BPAG1-E Restricts Keratinocyte Migration Through Control Of

ORIGINAL ARTICLE

BPAG1-e Restricts Keratinocyte Migration through Control of Adhesion Stability Magdalene Michael1,2,3,RumenaBegum1,2,KennethFong2, Celine Pourreyrone4,AndrewP.South4, John A. McGrath2 and Maddy Parsons1

Bullous pemphigoid antigen 1 (BPAG1-e, also known as BP230) is a member of the plakin family of hemidesmosome cytoskeletal linker proteins that is encoded by an isoform of the dystonin (DST) gene. Recently, we reported two unrelated families with homozygous nonsense mutations in this DST isoform that led to ultrastructural loss of hemidesmosomal inner plaques and clinical features of trauma-induced skin fragility. We now demonstrate that keratinocytes isolated from these individuals have signiﬁcant defects in adhesion, as well as increased cell spreading and migration. These mutant keratinocytes also display reduced levels of b4 integrins at the cell surface but increased total protein levels of keratin-14 and b1 integrins. These alterations in cell behavior and protein expression were not seen in control keratinocytes in which BPAG1-e expression had been silenced by stable expression of short hairpin RNA to target DST. The failure of knockdown approaches to recapitulate the changes in morphology, adhesion, and migration seen in patient cells therefore suggests such approaches are not appropriate to study loss of this protein in vivo. The contrasting ﬁndings in keratinocytes harboring naturally occurring mutations, however, demonstrate a previously unappreciated key role for BPAG1-e in regulating keratinocyte adhesion and migration and suggest a requirement for this protein in controlling functional switching between integrin types in epithelial cells. Journal of Investigative Dermatology (2014) 134, 773–782; doi:10.1038/jid.2013.382; published online 17 October 2013

INTRODUCTION known as BP230), a key component of hemidesmosomes Epidermolysis bullosa (EB) represents a diverse collection of and a member of the plakin family with cytoskeletal linker autosomal dominant and autosomal recessive conditions with properties (Borradori and Sonnenberg, 1999; Leung et al., varying degrees of trauma-induced skin and mucous membrane 2001a; Litjens et al., 2006; Sonnenberg and Liem, 2007). fragility and, in some cases, extracutaneous abnormalities (Fine BPAG1-e is encoded by the dystonin (DST) gene, the et al., 2008; Fine and Mellerio, 2009a, b). The disease spectrum alternative splicing of which may give rise to multiple tissue of EB (or EB-like disorders) encompasses inherited disorders of isoforms with variable expression in the skin, neurons, hemidesmosome attachment complexes, keratin intermediate muscles, and the central nervous system (Leung et al., filaments, focal adhesions, desmosome cell junctions, keratino- 2001b; Jefferson et al., 2006; Young and Kothary, 2007, cyte vesicle transport, and enzymes involved in cornification, 2008). The main isoform expressed in the central nervous and currently involves pathogenic mutations in at least 18 system is BPAG1-a, although a further neural variant, BPAG1- different genes (Bruckner-Tuderman et al., 2013). n, may also exist; whether BPAG1-n is expressed in vivo, One protein that was recently implicated in the pathogen- however, is uncertain (Leung et al., 2001b). In muscle, the esis of EB, is bullous pemphigoid antigen 1 (BPAG1-e, also main isoform is BPAG1-b. In the skin, there is a predominance of BPAG1-e, and some BPAG1-a and perhaps BPAG1-n may 1Randall Division of Cell and Molecular Biophysics, King’s College London, also be present (Sonnenberg and Liem, 2007). Guy’s Campus, London, UK; 2St Johns Institute of Dermatology, King’s College We recently reported two unrelated cases with a form of London, Guys Campus, London, UK; 3Division of Molecular Cell Biology, autosomal recessive EB simplex because of homozygous Institute for Molecular Bioscience, The University of Queensland, St Lucia, nonsense mutations in BPAG1-e (Groves et al., 2010; Liu Brisbane, Queensland, Australia and 4Jacqui Wood Cancer Centre, University of Dundee, Ninewells Hospital and Medical School, Dundee, UK et al., 2012). The mutations detected were c.3478C4Tand Correspondence: Maddy Parsons, Randall Division of Cell and Molecular c.3853A4T (GenBank NM_001723.4), which result in the Biophysics, King’s College London, Guy’s Campus, London SE11UL, UK. amino-acid changes p.Gln1124X and p.Arg1249X, E-mail: [email protected] respectively. Clinically, affected individuals had trauma- Abbreviations: BPAG1, bullous pemphigoid antigen 1; DST, dystonin; EB, induced blistering (generalized but mainly on the feet), and epidermolysis bullosa; KD, knockdown; K14, keratin 14; PBS, phosphate- skin biopsy revealed a lack of hemidesmosomal inner plaques buffered saline; shRNA, short hairpin RNA; WT, wild type on transmission electron microscopy and a complete absence Received 31 May 2013; revised 12 August 2013; accepted 27 August 2013; accepted article preview online 11 September 2013; published online of immunostaining for the BPAG1-e protein using two 17 October 2013 isoform-specific monoclonal antibodies.

& 2014 The Society for Investigative Dermatology www.jidonline.org 773 MMichaelet al. BPAG1-e Regulation of Keratinocyte Migration

In contrast to the clinicopathologic consequences of homo- antibody used to detect BPAG1-e in these experiments zygous nonsense mutations in BPAG1-e, a recent study recognized an epitope within the C-terminal region of the demonstrated that depleting BPAG1-e levels in keratinocytes protein, beyond the mutation site identified in both DST resulted in rather subtle defects in cell polarity and no patients. Owing to the lack of reliable antibodies specifically apparent changes in migration speed or adhesion (Hamill recognizing BPAG1-e N-terminal epitopes, we were unable to et al., 2009). Thus, the role of BPAG1-e in controlling various determine whether a truncated version of the BPAG1-e protein aspects of keratinocyte behavior is unclear. Here, we show remained in DST1 and DST2 patient cells. Further analysis of that cells derived from two individuals with loss-of-function these cells was conducted using in vitro organotypic assays in mutations in DST show significant defects in cell morphology, which keratinocytes are grown on top of collagen/Matrigel adhesion, and migration. Contrary to previous knockdown three-dimensional scaffolds containing human fibroblasts and (KD) studies, however, these cells show increased migration cultured at an air–liquid interface for 2 weeks to mimic a that results from a loss of surface b4 integrin and increased tissue-like environment (Nystrom et al., 2005). Analysis of active b1 integrin. Our findings confirm a key role for organotypic cultures post-fixation revealed a flattened BPAG1-e in controlling keratinocyte adhesion and motility morphology of basal keratinocytes in contact with the under- and further suggest this protein has an important role lying basement membrane in both DST patients compared in coordinating maturation of hemidesmosomes and with controls (Supplementary Figure S1i online). However, suppression of focal adhesion dynamics. there were no gross changes in the localization of differentiation markers observed (Supplementary Figure S1j online), in RESULTS agreement with previous findings in the patient skin. These DST mutant keratinocytes display altered adhesion and data demonstrate that loss of BPAG1-e expression in kerati- migration nocytes leads to alterations in basal keratinocyte morphology As the DST gene is known to give rise to multiple isoforms, that may be a consequence of altered stability or adhesion to gene expression analysis of known tissue-specific isoforms basement membrane. were analyzed by quantitative PCR in wild-type (WT) kerati- In order to analyze these potential attachment defects in nocytes. In comparison with total DST gene expression (total more detail, we quantified adhesion of WT or DST mutant BPAG1), nearly undetectable levels of the muscle-specific and keratinocytes to the basement membrane matrix Matrigel. neuronal isoforms, BPAG1-b and BPAG1-a, respectively were Both DST mutant keratinocyte cell lines showed a significant detected; however, high expression of BPAG1-e was observed delay in adhesion at 30 and 60 minutes post-plating compared confirming it as the predominant BPAG1 isoform expressed in with control cells (Figure 1a and b). Similarly, analysis of cell keratinocytes (Supplementary Figure S1a online). Comparison detachment demonstrated a significant increase in DST mutant of WT keratinocytes with those isolated from the first DST cell detachment rates compared with controls, suggesting that patient (DST1) revealed a reduction in total BPAG1 isoform adhesions formed in these cells are less stably connected to expression as well as a significant loss of BPAG1-e isoform the substratum (Figure 1c). Adhesion strengthening is known expression in DST1 keratinocytes (Supplementary Figure S1b to be required for normal keratinocyte migration (Margadant and c online) confirming the impact of the homozygous et al., 2010). We therefore analyzed random migration nonsense mutation on BPAG1-e isoform expression. To con- behavior of control and DST mutant cells plated at low firm the loss of BPAG1-e in both individuals with loss-of- density by time-lapse microscopy. Tracking single cells over function mutations in DST and assess potential changes in time revealed an unexpected increase in migration speed and adhesion protein localization, immunostaining for BPAG1-e, directional migration persistence in both DST mutant lines plectin, and collagen XVII, along with differentiation markers, (Figure 1d, e and f). This is in contrast to a previous study keratin 14 (K14) and involucrin, was performed on patient showing no change in migration in BPAG1-e KD cells in vitro skin sections (DST1 and DST2) compared with samples taken (Hamill et al., 2009). In order to compare silencing of BPAG1- from healthy volunteers. A total loss of BPAG1-e immuno- e with DST mutant cells, we generated keratinocytes stably reactivity was observed in both patients as expected but no expressing BPAG1-e short hairpin RNA (shRNA) to deplete overall changes in hemidesmosome localization or differentia- BPAG1-e levels by B95% (Figure 2a), and analyzed adhesion tion markers were detected (Supplementary Figure S1d, e and and migration under identical conditions to those above. In f online, see also Groves et al., 2010; Liu et al., 2012). agreement with Hamill et al. (2009), we also found no Western blot analysis confirmed a loss of BPAG1-e expression difference in adhesion (Figure 2b and c) or migration speed in keratinocytes isolated from both DST patients (Supple- in the BPAG1-e-depleted cells but a small decrease in mentary Figure S1g and h online). Notably, however, the directional persistence of these cells compared with control

Figure 1. Loss of bullous pemphigoid antigen 1 (BPAG1-e) results in decreased adhesion and increased migration speeds. (a, b) Wild-type (WT) and dystonin (DST) mutant cells were plated and fixed after 30, 60, and 120 minutes. (a) Immunofluorescence images of adherent cells stained with phalloidin A568. Scale bar ¼ 100 mm. (b) Quantification of number of adherent cells at indicated time points. (c) Quantification of cell detachment following trypsin treatment. (d–f) Analysis of random migration of WT and DST mutant keratinocytes. (d) Migration tracks of keratinocytes were analyzed for (e)speedand(f) persistence. For all quantification, data represent three independent experiments, error bars ¼ SEM; *Po0.05, **Po0.01, and ***Po0.001.

774 Journal of Investigative Dermatology (2014), Volume 134 MMichaelet al. BPAG1-e Regulation of Keratinocyte Migration

30 Minutes 60 Minutes 120 Minutes WT DST1 DST2

600 120 WT 500 100 * DST1 400 80 DST2 300 60 * 200 ** 40 % Of detached cells

Cells per field of view ** 100 *** 20 0 0 30 60 120 WT DST1 DST2 Minutes

40 40 40

20 20 20

0 0 0

–20 –20 –20

WT DST1 DST2 –40 –40 –40 0 0 0 10 20 30 40 10 20 30 40 10 20 30 40 –20 –10 –20 –10 –20 –10

5 0.92 ** 0.88 4 ** * ** 0.84 3 0.8 2 0.76 Persistence

1 0.72 Speed ( μ m per minute) 0 0.68 WT DST1 DST2 WT DST1 DST2

www.jidonline.org 775 MMichaelet al. BPAG1-e Regulation of Keratinocyte Migration

30 Minutes 60 Minutes 120 Minutes

Control BPAG1-e KD

BPAG1-e

Hsc70 Control KD

350 Control KD 300 BPAG1-e KD 250 200 150 100 50 Cells per field of view 0

30 60 120 KD BPAG1-e Minutes

50 40 1.2 20 0.7 1 0.6 * 0 0.5 0 0.8 0.4 20 0.6 0.3 40 0.4 0.2 Persistence 50 0.2 0.1 60

Speed ( μ m per minute) 0 0 80 Control KD BPAG1-e KD Control KD BPAG1-e KD

60 40 2006 20 40 60 040420 200 0 Control KD BPAG1-e KD

160 Control 140 BPAG1-e KD 120 100 Control BPAG1-e KD Control BPAG1-e KD 80 K14 KRT14 60

% Integrin levels 40 Hsc70 GAPDH 20 0 β4 β1 β1 (A)

Figure 2. Knockdown (KD) of bullous pemphigoid antigen 1 (BPAG1-e) does not recapitulate dystonin (DST) mutant adhesion and migration defects. (a) Western blot analysis of BPAG1-e protein levels in control and BPAG1-e KD keratinocytes. (b, c) Adhesion of BPAG1-e KD cells following 30, 60, and 120 minutes of plating. (b) Immunofluorescence images of cells stained with phalloidin A568. Scale bar ¼ 100 mm. (c) Quantification of number of adherent cells. (d–f) Random migration analysis of control and BPAG1-e KD cells. (d) Migration tracks of keratinocytes were analyzed for (e)speedand(f) persistence. (g) Western blot analysis of control and BPAG1-e KD keratinocytes with keratin 14 antibodies and (h) assessment of KRT14 mRNA levels by reverse-transcriptase–PCR (RT- PCR), with glyceraldehyde-3-phosphate dehydrogenase (GAPDH) as a control. (i) FACS analysis of surface levels of b4, b1 total, and b1 active (A) integrins in control and BPAG1-e KD cells. For all quantification, data represent three independent experiments; error bars ¼ SEM; *Po0.05.

shRNA-expressing cells (Figure 2d–f). These ﬁndings indicate result in alterations in cellular signaling that are not recapitu- that BPAG1-e controls adhesion strengthening and slows lated by simple depletion of BPAG1-e expression. Alterna- keratinocyte migration, potentially through control of adhe- tively, as discussed in relation to Figure 1, it is possible that a sion complex stability. Importantly, these effects are only truncated version of BPAG1-e remains in the DST1 and DST2 revealed in the DST mutant condition suggesting that the cells that we cannot detect using currently available anti- homozygous nonsense mutation in the DST keratinocytes may bodies and that this also contributes to the different

776 Journal of Investigative Dermatology (2014), Volume 134 MMichaelet al. BPAG1-e Regulation of Keratinocyte Migration

phenotypes observed in the DST cells compared with BPAG1- growth conditions (Figure 3a and b). Western blotting showed e silencing. Similarly, a low level of persisting BPAG1-e no change in levels of the cytoskeletal proteins actin or tubulin protein in shRNA silenced cells could suffice to maintain (data not shown) but revealed a significant increase in levels of some functionality and/or impart a different phenotype to that the intermediate filament protein K14 in cells from both DST seen in DST mutant cells. patients in comparison with two control keratinocyte cell lines (Figure 3c and Supplementary Figure S2a online). A similar Increased levels of K14 in DST mutant keratinocytes increase was also seen at the KRT14 mRNA level by reverse- Given the defects in cell adhesion and migration in BPAG1-e transcriptase–PCR (Figure 3d). Incidentally, protein levels of mutant keratinocytes, we speculated that loss of BPAG1-e may keratin 5 (K5), the associated partner of K14 was unaffected lead to disruption in other key cytoskeletal proteins and thus (Figure 3e). Conversely, there was no change in K14 protein or reduced cell adhesion. Indeed, morphometric analysis of cell mRNA levels in BPAG1-e KD cells, again demonstrating that monolayers revealed that both DST mutant cell lines showed these cells do not recapitulate the abnormal cellular pheno- significantly larger cell area in monolayers under normal types observed in patient cells (Figure 2g and h). Confocal

Keratin14 Phalloidin Merge

9,000 * 8,000 7,000 ***

WT 6,000 5,000 4,000 3,000 Cell area (AU) 2,000 1,000 0 WT DST1 DST2

WT DST1 DST2 DST1

WT DST1 K14 KRT14

GAPDH Hsc70 DST2

DST1 600 WT

500 DST1

DST/K14KD2 WT DST1 WT – K14KD1 K14KD2 400 300 K5 K14 200 * ** Hsc70 *** Cells per field of view 100 ** ** Hsc70 0 30 60 120 Minutes

Figure 3. Keratin 14 (K14) upregulation does not contribute to adhesion defect in dystonin (DST)mutantcells.(a) Immunoﬂuoresence of wild-type (WT) and DST keratinocytes with K14 antibodies and phalloidin A568. Scale bar ¼ 10 mm. (b) Cell area measurements of WT and DST keratinocytes monolayers. (c, e, f) Western blot analysis of (c)K14and(e) K5 protein expression in WT and DST mutant cell lines and (f) K14 expression in WT, DST1, and DST1 cells stably expressing two different K14 short hairpin RNA (shRNA) (K14KD1 and K14KD2). Hsc70 was used as a loading control. (d) Assessment of KRT14 mRNA expression by reverse-transcriptase–PCR (RT-PCR) in WT and DST1 cells, using glyceraldehyde-3-phosphate dehydrogenase (GAPDH) as a control. (g)Adhesion analysis of WT, DST1, and DST1 expression K14KD2 shRNA after 30, 60, and 120 minutes of plating. For all quantiﬁcation, data represent three independent experiments, error bars ¼ SEM; *Po0.05, **Po0.01, and ***Po0.001. AU, arbitrary units.

www.jidonline.org 777 MMichaelet al. BPAG1-e Regulation of Keratinocyte Migration

analysis demonstrated highly disorganized bundles of K14 observed phenotypes are due to the loss of BPAG1-e function filaments within the cytoplasm of BPAG1-e mutant cells in both cases as opposed to indirect effects via the Notch compared with more elaborated networks in controls pathway. (Figure 3a). Given that alterations in K14 function are also implicated in EB simplex (Coulombe et al., 1991), we sought BPAG1-e loss-of-function results in increased activation of b1 to determine whether the K14 upregulation in BPAG1-e- integrins deficient cells contributed to the adhesion and migration The defects in cell adhesion and spreading in the DST mutant defects observed. We stably expressed control shRNA or cells prompted us to analyze whether loss of BPAG1-e sequences targeting KRT14 in DST mutant cells to deplete function resulted in defects in integrin levels or localization. K14 levels to those seen in control keratinocytes (Figure 3f). Analysis of the hemidesmosomal integrins a6andb4 in skin in Analysis of adhesion and migration in these K14 ‘rescued’ control or DST samples showed no changes in localization of cells demonstrated no change compared with control DST these proteins at the dermal–epidermal junction in skin cells suggesting that the defective adhesion and motility (Figure 4a). Similarly, total b4 integrin protein levels as observed is not directly because of upregulated K14 in these assayed by western blotting were identical between control cells (Figure 3g and data not shown). and DST mutant cells (Figure 4b). However, analysis of cell It was previously reported that one of the patients (DST1) surface receptor levels of integrins by FACS revealed a also harbored a mutation in the NOTCH3 gene (Groves et al., significant decrease in surface levels of b4 integrins in the 2010). We confirmed this mutation was not present in the DST mutant cell lines (Figure 4c), which was also evident second DST mutant patient (Liu et al., 2012). Reverse- using surface antibody staining of non-permeabilized cells and transcriptase–PCR analysis of the NOTCH3 downstream confocal analysis (Supplementary Figure S2d online). More- target HES1 mRNA levels in cells treated with the Notch over, analysis of the focal adhesion-associated b1 integrin in inhibitor N-[N-(3,5-difluorophenacetyl)-L-alanyl]-S-phenylgly- parallel revealed increases in total protein levels in DST1 and cine t-butyl ester (DAPT) revealed a decreased response in DST2 cells by western blotting (Figure 4b and Supplementary DST1 cells suggesting a constitutive activation of this pathway Figure S2c online) as well as higher cell surface levels of total (Supplementary Figure 2b online). However, treatment of the and active versions of this integrin as assayed by FACS WT or DST1 cells with DAPT did not impact on the altered cell (Figure 4d) and immunofluorescence (Supplementary Figure adhesion or spread area (data not shown) suggesting that these S2e online). These changes in integrin levels (either total or

WT DST1 DST2

β4 Integrin β 4 Integrin

β1 Integrin

Hsc70 α 6 Integrin

200 500 WT WT 400 * 150 DST1 ** DST1 DST2 300 ** DST2 100 ** ** ** ** 200 50 100 % Integrin levels % Integrin levels 0 0 β4 α6 β1 β1 (A)

Figure 4. Dystonin (DST) mutation results in an integrin switch in keratinocytes. (a) Immunostaining of wild-type (WT) and DST mutant skin sections for b4and a6 integrin. Scale bar ¼ 30 mm. (b) Western blot analysis of b4 integrin and b1 integrin in WT and DST mutant cell lysates. Hsc70 was used as a loading control. (c, d) FACS analysis of (c) b4anda6 integrin surface levels and (d) b1 integrin total and active (A) surface levels in WT and DST mutant cells. For all quantiﬁcation, data represent three independent experiments, error bars ¼ SEM; *Po0.05, **Po0.01.

778 Journal of Investigative Dermatology (2014), Volume 134 MMichaelet al. BPAG1-e Regulation of Keratinocyte Migration

surface exposed receptor) were not seen in BPAG1-e KD cells migratory behavior of keratinocytes on basement membrane (Figure 2i) further supporting our previous suggestions that proteins in the absence of stimuli, which requires the shRNA depletion of this protein is not sufficient to mimic the coordination of protrusive signals and basement membrane DST mutant phenotypes. Reduced cell surface b4 integrin adhesion (Ridley et al., 2003). Given the change in integrin levels and corresponding increases in b1 integrin surface balance identified here, analyzing the cell–cell versus cell– levels seen in keratinocytes harboring the DST mutation is matrix adhesion strengthening in BPAG1-e null cells will be also consistent with the concept of a switch in adhesion types an important area to pursue in future studies. upon loss of this hemidesmosomal component at the cell membrane. Our data suggest that de-stabilization of hemi- DST mutation alters K14 expression in basal keratinocytes desmosomes may result in a phenotypic switch to focal Integrins a6b4anda3b1 share binding to common extra- adhesion use, and that this in turn reduces initial stable cellular ligands although their cytoplasmic regions interact adhesions, promotes cell spreading, and cell migration. with different cytoskeletal structures. a6b4 Connects the basal keratinocyte adhesions to the keratin cytoskele- DISCUSSION ton whilst a3b1 is linked indirectly to actin filaments. Mutations in BPAG1-e inhibits keratinocyte adhesion and Adhesion via a6b4 is stabilized through linkages to K5 and enhances migration through regulation of integrin levels K14 filaments, mediated though an association with plectin Our analysis of two different human BPAG1-e-deficient cells and BPAG1-e (Tsuruta et al., 2011). Here, we report a role has revealed a previously unidentified role for this protein in for BPAG1-e in the regulation of keratin filament assembly in controlling keratinocyte cell adhesion and migration, poten- epithelial cells. Ultrastructural analysis of DST mutant skin tially through modulation of integrin stability at adhesion sites. revealed detachment of keratin filaments from adhesive Loss of BPAG1-e resulted in no change in total b4 integrin structures leading to poorly formed hemidesmosomal inner protein levels, but a significant reduction at the cell surface, plaques (Groves et al., 2010; Liu et al., 2012). Interestingly, suggesting receptor internalization or defects in recycling as a DST mutant cells exhibited an aberrant upregulation of result of BPAG1-e loss. Notably, a3b1anda6b1 integrins, KRT14 gene and K14 protein expression suggesting that components of focal adhesions, are expressed in basal loss of linkage between the K14 filaments and the associated keratinocytes and can bind basement membrane extracellular membrane adhesion components results in altered signaling matrix proteins to mediate basal cell adhesion (Margadant cues that stimulate keratin expression, possibly as a et al., 2010). However, mouse knockout studies have reported compensatory mechanism. Therefore, it could be inferred that the severe adhesion defect exhibited by loss of b4 integrin that establishment of a continuous linkage between the could not be compensated by upregulation of a6b1 (DiPersio adhesion components and the keratin filament network et al., 2000; deHart et al., 2003), highlighting the importance generates a negative feedback signal to regulate KRT14 of the a6b4 hemidesmosomal complex in epidermal adhesion gene expression. Interestingly, a recent study revealed that stabilization despite the presence of alternative extracellular loss of keratin resulted in increased adhesion rates (Seltmann matrix–binding integrins. Conversely, the significant increase et al., 2013); however, our data show that the defective in activation of surface b1 integrins in the DST mutant cells adhesion rates in DST mutant cells are not rescued in cells would seem highly likely to have a key role in the weakened where K14 was reduced to endogenous levels, indicating basal adhesions and favor a more migratory phenotype. that the increased K14 protein levels do not directly b1 Integrins have a well-established role in keratinocyte contribute to this phenotype but instead structural spreading and migration and the consequent upregulation abnormalities may exist in the keratin cytoskeleton in DST of b1 could promote the increased spreading and migration mutant cells. phenotype observed in the DST mutant cells (Grose et al., 2002). Therefore, a crucial balance of hemidesmosome Homozygous nonsense mutations in DST do not recapitulate the versus focal adhesion integrins appears to impact on keratino- BPAG1-e KD phenotype cyte properties, determining adhesion stabilization in an intact A surprising finding in this study was the observation that the epithelium, or alternatively, permitting a migratory response DST mutant cells do not display the same phenotype as during wound healing and repair. BPAG1-e KD keratinocytes. Our data further confirm results The enhanced migration in DST mutant keratinocytes may reported by Hamill et al. (2009) that keratinocytes stably appear to challenge previous suggestions that BPAG1-e is a transduced with lentiviral-mediated shRNA-targeting positive regulator of migration in mice (Guo et al., 1995), BPAG1-e exhibit no alterations in integrin levels, adhesion although these disparities may be explained by the different impairment, or changes in migration speeds, albeit with a approaches used to analyze cell migration. Guo et al. (1995) small decrease in persistence (Hamill et al., 2009). As in the assessed the collective, directional migration of the epidermal case of Hamill et al. (2009), a robust depletion of 490% layer in Bpag1 / mice in response to wounding of the skin, BPAG1-e protein was achieved by lentiviral-mediated a complex process involving the establishment of polarity in transduction in this study. However, the contrasting the direction of migration, maintenance of intercellular phenotypes displayed by BPAG1-e KD keratinocytes adhesion, and regulation of basement membrane interactions compared with DST mutants, suggests that the remnant (Guo et al., 1995; Friedl and Gilmour, 2009). This study in BPAG1-e expressed because of incomplete depletion in the contrast utilized an in vitro approach to analyze the random shRNA KD system could be sufficient to preserve the basic

www.jidonline.org 779 MMichaelet al. BPAG1-e Regulation of Keratinocyte Migration

keratinocyte properties and function. Although the previous peroxidase-conjugated secondary antibodies (Dako, Ely, UK), Alexa report on the first DST patient skin demonstrated that the 488, Alexa 568-conjugated anti-rabbit and anti-mouse (Invitrogen, homozygous nonsense mutation does not lead to nonsense- Paisley, UK), Cy5-conjugated anti-rabbit and mouse (Zymed, mediated decay of the mRNA, it is still uncertain if the Invitrogen). Other reagents: 4,6-diamidino-2-phenylindole and Alexa putative truncated protein is indeed expressed in the DST 568-conjugated phalloidin (Invitrogen), DAPT (Calbiochem, Watford, mutant tissue because of lack of suitable antibodies to detect UK), and puromycin (Sigma, Dorset, UK). the N-terminal region of BPAG1-e (Groves et al., 2010; Liu et al., 2012). It remains plausible that although gene products Cell culture and shRNA KDs upstream of the nonsense mutation were detected, the All keratinocyte cell lines were cultured in keratinocyte serum-free resultant truncated protein could be unstable and subject medium media (Invitrogen) supplemented with penicillin and strep- to degradation. Furthermore, our observations that gene tomycin. DST1 and DST2 cell lines were obtained by isolation of expression of all BPAG1 isoforms, determined by ampli- primary keratinocytes from EB simplex patient biopsies and human fication of a region upstream of the BPAG1-e nonsense papillomavirus immortalization (Groves et al., 2010; Liu et al.,2012). mutations, is reduced in the DST mutant keratinocyte The WT1 human keratinocyte cell line, NEB-1, was a gift from Dr implies that there may also be reduced gene expression of Graham Neill (Morley et al., 1995) and an additional control the N-terminal regions of BPAG1-e, as it is the predominant keratinocyte cell line (K17; WT2) was generated by isolation of isoform in keratinocytes. Alternatively, the putative truncated keratinocytes from a 51-year-old female breast and human BPAG1-e, which lacks the keratin binding C-terminal region papillomavirus immortalization in an identical manner to both of but still harbors binding sites for b4 integrin and collagen the DST mutant patient cell lines. Following written and informed XVII (Koster et al., 2003), could exert a dominant-negative consent, blood and skin samples were obtained from the patient in effect on the hemidesmosomal complex resulting in the adherence to the Declaration of Helsinki Principles. Ethical approval observed alterations in integrin signaling and adhesion for the study was granted by Guy’s and St Thomas’ NHS Foundation dynamics. However, lack of antibodies to specifically Trust and King’s College London. HEK293T cells for lentivirus report on levels of a potential truncated BPAG1-e protein production were cultured in RPMI media supplemented with 10% make it currently impossible to determine which of these fetal calf serum, penicillin, streptomycin, and L-glutamine. BPAG1-e possibilities exists in either DST1 or DST2 cell lines. It should KD and control cell lines were generated by lentiviral-mediated be noted that relative expression of both the muscle- and transduction of WT keratinocytes with BPAG1-e shRNA and control brain-specific isoforms were increased in DST mutant constructs. Infected cells were selected using 2 mgml–1 puromycin. keratinocytes compared with WT (data not shown), Similarly, the KRT14KD cell line was made by lentiviral infection of however, as these isoforms are expressed at very low levels the DST1 cell line with the KRT14 shRNA construct and puromycin in WT cells it hard to envisage their contribution to the selection. phenotype observed in the DST cells through a compensatory mechanism. Future investigations to specifically test the Flow cytometry effect of expressing this truncated form of BPAG1-e in Cultured WT, DST1, and DST2 were scraped off with phosphate- BPAG1-e KD keratinocytes may help to provide further buffered saline (PBS) and fixed with 4% paraformaldehyde for insight into the differences in the phenotypes observed. 20 minutes. Cells were then blocked for 60 minutes with 2% BSA in In summary, these findings point to a previously un- PBS and incubated with primary antibody for 90 minutes. Cells were appreciated critical role for BPAG1-e in the regulation of washed three times in PBS, incubated with secondary antibody for keratinocyte hemidesmosomal adhesion through the control of 45 minutes, washed again and resuspended in PBS for FACS analysis. integrin signaling and keratin organization. This study also As a negative control, the primary antibody was omitted in the highlight the importance of analyzing potential roles of experiment. Mean fluorescent values were subtracted against the BPAG1-e and related BPAG1 isoforms in other skin blistering negative control and normalized to the WT values. or fragility disorders in future. Immunofluorescence and microscopy MATERIALS AND METHODS Keratinocytes were grown in keratinocyte grown medium media Plasmids and reagents (MEM Alpha Eagle’s medium supplemented with insulin, epidermal pLKO.1 lentiviral shRNA libraries for BPAG1-e and KRT14 were growth factor, adenine, hydrocortisone, penicillin, and streptomycin) purchased from Open Biosystems (Thermo Scientific, Northumberland, on Matrigel (BD Biosciences, Oxford, UK)-coated glass coverslips for UK). The shRNA sequences used for BPAG1-e KD was 50- 48 hours to form confluent monolayers. Cells were then fixed for TAATTCTGAGTAATACAGAGC-30 and 50-AAACCCAGGGCTGCCTT 20 minutes with 4% paraformaldehyde, permeabilized for 2 minutes GGAAAAG-30 and 50-AAACCCAGGGCTGCCTTGGAAAAG-30 were with 0.1% Tx-100 in PBS, blocked for 60 minutes in 10% fetal calf used for K14KD1 and K14KD2, respectively. Lentiviral packaging serum in PBS, and incubated with primary antibody for 90 minutes. plasmids: pMDG, and pD8.91. Primary antibodies human-anti-BP230, Cell were washed in PBS and incubated with secondary antibody for plectin, collagenXVII, a6 integrin, b4 integrin and K14 were used 45 minutes and washed again before mounting on glass slides using previously in (Groves et al., 2010). Other primary antibodies: mouse FluorSafe reagent. For skin immunofluorescence microscopy, the anti-BP230 (2B Scientific, Oxford, UK), total b1 integrin (Millipore, tissue was fixed for 30 minutes, permeabilized for 5 minutes, blocked Watford, UK), active b1 integrin (clone 12G10, Millipore), and for 1 hour before overnight incubation with the primary antibody. involucrin (Serotec, Oxford, UK). Secondary antibodies: horseradish Samples were washed in PBS, incubated with the secondary antibody

780 Journal of Investigative Dermatology (2014), Volume 134 MMichaelet al. BPAG1-e Regulation of Keratinocyte Migration

for 90 minutes, washed again, and mounted. Confocal microscopy CT-30)andGAPDH (forward 50-AGAAGGCTGGGGCTCATTTG-30, was performed on the Zeiss LSM510 point scanning microscope (Carl reverse 50-AGGGGCCATCCACAGTCTTC-30). Zeiss, Oberkochen, Germany). Wide-field microscopy as performed on the Olympus IX81 (Olympus KeyMed, Southend, UK) (Meta- CONFLICT OF INTEREST morph, Sutter filter wheels, ASI X-Y stage, Photometrics CascadeII The authors state no conflict of interest. 512B camera). ACKNOWLEDGMENTS Organotypic cultures This research was funded by the Dunhill Medical Trust, the Royal Society, and the British Skin Foundation. The work was also supported by the UK National The development of organotypic cultures was performed similar Institute for Health Research (NIHR) Biomedical Research Centre based at to the protocol used in Nystrom et al. (2005). Briefly, keratino- Guy’s and St Thomas’ NHS Foundation Trust and King’s College London. The cytes were grown on a matrix of collagen and Matrigel (1:1) views expressed are those of the authors and not necessarily those of the NHS, the NIHR, or the UK Department of Health. embedded with human dermal fibroblasts. After 48 hours, keratinocytes on the gel was raised onto an air–liquid interface and grown for 11 days, followed by paraformaldehyde fixation and paraffin SUPPLEMENTARY MATERIAL embedding. Supplementary material is linked to the online version of the paper at http:// www.nature.com/jid Cell adhesion and detachment assay For the adhesion assay, 100,000 cells per well were plated onto REFERENCES matrigel-coated coverslips placed in 24-well dishes. Adherent cells Borradori L, Sonnenberg A (1999) Structure and function of hemidesmosomes: were then fixed after 30, 60, 120 minutes by rinsing twice with PBS more than simple adhesion complexes. J Invest Dermatol 112:411–8 and paraformaldehyde fixation for 20 minutes. Cells were stained Bruckner-Tuderman L, McGrath JA, Robinson EC et al. (2013) Progress in epidermolysis bullosa research: summary of DEBRA international research with 4,6-diamidino-2-phenylindole and phalloidin-Alexa568 and conference 2012. J Invest Dermatol 133:2121–6 5 5 tilescan images were acquired on a wide-field microscope. Coulombe PA, Hutton ME, Letai A et al. (1991) Point mutations in human Images were quantified using CellProfiler image analysis software keratin 14 genes of epidermolysis bullosa simplex patients: genetic and (Broad Institute, Cambridge, MA). For the cell detachment assays, cells functional analyses. Cell 66:1301–11 were plated at 50,000 cells per well in 24-well dishes and grown for deHart GW, Healy KE, Jones JCR (2003) The role of a3b1 integrin in 48 hours. The number of detached cells was then counted following determining the supramolecular organization of laminin-5 in the extracellular matrix of keratinocytes. Exp Cell Res 283:67–79 5 minutes of trypsin-EDTA incubation and normalized against the DiPersio CM, van der Neut R, Georges-Labouesse E et al. (2000) alpha3beta1 number of detached cells after 15 minutes (total cell count). and alpha6beta4 integrin receptors for laminin-5 are not essential for epidermal morphogenesis and homeostasis during skin development. Migration assays JCellSci113:3051–62 Phase contrast time-lapse imaging of cells were performed on a Zeiss Fine JD, Eady RA, Bauer EA et al. (2008) The classification of inherited Axio100 microscope equipped with Sensicam CCD camera (PCO epidermolysis bullosa (EB): Report of the Third International Consensus Meeting on Diagnosis and Classification of EB. J Am Acad Dermatol Cooke), motorized stage (Ludl) and excitation/emission filters 58:931–50 (Chroma) and filter wheels (Ludl). Images were acquired using a Fine JD, Mellerio JE (2009a) Extracutaneous manifestations and complications 10x phase objective. For random migration analysis, cells were plated of inherited epidermolysis bullosa: part I. Epithelial associated tissues. for 4 hours at 8,000 cells per well in 12-well dishes. Time-lapse J Am Acad Dermatol 61:367–84 movies were taken at 5 minutes per frame for 8 hours. The cell movies Fine JD, Mellerio JE (2009b) Extracutaneous manifestations and complications were then tracked using the AndorIQ (Andor, Belfast, UK) software of inherited epidermolysis bullosa: part II. Other organs. JAmAcad and migration speeds and persistence was analyzed using in- Dermatol 61:387–402 house Mathematica6 (Wolfram) notebooks written and designed by Friedl P, Gilmour D (2009) Collective cell migration in morphogenesis, regeneration and cancer. Nat Rev Mol Cell Biol 10:445–57 Professor Graham Dunn (King’s College London, London, UK). Grose R, Hutter C, Bloch W et al. (2002) A crucial role of b1 integrins for keratinocyte migration in vitro and during cutaneous wound repair. Reverse-transcriptase–PCR Development 129:2303–15 WT and DST1 keratinocytes were subject to RNA extraction (Qiagen, Groves RW, Liu L, Dopping-Hepenstal PJ et al. (2010) A homozygous nonsense Manchester, UK) followed by reverse transcription (Superscript III, mutation within the dystonin gene coding for the coiled-coil domain of the Invitrogen) according to the manufacturer’s instructions. Gene expres- epithelial isoform of BPAG1 underlies a new subtype of autosomal recessive epidermolysis bullosa simplex. J Invest Dermatol 130:1551–7 sion analysis of the BPAG isoforms in WT and DST1 cells were Guo L, Degenstein L, Dowling J et al. (1995) Gene targeting of BPAG1: performed using Taqman gene expression assays (Applied Biosystems, abnormalities in mechanical strength and cell migration in stratified Life Technologies, Paisley, UK). The primer/probe set used for each epithelia and neurologic degeneration. Cell 81:233–43 isoform are as follows: total BPAG1 (Hs00156137_m1), BPAG1-e Hamill KJ, Hopkinson SB, DeBiase P et al. (2009) BPAG1e maintains (Hs01094846_g1), BPAG1-b (Hs01094902_m1), and BPAG-a keratinocyte polarity through beta4 integrin-mediated modulation of (Hs01094853_m1). Semiquantitative reverse-transcriptase–PCR Rac1 and cofilin activities. Mol Biol Cell 20:2954–62 was performed using primers against HES1 (forward 50-CGGACATT Jefferson JJ, Leung CL, Liem RK (2006) Dissecting the sequence specific functions of alternative N-terminal isoforms of mouse bullous pemphigoid CTGGAAATGACA-30,reverse50-TACTTCCCCAGCACACTTGG-30), antigen 1. Exp Cell Res 312:2712–25 NOTCH3 (forward 50-GTGTGTGTCAATGGCTGGAC-30,reverse50- Koster J, Geerts D, Favre B et al. (2003) Analysis of the interactions between 0 0 GTGACACAGGAGGCCAGTCT-3 ), KRT14 (forward 5 -CCCTACTTC BP180, BP230, plectin and the integrin alpha6beta4 important for AAGACCATTGAGGAC-30,reverse50-GTGACACAGGAGGCCAGT hemidesmosome assembly. JCellSci116:387–99

www.jidonline.org 781 MMichaelet al. BPAG1-e Regulation of Keratinocyte Migration

Leung CL, Liem RK, Parry DA et al. (2001a) The plakin family. JCellSci Nystrom ML, Thomas GJ, Stone M et al. (2005) Development of a quantitative 114:3409–10 method to analyse tumour cell invasion in organotypic culture. J Pathol Leung CL, Zheng M, Prater SM et al. (2001b) The BPAG1 locus: Alternative 205:468–75 splicing produces multiple isoforms with distinct cytoskeletal linker Ridley AJ, Schwartz MA, Burridge K et al. (2003) Cell migration: integrating domains, including predominant isoforms in neurons and muscles. J Cell signals from front to back. Science 302:1704–9 Biol 154:691–7 Seltmann K, Roth W, Kroger C et al. (2013) Keratins mediate localization of Litjens SH, de Pereda JM, Sonnenberg A (2006) Current insights into the hemidesmosomes and repress cell motility. J Invest Dermatol 133:181–90 formation and breakdown of hemidesmosomes. Trends Cell Biol 16: Sonnenberg A, Liem RK (2007) Plakins in development and disease. Exp Cell 376–83 Res 313:2189–203 Liu L, Dopping-Hepenstal PJ, Lovell PA et al. (2012) Autosomal recessive Tsuruta D, Hashimoto T, Hamill KJ et al. (2011) Hemidesmosomes and focal epidermolysis bullosa simplex due to loss of BPAG1-e expression. J Invest contact proteins: functions and cross-talk in keratinocytes, bullous Dermatol 132:742–4 diseases and wound healing. J Dermatol Sci 62:1–7 Margadant C, Charafeddine RA, Sonnenberg A (2010) Unique and redundant Young KG, Kothary R (2007) Dystonin/Bpag1–a link to what? Cell Motil functions of integrins in the epidermis. FASEB J 24:4133–52 Cytoskeleton 64:897–905 Morley SM, Dundas SR, James JL et al. (1995) Temperature sensitivity of the Young KG, Kothary R (2008) Dystonin/Bpag1 is a necessary endoplasmic keratin cytoskeleton and delayed spreading of keratinocyte lines derived reticulum/nuclear envelope protein in sensory neurons. Exp Cell Res from EBS patients. JCellSci108(Pt 11):3463–71 314:2750–61

782 Journal of Investigative Dermatology (2014), Volume 134