Research Article 2705 The MAGUK-family CASK is targeted to nuclei of the basal epidermis and controls keratinocyte proliferation

Nkemcho Ojeh, Vanja Pekovic, Colin Jahoda and Arto Määttä* School of Biological and Biomedical Sciences, Durham University, Durham DH1 3LE, UK *Author for correspondence (e-mail: [email protected])

Accepted 30 May 2008 Journal of Cell Science 121, 2705-2717 Published by The Company of Biologists 2008 doi:10.1242/jcs.025643

Summary The Ca2+/calmodulin-associated Ser/Thr (CASK) binds cell cycling in CASK knockdown cells is associated with syndecans and other cell-surface through its PDZ upregulation of Myc and hyperphosphorylation of Rb. domain and has been implicated in synaptic assembly, epithelial Moreover, CASK-knockdown cells show increased polarity and neuronal transcription. We show here that hyperproliferative response to KGF and TGFα, and accelerated CASK regulates proliferation and adhesion of epidermal attachment and spreading to the collagenous matrix. These keratinocytes. CASK is localised in nuclei of basal keratinocytes functions are reflected in wound healing, where CASK is in newborn rodent skin and developing hair follicles. Induction downregulated in migrating and proliferating wound-edge of differentiation shifts CASK to the cell membrane, whereas keratinocytes. in keratinocytes that have been re-stimulated after serum starvation CASK localisation shifts away from membranes upon entry to S phase. Biochemical fractionation demonstrates that Supplementary material available online at CASK has several subnuclear targets and is found in both http://jcs.biologists.org/cgi/content/full/121/16/2705/DC1 nucleoplasmic and nucleoskeletal pools. Knockdown of CASK by RNA interference leads to increased proliferation in cultured Key words: CASK, Keratinocytes, Nucleus, Proliferation, Wound keratinocytes and in organotypic skin raft cultures. Accelerated healing

Journal of Cell Science Introduction CASK belongs to the membrane-associated guanylate kinase Epidermal differentiation and wound healing are regulated by (MAGUK) family of scaffolding molecules that are associated with several signalling pathways that form an interconnected network intercellular junctions. Similar to the other MAGUK-family (Blanpain et al., 2006; Watt et al., 2006). Growth factors and members, CASK consists of an N-terminal PDZ domain, a central morphogens that are involved in epidermal biology often require SH3 domain and a C-terminal guanylate-kinase homology domain heparan sulfate (HS) as co-receptor for signalling. Examples of (reviewed by Funke et al., 2005). In addition, CASK is unique in HS-dependent signalling proteins include members of the that it also contains an N-terminal Ca2+/calmodulin-dependent fibroblast growth factor (FGF) family (Rapraeger et al., 1991; protein kinase II (CaMKII) domain (Hata et al., 1996). CASK has Guimond et al., 1993; Loo and Salmivirta, 2002), Wnt proteins been independently identified in Caenorhabditis elegans as LIN- (Tsuda et al., 1999; Alexander et al., 2000; Baeg et al., 2001) and 2, a protein required for EGF-receptor localisation during vulval members of the transforming growth factor β (TGFβ) and bone development, and in rat brain as a PDZ-domain protein that binds morphogenetic protein (BMP) families (Lyon et al., 1997; Rider, and syndecans (Hata et al., 1996; Hsueh et al., 1998). 2006). Moreover, the C-terminal guanylate-kinase domain of CASK is a Syndecans are cell-surface transmembrane HS proteoglycans pseudokinase that is involved in re-targeting the protein to the and are likely to be modulators of the signalling by these growth nucleus where it, at least in the case of neuronal cells, interacts with factors. Syndecans influence cellular signalling by two distinct the T-brain (TBR1) transcription factor and the CASK-interacting mechanisms. First, HS chains are required for the formation of nucleosome-assembly protein (CINAP), which regulate neuronal signalling complex between several different growth factors and (Hsueh et al., 2000; Wang et al., 2004). Thus, CASK their cognate receptors. Second, the short conserved cytoplasmic has a dual function as a membrane-associated scaffold protein and domains of syndecans interact with proteins that participate in as a transcriptional co-regulator. intracellular signalling and transcriptional control. The best In the current study, we investigated the role of CASK in the characterised syndecan-interacting proteins are PDZ (PSD-95, discs epidermis. It has been previously shown that syndecan 1 and large, ZO1) domain proteins, such as syntenin (Grootjans et al., syndecan 4 are involved in epidermal differentiation and repair. 1997) and the Ca2+/calmodulin-associated Ser/Thr kinase (CASK) Analysis of corneal wound healing in syndecan 1 knockout (KO) (Cohen et al., 1998; Hsueh et al., 1998; Hsueh and Sheng, 1999), mice revealed a failure to upregulate cell proliferation after which both bind to the conserved EFYA sequence in the C-terminus wounding as well as impairment in corneal cell migration (Stepp of all syndecans. et al., 2002). Overexpression of syndecan 1 in the basal layer of 2706 Journal of Cell Science 121 (16)

the epidermis by using the keratin 14 promoter leads to temporary Results hyperproliferation of newborn mouse epidermis but impairs cell Nuclear localisation of CASK in developing epidermis and hair proliferation in the newly re-epithelialised epidermis after wounding follicles (Ojeh et al., 2008). Impaired wound healing is also observed in We and others have previously shown that changes in the expression mice that overexpress syndecan 1 systemically in several tissues of syndecan 1 influence the regulation of proliferation and under the CMV promoter (Elenius et al., 2004). However, an differentiation of epidermal keratinocytes (Stepp et al., 2002; unresolved question is whether CASK, which binds to the Elenius et al., 2004; Ojeh et al., 2008). To investigate whether cytoplasmic tail of syndecans, influences the regulation of epidermal CASK, a MAGUK-family scaffolding protein that interacts with development and repair. the conserved cytoplasmic C-termini of syndecan-family members, We report here that CASK regulates proliferation and adhesion is involved in keratinocyte proliferation and/or differentiation, we of epidermal keratinocytes. CASK is a nuclear protein in developing investigated CASK localisation and function in the epidermis and hair follicles, basal keratinocytes of actively proliferating epidermis cultured keratinocytes. We first examined CASK expression in of newborn rodents, and cultured human and rodent keratinocytes. developing and adult mouse epidermis by immunofluorescence In these cells, CASK acts to restrict proliferation as knockdown staining (Fig. 1). In skin of newborn mice, CASK was using small interfering RNA (siRNA) that targets CASK shortens predominantly found in the nuclei of cells situated in the basal layer, cell-cycle time, amplifies responses to growth factors such as and also in a few cells within the spinous layers. Double staining keratinocyte growth factor (KGF) and TGFα and causes for the nucleoskeletal protein lamin A and CASK revealed a nuclear hyperproliferation in organotypic keratinocyte cultures. Knockdown colocalisation of both proteins in the basal layer of the epidermis of CASK also accelerates cell adhesion to collagen. These data (Fig. 1A, arrows). Some cells that expressed CASK stained also suggest that CASK is an important regulator of epidermal progenitor positive for the marker protein of proliferative cells Ki67 – mainly cells and participates in the maintenance of epidermal homeostasis. in the basal layer of the epidermis (Fig. 1B, arrows).

Fig. 1. Localisation of CASK expression in skin epidermis and developing hair follicles. (A,B) Immunofluorescence staining of the skin of newborn mice using a rabbit polyclonal anti-CASK antibody (A, B, green channel) and monoclonal JoL-4 antibody against lamin-A (red channel, AI) or a monoclonal Ki67 antibody (red channel, BI). Note the predominant nuclear staining for CASK in the basal layer of epidermis. Arrows indicate colocalisation of CASK and Journal of Cell Science lamin A (AII) and CASK and Ki67 (BII); dashed lines indicate basement-membrane zone. SC, stratum corneum; SG, stratum granulosum; SS, stratum spinosum; SB, stratum basale; HF, hair follicle. (C,D) Immunofluorescence staining of CASK in (C) adult mouse skin and (D) human skin. Dashed lines indicate basement-membrane zone. (E) CASK expression in mouse primary keratinocytes within nuclei, cytoplasm and at cell borders (arrowheads). (F) Immunoblotting of whole- cell protein (W), nuclear (N) and cytosolic (C) extracts from mouse primary keratinocytes probed with antibodies against CASK and α-tubulin. (G-K) Strong nuclear CASK expression is seen during rat hair- follicle development. Skin from whisker pads of E18 rats was immunostained for CASK (green) and the adherens-junction marker protein β-catenin (red). (G) Epithelial placode. (H,I) Hair germs. (J) Hair peg. (K) Developing follicle. (L) Western blot analysis of epidermal protein extracts from wild-type (WT), K14- human syndecan-1-transgenic (Tg) and syndecan-1 null (KO) mice probed with antibodies against CASK and involucrin. (M) CASK staining of skin from WT, Tg and KO mice. Notice the increased expression in the skin of Tg mice. Scale bars, 50 μm. CASK regulates epidermal proliferation 2707

Interestingly, adult mouse epidermis, which is not the expression of involucrin in human primary keratinocytes within hyperproliferative similar to the skin of newborn mice, showed a 3 days of incubation in the differentiation medium (Fig. 2D). Under shift in CASK localisation to cell borders within the basal layer of low-Ca2+ conditions, CASK staining was predominantly nuclear the epidermis (Fig. 1C). A similar pattern of expression was seen (Fig. 2B) whereas under high-Ca2+ conditions, the amount of nuclear in adult human skin, where CASK was detected strongly in the CASK reduced concomitantly with the increase in the plasma- cytoplasm and at the cell-cell junctions within the epidermal basal membrane-associated pool of CASK (Fig. 2B). This was confirmed layer (Fig. 1D). by immunoblotting of fractionated cell extracts. Under low-Ca2+ To further study CASK-localisation patterns in epidermal cells, conditions, approximately equal proportions of CASK were found we examined the localisation of CASK in cultured newborn mouse in cytoplasmic and nuclear extracts, whereas the proportion of primary keratinocytes. CASK was distributed in the nucleus, the nuclear CASK was markedly reduced in cells grown in cytoplasm and at the intercellular junctions of cells (Fig. 1E, differentiation medium (Fig. 2C). These results suggest that CASK arrowheads). To verify this observation, western blot analysis was is predominantly a nuclear protein in undifferentiated keratinocytes. carried out on nuclear and cytosolic protein fractions obtained from However, these experiments do not explain the apparent variation these cells. CASK was present in both the nuclear and cytosolic in CASK expression within the basal layer of epidermis of newborn extracts. The cytoplasmic protein tubulin, used as a control, was mice (Fig. 1) or provide clues to the prominent cell-surface staining present in both the whole-cell protein and cytosolic extracts but in the basal adult human and mouse epidermis. Therefore, we next was absent from the nuclear extracts (Fig. 1F). examined the distribution of CASK in HaCaT cells during the cell CASK expression during rat hair follicle development was cycle. investigated by staining skin sections taken from the whisker pads Serum-starved HaCaT cells in the quiescent state (G0) can be of rat embryos at gestational age 18 (E18). Strong nuclear stimulated to re-enter the cell cycle by adding serum. This results localisation of CASK was found in the epidermis, epithelial placode in progress through the G1 phase (lasting 0-14 hours), S phase (Fig. 1G), hair germ (Fig. 1H), hair pegs (Fig. 1I,J) and the (lasting 14-22 hours with DNA synthesis peaking between 18 and developing follicle (Fig. 1K). The adherens-junction marker β- 22 hours after release from quiescence) and, thereafter, the G2-M catenin was used to delineate the epithelial structures of the phase (Geng and Weinberg, 1993). To confirm this, we performed developing hair follicles (Fig. 1G-J). cell-cycle analysis by using flow cytometry on quiescent cells (G0) To investigate whether CASK expression levels are regulated by and stimulated cells again with serum after 6, 12, 22 and 30 hours syndecan 1, immunoblotting was performed using epidermal (Fig. 3A). To see the distribution of CASK during the various phases extracts from wild-type mice, transgenic mice expressing human of the cell cycle, we serum-starved HaCaT cells for 3 days after syndecan 1 under the keratin 14 (KRT14, hereafter referred to as which we found that >90% cells were in G0, as shown by the FACS K14) promoter (K14 human syndecan 1 transgenic mice) (Ojeh et analysis (Fig. 3A). During G1 phase, following re-stimulation (6 al., 2008) or syndecan-1-KO animals. CASK expression levels were hours and 12 hours following serum re-stimulation), CASK was found to be higher in the epidermal extracts obtained from mice detected at sites of cell-cell junctions and in nuclei. After 22 hours, that express human syndecan 1 (Fig. 1L). These transgenic mice CASK was found predominantly in the nucleus and was lost at cell- overexpress syndecan 1 in the basal epidermis, and the cell junctions. The increased presence of CASK in the nucleus

Journal of Cell Science overexpression of CASK in these mice indicates that expression correlated with the increased expression of the proliferating-cell levels of CASK can be regulated by its cognate cell-surface binding marker protein Ki67 (Fig. 3A), as well as with a higher number of partners. The levels of the terminal differentiation marker involucrin, cells that had incorporated 5-bromo-2-deoxyuridine (BrdU) – an a constituent of cornified envelope within the epidermal cell, indication that cells had been dividing during S phase (Fig. 4A). remained unchanged in the epidermis of these animals. By 30 hours post serum addition, a few mitotic cells were clearly Immunofluorescence studies also revealed an increase in CASK visible (Fig. 3A, arrowheads). This was also confirmed by staining staining, predominantly in the basal layer but also in the suprabasal for phosphorylated histone H3 (H3-P) of mitotic cells. At 30 hours layers of the K14 human syndecan 1 transgenic mouse epidermis mitotic cells were seen that often showed prominent CASK staining compared with those of the other animals (Fig. 1M). In addition, in the surrounding cytoplasm, but no colocalisation with H3-P- increased dermal staining of CASK was seen in the syndecan-1- positive chromatin was observed (Fig. 4B). To analyse CASK levels overexpressing mice, which is in line with the previously reported through the cell cycle, western blotting was performed on whole- increased of dermal cellularity in these mice (Ojeh et al., 2008). cell protein extracts. An increase in the level of CASK expression However, loss of syndecan 1 did not markedly compromise was seen 6 hours after re-stimulation. This was followed by a localisation and expression of CASK (Fig. 1M). decrease in CASK levels at 22 and 30 hours (Fig. 3B) which coincided with the loss of membrane-associated CASK (Fig. 3A Differentiation status and cell-cycle status regulate the nuclear and Fig. 4A) and with the appearance of, first, cyclin A at 24 hours localisation of CASK and, second, cyclin B by 30 hours (Fig. 3C). The predominantly nuclear localisation of CASK in basal To investigate CASK localisation at specific phases of the cell keratinocytes and during hair-follicle development prompted us to cycle, we synchronised HaCaT cells by using drugs that block the investigate more closely whether the localisation patterns of CASK progression of the cell cycle. Aphidicolin treatment inhibits DNA change during differentiation. We used the human keratinocyte cell replication and results in a cell population that is synchronised in line HaCaT, which has retained capacity for differentiation in vitro. G1-S phase, although with the side effect of DNA damage (Kurose A switch from low-Ca2+ (0.05-0.1 mM) to high-Ca2+ culture et al., 2006). The majority of Aphidicolin-treated HaCaT cells were conditions (1.2 mM) has been shown to induce differentiation of at the G1-S transition – without any mitotic cells – and all cells these cells (Hennings et al., 1980). This was confirmed by epithelial that stained positive for the S-phase marker proliferating-cell cell-cell adhesion as shown by the accumulation of E-cadherin at nuclear antigen (PCNA) showed nuclear localisation of CASK (Fig. cell borders within 24 hours after the Ca2+ switch (Fig. 2A), and 4C). Nocodazole treatment arrests cells at the mitotic spindle 2708 Journal of Cell Science 121 (16)

Fig. 2. CASK localisation is altered during keratinocyte differentiation. (A) HaCaT cells were grown in low-Ca2+ (0.06 mM) DMEM or in high-Ca2+ (1.6 mM) DMEM for 1 day (D1) and double stained with CASK (green channel) and a monoclonal antibody against E-cadherin (red channel) as indicated. (B) HaCaT cells were grown in low-Ca2+ or high-Ca2+ DMEM for 6 hours (6H), 1 (D1), 3 (D3) or 5 (D5) days and stained using a rabbit polyclonal antibody against CASK. Notice the gradual disappearance of nuclear CASK in cells grown in high-Ca2+ DMEM over time. (C) Nuclear (N) and cytosolic (C) extracts from HaCaT cells grown under low or high-Ca2+ conditions for 1 day (D1) and 5 days (D5), and immunoblotted using antibodies against CASK and α-tubulin. (D) Human primary keratinocytes were grown in low-Ca2+ defined keratinocyte serum-free (SFM) medium or in high-Ca2+ SFM medium for 1 day (D1) or 3 days (D3) and stained with rabbit polyclonal antibody against CASK (green channel) and a monoclonal involucrin antibody (red channel). Scale bars, 50 μm. Journal of Cell Science checkpoint and resulted in >70% of HaCaT cells being in G2 and/or (~10%) appearing in the soluble fraction after ammonium sulfate M phase (Fig. 4D). In nocodazole-treated cells, strong CASK extraction (Fig. 5A). staining was frequently observed in the cytoplasm surrounding H3- To investigate changes in solubility properties of nuclear CASK, P-stained chromatin. These data are in accordance with the results HaCaT cells were also subjected to nuclear-matrix extraction in situ obtained following serum re-stimulation of quiescent cells. and examined by immunofluorescence microscopy using antibodies against CASK and lamin A/C. Fig. 5B shows that, without nuclear Subnuclear distribution of CASK reveals nucleoplasmic and extraction (stage I), CASK is found in both cell nuclei and nucleoskeletal populations cytoplasm. After CSK 0.1% Triton X-100 (stage II) and CSK 0.5% Since CASK was localised to the nuclei of cells especially under Triton X-100 (stage III) extraction, all cells retained CASK strongly low-Ca2+ conditions, we decided to further investigate its subnuclear in the nucleus, whereas cytoplasmic CASK was greatly reduced. distribution by examining the solubility properties of CASK in After DNase I treatment (stage IV) and ammonium sulfate extraction HaCaT cells grown under low-Ca2+ conditions using a nuclear- (stage V), CASK was retained in nucleoli (confirmed by anti- matrix-extraction protocol (Dyer et al., 1997; Pekovic et al., 2007) nucleolin antibody staining; Fig. 5D), and in the nuclear lamina. (see Materials and Methods for further details). Fig. 5A shows that, Lamin A/C strongly stained the nuclear lamina at all stages of in these cells, CASK was present in all the insoluble pellets after extraction. These results were also similar when a GFP-CASK- each step of extraction in decreasing amounts. Approximately 40% transfected HaCaT cells were subjected to the same in situ nuclear- of CASK first appeared in the soluble fraction after CSK-0.1% matrix extraction (Fig. 5C,D). These data provide compelling Triton X-100 treatment (S2) and ~10% of CASK was present in evidence that a population of CASK forms part of the nucleoskeleton the soluble fraction after CSK-0.5% Triton X-100 treatment (S3). in HaCaT keratinocytes. CASK was absent in subsequent soluble fractions after treatment with DNAse I (S4) and ammonium sulfate (S5) resulting in its Knockdown of CASK accelerates cell proliferation and retention in the insoluble pellets P4 and P5. As a control, we used increases growth-factor responses lamin A/C because its solubility properties following sequential To investigate the functional role of CASK on cell proliferation, extraction has been described in detail (Dyer et al., 1997; Pekovic we used siRNA to knockdown CASK in HaCaT cells. Cells were et al., 2007). As expected, lamin A/C was present in all insoluble transfected with siRNA targeting CASK or with scrambled siRNA pellets after each step of extraction, with only a small amount as a control. Western blot analysis revealed an ~80-90% CASK regulates epidermal proliferation 2709

Fig. 3. CASK expression in HaCaT keratinocytes upon cell- cycle re-entry. (A) Serum-starved HaCaT cells were re-stimulated to enter the cell cycle by adding serum for 0, 6, 12, 22 or 30 hours, Journal of Cell Science and cell-cycle analysis was performed at each time point to verify the stage of the cell cycle (graphs on right). Cells were co- stained with a rabbit polyclonal anti-CASK antibody (green channel) and a mouse monoclonal anti-Ki67 antibody (red channel) at each time point. Mitotic cells were seen at 30 hours (arrowheads). (B) Cells were harvested at the indicated times after each serum stimulation and immunoblotting was performed using antibodies against CASK (112 kDa) and α-tubulin (50 kDa). (C) Expression of cyclin A (60 kDa) and cyclin B (62 kDa) at indicated times after re- stimulation. Scale bars, 50 μm.

downregulation of CASK 96 hours post transfection in CASK- the observed differences in cell number further, we analysed serum- siRNA-transfected cells compared with control-siRNA-transfected starved and re-stimulated cells at 0 hours (time of siRNA cells (Fig. 6A). We next measured cell proliferation in an MTT transfection) or 24, 48, 72 and 96 hours following siRNA assay using siRNA-transfected HaCaT cells. Preliminary transfection. We found that knockdown of CASK leads to a experiments on non-synchronised cells grown with 10% FBS significant increase in the growth rate of cells (Fig. 6C). showed a statistically significant increase in HaCaT cell numbers Furthermore, monitoring these cells by phase-contrast microscopy 72 hours after transfection. The same result was obtained with two confirmed that CASK-siRNA-transfected cultures had increasingly different CASK-siRNA oligonucleotides (Fig. 6B). To investigate more cells at 24, 48, 72 and 96 hours after transfection than those 2710 Journal of Cell Science 121 (16)

of control-siRNA-transfected cultures (Fig. 6D). Interestingly, the phosphorylated forms of the protein. Whereas the level of non- phosphorylated form of the retinoblastoma protein Rb, which phosphorylated Rb remained unchanged, expression of the controls cell cycle progression, was increased in CASK-siRNA- phosphorylated forms was increased in CASK-siRNA-transfected transfected cells at 72 hours post transfection (Fig. 6E). This was cells (Fig. 6E). confirmed by immunoblotting using an antibody against all Rb To confirm these findings on cell proliferation, we performed isoforms, which detects both phosphorylated and non- cell-cycle analysis using flow cytometry on siRNA-transfected

Fig. 4. CASK expression in HaCaT cells following release from cell cycle blockers. (A,B) Serum- starved HaCaT cells were re-stimulated to enter the cell cycle by adding serum. (A) Cells were treated with 10 μM BRDU for 45 minutes prior to staining with rabbit polyclonal anti- CASK antibody (green Journal of Cell Science channel) and mouse monoclonal anti-BrdU antibody (red channel) at 12 and 22 hours. (B) Cells were co-stained with anti- CASK antibody (ZYMED; green channel) and mouse monoclonal anti-H3-P antibody for mitotic cells (red channel) at 22 and 30 hours. (C) Cells synchronised with Aphidicolin were stained with a rabbit polyclonal anti-CASK antibody (ZYMED; green channel) and mouse monoclonal anti-PCNA antibody (red channel). Corresponding cell-cycle analysis is shown in the graph at right. (D) Cells were synchronised with nocodazole prior to staining with a rabbit polyclonal anti-CASK antibody (ZYMED; green channel) and a mouse monoclonal anti-H3 antibody (red channel). Corresponding cell cycle analysis is shown at right. Scale bars, 50 μm. CASK regulates epidermal proliferation 2711

cells. Serum-starved, quiescent cells were transfected with siRNA, CASK-siRNA-transfected cells, which was found to be negligible re-stimulated by addition of 10% FBS and harvested at 0 and 60 in both (Fig. 7C), confirming that the differences in cell number hours post transfection for FACS analysis. Flow cytometric DNA counted in MTT proliferation assays were not affected by apoptosis. analysis of the cell cycle revealed that CASK knockdown resulted The expression levels of subunits β1, β4 and α6 of basal integrins in a large increases of cells in S phase (36.3%) and G2-M phases were also investigated by flow cytometry and found to be very (21.3%) and a concomitant reduction of CASK-siRNA-transfected similar (Fig. 7D) suggesting that the difference in proliferation is cells in G1 phase (44.8%) compared with control-siRNA- not due to changes in basal integrin expression. transfected cell populations, of which ~22.8% were in S phase, To evaluate the mitogenic response of CASK-knockdown cells 13.8% in G2-M phases and 64.5% in G1 phase (Fig. 6F). Likewise, to growth factors, siRNA-transfected cells were serum starved cell-cycle analysis of siRNA-ransfected non-synchronised cells and left untreated, or were subsequently treated with 10 ng/ml indicated that there were more CASK-siRNA-transfected than KGF or 10 ng/ml TGFα for 24 hours. Both KGF and TGFα control-siRNA-transfected cells in S phase 60 hours after elicited a mitogenic effect on both control-siRNA- and CASK- transfection (Fig. 6G). These results were also confirmed using siRNA-transfected cells. Notably, cell proliferation was enhanced cell-cycle analysis based on 5-ethynyl-2Ј-deoxyuridine (EdU) in CASK-knockdown cells compared with control-siRNA- incorporation (Fig. 7A). Finally, by extrapolating data from the transfected cells after a 24-hour treatment with growth factors MTT proliferation graph, we calculated that cell-cycle time is (Fig. 7B). Taken together, these data indicate that CASK restricts approximately 3 hours faster in CASK-knockdown cells. We also keratinocyte proliferation and reduces responses to growth investigated the level of apoptosis in both control-siRNA- and factors. Journal of Cell Science

Fig. 5. Biochemical fractionation and in situ nuclear-matrix extraction of HaCaT cells. (A) Western blotting. HaCaT cells grown in low-Ca2+ medium were subjected to a sequential extraction by CSK 0.1% Triton X-100 (C/T, 0.1%), CSK 0.5% Triton X-100 (C/T, 0.5%), chromatin digestion by DNase I (DNAse) and a final extraction by 0.25 M ammonium sulfate. Whole-cell protein extracts (P1) were prepared and, following each step, insoluble fractions P2, P3, P4 and P5 were pelleted, and soluble fractions S2, S3, S4 and S5 were retained for immunoblotting with monoclonal anti-CASK antibody and JoL2 antibody to detect lamin A/C. (B) HaCaT cells were grown on glass coverslips in low-Ca2+ medium and then subjected to in situ nuclear-matrix extraction using five sequential treatments with CSK buffer only (no extraction; stage I), CSK 0.1% Triton X-100 (stage II), CSK 0.5% Triton X-100 (stage III), DNase I digestion (stage IV) and 0.25 M ammonium sulfate extraction (stage V), and processed for indirect immunofluorescence microscopy using antibodies against CASK (green channel) and lamin A/C (red channel). (C,D) HaCaT cells grown in low-Ca2+ DMEM were transfected with a full-length CASK-GFP construct and subjected to in situ nuclear-matrix extraction using sequential treatments up to the indicated stages. GFP-CASK is shown in the green channel and lamin A/C in the red channel (C). Stages IV and V extracted cell showing colocalisation (arrowheads) of the nucleolus marker, nucleolin (red channel) with GFP-CASK (green channel) (D). Scale bars, 50 μm (B), 20 μm (C,D). 2712 Journal of Cell Science 121 (16)

Fig. 6. RNAi of CASK leads to increased cell proliferation in HaCaT keratinocyte monolayer cultures. (A) Efficiency of CASK RNAi. HaCaT cells were transfected with control siRNA or three different siRNAs targeting CASK for 96 hours, and then harvested for immunoblotting and probed with antibodies against CASK and α-tubulin. (B) Growth of non-synchronised HaCaT cells transfected with control siRNA, CASK siRNA1 or CASK siRNA2. Cells were subjected to the MTT proliferation assay at 72 hours post siRNA transfection. Error bars indicate +s.d. (n=3). (C) Growth curves of control-siRNA- and CASK-siRNA- transfected HaCaT cells. Serum- starved and re-stimulated cells were transfected with control siRNA or CASK siRNA, and cells were subjected to MTT proliferation assay at 0, 24, 48, 72 and 96 hours post siRNA transfection. Error bars indicate ±s.d., (n=3). (D) Phase-contrast micrographs of control-siRNA and CASK-siRNA-transfected HaCaT cells at 24, 48, 72 or 96 hours post siRNA transfection. Scale bars, 100 μm. (E) Protein extracts from control- siRNA and CASK-siRNA-transfected HaCaT cells were immunoblotting and probed with antibodies against CASK, phosphorylated Rb (pRb), total Rb and actin. (F) FACS cell-cycle analysis of DNA content of control-siRNA and CASK-siRNA-transfected synchronised HaCaT cells at 0 and 60 hours post siRNA transfection. Percentages of cells in G1 and S

Journal of Cell Science phases, and at G2 or M phase are indicated. (G) FACS cell-cycle analysis of DNA content of control- siRNA and CASK-siRNA-transfected non-synchronised HaCaT cells at 60 hours post siRNA transfection. Percentages in G1 and S phases, and at G2 or M phase are indicated.

CASK knockdown increases keratinocyte proliferation in 3D keratin 6 (KRT6, hereafter referred to as K6), and basal markers raft cultures for the epidermis, such as keratins 5 (KRT5, hereafter referred to To confirm the effect of CASK knockdown on cell proliferation as K5) and K14, have been shown to be similar in raft cultures as and to evaluate any effects on differentiation of HaCaT they are in wound healing and hyperproliferative skin diseases keratinocytes, we seeded CASK-siRNA-transfected and control- (Waseem et al., 1999; Ojeh et al., 2001). As therefore expected, siRNA-transfected cells on raft cultures and grew them at air-liquid K6, K5 and K14 were similarly expressed in all epidermal layers interface for 7 days to allow stratification of keratinocytes. in both raft cultures (Fig. 8C). Moreover, expression levels of Immunoblotting of protein extracts from the epidermal component keratinocyte differentiation markers involucrin and filaggrin of the raft cultures demonstrated that CASK protein levels remained remained unchanged after CASK ablation (Fig. 8A,C). Even though downregulated for the entire culture period (Fig. 8A), which was the most intense involucrin staining appeared in the upper layers also confirmed by CASK-immunofluorescence staining (Fig. 8B). in the CASK siRNA raft epidermis compared with control rafts Levels of the S-phase marker PCNA (Fig. 6A), thickness of the raft (Fig. 8C), it can be concluded that downregulation of CASK does epidermis (Fig. 8B), numbers of Ki67 positive nuclei and H3-P- not prevent keratinocyte differentiation. The basal integrin subunits positive mitotic cells (Fig. 8C) were increased. Interestingly, we α6, β1 and β4 were present in the basal layer of both rafts cultures, also observed a marked increase in the protein levels of Myc (Fig. In the control rafts the integrin staining was mostly confined to 8A), a known regulator of keratinocyte proliferation (Arnold and basal cell membranes, whereas in CASK-siRNA rafts integrin Watt, 2001; Waikel et al., 2001; Zanet et al., 2005). The expression expression was detected also in apical and lateral surfaces of the of keratins that serve as markers for hyperproliferation, such as basal cell and, to a lesser extent, also in suprabasal cells (Fig. 8C). CASK regulates epidermal proliferation 2713

Fig. 7. Proliferation, growth factor responses, apoptosis and basal integrin expression in siRNA-transfected synchronised HaCaT cells at 60 hours post siRNA transfection. (A) Cell-cycle analysis of DNA content in cells pulse labelled with EdU. Percentages of cells in G1 and S phases, and at G2 or M phase are indicated. (B) Control-siRNA- and CASK-siRNA-transfected cells were serum starved and left untreated, or were treated with 10 ng/ml KGF or 10 ng/ml TGFα for 24 hours. Proliferation was substantially enhanced in CASK-siRNA-transfected cells treated with KGF or TGFα for 24 hours compared with control-siRNA-transfected cells treated with the same growth factors. (C) Percentage of apoptotic cells. Cells were harvested and subjected to staining for apoptosis using an apoptosis detection kit (BD Biosciences). Positive-control cells provided in the kit show a high percentage of apoptosis compared with negligible levels seen in either the control or CASK siRNA transfected cells. (D) Basal integrin expression. Cells were harvested, stained for α6, β1 or β4 integrin and subjected to analysis by flow cytometry. Journal of Cell Science CASK regulates keratinocyte cell adhesion development and differentiation, and by knocking down CASK in Finally, we investigated whether changes in the expression or cultured keratinocytes. Our results indicate that CASK restricts subcellular localisation of CASK take place during wound healing, keratinocyte proliferation. This conclusion is supported by where strictly controlled responses to growth factors lead to spatio- experiments that compare growth of HaCaT keratinocytes after temporally controlled proliferation and the migration of CASK-siRNA- or control-siRNA transfections, by measuring keratinocytes. We used immunofluorescence staining to investigate growth-factor (TGFα and KGF) responses in CASK-knockdown CASK expression levels in three locations of wounded epidermis cells and by establishing organotypic 3D raft cultures from CASK- indicated in the hematoxylin and eosin (H&E) staining of the edge knockdown cells. In all these experimental models, knockdown of of a punch biopsy wound (Fig. 9A). The epidermal sheet that CASK resulted in increased keratinocyte proliferation as measured migrated under the blood clot at the wound site had only a low, by cell numbers, cell-cycle-dependent analysis of DNA content and mostly cytoplasmic expression of CASK, whereas strong expression expression of cell proliferation markers. Notably, we did not was seen in the basal epidermis at the wound edge and in the observe any changes in the expression of epidermal differentiation epidermis further away from the wound site. The downregulation markers after CASK knockdown. of CASK expression in the migrating epidermal sheet suggested Previous studies have identified three major functions for that CASK is involved in keratinocyte migration or adhesion. mammalian CASK. First, CASK participates in the targeting of Attachment of suspended cells to collagen-coated surfaces was protein complexes to synapses, as inferred from its interactions with accelerated in the absence of CASK (Fig. 9B), which was also syndecan 2, and SynCAM (reviewed by Hsueh, 2006). reflected by more-pronounced spreading and the formation of focal Second, nuclear targeting of CASK and interactions with TBR1 adhesion sites of the attached cells as shown by vinculin staining (Hsueh et al., 2000) and CINAP (Wang et al., 2004) modulate (Fig. 9C). Thus, in addition to proliferation, CASK is involved in transcriptional regulation. Third, CASK contributes to epithelial the regulation of keratinocyte cell-matrix adhesion during wound polarity by interacting with syndecan 1, junctional adhesion healing. molecule (JAM) (Martinez-Estrada et al., 2001) and (Lehtonen et al., 2005). Intriguingly, both gene targeting (Atasoy Discussion et al., 2007) and insertional mutagenesis of the CASK locus (Laverty The aim of this investigation was to gain insights in the role of and Wilson, 1998) lead to cleft palate, implying a role for CASK CASK in the epidermis by studying CASK localisation during skin in embryonic development; but, owing to early neonatal death of 2714 Journal of Cell Science 121 (16)

CASK-knockdown mice (Atasoy et al., 2007), the loss-of-function Nuclear localisation of CASK in developing epidermis and phenotype of CASK is not yet fully characterised. Our results add cultured keratinocytes a novel main function to CASK as a modulator of epidermal Here, we have shown that, during rat embryonic hair follicle proliferation. Notably, mice that lack CASK display increased development and in skin of newborn mice, CASK is targeted to the apoptosis in the thalamus (Atasoy et al., 2007), possibly indicating nucleus. In adult skin, however, a shift occurs in the localisation of a different role for CASK in cell survival within the epidermis and CASK – to the cytoplasm and cell borders of cells in the epidermis. brain. Interestingly, this shift from nuclear to cytoplasmic CASK also takes Journal of Cell Science

Fig. 8. Knockdown of CASK leads to epidermal hyperproliferation in organotypic cultures. (A) Protein extracts from epidermal tissue were isolated from organotypic cultures, and prepared with control-siRNA- and CASK-siRNA-transfected HaCaT cells after 7 days at the air-liquid interface. Immunoblotting was performed using antibodies against CASK (112 kDa), PCNA (30 kDa), Myc (64 kDa), involucrin (120 kDa) and filaggrin (40 kDa), actin (42 kDa) was used as an internal control. Notice the reduced CASK protein level in CASK-siRNA-transfected epidermal extracts concomitant with the upregulation of PCNA and Myc. (B) H&E histology and immunofluorescence staining for CASK in control and CASK-knockdown HaCaT raft cultures. H&E histology shows a thicker epidermal morphology in the CASK-knockdown 3D cultures, consisting of several layers compared with the control-siRNA-transfected rafts. (C) Expression of keratins, proliferation and differentiation markers, and basal integrins in siRNA-transfected raft cultures. Notice the presence of Ki67-positive cells and H3-P-positive mitotic cells in the CASK-knockdown skin model, in both basal and suprabasal layers of the epidermis. Involucrin and filaggrin are expressed at similar levels in the suprabasal and differentiating layers of the epidermis. CASK knockdown skin model shows integrin expression also in lateral and apical membranes of the basal cells. All protein staining are shown in green and DAPI nuclear staining is shown in blue. Scale bars, 50 μm. CASK regulates epidermal proliferation 2715

place in neuronal development. Prominent nuclear expression of Lee et al., 2002) and also exhibits changes in its localisation CASK is seen in the embryonic cerebral cortex, whereas in brains from nucleus to cell membrane upon epidermal keratinocyte of six-week-old rats CASK could not be detected in nuclei of cerebral differentiation (Roberts et al., 2007). Moreover, studies on muscle cortex neurons (Hsueh et al., 2000). Taken together with our results, differentiation using the C2C12 myogenic cell line have also one can conclude that nuclear localisation of CASK is not required revealed that CASK is predominantly expressed in the nucleus of for the maintenance of mature neurons or for completion of terminal undifferentiated myoblasts, with a shift in location to the cytoplasm differentiation of the epidermis. However, in neurons the target in differentiated myotubes (Gardner et al., 2006). Our results indicate of the CASK-TBR1 complex include regulators of neuronal that CASK targeting to the plasma membrane or the nuclus can be differentiation and maturation, such as reelin (Hsueh et al., 2000). regulated by the cell-cycle and differentiation statuses of the cells. We have also shown using in vitro differentiation assay that there The subnuclear fractionation of CASK demonstrated that there is a shift in CASK localisation from the nucleus to the cytoplasm are distinct cytoplasmic, nucleoplasmic and nucleoskeletal pools of upon keratinocyte differentiation. Interestingly, human discs large CASK. Moreover, extraction of nucleoplasmic CASK also revealed homolog 1 (DLG1) protein, which also belongs to the MAGUK its presence in the nucleoli. ZO1, another MAGUK protein, is also , is known to interact with CASK (Nix et al., 2000; found in the nuclei of fibroblasts under proliferating conditions and is targeted to the nucleoli under pro-migratory conditions (Benezra et al., 2007). Another PDZ protein, syntenin 2, is also targeted to nucleus where it interacts with intranuclear phosphatidylinositol 4,5- bisphosphate pool in nucleoli and nuclear speckles (Mortier et al., 2005). Remarkably, RNA interference (RNAi) of syntenin 2 increases apoptosis and decreases cell divisions and BrdU incorporation in MCF-7 cells (Mortier et al., 2005). Thus, CASK and syntenin 2, which both interact with syndecans at the cell membrane, have opposite effect on epithelial cell proliferation.

CASK-mediated control of keratinocyte proliferation might involve Myc and Rb Increased proliferation of CASK-knockdown keratinocytes was linked with the upregulation of Myc. Several lines of evidence indicate that Myc is one of the key regulators of keratinocyte proliferation and, furthermore, promotes exit from epidermal stem- cell compartment (Arnold and Watt, 2001; Waikel et al., 2001; Zanet et al., 2005). A balanced level of Myc expression seems to be crucial for stem-cell maintenance. Transgene-driven overexpression of Myc results in hyperproliferation and depletion of the stem-cell

Journal of Cell Science compartment (Arnold and Watt, 2001; Waikel et al., 2001), whereas conditional knockout of Myc in the basal epidermis results in thin, fragile skin with a loss of cell proliferation (Zanet et al., 2005). Recently, a Myc-target gene that promotes keratinocyte proliferation, the RNA methyltransferase Misu, has been identified (Frye and Watt, 2006), providing a possible mechanism for Myc- induced hyperproliferation. In our study, we also have shown that loss of CASK in exponentially growing HaCaT cells is associated with an increase in the phosphorylated form of the retinoblastoma protein Rb. Rb regulates cell-cycle progression at the G1-S transition by negatively regulating the transcription factor E2F in a phosphorylation-dependent manner (Chellappan et al., 1991). Phosphorylation of Rb has been previously implicated in the regulation of keratinocyte proliferation (Munger et al., 1992).

Role of CASK during wound healing Wound repair is mediated in largely by growth factors that are released Fig. 9. CASK expression during skin-wound healing. (A) H&E histology of at the wound site and that have a fundamental role in wound healing, day 3 mouse back-skin full-thickness wound (top panel). Boxed areas that were including cellular migration, proliferation and extracellular matrix investigated for CASK expression are shown in magnification (bottom panel). (ECM) production (reviewed by Werner et al., 2007). Consequently, Wound site (left); wound edge (middle); away from wound (right). Frozen sections of wounded skin were immunostained using a rabbit polyclonal anti- the role of CASK in regulating keratinocyte proliferation and growth- CASK antibody. Notice the reduced staining for CASK in keratinocytes at the factor responses is likely to be important in wound healing responses. wound site (WS). (B,C) Knockdown of CASK results in accelerated cell Interestingly, we found changes in CASK expression at different sites adhesion and spreading. HaCaT cells were transfected with control-siRNA or of the epidermis during skin wound healing. CASK expression was CASK-siRNA oligonucleotides for 96 hours and seeded onto collagen-coated downregulated in the migrating epidermal sheet of mouse skin wounds substrates. After 30 minutes, 1 or 2 hours, attached cells were quantified by an adhesion assay (B) or stained for the focal adhesion marker vinculin (green) but was upregulated at the wound margin, suggesting that CASK is (C). DAPI nuclear staining is shown in blue. Scale bars, 50 μm (A), 20 μm (C). important for epithelial migration or adhesion. In another study, it 2716 Journal of Cell Science 121 (16)

has been shown that during mouse cutaneous repair, the leading cells Cell-cycle analysis was performed on a FACSCalibur flow cytometer (Becton Dickinson) and data were collected as DNA histograms from 10,000 single-cell events in the epidermal sheet migrating below the scab were positive for before analysis with the FlowJo software using the Dean/Jett/Fox model. In some syndecan 1 but the more intense signal was found adjacent to the experiments, cells were pulse labelled with EdU (5-ethynyl-2Ј-deoxyuridine), a wound margin at sites of rapid cell proliferation (Elenius et al., 1991). thymidine analog that is incorporated into DNA during S phase, similarly to BrdU. The PDZ domain of CASK is known to bind to the EFYA motif 10 μM EdU was added to DMEM and incubated with cells for 45 minutes before harvesting. Staining was performed according to the protocol of the Click-iTTM EdU within the C-terminus tail of syndecan (Cohen et al., 1998). Thus, it Alexa-Fluor®-488 cell-proliferation assay kit (Invitrogen). For growth curves, the appears that changes in CASK localisation can be modulated by proliferation of siRNA-transfected HaCaT cells was measured using the colorimetric syndecan expression during wound healing. This was also found to MTT assay (Promega) at 0 hour (time of siRNA transfection), and 24, 48, 72 and 96 hours post siRNA transfection. Absorbance was measured at 570 nm using a be the case in our transgenic mice studies where over expression of spectrophotometer plate reader (Antholucy 1). The effects of growth factors were syndecan 1 in mouse epidermis resulted in increased CASK studied in HaCaT cells transfected with siRNA for 72 hours and then starved for 24 expression. In support of this, subcellular distribution of CASK has hours in DMEM without serum. KGF and TGFα (Peprotech) were used at a been shown to be regulated by the balance of its various binding concentration of 10 ng/ml for a period of 24 hours. partners in different subcellular compartments. When syndecan 3 is Organotypic 3D skin cultures coexpressed with TBR1 and CASK in COS cells, CASK colocalises Organotypic cultures were prepared as described previously with modifications (Smola with syndecan 3 in the cytoplasm, whereas accumulation of CASK et al., 1998). Gels were prepared without fibroblasts or with 5ϫ105 fibroblasts per gel. Following an overnight incubation, 3ϫ105 siRNA-transfected HaCaT cells were is seen in the nucleus of cultured hippocampal neurons when co- seeded on top of each gel and allowed to attach for 24 hours. The cultures were then transfected with Tbr1 (Hsueh et al., 2000). raised to the air-liquid interface for 7 days. Co-cultures with fibroblasts were embedded We also observed changes in the rate of adhesion and spreading in TissueTek OCT compound, snap-frozen and processed for routine haematoxylin of CASK siRNA-transfected HaCaT cells compared with control- and eosin (H&E) histology and immunofluorescence staining. Mono-cultures (gels with HaCaT cells only) were harvested for protein extraction and subjected to western- siRNA-transfected cells when plated on collagen substrates using blot analysis. adhesion assays which measure both cell attachment and spreading. Knockdown of CASK led to faster adhesion and spreading. Cell adhesion assay Similarly, keratinocytes obtained from syndecan-1-null mice were For the measurement of cell attachment and cell spreading, HaCaT cells transfected with siRNA for 96 hours were trypsinised and plated on collagen-coated 24-well more proliferative and more adherent and migrated more slowly plates at densities of 1ϫ105 cells per well. At 30 minutes, 1 hour and 2 hours post than their wild-type counterparts when seeded onto various ECM seeding, cells were washed three times with PBS (containing 1 mM CaCl2 and 1 mM substrates, including collagen type I (Stepp et al., 2007). This was MgCl2) to remove unbound cells. Attached cells were then processed for immunofluorescence microscopy or subjected to the MTT assay. largely mediated by α6β4-integrin signalling and TGFβ1 signalling (Stepp et al., 2007). To conclude, we have found a novel role for Animal immunofluorescence studies CASK as a regulator of epidermal proliferation and adhesion that Normal skin tissue from 2-day-old newborn syndecan-1-transgenic mice (Ojeh et al., restricts growth-factor responses in keratinocytes. Thus, CASK is 2008), syndecan-1-knockout (KO) mice (Stepp et al., 2002) and wild-type C57 Black 6 mice was embedded in TissueTek OCT and snap-frozen. Full-thickness skin wounds likely to be a participant in both physiological and pathological on the backs of adult C57 Black 6 mice were created as described previously (Ojeh changes such as wound healing or psoriasis. et al., 2008). At day 3 post wounding, normal and wounded skin tissue was embedded in TissueTek OCT and snap-frozen. Cryosections were cut 7-μm thick, and processed for H&E histology and immunofluorescence staining. Mystacial pad regions from Materials and Methods E18 gestational age PVG hooded rats were removed, embedded and frozen in

Journal of Cell Science Cell culture TissueTek OCT for immunohistochemistry. Human dermal fibroblasts were isolated from skin and cultured as described previously (Ojeh et al., 2001). Cells were used between passages three and six. Mouse Antibodies and Immunostaining primary keratinocytes were isolated from skin of newborn C57 Black 6 mice and Immunofluorescence was performed according to standard laboratory procedures cultured as described previously (Ojeh et al., 2008). The human keratinocyte HaCaT (Long et al., 2006). The primary antibodies used and their dilutions are listed in cell line was cultured in Dulbecco’s modified Eagle’s medium (DMEM) supplemented supplementary material (Table S1). The three different anti-CASK antibodies were with 10% foetal bovine serum (FBS), 2 mM glutamine, 100 U/ml penicillin and 100 used as follows: Antibody from Zymed for human cells and tissues, antibody from μg/ml streptomycin. For Ca2+-differentiation experiments, HaCaT cells were grown Abcam for mouse and rat tissues and primary mouse keratinocytes and the Chemicon in high-Ca2+ medium [Ca2+-free DMEM supplemented with 1.6 mM Ca2+ and 10% antibody for immunoblotting. Secondary antibodies were Alexa-Fluor-488 green Chelex-treated (Ca2+ free) FBS] or low-Ca2+ medium (Ca2+-free DMEM supplemented (Invitrogen) conjugated anti-rabbit secondary antibody (1:800), Alexa-Fluor-488 green with 0.06 mM Ca2+ and 10% chelex-treated FBS). To induce quiescence through and 594 red (Invitrogen) conjugated anti-mouse secondary antibodies (1:800). DNA serum starvation, HaCaT cells were switched to DMEM without serum for 3 days. of cells was counterstained with DAPI (Sigma). Confocal images of the samples Induction of cell-cycle re-entry was achieved by the addition of 10% FBS to the were taken with Zeiss LSM 510 Zeiss or BioRad Radiance 2000 laser scanning culture medium. To induce a cell-cycle block at G1-S transition, cells were cultured confocal microscopes. Composite images were assembled using Adobe Photoshop in FBS-depleted DMEM for 24 hours followed by treatment with 2 ng/ml Aphidicolin 7.0 (Adobe® Systems) and LSM510 image browser software (Carl Zeiss). (Sigma) in DMEM containing 10% FBS for 20 hours. Synchronisation at mitotic For nuclear and cytosolic protein extracts, cell pellets were incubated in ice-cold spindle checkpoint was achieved by incubation of the cells with 200 ng/ml nocodazole hypotonic buffer (10 mM Tris pH 7.4, 10 mM KCl, 3 mM MgCl and 0.1% Triton (Sigma) in DMEM containing 10% FBS for 20 hours. 2 X-100) containing protease-inhibitor cocktail and lysed in a glass-glass homogeniser. Transient siRNA transfections The soluble cytoplasmic and the insoluble nuclear fractions were separated by Double-stranded siRNA oligonucleotides targeting CASK and a scrambled negative centrifugation at 4000 g for 5 minutes at 4°C. The nuclei-containing pellets were control were purchased from Ambion. The most effective siRNA sequence targeting extracted in the same buffer supplemented with 100 U/ml RNase-free DNase I (Sigma) human CASK was 5Ј-GGUAUUGGAAGAAAUUUCATT-3Ј. HaCaT cells were on ice for 10 minutes. Supernatants were precipitated with ice-cold methanol:acetone ϫ 5 ϫ 5 2 (1:1, v/v) at –20°C for 30 minutes, centrifuged at 12,000 g for 10 minutes and washed seeded at 1 10 cells/well in 6-well plates or 2.5 10 cell per 25-cm flasks in ϫ DMEM supplemented with 10% FBS without antibiotics and transfected with 100 in PBS. Samples were boiled in 2 Laemmli sample buffer for 5 minutes. nmol of siRNA oligonucleotides using Oligofectamine reagent (Invitrogen). At 24 Epidermal protein extracts were obtained from newborn syndecan-1-transgenic mice, syndecan-1-KO mice and wild-type C57 Black 6 mice as described previously hours after transfection, medium was replaced with fresh DMEM containing 10% ϫ FBS and antibiotics. Cells were assayed 48-96 hours after transfection, and transfection (Ojeh et al., 2008). Organotypic cultures were lysed in 2 Laemmli sample buffer. efficiency was determined by immunoblotting and imunofluorescence microscopy. Protein gel electrophoresis and immunoblotting were carried out as described previously (Long et al., 2006). FACS analysis, cell proliferation assay and growth-factor treatment siRNA-transfected cells were harvested at various time points, washed in PBS, and In situ nuclear matrix extraction and biochemical fractionation of fixed and permeabilised in 70% ice-cold ethanol. Cells were washed in PBS and keratinocytes incubated in PBS containing 100 μg/ml RNase and 50 μg/ml propidium iodide In situ nuclear-matrix extractions and biochemical fractionation using sequential overnight at 4°C. Prior to FACS analysis, cells were centrifuged and diluted in PBS. treatment with detergents, nucleases and salt in the presence of protease-inhibitor CASK regulates epidermal proliferation 2717

cocktail were performed according to published protocols (Dyer et al., 1997; Pekovic Hsueh, Y. P., Wang, T. F., Yang, F. C. and Sheng, M. (2000). Nuclear translocation and et al., 2007) with minor modifications. Briefly, HaCaT cells grown in low-Ca2+ transcription regulation by the membrane-associated guanylate kinase CASK/LIN-2. medium were washed in CSK buffer (10 mM PIPES pH 6.8, 10 mM KCl, 300 mM Nature 404, 298-302. sucrose, 3 mm MgCl2, 1 mM EGTA pH 8.0) then subjected to extraction in CSK Kurose, A., Tanaka, T., Huang, X., Traganos, F. and Darynkiewicz, Z. (2006). buffer containing 0.1% Triton X-100, CSK buffer containing 0.5% Triton X-100, Synchronization in the cell cycle by inhibitors of DNA replication induces histone H2AX 100 U/ml RNase-free DNase I (for in situ nuclear-matrix extraction) or 500 U/ml phosphorylation: an indication of DNA damage. Cell Prolif. 39, 231-240. RNase-free DNase I (for biochemical fractionation) in digestion buffer and a final Laverty, H. G. and Wilson, J. B. (1998). Murine CASK is disrupted in a sex-linked cleft extraction in extraction buffer (10 mM PIPES pH 8.3, 250 mM ammonium sulfate, palate mouse mutant. Genomics 53, 29-41. Lee, S., Fan, S., Makarova, O., Straight, S. and Margolis, B. (2002). A novel and 300 mM sucrose, 3 mM MgCl2, 1 mM EGTA pH 8.0). conserved protein-protein interaction domain of mammalian Lin-2/CASK binds and recruits SAP97 to the lateral surface of epithelia. Mol. Cell. Biol. 22, 1778-1791. We are grateful to Chris Hutchison (Durham University, UK) for Lehtonen, S., Ryan, J. J., Kudlicka, K., Iino, N., Zhou, H. and Farquhar, M. G. (2005). providing anti-JoL2, anti-JoL4 and anti-IF8 antibodies. We thank Cell junction-associated proteins IQGAP1, MAGI-2, CASK, spectrins, and alpha-actinin Georgia Salpingidou (Durham University, UK) for assistance with are components of the nephrin multiprotein complex. Proc. Natl. Acad. Sci. USA 102, confocal microscopy, Rebecca Stewart (Institute of Human Genetics, 9814-9819. Long, H. A., Boczonadi, V., McInroy, L., Goldberg, M. and Maatta, A. (2006). Bioscience Centre, Newcastle-upon-Tyne, UK), Fred Tholozan Periplakin-dependent re-organisation of keratin cytoskeleton and loss of collective (Durham University, UK) for help with FACS analyses and Ewa migration in keratin-8-downregulated epithelial sheets. J. Cell Sci. 119, 5147-5159. Markiewicz (Durham University, UK) for helpful discussions. This Loo, B. M. and Salmivirta, M. (2002). Heparin/Heparan sulfate domains in binding and work was supported by the Association for International Cancer signaling of fibroblast growth factor 8b. J. Biol. Chem. 277, 32616-32623. Lyon, M., Rushton, G. and Gallagher, J. T. (1997). The interaction of the transforming Research (AICR). growth factor-betas with heparin/heparan sulfate is isoform-specific. J. Biol. Chem. 272, 18000-18006. References Martinez-Estrada, O. M., Villa, A., Breviario, F., Orsenigo, F., Dejana, E. and Bazzoni, Alexander, C. M., Reichsman, F., Hinkes, M. T., Lincecum, J., Becker, K. A., G. (2001). Association of junctional adhesion molecule with calcium/calmodulin- Cumberledge, S. and Bernfield, M. (2000). Syndecan-1 is required for Wnt-1-induced dependent serine protein kinase (CASK/LIN-2) in human epithelial caco-2 cells. J. Biol. mammary tumorigenesis in mice. Nat. Genet. 25, 329-332. Chem. 276, 9291-9296. Arnold, I. and Watt, F. M. (2001). c-Myc activation in transgenic mouse epidermis results Mortier, E., Wuytens, G., Leenaerts, I., Hannes, F., Heung, M. Y., Degeest, G., David, in mobilization of stem cells and differentiation of their progeny. Curr. Biol. 11, 558-568. G. and Zimmermann, P. (2005). Nuclear speckles and nucleoli targeting by PIP2-PDZ Atasoy, D., Schoch, S., Ho, A., Nadasy, K. A., Liu, X., Zhang, W., Mukherjee, K., Nosyreva, domain interactions. EMBO J. 24, 2556-2565. E. D., Fernandez-Chacon, R., Missler, M. et al. (2007). Deletion of CASK in mice is Munger, K., Pietenpol, J. A., Pittelkow, M. R., Holt, J. T. and Moses, H. L. (1992). lethal and impairs synaptic function. Proc. Natl. Acad. Sci. USA 104, 2525-2530. Transforming growth factor beta 1 regulation of c-myc expression, pRB phosphorylation, Baeg, G. H., Lin, X., Khare, N., Baumgartner, S. and Perrimon, N. (2001). Heparan and cell cycle progression in keratinocytes. Cell Growth Differ. 3, 291-298. sulfate proteoglycans are critical for the organization of the extracellular distribution of Nix, S. L., Chishti, A. H., Anderson, J. M. and Walther, Z. (2000). hCASK and hDlg Wingless. Development 128, 87-94. associate in epithelia, and their src homology 3 and guanylate kinase domains participate Benezra, M., Greenberg, R. S. and Masur, S. K. (2007). Localization of ZO-1 in the in both intramolecular and intermolecular interactions. J. Biol. Chem. 275, 41192-41200. nucleolus of corneal fibroblasts. Invest. Ophthalmol. Vis. Sci. 48, 2043-2049. Ojeh, N. O., Frame, J. D. and Navsaria, H. A. (2001). In vitro characterization of an Blanpain, C., Lowry, W. E., Pasolli, H. A. and Fuchs, E. (2006). Canonical notch signaling artificial dermal scaffold. Tissue Eng. 7, 457-472. functions as a commitment switch in the epidermal lineage. Genes Dev. 20, 3022-3035. Ojeh, N., Hiilesvuo, K., Warri, A., Salmivirta, M., Henttinen, T. and Maatta, A. (2008). Chellappan, S. P., Hiebert, S., Mudryj, M., Horowitz, J. M. and Nevins, J. R. (1991). Ectopic expression of syndecan-1 in basal epidermis affects keratinocyte proliferation The E2F transcription factor is a cellular target for the RB protein. Cell 65, 1053-1061. and wound re-epithelialization. J. Invest. Dermatol. 128, 26-34. Cohen, A. R., Woods, D. F., Marfatia, S. M., Walther, Z., Chishti, A. H. and Anderson, Pekovic, V., Harborth, J., Broers, J. L., Ramaekers, F. C., van Engelen, B., Lammens, J. M. (1998). Human CASK/LIN-2 binds syndecan-2 and protein 4.1 and localizes to M., von Zglinicki, T., Foisner, R., Hutchison, C. and Markiewicz, E. (2007). the basolateral membrane of epithelial cells. J. Cell Biol. 142, 129-138. Nucleoplasmic LAP2alpha-lamin A complexes are required to maintain a proliferative Dyer, J. A., Kill, I. R., Pugh, G., Quinlan, R. A., Lane, E. B. and Hutchison, C. J. state in human fibroblasts. J. Cell Biol. 176, 163-172. (1997). Cell cycle changes in A-type lamin associations detected in human dermal Rapraeger, A. C., Krufka, A. and Olwin, B. B. (1991). Requirement of heparan sulfate for

Journal of Cell Science fibroblasts using monoclonal antibodies. Res. 5, 383-394. bFGF-mediated fibroblast growth and myoblast differentiation. Science 252, 1705-1708. Elenius, K., Vainio, S., Laato, M., Salmivirta, M., Thesleff, I. and Jalkanen, M. (1991). Rider, C. C. (2006). Heparin/heparan sulphate binding in the TGF-beta cytokine superfamily. Induced expression of syndecan in healing wounds. J. Cell Biol. 114, 585-595. Biochem. Soc. Trans. 34, 458-460. Elenius, V., Gotte, M., Reizes, O., Elenius, K. and Bernfield, M. (2004). Inhibition by Roberts, S., Calautti, E., Vanderweil, S., Nguyen, H. O., Foley, A., Baden, H. P. and the soluble syndecan-1 ectodomains delays wound repair in mice overexpressing Viel, A. (2007). Changes in localization of human discs large (hDlg) during keratinocyte syndecan-1. J. Biol. Chem. 279, 41928-41935. differentiation is associated with expression of alternatively spliced hDlg variants. Exp. Frye, M. and Watt, F. M. (2006). The RNA methyltransferase Misu (NSun2) mediates Cell Res. 313, 2521-2530. Myc-induced proliferation and is upregulated in tumors. Curr. Biol. 16, 971-981. Smola, H., Stark, H. J., Thiekotter, G., Mirancea, N., Krieg, T. and Fusenig, N. E. Funke, L., Dakoji, S. and Bredt, D. S. (2005). Membrane-associated guanylate (1998). Dynamics of basement membrane formation by keratinocyte-fibroblast regulate adhesion and plasticity at cell junctions. Annu. Rev. Biochem. 74, 219-245. interactions in organotypic skin culture. Exp. Cell Res. 239, 399-410. Gardner, K. L., Sanford, J. L., Mays, T. A. and Rafael-Fortney, J. A. (2006). CASK Stepp, M. A., Gibson, H. E., Gala, P. H., Iglesia, D. D., Pajoohesh-Ganji, A., Pal-Ghosh, localizes to nuclei in developing skeletal muscle and motor neuron culture models and S., Brown, M., Aquino, C., Schwartz, A. M., Goldberger, O. et al. (2002). Defects is agrin-independent. J. Cell Physiol. 206, 196-202. in keratinocyte activation during wound healing in the syndecan-1-deficient mouse. J. Geng, Y. and Weinberg, R. A. (1993). Transforming growth factor beta effects on Cell Sci. 115, 4517-4531. expression of G1 cyclins and cyclin-dependent protein kinases. Proc. Natl. Acad. Sci. USA 90, 10315-10319. Stepp, M. A., Liu, Y., Pal-Ghosh, S., Jurjus, R. A., Tadvalkar, G., Sekaran, A., Losicco, Grootjans, J. J., Zimmermann, P., Reekmans, G., Smets, A., Degeest, G., Durr, J. and K., Jiang, L., Larsen, M., Li, L. et al. (2007). Reduced migration, altered matrix and David, G. (1997). Syntenin, a PDZ protein that binds syndecan cytoplasmic domains. enhanced TGFbeta1 signaling are signatures of mouse keratinocytes lacking Sdc1. J. Proc. Natl. Acad. Sci USA 94, 13683-13688. Cell Sci. 120, 2851-2863. Guimond, S., Maccarana, M., Olwin, B. B., Lindahl, U. and Rapraeger, A. C. (1993). Tsuda, M., Kamimura, K., Nakato, H., Archer, M., Staatz, W., Fox, B., Humphrey, Activating and inhibitory heparin sequences for FGF-2 (basic FGF). Distinct requirements M., Olson, S., Futch, T., Kaluza, V. et al. (1999). The cell-surface proteoglycan Dally for FGF-1, FGF-2, and FGF-4. J. Biol. Chem. 268, 23906-23914. regulates Wingless signalling in Drosophila. Nature 400, 276-280. Hata, Y., Butz, S. and Sudhof, T. C. (1996). CASK: a novel dlg/PSD95 homolog with Waikel, R. L., Kawachi, Y., Waikel, P. A., Wang, X. J. and Roop, D. R. (2001). Deregulated an N-terminal calmodulin-dependent protein kinase domain identified by interaction with expression of c-Myc depletes epidermal stem cells. Nat. Genet. 28, 165-168. neurexins. J. Neurosci. 16, 2488-2494. Wang, T. F., Ding, C. N., Wang, G. S., Luo, S. C., Lin, Y. L., Ruan, Y., Hevner, R., Hennings, H., Michael, D., Cheng, C., Steinert, P., Holbrook, K. and Yuspa, S. H. Rubenstein, J. L. and Hsueh, Y. P. (2004). Identification of Tbr-1/CASK complex (1980). Calcium regulation of growth and differentiation of mouse epidermal cells in target genes in neurons. J. Neurochem. 91, 1483-1492. culture. Cell 19, 245-254. Waseem, A., Dogan, B., Tidman, N., Alam, Y., Purkis, P., Jackson, S., Lalli, A., Hsueh, Y. P. (2006). The role of the MAGUK protein CASK in neural development and Machesney, M. and Leigh, I. M. (1999). Keratin 15 expression in stratified epithelia: synaptic function. Curr. Med. Chem. 13, 1915-1927. downregulation in activated keratinocytes. J. Invest. Dermatol. 112, 362-369. Hsueh, Y. P. and Sheng, M. (1999). Regulated expression and subcellular localization of Watt, F. M., Lo, Celso, C. and Silva-Vargas, V. (2006). Epidermal stem cells: an update. syndecan heparan sulfate proteoglycans and the syndecan-binding protein CASK/LIN- Curr. Opin. Genet. Dev. 16, 518-524. 2 during rat brain development. J. Neurosci. 19, 7415-7425. Werner, S., Krieg, T. and Smola, H. (2007). Keratinocyte-fibroblast interactions in wound Hsueh, Y. P., Yang, F. C., Kharazia, V., Naisbitt, S., Cohen, A. R., Weinberg, R. J. and healing. J. Invest. Dermatol. 127, 998-1008. Sheng, M. (1998). Direct interaction of CASK/LIN-2 and syndecan heparan sulfate Zanet, J., Pibre, S., Jacquet, C., Ramirez, A., de Alboran, I. M. and Gandarillas, A. proteoglycan and their overlapping distribution in neuronal synapses. J. Cell Biol. 142, (2005). Endogenous Myc controls mammalian epidermal cell size, hyperproliferation, 139-151. endoreplication and stem cell amplification. J. Cell Sci. 118, 1693-1704.