|Home | About | Journals | Submit | Contact Us | Français|
Herein, we report the first evidence that c-SRC is required for retinoic acid (RA) receptor (RAR) signaling, an observation that suggests a new paradigm for this family of nuclear hormone receptors. We observed that CSK negatively regulates RAR functions required for neuritogenic differentiation. CSK overexpression inhibited RA-mediated neurite outgrowth, a result which correlated with the inhibition of the SFK c-SRC. Consistent with an extranuclear effect of CSK on RAR signaling and neurite outgrowth, CSK overexpression blocked the downstream activation of RAC1. The conversion of GDP-RAC1 to GTP-RAC1 parallels the activation of c-SRC as early as 15 min following all-trans-retinoic acid treatment in LA-N-5 cells. The cytoplasmic colocalization of c-SRC and RARγ was confirmed by immunofluorescence staining and confocal microscopy. A direct and ligand-dependent binding of RAR with SRC was observed by surface plasmon resonance, and coimmunoprecipitation studies confirmed the in vivo binding of RARγ to c-SRC. Deletion of a proline-rich domain within RARγ abrogated this interaction in vivo. CSK blocked the RAR-RA-dependent activation of SRC and neurite outgrowth in LA-N-5 cells. The results suggest that transcriptional signaling events mediated by RA-RAR are necessary but not sufficient to mediate complex differentiation in neuronal cells. We have elucidated a nongenomic extranuclear signal mediated by the RAR-SRC interaction that is negatively regulated by CSK and is required for RA-induced neuronal differentiation.
Retinoic acid (RA) is an active form of vitamin A. As a morphogen, it induces cellular differentiation in various cell types. RA or its derivatives have been shown to cause profound morphological differentiation in embryonic stem cells, embryonal carcinoma (EC) cell lines, and neuroblastoma (NB) cell lines (3, 6, 65). RA action is initiated through its binding to two members of the class II family of nuclear hormone receptors, RA receptors (RARs) (α, β, and γ) and retinoid X receptors (α, β, and γ) (2, 43, 44). RA associates with nuclear RARs to form a heterodimeric complex that binds to the RA-responsive element (RARE) in the promoter regions of the target genes (43). Characteristic effects of RA in tumor cells (growth inhibition and differentiation) are known to be mediated through the transcriptional activation of its signature genes (classical genomic effect) (34, 56). Other modes of RA action in malignant cells include the inhibition of the AP-1 protein (18, 33), the inhibition of c-Jun NH2-terminal kinase (38), the regulation of histone acetylation (53), the expression of transforming growth factor 2 (TGF-2) and insulin-like growth factor binding protein 3 IGFBP-3 (27), and the upregulation of the PEPCK gene (39). Although much work has been done on the effects of RA on gene expression, little is known concerning the potential for nongenomic extranuclear cellular signaling downstream of RAR engagement.
NB is the most common malignant extracranial solid tumor diagnosed in children and is responsible for 15% of pediatric cancer deaths (39, 45, 63, 70). As an embryonal tumor (12) of neural crest origin, the NB tumor consists of typically undifferentiated neuroectodermal cells (63) that have essentially lost their differentiation cues. NB cell lines continue to serve as a useful model for neuritogenic differentiation. Several studies have reported an improved prognosis for this disease following treatment of high-risk NB patients with retinoids (14). Moreover, RA is clinically one of the most effective inducers of differentiation in NB, and retinoids are routinely used in the treatment of high-risk NB (2, 67). Profound neuritogenesis observed in NB cell lines following RA administration is due to the activation of endogenous differentiation signaling and indicates the relevance of RA in this process. We and others reported an in vitro induction of postmitotic phenotypes in human NB cells lines following all-trans-RA (ATRA) administration (28-30, 59, 66). It is generally held that RARs function as DNA binding transcription factors to induce NB differentiation. This aspect of RAR function has been the major focus of studies devoted to the study of RA-RAR function in neuronal cells. The possibility that other functions may exist for the RAR protein, separate from the DNA binding domain, has not been explored to date. Herein, we have explored evidence that another domain within RAR may functionally interact with tyrosine kinases and play a critical role in neuronal differentiation.
RA-RAR signaling involves both the induction of genes required for neuronal differentiation and dramatic changes in cell shape and function normally ascribed to the posttranslational modification of proteins, e.g., phosphorylation, tubulin acetylation, actin polymerization, etc. Therefore, we hypothesized that there might be a more direct mechanism by which the RARs may regulate and/or coordinate nuclear and cytoplasmic events. One such modification would be the direct interaction with extranuclear protein kinase activities. This led to our investigation of the role for the src family of kinases (SFKs) and CSK in RARγ signaling. The SFKs are one of the oldest, largest (comprised of 11 members in humans, of which c-SRC, FYN, and YES are ubiquitously expressed), and most studied family of nonreceptor protein tyrosine kinases (8, 47, 68, 73). The activation of SFKs is known to occur by the “domain displacement” mechanism involving their SH2 and SH3 domains (9, 16, 46, 61). The catalytic activity of SFKs is tightly regulated by the state of phosphorylation of their conserved C-terminal regulatory tyrosine (Y527 in c-SRC) residue (20, 57, 60) by a family of nonreceptor protein tyrosine kinases comprised of C-terminal SRC kinase, CSK, and CSK-homologous kinase (19, 22, 41, 60, 72, 75). Phosphorylation by CSK leads to the conformational changes in the SFK molecule through intramolecular contacts involving the SH2 domain (72). Although enzymatic regulation of SFKs by CSK is well documented (41), the role of CSK in neuronal differentiation and/or nuclear hormone receptor signaling has not been studied.
Our data provide evidence for a direct and functionally relevant connection between RARγ, c-SRC, and CSK. Our results suggest a mechanism by which RARs might coordinate their interaction with cytoplasmic and membrane extranuclear events linked to the process of neuronal differentiation by directly binding to and activating protein tyrosine kinases en route to the nucleus, where they mediate transcription. Our results demonstrate that (i) the inhibition of SFKs by CSK or PP1 inhibits RA-induced neurite outgrowth in NB cell lines, (ii) RARγ binds to and catalytically activates c-SRC in an RA-dependent manner, (iii) CSK overexpression in LA-N-5 cells blocks the activation of c-SRC and RA-induced activation of the small GTPase RAC1, and (iv) a search of the RARγ amino acid sequence identified a highly conserved proline-rich region in the N terminus as being a potential binding site for the SFK-SH3 domain. Taken together, these data suggest that ligand-dependent signaling of the RARγ involves c-SRC and that SRC kinase is necessary for RA-induced neuritogenesis of NB cells. The results suggest a paradigm by which nuclear hormone receptors integrate membrane/cytoplasmic events in concert with nuclear transcriptional effects to orchestrate the complex differentiation program required for neuritogenesis.
ATRA, 9-cis-RA, 13-cis-RA, and monoclonal antibody against human β-actin were obtained from Sigma-Aldrich (St. Louis, MO). A pan-SFK kinase inhibitor, PP1, was purchased from Biomole (Plymouth Meeting, PA). Rabbit polyclonal antibodies against SRC (SC-19), FYN (SC-16), and YES (SC-14) were purchased from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA), and used for immunoprecipitation (IP), kinase assays, and Western blot analyses. Horseradish peroxidase-tagged anti-rabbit immunoglobulin G (IgG) and anti-mouse IgG were obtained from Amersham Biosciences (Buckinghamshire, England). Goat anti-mouse and anti-rabbit IgG (heavy plus light chains)-AP (human adsorbed) were obtained from Southern Biotechnology, Inc. (Birmingham, AL). Recombinant c-SRC and RARγ proteins were procured from Upstate Biotechnology (Lake Placid, NY) and Santa Cruz Biotechnology, Inc. (Santa Cruz, CA), respectively. This full-length recombinant RARγ of human origin (amino acids 1 to 454) is expressed in Escherichia coli as a 75-kDa tagged fusion protein. The product RARγ (corresponding to amino acids 1 to 454) was purified from bacterial lysates by glutathione agarose affinity chromatography (as specified by Santa Cruz Biotechnology, Inc.). Commercially obtained recombinant SRC (p60c-src) is an approximately 60-kDa protein that is expressed in Sf9 insect cells by recombinant baculovirus containing the human c-src gene. The protein is purified by sequential chromatography on hydroxyapatite (HA) and affinity columns (as specified by Upstate Biotechnology). PAK-1 PBD (RAC1 assay reagent, agarose for the pull-down of the activated RAC1) and monoclonal RAC1 antibody were obtained from Upstate Biotechnology (Lake Placid, NY). Rabbit antiserum (4301.3) raised against a CSK peptide fragment (31) was used for immunoblotting (1:1,000). Pansorbin was purchased from Calbiochem (La Jolla, CA). Nitroblue tetrazolium, 5-bromo-4-chloro-3-indolylphosphate (BCIP) p-toluidine salt, aprotinin, and bovine serum albumin (BSA) were obtained from Sigma (St Louis, MO). Geneticin (G418 sulfate) was procured from the Invitrogen Corporation (Carlsbad, CA). A protein assay kit was obtained from Bio-Rad (Hercules, CA). A chemiluminescence kit was obtained from Amersham Biosciences (Buckinghamshire, England). For immunofluorescence studies, rabbit polyclonal anti-RARγ (C-19) and mouse monoclonal anti-c-SRC (H-12) antibodies from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA), were used. For fluorescence visualization, rhodamine-conjugated goat anti-rabbit and fluorescein isothiocyanate-conjugated goat anti-mouse IgGs from Molecular Probes, Inc. (Eugene, OR), were used as secondary antibodies against RARγ and c-SRC primary antibodies, respectively. DAPI (4′,6′-diamidino-2-phenylindole) for counterstaining was also purchased from Molecular Probes, Inc. (Eugene, OR). SRC, FYN, and YES immunokinase assays were done using an SRC assay kit (catalog no. 17-131) from Upstate Biotechnology (Lake Placid, NY). Radioactive [γ-32P]ATP (specific activity of 3,000 Ci/mmol) was purchased from Perkin-Elmer Life and Analytical Sciences (Boston, MA). Affi-Gel 15 (activated affinity medium) was bought from Bio-Rad Laboratories (Hercules, CA). Lipofectamine 2000 reagent was procured from Invitrogen Corporation (Life Technologies, Carlsbad, CA). Protein G-agarose Fast Flow beads were purchased from Upstate Biotechnology (Lake Placid, NY). Anti-HA monoclonal antibody (HA.11, clone 16B12) and anti-HA polyclonal antibody (Y-11) were obtained from Covance, the Development Service Company (Berkeley, CA), and Santa Cruz Biotechnology, Inc. (Santa Cruz, CA), respectively. Anti-FLAG antibody (anti-FLAG-M2 monoclonal antibody peroxidase conjugate) and anti-FLAG polyclonal antibody were obtained from Sigma-Aldrich (St. Louis, MO). Protease inhibitor cocktail tablets were procured from Roche Diagnostics (Mannheim, Germany). Mutagenesis was carried out using a Stratagene (La Jolla, CA) QuikChange II XL site-directed mutagenesis kit.
All plasmid constructions were prepared according to standard procedures. The sequences and orientations of inserted DNA fragments in plasmid constructs were verified by restriction enzyme analysis and automated standard DNA sequencing. The wild-type human RARγ1 (kindly provided by Ron Evans) gene was amplified by PCR using two primers, 5′-GATCGCGGATCCATGTACCCATACGATGTTCCAGATTACGCTCTTGCGGCCACCAATAAGGAGCGACTC-3′ and 5′-CTAGCGGAATCTCAGGCTGGGGACTTCAGGC-3′ (with a 2× FLAG tag at the N terminus), and then subcloned into plasmid pcDNA3.1 between BamHI and EcoRI sites. Deletion of a proline-rich domain (from positions P75 to R85) in the N-terminal A/B domain of wild-type human RARγ was done by site-directed mutagenesis (QuikChange II XL site-directed mutagenesis kit; Stratagene) using primers 5′-CAGCTCAGAGGAGATGGTGGTCTACAAGCCATGCTTCGTG-3′and 5′-CACAAGCATGGCTTGTAGACCACCATCTCCTCTGAGCTG-3′ and confirmed by DNA sequencing (Davis Sequencing, Davis, CA). Chicken SRC (c-SRC) (from H. Fu) was in plasmid pCSA with an HA tag at the N terminus of the protein. A QuikChange II XL site-directed mutagenesis kit (Stratagene) was used to delete the proline-rich domain of RARγ (11 amino acids from positions 75 to 85). Oligonucleotides used for deletion were 5′-CAGCTCAGAGGAGATGGTGGTCTACAAGCCATGCTTCGTG-3′ and 5′-CACGAAGCATGGCTTGTAGACCACCATCTCCTCTGAGCTG-3′. The wild-type FLAG-RARγ gene in the pcDNA3.1 vector was used as the template. DNA sequencing ensured the successful deletion of the domain at positions 75 to 85 (Δ75-85) of RARγ. A point mutation of SRC (Y527F) was made by using a QuikChange II XL site-directed mutagenesis kit according to the manufacturer's conditions. Oligonucleotides containing the mutation were designed according to the manufacturer's instructions, and they were 5′-GACAGAGCCCCAGTTCCAGCCTG GAGAGAACC-3′ and 5′-GGTTCTCTCCAGGCTGGAACTGGGGCTCTGTC-3′. Wild-type HA-SRC in vector pCSA was used as the template. The mutation was confirmed by DNA sequencing (Davis Sequencing). The c-src gene was amplified by PCR using primers 5′-CATCGCGGATCCACTAGTAACGGCCGCCAG-3′ with a BamHI site and 5′-GTCATGCCATGGCGAGGTTCTCTCCAGGCTG-3′ with an NcoI site and subcloned into plasmid pcDNA3-EGFP (from our laboratory) between BamHI and NcoI sites with enhanced green fluorescent protein (EGFP) at the C-terminal end of c-SRC. For hRARγ, PCR was carried out using primers 5′-GATCCGGAATTCCATGGCCACCAATAAGGAGCG-3′, containing an EcoRI site, and 5′-CGGGATCCCCCGGGGAAATAAGTTAGCACAATCAT-3′, containing a BamHI site. The amplified gene was then inserted into vector pDsRed2-C1 (Invitrogen) between EcoRI and BamHI sites with DsRed protein at the N terminus of hRARγ. The integrity of all the constructs was confirmed by DNA sequencing (Davis Sequencing).
The human NB cell lines LA-N-5, LA-N-6, and SK-N-BE(2) were cultured in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum (FBS) (HyClone, UT) with 100 units/ml penicillin and streptomycin at 37°C in a humidified atmosphere containing 5% CO2. Cells were treated with 10−5 to 10−6 M ATRA in 100% ethanol or dimethyl sulfoxide (DMSO) (vehicles for ATRA) under amber-light conditions (29, 30). Cells were pretreated with pan-SRC inhibitor PP1 at a final concentration of 4 μM for 1 h. Culture medium (containing PP1 and ATRA) was changed every 48 h. Similar culture conditions and treatment regimens were maintained for morphological studies and kinase assays.
Stable clones of NB cells overexpressing wild-type CSK were produced using a retroviral construct (pLXSN). Cells were plated in log phase at a density of 5 × 106 cells per 10-cm dish. The fresh supernatants from stable virus producer PG13 cells were filtered (0.22 μm) and were used for infection. The infected cells were selected in G418 (1 mg/ml), and the expression levels of CSK in different clones derived from the bulk population were evaluated by Western blot analysis. The clones (5P, 10P, 15P, 8, 10, 1, and 2) were maintained in 300 to 500 μg/ml of G418 in cultures, and expression was confirmed by Western blot analysis before every experiment (data not shown).
For the neurite outgrowth assay, cells were seeded (3 × 106 cells) in the above-described growth medium in a 10-cm-diameter tissue culture treated dish and allowed to attach for 24 h. The cells were then either treated with ATRA (10−5 M) under amber-light conditions or left untreated in control medium (25 μl of 100% ethanol in 10 ml) as described elsewhere previously (30). The medium was changed every 48 h. Neurite-bearing cells were semiquantified morphologically after 7 to 9 days of culture after staining them with methylene blue or under a phase-contrast microscope (Nikon TMS inverted microscope using Kodak Ektachrome 100 film). Quantification of morphological differentiation was done on the basis of the number of cells showing outgrowth more than twice the diameter of the cell body. Ten independent frames were considered for determinations of statistical significance. Considering the typical networking pattern observed after the RA treatment, the outgrowths were measured as the percentage of cells showing the second/third degree of collaterals.
LA-N-5 cells were solubilized with 500 μl of lysis buffer (150 mM NaCl, 6 mM Na2HPO4, 4 mM NaH2PO4, 2 mM EDTA, 1% sodium deoxycholate, 1% NP-40, 0.1% sodium dodecyl sulfate [SDS], 1% aprotinin, 0.2 M sodium orthovanadate, and 0.1 M phenylarsineoxide). For IP of SRC, YES, and FYN, clarified lysates were assayed for total protein (Bio-Rad protein assay kit) using BSA as a standard. The clear lysates were immunoprecipitated by specific antibodies (1 μg protein) for 2 h after protein equilibration (2 to 4 mg protein). Immunoprecipitates were bound to pansorbin and resolved by 12.5% SDS-polyacrylamide gel electrophoresis (PAGE). Membranes were immunoblotted with anti-human β-actin to confirm equal loading. Individual bands were visualized by an enhanced chemiluminescence reagent combined with peroxidase-conjugated anti-rabbit or anti-mouse IgG using BCIP (50 mg/ml) and nitroblue tetrazolium (50 mg/ml). The relative density of the bands was plotted in arbitrary units using ImageJ, version 1.32j (NIH).
The glutathione S-transferase (GST) fusion protein corresponding to the human PAK-1 p21 binding domain (PBD) (residues 67 to 150) was expressed in E. coli. The final protein products were bound to glutathione agarose in a liquid suspension containing 300 μg of PAK-1 PBD in 333 μl of 50% agarose slurry of 20 mM phosphate-buffered saline (PBS) (pH 7.4) containing 50% glycerol. The pull-down assay was carried out at different time points following ATRA treatment in controls and CSK-overexpressing clones (21). In short, ATRA-treated cells were lysed with extraction buffer (25 mM HEPES [pH 7.5], 150 mM NaCl, 1% Igepal CA630, 10 mM MgCl2, 1 mM EDTA, 10% glycerol, 10 μg/ml leupeptin, 10 μg/ml aprotinin, and 1 mM NaF-1 mM sodium orthovanadate), and following centrifugation, 10 μl of PAK-1 PBD (1 μg/μl) was added per sample of lysate and incubated for 45 min at 4°C. For the positive control, lysates of each clone were treated with 10 mM EDTA and 100 μM GTP-γS and incubated for 15 min at 30°C. Before adding PAK-1 PBD, the reaction was stopped by adding 60 mM MgCl2 to the mixture. Agarose beads were resuspended in 30 μl Laemmli sample buffer to resolve protein by 15% SDS-PAGE. The membranes were probed with monoclonal RAC1 (1:1,000) antibody. The bands were quantified by densitometry (Eagle Eye II-Still Video system; Stratagene). The amount of total RAC1 protein in each lysate was quantitated as an additional loading control. We carried out an earlier time course of activation of RAC1 following ATRA administration. Cells were treated with ATRA (10−5 M) in the dark at different time points (5 min, 15 min, 30 min, 1 h, 3 h, and 6 h), and activated RAC1 was pulled down from the cell lysates as mentioned above.
The phosphotransferase activity of cell lysates or specific immunoprecipitates (SRC, FYN, or YES) was determined in vitro using a cell-free system, i.e., an SRC assay kit from Upstate Biotechnology (Lake Placid, NY), according to the manufacturer's instructions, as described previously (21). The in vitro kinase assay employs an exogenous SRC substrate. The data are expressed as changes (n-fold) in c-SRC, FYN, and c-YES kinase activities in LA-N-5 cells under different experimental conditions. The immunoprecipitates were resolved on SDS-PAGE to quantify c-SRC, FYN, and YES protein levels in IPs. Each experimental point was performed in triplicates. The time course of activation of c-SRC following ATRA administration was studied in LA-N-5 cells. Cells were treated either with ethanol or DMSO (vehicles for ATRA) or with ATRA (10−5 M) in the dark for different time points (5 min, 15 min, 30 min, 1 h, 3 h, and 6 h), and the in vitro kinase activity of c-SRC was determined from the immunoprecipitated c-SRC as mentioned above. In separate experiments, in order to test the effect of RA-dependent binding of RARγ to c-SRC on the activation of c-SRC in LA-N-5 cells, we immunoprecipitated RARγ and c-SRC from LA-N-5 cells and performed an in vitro SRC kinase assay in the presence and absence of ATRA (see Fig. Fig.7A).7A). Endogenous c-SRC and RARγ from the normalized clear lysates of LA-N-5 cells (growing in medium containing 10% FBS) were immunoprecipitated separately using the respective antibodies (rabbit polyclonal antibody for c-SRC and rabbit polyclonal antibody for RARγ). Individual immunoprecipitants were then used for the SRC kinase assay. An in vitro kinase assay for SRC was carried out according to the manufacturer's protocol, with little modification. In short, the pellets obtained following the washings in buffer following an hour of incubation with secondary antibodies (pansorbin) were washed (one time) in pan-kinase buffer [0.1 M NaCl, 1% (vol/vol) aprotinin, 10 mM piperazine-N,N′-bis(2-ethanesulfonic acid) (PIPES), pH 7.0]. The pellets were then resuspended in SRC kinase reaction buffer (100 mM Tris-HCl [pH 7.2], 125 mM MgCl2, 25 mM MnCl2, 2 mM EGTA, 0.25 mM sodium orthovanadate, and 2 mM dithiothreitol). The reaction mixture for the in vitro SRC kinase assay contained SRC kinase reaction buffer, SRC substrate peptide (150 to 375 μM/assay), immunoprecipitated c-SRC from LA-N-5 cells, immunoprecipitated RARγ from LA-N-5 cells, and [γ-32P]ATP. Freshly made ATRA (where mentioned) was added to the reaction mixture under light-protected conditions. Following 30 min of incubation in the dark at 30°C (with agitation), the reaction was stopped by pulse centrifugation at 10,000 rpm. The supernatant was used for the subsequent procedures according to the manufacturer's protocol.
LA-N-5 cells were seeded onto glass coverslips in 10-cm petri dishes and allowed to attach in culture medium containing 10% FBS. Cells were fixed with chilled 100% methanol (10 min), permeabilized with 0.1% Triton X-100, and washed three times in PBS. Nonspecific binding was blocked with 2% BSA in PBS for 30 min at 37°C. Staining was carried out using mouse monoclonal anti-c-SRC (1:50) and rabbit polyclonal anti-RARγ (1:50) antibodies. Primary antibodies were diluted in blocking buffer, and cells were incubated for 1 h at 37°C. After washing three times in PBS, cells were incubated with fluorescein goat anti-mouse IgG (secondary for mouse monoclonal anti-c-SRC [1:1,000 dilution in blocking buffer]) and tetramethylrhodamine goat anti-rabbit IgG (secondary for rabbit polyclonal anti-RARγ [1:1,000 dilution in blocking buffer]) antibodies in the dark for 45 min. Nuclei were counterstained with DAPI. Cells were visualized under a Zeiss epifluorescence microscope, and images were collected and merged using SPOT ADVANCE Fluorescence PC software. The mean ratio of cytoplasmic to nuclear intensity and the correlation between cytoplasmic intensity and nuclear intensity of RARγ in LA-N-5 cells were determined using the MetaMorph Imaging system (Universal Imaging Corp., Downingtown, PA).
The colocalization of c-SRC and RARγ was studied from the pattern of distribution of exogenously expressed c-SRC and RARγ in HEK293 cells. The transient expression of EGFP-tagged c-SRC and red fluorescent protein (RFP)-tagged RARγ was carried out in HEK293 cells for these colocalization studies. In short, exponentially growing HEK293 cells were plated onto 12-mm glass coverslips (Fisher Scientific, Pittsburgh, PA) in six-well plates (four coverslips per well), and cells were allowed to attach overnight in medium containing 10% FBS. The following day, cells were either transiently transfected with EGFP-tagged c-SRC (0.8 μg) or RFP-tagged RARγ (0.8 μg) or cotransfected with both EGFP-tagged c-SRC (0.4 μg) and RFP-tagged RARγ (0.4 μg) using Lipofectamine 2000 reagent according to the manufacturer's protocol. Twenty-four hours after transfection, cotransfected cells were treated with either ATRA (10−5 M) or DMSO (vehicle of ATRA) for 1 h under subdued-light conditions. Cells were fixed with warm PHEMO buffer (0.068 M PIPES, 0.025 M HEPES, 0.015 M EGTA, 0.003 M MgCl2, 10% DMSO [pH 6.8]) containing 3.7% formaldehyde, 0.05% glutaraldehyde, and 0.5% Triton X-100 for 10 min at room temperature following a warm PBS wash. Coverslips were then washed three times in PBS for 5 min and mounted onto glass slides using Gel Mount mounting medium (Biomeda Corp., Foster City, CA). Cells were imaged using a Zeiss (Thornwood, NY) LSM 510 Meta confocal microscope with a 63× (1.4-numerical-aperture) or 100× (1.4-numerical-aperture) Plan-Apochromat oil objective. All images were acquired using Zeiss LSM 510 software and processed using Adobe Photoshop 7.0.
A surface plasmon resonance (SPR)-based biosensor system, the BIAcore (Uppsala, Sweden) 3000 system, was used to measure the kinetic parameters for the interactions between soluble recombinant RARγ protein (analyte) and the immobilized recombinant His-tagged SRC protein (ligand). The binding of RARγ recombinant protein in the presence of ATRA to SRC was monitored in real time as described previously (71). Briefly, His-tagged SRC (9.23 nM) was covalently linked to the surface of a research-grade CM5 sensor chip via an amine-coupling reaction according to the manufacturer's instructions (BIAcore handbook), yielding a resonance signal of ~200 resonance units (RU). One flow cell was intentionally left underivatized to allow for corrections for refractive index changes. The binding of RARγ (1.85 nM) to immobilized SRC in the presence or absence of ATRA (20 μM) on the biosensor surface was determined by the change in the RU using the KINJECT function of the BIAcore control software (flow rate of 30 μl/min, with 3 min for association and 5 min for dissociation). A schematic representation of the sensor chip surface is shown in Fig. Fig.6A6A (i, bottom). First, we tested the binding capacity of the individual components alone (see Fig. 6Ai). Different concentrations of ATRA were then titrated against the RARγ concentration to determine if ATRA could induce the binding of RARγ to c-SRC (see Fig. 6Aii). Finally, we examined the effect of free recombinant SRC protein on the binding of RARγ to the immobilized SRC protein (see Fig. 6Aiii). We then tested the specificity of RARγ binding to SRC in the presence of 9-cis-RA (20 μM), ATRA (20 μM), or 13-cis-RA (20 μM) (see Fig. 6Aiii). Each experiment was repeated at least twice to ensure reproducibility. The data analysis was performed using Bia-evaluation software supplied by the vendor.
Recombinant c-SRC was linked to the activated affinity medium Affi-Gel 15 according to the manufacturer's instructions. Briefly, the linking was carried out by adding recombinant c-SRC protein to the beads (0.5-ml bed volume) prewashed in 10 mM sodium acetate buffer (pH 4.5). After reaction for 2 h at room temperature, the unreacted sites on beads were blocked using a solution containing 100 mM Tris-HCl (pH 8.0) and 350 mM NaCl for 1 h. The beads were then tested for SRC by immunoblot analysis (see Fig. Fig.6B,6B, lanes 11 to 13). Beads bound to SRC were incubated with recombinant RARγ with or without ATRA in the dark for 1 h at 37°C. The SRC-conjugated beads were then resolved by SDS-PAGE and immunoblotted for RARγ. Recombinant protein served as a positive control. Beads alone with (see Fig. Fig.6B,6B, lane 2) or without (lane 1) blocking were incubated with RARγ as negative controls. Recombinant c-SRC-coated and blocked beads (see Fig. Fig.6B,6B, lane 3) also served as negative controls. The binding of RARγ to unconjugated Affi-Gel beads in the presence of ATRA was tested as the negative control.
HEK293 cells were maintained in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum. The cells in 60-mm tissue culture dishes were either transiently cotransfected with FLAG-RARγ (0.2 μg to 0.4 μg) and HA-c-SRC (0.2 μg-0.4 μg) plasmid DNAs or transfected separately with HA-c-SRC (0.8 μg) and FLAG-RARγ (0.8 μg) plasmid DNAs. For coimmunoprecipitation experiments, cotransfections and transfections were done with 0.8 μg of plasmid DNA using Lipofectamine 2000 reagent according to the manufacturer's protocol. Empty vectors (0.8 μg) were used for mock transfections. Whole-cell extracts were prepared in lysis buffer (50 mM Tris-HCl [pH 8], 0.05% NP-40, 100 mM NaF, 1 mM EDTA, 1 mM EGTA, 150 mM NaCl, 0.08 mM phenylmethylsulfonyl fluoride, 0.01 mg/ml leupeptin, 0.01 mg/ml aprotinin, and protease inhibitor cocktail tablet) after 24 h of transfection. Where mentioned, IP was performed on normalized lysates from the transfected cells by incubating them first with anti-HA antibody (monoclonal HA.11 antibody, clone 16B12) overnight at 4°C and then with protein G-agarose continuously for 1 h. Proteins with or without prior IP were resolved by 10% SDS-PAGE, electrotransferred onto nitrocellulose membranes, and immunoprobed (immunoblotted) separately by anti-HA (anti-HA polyclonal antibody [1:1,000 dilution]) and anti-FLAG (anti-FLAG-M2 monoclonal antibody [1:1,000 dilution]) antibodies. FLAG-tagged proteins in normalized lysates from a parallel set of transfected cells were immunoprecipitated using anti-FLAG polyclonal antibody. Lysates were incubated first with anti-FLAG antibody (overnight at 4°C) and then with pansorbin (for 1 h). Resolved proteins were immunoblotted separately by anti-HA (anti-HA monoclonal antibody) and anti-FLAG (anti-FLAG monoclonal antibody) antibodies. Chemiluminescence was detected by standard enhanced chemiluminescence, Western blotting detection reagents, and an analysis system according to the protocol of Amersham Biosciences (see Fig. Fig.6C).6C). To strengthen our claim that RARγ binds c-SRC in vivo, we generated a deletion mutant of the wild-type RARγ protein (Δ75-85) and tested the binding of this mutant to wild-type c-SRC by coimmunoprecipitation studies. HEK293 cells were transiently transfected with the FLAG-tagged Δ75-85 mutant of RARγ and HA-tagged wild-type c-SRC plasmids. In a separate experiment, HA-tagged Y527-SRC was cotransfected with FLAG-tagged wild-type RARγ plasmids. IPs of the normalized lysates from transfected cells were carried out using anti-HA antibody (monoclonal HA.11, clone 16B12) overnight at 4°C as mentioned above. Resolved proteins were immunoblotted separately by anti-FLAG (anti-FLAG-M2 monoclonal antibody [1:1,000 dilution]) and anti-HA (anti-HA polyclonal antibody [1:1,000 dilution]) antibodies. The binding of the FLAG-tagged Δ75-85 mutant of RARγ and HA-tagged wild-type c-SRC was compared with the binding of FLAG-tagged wild-type RARγ and HA-tagged wild-type c-SRC. Similarly, the binding of HA-tagged Y527-SRC and FLAG-tagged wild-type RARγ was compared with the binding of HA-tagged Y527-SRC and the FLAG-tagged Δ75-85 mutant of RARγ.
ATRA treatment caused the characteristic growth inhibition and differentiation in all the NB cell lines tested. ATRA-induced differentiation in LA-N-5, LA-N-6, and SK-N-BE(2) cells showed the characteristic neurite outgrowth as shown in Fig. Fig.11 and and2.2. Typically, cells extend axon-like processes by day 4 and morphologically resemble a tightly knit web (Fig. (Fig.2C).2C). By day 7, cells formed a typical network of cellular extensions involving second/third-degree processes. This effect was found to be dose dependent from 10−5 M to10−7 M and was appreciable from days 5 to 7 of treatment, reaching a maximum at around days 10 to 11. Flow cytometry analyses in LA-N-5 cells showed a steady increase in the percentage of cells in G0/G1 phase from the 5th day (112% of the control) to the 11th day (128% of the control) of ATRA treatment (data not shown). This increase was associated with the concomitant induction of p27kip1 expression and cell counts (data not shown). Interestingly, different NB cell lines showed the characteristics response to ATRA in terms of the degree of neurite outgrowth, and out of three cell lines tested, LA-N-5 cells were found to be the most sensitive to ATRA treatment (Fig. (Fig.11).
In order to test the involvement of SFKs in retinoid-induced NB differentiation, the pan-SRC inhibitor PP1 was used in this study. ATRA-induced neurite outgrowth was semiquantified following the treatment of PP1 in LA-N-5, LA-N-6, and SK-N-BE(2) cells. At around days 7 to 9, 100% of the ATRA-treated LA-N-5 cells were differentiated, compared to 10 to 15% in vehicle-treated cells. Interestingly, the pretreatment of NB cell lines with PP1 completely abolished neurite outgrowth (<5% of the control), as shown in Fig. Fig.1.1. Although the percentages of neurite-bearing cells were less in LA-N-6 and SK-N-BE(2) cells, the inhibitory effect of PP1 was consistent in all NB cell lines.
Recently, we reported the negative role of CSK in the TrkA-mediated neural differentiation of PC12 cells (21). Since CSK negatively regulated SFKs in the PC12 system, we overexpressed wild-type CSK in LA-N-5, LA-N-6, and SK-N-BE(2) cells (Fig. (Fig.2).2). Stable clones of LA-N-5 cells (clones 5P, 10P, and 15P as shown in Fig. Fig.2,2, lanes 3, 4, and 5, respectively, compared to the wild type and vector control in lanes 1 and 2, respectively), LA-N-6 cells (clones 8 and 10 as shown in lanes 3 and 4, respectively, compared to the wild type and vector control in lanes 1 and 2, respectively), and SK-N-BE(2) cells (clones 1 and 2 as shown in lanes 3 and 4, respectively, compared to the wild type and vector control in lanes 1 and 2, respectively) overexpressed wild-type CSK as shown in the upper panels of Fig. 2A, B, and C, respectively. Data show a three- to fourfold (densitometry) increase in the expression of CSK in different clones (upper bar diagrams of Fig. 2A, B, and C) of LA-N-5, LA-N-6, and SK-N-BE(2) cells compared to the endogenous levels of their respective vector controls.
In order to test the effect of the physiological inhibition of SFKs on ATRA-induced neurite outgrowth in NB, we treated the clones of LA-N-5, LA-N-6, and SK-N-BE(2) cells overexpressing CSK with ATRA and semiquantified their neuritogenic response. Figure 2A, B, and C show the effects of CSK on ATRA-induced neurite outgrowth in LA-N-5, LA-N-6, and SK-N-BE(2) cells, respectively. ATRA-induced neuritogenesis was significantly blocked by the overexpression of wild-type CSK compared to the empty vector controls (pLXSN) in all three NB cell lines tested. Although the neuritogenic response to ATRA varied between the cell lines, the effect of CSK was found to be uniformly inhibitory in the three different cell lines.
To determine the specificity of PP1 and wild-type CSK action in NB cells, we examined their effect on the kinase activities of the members of SFKs that are present in LA-N-5 cells. Figure Figure3A3A shows the kinase activities of three ubiquitously expressed members of SFKs in LA-N-5 cells under regular culture conditions as described in Materials and Methods. Treatment of PP1 (4 μM) blocked both c-SRC and FYN kinase activity in these cells compared to untreated controls (Fig. (Fig.3B).3B). Similarly, the overexpression of CSK inhibited c-SRC and FYN kinase activity compared to vector controls (Fig. (Fig.3C).3C). Our data demonstrate that the inhibitory effect of CSK on the kinase activity of c-SRC was more pronounced than that on the kinase activity of FYN in LA-N-5 cells. Anti-c-SRC (lanes 1 and 2) and anti-FYN (lanes 3 and 4) immunoblots corresponding to effects of PP1 and CSK were shown below the bars representing their respective kinase activities in Fig. 3B and C, respectively. Results indicate that treatment with a pharmacological inhibitor (PP1) and the overexpression of a physiological inhibitor of SFKs (CSK) reduced c-SRC and FYN kinase activities in LA-N-5 cells. Importantly, the inhibition of SFKs under these conditions corresponded with the blockade of ATRA-induced neurite outgrowth in LA-N-5 cells (Fig. (Fig.11 and and2A).2A). In order to understand the role of SRC in retinoid signaling, we determined the time course of activation of c-SRC following ATRA administration in LA-N-5 cells. Results show that ATRA causes a transient activation of c-SRC (around a twofold increase in kinase activity compared to the untreated control) within 15 min of the treatment, which lasted for a total of 60 min (Fig. (Fig.3D3D).
Recent literature revealed a major role of the small GTPase RAC1 in the regulation of neuritogenesis, an effect mediated through its control over cytoskeleton rearrangements in different cell types (5, 37). In fact, RAC1 has been found to act downstream of receptor protein tyrosine kinases in the neurite outgrowth response of N1E-115 NB cells (25, 62). In order to substantiate the effect of CSK on neurite outgrowth as shown in Fig. Fig.2,2, we studied the effects of CSK expression on the ATRA induction of RAC1 activation in LA-N-5 cells. LA-N-5 cells were treated for different times with ATRA, and GTP-RAC levels were quantified using a GTP-RAC pull-down assay as described previously (21). ATRA treatment induced a strong and time-dependent increase in RAC1 activation in LA-N-5 cells (Fig. (Fig.4A).4A). Similar levels of activated RAC1 were observed 24 to 72 h following ATRA treatment (Fig. (Fig.4A,4A, lanes 2, 3, and 4). In order to find the state of RAC1 activation at the terminal differentiation of cells, we also measured the activation at 168 h (seventh day) of ATRA treatment. Activation of RAC1 returned to the nonstimulated level at 168 h (results not shown). Protein levels of RAC1 in the lysates used to measure GTP-RAC1 were determined to be equal by immunoblot analysis. We then determined the effect of CSK overexpression on the capacity of ATRA to induce the activation of RAC1 to its GTP-RAC1 form. The results shown in Fig. Fig.4B4B demonstrate that CSK overexpression completely abrogates the capacity of ATRA to activate RAC1. From these data, we conclude that CSK regulates the RAR-induced activation of the small G protein RAC1. This is correlated directly with the marked inhibition of neurite outgrowth shown in Fig. Fig.2.2. Furthermore, our results show that in LA-N-5 cells, the activation of RAC1 occurs as early as at 15 min of ATRA treatment (Fig. (Fig.4C).4C). Convincingly, the kinetics of RAC1 activation in LA-N-5 cells were comparable to those of the activation of c-SRC under similar conditions of ATRA stimulation (Fig. (Fig.3D3D).
Figure Figure55 shows immunofluorescence analyses of LA-N-5 cells probed with specific antibodies directed against human RARγ and c-SRC. Immunofluorescence of RARγ in LA-N-5 cells shows a cytoplasmic staining for rhodamine corresponding to the RARγ immunoreactivity. The mean ratio of cytoplasmic to nuclear intensity for RARγ (expressed as relative pixel intensity) was 2.2 ± 0.16 (standard error) (n = 18 to 20) as measured using the MetaMorph Imaging system (Universal Imaging Corp.). A correlation (R2 = 0.7597) between nuclear intensity and cytoplasmic intensity was also observed for RARγ staining. For colocalization studies, LA-N-5 cells were double immunostained with specific antibodies for RARγ and c-SRC. A merge of the RARγ and c-SRC images using the SPOT ADVANCE program showed coimmunolocalization, as evidenced by the color change (Fig. (Fig.5).5). These results demonstrate that both RARγ and c-SRC have distinct patterns of staining, and an overlap of colors in the merged image is not 100%. It can be noted from this analysis that a fraction of RARγ is present in the cytoplasm of LA-N-5 cells and is colocalized with c-SRC.
Boonyaratanakornkit et al. (10) found a direct interaction between the SH3 domain of SFKs and the proline-rich motif of the nuclear steroid hormone receptor progesterone. A search of the RARγ amino acid sequence revealed several proline-rich motifs in the N terminus (residues 74 to 86). Our immunofluorescence results demonstrated a “colocalization” between RARγ and SRC; hence, we sought to determine the ability of purified recombinant RARγ to bind purified recombinant c-SRC by an SPR technique. This method would permit real-time direct measurements of the association and dissociation kinetics of macromolecular interactions. Figure Figure6A6A represents (i) the basal binding of individual components to SRC, (ii) the RA dose-dependent induction of association between RARγ and SRC, and (iii) the capacity of free recombinant SRC protein to compete for SRC-RARγ-RA binding. Figure Figure66 shows the representative BIAcore sensorgrams for the binding between immobilized recombinant c-SRC and RARγ in the presence or absence of its ligands (ATRA, 13-cis-RA, and 9-cis-RA). The RU signal change was close to 250 RU when SRC bound to 1.85 nM RARγ in the presence of 20 μM of ATRA, compared to virtually no binding when 20 μM 13-cis-RA was used to replace ATRA. Lower concentrations of ATRA (2 μM to 0.002 μM) did not cause any significant change in the RU (Fig. 6Aii). A much lower binding affinity (~25 RU) was observed when 20 μM 9-cis-RA was used to replace ATRA. Interestingly, when c-SRC (9.25 nM) was mixed with 1.85 nM RARγ and 20 μM ATRA, a more-than-twofold decrease in binding affinity was observed (~80 RU), suggesting that SRC in solution was competing with immobilized SRC on the sensor chip surface to bind RAR (Fig. 6Aiii). These results provide experimental evidence showing that c-SRC is capable of directly binding to RARγ and that this interaction is RA ligand dependent.
In order to study the binding interaction between RARγ and c-SRC, we used Affi-Gel beads bound to c-SRC. Figure Figure6B6B shows the RARγ immunoblot for the reaction mixtures containing RARγ recombinant protein incubated with c-SRC linked to Affi-Gel 15 beads in the presence (lanes 7, 8, and 9) or absence (lanes 4, 5, and 6) of ATRA. Washed precipitates run on an SDS-PAGE gel were blotted for RARγ with recombinant RARγ as a positive control (lane 10). Beads alone with (Fig. (Fig.6B,6B, lane 2) or without (lane 1) blocking were incubated with RARγ as negative controls. Recombinant c-SRC-coated and blocked beads (lane 3) also served as negative controls. RARγ binding to c-SRC-coated and blocked beads in the presence of ATRA was significantly higher (Fig. (Fig.6B,6B, lanes 7, 8, and 9) than that in the absence of ATRA (Fig. (Fig.6B,6B, lanes 4, 5, and 6). Densitometry evaluation showed a six- to eightfold increase (P < 0.0005) of binding in the presence of ATRA compared to that in the absence of ATRA as represented in the RARγ immunoblot. Figure Figure6B6B (inset) shows a Western blot of beads conjugated in the presence of recombinant SRC (lane 11) and conjugated in the absence of SRC (lane 12) or recombinant c-SRC loaded onto the gel as a positive control (lane 13). As a control, no binding was observed between RARγ and unconjugated Affi-Gel beads in the presence of ATRA. The combined results demonstrate a direct RA ligand-dependent induction for RARγ binding to c-SRC.
Next, we sought to determine if SRC can bind to RARγ in vivo. Full-length human RARγ (FLAG tagged) and c-SRC (HA tagged) were transiently expressed in HEK293 cells. Whole-cell lysates from HA-tagged c-SRC- and/or FLAG-tagged RARγ-transfected cells were resolved by SDS-PAGE along with lysates from HA-tagged c-SRC- and FLAG-tagged RARγ-cotransfected (with increasing DNA concentrations) cells and immunoblotted separately with anti-HA and anti-FLAG antibodies as shown in Fig. 6Ci. Comparable amounts of expression of both the proteins were observed following cotransfections of 0.4 μg and 0.8 μg of DNA. For coimmunoprecipitation studies, both RARγ (FLAG tagged) and c-SRC (HA tagged) proteins were transiently coexpressed in HEK293 cells. To determine if SRC and RARγ interact in vivo, we immunoprecipitated HA-tagged c-SRC or FLAG-tagged RARγ separately as shown in Fig. Fig.6C6C (panels ii and iii, respectively). FLAG-tagged RARγ coimmunoprecipitated with HA-tagged c-SRC from lysates using an anti-HA antibody. Immunoblotting with anti-HA antibody showed that RARγ (as detected by anti-FLAG antibody) (upper panel) was present in c-SRC IPs as shown in lane 3 (lower panel) of Fig. 6Cii. No RARγ was detected in IPs of lysates following transfections of HA-tagged c-SRC and FLAG-tagged RARγ alone, as in lanes 4 and 5, respectively, at the upper panel of Fig. 6Cii. Lane 4 of Fig. 6Cii (lower panel) showed that HA-c-SRC was immunoprecipitated following the transfection of HA-tagged c-SRC alone. No interacting proteins were detected in the preimmune samples (lane 1), the mock-transfected samples (lane 2), or HEK293 cell lysates (lane 6). Western blot analysis of whole-cell lysates from transfected cells was carried out simultaneously to confirm the expression of c-SRC and RARγ in all transfections (full length and a deletion mutation of 75 to 85 amino acids) in the immunoprecipitated samples (lanes 7 to 12). Data show the expression of HA-tagged c-SRC (lower panel, lanes 8 and 9) and FLAG-tagged RARγ (upper panel, lanes 8, 10, and 12) in the respective lysates. In order to further confirm the coimmunoprecipitation study, we performed reverse IP on the lysates from similar cotransfections; HA-tagged c-SRC was coimmunoprecipitated with FLAG-tagged RARγ following IP by anti-FLAG polyclonal antibody as shown in Fig. 6Ciii. Immunoblotting with anti-FLAG antibody showed that c-SRC (as detected by anti-HA antibody) interacted with RARγ as shown in lane 3 (upper panel) of Fig. 6Ciii. Similar to the results shown in Fig. 6Cii, no interactions were seen upon IP of lysates following transfections of FLAG-tagged RARγ and HA-tagged c-SRC alone, as in lanes 4 and 5, respectively, of Fig. 6Ciii (upper panel). Lane 4 of Fig. 6Ciii (lower panel) showed that FLAG-tagged RARγ was immunoprecipitated following the transfection of FLAG-tagged RARγ alone. No binding was observed in the case of preimmune samples (lane 1), mock-transfected samples (lane 2), and lysates from mock-transfected HEK293 cells (lane 5). Western blot analysis of whole-cell lysates from transfected cells was carried out simultaneously to test the expression of c-SRC and RARγ (full-length and deletion mutation of 75 to 85 amino acids) in the immunoprecipitated samples (lanes 7 to 12). The data demonstrate the expression of HA-tagged c-SRC (upper panel, lanes 7, 9, 10, and 11) and FLAG-tagged RARγ (lower panel, lanes 7, 8, 10, and 12) in their respective lysates (Fig. 6Ciii) and clearly demonstrate an in vivo interaction of RARγ and c-SRC. To further characterize the interaction between c-SRC and RARγ, we have generated a deletion of 11 amino acids (Δ75-85) within a proline-rich region of RARγ and coexpressed this mutant in HEK293 cells with c-SRC. IP studies using lysates from HEK293 cells that were cotransfected with the FLAG-tagged Δ75-85 mutant of RARγ and HA-tagged wild-type c-SRC demonstrated an abrogation of binding between these proteins (Fig. 6Civ). Additionally, we tested the binding of Y527-SRC with both wild-type and Δ75-85 RARγ and compared the binding with wild-type SRC. Figure 6Civ shows that both wild-type c-SRC and Y527-SRC bind equally with wild-type RARγ (top, lanes 3 and 4), while this binding was abrogated when wild-type RARγ was replaced by the mutated RARγ (top, lanes 5 and 6). Control experiments show that no binding was observed in the case of preimmune samples (top panel, lane 1), mock-transfected samples (top panel, lane 2 [this lane represents IP out of lysates from mock-transfected cells]), and whole-cell lysates from mock-transfected HEK293 cells (top panel, lane 7). Western blot analysis of whole-cell lysates from transfected cells was carried out simultaneously to confirm the expression of c-SRC and RARγ (full length and deletion mutant) in the immunoprecipitated lysates (top panel, lanes 8 to 12). Data confirm the expression of HA-tagged c-SRC (middle panel, lanes 3, 4, 5, 6, 8, 9, and 11) and FLAG-tagged RARγ (top panel, lanes 8, 10, and 12) in the respective cell lysates. The immunoblot in the bottom panel shows the expression of FLAG-tagged RARγs (wild-type protein as shown in lanes 3 and 4 as well as mutated protein as shown in lanes 5 and 6, respectively) in the cell lysates that were used in the IP studies. Data presented in Fig. 6Civ further confirm our observation that RARγ binds to c-SRC in vivo.
In order to confirm our result showing that endogenous c-SRC colocalizes with RARγ (Fig. (Fig.5),5), we studied the colocalization of EGFP-tagged c-SRC and RFP-tagged RARγ by confocal microscopy. For this purpose, we first identified the subcellular distribution of exogenously expressed c-SRC and RARγ in HEK293 cells (Fig. 6Cv, top). In short, EGFP-tagged c-SRC and RFP-tagged RARγ were transiently transfected (separately or together) in HEK293 cells, and their subcellular distributions were examined. Confocal images from cells transfected with EGFP-tagged c-SRC and RFP-tagged RARγ (separately or together) showed a clear cytoplasmic distribution of c-SRC (images a and c of Fig. 6Cv, top). RARγ, on the other hand, was localized in both nuclear and cytoplasmic compartments of the cells (images d and f of Fig. 6Cv, top). Figure 6Cv shows that the pattern of subcellular distribution of c-SRC and RARγ as identified by confocal microscopy was similar to the pattern of distribution of endogenous c-SRC and RARγ as observed by immunofluorescence staining of these proteins in LA-N-5 cells (Fig. (Fig.5).5). When EGFP-tagged c-SRC was cotransfected with RFP-tagged RARγ, a change in color (yellow/orange) of the fluorescence was observed in the cytosol of the merged confocal images (images c and f of Fig. 6Cv, bottom), indicating that EGFP-tagged c-SRC was colocalized with RFP-tagged RARγ in that compartment of the cells. A clear increase in the trend of the merge between EGFP-tagged c-SRC and RFP-tagged RARγ was observed following the administration of ATRA (compare image c and image f of Fig. 6Cv, bottom). A similar increase in the trend of the nuclear distribution of RFP-tagged RARγ was observed following an hour of ATRA treatment in these cotransfected cells. Interestingly, we observed a characteristic morphological change in these cells following ATRA administration. Confocal images (images d and f) of Fig. 6Cv (bottom) show that following an hour of ATRA treatment, cells exhibit characteristic protrusions of plasma membranes resembling an “edge”-like focal adhesion structure (Fig. 6Cv, bottom). These discrete “edge”-like adhesion structures of the plasma membrane were observed in 100% of the ATRA-treated cells (data not shown). Moreover, only c-SRC (not RARγ) has been observed to be preferentially redistributed in these structures, as evident from their green fluorescence, and hence, no change of color in the merged images (with RFP-tagged RARγ) was seen in these “edge”-like focal adhesion structures.
In order to determine if RA-dependent binding to RARγ activated c-SRC in LA-N-5 cells, we immunoprecipitated RARγ and c-SRC from LA-N-5 cells and performed an in vitro SRC kinase assay in the presence and absence of ATRA (Fig. (Fig.7A).7A). Lane 4 of Fig. Fig.7A7A showed a significantly higher (P < 0.05) SRC kinase activity in reaction mixtures containing both RARγ and c-SRC incubated with ATRA than the control lanes (lanes 1, 2, and 3). To understand the role of CSK in the ligand-dependent interaction between SRC and RARγ, we studied the kinase activity of SRC immunoprecipitated from PP1-treated cells or LA-N-5 cells overexpressing CSK (Fig. (Fig.7B).7B). The kinase activity (counts per minute [cpm]) of SRC immunoprecipitated from these cells when incubated with RARγ in the presence of ATRA was markedly reduced compared to control untreated cells (Fig. (Fig.7B,7B, lanes 2 and 3 and lane 1, respectively).
The results generated using a number of knockout mouse models have confirmed a role for the nonreceptor protein tyrosine kinases and their downstream substrates in neural development. Neuronal developmental defects have been observed in SRC, FYN, CSK, ABL, and DAB knockout mice (7, 26, 32). Since nuclear hormone receptors are known to participate in neuronal development, we were prompted to examine the role of the nonreceptor protein tyrosine kinase CSK in nuclear hormone receptor signaling pathways. We focused on the RAR signaling pathway in NB cells, partly because NB is a malignancy currently treated with RA (49).
The results of our analysis demonstrate that RA-mediated differentiation in NB involves a ligand-dependent binding of RAR to the SFK (c-SRC), which leads to the enzymatic activation of this kinase. Our data provide evidence that c-SRC interacts directly with the RAR and that the CSK-SFK signaling axis regulates RA-induced neuritogenesis. These results establish the existence of a nongenomic signaling pathway involving SRC that is required for RAR-induced differentiation. Our data provide important insights into how the RAR orchestrates and transmits signals in both the nonnuclear and nuclear cell compartments to orchestrate neurite outgrowth and possibly neuronal differentiation. The physiologic and biochemical processes involved in neuronal differentiation include cytoskeletal changes, microtubule dynamics, cell cycle arrest, and the induction of specific genes required for specialized neuronal cell function. Our results, which implicate the tyrosine kinases CSK and SRC in RAR signaling, may have implications for strategies aimed at the pharmacologic control of neuronal differentiation.
In the present study, we have observed characteristic growth inhibition, cell cycle arrest, and neuritogenic differentiation in all three NB cell lines following ATRA exposure. In order to determine the role of CSK and SFKs in ATRA-induced neuritogenesis (neurite outgrowth), we overexpressed CSK in multiple NB cell lines. Interestingly, progesterone and estrogen receptors have recently been implicated to signal through members of the src family kinases (13, 17, 36, 50, 64). Boonyaratanakornkit et al. previously reported that the progesterone receptor causes an activation of SRC following binding to the SRC-SH3 domain through its proline-rich region (10). Other members of the nuclear hormone family, including the androgen, estrogen, and vitamin D receptors, were not found to display this signaling paradigm. Upon sequence alignment of the different RARs, we noted a group of conserved proline-rich amino acid sequences within the coding regions of the RARγ, RARβ, and RARα receptors. Within the N-terminal region of RARγ, a PxxP motif at amino acids 75 to 85 exists (Fig. (Fig.8).8). This observation, together with our initial observation that SFK inhibitors and CSK overexpression completely blocked RA-induced differentiation of NB cells, led to our efforts to determine if SFKs are somehow associated with RAR signaling in neuronal cells.
To begin to explore whether a physical interaction occurs between RARγ and SRC in NB cells, we utilized immunofluorescence microscopy. The subcellular dynamics of RAR have been examined previously by Kawata (35), and a transient modulation of cytoplasmic and nuclear expression of RARs in differentiating human NT2 cells was reported previously by Borghi et al. (11). We were able to demonstrate the immunolocalization of RARγ and c-SRC in LA-N-5 cells (Fig. (Fig.5).5). Other investigations showed that 20% of RAR is present in the cytoplasm of HL-60 cells and that RAR shuttles between cytoplasmic and nuclear compartments (48). In agreement with that report, we observed a cytoplasmic staining for RARγ and a correlation between cytoplasmic and nuclear intensity in LA-N-5 cells (Fig. (Fig.55).
To confirm the interaction between SFK and RARγ, we utilized several established biophysical methodologies combined with a biochemical analysis of SFK activity. Our results provide the first evidence that RARγ binds to c-SRC (Fig. (Fig.6).6). Ligand-dependent real-time binding between RARγ and c-SRC was observed by SPR analysis. Interestingly, only ATRA and not 9-cis-RA/13-cis-RA was found to mediate the binding of RARγ with c-SRC. Affi-Gel binding studies in vitro also showed an ATRA-dependent binding between these two proteins. Since RARγ has conserved proline-rich regions similar to the progesterone receptors, we argue that the RARγ interaction may be mediated through the SH3 domain of SRC, as reported previously by Boonyaratanakornkit et al. (10). We observed that the ligand-mediated binding of RARγ to c-SRC is direct and independent of RARE/DNA. Such an interaction is possible only if the receptor undergoes a conformational change following ligand binding in the absence of DNA. In line with our argument, studies described previously by Leng et al. showed that upon ligand binding, RAR acquires a specific conformation involving its ligand binding domain, and this can occur in the absence of DNA (40). Upon confirming the in vitro interaction of RARγ with c-SRC by SPR and Affi-Gel techniques, we went on to demonstrate the binding of RARγ to c-SRC in vivo. Our result established an interaction between RARγ and c-SRC in mammalian cells (Fig. (Fig.6C).6C). The in vivo IP results were consistent with our in vitro observation that RARγ binds directly to c-SRC. This binding was observed in the cases of both wild-type SRC and Y527F-mutated SRC (Fig. 6Civ). From sequence analyses of RARs, we predicted that this kind of direct binding involving the polyproline motif in the N-terminal A/B domain of the receptor is likely to occur. To test this idea, we deleted the proline-rich domain (from amino acids 75 to 85 [Δ75-85 RARγ]) in the A/B domain of RARγ (Fig. 6Civ). Coimmunoprecipitation experiments confirmed (Fig. 6Civ) that the deletion of the proline-rich domain of RARγ abrogates the binding of the RARγ protein to c-SRC. These results support a role for a direct binding of RARγ to c-SRC in vivo and suggest a molecular basis for the RARγ-SRC interaction. Since coimmunoprecipitation experiments were performed in the absence of ATRA, the possibility of the existence of a partial constitutive binding between these proteins cannot be ruled out. However, as the cells were cultured in medium containing10% FBS, and serum contains RA, it is likely that the observed binding between these proteins is mediated through RA present in FBS. Our experiments confirm the binding of SRC to the RARγ isoform. Considering the conservation of proline-rich regions in RARα and RARβ, it is likely that a similar interaction may also occur in other forms of RARs. Studies have been undertaken to characterize the binding of SRC to different isoforms of RAR by mutational analyses.
In addition to the demonstration of the cytoplasmic colocalization of endogenous c-SRC and RARγ proteins in LA-N-5 cells by immunofluorescence studies, we have tested the colocalization of these proteins using confocal microscopy. In HEK293 cells, EGFP-tagged c-SRC is colocalized with RFP-tagged RARγ in the cytoplasmic compartment of cells (Fig. 6Cv, bottom). This result (Fig. 6Cv) is in agreement with our observation that endogenous RARγ and c-SRC proteins are colocalized in the cytosol of LA-N-5 cells (Fig. (Fig.5).5). Data further confirm our finding that both RARγ and c-SRC are colocalized in the cytoplasm and favor the likelihood of an association/binding between them. The identification of c-SRC-rich “edge”-like adhesion structures of the plasma membrane in ATRA-treated cells is a novel observation. Since c-SRC is activated following ATRA treatment (Fig. (Fig.3D),3D), and c-SRC has been shown to be differentially localized in these structures, it implies that the morphological changes occurring in the cells following ATRA treatment are mediated through the intracytoplasmic relocalization of the activated c-SRC. The physiological significance of the observed relocalization of c-SRC to the periphery of the cell during ATRA stimulation remains an open question.
Since RARγ associates directly with SRC, we next determined whether the RARγ-SRC interaction leads to the catalytic activation of this tyrosine kinase. Since the activation of SFKs occurs by “domain displacement” interactions involving the binding of ligands to its SH2 and/or SH3 domains (9, 61), we postulated that the binding of RARγ to c-SRC activated the kinase in an ATRA-dependent way. Our results demonstrate that c-SRC immunoprecipitated from LA-N-5 cell was activated in the presence of RARγ plus ATRA. These results indicate that the RA-RARγ binding to c-SRC leads to the activation of the kinase (Fig. (Fig.7A).7A). The mediation of cellular functions involving a similar kind of an activation of SFKs via interactions with its SH2/SH3 domains by other steroid hormone receptors (progesterone, androgen, estrogen, and vitamin D receptors) has been reported in literature (10, 13, 17, 36, 50).
To further explore one potential mechanism for how the RARγ-SRC interaction induced by ATRA might contribute to neurite outgrowth in LA-N-5 cells, we examined downstream components of neurite outgrowth, the activation of Rho family GTPases. The fact that SRC from LA-N-5 cells is activated following its binding to RARγ in an RA-dependent way and that the physiologic inhibition of SRC activity blocked ATRA-induced neurite outgrowth in these cells prompted us to argue that the activation of SRC following ATRA administration presumably leads to a cascade of downstream signals culminating in the neurite outgrowth. Neurite outgrowth occurs through the rearrangement in cytoskeletal dynamics of actin (37, 51, 54), and the Rho family GTPases RAC1 and Cdc42 are cytoskeleton switches for neuritogenesis (4, 5, 24, 42, 52, 69). Recently, we reported that SFKs regulate the activation of small GTPases in PC12 neuritogenesis (21). Others showed previously that the inhibition of RAC1/Cdc42 abrogates neurite outgrowth in various cell types (15, 37, 55). The overexpression of dominant negative RAC1-N17 and constitutively active RAC1-V12 has been shown to block and induce ATRA-induced neurite outgrowth, respectively, in SH-SY5Y cells (58). Neuritogenesis in N1E-115 NB cells has been reported to involve RAC1 (23, 25, 62). We propose that SRC may regulate the ATRA-induced activation of RAC1 in our system. In agreement with data reported previously by Alsayed et al., we show that ATRA induces the activation of RAC1 in our NB model (1). The fact that PP1 treatment or CSK overexpression blocked the ATRA-induced activation of RAC1 vis-à-vis neurite outgrowth in LA-N-5 cells (Fig. (Fig.22 to to4)4) suggests that RAC1 may be involved in RAR-induced nonnuclear signals downstream of SRC in NB. A nongenomic mode of action of ATRA in its neuritogenic response (in NB) was supported by the demonstration of an early and transient activation of c-SRC following the administration of ATRA in LA-N-5 cells (Fig. (Fig.3D).3D). Furthermore, the kinetics of c-SRC activation by ATRA match the pattern of an early activation of RAC1 following ATRA treatment in LA-N-5 cells (Fig. (Fig.4C).4C). The activation of RAC1 within 15 to 30 min of treatment with ATRA suggests a rapid and hence genome-independent mode of signaling in retinoid function. Interestingly, the characteristic morphological changes observed under a confocal microscope following 1 h of ATRA administration (Fig. 6Cv, bottom) strengthen our argument for the role of c-SRC in the acute mode of signaling mediated by the RAR in NB cells. Taken together, our study identifies a direct binding of RARγ to c-SRC and provides the first evidence for the biological significance of this RARγ signaling pathway through the activation of SRC and its downstream effector RAC1 in the neuronal differentiation of NB cells.
In conclusion, we propose a new paradigm for RAR signaling in neuronal cells. In addition to its capacity to activate gene expression, RAR engages in a dual function, the capacity to activate SRC in the cytoplasm through a hormone-dependent direct binding to SRC, a process required for neuritogenesis (Fig. (Fig.8).8). In this manner, the RAR can coordinate and orchestrate complex cytoplasmic, membrane, and nuclear events required for neuronal differentiation. Why would there be a utility for the RAR to bind and activate cytoplasmic and nuclear effectors? Many signaling proteins, including membrane proteins, coordinate downstream and upstream signals by virtue of their multiple domains. Recent evidence suggests not only that the epidermal growth factor receptor binds ligand at the cell surface but also that a portion of its cytoplasmic domain is then cleaved to enter the nucleus to drive transcription (1). Similarly, we envision that a multifunctional RAR may exert its effects on cytoplasmic versus nuclear targets via different regions of the RAR protein (Fig. (Fig.8).8). We have now partly mapped the regions of SRC and RARγ required for this interaction in vivo. What is less clear at this time is to what extent the cytoplasmic nonnuclear functions of RAR (SRC activation) can be separated from the transcriptional functions of the retinoid receptor and how the coordination of these distinct functions is mechanistically achieved. Preliminary data generated in CSK-overexpressing NB cells demonstrates that the RA-RAR-induced cell cycle arrest and p27 induction responses are intact, suggesting that these nonnuclear events are not required for certain RAR functions. Many of the mechanisms for the nonnuclear RAR function remain to be explored. We will use our CSK-transduced NB cell lines to further explore these elements of RARγ structure and function (adapter protein interactions, etc.). Finally, the recent evidence that CSK and SRC regulate nongenomic androgen receptor signals (74) suggests that this signaling axis may have an impact on hormone-induced signals that relate to cellular transformation in epithelial cells.
Wild-type human RARγ1 was kindly provided by Ron Evans (Salk Institute, CA), and c-SRC was obtained from H. Fu (Emory University, GA). Mean ratios of cytoplasmic to nuclear intensity and correlations between cytoplasmic intensity and nuclear intensity were determined by Adam Marcus of the Confocal Microscope Facility, Winship Cancer Center, Emory University, Atlanta, GA. We thank K. Schafer-Hales (Cell Imaging and Microscopy Core, Winship Cancer Institute) for her help with the confocal microscope and image processing.
We also acknowledge the support of the NIH for funding this work, CA94233 to D.L.D. D.L.D. is supported by a Georgia Cancer Coalition grant. The work is supported by the Aflac Cancer Center and Blood Disorders Service.
Published ahead of print on 26 February 2007.