|Home | About | Journals | Submit | Contact Us | Français|
Numerous physiological functions, including a role as a cell surface receptor, have been ascribed to Alzheimer’s disease-associated amyloid precursor protein (APP). However, detailed analysis of intracellular signaling mediated by APP in neurons has been lacking. Here, we characterized intrinsic signaling associated with membrane-bound APP C-terminal fragments, which are generated following APP ectodomain release by α- or β-secretase cleavage. We found that accumulation of APP C-terminal fragments or expression of membrane-tethered APP intracellular domain results in adenylate cyclase-dependent activation of PKA and inhibition of GSK3β signaling cascades, and enhancement of axodendritic arborization in rat immortalized hippocampal neurons, mouse primary cortical neurons and mouse neuroblastoma. We discovered an interaction between BBXXB motif of APP intracellular domain and the heterotrimeric G-protein subunit GαS, and demonstrate that GαS coupling to adenylate cyclase mediates membrane-tethered APP intracellular domain-induced neurite outgrowth. Our study provides clear evidence that APP intracellular domain can have a non-transcriptional role in regulating neurite outgrowth through its membrane association. The novel functional coupling of membrane-bound APP C-terminal fragments with GαS signaling identified in this study could impact several brain functions such as synaptic plasticity and memory formation.
Alzheimer’s disease (AD) is pathologically characterized by the cerebral deposition of β-amyloid peptides (Aβ) in senile plaques. Aβ is released by the sequential proteolytic processing of amyloid precursor protein (APP), a type I transmembrane protein. Cleavage of full-length APP (APP-FL) by α- or β-secretase releases the entire ectodomain, leaving behind membrane bound C-terminal fragments (APP-CTF), made of the transmembrane and cytoplasmic domains (Lichtenthaler et al., 2011). While APP metabolism and contribution of Aβ to AD pathology has been the focus of intense investigation, the normal biological function(s) of APP in the nervous system is still not completely understood (Turner et al., 2003; Thinakaran and Koo, 2008; Guo et al., 2011). It has been proposed that APP can affect synaptic function by its dual roles via its cell adhesive properties or through its putative receptor-like intracellular signaling (Ando et al., 1999; Turner et al., 2003; Soba et al., 2005; Thinakaran and Koo, 2008). Moreover, several lines of evidence reveal that APP expression modulates neurite outgrowth in neuroblastoma cells and neurons (Allinquant et al., 1995; Perez et al., 1997; Ando et al., 1999; Small et al., 1999; Leyssen et al., 2005; Young-Pearse et al., 2008; Hoe et al., 2009). Mice lacking APP expression show progressive loss of presynaptic terminals, reduced dendritic length, impairment of synaptic plasticity, and deficit in learning and memory (Turner et al., 2003; Zheng and Koo, 2006; Hoe et al., 2010; Aydin et al., 2011; Guo et al., 2011). However, the molecular mechanisms underlying the above observations largely remain undefined.
APP cytosolic domain possesses conserved sequence motifs responsible for complex network of protein-protein interactions (Muller et al., 2008; Suzuki and Nakaya, 2008; Schettini et al., 2010; Aydin et al., 2011), which could account for a variety of cellular functions mediated by APP. The present work focuses on the modulation of APP cytosolic tail-mediated intracellular signaling, which underlies neurite outgrowth. We have designed a membrane-tethered APP intracellular domain construct (referred to as mAICD) to allow us to activate, in a constitutive manner, putative signaling associated with APP-CTF. We report here that accumulation of APP-CTF or membrane tethering of APP cytosolic sequence stimulates neurite outgrowth in mouse N2a neuroblastoma cells, rat H19-7 immortalized hippocampal cells, and mouse cortical primary neurons. Expression of mAICD initiates a previously unrecognized signaling pathway that involves a novel association between APP intracellular domain and the heterotrimeric G protein subunit GαS. This functional coupling leads to steady-state increase of phosphorylated PKA substrates such as CREB and GSK3β, which are likely to impact neuronal morphology and function.
Compound E was generously provided by Dr. Todd E. Golde (University of Florida, Gainesville, FL) (Seiffert et al., 2000). KT5720 and MDL-12,330A were obtained from Biomol (Plymouth Meeting, PA). Tetrodotoxin was purchased from Alomone Labs (Jerusalem, Israel). Cell culture and transfection reagents were purchased from Invitrogen. Unless indicated, all other reagents were purchased from Sigma (St. Louis, MO).
Rabbit polyclonal antibodies CT11, CTM1, and PS1nt have been described (von Koch et al., 1997; Thinakaran et al., 1998; Vetrivel et al., 2008). The following antibodies were purchased from commercial sources: polyclonal phospho-[Ser133] CREB, polyclonal phospho-[Ser9] GSK3β and goat anti-mouse CD147 (clone T-18) (Santa Cruz Biotechnology Inc.); mAb HA [clone 6E2] and polyclonal phospho-[Ser/Thr] PKA antibody (Cell Signaling Technology Inc.); mAb anti-HA [clone 12CA5] (Roche Diagnostics); polyclonal anti-MAP2 and mAb anti-Flag [clone F3165] (Sigma); mAb GAPDH (Abcam); polyclonal dimethyl Histone H3 [Arg 17] (Upstate Biotech Inc.); and mAb 82E1, which recognizes an epitope within residues 1–16 of Aβ (IBL International). Alexa-488- and Alexa-555-conjugated secondary antibodies were purchased from Molecular Probes. IRDye 680 and IRDye 800CW-conjugated secondary antibodies were purchased from LI-COR Biosciences.
Mouse N2a neuroblastoma cells were cultured in 1:1 Dulbecco’s modified Eagle’s medium (DMEM)-Opti-MEM medium supplemented with 5% fetal bovine serum. N2a pools stably expressing mAICD, PS1-wt, and PS1-D385A, or harboring an empty vector (EV), were generated by retroviral infections as described previously (Onishi et al., 1996; Vetrivel et al., 2009). Briefly, retroviral supernatants collected 48 h after transfection of Plat-E cells were used to infect N2a cells in the presence of 8 µg/ml polybrene. Stably transduced cells were selected in the presence of 1 µg/ml puromycin and pooled for further analysis. H19-7 rat embryonic immortalized hippocampal cell line was cultured in DMEM supplemented with 10% fetal bovine serum (FBS) at 33°C as described previously (Eves et al., 1992; Wu et al., 2004). Immediately after transfection, proliferating H19-7 cells were incubated at 39°C in a supplemented DMEM containing 50 ng/ml basic fibroblast growth factor to initiate differentiation and neurite outgrowth (Eves et al., 1992; Wu et al., 2004). COS cells were maintained in DMEM supplemented with 10% FBS. Primary cultures of cortical neurons were generated from E16 male and female embryos as described (Parent et al., 2005), and maintained at 37°C in Minimal Essential Medium supplemented with 1% glutamine, 5% horse serum, 0.5% D-glucose, 0.15% HCO3 and nutrients, in a humidified 10% CO2 incubator. N2a and H19-7 were grown in 6-well plates on a poly-L-lysine coated 18 mm-glass coverslips, whereas cortical neurons were cultured on 0.1% polyethylenimine coated-glass coverslips. N2a and H19-7 cells were transiently transfected using LipoD293™ according to manufacturer’s protocol (SignaGen Laboratories). Transient transfection of 7–11 DIV neurons was performed in Neurobasal™ medium using Lipofectamine™ 2000. After 3 h, transfection medium was replaced by 50% original medium and 50% supplemented Minimal Essential Medium without serum. Sixteen to twenty-four hours after transfection, cells were treated with Compound E (10 nM) and/or various drugs, and cultured for another 24 h before analysis.
Overlap extension PCR was used to generate a cDNA that encodes mAICD or mALID1 by replacing the extracellular and transmembrane domains of APP or APLP1 with the sequence encoding the membrane-targeting MyrPalm motif of the Src family tyrosine kinase Lyn (MGCIKSKRKDNLNDDGVDN). The C-terminus of mAICD was tagged with CT11 epitope (RFLEERP) of APLP1 (von Koch et al., 1997). Similarly, sequences encoding CT11-tagged C-terminal 50 residues of APP (AICD) was generated by overlap PCR. The resulting cDNAs were cloned into pMXpuro retroviral vector (kindly provided by Dr. Toshio Kitamura, University of Tokyo, Japan). To generate membrane-tethered control construct (mCtl), the 47 amino acid APP intracellular domain in mAICD was replaced with residues 1–47 of the cyan fluorescent protein. To generate mAICD-mutAAA, the residues RHLSK (corresponding to amino acids 672–676 of APP695) in mAICD were mutated to AALSA. PCR amplified regions in all constructs was verified by sequencing. APP-C99-6Myc (a gift from Dr. Alison Goate, Washington University), APP-FL, APP α-site cleavage mutant (APP-F615P; numbering corresponds to APP695), β-site cleavage mutant (APP-M596V) were generated by PCR mutagenesis. Plasmids encoding HA-GαS-wt and palmitoylation-deficient HA-GαS-C3S have been described previously (Wedegaertner et al., 1993; Dupre et al., 2007). Plasmids encoding Flag epitope-tagged dopamine D1 receptor (plasmid 15464) and β1 adrenergic receptor (plasmid 14698) were obtained from Addgene (Cambridge, MA).
Total cell lysates for immunoblotting were prepared in 150 mM NaCl, 50 mM Tris-HCl, pH 7.4, 0.5% NP-40, 0.5% sodium deoxycholate, 5 mM EDTA, 0.25% SDS, 0.25 mM phenylmethylsulfonyl fluoride and a protease inhibitor cocktail (Sigma) and briefly sonicated on ice. Co-immunoprecipitation was modified from previously described protocols (Biederer et al., 2002; Gong et al., 2010). Briefly, cells were lysed in ice-cold lysis buffer (25 mM HEPES, pH 7.4; 5 mM MgCl2, 1% NP-40, 125 mM potassium acetate, 10% glycerol and protease inhibitor cocktail). Lysates were clarified by centrifugation at 13,000 rpm for 20 min at 4°C, and aliquots of the supernatants were adjusted to 500 µl with lysis buffer and incubated overnight with the indicated antibodies at 4°C. Immune complexes were captured by incubating with 40 µl of protein-G agarose beads (Thermo Scientific) for 2 h at 4°C, and washed 3 times with 500 µl of lysis buffer at 4°C. Bound proteins were eluted in Laemmli buffer and analyzed by immunoblotting along with an aliquot of the total input lysate.
N2a cells were homogenized in buffer A, containing 5 mM HEPES, 1.5 mM MgCl2, 10 mM KCl, 0.125% Triton, 0.25 mM phenylmethylsulfonyl fluoride and protease inhibitor cocktail at pH 7.9, and centrifuged at 1000 × g for 10 min. The nuclear pellet (P1) was washed with buffer B, containing 20 mM HEPES, 25% glycerol, 0.5 mM NaCl, 0.5 mM MgCl2, 0.5 mM EDTA, 0.25% Triton X-100, 0.25 mM PMSF and protease inhibitor cocktail, and subsequently sonicated. The supernatant was centrifuged at 114,000 × g for 30 min to yield a membrane pellet (P2) and supernatant (S2) containing soluble cytosolic proteins. Each fraction was resuspended in equal volume of buffer B and analyzed by immunoblotting with CT11 antibody. Histone H3, CD147 and GAPDH were used as nuclear, membrane and cytosolic markers, respectively. Lipid rafts from cultured cells were isolated as described previously (Vetrivel et al., 2004). Briefly, cells were lysed on ice in a buffer containing 0.5% Lubrol WX (Lubrol 17A17; Serva) and lysates were subject to centrifugation on discontinuous flotation density gradients and aliquot of fractions were analyzed by Western blotting.
Cells were maintained for 3 h in HEPES buffer prior to FSK stimulation (50 µM, 30 min at 37°C), as previously described (Barnes et al., 2008). To reduce neuronal activity, cortical cultures were serum deprived for 3 h using HEPES buffer supplemented with tetrodotoxin (1 µM). In order to examine the signaling cascade activation, pharmacological inhibitors were added 30 min before FSK stimulation (2 µM KT5720 or 10 nM MDL-12,330A). In order to stimulate or inhibit neurite extension, cells were exposed to 1 µM FSK, 10 nM MDL-12,330A, or 10 nM Compound E for 24 h. Following the treatment, cells were fixed for 30 min at 4°C with 4% paraformaldehyde/ 4% sucrose and permeabilized on ice with 0.2% Triton X-100 solution for 8 min. Fixed cells were incubated overnight at 4°C with antibodies of interest (1:20,000 CT11, 1:100 phospho-[Ser/Thr] PKA, 1:100 phospho-[Ser133] CREB, 1:250 phospho-[Ser9] GSK3β, 1:5,000 MAP2, 1:5,000 HA clone 6E2). Subsequently, cells were incubated with Alexa-488- and Alexa-555-conjugated secondary antibodies (1:250) for 1 h at room temperature, and mounted using Permafluor mounting medium (Thermo Scientific).
Labeled neurons were imaged using a motorized Nikon TE 2000 microscope and Cascade II:512 CCD camera (Photometrics, Tucson, AZ) using 20X, 60X (N.A. 1.49) or 100X (N.A. 1.45) objectives. Images were acquired as 200 nm z-stacks and processed using MetaMorph software (Molecular Devices Corporation). For visualization and analysis of colocalization, deconvolved z-stack images were generated using Huygens software (Scientific Volume Imaging). NeuronJ plugin (version 1.29) of ImageJ software was used to measure neurite number and length of cortical neurons (Abramoff et al., 2004; Deyts et al., 2009). Total neurite length was evaluated by manually tracing all neurites in each neuron. In neurons, total neurite area was evaluated using thresholding feature of MetaMorph software to isolate the cell of interest, followed by somatic area removal from YFP expressing cells. Similarly, total neurite area was evaluated in H19-7 and N2a cells through image subtraction of a predefined somatic area applied to each cells. Total axonal area was determined by subtraction of total dendritic area (as seen as MAP-2 labeling) from total neurite area (as seen as YFP fluorescence). To quantify the level of fluorescence, identical parameters and photomultiplicator values were used to acquire the images. Raw images were first set to the same threshold level to eliminate nonspecific fluorescence, and then the gray intensity level was determined and divided by the number of pixels per area (calculated from YFP fluorescence image). Colocalization was evaluated by the degree of fluorescence-overlapping area between two fluorophores using Pearson’s coefficient analysis from JACoP plugin of ImageJ.
Each experiment was performed using at least three independent sets of cultures. Data are presented as mean ± SEM. Statistical significance was determined by ANOVA Kruskal-Wallis test with independent post hoc Dunn’s multiple comparison analysis using GraphPad prism software (San Diego, CA). *P<0.05, **P<0.001 compared to empty vector (EV) transfected cells and #P<0.05 and ##P<0.001 compared to untreated cells within the same transfected conditions.
Proteolytic processing of APP-FL by α- and β-secretases generates membrane-bound APP-CTFs, which preferentially localize to lipid raft microdomains in cultured cell lines and in the brain [reviewed in (Cheng et al., 2007)]. These CTFs are subsequently cleaved by γ-secretase at the γ- and ε-sites, to release Aβ and AICD respectively, from the membrane [reviewed in (Thinakaran and Koo, 2008; Lichtenthaler et al., 2011)]. To investigate intracellular signaling mediated by APP and APP-CTF at the membrane, we generated a fusion protein, termed mAICD, in which the extracellular and transmembrane domains of human APP were replaced by 17 residues corresponding to the N-terminal myristoylation and palmitoylation (MyrPalm) motif of the Src family tyrosine kinase Lyn (Fig. 1a) (Zacharias et al., 2002). Fractionation of transfected mouse N2a neuroblastoma cells revealed membrane association of mAICD and a 47 amino acid control polypeptide with an N-terminal MyrPalm motif (mCtl) as expected, and confirmed the presence of AICD in the nuclear fraction (Fig. 1b). Sucrose density gradient fractionation analysis further revealed predominant membrane raft association of mAICD (Fig. 1c), which faithfully mimics preferential raft localization of APP-CTFs derived from processing of APP-FL in cultured cell lines and in mouse brain (Vetrivel et al., 2005). Thus, subcellular localization studies establish that mAICD is retained at the membrane, especially in membrane microdomains. Membrane microdomains are known to compartmentalize numerous membrane-associated receptor-activated signaling cascades, including neurotransmitter signaling (Allen et al., 2007). Consistent with the recruitment of signaling molecules, expression of mAICD in N2a cells significantly increased total neurite area as compared with empty vector control (EV). However, expression of APP-FL, AICD or membrane-tethered mCtl did not affect the level of neurite outgrowth (Fig. 1e, f).
To examine whether CTFs generated by processing of APP-FL by α- and β-secretases also stimulate neurite outgrowth, we incubated cells with the γ-secretase inhibitor Compound E (Seiffert et al., 2000), which produced an accumulation of APP-CTFs in cells transfected with APP-FL (Fig. 1d). This accumulation of APP-CTFs caused a marked increase of neurite extension (Fig. 1e, f). As expected, treatment with Compound E did not have noticeable effect on the steady-state levels of mCtl, AICD or mAICD. Expression of mCtl and AICD did not affect N2a cell morphology, and the treatment with Compound E had no additional influence of neurite outgrowth in cells expressing mAICD (Fig. 1e, f). Thus, enhanced neurite outgrowth in these experiments is entirely attributed to membrane accumulation of APP-CTF.
We evaluated the linear relationship between the level of mAICD expression and the total cellular area. As shown in Fig. 1g, increase of N2a area correlate significantly with the level of mAICD expression, either revealed by cotransfected YFP fluorescence (y=0.887x+0.678; r2=0.560; P<0.001) or immunostaining of tagged mAICD polypeptide (y=0.939x+0.709; r2=0.626; P<0.001). Expression levels of YFP alone did not show any significant correlation with cellular area (y=0.015x+0.972; r2=0.001; P=0.899). Taken together, our results support the idea that accumulation of membrane-tethered APP-CTF, either by inhibiting the γ-secretase-dependent processing of APP-CTF or by targeting APP intracellular domain to the membrane using the motifs that direct myristylation/palmitoylation, influences neurite formation.
Next we assessed whether membrane accumulation of APP intracellular domain affects neurite extension in primary mouse cortical neurons. We observed that MyrPalm motif containing polypeptides (mAICD and mCtl) localized to punctate membranous structures in the cell body as well as dendrites and axons, whereas AICD expression was mainly found in the nucleus and within larger caliber primary dendrites (Fig. 2a). As in N2a cells, expression of mAICD in pyramidal neurons significantly increased total neurite area as compared with EV or mCtl (Fig. 2b, 2c). Notably, expression of APP-FL and AICD did not affect the level of neurite outgrowth. Expression of mAICD caused a two-fold increase of the total neurite length (Fig. 2d; P<0.001) as well as neurite number per branches (Fig. 2e; P<0.001). Further analysis revealed that mAICD expression increased both the axonal and dendritic network (Fig. 2f and 2g; respectively) without an apparent bias towards either neuronal compartment (Fig. 2h). Taken together, these results provide the first evidence that membrane-tethered APP cytosolic domain affects neurite outgrowth in cultured neurons.
To investigate the signaling events associated with mAICD-induced neurite outgrowth we employed immortalized hippocampal H19-7 cell line, which exhibits neuronal phenotype, conditional proliferation, and a capacity for differentiation after cessation of division (Eves et al., 1992; Wu et al., 2004) (Fig. 3). As we observed in primary neurons and N2a neuroblastoma cells, expression of mAICD stimulated neurite extension in hippocampal H19-7 cells (Fig. 3a top panel). Quantitative analysis of total neurite area using YFP fluorescence, revealed that neurite outgrowth was more than two fold increased in mAICD expressing cells, whereas expression of APP-FL had no effect over vector control (Fig. 3b).
We tested whether mAICD expression leads to activation of cAMP/PKA signaling because basic neuronal morphology as well as initial steps in neuronal elongation could be dependent on the activation of cAMP/PKA pathway (Sanchez et al., 2001; Hutchins, 2010; Shelly et al., 2010). When transfected cells were immunostained with an antibody that specifically recognizes substrates phosphorylated by PKA (pPKA substrates; characterized by Arg at position −3 relative to the phosphorylated Ser or Thr) (Gronborg et al., 2002; Barnes et al., 2008), we observed a significant increase of pPKA substrate staining intensity in cells expressing mAICD relative to cells transfected with EV and APP-FL expressing cells (Fig. 3a bottom panel, c; EV: 1.00±0.05; APP-FL: 1.12±0.05; mAICD: 3.18±0.23; P<0.001). Following treatment with forskolin (FSK), an adenylate cyclase activator, the intensity of pPKA substrates staining in EV and APP-FL markedly increased by more than three-fold, while no further increase was observed in mAICD expressing cells (Fig. 3c).
To examine putative signaling pathways associated with accumulation of membrane-bound APP-CTF and enhanced neurite outgrowth, we also immunostained N2a cells with Ser/Thr pPKA antibody. As observed in H19-7 cells, expression of mAICD resulted in enhanced phosphorylation of PKA substrates in N2a cells (Fig. 4a, b). When APP-FL expressing cells were incubated with Compound E, a treatment that caused accumulation of APP-CTF and increase of neurite outgrowth, the intensity of pPKA substrate staining also increased by more than two-fold (Fig. 4b; P<0.001). However, as expected from the lack of neurite outgrowth (outlined above), Compound E treatment failed to affect PKA-dependent signaling in mCtl or cells expressing AICD. Furthermore, treatment with Compound E did not significantly show any additive effect in the level of pPKA substrate intensity in mAICD expressing cells as compared to untreated cells, which is in agreement with the neurite outgrowth data.
To establish the involvement of cAMP/ PKA signaling in the observed increase of pPKA substrate staining intensity in mAICD expressing cells, we employed pharmacological inhibition of PKA and adenylate cyclase. As shown in Fig. 4c, we observed that acute incubation with KT5720, a specific PKA inhibitor, abolished the cAMP dependent increase of pPKA staining in FSK treated cells transfected with EV, mCtl or APP-FL. Similarly, incubation with MDL-12,330A, a specific adenylate cyclase inhibitor also abrogated FSK-mediated increase of pPKA staining (Fig. 4d). More interestingly, mAICD expression-related increase of pPKA staining was also abolished by incubation with either KT5720 (Fig. 4c; P<0.05) or MDL-12,330A (Fig. 4d; P<0.05).
We also quantified the levels of phosphorylated cAMP Response Element Binding protein-1 (pCREB), which is one of the key neuronal targets of activated PKA and axonal outgrowth (Lonze and Ginty, 2002). At steady state, we observed higher levels of pCREB staining in N2a cells expressing mAICD as compared with those expressing APP-FL or the EV control (Fig. 4e, f; p<0.001). Following acute stimulation with FSK, the intensity of pCREB staining was enhanced in EV and APP-FL expressing cells (Fig. 4f). No further increase in pCREB-staining was found in mAICD expressing cells following FSK stimulation. Consistent with the involvement of membrane-bound APP-CTFs in activating PKA-dependent signaling, treatment with Compound E increased pCREB levels by more than four-fold in cells expressing APP-FL (Fig. 4f; p<0.001), an effect that was not observed in EV or mAICD expressing cells. Western blot analysis confirmed that levels of pCREB are increased by about two-folds in FSK treated cells as well as in cells stably expressing mAICD as compared with EV (Fig. 4g, h). As seen by immunostaining, we observed no further increase of pCREB in FSK treated mAICD cells (Fig. 4h). Collectively, these results confirm the involvement of pPKA substrates in the sequence of events leading up to enhanced neurite outgrowth following mAICD expression or the accumulation of membrane-tethered APP-CTF generated from processing of APP-FL.
It has been shown that activation of cAMP/PKA and GSK3β signaling pathways are both necessary for the initial elongation of neurites in neuronal cell types (Sanchez et al., 2001; Hutchins, 2010). Interestingly, Ser9 in GSK3β is a physiological substrate of PKA, and phosphorylation at this residue leads to inactivation of GSK3β (Fang et al., 2000; Jope and Johnson, 2004; Shelly et al., 2010). Since GSK3β activation is associated with neurite retraction, and inhibition of GSK3β promotes axonal elongation (Yoshimura et al., 2006; Hur and Zhou, 2010; Shelly et al., 2010), we considered the possibility that neurite outgrowth stimulated by expression of membrane-tethered APP-CTF might involve a crosstalk between PKA and GSK3β signaling cascades. To assess GSK3β activity in individual cells, we stained transfected N2a cells and cultured cortical neurons using an antibody that recognizes inactive Ser9 phosphorylated GSK3β (pGSK3β). Overexpression of mAICD caused a two- to five-fold increase in pGSK3β levels in the soma and neurites of pyramidal neurons and N2a neuroblastoma cells, respectively, whereas expression of AICD had no effect (Fig. 5). FSK-induced activation of cAMP/PKA was also associated with an increase of pGSK3β fluorescence intensity in mCtl and AICD expressing cells consistent with previous reports (Fang et al., 2000; Jope and Johnson, 2004; Shelly et al., 2010), but there was no further increase in cells expressing mAICD (Fig. 5b, d). More importantly, PKA inhibitor (KT5720) and adenylate cyclase inhibitor (MDL-12,330A) completely abolished mAICD-induced increase in pGSK3β levels in N2a cells (Fig. 5a, b; untreated: 4.89±0.37; KT5720: 1.16±0.07; MDL-12,300A: 1.15±0.09), as well as in cortical neurons (Fig. 5e, f; untreated: 1.98±0.03; MDL-12,300A: 1.05±0.08). Consistent with the immunofluorescence results, Western blot analysis of stably transduced N2a pools confirmed that the basal levels of pGSK3β were about five-fold higher in stable mAICD pools as compared with the EV cells (Fig. 5c, d). Furthermore, treatment with FSK induced phosphorylation of GSK3β by more than four-fold in EV cells, whereas there was no further increase in cells stably expressing mAICD. Taken together, our results support the idea that targeting APP intracellular domain at the membrane is closely associated with GSK3β inactivation through adenylate cyclase/PKA cascades. Since inhibition of GSK3β promotes axon outgrowth (Yoshimura et al., 2006; Hur and Zhou, 2010), these results indicate that control of GSK3β activity via Ser9 phopshorylation might also contribute to the increase of neurite formation seen in mAICD expressing cells.
Based on the results detailed above, we predicted that accumulation of membrane-tethered APP-CTF or expression of mAICD is associated with activation of adenylate cyclase signaling cascade, which could be required for neurite outgrowth. Accordingly, we investigated the requirement of adenylate cyclase for neurite outgrowth associated with membrane-bound APP-CTFs in N2a cells and cortical neurons by treating cells for 24 h with an adenylate cyclase activator (FSK) or the inhibitor (MDL-12,330A). As expected from previous reports (Sanchez et al., 2001; Hutchins, 2010; Shelly et al., 2010), chronic activation of cAMP pathway enhanced neurite outgrowth in control N2a cells (Fig. 6a, b) and cortical neurons (Fig. 6c, d). However, treatment with FSK did not show any additive effect in cells expressing mAICD (Fig. 6b, d), suggesting that mAICD expression acts upstream of adenylate cyclase activation and that cAMP-dependent signaling might have reached maximal physiological outcome. As expected, treatment with the adenylate cyclase inhibitor MDL-12,330A eliminated FSK-induced neurite extension in EV transfected N2a cells and neurons (Fig. 6b and 6d; respectively). More importantly, mAICD expression-induced neurite extension was abolished following MDL-12,330A treatment in N2a cells (Fig. 6a bottom right and 6b; untreated: 3.04±0.26; MDL-12,330A: 1.05±0.09; P<0.001) and in cortical neurons (Fig. 6c bottom right and 6d; untreated: 2.56±0.18; MDL-12,330A: 1.00±0.07; P<0.001). Together, these results indicate that mAICD-induced neurite outgrowth takes place downstream of adenylate cyclase activation. Therefore, it becomes apparent that activation of adenylate cyclase is necessary and sufficient to initiate morphological changes associated with membrane accumulation of APP intracellular domain.
Next, in order to characterize adenylate cyclase associated signaling molecule(s) that function downstream of APP-CTF, we focused our attention on guanine nucleotide-binding protein G(s) subunit alpha (GαS), the canonical signal transducer that links receptor-ligand interactions with the activation of adenylyl cyclase. In stably transfected N2a cells, mAICD localized to cellular membranes as clusters. Transiently expressed wild-type GαS (GαS-wt) also localized as clusters on the plasma membrane and other organelles, and strongly overlapped with mAICD, especially in neurite extensions (Fig. 7a). Interestingly, a dominant negative GαS mutant that could not be palmitoylated (GαS-C3S) showed poor colocalization with mAICD as visualized by line-scan histogram (Fig. 7a). Pearson’s correlation coefficient analysis confirmed greater colocalization of mAICD with GαS-wt as compared to GαS-C3S mutant (GαS-wt: 0.591±0.030; GαS-C3S: 0.369±0.047, P<0.01). In agreement with these findings, mAICD efficiently co-immmunoprecipitated with GαS from non-denaturing lysates of transfected N2a cells (Fig. 7b), under conditions that preserved the interaction of GαS with D1-dopamine and β1-adrenergic receptors, two canonical receptors that signal through GαS (Fig. 7c). Consistent with poor colocalization, GαS-C3S mutant failed to co-immunoprecipitate with mAICD (Fig. 7b). Finally, analysis of neurite outgrowth revealed that co-expression of GαS-wt had no effect whereas expression of the dominant-negative GαS-C3S mutant significantly inhibited mAICD-induced neurite extension in N2a cells (Fig. 7d; mAICD+GαS-wt: 2.80±0.15; mAICD+GαS-C3S: 1.46±0.08, P<0.001) and in primary neurons: (Fig. 7e; mAICD+GαS-wt: 2.82±0.11; mAICD+GαS-C3S: 1.20±0.07; P<0.001). Together, these findings indicate that GαS protein coupling is an essential part of the signaling pathway that translates mAICD expression to promote neurite outgrowth.
To further establish a direct link between APP C-terminal domain and GαS protein, we scanned the amino acid sequence of APP intracellular domain for the presence of specific motifs such as BBXXB, BXBXB, BBXB, BXBB, or BXB (where B represents a basic amino acid residue and X a non-basic residue), which have been identified as G-protein binding sites within the intracellular domains of G-protein coupled receptors (Okamoto et al., 1990; Wu et al., 1995; Pauwels et al., 1999; Zhou and Murthy, 2003). APP intracellular domain possesses a single highly conserved BBXXB motif: 672RHLSK676 (APP695 isoform numbering). To test the possibility that RHLSK sequence on APP C-terminal domain is the critical site of interaction with GαS subunit, we introduced alanine substitution of the basic residues in mAICD to convert the BBXXB motif to AALSA (named mAICD-mutAAA; Fig. 8a). Co-immunoprecipitation experiments confirmed lack of interaction between GαS and mAICD-mutAAA (Fig. 8b). Consistent with its inability to associate with GαS, expression of mAICD-mutAAA had no effect on neurite extension in N2a cells (Fig. 8c, d; P<0.001) and cortical neurons (Fig. 8e, f; P<0.001). These results clearly establish that APP-CTF possess a sequence motif for interaction with GαS, and functional coupling with GαS transduces signaling downstream of mAICD expression to induce neurite outgrowth.
APP possesses two homologs (APLP1 and APLP2) that undergo secretase cleavages similar to APP cleavage releasing also an intracellular cytoplasmic domain [reviewed in (Aydin et al., 2011; Guo et al., 2011)]. It has been suggested that these homolog might possess very similar physiological properties based on their structural similarities, which was also supported by gene knock down studies. Additional studies indicate that APP and its homolog may form heterocomplexes that required the presence of each other to favor their cell-adhesive property (Soba et al., 2005). To determine the selectivity of mAICD-induced neurite outgrowth and associated GαS protein signaling, we explored the possibility that APLP1 homolog might possess similar property. Indeed, we found that N2a cells and cortical neurons exert as well an increase of neurite outgrowth following expression of APLP1 membrane-tethered intracellular domain (named mALID1; Fig. 9a–d). Statistical analysis revealed a significant increase of total neurite area in N2a cells (Fig. 9a–b; P<0.001) and mouse cortical neurons (Fig. 9c–d; P<0.001) expressing mALID1 as compared to EV or mCtl expressing cells. In agreement with these findings, we observed that mALID1 also co-immunoprecipitated with GαS couple proteins from non-denaturing lysates of transfected cells (Fig; 9e). Collectively, these results strongly support that APLP1-CTF promotes as well neurite outgrowth through its membrane association with GαS protein coupling.
Next, in order to establish the critical role of APP secretase activity and APP-CTF accumulation in APP function, we first examined neurite outgrowth induced by accumulation of APP-CTF through pharmacological γ-secretase inhibition (Fig. 10a–d). As described earlier (Fig. 1e and f), inhibition of γ-secretase using Compound E stimulated neurite outgrowth in N2a cells and cortical neurons expressing APP-FL. This increase was abrogated by treatment with MDL-12,330A in N2a cells (Fig. 10a bottom and 10b) and cortical neuron cultures (Fig. 10c bottom and 10d). Thus, as in the case of mAICD expression discussed above (Fig. 6), adenylate cyclase activity is required for the activation of neurite outgrowth following accumulation of APP-CTF derived from proteolytic processing of APP-FL (Fig. 10a–d).
In order to substantiate the finding from pharmacological inhibition studies, we performed additional experiments using N2a cells stably expressing an experimental γ-secretase loss-of-function PS1 mutant (PS1-D385A). As expected from previous reports (Wolfe et al., 1999; Vetrivel et al., 2005), stable PS1-D385A cells accumulate higher levels of APP-CTFs as compared with stable PS1-wt cells, following transient transfection with APP-FL (Fig. 10e). This APP-CTF accumulation parallels an increase of neurite outgrowth (Fig. 10f, g). Interestingly, expression of APP-C99 (comparable to the level of mAICD expression) was sufficient to induce neurite outgrowth in PS1-wt expressing cells. Nevertheless, the stimulation of neurite extension following APP-C99 transfection was comparable in PS1-wt and D385A cells, despite the relatively higher accumulation of APP-C99 in D385A cells. This observation, along with the data from mAICD with FSK experiments described above, is consistent with the notion that the signaling mediated by APP intracellular domain has reached a point of saturation.
Processing of APP-FL by α- and β-secretases generates α- and β-CTFs, respectively, differing at their extreme N-terminus but sharing the same transmembrane and intracellular domains. In order to test whether accumulation of either α- or β-CTFs differentially regulates neurite extension, we stably expressed APP-FL β-site cleavage mutant (M596V) or APP-FL α-site cleavage mutant (F615P) in N2a cells and examined their effect on neurite outgrowth. As shown in Fig. 10h, the overall levels of APP-CTFs were comparable in cells expressing wt APP-FL and α- or β-secretase cleavage site mutants at basal levels and following inhibition of γ-secretase activity. However, as expected from previous studies (Sisodia, 1992; Citron et al., 1995; Vetrivel et al., 2011), fractionation of lysates on high resolution Tris-Tricine gels reveal that β-CTFs were not generated by APP-M596V mutant; whereas APP-F615P mutant generated higher levels of β-CTF at the expense of α-CTF production (Fig. 10i). Further analysis of N2a cells revealed that expression of APP-FL wt or APP-FL mutants uniformly stimulated neurite outgrowth following treatment with Compound E (Fig. 10j). Collectively, these results indicate that α- or β-secretase processing of APP does not differentially stimulate neurite outgrowth in N2a cells.
In this study, we investigated the potential function of membrane-tethered APP intracellular domain in neurons using a fusion protein mAICD in which the extracellular and transmembrane domains of APP were deleted and replaced by sequences encoding membrane-targeting MyrPalm motif. This strategy allowed us to recruit APP-CTF to lipid raft membrane microdomains, and potentially activate in a constitutive manner putative signaling associated with the APP intracellular domain. We found that accumulation of membrane-tethered APP-CTF, resulting from the inhibition of γ-secretase processing of APP-FL or by expression of mAICD, produced a marked increase of neurite extension in primary cortical neurons, H19-7 hippocampal cells and N2a neuroblastoma cells. Moreover, we report that membrane accumulation of APP intracellular domain is intimately coupled to adenylate cyclase signaling, which mediates enhanced neurite formation. Our results show that GαS coupling to adenylate cyclase is a necessary step in mAICD-induced neurite outgrowth in primary neurons as well as neuronal cell lines. Finally, we demonstrate that a BBXXB motif (672RHLSK676) within APP-CTF as the functional site of interaction with GαS protein, and that this interaction is critical to promote neurite extension.
Several reports have previously linked APP expression to neurite outgrowth (Allinquant et al., 1995; Perez et al., 1997; Ando et al., 1999; Small et al., 1999; Leyssen et al., 2005; Young-Pearse et al., 2008; Hoe et al., 2009; Hoe et al., 2010). Indeed, cell-surface APP, its soluble ectodomain cleavage product, and APP-CTF have been reported to regulate neurite outgrowth. Several mechanisms have been proposed including a role of Aβ/APP receptor-mediated signaling (Ando et al., 1999; Young-Pearse et al., 2008; Shaked et al., 2009; Sola Vigo et al., 2009). The theory that APP may function as a receptor has been proposed based mainly on structural similarity to type I membrane receptors, a concept that has been debated for more than a decade. Indeed, our findings clearly support the idea that APP is acting as a G-protein coupled receptor, a model in which membrane-bound APP-CTF facilitates interactions and recruitment of cytosolic adaptors and proteins that exacerbate intracellular signaling and influence a variety of cellular functions, including neurite outgrowth (Fig. 11). It was previously reported that APP could interact with Go complex (Nishimoto et al., 1993; Brouillet et al., 1999; Shaked et al., 2009). However, neither report established that the interaction of APP-CTF with Go complex is functional or has direct implications for the developing nervous system. In our study, we outline our characterization of APP with another G-protein coupled subunit, GαS. But, most importantly, we demonstrate that interaction of APP-CTF with GαS is necessary and sufficient to induce neurite formation in mammalian neurons, ruling out major contribution of other putative G-protein interactions in neurite outgrowth. Using modulators of cAMP/PKA activities, we establish that adenylate cyclase activation plays a pivotal role in APP-CTF-induced neurite outgrowth. Furthermore, our results demonstrate that direct coupling of membrane-bound APP intracellular domain to GαS is a previously unrecognized essential step that mediates this function.
Several years ago, it was reported that residues 657–676 of APP cytoplasmic domain (which includes 672RHLSK676 domain) possess affinity for GαO protein (also referred to as GO); this interaction was found to activate GαO protein reconstituted in phospholipid vesicles with synthetic APP-CTF peptides (Nishimoto et al., 1993). However, a subsequent study reexamined this issue using neuronal membranes prepared from E19 rat embryonic brain tissue and reported that physiological interaction with APP cytoplasmic tail inhibited GαO GTPase activity (Brouillet et al., 1999). While both studies identified residues 657HH658 as critical for GαO interaction (Nishimoto et al., 1993; Brouillet et al., 1999), neither study examined the functional outcome of APP interaction with GαO protein. Nevertheless, based on stimulation or inhibition of GαO GTPase activity following the application of mAb22C11 which binds to the extracellular domain of APP, these studies simply speculated that APP functions as a receptor and GαO protein binding likely is involved in signal transduction. Therefore, by analogy to other G-protein coupled receptors such as the adrenergic and serotoninergic receptors (Eason et al., 1992; Lefkowitz, 1998; Xiao, 2001; Pindon et al., 2002), it is possible that signaling through APP-CTF involves stochastic or simultaneous binding of GαO and GαS on the same APP molecule. Referred as G-protein coupling “promiscuity”, this attribute could indeed account for limiting the spatio-temporal signaling of G-protein coupled receptors (Woehler and Ponimaskin, 2009). Further studies will be needed to fully understand the complexity of this paradigm with specific emphasis on APP.
G-protein coupled receptors are known to activate G-proteins from more than one family. Following G-protein coupled receptor activation, GαS and Gαi/O class of G-proteins mediate stimulation or inhibition of adenylate cyclase activity; respectively. Because expression of mAICD correlated with the activation of adenylate cyclase, we focused our attention on GαS. Our functional study on neurite outgrowth defines BBXXB motif as the binding site for the previously unrecognized GαS interaction with APP C-terminal domain. The BBXXB motif and sequences with similar characteristics are known to be involved in the G-protein activity of several cell surface membrane receptors (Okamoto et al., 1990; Okamoto and Nishimoto, 1992; Wu et al., 1995; Pauwels et al., 1999; Zhou and Murthy, 2003; Ulloa-Aguirre et al., 2007; Peverelli et al., 2009). Our results also demonstrate that palmitoylation of GαS is a prerequisite for interaction with the BBXXB motif in APP.
Here, we show that membrane-bound APP intracellular domain regulates neurite formation by G-protein coupling, eliciting adenylate cyclase-dependent intracellular signaling, independent of the transcriptional role ascribed to AICD (Muller et al., 2008; Schettini et al., 2010). Although our studies focused on neurite formation, adenylate cyclase activation resulting from APP-CTF coupling to GαS could play a critical role in regulating short and long-term synaptic plasticity (Malenka and Nicoll, 1999; Malinow and Malenka, 2002), and memory formation (Arnsten et al., 2005; Abel and Nguyen, 2008). In support to this idea, we have previously reported increase of synaptic neurotransmission associated with cAMP/PKA pathway activation in presenilin 1-deficient neurons (Parent et al., 2005; Barnes et al., 2008), a condition where APP-CTF accumulates at the cell surface (Leem et al., 2002; Barnes et al., 2008) and in lipid raft (Vetrivel et al., 2005). More interestingly, our studies show that accumulation of APP-CTF at the membrane produces dual PKA activation and GSK3β inhibition in neurons and neuroblastoma cells. Clear evidence in the literature suggests that these dual signaling events would favor axodendritic formation and function (Lonze and Ginty, 2002; Jope and Johnson, 2004; Yoshimura et al., 2006; Hur and Zhou, 2010; Hutchins, 2010; Shelly et al., 2010). GSK3β activation is a known pathological hallmark of AD (Hooper et al., 2008; Hernandez et al., 2010). Therefore, a relationship between signaling initiated by APP-CTF and GSK3β inhibition highlights an aspect that is extremely important as a new perspective to AD etiology and therapeutics.
Further studies are needed to address the possibility that expression of the CTFs derived from APP/APLP homologues might be sufficient to rescue reduced neurite outgrowth and synaptic defects previously reported in APP-deficient neurons (Allinquant et al., 1995; Zheng and Koo, 2006; Hoe et al., 2009; Hoe et al., 2010; Lee et al., 2010; Aydin et al., 2011; Guo et al., 2011). Moreover, we need to consider the possibility that lack of APP expression alone might not produce a “loss-of-function” pertaining to neurite outgrowth downstream of G-protein coupled signaling mediated by APP/APLPs. To further characterize the physiological relevance of the signaling that we have discovered, study in APP/APLP1 or APP/APLP2 double knock-out neurons could be necessary. Still, much needs to be learned regarding the subcellular sites where APP-CTFs reside and function in GαS-signaling in neurons before the intracellular domain is released from the membrane by γ-secretase processing. For example, a recent study showed that synaptic activity only regulates processing of a subset of APP-CTFs in Arc-containing endosomes in dendritic structures (Wu et al., 2011). Therefore, although we used neurite outgrowth as our assay in the overexpression experiments, the physiological functions regulated by APP/APLP-CTF signaling through GαS coupling in neurons could be subtler to discern.
In recent years, G-protein coupled receptor signaling cascades have been proposed as a therapeutic target for AD by several investigators [reviewed in (Thathiah and De Strooper, 2011)]. Therefore, the novel functional coupling of APP-CTF with GαS-protein at the neuronal membrane identified in this study has important implications because accumulation of APP-CTF at the membrane is an invariable outcome of therapeutic inhibition of γ-secretase processing of APP aimed at reducing cerebral amyloid burden.
This work was supported by the National Institutes of Health grants NS055223 (ATP) and AG019070 (GT), and the IIRG-06-26148 grant from the Alzheimer’s Association (ATP). We thank Dr. Mitchell Villereal for providing H19-7 cells. We are grateful to Stacy Herrera, Breanne Kassarjian, Rafael Marquez, and Megan Rawson for technical assistance. We thank Drs. Eric Norstrom and Hyun-Ju Kim, and members of the Thinakaran lab for helpful discussions and advice.
Conflict of Interest: The authors declare no competing financial interests.