Dermatol 127:792–801 Fischer U, Huber J, Boelens WC et al. (1995) The HIV-1 Rev activation domain is a nuclear export signal that accesses an export pathway used by specific cellular RNAs. Cell 82:475–83 Gallicano GI, Kouklis P, Bauer C et al. (1998) Desmoplakin is required early in development for assembly of desmosomes and cytoskeletal linkage. J Cell Biol 143:2009–22 Getsios S, Waschke J, Borradori L et al. (2010) From cell signaling to novel therapeutic concepts: international pemphigus meeting on advances in pemphigus research and therapy. J Invest Dermatol 130:1764–8 Heupel W-M, Engerer P, Schmidt E et al. (2009) Pemphigus vulgaris IgG cause loss of desmoglein-mediated adhesion and keratinocyte dissociation independent of epidermal growth factor receptor. Am J Pathol 174: 475–85 Huber O, Korn R, McLaughlin J et al. (1996) Nuclear localization of b-catenin by interaction with transcription factor LEF-1. Mech Dev 59:3–10 Kalderon D, Richardson WD, Markham AF et al. (1984) Sequence requirements for nuclear location of simian virus 40 large-T antigen. Nature 311: 33–8 Kro¨ger C, Loschke F, Schwarz N et al. (2013) Keratins control intercellular adhesion involving PKC-a–mediated desmoplakin phosphorylation. J Cell Biol 201:681–92 Li D, Zhang W, Liu Y et al. (2012) Lack of plakoglobin in epidermis leads to keratoderma. J Biol Chem 287:10435–43 Mao X, Choi EJ, Payne AS (2009) Disruption of desmosome assembly by monovalent human pemphigus vulgaris monoclonal antibodies. J Invest Dermatol 129:908–18 Mao X, Li H, Sano Y et al. (2013) MAPKAP kinase 2 (MK2)-dependent and independent models of blister formation in pemphigus vulgaris. J Invest Dermatol 134:68–76 Mao X, Sano Y, Park JM et al. (2011) p38 MAPK activation is downstream of the loss of intercellular adhesion in pemphigus vulgaris. J Biol Chem 286:1283–91 Moffat J, Grueneberg DA, Yang X et al. (2006) A lentiviral RNAi library for human and mouse genes applied to an arrayed viral high-content screen. Cell 124:1283–98 Ruiz P, Brinkmann V, Ledermann B et al. (1996) Targeted mutation of plakoglobin in mice reveals essential functions of desmosomes in the embryonic heart. J Cell Biol 135:215–25 Saito M, Stahley SN, Caughman CY et al. (2012) Signaling dependent and independent mechanisms in pemphigus vulgaris blister formation. PLoS One 7:e50696 Simcha I, Shtutman M, Salomon D et al. (1998) Differential nuclear translocation and transactivation potential of b-catenin and plakoglobin. J Cell Biol 141:1433–48

www.jidonline.org 1663

V Spindler et al. Plakoglobin and p38MAPK-Dependent Cell Adhesion

Smith EA, Fuchs E (1998) Defining the interactions between intermediate filaments and desmosomes. J Cell Biol 141:1229–41 Spindler V, Drenckhahn D, Zillikens D et al. (2007) Pemphigus IgG causes skin splitting in the presence of both desmoglein 1 and desmoglein 3. Am J Pathol 171:906–16 Spindler V, Rotzer V, Dehner C et al. (2013) Peptide-mediated desmoglein 3 crosslinking prevents pemphigus vulgaris autoantibody-induced skin blistering. J Clin Invest 123:800–11 Stanley JR, Amagai M (2006) Pemphigus, bullous impetigo, and the Staphylococcal scalded-skin syndrome. N Engl J Med 355:1800–10 Thomason HA, Scothern A, McHarg S et al. (2010) Desmosomes: adhesive strength and signalling in health and disease. Biochem J 429:419–33 Vasioukhin V, Bowers E, Bauer C et al. (2001) Desmoplakin is essential in epidermal sheet formation. Nat Cell Biol 3:1076–85 Wallace L, Roberts-Thompson L, Reichelt J (2012) Deletion of K1/K10 does not impair epidermal stratification but affects desmosomal structure and nuclear integrity. J Cell Sci 125:1750–8 Waschke J (2008) The desmosome and pemphigus. Histochem Cell Biol 130:21–54

1664 Journal of Investigative Dermatology (2014), Volume 134

Waschke J, Bruggeman P, Baumgartner W et al. (2005) Pemphigus foliaceus IgG causes dissociation of desmoglein 1-containing junctions without blocking desmoglein 1 transinteraction. J Clin Invest 115:3157–65 Williamson L, Raess NA, Caldelari R et al. (2006) Pemphigus vulgaris identifies plakoglobin as key suppressor of c-Myc in the skin. EMBO J 25:3298–309 Yamamoto Y, Aoyama Y, Shu E et al. (2007) Anti-desmoglein 3 (Dsg3) monoclonal antibodies deplete desmosomes of Dsg3 and differ in their Dsg3-depleting activities related to pathogenicity. J Biol Chem 282: 17866–76 Yang L, Chen Y, Cui T et al. (2012) Desmoplakin acts as a tumor suppressor by inhibition of the Wnt/b-catenin signaling pathway in human lung cancer. Carcinogenesis 33:1863–70 Zhurinsky J, Shtutman M, Ben-Ze’ev A (2000a) Plakoglobin and beta-catenin: protein interactions, regulation and biological roles. J Cell Sci 113: 3127–39 Zhurinsky J, Shtutman M, Ben-Ze’ev A (2000b) Plakoglobin and beta-catenin: protein interactions, regulation and biological roles. J Cell Sci 113 (Pt 18):3127–39