ABSTRACT
The p75 neurotrophin receptor (p75NTR, also known as tumor necrosis factor receptor superfamily member 16) is implicated in diverse cellular events, but fundamental aspects of its signaling mechanisms remain unclear. To address this, we have established a novel bioassay to characterize signaling cascades activated by p75NTR. We show that in COS7 cells, p75NTR expression causes a large increase in cell surface area that relies on the activation of Rac1, and we demonstrate that the p75NTR-dependent COS7 phenotype is dependent on ADAM17- and γ-secretase-dependent cleavage of p75NTR and generation of the p75NTR intracellular domain (p75NTRICD). We show that the p75NTR adaptor protein NRAGE (also known as MAGED1) acts downstream of the p75NTRICD in this cascade and, through a yeast two-hybrid screen, identify NEDD9, a Cas family adaptor protein, as a novel NRAGE-binding partner that mediates p75NTR-dependent Rac1 activation and cell spreading. Our results demonstrate a crucial role for p75NTR cleavage in small GTPase activation and define a novel Rac1 activation pathway involving the p75NTRICD, NRAGE and NEDD9.
INTRODUCTION
The p75 neurotrophin receptor (p75NTR, also known as tumor necrosis factor receptor superfamily member 16), a member of the tumor necrosis factor receptor (TNFR) superfamily, participates in an array of cellular events that include apoptosis, survival signaling, differentiation, neurite outgrowth and growth cone collapse (reviewed in Schecterson and Bothwell., 2010). Some of these are activated by the neurotrophin binding to p75NTR and, in others, p75NTR acts as an accessory receptor and signaling component for other ligand-binding receptors (reviewed in Reichardt., 2006). Several studies have demonstrated the importance of small GTPases, most notably RhoA and Rac1 (Yamashita et al., 1999, Yamashita et al., 2002, Harrington et al., 2002, Yamashita and Tohyama., 2003, Domeniconi et al., 2005, Harrington et al., 2008, Coulson et al., 2008, Park et al., 2010, Sun et al., 2012), in p75NTR signaling cascades but the precise mechanisms by which these and other downstream elements are activated remain uncertain.
In neuronal growth inhibition, a receptor complex containing p75NTR, Nogo (also known as reticulon-4) receptor and Lingo1 responds to myelin-based inhibitors (MBIs), and p75NTR functions as the receptor component that induces RhoA activation (Yamashita et al., 2002, Domeniconi et al., 2005, Harrington et al., 2008, Park et al., 2010, Sun et al., 2012). Kalirin-9, which is a dual Rho and Rac guanine exchange factor (GEF), and Rho GDP-dissociation inhibitor (Rho-GDI) share a binding site on p75NTR, and it has been proposed that MBIs shift the p75NTR binding preference from kalirin-9 to Rho-GDI (Harrington et al., 2008). This, in turn, acts to inhibit Rho-GDI activity and promote RhoA action. p75NTR can also mediate Rac1 regulation. In some settings, this links the activated receptor to the JNK signaling cascade and promotes apoptosis (Harrington et al., 2002), and in others, p75NTR collaborates with Par3 to localize Rac1 to the axon–glial interface and thereby promote myelination (Tep et al., 2012).
p75NTR undergoes a two-step cleavage event, known as regulated intramembrane proteolysis (RIP), in which the extracellular juxtamembrane domain is cleaved by ADAM17, followed by cleavage of the transmembrane domain through the γ-secretase complex. These sequential cleavage events release the intracellular domain (ICD) from its transmembrane tether. The untethered p75NTRICD fragment that is generated has been implicated in cell migration (Wang et al., 2008), enhancement of pro-survival signaling (Ceni et al., 2010, Matusica et al., 2013), induction of apoptosis (Kenchappa et al., 2006, Kenchappa et al., 2010) and changes in cell morphology (Domeniconi et al., 2005).
p75NTR is not catalytically active and therefore downstream signaling events rely on its interaction with cytosolic adaptor proteins. Previous studies have identified NRAGE (neurotrophin receptor interacting MAGE homolog, also known as MAGED1) as a p75NTR interactor that activates a JNK and caspase-3-dependent apoptotic pathway in vitro and in vivo (Salehi et al., 2000, Salehi et al., 2002, Bronfman et al., 2003, Bertrand et al., 2008). Interestingly, NRAGE has also been shown to play a role in regulating homotypic cell adhesion (Xue et al., 2005) and in cell movements that include epithelial to mesenchymal transitions (EMT) in mammary epithelia (Kumar et al., 2011).
Taken together, available data suggest that p75NTR and the soluble p75NTRICD fragment can participate in a wide array of functions in different contexts. The mechanisms that allow for the activation of distinct signaling paths remain uncertain, and specific signaling cascades that are selectively activated by the p75NTRICD have not been identified. This is, in part, due to a lack of reliable in vitro assays for analyzing p75NTR and p75NTRICD signaling cascades. Here, we describe a novel COS7-based cell spreading assay that provides a robust output for analyzing a subset of p75NTR-dependent signaling events. Using this assay, we show that p75NTR-dependent Rac1 activation only occurs after ADAM17-dependent cleavage of p75NTR and liberation of the p75NTRICD. We demonstrate that activation of Rac1 by the p75NTRICD relies on an NRAGE-dependent pathway and, through yeast-two hybrid screening, we identify NEDD9 as a novel NRAGE cofactor that is also required for p75NTR-dependent Rac1 activation and cell spreading. Our results demonstrate a crucial role for p75NTR cleavage in small GTPase activation and define a novel Rac1 activation pathway involving the p75NTRICD, NRAGE and NEDD9.
RESULTS
p75NTR induces COS7 cell spreading
To gain insight into the role of p75NTR cleavage and proximal signaling events, we established a bioassay based loosely on a COS7 contractility assay used previously to study semaphorin-induced signaling events (Takahashi et al., 1999, Takahashi and Strittmatter, 2001, Zanata et al., 2002, Mitsui et al., 2002). COS7 cells were transfected with EGFP together with p75NTR constructs (Fig. 1A,B), then plated at low density on glass coverslips that had been pre-coated with laminin. After 24 hours, cells were fixed, left non-permeabilized and exposed to wheat germ agglutinin conjugated to Cy3 to visualize cell surfaces. Fig. 1C,D shows that overexpression of full-length wild-type p75NTR or the p75NTRICD resulted in significant COS7 cell spreading, giving a mean cell surface area almost double that of control cells. Interestingly, overexpression of a cleavage-resistant form of p75NTR did not promote cell spreading, suggesting that COS7 cell spreading might provide a bioassay for p75NTR signaling activities that are dependent on receptor cleavage.
p75NTR acts through Rac1 to promote cell spreading
The Rho family of small GTPases plays a prominent role in transducing signals from plasma membrane receptors to the actin cytoskeleton, and p75NTR has previously been shown to activate RhoA and Rac1. Because Rac1 is necessary for fibroblast and macrophage cell spreading (Wells et al., 2004, Guo et al., 2006), we asked whether Rac1 was required for the cell spreading induced by p75NTR overexpression. To do this, COS7 cells were transfected with RacN17, a dominant-negative Rac1 isoform, and with full-length p75NTR or with the p75NTRICD. Fig. 2A,B shows that cell spreading evoked by the p75NTR constructs was blocked in cells expressing RacN17.
As an alternative approach, we asked whether NSC 23766, a Rac1 inhibitor that blocks the interaction of Rac1 with its cognate GEFs, reduced the amount of cell spreading induced by p75NTR. COS7 cells transfected with full-length p75NTR or the p75NTRICD were plated on glass coverslips precoated with laminin and, after 24 hours, were treated with NSC 23766 (100 µM) for 6 hours. Fig. 2C,D shows that the Rac1 inhibitor effectively blocked cell spreading evoked by full-length p75NTR and by the p75NTRICD.
In parallel experiments, we tested whether a dominant-negative form of RhoA (N19) altered spreading induced by full-length p75NTR or the p75NTRICD, but we found that it had no effect in this assay (Fig. 2E). Taken together, these data demonstrate that Rac1, but not RhoA, plays a crucial role in evoking p75NTR-induced cell spreading in COS7 cells.
ADAM17-mediated p75NTR cleavage is a prerequisite for cell spreading
p75NTR undergoes a two-step cleavage event, known as regulated intramembrane proteolysis (RIP), to generate the p75NTRICD. To determine whether RIP-dependent generation of the p75NTRICD is required for the changes in COS7 cell morphology, we first determined whether GM6001 and BB94, metalloprotease inhibitors capable of targeting ADAM proteases, blocked the cell spreading induced by overexpression of full-length p75NTR. COS7 cells transfected with full-length p75NTR or the p75NTRICD were plated on glass coverslips pre-coated with laminin and, after 24 hours, were treated with GM6001 (10 µM) and BB94 (200 nM) for 1 hour. Fig. 3A,B shows that GM6001 and BB94 sharply reduced spreading in COS7 cells expressing full-length p75NTR and that these compounds had no effect on cell spreading induced by the p75NTRICD. This suggests that p75NTR cleavage mediated by cell-surface ADAMs plays a crucial role in generating the cell spreading phenotype.
We have previously shown that ADAM17 mediates the initial extracellular cleavage of p75NTR (Kommaddi et al., 2011, Ceni et al., 2010), and therefore characterized its effect on cell spreading. Fig. 3C shows that ADAM17 mRNA is readily detected in COS7 cells and its levels are not altered by overexpression of full-length p75NTR, the cleavage-resistant form of p75NTR or the p75NTRICD. To determine whether ADAM17 is required for the effect of p75NTR on cell spreading, small interfering RNA (siRNA) was used to reduce endogenous ADAM17 levels in COS7 cells overexpressing full-length p75NTR and p75NTRICD. ADAM17 siRNA effectively reduced ADAM17 expression in COS7 cells, and accumulation of the p75NTRICD was blocked when ADAM17 levels were suppressed (Fig. 3D). To test the effect of ADAM17 knockdown on p75NTR-induced cell spreading, COS7 cells were transfected with ADAM17 siRNA or with control siRNA, together with p75NTR+EGFP or EGFP alone, and assessed for cell area 48 hours later. ADAM17 knockdown effectively blocked cell spreading induced by p75NTR overexpression but had no effect on spreading induced by the p75NTRICD (Fig. 3E,F). Taken together, these data demonstrate that by initiating p75NTR cleavage and allowing generation of the p75NTRICD, ADAM17 plays a crucial role in elaborating the cell spreading phenotype induced by p75NTR overexpression.
The p75NTRICD drives cell spreading
The p75NTRICD is generated in COS7 cells when the full-length receptor is overexpressed. The production of this fragment is blocked by the metalloprotease inhibitors GM6001 and BB94 as well as by compound XXI, a γ-secretase inhibitor (Fig. 4A). Consistent with previous results, γ-secretase inhibition resulted in accumulation of a 25-kDa C-terminal fragment (CTF) of p75NTR (Fig. 4A, lane 2). Several studies have suggested that the CTF and the p75NTRICD have distinct biological activities (Domeniconi et al., 2005, Kenchappa et al., 2006, Underwood et al., 2008, Kenchappa et al., 2010), and we therefore addressed whether the CTF could mediate p75NTR-dependent cell spreading. COS7 cells overexpressing p75FL, p75NTRICD or an EGFP control were plated on glass coverslips and treated with compound XXI (10 µM) or vehicle for 1 hour. Fig. 4B,C shows that compound XXI had no effect on cells expressing EGFP alone or expressing the p75NTRICD yet it blocked the spreading of cells expressing full-length p75NTR. Because compound XXI causes accumulation of the p75NTR CTF while preventing p75NTRICD accumulation, this indicates that the p75NTRICD is the relevant fragment required for p75NTR-dependent cell spreading in COS7 cells. Therefore, the p75NTRICD, generated through sequential ADAM17- and γ-secretase-dependent p75NTR cleavage, is required to drive changes in COS7 cell shape.
p75NTR binds to the four neurotrophins present in mammals and in our next experiments, we addressed whether these ligands have an impact on the COS7 spreading assay. The addition of nerve growth factor (NGF) or brain-derived neurotrophic factor (BDNF) (each at 25 ng/ml), for 1 hour or for 24 hours, had no effect on p75NTR cell spreading (Fig. 4D). To determine whether endogenous neurotrophins might play a role, we used RT-PCR to establish whether neurotrophins or their receptors are expressed in COS7 cells. Neither p75NTR nor the Trk receptor mRNA were detected (data not shown) but mRNA encoding each of the four neurotrophins was present in this line (Fig. 4E). The capacity of full-length p75NTR to drive cell spreading was not significantly changed when cells were maintained in REX (Fig. 4F), an antibody directed against the p75NTR extracellular domain that blocks ligand binding (Clary et al., 1994). We conclude that neurotrophin binding to p75NTR is not required for the cell spreading phenotype observed in this line and that constitutive, rather than ligand-induced, p75NTR cleavage drives the cell spreading phenotype.
NRAGE acts downstream of p75NTR to mediate cell spreading
The p75NTRICD does not have intrinsic enzymatic activity and therefore relies on interaction with cytosolic binding proteins to elicit downstream effects. One of these adaptors, termed NRAGE, mediates morphological changes in transformed human mammary epithelial cells and U2OS cells (Xue et al., 2005, Kumar et al., 2011), and we therefore asked whether NRAGE is required for p75NTR-dependent COS7 cell spreading. After confirming that NRAGE siRNAs suppressed NRAGE protein expression in COS7 cells (Fig. 5A), we asked whether NRAGE depletion altered cell spreading induced by overexpression of full-length p75NTR or the p75NTRICD. NRAGE knockdown had no effect on the area occupied by control cells but it eliminated the increase in cell spreading induced by p75NTR or the p75NTRICD (Fig. 5B,C). NRAGE could mediate these effects by binding to the p75NTRICD and playing a direct role in downstream signaling or by facilitating the generation or maintenance of the p75NTRICD. However, levels of full-length p75NTR or the p75NTRICD were unchanged by NRAGE knockdown (data not shown), ruling out the latter possibility. We therefore focused our efforts on the hypothesis that NRAGE links the p75NTRICD to downstream signaling partners.
NRAGE interacts physically and functionally with NEDD9
To identify NRAGE-interacting proteins that could link the p75NTRICD to Rac1 activity, we performed a cytosolic yeast two-hybrid screen, using an NRAGE–SOS fusion protein as bait with a human fetal brain library of cDNAs fused to a myristoylated membrane-localization signal as a source of potential binding partners. Interestingly, four of the positive clones to emerge from this screen encoded distinct overlapping regions of NEDD9, a member of the Cas family (Fig. 6A). NEDD9 is a key player in the regulation of cell shape and cell migration and has recently emerged as a key player directing EMT in melanoma, lung and breast cancer (Kim et al., 2006, Izumchenko et al., 2009, Miao et al., 2013, Kondo et al., 2012, Little et al., 2014).
All four of the NRAGE-binding clones contained a NEDD9 fragment that started at amino acid 637 and ended at the C-terminus of the protein (amino acid 834), and Fig. 6B shows that a GST–NEDD9 fusion protein containing this 198-amino-acid fragment was capable of associating with NRAGE expressed by in vitro translation, indicating that NRAGE directly binds to this NEDD9 fragment. To identify the minimal region within NEDD9 required for NRAGE binding, pulldowns were performed using progressively smaller fragments of NEDD9 and, from this, a fragment of NEDD9 stretching from amino acids 704 to 765 emerged as the NRAGE interaction domain (Fig. 6B). Examination of the primary sequence of this 62-amino-acid region revealed a putative helix-loop-helix domain. To test the relevance of the helical domains in this region, we disrupted each of the two helices by substituting proline residues in helix 1 (G722P) or in helix 2 (G744P). The G722P substitution had no effect on the NRAGE–NEDD9 interaction, whereas the G744P mutation abolished the interaction between the two proteins (Fig. 6C). Taken together, these data indicate that the helix-loop-helix present in NEDD9 is required to bind to NRAGE and that the second helix of this domain plays a crucial role in their association.
To determine whether the NRAGE–NEDD9 interaction occurred in mammalian cells and to identify the domain in NRAGE that bound to NEDD9, we fused amino acids 637–834 of NEDD9 to GST and then expressed this with FLAG-tagged NRAGE, or with several FLAG-tagged NRAGE deletion mutants (shown schematically in Fig. 7A), in HEK293T cells and performed pull downs. Full-length NRAGE showed robust binding to the GST–NEDD9 fusion protein but not to GST alone (Fig. 7B). The main structural motifs present in this section of NRAGE are the interspersed repeat domain, the MAGE homology domain and a putative coiled-coil domain. Screening the NRAGE deletion mutants revealed that the NEDD9 interaction relied on a region within amino acids 666–775 that is C-terminal to the MAGE homology domain. Deletion of the coiled-coil domain or the C-terminal 48 amino acids reduced, but did not completely abrogate, the NRAGE–NEDD9 interaction. Therefore, much of this 110-amino-acid fragment of NRAGE is required to sustain a robust interaction with NEDD9.
We then asked whether NEDD9 is required for p75NTR-induced spreading, using NEDD9-specific siRNA to reduce its levels in COS7 cells (Fig. 8A). Interestingly, NEDD9 knockdown had no effect on cells transfected with EGFP alone but strongly inhibited cell spreading mediated by overexpression of full-length p75NTR or by the p75NTRICD (Fig. 8B,C). We conclude that NEDD9 is a downstream effector of a p75NTR–NRAGE pathway that mediates the cell spreading phenotype.
Finally, to determine whether p75NTR activates Rac1 through an NRAGE- and NEDD9-dependent pathway, we assessed Rac1 activity in COS7 cells using the fluorescence resonance energy transfer (FRET) biosensor described by Hodgson and colleagues (2010). p75NTR overexpression produced a robust increase in Rac1 activity and, significantly, the p75NTR-dependent increase was blocked when NRAGE or NEDD9 were depleted using siRNA (Fig. 8D). Taken together, these findings demonstrate that NRAGE and NEDD9 function downstream of p75NTR to activate Rac1.
DISCUSSION
Here, we show that p75NTR induces cell spreading through activation of the small GTPase Rac1. We demonstrate that p75NTR must be cleaved in a proteolytic process involving ADAM17 and γ-secretase for this effect to be manifested, and show that the p75NTRICD is the relevant signaling moiety in this context. We demonstrate that NRAGE participates downstream of the p75NTRICD in producing the cell spreading phenotype and identify NEDD9 as a novel NRAGE-binding protein that participates in this cascade. Thus, our data suggest that p75NTR-dependent cell spreading is dependent on generation of the p75NTRICD, which in turn drives NRAGE- and NEDD9-dependent activation of Rac1. Although generation of the p75NTRICD, NRAGE signaling and Rac1 activation have all been proposed to play important roles downstream of p75NTR (Salehi et al., 2002, Harrington et al., 2002, Domeniconi et al., 2005, Kenchappa et al., 2006, Bertrand et al., 2008, Wang et al., 2008, Kenchappa et al., 2010, Ceni et al., 2010, Tep et al., 2012, Matusica et al., 2013), this is the first study that links these events in a single cascade.
RIP is a conserved and well-established mechanism that affects the function of diverse membrane-anchored proteins, such as APP, Notch and Delta. RIP of p75NTR is characterized by the dual cleavage of the receptor by ADAM17 and γ-secretase, which then releases the p75NTRICD into the cytosol (Kanning et al., 2003, Zampieri et al., 2005). Our data demonstrate that ADAM17 is responsible for the initial cleavage of p75NTR in COS7 cells and show that this is a prerequisite for subsequent γ-secretase-dependent proteolysis and release of the p75NTRICD. Because knockdown of ADAM17 inhibited production of the p75NTRICD and prevented cell spreading in cells transfected with full-length p75NTR, but did not block cell spreading of cells transfected with the p75NTRICD, we conclude that the p75NTRICD is the active signaling component required for Rac1 activation and cell spreading.
In previous work, we have shown that NGF-dependent TrkA (also known as NTRK1) activation activates an Erk-dependent signaling pathway to generate the p75NTRICD, which in turn facilitates NGF-dependent Akt signaling and cell survival (Kommaddi et al., 2011). Consistent with this, another recent study has shown that p75NTR cleavage facilitates Akt signaling and Erk signaling in sympathetic neurons (Matusica et al., 2013). Other works have shown that p75NTR cleavage is necessary for BDNF-induced sympathetic neuron death, for myelin-associated-glycoprotein-induced growth cone collapse (Kenchappa et al., 2006; Domeniconi et al., 2005) and for glioma migration (Wang et al., 2008). Here, we show that the addition of exogenous neurotrophin has no effect on p75NTR-induced cell spreading, suggesting that this effect might be ligand independent. In sympathetic neurons, expression of ADAM17 mRNA is induced through a p75NTR-dependent signaling cascade but we found that, in COS7 cells, p75NTR overexpression does not alter ADAM17 mRNA levels. All four neurotrophins are expressed in COS7 cells, and it is conceivable that they bind to p75NTR and induce its cleavage in this setting. However, we found that incubation with REX, an antibody that blocks neurotrophin binding to p75NTR, had no effect on cell spreading. Therefore, consistent with our previous results (Ceni et al., 2010), we conclude that ligand binding to p75NTR is not required for the cleavage of the receptor. Determining the precise mechanisms that regulate p75NTR cleavage under physiological circumstances remains an interesting challenge.
NEDD9 (also called HEF1 or CasL) belongs to the Cas family of adaptor proteins that also includes p130Cas, Efs and HEPL (also known as BCAR1, Sin and CASS4, respectively). NEDD9 plays important roles in cell migration and cell adhesion (reviewed in Bouton et al., 2001, Guerrero et al., 2012), can stabilize focal adhesions and induce cell spreading (Bradshaw et al., 2011, Zhong et al., 2012, Baquiran et al., 2013), and has emerged as pro-metastasis factor in melanoma (Kim et al., 2006, Ahn et al., 2012), breast cancer (Izumchenko et al., 2009, Little et al., 2014), lung cancer (Miao et al., 2013, Kondo et al., 2012) and glioblastoma (Natarajan et al., 2006). This is the first study to establish a link between NRAGE and NEDD9 but it is interesting to note that other works have implicated NRAGE (Kumar et al., 2011) and NEDD9 (Kong et al., 2011, Ahn et al., 2012) in the negative regulation of E-cadherin–β-catenin and in EMT. Given that other studies have demonstrated that Rac1 plays an essential regulatory role at the E-cadherin–β-catenin complex (Pujuguet et al., 2003, Wu et al., 2008, Frasa et al., 2010, Zhu et al., 2012), it will be important to determine whether an NRAGE–NEDD9–Rac1 cascade functions in EMT or similar contexts.
The precise mechanism by which the p75NTRICD–NRAGE–NEDD9 cascade induces Rac1 activity is not known but activation of a Rac1 GEF seems likely to be important. Interestingly, a Rac1 GEF termed Trio has been reported to associate with an intact p75NTR–SorCS2 complex in hippocampal neurons and to facilitate Rac1 activity and neuronal growth. In this scenario, the association of proNGF with p75NTR results in Trio dissociation, which in turn results in Rac1 inactivation and hippocampal growth cone collapse (Deinhardt et al., 2011). Another Rac1 GEF, DOCK3, has been reported to directly bind to NEDD9 and activate Rac1-dependent melanoma migration (Sanz-Moreno et al., 2008).
In conclusion, we have used a novel cell-based assay to dissect the signaling events lying downstream of p75NTR. We show that generation of the p75NTRICD plays a crucial role in the activation of Rac1, identify NEDD9 as a NRAGE-interacting protein and show that a p75NTRICD–NRAGE –NEDD9 complex is required to effect morphological changes.
MATERIALS AND METHODS
Reagents and plasmids
Polyclonal antibodies directed against p75NTR and NRAGE were as described previously (Barker et al., 1994, Salehi et al., 2000), antibodies against ADAM17 were purchased from Cedarlane Laboratories Ltd (E8404, Burlington, ON, Canada), antibodies against NEDD9 were obtained from Thermo Fisher Scientific (MA1-5784, Ottawa, ON, Canada) and the M2 antibody (directed against the FLAG epitope) was obtained from Sigma (F3165, Oakville, ON, Canada). The monoclonal antibody against β-actin was purchased from MP Biomedicals (691001, Irvine, CA). The REX antibody was provided by Louis Reichardt (University of California at San Francisco, CA). NGF and BDNF were obtained from Alomone laboratories (Jerusalem, Israel). Compound XXI, ilomastast (GM6001) and epoxomicin were from Calbiochem (San Diego, CA); batimastat (BB94) and NSC 23766 were from Tocris Bioscience (Ellisville, MO). Laminin was purchased from BD Biosciences (Mississauga, ON, Canada). Poly-D-lysine (PDL) was obtained from Sigma (Oakville, ON, Canada). Rhodamine-conjugated wheat germ agglutinin (WGA) was purchased from Vector Laboratories (Burlington, ON, Canada) and horseradish-peroxidase-conjugated secondary antibodies were purchased from Jackson ImmunoResearch (West Grove, PA). All cell culture reagents were from Wisent Bioproducts (Saint Bruno, QC, Canada). Dako anti-fading mounting medium was purchased from Cedarlane Laboratories Ltd (Burlington, ON, Canada). The Rac biosensor was obtained from Addgene (Cambridge, MA) and was as described previously (Hodgson et al., 2010). Glutathione plasmids encoding full-length p75NTR, cleavage-resistant p75NTR and the p75NTRICD were as described previously (Kommaddi et al., 2011). Plasmids encoding RacN17 and RhoAN19 were kindly provided by Dr Peter McPherson (McGill University, QC, Canada).
Cell culture and transfection
COS7 cells and HEK293T cells were maintained in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum, 2 mM L-glutamine and 100 mg/ml penicillin-streptomycin, under 5% CO2 at 37°C. COS7 cells were co-transfected with EGFP cDNA (0.5 µg) in the absence or presence of plasmids encoding the various p75NTR isoforms (2 µg) using the calcium phosphate transfection method. For experiments involving dominant-negative Rac (RacN17) or RhoA (RhoAN19), COS7 cells were co-transfected with p75NTR plasmids in the absence or presence of plasmids encoding the dominant-negative GTPases, with EGFP (0.5 µg) co-transfected in all cases. Cells transfected with the different constructs were maintained for 48 hours at 37°C before either plating on coverslips for the spreading assay or lysing in Laemmli sample buffer for analysis by immunoblotting.
NRAGE, NEDD9 and ADAM17 siRNA sequences directed against the respective simian mRNAs were designed using the Invitrogen Stealth RNAiTM siRNAs prediction algorithm (specific sequences available on request). For knockdown experiments, cells were transfected with p75NTR constructs using the calcium phosphate transfection and, 48 hours later, were transfected with the different siRNAs in antibiotic-free medium using Lipofectamine 2000 (Invitrogen, Carlsbad, CA), as per the manufacturer's instructions. A non-specific siRNA was used for control knockdowns. Cells were maintained for 48 hours at 37°C then plated on coverslips for the spreading assay or lysed in Laemmli sample buffer for analysis by immunoblotting.
Cell spreading assay
Glass coverslips (12 mm, Fisherbrand, Fisher Scientific, Ottawa, ON, Canada) were coated for 30 minutes with poly-D-lysine (0.5 µg/ml), washed with sterile water, coated with laminin (0.5 µg/ml) for 2 hours at 37°C and again washed with sterile water. Transfected COS7 cells were plated on coverslips (4000 cells/slip) then incubated at 37°C under 5% CO2 for 24 hours. Medium was then removed and replaced with 4% paraformaldehyde in PBS for 15 minutes at 37°C. After washing with PBS (three 5-minute washes), cells were incubated with Rhodamine-tagged WGA (5 µg/ml) in PBS for 10 minutes and then washed with PBS (two 5-minute washes) and with water (one 5-minute wash). Coverslips were mounted in anti-fading mounting media (Dako), and kept at 4°C until imaging was performed. Imaging was performed using a 40× objective (NA 1.4) on a Zeiss Axio observer fluorescent inverted microscope equipped with Xenon illumination, and images were captured using Zen software (Zeiss) with an AxioCam MRm Rev.3 camera. The cell surface area of GFP-expressing cells was quantified with the NIH ImageJ software. At least 100 cells were counted per condition in each experiment.
In the case of treatment with GM6001, BB94 and compound XXI, cells were treated with the different compounds for 1 hour prior to fixation and staining. For NGF or BDNF, cells were serum starved for 2 hours then treated with the different neurotrophins (25 ng/ml) for 1 hour or 24 hours prior to fixation and staining. In the case of the REX treatment, cells were treated with the REX antibody (1∶100) for 24 hours prior to fixation and staining.
RNA extraction and RT-PCR
mRNA was extracted from COS7 cells using Qiagen RNeasy Mini Kit according to the manufacturer's instructions (Valencia, CA). cDNA was produced using the Ominscript RT kit (Qiagen) with random hexamers (GE Healthcare, Mississauga, ON, Canada) as primers. PCR was then performed using the GoTaq green master mix reagent (Fisher, Ottawa, ON, Canada) with primers targeting the simian neurotrophins NGF, BDNF, NT-3 and NT-4, as well as the neurotrophin receptors, p75NTR, TrkA, TrkB and TrkC. In the case of ADAM17 mRNA expression, COS7 cells were transfected with p75FL, CR or ICD, or left untransfected, and mRNA was extracted as described above. PCR was then performed using GoTaq green master mix reagent with primers against simian ADAM17. The primer sequences are available upon request. The PCR run was performed at an annealing temperature of 55°C for 35 cycles in the case of neurotrophins and neurotrophin receptors and 28 cycles for ADAM17. Actin was used as an internal control. PCR products were run on a 1.5% agarose gel and bands were visualized with ethidium bromide.
Yeast two-hybrid screening
NRAGE-interacting proteins were identified through a cytosolic yeast two-hybrid screen. For this, the NRAGE open reading frame was cloned into the pSOS vector to produce a NRAGE–SOS fusion that was then expressed in a cdc25H yeast strain. Screening was performed using a library of human fetal brain cDNAs cloned into the pMyr vector, which anchors fusion proteins to the yeast plasma membrane, as per the manufacturer's instructions (Stratagene/Agilent, CA). From 2.5×106 clones analyzed, 70 supported growth of the cdc25H yeast strain at 37°C and four of these contained distinct fragments of human NEDD9 cDNAs. The longest of these encoded amino acids 637–834, which represents the terminal 198 amino acids of NEDD9.
In vitro translation and pulldown experiments
The 198-amino-acid NEDD9 fragment obtained from the yeast two-hybrid screen was cloned into a pGEX4-1 to produce a GST fusion protein. This region of NEDD9 contained a putative helix-loop-helix, and deletion and site-directed mutants of this region were produced using PCR overlap. Full-length NRAGE was produced using an in vitro transcription-translation kit (Promega, WI) and pulldowns were performed using the GST–NEDD9 fragments using the methodology described previously (Salehi et al., 2000). Levels of NRAGE or its deletion fragments that associated with the GST–NEDD9 fragment were determined by immunoblotting with anti-NRAGE antisera.
To identify the region of NRAGE that bound to NEDD9, the GST fusion containing the 198-amino-acid NEDD9 fragment (amino acids 637–834) was transferred to a mammalian expression vector and co-expressed with FLAG-tagged NRAGE or FLAG-tagged NRAGE deletion mutants in HEK293T cells, using the calcium phosphate transfection. Cells were lysed in NP40 buffer (10 mM Tris-HCl pH 8.0, 150 mM NaCl, 1% NP40, 10% glycerol) after 24 hours and then incubated with 20 µl glutathione-conjugated beads (GE Healthcare, Baie d'Urfe, QC, Canada) for 1 hour at 4°C. Beads were washed three times in NP40 buffer, resuspended in Laemmli sample buffer, and incubated at 100°C for 5 minutes. Levels of NRAGE or its deletion fragments that associated with the GST–NEDD9 fragment were determined by immunoblotting with M2.
Western blot analysis and immunoblotting
Cells were harvested in 2× Laemmli sample buffer and boiled for 5 minutes prior to loading on SDS-PAGE gels. In the case of ADAM17 detection, BB94 (0.2 µM) was added to the 2× Laemmli sample buffer before lysing the cells. For p75NTRICD detection, cells were pre-treated with epoxomicin (1 µM) for 6 hours before lysis to block proteasomal degradation of the ICD, as described previously (Ceni et al., 2010). After SDS-PAGE and transfer to nitrocellulose, membranes were rinsed in PBS then blocked in TBST (10 mM Tris-HCl pH 8.0, 150 mM NaCl, 2% Tween 20) that was supplemented with 5% (w/v) dried skimmed-milk powder. Primary and secondary antibody incubation was performed in TBST containing 2.5% (w/v) dried skimmed-milk powder blocking solution, with primary incubations performed overnight at 4°C and secondary incubations performed for 1 hour at room temperature. Membranes were extensively washed in TBST after each incubation. Immunoreactive bands were detected using enhanced chemiluminescence solution kit (Perkin-Elmer Life Sciences, Norwalk, CT), as per the manufacturer's instructions.
Rac activity using FRET
Rac activity was assessed using the Rac FRET biosensors described previously (Hodgson et al., 2010). Briefly, the Rac FLAIR biosensor consists of a dual chain in which the donor CFP and the acceptor YFP are on different chains. The ECFP is replaced by CyPet (CyPet–Rac1) and EYFP is replaced by pYPet–PBD (PAK binding domain). Once Rac1 is activated, PBD interacts with Rac1 and this interaction results in a FRET signal.
COS7 cells were transfected or not with p75FL, together with CyPet–Rac1 and pYPet–PBD (2 µg) using calcium phosphate transfection and were then incubated for 48 hours at 37°C under 5% CO2. In order to correct for donor and acceptor bleed through, cells were transfected with either of the biosensors (CyPet–Rac1 or pYPet–PBD). For NRAGE and NEDD9 siRNA transfection, cells were transfected or not with p75FL as well as the Rac biosensors and, 48 hours later, NRAGE and NEDD9 were knocked down with NRAGE and NEDD9 siRNA transfected into cells using Lipofectamine 2000 in antibiotic-free medium. Cells were kept for 48 hours at 37°C under 5% CO2, then plated on glass coverslips precoated with poly-D-lysine for 30 minutes and laminin for 2 hours at a density of 10,000 cells/ml. Cells were kept for 24 hours at 37°C under 5% CO2, then fixed with 4% paraformaldehyde for 15 minutes at 37°C. They were then washed with PBS (two 5-minute washes) and sterile water (one 5-minute wash) and mounted on slides. Slides were kept at 4°C until imaging.
Statistical analysis
Statistical analyses were performed using one-way ANOVA and Tukey post-hoc analysis. All values are expressed as the mean±s.e.m.
Acknowledgements
We thank Vincent Soubannier for assistance with FRET and comments on the manuscript, Genevieve Dorval for help with transfections and Inna Ermeichouk for help in imaging and quantification.
Author contributions
P.A.B. conceived of the project, and P.A.B., A.S. and M.Z. designed the overall experimental plan. M.Z. performed all of the experiments except for the yeast two-hybrid screen, NEDD9 and MAGE-D1 mutagenesis and related interaction assays, which were performed by A.S. V.R. collected, analyzed and quantified microscopy data. M.Z. and P.A.B. wrote the manuscript.
Funding
This project was supported by the Canadian Institute of Health Research [grant number MOP62827].
References
Competing interests
The authors declare no competing or financial interests.