[1] 13w A neural-specific splicing event generates an active form of the Wiskott-Aldrich syndrome protein

[1] 145w Actin polymerization is required for cellular events such as podosome, lamellipode or filopode formation in migrating cells, and members of the Wiskott-Aldrich syndrome protein (WASP) family have essential roles in regulating actin dynamics at the cell leading edge. However, WASP proteins need first to be activated in order to be able to target actin polymerization. Here, we show the occurrence of a neural-specific splicing event, which is favoured by the nuclear orphan receptor chicken ovalbumin upstream promoter-transcription factor I, and generates a truncated WASP protein deleted of exon 2-encoded amino acids. This deletion relocates the protein to the plasma membrane and induces the formation of actin-rich podosome-like structures that also contain paxillin and vinculin. Furthermore, expression of the truncated protein in PC12 cells, as well as in primary neurons, stimulates neuritogenesis. These data underscore the importance of the neural-specific splicing of WASP RNA during development.

[1] 70w In the past ten years, proteins of the Wiskott-Aldrich syndrome protein (WASP) family (i.e. WASP, N-WASP and three WASPfamily verprolin homologous protein (WAVE)/Scar isoforms) have emerged as key regulators of actin polymerization. These proteins are involved in the formation of cellular structures required for cell locomotion (Linder et al, 1999;Takenawa & Miki, 2001), phagocytosis and endocytosis (Lorenzi et al, 2000;Kessels & Qualmann, 2002) and vesicle transport (Taunton et al, 2000).

[2] 183w WASP proteins are known to interact through distinct conserved domains, with a large variety of cell-signalling molecules implicated in actin filament rearrangement. The amino-terminal part of WASP proteins carries a WASP homology 1 (WH1) domain (aa 39-149), which shows similarity to pleckstrin homology (PH) domains (Miki et al, 1996). A central domain has been shown to bind Rho family GTPases and is thus called the GTPase-binding domain (GBD). The GBD is also able to contact the carboxy-terminal part of WASP, thus generating an autoinactive form of WASP that is unable to bind the actin nucleation complex Arp2/3 (Kim et al, 2000). Activation can, however, occur through binding of Cdc42 and/or PIP 2 (Prehoda et al, 2000;Rohatgi et al, 2000). WASP proteins share a common C-terminus that consists of one or two verprolin homology regions (V), followed by a cofilin homology and acidic region (CA). V and CA motifs are able to interact with monomeric actin (Miki & Takenawa, 1998) and Arp2/3, respectively (Machesky & Insall, 1998). Thus, the VCA domain is likely to be involved directly in actin nucleation (Yamaguchi et al, 2002).

[3] 104w Neuronal actin dynamics involve at least N-WASP and the WAVEs. Indeed, N-WASP is able to enhance neurite outgrowth from differentiating PC12 cells (Banzai et al, 2000), whereas WAVE proteins are found in growth cones of NG108 cells (Nozumi et al, 2003). As for the WASP gene, it is known to be mainly expressed in the haematopoietic lineage, in which it has an active role, as Wiskott-Aldrich syndrome (WAS) patients show severe phenotypes such as thrombocytopenia and immunodeficiency (Wiskott, 1936;Sullivan et al, 1994). In this report, we show that the WASP gene can generate a neural-specific form of WASP that is able to stimulate neuritogenesis.

[1] 104w Next, we tested the effects of the different fusion proteins on neurite extension during PC12 cell neuronal differentiation. This cell line produces a homogeneous population of sympathetic neurons when treated with nerve growth factor (NGF), and neurite extension in the differentiating cells is known to involve WASPlike proteins (Banzai et al, 2000). This system is thus more suitable for analysing the impact of the various forms of WASP on neurite extension than is the P19 cell system, which produces numerous Involvement of WASP in neural differentiation Y.L. Page et al &2004 EUROPEAN MOLECULAR BIOLOGY ORGANIZATION EMBO reports VOL 5 | NO 9 | 2004

[1] 367w In this study, we show that WASP expression is not only restricted to haematopoietic cells but also significant in neural cells. Indeed, WASP mRNA and proteins are detected in the brain tissue and in differentiating P19 cells. These results are consistent with those of She et al (1997), who reported northern blot detection of WASP mRNA in several tissues, notably in the brain. Thus, neuronal cells express several members of the WASP family, namely N-WASP, WASP and WAVEs. In vitro, all these proteins act similarly by stimulating actin polymerization. Why do neuronal cells then need so many different WASP proteins if they fulfill similar functions? One explanation would be that each WASP protein exerts specific functions within the cell, either because of a specific subcellular localization or because WASP proteins are specifically activated through individual signalling pathways. Experimental data are in favour of at least the first hypothesis. Indeed, WASP and N-WASP are known to be differentially addressed in Cos-7 cells, whereas WASP is mostly perinuclear (Kato et al, 1999) and N-WASP is at the plasma membrane (Miki & Takenawa, 1998;Yamaguchi et al, 2000). Furthermore, WAVE1, WAVE2 and WAVE3 show different distributions in the growth cones of NG108 neural cells (Nozumi et al, 2003). We now show that exon 2 skipping in the WASP RNA generates a form that has a different subcellular localization. The deletion specifically targets the WH1 domain, a domain that provides important protein-protein and protein-lipid interfaces, including the one allowing WIP recruitment (Prehoda et al, 1999;Rohatgi et al, 2000;Rong & Vihinen, 2000, Volkman et al, 2002). Deleting part of the WH1 domain should thus generate a protein that lacks interaction with WIP. This is substantiated by the fact that WASPDEx2 is unable to localize WIP in F-actin-containing podosome-like structures in Cos-7 cells. WIP is known to inhibit the N-WASP/Cdc42-activated actin polymerization mediated by Arp2/3 (Martinez-Quiles et al, 2001), probably through stabilization of the inactive conformation of WASP. As WIP is expressed in the brain cells (data not shown), one would expect that WASPDEx2 stimulates neuritogenesis owing to a higher ability to reach the activated state. Our data support such a model and further highlight the complex regulation of actin dynamics in neural cells.

[1] 286w Plasmid construction. The mouse WASP cDNA was transferred in-frame from Bluescript to pEGFP-C1 or C2 (Clontech) by XhoI digest for GFP-WASP FL, HindIII/BamHI double digest for GFP-WASP1-254 and BamHI digest for GFP-WASP255-521. The deletion of exon 2 (47-93 aa) sequence was generated by sitedirected mutagenesis and restriction-enzyme-mediated excision of the DNA segment. For this purpose, the QuickChange TM sitedirected mutagenesis kit (Stratagene, USA) was used with the following primers: mut1up 5 0 -GCTTGGCCGAAAGTACTGGA CACTGGCTACC-3 0 ; mut1down 5 0 -GCCAGTGTCCAGTACTTTCG GCCAAGAAGCTCAAAG-3 0 and mut2up 5 0 -ATGGCCTACTGTCG GCTACTCTGGGAACAGGAGC-3 0 , and mut2down 5 0 -CCCAGAG Involvement of WASP in neural differentiation Y.L. Page et al EMBO reports VOL 5 | NO 9 | 2004 &2004 EUROPEAN MOLECULAR BIOLOGY ORGANIZATION RT-PCR and western blot analysis. A 2 mg protein of total RNA was used as template for M-MLV reverse transcriptase and Pd(N)6 random hexamer (Amersham Pharmacia). PCR amplification was performed on 2 ml of the reverse transcription reaction, with 1 mM of the following primers: mWASP up 5 0 -CAGAGACTCTTTGAGC-3 0 ; mWASP down 5 0 -AGTGTGGAAGAACGG-3 0 ; mN-WASP up 5 0 -AACGAGTCTCTCTTCTCCTTCC-3 0 ; mN-WASP down 5 0 -AGAT GTTGTTGACTTGTGAACC-3 0 ; PO up 5 0 -CAGCTCTGGAGAAAC TGCTG-3 0 ; PO down 5 0 -GTGTACTCAGTCTCCACAGA-3 0 . For radioactive PCRs, 2 mCi [ 32 P-a]dCTP was added to the amplification mix. When assaying the activity of the minigene, the constructs were co-transfected in Cos-7 cells together with a b-galactosidase expression vector, and the amount of RNA used for reverse transcription was normalized for b-galactosidase activity. For western blot analysis, whole-cell extracts were run in a 12.5% acrylamide/ 0.085% bisacrylamide gel. After blotting, the membrane was incubated with a WASP antibody (B-9) from Santa Cruz, Inc.

[2] 411w Cell culture and transfections. PC12 cells were grown in Dulbecco's modified Eagle's medium (DMEM) supplemented with 7% horse serum and 3% fetal calf serum (FCS). In all, 2 mg of each recombinant plasmid was transfected into PC12 cells with FuGENE TM 6 transfection reagent (Roche). At 24 h after transfection, differentiation was induced with 50 ng/ml recombinant NGF (ICN) and allowed to proceed for 48 h in DMEM supplemented with 1% horse serum. P19 embryonal carcinoma cells and Cos-7 cells were grown in DMEM supplemented with 10% FCS. Stable P19 cell lines and RA-induced neuronal differentiation were as described (Adam et al, 2000). Cos-7 cells were grown in the same conditions as P19 cells. P5 cerebellar neurons were obtained from mice as described (Loeffler et al, 1990). Cells were seeded on polylysine-coated coverslips in minimum essential medium (MEM)/10% FCS. At 1 h after plating, cells were transfected with the various vectors and Lipofectamine 2000 (Invitrogen). F-actin visualization and immunocytochemistry. Cos-7 cells were transfected with the various GFP fusion vectors. After 48 h, cells were fixed with 4% paraformaldehyde and 10% saccharose in phosphate-buffered saline (PBS), and then washed twice with 100 mM glycine in PBS. Labelling of F-actin was performed with 1% Tween 20 and 3 mM tetramethylrhodamine isothiocyanate (TRITC)-phalloidin (SIGMA) in PBS. For immunocytochemistry, Cos-7 cells were grown on polylysine-coated coverslips and transfected with GFP-WASP expression vectors. Cells were then fixed and incubated with primary antibodies (anti-paxillin from BD Biosciences; anti-vinculin (H-300) and anti-WIP (G-20) from Santa Cruz, Inc.), followed by incubation with rhodamine- b coupled secondary antibodies (Jackson Immunoresearch Laboratories, Inc.). Images were acquired on a DMRXA2 microscope (Leica) using a CCD camera (CoolSnap HQ, Ropper Scientific) and subsequently processed by the Metamorph software (Universal Imaging). For confocal microscopy, a TCS NT microscope (Leica) was used. Neurite analysis and statistical methods. PC12 cells were directly observed in culture plates with an epifluorescence microscope coupled to a digital camera. P5 cerebellar neurons were grown on polylysine-coated coverslips and mounted on glass slides before observation. GFP-positive cells with neurites longer than the diameter of two cell bodies were scored as 'neurite bearing'. Neurite length was estimated with the Scion Image software (Scion Corp.). Two separate experiments were carried out, and similar results were recorded. At least 200 cells for each treatment were counted. Statistical analyses were performed using the w 2test, Po0.05 for PC12 cells and the Student's t-test for P5 neurons. Supplementary information is available at EMBO reports online (http://www.emboreports.org).

[1] 408w Previous data have shown that the nuclear orphan receptor chicken ovalbumin upstream promoter-transcription factor I (COUP-TFI), when overexpressed during retinoic-acid-induced differentiation of P19 embryonal carcinoma cells, is able to stimulate cell migration as well as neuritogenesis (Adam et al, 2000); two phenomena that are linked to actin dynamics and WASP protein activity. We first ran semiquantitative reversetranscription-polymerase chain reaction (RT-PCR) experiments in control P19 cells, as well as in P19 cells stably expressing COUP-TFI (Adam et al, 2000), to examine whether COUP-TFI could influence WASP and/or N-WASP expression. Fig 1A shows that both genes are expressed in P19 cells and that retinoic acid did not modify N-WASP RNA levels. The WASP gene showed a more complex expression pattern because two RNAs were detected, the longer one being mainly expressed in undifferentiated cells, and the shorter one mainly in retinoic acid (RA)-treated cells. In P19 cells stably expressing COUP-TFI, N-WASP RNA levels remained scientificreport scientific report unaffected by RA (data not shown), whereas only the short WASP message was detected on addition of RA (Fig 1B). Stable expression of a mutated COUP-TFI unable to bind DNA had no effect (Fig 1B). We next examined WASP expression by RT-PCR on RNAs prepared from different tissues and regions of the nervous system (Fig 1C). Interestingly, the short WASP RNA was seen only with samples derived from neural tissues. The corresponding complementary DNAs (cDNAs) were cloned and sequenced, revealing that the 250 bp fragment was the expected DNA containing the WASP first three exons, whereas exon 2 was absent from the 100 bp fragment. Thus, exon 2 skipping in WASP RNA is likely to be a hallmark of neural cells. In silico translation of WASP RNA deleted from exon 2 revealed that no frameshift or stop codon was introduced by this event, and that the corresponding protein would present a 47 aa deletion (47-93 aa) in the WH1 domain. Western blot analysis revealed the presence of one WASP protein in the spleen extract, migrating at the expected position, whereas P19 cell extracts contained two WASP proteins: the full-length (FL) WASP and a shorter form lacking about 5 kDa when compared with FL (Fig 1D). Cerebellar neurons from a 5-day-old mouse (P5) also expressed both short and FL WASP proteins (Fig 1E). Knowing that the theoretical difference in molecular weight between WASP FL and WASPDEx2 is 5.3 kDa, the shorter form detected by western blot might result from the exclusion of exon 2.

[2] 93w To determine if exon 2 skipping could be regulated by COUP-TFI, we constructed a WASP minigene containing 2 kb of promoter sequence as well as gene sequence up to intron 4 (Fig 2). When this minigene was transfected alone in Cos-7 cells, the resulting RNA was detected as a single amplicon including exon 2. Notably, co-transfection of the minigene with a COUP-TFI expression vector in Cos-7 cells resulted in a partial exclusion of exon 2 from the transcript, indicating a crucial role for the orphan receptor in regulating splice site selection (Fig 2).

[3] 169w Green fluorescent protein (GFP)-fused WASP domains were then expressed in Cos-7 cells to analyse the subcellular distribution of the fusion proteins in comparison with F-actin (Fig 3). Examination of the transfected cells showed that GFP-WASP was mainly found in a large perinuclear domain (first set of GFP-WASP images, Involvement of WASP in neural differentiation Y.L. Page et al EMBO reports VOL 5 | NO 9 | 2004 &2004 EUROPEAN MOLECULAR BIOLOGY ORGANIZATION Nonetheless, the typical ring-shaped distribution of paxillin and vinculin usually found in podosomes was not observed. Interestingly, WIP, a protein interacting with the WH1 domain of WASP and N-WASP (Ramesh et al, 1997;Volkman et al, 2002) and which has been shown to localize in podosomes (Moreau et al, 2003), was not detected in WASPDEx2-containing structures (Fig 4). These results indicate that WASPDEx2 is able to induce actin polymerization, as well as paxillin and vinculin accumulation, in ventral structures, which may differ from classical podosomes, and that the truncated WASP may not be able to interact with WIP.