Integrins have been implicated in key cellular functions, including cytoskeletal organization, motility, growth, survival, and control of gene expression. The plethora of integrin α and β subunits suggests that individual integrins have unique biological roles, implying specific molecular connections between integrins and intracellular signaling or regulatory pathways. Here, we have used a yeast two-hybrid screen to identify a novel protein, termed Nischarin, that binds preferentially to the cytoplasmic domain of the integrin α5 subunit, inhibits cell motility, and alters actin filament organization. Nischarin is primarily a cytosolic protein, but clearly associates with α5β1, as demonstrated by coimmunoprecipitation. Overexpression of Nischarin markedly reduces α5β1-dependent cell migration in several cell types. Rat embryo fibroblasts transfected with Nischarin constructs have “basket-like” networks of peripheral actin filaments, rather than typical stress fibers. These observations suggest that Nischarin might affect signaling to the cytoskeleton regulated by Rho-family GTPases. In support of this, Nischarin expression reverses the effect of Rac on lamellipodia formation and selectively inhibits Rac-mediated activation of the c-fos promoter. Thus, Nischarin may play a negative role in cell migration by antagonizing the actions of Rac on cytoskeletal organization and cell movement.
The integrin family of cell surface glycoproteins plays a major role in the interaction of cells with the extracellular matrix (Aplin et al. 1998; Giancotti and Ruoslahti 1999). Integrins exist as α/β heterodimers and each subunit has a large extracellular domain, a single helical transmembrane domain, and, typically, a relatively short cytoplasmic domain. At specialized sites of cell–matrix adhesion, termed focal contacts, integrin cytoplasmic domains articulate, directly or indirectly, with various proteins, including talin, α-actinin, vinculin, paxillin, tensin, and focal adhesion kinase (FAK), that are involved in coupling between integrins and the actin cytokeleton (Burridge and Chrzanowska-Wodnicka 1996). Integrin–cytoskeletal linkages play a critical role in cell adhesion, determination of cell shape, and cell motility (Burridge and Chrzanowska-Wodnicka 1996; Miyamoto et al. 1998). Integrins also play an important role in signal transduction processes, either by directly generating signals or by modulating signals generated by other receptors (Clark and Brugge 1995; Schwartz 1997; Aplin et al. 1998, Aplin et al. 1999a; Giancotti and Ruoslahti 1999). Integrin modulation of signaling affects control of the cell cycle (Assoian 1997) and regulation of programmed cell death (Frisch and Ruoslahti 1997).
The cytoplasmic domains of integrins play a key role in their function. Thus, the β chain cytoplasmic tail has been implicated in the recruitment of integrins to focal contacts (Reszka et al. 1992), activation of FAK (Akiyama et al. 1994), and determining the affinity of integrins for their ligands (Wang et al. 1997). Similarly, the α subunit cytoplasmic tail has been implicated in regulation of integrin affinity (O'Toole et al. 1994) and control of cell motility (Chan et al. 1992; Bauer et al. 1993).
Integrins can interact with a variety of partner proteins, including various membrane receptors that bind to the extracellular and transmembrane domains of integrins (Hemler 1998; Porter and Hogg 1998), as well as intracellular proteins that associate with integrin cytoplasmic tails (Aplin et al. 1998). Yeast two-hybrid techniques have been used to identify several proteins that interact with integrin β subunit cytoplasmic domains and that have interesting and important biological functions (Kolanus et al. 1996; Biffo et al. 1997; Chang et al. 1997; Kashiwagi et al. 1997; Delcommenne et al. 1998; Li et al. 1999; Zhang and Hemler 1999). Fewer proteins have been reported to interact with α chain cytoplasmic domains. Thus, calreticulin has been reported to bind the conserved GFFKR motif found in all α chains and to modulate integrin affinity (Coppolino et al. 1997), whereas calcein integrin binding protein (CIB) is a calcium-binding protein that associates specifically with the cytoplasmic domain of αIIb, possibly playing a role in activation of the αIIbβ3 integrin (Naik et al. 1997).
The α5β1 integrin, a receptor for fibronectin, seems to play a special role in regulating growth and survival in some cell types. Thus, high expression of α5β1 has been linked with reductions in tumor cell growth rates both in vitro and in vivo (Giancotti and Ruoslahti 1990; Schreiner et al. 1991; Varner et al. 1995). Surprisingly, α5β1 also plays a unique role in protecting cells against apoptosis triggered by mitogen deprivation (Zhang et al. 1995; O'Brien et al. 1996; Lee and Juliano 2000). In addition, α5β1 has been reported to regulate muscle cell growth and differentiation (Sastry et al. 1999). These data suggest that certain effects of α5β1 on growth or apoptosis may be α5 specific, and thus, there may be intracellular proteins that selectively interact with the α5 cytoplasmic tail to mediate these events. Accordingly, we have made use of the yeast two-hybrid system to identify proteins that bind to the α5 cytoplasmic domain. We have identified a novel protein that associates with the cytoplasmic tail of the α5 subunit, and, to a minor degree, with cytoplasmic domains of other α subunits, and that strongly affects cell migration and influences cytoskeletal organization. We named this novel protein Nischarin, which is derived from a term in classic Sanskrit that connotes slowness of motion. This designation is based on the finding, shown below, that overexpression of Nischarin dramatically impairs cell migration.
Materials and Methods
Yeast Two-Hybrid Screening
Here, L40 (Mata his3Δ200 trp1-901, 112 ade2 LYS2::(lexAop)4-HIS3 URA3(lexAop)8-LacZ Gal4) and AMR70 (Mata his3 lys2 trp1 leu2 URA3::(lexAop)8-LacZ Gal4) yeast strains were used (gifts from Dr. Stan Hollenberg, Vollum Institute, Oregon Health Sciences University, Portland, OR). Yeast two-hybrid screening was conducted as previously described (Vojtek et al. 1993). The pBTM α5 plasmid, which has the α5 cytoplasmic domain fused to the LexA DNA-binding domain and with a tryptophan marker, was transformed into yeast strain L40 and selected for tryptophan prototrophy. Plasmids (pVp16) containing mouse embryonic cDNA libraries of 9.5 and 10.5 d fused to the VP16-transactivating domain and a leucine marker were transformed into the L40 strain containing the bait plasmid and screened for leucine, tryptophan, and histidine prototrophy. A total of 1.7 × 107 transformants were screened for positives. The libraries and vectors were gifts from Dr. Stan Hollenberg. Histidine-positive colonies were further tested for LacZ activation. Dual positives were further confirmed for specificity of the interaction using various baits and included integrin β1, α2, and αv cytoplasmic domains, as well as lamin, an irrelevant protein in this context. Specificity of the interaction was confirmed by mating experiments. The pBTM bait plasmids were “cured” from dual-positive clones by growth in nonselective medium. The presence of the library plasmids with inserts in the “cured” clones was confirmed by PCR using vector-specific primers. AMR 70 strain cells were transformed separately with pBTM α5, pBTM β1, pBTM αv, pBTM α2, pBTM lamin, or pBTM vector alone. These transformed cells were mated with the “cured” L40 cells that contained positive pVP16 library plasmids.
Cloning of Full-Length Nischarin
To clone full-length Nischarin, we screened a mouse brain library in the lambda Zap II vector (Stratagene). Using a colony hybridization technique, ∼30,000 plaques were screened with a 32P-labeled PCR product consisting of 0.45 kb of the integrin-binding region of Nischarin. From this screen, one strong positive plaque was identified and confirmed in two further rounds of screening. Sequence analysis of this clone (A3.1) indicated that the sequence was incomplete at the 3′ end. Using a different PCR probe, the lambda Zap library was screened again to obtain the remainder of the Nischarin cDNA. This screen gave several positives, and the longest clone (clone 14.2) was picked. Clones A3.1 and 14.2 provided the complete open reading frame (ORF) of Nischarin.
DNA Constructions and Transfection
The construction of two-hybrid bait plasmids, GST chimeras, and partial- and full-length myc-tagged Nischarin mammalian expression constructs followed standard recombinant DNA procedures. Clones A3.1 and 14.2, mentioned above, were used to make full-length expression constructs. Full details are available upon request. A chimera comprised of full-length Nischarin and GFP was prepared by an inframe insertion of the coding region of Nischarin into the pE-GFP-N1 vector (CLONTECH Laboratories, Inc.). Expression plasmids for CD4, human α2, and αv integrin subunits were obtained from Drs. R. Nicholas (University of North Carolina-Chapel Hill, Chapel Hill, NC), L. Parise (University of North Carolina-Chapel Hill), and David Cheresh (The Scripps Research Institute, La Jolla, CA), respectively. Transfection of mammalian cell lines was usually done with Lipofectamine (GIBCO BRL) or Superfect (QIAGEN), according to the manufacturer's specifications.
Northern Blot Analyses
A mouse multiple tissue Northern blot (CLONTECH Laboratories, Inc.) was probed with a 0.45-kb fragment of the α5 integrin–binding region of Nischarin (probe 1) (nucleotides 1,305–1,743), with fragments from the far 5′ end (probe 2) (nucleotides −334–+56], or the 3′ end (probe 3)(nucleotides 2,936–3,748). RNA was isolated from various cell lines, run on agarose-formaldehyde gels, and hybridized with probes 1–3, using previously described techniques (Alahari et al. 1996).
The predicted ORF of Nischarin was used to design two peptides represented at the far COOH terminus of the protein. The peptides (EALCGRELPVELTGA-C and LDDGRRVRDLDRVL-C) were obtained from the University of North Carolina-Protein Core Laboratory. Both peptides were conjugated to keyhole limpet hemocyanin (Pierce Chemical Co.) and sent to Aves Laboratories for production of chicken pAbs. Anti-α5 cytoplasmic domain pAb was a gift from Richard Hynes (Massachusetts Institute of Technology, Cambridge, MA). Anti-myc mAbs and pAbs were purchased from Babco. pAbs to αv cytoplasmic domain were provided by Guido Tarone (University of Torino, Torino, Italy). Rat anti–mouse α5 mAb, and control rat IgG were purchased from PharMingen and Sigma-Aldrich, respectively. mAbs to vinculin and phosphotyrosine were purchased from Sigma-Aldrich and Upstate Biotechnology. Fluorescent phalloidin was bought from Sigma-Aldrich. A partially purified preparation of the human α5β1 integrin (Chemicon) was sometimes used as a control.
Binding To GST Fusion Proteins
GST–Nischarin fusion proteins expressed from pGEX vectors (Amersham Pharmacia Biotech) were prepared in a standard manner and bound to glutathione-Sepharose 4B beads for “pull down” experiments. CHO cells (clone B2α27), which overexpress the human α5 integrin subunit, were used as the source of integrins (Bauer et al. 1993). α5-deficient cells (CHO B2) were used as controls. CHO cells were lysed in a buffer containing nonionic detergent and protease inhibitors. The CHO lysate was added to the GST protein–containing beads, incubated for 1 h at 4°C, and washed four times with buffer. Bound CHO proteins were eluted by boiling in 2× SDS sample buffer and analyzed by Western blotting using anti-α5 cytoplasmic domain antibody.
CHO B2α27 and B2 cells were transiently transfected with myc vector, myc-Nischarin (434–581), or myc-Raf. After 48 h of transfection, cells were lysed in a 0.5% Triton X-100 buffer. These lysates were immunoprecipitated with anti-myc antibody, resolved by 7% SDS-PAGE, electrophoretically transferred to nylon membranes, and Western blotted with anti-α5 cytoplasmic domain antibody. In further studies with full-length Nischarin, cells were lysed in a buffer containing 0.1% Triton X-100 (Borowsky and Hynes 1998). In one set of experiments, lysates of Cos7 cells cotransfected with myc-Nischarin and α5, αv, or CD4, were immunoprecipitated with anti-myc and blotted with anti-α5 extracellular domain antibody (Transduction Laboratories), anti-αv, or anti-CD4 antibodies (Santa Cruz Biotechnology, Inc.). For mouse NB41A3 cells, endogenous α5 was immunoprecipitated with rat anti-α5 mAb and the immunoprecipitate was blotted for endogenous Nischarin using the chicken anti-Nischarin pAb described above.
Cell Migration Experiments
Wound-type cell migration experiments were performed as described previously (Bauer et al. 1992, Bauer et al. 1993), with minor modifications. In brief, 3T3 cells were cotransfected with 1 μg of β-galactosidase plasmid and various amounts of full-length myc-Nischarin construct and plated on gridded tissue culture dishes. After 48 h, the cell layer was scraped along the center of the dish with a sterile razor blade. After overnight incubation at 37°C in serum-containing medium, cells migrating into the scraped area were detected by staining for β-galactosidase. The percent of transfected cells that migrated across the line and into the denuded area was calculated by counting blue cells in several gridded fields from the unscraped area and in several fields from the scraped area. At least 50 migrant cells were counted for each condition. The ratio of migrant transfected cells to total transfected cells ×100 was taken as the percent migration.
Cell migration studies using Nischarin-transfected or control-transfected 3T3 cells or CHO cells were also performed using a transwell assay, according to a previously described procedure (Keely et al. 1997). The transfected cells were marked by use of a GFP expression plasmid. Fibronectin or other matrix proteins were coated on the underside of the transwell, the cells were plated on the upper surface, and the percent of Nischarin or control transfectants migrating across the 8-μm pore size membrane was determined by visual inspection in a fluorescence microscope after overnight incubation in BSA-containing medium. Transwell experiments were performed with wild-type 3T3 cells, 3T3 sublines overexpressing human α5 or α2 subunits (Aplin et al. 1999b), and CHO B2 cells lacking α5, as well as CHO B2a27, its α5 transfectant (Bauer et al. 1993).
Cos7 cells transfected with myc-Nischarin and untransfected Neuro 2A cells were subjected to subcellular fractionation, as described previously (Gu and Majerus 1996), with minor modifications. Cell lysates were centrifuged briefly at 1,500 rpm to remove nuclei and intact cells. The supernatant was further spun at 100,000 g for 30 min at 4°C; this supernatant was considered to be the cytosolic fraction. The pellet was solubilized in a 1% Triton X-100–containing solution and centrifuged at 100,000 g for 30 min; this supernatant was considered to be the membrane fraction. Membrane and cytosolic fractions were resolved by 8% SDS-PAGE, electrophoretically transferred onto a nylon membrane, and blotted with anti-Nischarin antibodies, as described above.
Immunofluorescence studies with antibodies to integrins or focal contact proteins were conducted according to procedures described previously (Burridge et al. 1992). Rat embryonic fibroblasts (REFs) were cotransfected with 1 μg of GFP plasmid and 2 μg of myc-Nischarin plasmid, or 1 μg of GFP alone, per well on six-well plates. After 48 h, cells were trypsinized and plated onto fibronectin-coated cover slips for 3 h in serum-containing medium. The cells were washed three times with cold PBS, fixed for 10 min in 0.37% formaldehyde, and permeabilized in 1%Triton X-100 for 5 min. Then, cells were washed several times and blocked in 2% BSA for 1 h at room temperature. Primary antibody incubation was done in a moist chamber overnight in a cold room. Anti-tubulin, anti-PY, anti-vinculin, and anti-vimentin antibodies were used at a dilution of 1:100. After rinsing in PBS, coverslips were incubated with an appropriate TRITC-conjugated secondary antibody for 1 h at room temperature. For actin staining, cover slips were incubated with TRITC-phalloidin (1:1,000) for 15 min.
In some cases, the subcellular distribution of Nischarin was evaluated using the full-length Nischarin–GFP chimera described above. This was transfected into 3T3 cells at a level of 2 μg per well (it should be noted that levels of expression of Nischarin–GFP chimeric protein were substantially lower than expression of myc-Nischarin protein when equivalent amounts of plasmid were transfected). After 48 h of transfection, cells were plated onto fibronectin coverslips, as described above, and incubated with antibodies to vinculin or integrins, and then with TRITC-conjugated secondary antibody, or with TRITC-phalloidin to visualize actin. In all cases, coverslips were observed on a ZEISS Axioscop fluorescence microscope using a 40× oil immersion objective. Images were recorded using a CCD camera and a computer with Metamorph image analysis software.
Rho GTPase Experiments
For studies on Rho-mediated signaling, NIH 3T3 cells were cotransfected with 1 μg of luciferase reporter under the control of the c-fos promoter (c-fos–Luc) (Hill et al. 1995), 3 μg of pAX142 vector, pAX142 Rac Q61L (Whitehead et al. 1988), or an activated MEK construct (pFC-MEK1; Stratagene) and various amounts of pcDNA myc-Nischarin or pcDNA vector, using Superfect. The pAX142 vectors were provided by Drs. I. Whitehead and C. Der (University of North Carolina-Chapel Hill). After 4 h of transfection, cells were washed with PBS, maintained in 0.5% serum for 24 h, and lysed in luciferase buffer, as described above. Additional experiments were done with commercial luciferase reporter systems (Stratagene) using either Rac-driven c-Jun transcriptional activation or protein kinase A–driven activation of the cyclic AMP–response element (CRE)-response element. In all transfections, DNA quantities were normalized with the pcDNA vector. Luciferase activity was measured by normalizing for total protein content or by coexpression of Renilla luciferase.
To study the effect of Rho-family GTPase on the cytoskeleton, 3T3 cells were transfected with a plasmid expressing an activated (Q61L) form of Rac and with a Nischarin plasmid or with a vector control. A small amount of a GFP-expressing plasmid was used to mark the transfectants. After 48 h, the actin filaments were stained with TRITC-phalloidin and the cells were observed by fluorescence microscopy, as described above.
Detection of a Novel α5-interacting Protein
We used a yeast two-hybrid screen to identify proteins that interact with the α5 cytoplasmic domain. A protein composed of the complete cytoplasmic tail of α5 fused with the DNA-binding domain of Lex A was expressed from the yeast plasmid pBTMα5. We searched mouse embryonic libraries for proteins that interact with the α5 cytoplasmic tail. Protein domains from the libraries were fused to the VP16-transactivating domain and expressed from the yeast plasmid pVP16. Cotransformants of pBTMα5 and pVP16 in the yeast L40 were screened for conversion to histidine prototrophy. Out of 1.7 × 107 transformants screened, 120 colonies were positive for histidine; of those, 45 were also positive for LacZ activation. To determine specificity, several other “baits” were tested for interaction with the α5-binding library protein(s). In particular, we tested for interactions with the cytoplasmic domains of the α2, αv, or β1 integrin subunits, or with the irrelevant protein lamin. As seen in Fig. 1 and Table, the α5 bait strongly interacted with the library protein and activated expression of the histidine and LacZ markers. The αv and α2 baits only weakly activated the histidine reporter, and were unable to activate LacZ. These data suggest that the α5-positive library protein may interact weakly with several integrin α subunits, but binds strongly to the α5 subunit.
Restriction enzyme analysis and DNA sequencing revealed that the insert in all positives tested was comprised of the identical 450 nucleotide sequence. This sequence comprises an ORF, but no start or stop codons. To find the full-length sequence of this cDNA, we screened a mouse brain library in the lambda zap II vector. We identified two overlapping cDNA fragments; and sequence analysis of these fragments revealed the complete ORF for this gene. This consists of 4,062 nucleotides and codes for a novel protein of 1,354 amino acids, with a predicted molecular weight of 148,053. As mentioned above, we have termed this protein Nischarin, a name which describes its effects on cell migration. The predicted amino acid sequence of Nischarin is shown in Fig. 2. The validity of this ORF as a protein-coding region is suggested by the presence of a well conserved Kozak sequence immediately 5′ of the ATG, the presence of a poly A tail in the 3′ untranslated region, the fact that the observed message size (5.5 kb, see below) is consistent with the cDNA size, and the fact that several mouse expressed sequence tags, which have the correct reading frame, overlap our predicted ORF region.
Homologies with Other Proteins
BLAST analysis indicates that Nischarin lacks significant homology with any known protein with a well-described function. A close human homologue of Nischarin protein has been reported in Genbank (sequence data available from GenBank/EMBL/DDBJ under accession no. AF082516). It has been suggested, based on limited evidence (Ivanov et al. 1998), that this protein is a neuroreceptor for imidazoline compounds. The human protein is very similar to Nischarin. Thus, there is 82% identity in a 792 NH2-terminal region that includes the 150 residue integrin-binding region originally identified by two-hybrid screening. There is also 84% identity in a 400 residue COOH-terminal region. However, there is less homology in the central proline-rich region (see below). The NH2-terminal region of Nischarin also displays a strong homology to a Caenorhabditis elegans protein (sequence data available from GenBank/EMBL/DDBJ under accession no. Z69383) of unknown function that resembles the regulatory subunits of protein phosphatases. In addition, there are two Drosophila homologues of the NH2-terminal region (AE003611 and AE003811). When BLAST searching is performed with the filter for low complexity regions inactivated, the central region of Nischarin, which has multiple repeats of proline, alanine, and glutamic acid residues, is found to resemble regions seen in neurofilament proteins (e.g., accession no. Z31012) and dermal proteins (accession nos. P17437 and X51394). Despite the existence of a highly proline-rich region, no consensus-binding sites for SH3 domains (Sparks et al. 1996) were identified in this region (however, see below). The integrin-binding region of Nischarin identified in the two-hybrid screen does not have any close homologues other than in its human direct counterpart. We have used several software programs to look for specific functional motifs that might be present in Nischarin. Aside from the few exceptions noted below, we were unable to find well-known protein functional motifs in Nischarin. There is a cytochrome p450 motif (FHADLRSCFA) 558–567 that may be indicative of an enzymatic function; there are two leucine zipper repeats 449–470 and 1,211–1,232 (LGADEDFLLEHIRILKVLWCFL and LGRGRGPLRPKTLLLTSAEIFL) that may be predictive of protein–protein interaction sites. There are also potential SH3 domain–binding motifs, two in the NH2-terminal region and two in the COOH-terminal region. Thus, the primary sequence of Nischarin offers few clues to the biological function of this protein.
Nischarin Binds Integrin α5 Subunit In Vitro and In Vivo
To confirm that the interaction between the α5 cytoplasmic tail and Nischarin detected by two-hybrid analysis also occurs in vitro, two GST–Nischarin constructs were made, GST–Nisch (435–582) and GST–Nisch (33–588). GST–Nisch (435–582) corresponds to the integrin-binding region of Nischarin identified in the two-hybrid system, whereas the second construct contains additional NH2-terminal residues. The Nischarin–GST fusion proteins were immobilized on a glutathione-agarose matrix and incubated with purified α5β1 protein, with a cell lysate of CHO B2α27 (human α5–transfected cells) or CHO B2 (α5-deficient cells). The proteins retained on the glutathione matrix were analyzed by SDS-PAGE and immunoblotting for α5. Consistent with the yeast data, the GST–Nischarin fusion proteins, but not GST alone, were able to interact with purified α5β1 or with α5β1 from a cell lysate (Fig. 3 A). These data indicate Nischarin binds the α5 integrin subunit in vitro.
Several types of coimmunoprecipitation experiments were performed to confirm that Nischarin interacts selectively with the α5 integrin subunit in mammalian cells. First, CHO B2α27 cells were transiently transfected with a construct expressing a truncated myc epitope–tagged segment of Nischarin (435–582), with myc-tagged Raf, or with myc vector alone. Cells were lysed in a buffer containing nonionic detergent, immunoprecipitated with anti-myc antibody, and blotted with anti-α5 or anti-myc antibodies. Western blotting with anti-α5 antibody indicated that α5 coimmunoprecipitated with anti-myc antibody only in cells transfected with myc-Nischarin (435–582), but not in cells transfected with myc-Raf or myc vector alone (Fig. 3 B). Expression of similar amounts of myc-Nischarin and myc-Raf was confirmed by Western blotting (data not shown). In experiments with CHO B2 cells, which lack α5, Nischarin did not coimmunoprecipitate a band in the α5 region. This indicates that myc-tagged Nischarin (435–582), but not an irrelevant myc-tagged protein, can bind α5β1 integrin.
Further coimmunoprecipitation experiments were done using full-length Nischarin. To show that Nischarin interacts preferentially with α5 integrins and not with other transmembrane proteins, coimmunoprecipitations were done with CD4, another protein having a single transmembrane domain. Cos 7 cells were transiently cotransfected with myc-Nischarin and pcDNA-α5 or myc-Nischarin and pcDNA-CD4. These cells were lysed and myc immunoprecipitates were processed for blotting with anti-α5 or anti-CD4 antibodies. As shown in Fig. 3 C, myc-Nischarin specifically immunoprecipitated α5, but not CD4.
To further examine the alpha subunit selectivity of Nischarin, Cos 7 cells were cotransfected with myc-Nischarin and plasmids expressing the human α5 or αv subunits. These cells were lysed and myc immunoprecipitates were processed for blotting with anti-α5 or anti-αv antibodies. As seen in Fig. 3 D, Nischarin coimmunoprecipitated substantial amounts of α5, but barely detectable αv subunits. Thus, the strong preference of Nischarin for the α5 subunit, which was detected in the two-hybrid analysis, seems to be borne out in the cellular setting.
Coimmunoprecipitation experiments were also done for endogenous Nischarin and endogenous α5 subunit. NB41A3 cells, a mouse neuronal-derived line, were lysed and immunoprecipitates formed using rat anti–mouse α5 mAb, or rat IgG as a control. The immunoprecipitates were Western blotted using chicken anti-Nischarin pAb. As seen in Fig. 3 E, a band for Nischarin was detected in the α5 immunoprecipitate, but not in the control. It was not possible to examine other integrin α–subunit imunoprecipitates, since only low levels of expression were found in NB41A3 cells for other integrins for which anti–mouse mAb are available.
The experiments shown in Fig. 3 A–E indicate: (a) full-length Nischarin or truncated Nischarin (435–582) can bind to the α5 integrin subunit, (b) other proteins (e.g., myc-Raf) do not bind α5 under the conditions used for immunoprecipitation, and (c) full-length Nischarin does not bind to coexpressed irrelevant proteins. Furthermore, Nischarin seems to prefer the α5 subunit compared with other tested α subunits. These findings demonstrate a selective interaction between Nischarin and the α5 integrin subunit in mammalian cells. This interaction is able to occur under physiological conditions, as indicated by the coimmunoprecipitation of endogenous α5 and Nischarin.
Tissue Distribution of Nischarin
To determine the tissue distribution of Nischarin mRNA, a mouse multiple tissue Northern blot was hybridized with a PCR probe from the Nischarin integrin-binding region. A single mRNA of ∼5.5 kb was detected. In further analysis (not shown), three probes from different regions of the cDNA (extreme 5′ end, integrin-binding region, and extreme 3′ end of the ORF) were used; all three probes detected the same message. Nischarin mRNA expression was highest in brain and kidney, expression levels were lower in heart, liver, lung, and skeletal muscle, whereas no expression was seen in spleen and testis (Fig. 4 A). To address the expression of Nischarin in various cell types, we performed Northern blot analysis on several cell lines. Nischarin message was present in various rodent epithelial, fibroblast, and neuronal cell lines, though higher levels tended to be present in neuronal cells (Fig. 4 B and data not shown). Analysis of a mouse embryo RNA blot indicated the presence of Nischarin message as early as 7 d of development (Fig. 4 C).
Expression of Nischarin in Cells and Its Subcellular Localization
The expression of transfected myc-tagged full-length Nischarin and expression of endogenous Nischarin were evaluated by Western blotting of cell lysates with anti-myc antibodies, as well as chicken pAbs directed against polypeptides from the COOH-terminal region of the predicted Nischarin ORF. In Cos7 cells transfected with myc-tagged full-length Nischarin, both antibodies recognized the same band of ∼190 kD (Fig. 5A and Fig. B); this is somewhat larger than the predicted molecular weight of expressed Nischarin, but the reason for this is unclear. The expression of endogenous Nischarin was evaluated in detergent lysates of several cell lines by Western blotting with chicken anti-Nischarin antibodies (Fig. 5 A). In rat intestinal epithelial (RIE) cells (Oldham et al. 1996), a protein band of ∼190 kD was detected; this band comigrated with the immunoreactive band in Nischarin-transfected Cos7 cells. In NIH 3T3 cells, the chicken anti-Nischarin antibody detected a very weak band of 190 kD (data not shown). In several mouse neuronal cell lines (NIE 119, NB41A3, BC(3)H1, and Neuro 2A), an immunoreactive band of somewhat higher apparent molecular weight was detected (Fig. 5 A, and data not shown). The immunoreactivities of the bands detected in Nischarin-transfected Cos7 cells, or in nontransfected RIE cells, were competed out upon addition of the Nischarin COOH-region peptide used for chicken immunization (Fig. 5 C). Thus, the bands detected by the chicken antibody in the various rodent cell lines or in the transfected Cos7 cells are very likely to represent Nischarin. The detection of immunoreactive bands of differing apparent molecular weights in the various cell types examined suggests differences in splicing or posttranslational modifications. The fact that only a single Nischarin message is seen by Northern analysis seems to militate against the possibility of splicing differences. However, alternate splicing of a short region of the mRNA might not be readily detected. The data of Fig. 5A–C, indicate that Nischarin is widely expressed in various rodent cell lines, with higher levels found in cells of neuronal origin.
Both biochemical and fluorescence microscopy techniques were used to evaluate the subcellular localization of Nischarin. Overall distribution of Nischarin was assayed by subcellular fractionation. Cos7 cells were transfected with plasmids expressing myc-tagged full-length Nischarin. These cells, as well as untransfected Neuro 2A cells, were fractionated as described in Materials and Methods. Fractions containing membranes or 100,000 g supernatant were prepared, resolved by SDS-PAGE, and subjected to Western blotting with chicken anti-Nischarin antibodies (Fig. 6 A). Both transfected and endogenously expressed Nischarin were primarily found in the 100,000 g supernatant fraction, indicating that Nischarin is a soluble rather than a transmembrane protein.
We have used the Nischarin–GFP chimera described in Materials and Methods and fluorescence microscopy to obtain further information on the subcellular localization of Nischarin (Fig. 6 B). NIH 3T3 cells expressing the Nischarin–GFP chimera were counterstained with antibodies to the focal contact protein vinculin (Fig. 6 B). Nischarin–GFP did not enter the nucleus and was primarily found diffusely distributed in the cytosol. However, a greater concentration of Nischarin was seen in the perinuclear region partially associated with punctate structures that may be endomembrane vesicles. There was no evidence of Nischarin–GFP accumulation at vinculin-rich focal adhesion sites. Staining of focal contacts and fibrillar structures with anti-α5 antibody also failed to reveal any obvious colocalization with Nischarin (not shown). Similar studies using an antiphosphotyrosine antibody to detect focal contact sites yielded the same result (data not shown). Thus, Nischarin is found primarily in the cytosol, with some concentration in the perinuclear area, but does not concentrate at “classical” focal adhesion sites.
The observations in Fig. 6 indicate that Nischarin is largely a cytosolic protein, and its overall distribution does not coincide with focal adhesion structures. Since Nischarin can clearly associate with the α5 subunit, as demonstrated by coimmunoprecipitation, this may indicate that only a small fraction of the total cellular pool of Nischarin is associated with the α5β1 integrin at any given time.
Nischarin Inhibits Cell Migration
To investigate the biological role(s) of Nischarin, we focused on the finding that Nischarin seems to interact most strongly with the α5 subunit and on the knowledge that α5β1 plays an important role in cell motility. The effect of Nicharin on cell migration was initially evaluated using a monolayer “wounding” assay (Bauer et al. 1993). NIH 3T3 cells were cotransfected with various amounts of a plasmid expressing full-length Nischarin and with a β-galactosidase marker plasmid. After transfection and recovery, the transfected cell monolayers were scraped with a razor blade. The migration of transfected (β-galactosidase positive) cells across the wound boundary was quantitated, as described in Materials and Methods. As seen in Fig. 7 A, cells overexpressing Nischarin showed significant inhibition of migration compared with cells transfected with β-galactosidase plasmid alone. Increasing the dose of Nischarin plasmid resulted in a progressive decrease in migration. This is unlikely to be due to toxicity since increasing the amount of Nischarin transfected did not reduce the survival of the transfectants, as judged by the fraction of β-galactosidase–positive cells in the total cell population.
Nischarin effects on cell movement were also evaluated using a transwell assay (Keely et al. 1997) where the cells migrate across a membrane containing 8 μm pores. These studies also undertook to examine whether the effect of Nischarin on migration displayed any specificity in terms of the integrin involved in migration. One set of experiments used CHO sublines differing in α5β1 expression (Bauer et al. 1993). CHO B2 cells lack α5β1 and do not adhere to fibronectin, but express other integrins that allow adhesion and migration on vitronectin or other matrix proteins. CHO B2a27 cells derive from B2, but have been stably transfected with human α5; these cells migrate on fibronectin in a completely α5β1-dependent manner. The transwell membranes for these assays were coated with either fibronectin or vitronectin. As seen in Fig. 7 B, transfection of B2a27 cells with Nischarin led to a major reduction of their migration on fibronectin-coated membranes, but only a modest reduction on vitronectin-coated membranes. Furthermore, migration of the B2 cells on vitronectin-coated membranes was not at all affected by expression of Nischarin. Treatment of cells with cytochalasin D completely abolished migration of either cell type.
In another set of experiments, transwell migration assays were performed with 3T3 cells stably transfected with human α5 or α2 subunits (Aplin et al. 1999b). The adhesion, and presumably migration, of these cells is strongly influenced by the transfected integrin subunit. As seen in Fig. 7 C, transfection with Nischarin markedly inhibited migration of the α5-overexpressing cells on fibronectin, but had little effect on the migration of the α2-overexpressing cells on collagen.
Thus, Nischarin overexpression can profoundly inhibit cell migration. This effect displays substantial integrin subunit specificity and is much more dramatic for migration on fibronectin mediated by α5β1 than for migration on other matrix proteins mediated by other integrins.
Effects of Nischarin on the Cytoskeleton
Since overexpression of Nischarin resulted in substantial alterations in cell migration, we wished to determine if this was accompanied by changes in the organization of the cytoskeleton. REF cells were transfected with plasmids expressing full-length Nischarin and GFP. Cells were probed with the actin-binding reagent phalloidin or were immunostained for phosphotyrosine, vinculin, tubulin, or vimentin. Phalloidin staining (Fig. 8 A) indicated that many of the transfected REF cells had a unique phenotype, with a more or less circular shape and having actin filaments arranged in “basket” structures around the periphery, rather than as the linear stress fibers commonly seen in adherent fibroblasts. Although this phenotype was not universal, ∼60% of the REF cells cotransfected with Nischarin and GFP showed the basket-like actin structures. In contrast, only a few percent of the control GFP transfectants had this phenotype. We next looked for the effects of Nischarin on focal adhesions by staining for vinculin and phosphotyrosine. As seen in Fig. 8 B, vinculin-containing focal contacts and phosphotyrosine in focal contacts were somewhat reduced in REF cells transfected with Nischarin compared with control cells. Staining, with anti-tubulin or anti-vimentin antibodies, suggested that Nischarin expression had little effect on the organization of microtubules or intermediate filaments in REF cells (not shown). Similar effects were observed in WI-38 cells, another well spread cell line (not shown). However, the “basket” phenotype was not apparent in Nischarin-transfected NIH 3T3 cells or in Cos7 cells, both of which are less well spread. The highly organized actin filaments of REFs may allow easier visualization of the relatively subtle effects of Nischarin on cytoskeletal architecture.
Effects of Nischarin on Rac GTPase-mediated Signaling
Since the organization of the actin cytoskeleton, as well as cell motility, are strongly influenced by the activity of the Rac GTPase (Mackay and Hall 1998), we decided to evaluate whether Nischarin might affect Rac-mediated functions. As an initial test of the effects of Nischarin on signaling by Rac, NIH 3T3 cells were transfected with a reporter plasmid that uses the c-fos promoter to drive luciferase expression. Cells were then cotransfected with plasmids expressing activated versions of Rac, as well as with various amounts of a plasmid expressing full-length Nischarin. As seen in Fig. 9 A, transfection with Nischarin plasmid did not affect levels of expression of cotransfected Rac. Activated Rac strongly stimulated luciferase expression (Fig. 9 B), most likely through well-known effects on transcription factors that recognize the serum response element (SRE) in the c-fos promoter (Hill et al. 1995). However, coexpression of Nischarin blocked Rac-mediated stimulation of the c-fos–Luc reporter in a dose-dependent manner (similar effects were observed with activated CDC42, whereas much more modest effects of Nischarin were observed with Rho, not shown). This inhibition was not due to nonspecific effects on transcription or translation, since stimulation of c-fos–Luc by a plasmid expressing constitutively active MEK was only weakly affected by coexpression of Nischarin (Fig. 9 C). Thus, Nischarin strongly inhibits signaling mediated by Rac, but is less effective in blocking signaling mediated by an effector in the Erk/MAPK cascade. The amount of Nischarin plasmid required to strongly block Rac-induced activation of the c-fos reporter correlates well with the amount that produces a dramatic reduction in cell migration (compare Fig. 7 and Fig. 9). A similar strong inhibition of reporter gene expression by Nischarin was also observed using a commercial system that detects Rac-driven activation of c-Jun. Furthermore, Nischarin only modestly inhibited expression of a reporter gene that is responsive to protein kinase A–mediated activation of a CRE element (Reddig, P., and R.L. Juliano, unpublished observations). Thus, overexpression of Nischarin seems to preferentially affect signaling through Rac-driven pathways rather than other well-known signaling pathways.
To further evaluate the possible interplay between Nischarin and the Rac GTPase, we examined the effect on Nischarin on a characteristic cytoskeletal function of Rac, namely the enhanced formation of lamellipodia (Mackay and Hall 1998). NIH 3T3 cells were transfected with a plasmid expressing an activated form of Rac (Q61L), and, in some cases, were cotransfected with full-length Nischarin. A small amount of a GFP plasmid was used to mark the transfected cells. As seen in Fig. 10 A, expression of activated Rac produced the expected enhancement of lamellipodia formation, with >80% of the transfectants showing large and distinct areas of membrane ruffling. When cells were cotransfected with active Rac and Nischarin there was a strong inhibition of the Rac effect, with only ∼35–40% of the cells displaying large lamellipodia when higher doses of Nischarin were transfected (Fig. 10 C). Some of the Rac plus Nischarin–transfected cells resembled untransfected 3T3 cells (Fig. 10 B). However, various cells at each dose of Nischarin displayed intermediate phenotypes with partial ruffling and incomplete lamellipodia (not shown). Thus, overexpression of Nischarin can inhibit lamellipodia formation, one of the most characteristic effects of Rac on the cytoskeleton and associated with cell movement. These findings suggest that Nischarin inhibits cell migration, at least in part, through its actions on pathways regulated by the Rac GTPase.
Recent studies suggest that individual integrin α/β heterodimers can play unique roles in the regulation of cell migration, growth, survival, and differentiation (Pozzi et al. 1998; Farrelly et al. 1999; Liu et al. 1999; Lochter et al. 1999; Sastry et al. 1999; Lee and Juliano 2000). This may come about via specific interactions between the cytoplasmic domains of individual integrins and intracellular proteins involved in signal transduction or other aspects of cell regulation. The α5β1 integrin is particularly interesting in this regard, since it has been implicated in the control of both cell growth and programmed cell death (Varner et al. 1995; Zhang et al. 1995; O'Brien et al. 1996; Sastry et al. 1999; Lee and Juliano 2000). Here, we have reported the identification and characterization of mouse Nischarin, a soluble intracellular protein that is capable of interacting with the cytoplasmic domain of the integrin α5 subunit. Using both two-hybrid analysis and coimmunoprecipitation of expressed proteins in cells, Nischarin was found to interact with the α5 subunit much more strongly than with the other two α subunits tested. Furthermore, immunoprecipitation of endogenous α5β1 from a mouse neuronal cell line resulted in coimmunoprecipitation of endogenous Nischarin. Thus, current evidence suggests that Nischarin interacts preferentially with the α5 subunit, and this interaction can occur under physiological conditions. However, we cannot rule out the possibility that Nischarin may interact with other examples of the many known α subunits.
Nischarin bears limited resemblance to identified proteins with a well-known functions. Only two close homologues of Nischarin have been reported in the DNA data bases. The human homologue of Nischarin has been described as a putative imidazoline receptor (Ivanov et al. 1998) (sequence data available from GenBank/EMBL/DDBJ under accession no. AF082516). Imidazolines are thought to be neurotransmitters and, thus, their receptors would presumably be transmembrane proteins. However, our results clearly show that endogenous Nischarin is primarily a soluble protein and most likely rules out a role as a transmembrane neurotransmitter receptor. A second close homologue of the NH2-terminal region of Nischarin has been reported from the C. elegans genome project (accession no. Z69383) and two similar proteins are found in Drosophila. However, the function of these proteins is completely unknown. The human protein contains a segment that is highly homologous to the integrin-binding region of mouse Nischarin that we detected by two-hybrid screening. There are also substantial homologies between the central proline-rich region of Nischarin and portions of several neurofilament proteins, but the functional significance of this is unclear. Examination of the primary sequence of Nischarin, using programs that search for common protein structural or functional motifs, yielded few clues as to the biological role of this molecule.
Nischarin is expressed in many cell types and is found both in the adult mouse and in the developing embryo. Higher amounts of Nischarin are found in neuronal-derived cell lines than in epithelial cells or fibroblasts, but some Nischarin is expressed in all of these cell types. Western blotting of various cell lines for endogenous Nischarin revealed proteins of two distinct sizes; in some cells, an ∼190 kD form is found that comigrates with expressed Nischarin. However, in neuronal cells, a larger form of the protein is seen. The basis for this difference is currently unknown, but may reflect cell-type specific alternative splicing, use of alternate start codons, or posttranslational modification.
Immunofluorescence and biochemical fractionation studies indicate that Nischarin is largely a soluble cytosolic protein. It is clear from fluorescence microscopy studies that Nischarin is not concentrated in vinculin-rich focal contacts. Although, one might expect a protein that interacts with α5β1 integrin to be localized to focal contacts, this is not always the case. For example, members of the TM4 family of proteins clearly interact specifically with certain integrins, but TM4 proteins are not found in “classic” focal contacts (Porter and Hogg 1998; Berditchevski and Odintsova 1999). This is also true of calveolin, which has been found to interact with certain integrins (Wary et al. 1996). At present it is unknown whether there is any physiological regulation of the association between the α5β1 integrin and Nischarin that might affect its subcellular distribution.
It seems clear that Nischarin can selectively bind to the cytoplasmic domain of the α5 integrin subunit, based both upon two-hybrid analysis and coimmunoprecipitation of full-length Nischarin with native α5β1 in mammalian cells. However, at any given time, only a small fraction of the total Nischarin in a cell is likely to be bound to the integrin, since most Nischarin is found in the cytosolic fraction. This type of situation is often seen in signaling pathways, where only a minority of a cytosolic effector molecule associates with its membrane-bound partner molecule. The well-known association between Ras and Raf-1 is a good example, where Raf is primarily found in the cytosol, despite its clear ability to interact with membrane-bound Ras (Wartmann et al. 1997; Campbell et al. 1998).
Overexpression of full-length Nischarin results in major changes in cell behavior and also affects cytoskeletal organization. The most dramatic aspect is the profound inhibition of cell migration caused by Nischarin. At this point, it is unclear whether the reduced cell migration observed in the “wounding” and transwell assays used here is due to a reduction in innate motility or to an impairment of directional movement (Gu et al. 1999). This will be an important issue for further investigation. The effects of Nischarin on cell migration are quite α-subunit selective. Thus, α5β1-dependent migration on fibronectin is inhibited far more strongly than migration on other substrata mediated by other integrins.
The overexpression of Nischarin in certain fibroblasts leads to substantial changes in focal contact and actin filament organization. Thus, Nischarin-transfected REF cells display fewer linear stress fibers and a reduction in mature, vinculin-positive focal contacts. Instead, the actin filaments form unusual “basket” structures around the cell periphery. These effects are clearly seen in well spread fibroblasts such as REF and WI-38 cells, but are much less apparent in cell lines such as 3T3 and Cos. The dramatic effects of Nischarin on cell migration and actin filament organization described here may be due, at least in part, to the fact that the transfected molecule is expressed at substantially higher levels than the normal amount of endogenous Nischarin. However, even the somewhat skewed effects triggered by overexpression may provide important clues in eventually ascertaining the physiological role of Nischarin.
The observed effects of Nischarin on cell motility and cytoskeletal organization suggested that Nischarin might impact the pathways used by some Rho-family GTPases to regulate individual pools of actin filaments (Mackay and Hall 1998). Thus, it was satisfying to find that overexpression of Nischarin strongly blocked the ability of active forms of Rac to drive reporter gene expression from the serum-response element of the c-fos promoter. Failure to strongly block MEK-induced activation of this same promoter indicates that Nischarin acts preferentially on Rac-mediated events rather than other signaling cascades. The theme that Nischarin can block Rac functions clearly associated with cytoskeletal organization and cell motility was extended by the observation that Nischarin can inhibit or reverse the well-known action of Rac in promoting lamellipodia formation. Thus, it seems likely that Nischarin can have a substantial impact on the signaling and cytoskeletal functions of Rac. There is little in the primary sequence of Nischarin to suggest a mechanism for its influence on Rac GTPase pathways, that is, no obvious homologies to exchange factor, GAP, or GDI domains (Sasaki and Takai 1998) are apparent.
Lately, the mechanistic basis underlying integrin-mediated cell movement has received a good deal of attention. It is clear that FAK is a key regulator of cell migration in most cells (Ilic et al. 1995; Cary et al. 1998; Sieg et al. 2000). Overexpression of the PTEN tumor suppressor, a dual specificity phosphatase, results in the dephosphorylation of FAK and a reduction in directional cell motility (Tamura et al. 1998; Gu et al. 1999). The focal contact protein p130Cas is tyrosine phosphorylated by FAK; subsequently, the adaptor protein Crk can bind phosphotyrosyl sites on Cas. The Cas–Crk complex has been implicated in the control of cell migration, whereas the Rac GTPase seems to be a downstream mediator of the Cas/Crk pathway (Cary et al. 1998; Klemke et al. 1998). In epithelial cells, an important connection has been made between cell motility and a signaling pathway involving phosphatidyl inositol-3-kinase and the Rac and CDC42 GTPases (Keely et al. 1997; Shaw et al. 1997). Interestingly, integrin-mediated cell adhesion has been shown to directly activate Rac and CDC42 (Price et al. 1998), whereas Ras, CDC42, Rac, and Rho have all been implicated in cooperative regulation of cell movement (Clark and Brugge 1995; Nobes and Hall 1999). Thus, though many of the molecular details remain to be determined, it seems clear that a pathway (perhaps branched) involving integrins, FAK, Cas/Crk, phosphatidyl inositol-3-kinase, and Rho-family GTPases positively regulates cell motility. However, other than the role of PTEN in FAK dephosphorylation, there has been little evidence, to date, of physiological inhibitors of cell motility. In this context, Nischarin may play an important role by negatively impacting cell motility pathways controlled by Rac.
In summary, we have identified and characterized a novel protein that we have named Nischarin. This protein can bind selectively to the cytoplasmic tail of the integrin α5 subunit. Overexpression of Nischarin has potent effects in terms of retarding cell migration, and it acts preferentially on migration mediated by the α5β1 integrin. Nischarin overexpression also influences actin filament organization in some cell types. These effects may be mediated through Nischarin's selective action on pathways regulated by the Rac GTPase. Thus, one important aspect of Nischarin's biological role may be to counterbalance the effects of Rac in promoting directed cell movement.
We would like to thank Mike Fisher for help with tissue culture work, Dr. Peter Reddig for the GFP–Nischarin construct and experiments with reporter gene assays, and Dr. Andrew Aplin for the α5- and α2-positive 3T3 cell lines. We also thank Drs. U. Naik and T. Griffith for their suggestions concerning yeast two-hybrid techniques, and Dr. K. Burridge for his critical reading of the manuscript. In addition, the authors thank Brenda Asam for outstanding secretarial assistance.
This work was supported by a grant from the National Institutes of Health to R.L. Juliano (CA 74966).
Abbreviations used in this paper: FAK, focal adhesion kinase; GFP, green fluorescent protein; ORF, open reading frame; REF, rat embryonic fibroblast; RIE, rat intestinal epithelial.