Integrins Regulate Centrosome Integrity and Astrocyte Polarization Following a Wound Huashan Peng,1 Yen May Ong,1 Waris Ali Shah,1 Paul C. Holland,2 Salvatore Carbonetto1 1
Centre for Research in Neuroscience, McGill University Health Centre, Montreal, Quebec H3G 1A4, Canada
2
Montreal Neurological Institute, McGill University, Montreal, Quebec H3G 1A4, Canada
Received 18 May 2012; revised 14 August 2012; accepted 27 August 2012
ABSTRACT: In response to a wound, astrocytes in culture extend microtubule-rich processes and polarize, orienting their centrosomes and Golgi apparatus woundside. b1 Integrin null astrocytes fail to extend processes toward the wound, and are disoriented, and often migrate away orthogonal, to the wound. The centrosome is unusually fragmented in b1 integrin null astrocytes. Expression of a b1 integrin cDNA in the null background yields cells with intact centrosomes that polarize and extend processes normally. Fragmented centrosomes rapidly assemble following integrin ligation and cell attachment. However, several experiments indicated that cell adhesion is not necessary. For example, astrocytes in suspension expressing a chimeric b1 subunit that can be activated by an antibody assemble centrosomes suggesting that b1 activation is sufficient to cause centrosome assembly in the absence of cell adhesion. siRNA knock-
INTRODUCTION Centrosomes were among the first organelles identified and were recognized early on as important in cell division and cytoskeletal organization. Now routinely referred to as the major microtubule organizing center (MTOC) in animal cells (Luders and Stearns, 2007), we know that the centrosome is an aggregate of hunCorrespondence to: S. Carbonetto (
[email protected]). Contract grant sponsors: NSERC, CIHR. ' 2012 Wiley Periodicals, Inc. Published online 4 September 2012 in Wiley Online Library (wileyonlinelibrary.com). DOI 10.1002/dneu.22055
down of PCM1, a major centrosomal protein, inhibits cell polarization, consistent with the notion that centrosomes are necessary for polarity and that integrins regulate polarity via centrosome integrity. Screening inhibitors of molecules downstream of integrins indicate that neither FAK nor ILK is involved in regulation of centrosome integrity. In contrast, blebbistatin, a specific inhibitor of non-muscle myosin II (NMII), mimics the response of b1 integrin null astrocytes by disrupting centrosome integrity and cell polarization. Blebbistatin also inhibits integrin-mediated centrosome assembly in astrocytes attaching to fibronectin, consistent with the hypothesis that NMII functions downstream of integrins in regulating centrosome integrity. ' 2012 Wiley Periodicals, Inc. Develop Neurobiol 73: 333–353, 2013
Keywords: cell polarity; integrins; centrosome integrity; NMII
dreds of proteins (Andersen et al., 2003; BettencourtDias and Glover, 2007; Muller et al., 2010) and includes signaling as well as structural complexes (Doxsey et al., 2005). These centrosomal protein complexes impinge on many cellular events and their organization within scaffolds is critical for their function (Mikule et al., 2007). This is most obvious in cell division where centrosomes replicate in tune with the cell cycle to give rise to one centrosome at each pole of the mitotic spindle. The contribution of centrosomes to the spindle apparatus is required for equal partitioning of chromosomes to daughter cells (Piel et al., 2001; Raff, 2003) as well as control of the cell cycle (Doxsey et al., 2005). When centrosome 333
334
Peng et al.
organization is disrupted function can be perturbed (Mikule et al., 2007; Sir et al., 2011). Centrosome disorganization is variously described as amplified, split, diffuse, unclustered, and immature; abnormalities that are hallmarks of cancerous cells (Nigg, 2002). In post-mitotic cells centrosomes are critical for orientation of the nucleus, Golgi apparatus, and endoplasmic reticulum as well as for cell polarity. In neural cells, the most highly polarized of all cell types, the direction of neuroblast migration or of subsequent axonal outgrowth is thought to be controlled by the centrosome (Solecki et al., 2004; de Anda et al., 2005, 2010; c.f. Distel et al., 2010; Stiess et al., 2010). Similarly, astrocytes polarize to form radial glia in the developing brain that guide migrating neuroblasts and are also neuronal precursors (Mori et al., 2005). In the adult brain astrocytes are polarized with respect to the vasculature (Petzold and Murthy, 2011), synapses (Hamilton and Attwell, 2010; Henneberger et al., 2010), and extend processes forming a glial scar around a wound (Sofroniew, 2009). Once initiated, events in the cell cycle, including centrosome dynamics, are widely viewed as cell autonomous. However, the plane of cell division can be modulated by cell adhesion to the extracellular matrix (ECM) (Thery et al., 2005; Kwon et al., 2008; Sofroniew, 2009). Other experiments implicate ECM receptors of the integrin superfamily in the regulation of centrosomes during mitotic spindle assembly, and cytokinesis (Fielding et al., 2008; LaFlamme et al., 2008). Here, we show that integrins control the integrity of the centrosome. Post-mitotic astrocytes in culture can be induced to polarize and respond to a wound (Etienne-Manneville and Hall, 2001). Lack of b1 integrin expression or inhibitory b1integrin antibodies disrupt centrosomal integrity. This correlates with disorientation and rapid migration of astrocytes into the wound. Binding of integrins to ligands immobilized to culture dishes mediates cell attachment and rapid centrosome assembly. Several observations indicate that adhesion per se is not necessary for integrin regulation of centrosome integrity. Thus centrosome assembly occurs on polylysine-coated substrates and is blocked by integrin antibodies although polylysine does not activate focal adhesion kinase (FAK). More importantly, the b1 subunit is sufficient to trigger centrosome assembly when activated by aggregation independently of cell-substratum adhesion. Perturbation of FAK or of integrinlinked kinase (ILK), two major intermediates in the integrin signaling pathway, does not affect Developmental Neurobiology
centrosome integrity. In contrast, inhibition of nonmuscle myosin II (NMII) with blebbistatin or siRNA knockdown disrupts centrosome integrity and cell polarity similar to integrin perturbation suggesting that NMII, which has been previously implicated in cell polarity and migration, is also important in centrosome integrity. Blebbistatin also inhibits centrosome assembly mediated by integrins attaching to fibronectin. Together these data suggest a novel pathway of regulation of cell polarization via integrins acting on centrosome integrity.
MATERIALS AND METHODS Reagents Anti-c-tubulin (ab11317) and anti-ILK (ab2283) antisera were obtained from AbCam, Cambridge, MA. Anti-pericentrin (PRB-432C) antibodies were from Covance, Princeton, NJ. Antibodies against b-tubulin and vinculin were obtained from Sigma-Aldrich Canada, Oakville, Ontario. Anti-FAK397 (44-624G) which specifically reacts with activated FAK phosphorylated at tyrosine 397 (Y-397) was from Invitrogen Canada, Burlington, ON. JG22, a chick specific b1-integrin, function-blocking antibody was gift from Dr. David Gottlieb, (Washington University). Function-blocking antibodies to the a6 subunit were obtained from Santa Cruz Biotechnology, Santa Cruz, CA. Antibodies to the a1 subunit were generated in the lab and described previously (Turner et al., 1989; Tawil et al., 1990). Function blocking rabbit antiserum directed against the b1-integrin subunit and non-function blocking antiserum directed against a synthetic peptide corresponding to a b1 subunit cytoplasmic domain sequence were described previously (Tawil et al., 1993). The secondary antibodies used were donkey anti-mouse Alexa Flour 488, 555 (Invitrogen Canada, Burlington, ON) and goat anti-mouse rhodamine (Jackson Laboratories, Bar Harbor, Maine). Other reagents included RITC-, FITC- and CPITC-(Rhodamimine Flouroscein and Coumarin Isothiocyanate) conjugated phalloidin, arg-gly-asp (RGD) peptide, arg-gly-glu (RGE; control) peptide were obtained from Sigma-Aldrich Canada, Oakville, Ontario.
Cell Culture, Scratch-Wound Assay, and Immunocytochemistry Mouse ES cells were differentiated into astrocytes as previously described (Peng et al., 2008). Primary astrocytes were cultured from E17-18 embryos of Sprague-Dawley rats (Tawil et al., 1993). Chick brain cells were cultured by methods similar to ones previously published (Carbonetto and Fambrough, 1979). The cultures were fixed and process-bearing cells identified as neurons by immunocytochemistry with antibodies to neurofilaments. Highly
Integrin-Centrosome Pathway in Cell Polarity enriched cultures of chick myoblasts were prepared by conventional methods (Devreotes and Fambrough, 1976) as were enriched cultures of skin fibroblasts (Seluanov et al., 2010). For scratch wound and other assays, acid-washed, glass coverslips were sterilized by UV irradiated for 20 min and coated with either laminin at 20 lg/mL, Type 1 collagen, fibronectin at 10 lg/mL, or polylysine at 100 lg/mL overnight at 378 (all from Sigma-Aldrich Canada, Oakville, Ontario, except fibronectin which was from BD Canada, Mississauga, Ontario). Astrocytes were harvested in serumfree medium, seeded on the coated glass coverslips in 24well tissue culture dishes, and incubated at 378, 5% CO2 until they were confluent. Scratch wounding of astrocyte monolayers and the subsequent estimation of centrosome reorientation were carried out as previously described (Etienne-Manneville, 2006; Peng et al., 2008). Cells were fixed with 4% paraformaldehyde (PFA) containing 4% sucrose. To visual the centrosome, cells were fixed with methanol at 208 for 10 min followed by three washes with PBS with the first wash being 10 min to allow the cells to rehydrate after methanol fixation. PFA fixed astrocytes were permeabilized with 0.25% Triton-X-100 and washed three times with PBS. Cells on coverslips were then blocked with 10% BSA for 1 h at 238C or 48C overnight, followed by three washes with PBS before incubating the coverslips with the primary antibody in 3% BSA for 1 h at room temperature or at 48C overnight. The primary antibody was then removed and the coverslips washed three times with PBS before addition of the secondary antibody in 3% BSA for 1 h at room temperature followed by 40 ,6-diamidino-2-phenylindole (DAPI, 0.1 lg/mL) for 10 min at 238C. Coverslips were washed two times with PBS and then water, and mounted, on slides with Slow Fade Gold (Invitrogen) and sealed with clear nail polish. Experiments with blebbistatin were carried out essentially as described above except that after wounding cells were treated with 5–25 lM blebstatin; a range of concentrations that has been shown highly selective for inhibition of NMII (Duxbury et al., 2004; Schenk et al., 2009).
Organ Culture of Chick Retinas Retinas were dissected from E14 chick embryos and cut into quadrants. The quadrants were cultured floating in 24well plates in which the medium had been equilibrated with 5% CO2. In some cultures, retinas were treated for 2 or 4 h with mab JG22. At the end of this period the tissue was washed with PBS, fixed with 4% PFA in 100 mM phosphate buffer (pH 7.3) for 15 min at 238C and sequentially through 5%, 10%, 20%, and lastly 30% sucrose solution. Segments were cryoprotected in 30% sucrose solution for at least 24 h at 48C, embedded in optimum cutting temperature mounting medium (O.C.T., Sakura Finetek, Torrance, CA) and cut transversally (10 lm sections) on a cryostat. The retinal sections were permeabilized by treatment with 0.25% Triton X-100 in PBS for 5 min, blocked by incubation in 10% BSA for 30 min, and washed 33 5 min in PBS. Sections
335
were incubated with primary antibody in 3% BSA, PBS for 60 min (or over night at 48C depending on antibody concentration and the accessibility of the antigen) and washed three times with PBS. Sections were then incubated with secondary antibodies in 3% BSA for 60 min with 0.1 lg/ mL DAPI for 10 min and washed three times with PBS.
Plasmids, Recombinant Adenovirus, and siRNAs Recombinant adenovirus containing a b1 integrin subunit expression cassette was engineered and used similar to methods previously described (Tremblay and Carbonetto, 2006). Dr. Susan LaFlamme generously supplied us with an expression vector for a chimeric integrin b1 subunit that consisted of the extracellular and transmembrane domains of the small subunit of the human interleukin-2 receptor (tac) fused to the b1 cytoplasmic domain (LaFlamme et al., 1994). Astrocytes cultured to 60% confluence, were infected with 1 MOI of the tac b1 integrin adenovirus in DMEM and vitamins for 2 h. The infected cells were cultured in 10% FBS DMEM medium for 3 days to allow expression of tac b1 chimeric integrin. The cells were then detached from the culture dish by trypsinization, and kept in suspension for 2 h to allow the centrosome to disassemble (Fig. 3). The suspended cells were treated with an antibody to tac for 45 min at 48C and then for 15 min at 378C to cluster the tac b1 chimeric integrin. The cells were subsequently washed, fixed with 4% PFA, and permeabilized with TritonX-100 for immunocytochemistry. Myc-tagged FRNK expression vector was a gift from Dr. J.T. Parsons (University of Virginia, Charlottesville). Recombinant adenovirus containing a b1 integrin subunit expression cassette was engineered and used as previously described (Tremblay and Carbonetto, 2006). StealthTM Select siRNAs (Invitrogen Canada) were used to knock down PCM1 expression (Catalog nos. RSS 330828; 330829; 330830) and ILK expression (Catalog nos. RSS 301329; 301330; 301331). BLOCK-iTTM Alexa Fluor1 Red Fluorescent Control oligonucleotide (Invitrogen Cat no. 14750100) was used to test transfection efficiency and Stealth Universal Control siRNA with medium GC content (Invitrogen Cat no. 12935-300) was used as a negative (scrambled oligonucleotide) control. Cells were transfected with oligonucleotides and plasmids using Lipofectamine 2000 (Invitrogen Canada Inc.) according to the manufacturer’s instructions.
Fluorescence Microscopy Images were made with a Zeiss Axiophot microscope and captured with a Qimaging Retiga1300 10-bit digital camera under normalized exposures. Data were observed by at least two individuals including one who was uninformed about the samples being viewed. All quantification of morphometric data were generated after methods previously published (Gee et al., 1994) using Northern Eclipse software (Empix). For statistics, we performed multiple comparisons Developmental Neurobiology
336
Peng et al.
using unbalanced one-way ANOVA followed by pair wise post hoc comparisons (Tukey–Kramer and Fisher’s tests). Data are presented as means 6 the standard deviation. All probability (p) values stated in the text are from analysis by ANOVA. All statistical analyses were generated using StatView 4.5 package (Abacus). All images and figures were processed using Adobe Photoshop and Illustrator.
RESULTS The b1 Integrin Subunit is Necessary for Centrosome Integrity and Polarization of Astrocytes Following a Wound Monolayers of astrocytes extend microtubule-rich processes and polarize when wounded, orienting their centrosomes and Golgi apparatus woundside. Our previous studies on astrocyte wound healing have implicated integrins and dystroglycan, two major types of ECM receptor, in the astrocytic response to a wound (Peng et al., 2008). Process extension is distinct from cell polarization and involves positioning of the centrosome and Golgi apparatus woundside (Etienne-Manneville and Hall, 2001). Dystroglycan contributes to process extension but lack of its expression or blocking its function has little effect on astrocyte polarization. We generated b1 integrin null astrocytes, from b1 null ES cells (Fassler et al., 1995) [Fig. 1(a)], and compared their response to a wound with astrocytes differentiated from wild-type ES astrocytes [Fig.1(b)]. We found that wild type astrocytes respond much like primary cultures of astrocytes (Etienne-Manneville, 2006; Peng et al., 2008) whereas b1 integrin null astrocytes fail to extend processes toward the wound and instead migrate rapidly out of the monolayer and into the wound area [Fig. 1(b)]. Moreover, the b1 null astrocytes were disoriented and had lost their polarity relative to the wound [Fig. 1(b), arrowheads]. The centrosome, as the MTOC, is thought to be essential for induced polarization and together with the Golgi apparatus defines the polarity of the cell (Etienne-Manneville and Hall, 2001). Hence we sought to map the position of the centrosomes with antibodies to c-tubulin and pericentrin, two proteins essential for centrosomal integrity (Mikule et al., 2007). We were surprised to find that the b1 integrin-null astrocytes lacked any well-defined structure we could identify as a centrosome [Fig. 1(b)]. Instead, we saw a diffuse pattern of staining with many small foci, indicating a loss of centrosome integrity [Fig. 1(b,c)]. This phenotype was due to lack of expression of the b1 integrin subunit and not genetic divergence in the ES cell lines since expression of a full length b1 integrin subunit Developmental Neurobiology
cDNA in b1-null astrocytes [Fig. 1(a)] rescued cell polarity, process extension, and the normal structure of the centrosome [Fig. 1(b)].
Evidence that an Intact Centrosome is Necessary for Astrocyte Polarization Centrosomes have been reported to regulate the position of process outgrowth from the cell bodies of differentiating neurons as they polarize to extend processes (de Anda et al., 2005); however recent data suggest that this may not be invariably the case (Distel et al., 2010). The organelle dynamics during woundinduced polarization are an area of current investigation (Luxton and Gundersen, 2011), although the requirement for an intact centrosome has not been directly addressed. To investigate this further we knocked down the expression of the centriolar satellite protein PCM1 (Fig. 2). Depletion of PCM 1 has previously been shown to reduce targeting to the centrosome of several centrosomal proteins, e.g., pericentrin, centrin, and ninein (Dammermann and Merdes, 2002). Inhibition of PCM1 expression resulted in a marked loss of PCM1 staining and a partial loss of pericentrin staining [Figs. 2 and 3(a), right panels]. As previously reported (Dammermann and Merdes, 2002), c-tubulin was unaffected by PCM1 knockdown [Fig. 3(a)]. PCM1 knockdown followed by a scratch wound resulted in a significant inhibition of centrosome orientation indicative of cell polarization [Fig. 3(b)]. The position of the centrosome was assayed with labeling for c-tubulin since it is unaffected by PCM 1 knockdown. Prior reports demonstrated a role for integrins in astrocyte polarization (Etienne-Manneville and Hall, 2001; Peng et al., 2008). The extent of centrosome disruption visible following knockdown of PCM1, one of hundreds of centrosomal proteins, is not nearly so great as in b1 integrin null cells. Nevertheless, these data reinforce the importance of centrosomes for astrocyte polarization. They are also consistent with the hypothesis that this class of integrins contributes to cell polarization by regulating the centrosome.
Centrosome Assembly and Disassembly with Integrin-Mediated, Cell Attachment Integrins of the b1 subclass are well known to mediate adhesion of cells to laminin, fibronectin, and collagen (Takada et al., 2007). We used relatively rapid cell attachment assays to examine centrosome dynamics on different integrin ligands. Astrocytes maintained in suspension for 4 h have no well-defined cen-
Integrin-Centrosome Pathway in Cell Polarity
trosome and show only diffuse pericentrin or c-tubulin staining throughout the cell (Fig. 4). These astrocytes rapidly assemble a centrosome when allowed to attach to laminin, collagen, fibronectin, or to polylysine (Fig. 4). On all substrates, coalescence of pericentrin staining into a single organelle could be seen
337
as early as 10 min after plating and by 4 h a single focus of pericentrin staining is present in close to 100% of cells examined on all substrates tested [Fig. 4(b)]. Given that b1 integrins are required for centrosome integrity (Fig. 1), it was unexpected that centrosome assembly in astrocytes plated on polylysine pro-
Figure 1 (See legend on following page.) Developmental Neurobiology
338
Peng et al.
ceeded as on ECM proteins. Cell attachment and intracellular signaling typically follow binding of ligands of the b1 integrins to their cognate receptors. In contrast, polylysine, which is highly positively charged, mediates cell attachment as a result of its interaction with negatively charged components of the cell membrane (Miller and Boettiger, 2003); a process which is presumably not dependent on ligation of an integrin. We next asked whether particular integrins affect centrosome assembly and/or the integrity of preassembled centrosomes. For experiments on centrosome assembly cells were maintained in suspension for 4 h at which point the vast majority of cells have disorganized centrosomes (Fig. 4). We either selectively inhibited integrin function, using RGD peptide to target av and a5b1 integrins or with function-blocking antibodies directed against a1b1 or a6b1 integrins. Selective inhibition had no effect on centrosome formation [Fig. 5(a,b)], whereas antibodies to the b1 subunit blocked centrosome assembly. Keeping in mind that the integrins targeted by these reagents have distinct ligand selectivity (a1b1, laminin/collagen; a5b1, fibronectin; a6b1, laminin) the fact that wild-type astrocytes adherent to single ECM components have well-formed centrosomes [Fig. 5(b)] supports the concept of functional redundancy, i.e., any ligand evoking a signal through the b1 subunit is sufficient to maintain centrosome
integrity. The effect of b1 integrin function-blocking antibodies on centrosome assembly continued beyond the initial rapid phase (