Kazrin regulates keratinocyte cytoskeletal networks, intercellular junctions and differentiation

The outermost layers of the epidermis, the stratum corneum, form a protective barrier between the body and the external environment. Cells in the stratum corneum assemble a structure known as the cornified envelope at the inner surface of the plasma membrane (Candi et al., 2005). The cornified envelope consists of a meshwork of insoluble, transglutaminase cross-linked proteins and lipids essential for the mechanical integrity and water impermeability of the skin. As stratum corneum cells detach from the epidermal surface they are replaced by proliferation of undifferentiated basal layer cells, whose progeny terminally differentiate as they move through the suprabasal layers. The balance between proliferation and differentiation in the epidermis is influenced by signals emanating from the basement membrane, intercellular junctions and their associated cytoskeletal networks (Watt, 2001; Wheelock and Johnson, 2003; Yin and Green, 2004).

One of the proteins incorporated into the cornified envelope is periplakin (Ruhrberg et al., 1997), a member of the plakin family of proteins that link cytoskeletal networks to each other and to membrane-associated adhesive sites (Jefferson et al., 2004). The periplakin C terminus directly binds intermediate filaments, while the N terminus associates with cortical actin and desmosomes (DiColandrea et al., 2000). We recently reported that the N terminus of periplakin binds a novel, highly conserved protein, kazrin (Groot et al., 2004). Thus far, four transcript variants of kazrin have been identified encoding three proteins with different N termini (kazrinA, B and C). Kazrin localizes to the nucleus, cell membrane, desmosomes and adherens junctions (Gallicano et al., 2005; Groot et al., 2004).

The direct binding of intermediate filaments by periplakin, together with the subcellular distributions of periplakin and kazrin, suggest that both proteins participate in the known interplay between desmosomes and adherens junctions (Huen et al., 2002; Lewis et al., 1997; Vasioukhin et al., 2001). Kazrin is widely expressed, suggestive of a function that extends beyond assembly of the cornified envelope (Groot et al., 2004). In the present article, we have uncovered new roles for kazrin in regulating cytoskeletal organization, intercellular junction assembly and differentiation.

Our previous data showed that kazrin localizes to desmosomes, the interdesmosomal plasma membrane and the nucleus of epidermal keratinocytes both in vitro and in vivo (Groot et al., 2004). However, the expression and subcellular localization of kazrin during the early stages of intercellular junction assembly and terminal differentiation have not been investigated. To examine this, keratinocytes were cultured to confluence in low Ca²⁺ medium (containing 0.09 mM Ca²⁺), and then transferred to standard (`high Ca²⁺') medium containing 1.8 mM Ca²⁺ to stimulate intercellular junction assembly and accumulation of terminally differentiated cells (Green et al., 1987; Vasioukhin et al., 2000; Watt et al., 1984).

We found that endogenous kazrin protein levels increased twofold following 24 hours incubation in high Ca²⁺ medium (Fig. 1A). Levels were also elevated at 48 hours. However, there was a decrease at 72 hours, which probably reflects kazrin becoming insoluble as it is incorporated into the cornified envelope (Groot et al., 2004). Within 6 hours of addition of Ca²⁺, kazrin had accumulated at cell-cell borders where it colocalized with the desmosomal marker, desmoplakin (Fig. 1B,C). By 24 hours, kazrin and desmoplakin levels at cell-cell borders were further increased (Fig. 1D,E). At all time points kazrin was also detected in the nucleus (Fig. 1B-E). We conclude that although kazrin is expressed in all the living layers of the epidermis (Groot et al., 2004), it is upregulated in suprabasal, differentiating cells.

As we observed an increase in kazrin expression during keratinocyte differentiation, we evaluated the consequences of overexpressing kazrin in undifferentiated primary human keratinocytes. When expressed at moderate levels in all cells through retroviral infection, N-terminally HA-tagged kazrinA was present in the nucleus and colocalized with periplakin at cell-cell borders and the interdesmosomal plasma membrane (Fig. 2A-C). Thus the distribution of overexpressed kazrinA was the same as that of the endogenous protein (Fig. 1B-E) (Groot et al., 2004).

The consequences of higher levels of kazrinA overexpression were evaluated by transient transfection of human keratinocytes. Cells were transfected and incubated overnight in low Ca²⁺ medium, and then transferred to high Ca²⁺ medium for 2 hours in order to stimulate assembly of adherens junctions and desmosomes (Green et al., 1987; Vasioukhin et al., 2000; Watt et al., 1984). Cells transiently overexpressing N-terminally HA-tagged kazrinA exhibited profound changes in cell shape compared to cells transfected with a control GFP construct (Fig. 2D-F). KazrinA overexpression resulted in loss of the normal cuboidal shape characteristic of epithelial cells, and this was reflected in a statistically significant decrease in shape factor [defined as 4π× (area/perimeter²); Fig. 2G-I]. Although the average cell area available for contact with the substrate was not significantly affected, the average perimeter of kazrinA-overexpressing cells was dramatically increased, reflecting the kazrinA-induced shape changes (data not shown). Keratinocytes expressing higher levels of kazrinA (Fig. 2H, arrowheads) exhibited more dramatic changes in shape than cells with lower expression levels (Fig. 2H, asterisks). KazrinA-overexpressing cells extended membrane projections over neighbouring cells (Fig. 2E,F, arrowheads), often appearing to be in the process of stratifying. Overexpression of kazrinA, B and C isoforms resulted in similar changes in keratinocyte shape (Fig. 2J). We therefore focused on the most evolutionarily conserved isoform, kazrinA, in subsequent experiments.

Compared with control cells transfected with GFP (Fig. 3A), cells overexpressing kazrinA exhibited abnormal spreading and stretching (Fig. 3B,C,E-H). KazrinA overexpression impaired formation of the cortical actin band normally present in keratinocytes (see Fig. 3A) and led to an overall decrease in filamentous actin (F-actin), detected by phalloidin (Fig. 3B,C,E,F). This was confirmed by determining the ratio of F- to G-actin in lysates of keratinocytes transduced with GFP or kazrinA retroviral vectors (Fig. 3D). The F-actin:G-actin ratio in cells overexpressing kazrinA was 0.12, compared with 0.3 in cells overexpressing GFP (Fig. 3D). Lysates were incubated with phalloidin as a positive control for accumulation of F-actin.

In contrast to the effects of kazrinA overexpression on the actin cytoskeleton, the microtubule network was intact in kazrinA transfectants (Fig. 3G,H). However, keratin filaments were reorganized and did not extend into cell-cell contacts (Fig. 3I-M). Similar effects on cell shape and the cytoskeleton were observed in cells transfected with C-terminally HA-tagged or N- or C-terminally FLAG-tagged kazrinA, B or C constructs (data not shown).

We next investigated whether kazrinA overexpression affected intercellular junction assembly. As before, keratinocytes were transiently transfected in low Ca²⁺ medium and then transferred to standard medium for 2 hours prior to analysis. Overexpression of GFP did not prevent the accumulation of desmoplakin at desmosomes or E-cadherin at adherens junctions (Fig. 4A-D). By contrast, cells overexpressing kazrinA exhibited a dramatic reduction in desmoplakin staining at cell-cell borders (Fig. 4E-H,K). Although kazrinA transfection did not prevent E-cadherin localization to cell-cell borders, E-cadherin organisation did appear to be altered relative to cells transfected with GFP (Fig. 4I,J,L).

We examined whether modulation of cell signalling pathways known to regulate the cytoskeleton and cell-cell adhesion would affect kazrinA-induced cell shape changes. Keratinocytes were incubated following transfection and during the 2 hour Ca²⁺ switch in the presence or absence of small molecule inhibitors of Rho kinase (ROCK; Fig. 5A-D), PI 3-kinase and the EGF receptor (data not shown). Cell shape, as quantified by calculating the shape factor, of GFP transfectants was not significantly affected by the inhibitor treatments (Fig. 5A,B and data not shown). Treatment with inhibitors of PI 3-kinase and the EGF receptor did not result in statistically significant changes in the shape factor of kazrinA transfectants (data not shown). By contrast, treatment with the ROCK inhibitor Y27632 resulted in a dramatic exacerbation of the kazrinA-induced cell shape changes, the shape factor decreasing from 0.48 in untreated kazrinA transfectants to 0.25 in treated cells (Fig. 5C,D).

ROCK is activated by GTP-bound RhoA and RhoC and promotes the formation of contractile actin stress fibres (Riento and Ridley, 2003). The observations of decreased F-actin, reorganized intermediate filaments, defects in intercellular junction assembly and hypersensitivity to ROCK inhibition upon kazrinA overexpression suggested a reduction in Rho activity (Braga et al., 1997; Jaffe and Hall, 2005; Waschke et al., 2006). To directly assess levels of active Rho, we stably transduced keratinocytes with retroviral vectors expressing GFP or kazrinA and performed pull-downs of cell lysates with the Rho binding domain of Rhotekin fused to GST (Fig. 5E,G). KazrinA overexpression resulted in an average decrease of 40% in levels of active Rho relative to the GFP control (Fig. 5E,G). By contrast, there was no detectable effect of kazrinA overexpression on levels of active GTP-bound Rac1 (Fig. 5F), or on activation (as judged by tyrosine phosphorylation and membrane recruitment) of p190Rho GAP, which occurs downstream of Rac1 activation (data not shown) (Nimnual et al., 2003).

To test whether increased Rho activity could block kazrinA-induced changes in cell shape, actin organisation and desmosome assembly, keratinocytes were co-transfected with kazrinA and either GFP or dominant-active V14RhoA (Fig. 5H-P). Keratinocytes expressing kazrinA alone or together with GFP displayed the expected shape alterations and decreased cortical actin (Fig. 5I-L, open arrows). Keratinocytes expressing V14RhoA alone displayed an increase in cortical actin and stress fibres (Fig. 5K,L, filled arrows). Over 80% of cells expressing both kazrinA and V14RhoA retained the cortical actin band and cuboidal shape characteristic of untransfected keratinocytes (Fig. 5H,K,L, arrowheads). In addition, relative to cells overexpressing kazrinA alone, those co-expressing V14RhoA displayed increased desmoplakin at cell-cell borders (Fig. 5M-P). We conclude that overexpression of kazrinA results in a decrease in levels of active Rho and that this is responsible for the changes in cell shape, cytoskeletal organization and desmosome assembly.

The morphology of kazrinA-overexpressing keratinocytes was reminiscent of either highly motile cells or cells undergoing terminal differentiation (Magee et al., 1987; Watt and Green, 1982). To examine the effects of kazrin on cell motility, we performed video microscopy on keratinocytes transiently transfected with GFP-kazrinA. GFP-kazrinA transfectants did not display increased motility. The cells stretched and elongated but did not migrate far from their initial positions (Fig. 6A-F, green cells, black and white arrows; see Movie 1 in supplementary material). The GFP-kazrinA-expressing keratinocytes did not exhibit increased apoptosis, as assessed by TUNEL and annexin V staining (data not shown).

To determine whether the kazrinA-induced changes in cell shape reflected induction of terminal differentiation, transiently transfected cells were examined for expression of the cornified envelope protein involucrin (Fig. 6H,I). At 48 hours post transfection, an increased proportion of kazrinA-overexpressing cells stained positive for involucrin (Fig. 6I) relative to those expressing GFP (Fig. 6H). Similar results were obtained with the terminal differentiation marker transglutaminase 1 (data not shown). The morphological changes induced by kazrinA, including stretching and extension of processes over neighbouring cells, were strikingly similar to the morphology of keratinocytes undergoing spontaneous terminal differentiation (Fig. 6H,I).

For further analysis of the effects of kazrin on terminal differentiation, we examined keratinocytes stably transduced with the kazrinA retroviral vector. In clonogenicity assays, kazrinA-infected cells produced substantially fewer colonies than cells transduced with GFP, with a substantial reduction in colony-forming efficiency observed in three independent experiments (Fig. 6G and data not shown). Moreover, kazrinA-infected colonies were smaller than GFP control colonies, reflecting decreased proliferative potential. KazrinA-infected cells within the colonies were enlarged, which is indicative of terminal differentiation (Fig. 6G).

Keratinocytes undergoing terminal differentiation exhibit increased forward and side scatter, measured by flow cytometry, reflecting their increase in size and granularity, respectively (Jones and Watt, 1993). Irrespective of whether the cells were grown to confluence in the absence of feeders in low Ca²⁺ (Fig. 6J, upper panels) or standard (Fig. 6J, lower panels) medium, kazrinA-infected cells showed an increase in the population with high forward and side scatter relative to GFP control cells. Consistent with the increase in forward and side scatter, kazrinA-infected cultures had a higher proportion of cells expressing the terminal differentiation markers involucrin and transglutaminase 1 than cells infected with GFP (Fig. 6K,L). Thus overexpression of kazrinA stimulates terminal differentiation, as evaluated by four different criteria (reduced clonal growth, increased forward and side scatter, and induction of involucrin and transglutaminase 1 expression).

To examine the effects of decreased kazrin expression, two different siRNA (siRNA426 and 1077) sequences specific for kazrin were stably expressed in primary human keratinocytes via retroviral infection. siRNA426 reduced kazrin mRNA levels by an average of 60% relative to the scrambled siRNA control (Fig. 7A). siRNA1077 was less effective, causing an average decrease in kazrin mRNA expression of approximately 30% (Fig. 7A). Keratinocytes expressing either siRNA construct had decreased kazrin protein levels, with siRNA426 being more effective than siRNA1077 (Fig. 7B). Although kazrinA overexpression reduced levels of GTP-bound RhoA (Fig. 5E,G), we did not detect significant changes in Rho activity in keratinocytes expressing either siRNA construct relative to the control (Fig. 7C). Consistent with this, kazrin knockdown did not affect the levels of desmoplakin at cell-cell borders when monitored at 15 minute intervals for 2 hours following a Ca²⁺ switch (Fig. 7D,E and data not shown).

Although kazrin knockdown did not affect levels of active Rho, keratinocytes expressing either siRNA construct exhibited a dramatic increase in proliferation relative to control cells (Fig. 8A). We also observed a reproducible and statistically significant increase in the proportion of keratinocytes with low forward and side scatter (undifferentiated) in populations expressing siRNA426 relative to cells expressing the scrambled siRNA control (paired Student's t-test, P<0.02, n=5; Fig. 8B and data not shown). This correlated with statistically significant decreases in the proportions of cells expressing involucrin (paired Student's t-test, P<0.02, n=5; Fig. 8C and data not shown) and transglutaminase 1 (paired Student's t-test, P<0.05, n=5) (Fig. 8D and data not shown). siRNA1077, which was less effective at reducing kazrin levels (Fig. 7A,B) did not cause significant changes in forward and side scatter or transglutaminase 1 expression but did lead to a significant reduction in involucrin-positive cells (paired Student's t-test, P<0.05, n=5; data not shown). The effects of siRNA426 on keratinocyte differentiation were specific as they were rescued by co-expression of a mutated construct of kazrinA (KA-mut) which is resistant to the siRNA (Fig. 8E,F). Taken together with the overexpression data, these results indicate that kazrin plays an important role in regulating keratinocyte terminal differentiation.

We have found that overexpression of kazrin isoforms A, B or C by transient transfection resulted in striking changes in cell morphology and cytoskeletal organization in primary human epidermal keratinocytes and led to reduced accumulation of desmoplakin and altered organization of E-cadherin at cell-cell borders. The level of GTP-bound active Rho, but not Rac1, was reduced in keratinocytes overexpressing kazrinA and the effects of kazrinA on cell shape, cytoskeletal organization and desmosome assembly were rescued by co-expression of dominant active V14RhoA. Our results are consistent with the known role of RhoA in regulating the organization of both the actin (for a review, see Jaffe and Hall, 2005) and the intermediate filament cytoskeleton (Waschke et al., 2006).

The levels of Rho GTPase activity are tightly controlled during establishment of cell-cell adhesions (Yamada and Nelson, 2007), and dysregulation can interfere with the assembly or the maintenance of intercellular junctions (Braga et al., 1997; Takaishi et al., 1997; Zhong et al., 1997). Although the mechanism by which kazrin reduces the levels of GTP-bound Rho remains to be determined, one possibility is that kazrin recruits a GTPase-activating protein to the membrane, which would stimulate GTP hydrolysis and result in inactivation of Rho. The inhibitory effect of kazrin on desmosome assembly is in good agreement with a recent study which found that dissociation of desmosomes by pemphigus vulgaris and foliaceus autoantibodies is mediated by a decrease in GTP-bound RhoA (Waschke et al., 2006).

The effects of kazrin overexpression on desmosome assembly parallel those observed when desmoglein 1 is overexpressed in transformed human keratinocytes (Hanakawa et al., 2002; Norvell and Green, 1998). Thus low levels of desmoglein 1 or kazrin can incorporate into junctions, whereas high levels disrupt cell-cell adhesion (Hanakawa et al., 2002; Norvell and Green, 1998). It is possible that upregulation of kazrin and desmoglein 1 as cells differentiate results in remodelling of basal layer desmosomes, facilitating cell movement from the basal to the suprabasal layers.

The changes in cell shape induced by kazrin are highly reminiscent of those that occur when keratinocytes leave the basal layer at the onset of terminal differentiation (Magee et al., 1987; Watt and Green, 1982) and this, together with the increase in kazrin protein levels during keratinocyte differentiation, led us to investigate whether kazrin modulates differentiation. Kazrin overexpression stimulated differentiation whereas knockdown resulted in increased proliferation and a decreased proportion of differentiated cells. Kazrin is notable for being the only cornified envelope precursor thus far identified that regulates terminal differentiation (Candi et al., 2005).

In primary mouse keratinocytes inhibition of Rho activity has been linked to induction of terminal differentiation, whereas dominant active V14RhoA can maintain the cells in an undifferentiated state (Grossi et al., 2005). By contrast, the inhibition of terminal differentiation resulting from decreased kazrin expression was not correlated with an overall increase in active Rho, although there remains a possibility that levels of GTP-bound Rho were affected at specific cellular locations in the kazrin-knockdown cells. We did not detect a significant effect of kazrin knockdown on intercellular junction assembly during the first 2 hours after a Ca²⁺ switch; at later times (24 and 48 hours), stratification was reduced, but this could reflect the inhibition of terminal differentiation, rather than an effect on junctional dynamics (data not shown). Our current hypothesis is that kazrin regulates desmosome assembly through inhibition of Rho, and stimulates commitment to differentiation through a Rho-independent pathway.

Recent studies in Xenopus tropicalis embryos demonstrate effects of kazrin that are independent of periplakin binding. Knockdown of kazrin impairs axial elongation, cell differentiation and epidermal morphogenesis in the developing embryo (Sevilla et al., 2008). A kazrin construct lacking the periplakin-binding domain can partially rescue the developmental defects associated with knockdown of kazrin. All kazrin isoforms have a nuclear localization sequence (Groot et al., 2004) and it will be interesting to investigate whether the periplakin- and Rho-independent effects of kazrin are dependent on nuclear accumulation.

Extensive studies in keratinocytes have shown that cell-substratum adhesion, mediated by integrins with involvement of Rac1, negatively regulates terminal differentiation (Benitah et al., 2005; Watt, 2001). Other more recent studies have indicated that cell-cell adhesion mediated by desmosomes and adherens junctions contributes to epidermal cell fate decisions (Elias et al., 2001; Hardman et al., 2005; Merritt et al., 2002; Tinkle et al., 2004). It is tantalizing to speculate that with its dual effects on differentiation and desmosome assembly, kazrin coordinates the onset of terminal differentiation with movement out of the epidermal basal layer.

The constructs encoding N-terminally HA-tagged kazrinA, kazrinB and kazrinC in the EFplinkHA vector have been described previously (Groot et al., 2004). pCIneo kazrinA was used in all transient transfections of kazrinA with the exception of those in Fig. 2J. This construct was generated by ligating the XhoI fragment from EFplinkHA-kazrinA into similarly digested pCIneo. The EFplinkHA-kazrinA construct was digested with XhoI to release the HA-kazrinA encoding fragment, which was then blunt ended with Klenow, digested with SalI and inserted into the SnaBI-SalI sites of the pBABEpuro retroviral vector. The GFP-kazrinA fusion was created by subcloning PCR-amplified kazrinA into the EcoRI and BamHI sites of pEGFPC2 (Clontech Laboratories, Inc.) using the following primers: forward 5′-GCGGAATTCATGATGGAAGACAATAAG-3′, reverse 5′-GCGGGATCCCCAGTCCGCGTCCTCCTC-3′. The N-terminally myc-tagged V14RhoA construct in EFplink2 was a gift from R. Treisman (CRUK London Research Institute, UK).

Primary human keratinocytes from newborn foreskin (strains kv and kt, passages 2-4) were cultured and transfected as previously described (DiColandrea et al., 2000). In some cases, following a 1-2 hour recovery period after removal of the transfection mixture, the following inhibitors (Calbiochem) were added: Y27632 (15 μM), LY294002 (20 μM), AG1478 (20 μM) or 4557W (5 μM). Keratinocytes were infected with retroviral vectors as described previously (Gandarillas and Watt, 1997) and selected with 2 μg/ml puromycin (MP Biomedicals, Irvine, CA).

The following mouse monoclonal antibodies were used at the dilutions stated: 9E10 (1:100; anti-human myc; a gift from G. Evan, UCSF, CA), HA.11 (1:1000; anti-HA, Covance Research Products, Inc.), AC40 (1:500; anti-actin, Sigma-Aldrich), LP34 (1:100; anti-pan-keratin), clone 2-28-33 (1:5000; anti-β-tubulin, Sigma-Aldrich), 115F (1:50; anti-desmoplakin, a gift from D. R. Garrod, University of Manchester, UK), HECD-1 (1:100; anti-human E-cadherin, a gift from M. Takeichi, Riken Institute, Kobe, Japan), SY5 (1:100; anti-human involucrin), BC.1 (1:100; anti-transglutaminase 1, a gift from R. Rice, University of California, CA), Clone 55 (3 μg/ml; anti-RhoA, B and C; Millipore), Clone 26C4 (1:200; anti-RhoA; Santa Cruz Biotechnology). The rabbit antiserum Y11 (Santa Cruz Biotechnology) specific for HA was used at 1:50.

A peptide corresponding to amino acids 329-348 (DRSSTPSDINSPRHRTHSLC) of human kazrinA was used to generate antiserum LS7. LS7 was affinity purified on a column containing its immunogen coupled to Amino-Link Coupling Gel (Perbio, Erembodegem, Belgium).

For flow cytometry, SY5 and BC.1 were directly conjugated to Alexa Fluor 633 using the Protein Labeling Kit from Invitrogen. HA.11 coupled to Alexa Fluor 488 was from Covance Research Products, Inc. For immunofluorescence, Alexa Fluor 488-, 594- or 633-conjugated goat anti-rabbit or mouse IgG (Invitrogen) were diluted at 1:1000 and used for detection of primary antibodies. For immunoblotting and immunohistochemistry, HRP-conjugated donkey anti-rabbit or anti-mouse IgG was used at 1:4000 (Amersham Biosciences).

Immunofluorescence staining was performed as described previously (DiColandrea et al., 2000). To visualize F-actin, TRITC- or coumarin-conjugated phalloidin (Sigma) was included in the secondary antibody incubation. DAPI (Invitrogen) was included in the secondary antibody incubation to stain nuclei. All immunolabelled samples were mounted using Mowiol (Calbiochem). Images of fluorescently labelled specimens were taken at room temperature using a LSM 510 laser scanning confocal microscope equipped with 405 nm, 488 nm, 594 nm and 633 nm lasers (Carl Zeiss, Inc.). LSM 510 software was used to acquire images using the following immersion lenses (Carl Zeiss, Inc.): 10× C-Aprochromat (numerical aperture 0.45; water correction), 25× Plan-Neofluar (numerical aperture 0.80; immersion correction), 40× C-Aprochromat (numerical aperture 1.2; water correction).

Three-dimensional reconstructions of z-stacks were carried out using LSM 510 software. All images were further processed using Adobe Photoshop CS2 and compiled using Adobe Illustrator CS2. For quantifications, samples were scored blindly. Metamorph 7 (Molecular Devices) was used to calculate the fluorescence pixel intensity of desmoplakin and E-cadherin at cell-cell borders. This value was determined by multiplying the average pixel intensity by the area of the defined border divided by the border length.

Four slides with three coverslips each of immunolabelled transfected keratinocytes were automatically examined per experiment, using the high content imaging system Discovery-1 (Molecular Devices) with a Photometrics CoolSNAP HQ camera. Fluorescent images were acquired from four fields on each coverslip using a Nikon 20′ objective (numerical aperture 0.45 lens, air correction). DAPI, FITC and TRITC fluorochromes were excited by 360 nm, 470 nm and 535 nm light, respectively. Images were interactively analyzed using Metamorph 7 (Molecular Devices) software to determine cell outlines and shape factors [4π×(area/perimeter²)] for individual cells.

Analysis of cellular actin was performed using the G-actin/F-actin In Vivo Assay Kit (Cytoskeleton Inc., Denver, CO) in accordance with the manufacturer's instructions.

GST pull-downs for active Rho were performed as described previously (Wells et al., 2005), using the Rho binding domain of Rhotekin fused to GST coupled to glutathione agarose (Millipore). GST pull-downs for active Rac1 were performed using the Rac1/Activation Assay Kit (Millipore). Immunoblotting was performed as described previously (Groot et al., 2004).

Time-lapse recordings of cells transiently transfected with GFP-kazrinA were made with a Zeiss Axiovert 135T inverted microscope equipped with a warm box (37°C) with a CO₂ source, Hamamatsu Orca ER firewire CCD camera, xyz motorized stage, a piezo z control and a 10× PlanAPOCHROMAT objective (numerical aperture 0.32; air correction; Carl Zeiss, Inc.). Images were obtained at 5-minute intervals using AQMAdvance6 Kinetic Acquisition Manager (Medical Solutions, PLC) and subsequently prepared with Adobe ImageReady. Cell tracking was performed using Metamorph software (Molecular Devices).

Intracellular staining of keratinocytes for involucrin and transglutaminase 1 was performed as described previously (Gandarillas and Watt, 1997), except that cells were fixed in 4% paraformaldehyde for 10 minutes at room temperature and antibodies used were directly conjugated to fluorophores. Flow cytometry was performed with the BD FACSCalibur System (BD Biosciences) using BD CellQuest software. FlowJo software (Tree Star) was used to further analyze the flow cytometry data.

Two strategies were used to find sequences suitable for siRNA-mediated knockdown of kazrin. The first was independent design of oligonucleotides following published guidelines, which resulted in the identification of the sequence targeted by siRNA426 (AGCTGATCGGAAGCGCTTA). This sequence was cloned into the pRetroSuper (Brummelkamp et al., 2002) vector (H1 promoter). The second method was screening Silencer® pre-designed siRNAs (Ambion) specific for human kazrin via transfection in 293 cells using Lipofectamine 2000 (Invitrogen). This resulted in the identification of the sequence targeted by siRNA1077 (CCTGCACAACCCTATTGTA), which was cloned into pSilencer 5.1-H1 Retro (Ambion). The negative controls were the sequence for Scrambled 2 from Ambion cloned into pSilencer 5.1-H1 Retro or pRetroSuper (GTACTGCTTACGATACGG). Kazrin knockdown was assessed using quantitative RT-PCR and immunoblotting.

Silent mutations were generated in the cDNA encoding the region of kazrinA recognized by siRNA426 using the QuickChange II Site-Directed Mutagenesis Kit (Stratagene). The following primers were used to introduce four base changes, which are indicated in lowercase letters: CCTTGCAGGCCATGAAAGCcGAcCGaAAaCGCTTAAAGGGCGAGAAGA and TCTTCTCGCCCTTTAAGCGtTTtCGgTCgGCTTTCATGGCCTGCAAGG.

RNA from cultured keratinocytes was isolated using TriReagent (Sigma) and cDNA was generated with the Superscript First-Strand Synthesis System for RT-PCR (Invitrogen). Quantitative PCR was performed using pre-designed TaqMan gene expression assays (Applied Biosystems) with the 7900HT Fast Real-Time PCR System (Applied Biosystems). Quantification was based on ΔΔCt calculations, and all samples were compared against β2 microglobulin as an endogenous control.

Keratinocytes were seeded at a density of 2000 cells/well, in triplicate, in a 96-well plate and cultured in keratinocyte-serum-free medium for 1, 2, 5 or 7 days prior to assessment of proliferation using the MTT assay (Sigma-Aldrich) according to the manufacturer's instructions.

This work was supported by Cancer Research UK and the Wellcome Trust. L.M.S. is the recipient of a National Institutes of Health fellowship (HD42379-02). We acknowledge the support of the University of Cambridge and Hutchinson Whampoa Ltd. We are most grateful to the Light Microscopy Department in the London Research Institute for providing assistance with image acquisition and analysis. We thank Erik Sahai for helpful discussions.