|Home | About | Journals | Submit | Contact Us | Français|
The amyloid precursor protein (APP) is the source of Aβ fragments implicated in the formation of senile plaques in Alzheimer’s Disease (AD). APP-related proteins are also expressed at high levels in the embryonic nervous system and may serve a variety of developmental functions, including the regulation of neuronal migration. To investigate this issue, we have cloned an orthologue of APP (msAPPL) from the moth, Manduca sexta, a preparation that permits in vivo manipulations of an identified set of migratory neurons (EP cells) within the developing enteric nervous system. Previously, we found that EP cell migration is regulated by the heterotrimeric G protein Goα: when activated by unknown receptors, Goα induces the onset of Ca2+ spiking in these neurons, which in turn down-regulates neuronal motility. We have now shown that msAPPL is first expressed by the EP cells shortly before the onset of migration, and that this protein undergoes a sequence of trafficking, processing, and glycosylation events that correspond to discrete phases of neuronal migration and differentiation. We also show that msAPPL interacts with Goα in the EP cells, suggesting that msAPPL may serve as a novel G protein-coupled receptor capable of modulating specific aspects of migration via Goα-dependent signal transduction.
A hallmark feature of Alzheimer’s Disease (AD) is the accumulation of amyloid plaques in the brain (Glenner and Wong, 1984; Masters et al., 1985), which are formed predominantly by short β-amyloid fragments (Aβ) that are derived by proteolytic cleavage of the amyloid precursor protein (APP; Kang et al., 1987; Suh and Checler, 2002). APP is a member of a family of transmembrane glycoproteins with structural similarity to cell surface receptors: they possess large extracellular domains with several potential protein interaction motifs, plus short cytoplasmic domains containing motifs that can interact with a variety of adapter and intracellular signaling molecules (Annaert and De Strooper, 2002; Turner et al., 2003). At least 10 distinct isoforms of APP may be generated via alternative mRNA splicing from a single gene (Sandbrink et al., 1994a; Sandbrink et al., 1994b), but the predominant isoform expressed in the nervous system is APP695 (Kitaguchi et al., 1988). Proteolytic cleavage of APP by a combination of juxtamembrane and intramembrane secretases results in the release of the Aβ fragment, which can then aggregate into amyloid plaques (Selkoe, 1998). Both the soluble Aβ monomer (Tong et al., 2004; Yankner, 1996) and its aggregates (Klein, 2002; Selkoe, 2001; Stern et al., 2004) are capable of inducing a variety of neurotoxic effects, providing support for the “amyloid hypothesis” that this cleavage product of APP mediates the neurodegenerative aspects of Alzheimer’s disease (Hardy and Allsop, 1991; Hardy and Selkoe, 2002).
However, more recent studies on the biological actions of APP have impelled a modified view of this hypothesis. APP is expressed at high levels in the developing nervous system and is localized to regions of neuronal motility and synapse formation (Akaaboune et al., 2000; Loffler and Huber, 1992; Masliah et al., 1992). APP has been detected within growth cones and developing synapses in the mammalian brain (Akaaboune et al., 2000; Clarris et al., 1995; Kirazov et al., 2001; Lahiri et al., 2002) and is upregulated in regenerating olfactory neurons (Struble et al., 1998). In cell culture, APP can also promote neurite outgrowth (Beher et al., 1996; Qiu et al., 1995) and cell-substratum adhesion (Chen and Yankner, 1991; Kibbey et al., 1993). These observations have led to the proposal that APP is part of a large protein complex normally involved in the regulation of such processes as neuronal migration, axonal outgrowth, and synaptic differentiation (Sabo et al., 2001; Turner et al., 2003), albeit via poorly defined mechanisms.
More provocative is the co-localization of APP with a number of molecules in focal complexes at the leading edges of migratory cells. These include β-integrins (Yamazaki et al., 1997), which may provide a functional link between APP and the actin cytoskeleton, and the adapter proteins Fe65, mDAB1, and members of the X11/Mint family, each of which can bind to the cytoplasmic domain of APP (Sabo et al., 2001). APP also directly interacts with the heterotrimeric G protein Goα (Nishimoto et al., 1993; Giambarella et al., 1997; Okamoto et al., 1996), a signal transduction molecule that comprises up to 1% of all membrane-associated protein in the developing nervous system (Edmond et al., 1990) but whose developmental functions remain poorly understood. Indeed, APP has been postulated to act as a novel class of G protein-coupled receptor (Brouillet et al., 1999; Giambarella et al., 1997), although the importance of this interaction with respect to the control of neuronal growth and survival are still controversial (Turner et al., 2003). Both APP and its proteolytic products are capable of exerting a variety of actions on cell growth, signal transduction, and neural plasticity (Kamenetz et al., 2003; Li et al., 1997; Mucke et al., 1994; Walsh et al., 2002), leading to the proposal that APP may serve as a “molecular hub” at the center of many converging regulatory pathways (Turner et al., 2003). Thus, besides the generation of excessive Aβ, aberrant processing of APP might disrupt one or more of its normal functions in the brain, which might in turn contribute to the abnormalities detected in AD. Yet despite a plethora of studies on APP and its fragments in a variety of assays, the developmental and physiological roles of APP have remained unclear.
Analysis of the intrinsic functions of APP in vertebrates has been complicated by the identification of two closely related proteins (APLP1 and APLP2) which share partially overlapping functions with APP (Heber et al., 2000; von Koch et al., 1997). In contrast, only one APP-related gene is expressed in Drosophila (appl; Rosen et al., 1989) and Caenorhabditis (apl-1; Daigle and Li, 1993), which has facilitated investigations into their normal functions in vivo. The proteins encoded by these genes share a number of structural similarities with the vertebrate APP family (Coulson et al., 2000), including the conserved intracellular domain required for Goα interactions (Nishimoto et al., 1993; Torroja et al., 1999b). Moreover, transgenic studies in Drosophila have shown that human APP695 can rescue at least some of the deficits caused by the deletion of APPL (Luo et al., 1992), demonstrating that these proteins are functionally as well as structurally conserved. In Drosophila, APPL has been implicated in variety of developmental functions, including the acquisition of phototactic behavior (Luo et al., 1992), axonal transport (Gunawardena and Goldstein, 2001; Torroja et al., 1999a), and postembryonic differentiation of neuromuscular synapses (Ashley et al., 2005; Torroja et al., 1999b). In particular, Torroja et al. found that selective mutations of the putative Goα-binding domain of APPL caused specific alterations in the growth of presynaptic boutons at the neuromuscular junctions (Torroja et al., 1999b), lending support to the model that APP-related proteins may functionally interact with Goα in developing neurons.
Surprisingly, genetic deletions of APPL in Drosophila produced only subtle effects on the developing nervous system (Li et al., 2004; Luo et al., 1992; Zambrano et al., 2002), possibly due to compensatory interactions by other functionally related proteins that may have masked the normal contribution of APPL to neural development. As an alternative means of addressing this issue, we have examined the developmental role of APPL during the formation of the enteric nervous system (ENS) in the moth Manduca sexta. The ENS in this species contains a distributed population of neurons (the EP cells) that occupy a branching nerve plexus (the enteric plexus) spanning the foregut-midgut boundary (Copenhaver and Taghert, 1988; Copenhaver and Taghert, 1989b; see Fig. 1). In previous work, we showed that the EP cells achieve their mature distributions in the enteric plexus via directed migration along identified sets of visceral muscles that form on the surface of the gut. After delaminating from the foregut epithelium, small groups of EP cells align with one of eight longitudinal muscle bands on the midgut or with specific radial muscles on the foregut (Fig. 1, 55% of development). Subsequently, the neurons migrate rapidly along these pathways (between 55–65% of development) before completing their differentiation (Copenhaver et al., 1996; Copenhaver and Taghert, 1989b). Of note is that throughout their development, the EP cells grow on the most superficial layer of the developing gut musculature; thus, both the neurons and their pathways remain accessible to direct observations and acute manipulations in embryonic culture (Horgan et al., 1995; Wright and Copenhaver, 2000).
The guidance of EP cell motility is also precisely regulated: although they continually extend short filopodial processes onto the adjacent interband musculature (Fig. 1, 55–60% of development), both the migrating neurons and their growing axons remain closely apposed to their muscle band pathways, only branching laterally to innervate the interband regions towards the end of their differentiation (Fig. 1, 100% of development). In part, the precise guidance of the EP cells is mediated by fasciclin II (mFas II), a homophilic adhesion receptor that is expressed by both the neurons and transiently the muscle bands during the migratory period (Wright and Copenhaver, 2000; Wright and Copenhaver, 2001). However, strong repulsive interactions have also been found to prevent EP cell migration onto the interband regions (Copenhaver et al., 1996), a process that may be mediated by the G protein Goα. All of the neurons commence the expression of Goα just prior to the onset of migration, and selective activation of Goα within individual EP cells inhibits their motility in a Ca2+-dependent manner (Horgan et al., 1994; Horgan et al., 1995). However, the identity of candidate Goα-coupled receptors expressed capable of inducing this response in the EP cells has remained unknown.
In the present study, we have isolated and sequenced the gene encoding APPL in Manduca (designated msAPPL), and we have analyzed its developmental expression and processing in the EP cells with respect to their migration and differentiation. We have also investigated whether msAPPL and Goα interact within the EP cells during their migration. The results of these studies suggest a novel role for APP-related proteins in the control of neuronal motility within the developing nervous system.
Developmentally synchronized embryos were collected from a colony of Manduca sexta and maintained at 25°C; at this temperature, embryogenesis is completed in 100 hr (1% development = 1 hr). Developmental staging, immunoblots, and whole-mount immunohistochemistry were performed as previously described (Horgan and Copenhaver, 1998; Wright et al., 1998). Antibodies were used at the following concentrations: anti- mFas II: 1: 20,000 (Wright et al., 1999); anti-ELAV 1:500 (Robinow and White, 1991; gift from Dr. Kalpana White); anti-Drosophila APPL: 1:1000 (dAPPL; gift from Drs. Kalpana White and Vivian Budnik); and affinity-purified anti-Goα: 1:100 (Horgan et al., 1995). We also generate two antibodies against msAPPL: anti-msAPPL-cyt against the C-terminal (CZ)YENPTYKYFEVKE sequence (used at 1:5000); and anti- msAPPL-ect against the extracellular domain (CZ)EDDDYTDADDSAWPRPES sequence (used at 1:1000). For whole-mount immunostaining, primary antibodies were detected using the HRP-based ABC protocol of Vector laboratories (Burlingame, CA) or using fluorochrome-conjugated secondary antibodies (Cy3-conjugates from Jackson Labs, West Grove, PA; and Alexa-488 and −568 conjugates from Molecular Probes, Eugene, OR). Whole-mount immunostained preparations were then photographed by conventional microscopy or imaged by confocal microscopy, either with an Olympus Fluoview 300 laser scanning confocal head mounted on an Olympus BX51 microscope, or with a BioRad 1024 ES laser scanning confocal microscope.
Primary neuronal cultures were prepared after the methods of Hayashi et al. (Hayashi and Hildebrand, 1990; Hayashi and Levine, 1992). Briefly, desheathed ganglia were harvested from P0–P1 stage Manduca pupae, enzymatically dissociated with trituration, and plated onto glass coverslips coated with Concanavalin A plus Laminin (4–5 ganglia per dish). After 4 days in L-15-based culture medium (Gibco; Carlsbad, CA) plus 10% fetal bovine serum (FBS), the neurons were rinsed and grown for 2 additional days in reduced-protein L-15 medium (L-15 plus 1 μg/ml 20-hydroxyecdysone, 100 U/ml penicillin, 100 U/ml streptomycyin; from Sigma; St. Louis, MO). The medium from the cultures was then harvested and concentrated to a final volume of 30 μl (using a 30 kDa cut-off Centricon spin concentrator; Amicon; Beverly, CA), while the neurons were harvested in RIPA buffer (150 mM NaCl, 1% IGEPAL, 0.5% deoxychoic acid, 1% SDS, 50 mM Tris pH 8.0) for immunoblot analysis.
For immunoblots, soluble protein was extracted from pooled tissue samples collected from synchronous groups of embryos in lysis buffer, or from primary neuronal cultures as described above. The protein samples were then separated by polyacrylamide gel electrophoresis (after our published methods; Wright and Copenhaver, 2000) and transferred to nitrocellulose. Replicate blots were incubated either with the primary antibodies listed above, with pre-immune sera, or with aliquots of the anti-msAPPL antibodies that had been pre-adsorbed against the synthetic peptides used as their antigens (in a 10:1 molar ratio). Aliquots of each anti-msAPPL antibody were also cross-pre-adsorbed with the peptide epitope used to generate the other antibody, as an additional control for the specificity of our adsorption protocol. Bound antibodies were then detected either with alkaline phosphatase-conjugated secondary antibodies (1:2500; KPL; Gaithersburg, MD) and detected by a colorometric reaction, or with HRP-conjugated secondary antibodies (anti-chicken at 1:10K from Promega, Madison, WI; other HRP-conjugated secondary antibodies at 1:10K from Jackson Immunoresearch, West Grove, PA), and detected by chemiluminescence using the Supersignal WestPico kit from Pierce (Rockford, IL) or the enhanced Visualizer kit from Upstate (Lake Placid, NY).
A partial clone of Manduca APPL (msAPPL) was obtained from embryonic cDNA, using degenerate PCR primers against evolutionarily conserved domains of APP-related proteins from other species (Daigle and Li, 1993; Rosen et al., 1989; Wasco et al., 1992; Wasco et al., 1993). 32P-labeled probes derived from this clone were then used to isolate clones from a cDNA library (gift of Dr. James Nardi) that contained the full-length coding sequence for msAPPL. Nucleotide sequencing was performed on an ABI 377 automated fluorescence sequencer and analyzed using the DNASTAR suite of programs. For Northern blots, 32P-labeled probes were prepared from a cDNA clone containing the 1.6 kb open reading frame of msAPPL and used to label blots of poly A(+) mRNA that had been isolated from embryos with the MicroPoly(A)Purist kit from Ambion. The mRNA was separated on a 1% agarose gel, transferred to a nylon membrane, hybridized with probe overnight at 80°C, and exposed to film after extensive washing. Whole-mount in situ hybridization histochemical staining of staged embryos was performed following published methods (Wright et al., 1999; Patel and Goodman, 1992), using digoxigenin-labeled probes generated against the 1.6 kb cDNA clone encoding msAPPL. Bound probe was detected in filleted embryos using an alkaline phosphatase-conjugated anti-digoxigenin antibody (1:2000; Roche).
For co-immunoprecipitation of APPL and Goα, tissues were collected from developmentally synchronous groups of embryos or from desheathed pupal ganglia and frozen on dry ice. The samples were then homogenized in RIPA buffer for 20 min at room temperature, spun briefly, and the supernatant pre-cleared with Gammabind Plus Sepharose beads (25 μl in 250 μl ; from Pharmacia). Replicate samples were then incubated either with 1–10 μl of each anti-msAPPL antibody, pre-immune sera, or with aliquots of the anti-msAPPL antibodies that had been pre-adsorbed against their synthetic peptides, as described above. Antibodies were added to the tissue homogenates for 3–4 hr at room temperature, followed by the addition of 25 μl washed beads (pre-adsorbed with 1% IgG-free bovine serum albumin) for 1 hr at room temperature. Beads with bound antibody were then pelleted by centrifugation, washed three times in chilled RIPA buffer, and then resuspended in SDS sample buffer (Wright et al., 1999). After boiling for 5 min, to total volume of containing the resuspended pellets or 30 μl aliquots of the residual supernatants from each immunoprecipitation reaction were separated on 12% SDS polyacrylamide gels, transferred to nitrocellulose, and immunoblotted with anti-Goα antibodies. Labeled bands were visualized using HRP-conjugated secondary antibodies detected with the Visualizer kit (Upstate).
For removal of N-linked glycoproteins from msAPPL, staged groups of embryos (at 65% of development) were homogenized in RIPA buffer plus protease inhibitors at room temperature for 15 min. Alternatively, the tissue samples were homogenized in hot extraction buffer (10 % glycerol; 1% mM SDS, 50 mM Tris pH 6.7) plus inhibitors for 5 min at 100°C. The samples were cleared by brief (5 min) microcentrifugation, and the soluble fractions incubated with the deglycosylating enzyme PNGase F (Peptide: N-Glycosidase F; 500 U/μl; New England Biolabs; Beverly, MA) following the manufacturer’s instructions for 16–24 hr at 37°C. Control samples were treated in an identical manner but without the addition of PNGase. Replicate samples were also reacted with α1–2 Fucosidase (10 U/μl; New England Biolabs) or O-glycanase (1.25 mU/μl; Prozyme, San Leandro, CA) as per the manufacturers’ suggested protocols. The resultant protein samples were then separated on polyacrylamide gels, transferred onto nitrocellulose, and immunoblotted with our antibodies against either the cytoplasmic domain (anti-msAPPL-cyt) or extracellular domain (anti-msAPPL-ect) of msAPPL, or with antibodies that had been pre-adsorbed against their peptide epitopes. Immunoblots that had been reacted with one antibody were routinely stripped and re-probed with the other antibody for an exact comparison of the protein bands recognized by antibodies specific for the different domains of msAPPL.
As an initial means of examining whether APPL-related proteins are expressed by the EP cells during their migration, we used an antiserum against Drosophila APPL (dAPPL) to immunostain embryos at 60% of development, a stage when the EP cells are still migrating but have also begun to extend axonal processes (see Fig. 1). Strong immunostaining was detected in all of the EP cells as they migrated onto the midgut muscle bands (Fig. 2A, arrows), including robust staining within their leading processes (Fig. 2A, arrowheads). To verify that these neurons express authentic APPL, we isolated a cDNA clone encoding the Manduca homologue of APPL (designated msAPPL; nucleotide and deduced amino acid sequences are shown in Suppl. Fig. 1). This cDNA clone contained a single open reading frame encoding a 732 AA protein that shares 51% overall sequence identity with dAPPL and 33% overall identity with human APP695. The predicted structure of msAPPL is closely similar to other members of the APP family: it is a type-one transmembrane protein with several potential sites for N-linked glycosylation (see annotations in Suppl. Fig. 1). It also has the characteristic extracellular domains (E1 and E2) and cytoplasmic domain (C) found in all APP-related proteins (Bush et al., 1994; Coulson et al., 2000), although like dAPPL in Drosophila and APL-1 in C. elegans, msAPPL does not share sequence conservation with the Aβ region of human APP (Suppl. Fig. 2). MsAPPL also lacks the large inserted sequences found between the E1 and E2 domains of Drosophila (Suppl. Fig. 2), so that the predicted molecular weight of its core protein (81 kDa) is closer to that of APP695.
Sequence conservation between msAPPL and other APP-related proteins was found to be especially strong within its extracellular (E1 & E2) and cytoplasmic (C) domains (Fig. 2B). The E1 domain of msAPPL is 66% identical to that of dAPPL and 43% identical with human APP695, while the E2 domain is 60% identical with that of dAPPL and 36% identical with APP695. Of particular note is the high degree of sequence similarity between the cytoplasmic (C) domains of msAPPL and the equivalent domains in both fly (88%) and human APP695 (77%), including a 100% conserved TPEERH sequence within the Goα binding domain of APP695 (Nishimoto et al., 1993) and a 100% conserved NGYENPTYK sequence within the internalization motif required for interactions between APP695 and the adapter proteins FE65, X11/Mint, and DAB1 (Turner et al., 2003).
To examine the expression pattern of msAPPL-specific mRNA, 32P-labeled riboprobes were generated from our msAPPL cDNA clone and used in Northern blots of poly A+ mRNA isolated from developing embryos. As shown in figure 3A, these probes recognized a single mRNA species at ~6.0 kb, similar to the 6.5 kb transcript reported for dAPPL mRNA (predicted to encode the slightly larger dAPPL protein; Rosen et al., 1989). Digoxigenin-labeled riboprobes were then used for in situ hybridization histochemical staining of staged embryos throughout the periods of EP cell neurogenesis, migration and differentiation (Fig. 4). We previously showed that the EP cells first emerge from a neurogenic placode in the posterior dorsal lip of the foregut epithelium, a process that occurs between 30–40% of development (Copenhaver and Taghert, 1990). As shown in figure 4 (at 35% of development), no detectable msAPPL-specific mRNA was apparent in the EP cells during their initial delamination (arrows). In contrast, strong expression levels were apparent at this stage in neurons within the frontal ganglion on the anterior foregut (FrGl; arrowhead), which are born at an earlier period of development (Copenhaver and Taghert, 1991). By 40% of development, a stage when the EP cells are post-mitotic but have not yet begun to migrate (Copenhaver and Taghert, 1989b), faint levels of msAPPL mRNA could first be detected within the clustered neurons (Fig. 4, 40%). Staining intensities in the EP cells then steadily increased over the next 15% of development (40–55%), when the neurons spread circumferentially around the foregut-midgut boundary (black hatch marks in Fig. 4; see also Fig. 1 for orientation). All of the neurons continued to express robust levels of msAPPL-specific mRNA during their subsequent phase of fast migration onto the midgut muscle bands (Fig. 4, 55–65% of development; arrows indicate the progression of EP cell migration onto the mid-dorsal muscle bands). This pattern persisted throughout their subsequent periods of axon outgrowth (65–85%) and target innervation (75–100%; Copenhaver and Taghert, 1989a; Copenhaver and Taghert, 1989b).
Elsewhere in the enteric nervous system, the expression levels of msAPPL mRNA remained high in all post-mitotic neurons that had migrated anteriorly along the course of the esophageal nerve (EN) to form the frontal ganglion (FrGl) and hypocerebral ganglion (HcGl) of the foregut (Fig. 5A). Throughout embryogenesis, strong msAPPL-specific staining was also detectable in neurons within the ganglia of the central nervous system (Fig. 5B) and in clusters of peripheral neurons on the body wall (Fig. 5B, arrowheads). In contrast, no detectable staining was seen in either central or peripheral glial cells, nor in any other non-neural cell type within the embryo. These results indicate that msAPPL is a neuron-specific protein whose expression commences only after cells have completed their terminal mitosis. In the ENS, msAPPL expression in the EP cells coincides with the onset of their migratory behavior and persists throughout their subsequent phases of outgrowth and synapse formation.
In Drosophila, dAPPL has been shown to be expressed predominantly as a single glycosylated protein species of ~145 kDa, from which a ~130 kDa secreted fragment is rapidly processed by one or more secretase-like proteases (Luo et al., 1990). Immunoblots of protein extracted from the post-embryonic CNS (Fig. 3B, lane 1) revealed a larger immunoreactive species at ~135 kDa and a smaller species at ~120 kDa. A similar pattern was seen in embryonic extracts (Fig. 3B, lane 2), although the 120 kDa band was substantially more abundant than 135 kDa band. As described below, this smaller, 120 kDa band most likely represents the soluble ectodomain of msAPPL that is released after cleavage of the holoprotein by secretase activity, similar to the pattern seen in Drosophila. The apparent sizes of these proteins in Manduca are slightly smaller than those seen in Drosophila extracts (Fig. 3B, lane 3), as expected from the predicted amino acid sequences of the moth versus fly proteins (80 AA versus 98 AA; Luo et al., 1990) (see also Suppl. Fig. 2).
To examine whether the smaller, 120 kDa band does represent the secreted ectodomain of msAPPL, we generated anti-peptide antibodies against specific sequences within the N- and C-terminal domains of the predicted protein (designated anti-msAPPL-ect and anti-msAPPL-cyt, respectively). The anti-msAPPL-ect antibody would be predicted to recognize both the holoprotein and any ectodomain fragments released by juxtamembrane secretase activity, while the anti-msAPPL-cyt antibody should recognize the holoprotein but not secreted ectodomains (see Suppl. Fig. 3). We then used primary cultures of Manduca CNS neurons to test whether the 120 kDa band was selectively released from the neurons into the surrounding medium. As shown in figure 3C, the anti-msAPPL-ect antibody recognized both the 135 kDa and 120 kDa bands in extracts prepared from intact CNS (lane 1) and in extracts of the primary neurons (lane 2; harvested 6 days after plating), but only the smaller, 120 kDa band was detected in the surrounding medium (Fig. 3C, lane 3). All immunostaining was eliminated when we pre-adsorbed the antibody with the N-terminal domain peptide against which it was generated (Fig. 3D), whereas pre-adsorption of this antibody with the peptide used to generate the msAPPL-cyt antibody had no such effect (not shown).
In contrast, the anti-msAPPL-cyt antibody recognized only the larger 135 kDa band but not the 120 kDa band in extracts of both CNS (Fig. 3E, lane 1) and primary neurons (lane 2). This antibody also labeled several larger isoforms (discussed below) and a band at ~117 kDa (Fig. 3E, arrowhead), which may be an immature form of msAPPL or a degradation product. However, none of these bands were detected in the surrounding medium (Fig. 3E, lane 3). As with the N-terminal-specific antibody, pre-absorption of the anti-msAPPL-cyt antibody with its peptide epitope eliminated all detectable staining (Fig. 3F), whereas pre-adsorption with the peptide used to generate the anti-msAPPL-ect antibody had no effect (not shown). These results support the conclusion that the 120 kDa band detected with either the anti-dAPPL and anti-msAPPL-ect antibodies (both of which recognize epitopes in the extracellular regions of msAPPL) represents the secreted ectodomain of msAPPL, consistent with the release of the extracellular portions of other APP family proteins by juxtamembrane secretases (De Strooper and Annaert, 2000; Luo et al., 1990).
To examine the developmental expression of msAPPL over the course of embryonic development, we isolated body wall tissues (containing the CNS and PNS) and midgut tissue (containing the ENS) from embryos starting at 40% of embryogenesis, at which stage the EP cells have just begun to express the msAPPL gene. Equal amounts of total protein were then examined in immunoblots stained with antisera recognizing the N-terminal domain of msAPPL. In extracts of the body wall (Fig. 6A; containing protein from the CNS and PNS), msAPPL-related protein could be detected at 40% of development, corresponding to a period of active axonal and dendritic outgrowth by both central and peripheral neurons. Initially, both the 135 kDa and 120 kDa bands were of equal intensity, but as development progressed, the 120 kDa band became markedly more intense (Fig. 6A, 65–85%). Given our data suggesting that the 120 kDa band represents the secreted ectodomain of msAPPL (Fig. 3), these results indicate that most of the protein expressed in the CNS was rapidly processed by one or more secretases during this period. However, by the completion of embryogenesis (100%), the relative abundance of the two isoforms had returned to approximately equal levels (as also shown in Fig. 3B, lane 1).
A similar trend was also detected in extracts containing proteins from the ENS (Fig. 6B). At 40% of development, soon after the onset of msAPPL expression by the EP cells (see Fig. 4), both the 135 kDa and 120 kDa isoforms could be detected in extracts containing these neurons. In addition, however, we also detected three larger bands with apparent molecular weights of 155–180 kDa (Fig. 6B, arrowheads) that were not seen in post-embryonic tissue extracts. These larger protein species remained abundant throughout subsequent phases of EP cell migration (55–65%) and outgrowth (70–85%), while the 135 kDa band was reduced in relative intensity throughout much of this period. Only towards the end of embryogenesis did the 135 kDa band become more apparent. By 100% of development (hatching), the higher molecular weight bands were no longer detectable, leaving only the 135 kDa and 120 kDa bands (as seen in the CNS).
Previous studies have shown that both human APP695 and fly dAPPL isoforms are often glycosylated (Weidemann et al., 1989; Luo et al., 1990), and as noted above, the predicted sequence for msAPPL includes at least three potential sites for N-linked glycosylation, as well as potential sites for O-glycosylation and fucosylation (Suppl. Fig. 1). To test whether the larger protein species seen in figure 6B (arrowheads) might represent hyperglycosylated isoforms of msAPPL, we reacted embryonic lysates with PNGase F to remove N-linked glycoproteins (Plummer and Tarentino, 1991), and then examined them in immunoblots labeled with a combination of our anti-msAPPL antibodies to reveal all potential isoforms. When we used embryo extracts prepared in RIPA buffer, we found that both the 135 kDa and 120 kDa bands (putatively the full-length and secreted ectodomains of msAPPL) were shifted to lower apparent molecular sizes by 3–4 kDa (compare lanes 2 & 3 in Fig. 6C), consistent with the effects of perturbing endoglycosylation in other APP family proteins (Luo et al., 1990; Oltersdorf et al., 1990). A larger shift was seen in the apparent sizes of the higher molecular weight bands, each being reduced by ~10–12 kDa. When we prepared embryonic lysates with hot extraction buffer (as per the manufacturer’s suggestion), we also observed the 155–180 kDa trio of bands plus the 135 and 120 kDa isoforms (Fig. 6C, lane 4). However, treatment of this preparation with PNGase completely eliminated the higher molecular weight bands, while the intensity of the 135 kDa band was enhanced (Fig. 6C, lane 5), suggesting that the higher molecular weight isoforms might represent hyperglycosylated forms of full-length msAPPL. In contrast, only minimal reductions were seen in the size of the 135 and 120 kDa bands with this treatment. Pre-incubation of the antisera with their peptide epitopes eliminated labeling of the trio of higher molecular weight proteins as well as the 135 and 120 kDa isoforms, (Fig. C, lane 6; compare with lane 1).
Because the residual size of the deglycosylated bands in either of these reaction conditions was still substantially larger than the predicted mass of the core protein of msAPPL (~81 kDa), we also reacted embryo extracts with O-glycanase (to remove O-linked glycoproteins) or α1–2 Fucosidase (to remove fucosyl side groups); however, neither of these treatments further reduced the apparent size of the msAPPL isoforms (not shown). Additional modifications that may contribute to this discrepancy in size are discussed below. Nevertheless, these results indicate that msAPPL undergoes a dynamic sequence of post-translational modification and processing in the EP cells over the course of their migration and subsequent differentiation.
To gain a better understanding of how msAPPL is distributed in the EP cells during their migration, we immunostained identically staged embryos either with antibodies against the cell adhesion receptor fasciclin II (mFas II; to show the full extent of EP cell outgrowth; Wright et al., 1999) or with our anti-APPL antibodies. A third set of matched preparations was stained with an antibody against ELAV (Robinow and White, 1991), an unambiguous neuronal marker that we have shown labels the EP cells but not enteric glial cells (Copenhaver, 1993). As previously described (Copenhaver and Taghert, 1990), when the EP cells first delaminate onto the foregut, they form a packet of post-mitotic but pre-migratory neurons (Fig. 7, arrows) positioned adjacent to a rostral cluster of glial precursor cells (arrowheads) that are associated with the esophageal nerve (EN). By 40% of development, all of the EP cells had begun to express ELAV; in contrast, msAPPL expression was not detectable in these neurons until several hours later, first appearing in the EP cells as they commenced their slow, circumferential spreading phase of migration around the foregut (Fig. 7, 40–55% of development). Like ELAV, msAPPL expression was restricted to the neurons of the ENS, being absent from the enteric glia or underlying visceral musculature. Throughout this period of development, msAPPL immunoreactivity was distributed in a punctate manner throughout the cytoplasm of the EP cells, as well as within their short filopodial processes extending on the adjacent foregut and midgut surfaces (described below). Subsequently, during their fast phase of migration onto the midgut muscle bands (55–65% of development; Fig. 8), all of the EP cells continued to express robust levels of APPL-related protein. As with mFas II, msAPPL immunoreactivity could be clearly seen in their leading processes and growing axons (Fig. 8, arrows), in contrast to ELAV, which was largely confined to the neuronal cell bodies.
The co-localization of msAPPL with mFas II was better demonstrated in confocal images of preparations that had been double-labeled with antibodies against these two proteins (Fig. 9). In actively migrating neurons (Fig. 9A; 58% of development), mFas II immunostaining (red) defined the outline of the EP cells (arrows) and their leading processes (arrowheads), and more faintly labeled their underlying muscle band pathways (asterisks; as described in Wright and Copenhaver, 2000). During this phase of migration, robust APPL immunoreactivity (green) was associated with the leading processes of the EP cells (Fig. 9A, arrowheads), as well as a more diffuse pattern of punctate staining throughout their somata (arrows). At higher magnification (Fig. 9B, different preparation than in 9A), co-localization of msAPPL and mFas II could clearly be seen within filopodia of leading processes of the EP cells (arrowheads); intriguingly, msAPPL was distributed all the way to the tips of some filopodia but not others, suggesting that the trafficking of this protein is dynamically regulated in these highly motile structures.
By 95% of development (Fig. 9C), a stage when the EP cells have completed their differentiation (see Fig. 1), the distributions of both proteins had changed substantially. mFas II immunoreactivity was primarily localized to the axons of the EP cells that had grown around the cell bodies of neighboring neurons, although some of this staining was associated with the enteric glial cells that ensheath the EP cell bodies (Wright and Copenhaver, 2000). APPL immunoreactivity was also still expressed in the axons of the EP cells (Fig. 9C) and could be faintly detected in their diffuse terminal branches on the interband musculature (not shown). However, the cytoplasmic component of msAPPL was now confined to a ring of large, perinuclear vesicles that did not contain mFas II (Fig. 9C, arrows). These results indicate that in addition to the developmentally regulated processing of msAPPL shown in figure 6, msAPPL (or its proteolytic fragments) undergoes differential trafficking in the EP cells, coincident with the different stages of neuronal migration and differentiation.
Previously, we showed that the EP cells first express the gene encoding the heterotrimeric G protein Goα just prior to the onset of migration (Horgan et al., 1995), and that Goα is the only G protein detectable in the leading processes of these neurons during their subsequent migration and outgrowth (Horgan et al., 1994; Horgan et al., 1995). Since human APP695 has been shown to interact directly with Goα in vertebrate cells (Brouillet et al., 1999; Nishimoto et al., 1993), we used confocal imaging of embryos that had been double- labeled with antibodies against msAPPL and Goα to examine the relative distributions of these two proteins in vivo. As shown in figure 10A, the two proteins were spatially co-localized within both the somata (arrows) and growing axons (arrowheads) of the EP cells. Goα is also transiently expressed by the midline cells (ml) lying between the muscle band pathways that are followed by the migratory neurons (Horgan et al., 1994), but these midline cells were devoid of msAPPL expression.
Because of relatively high background levels associated with our anti-Goα antibody in whole-mount preparations, we were unable to resolve the fine filopodial processes of the EP cells in these preparations. As an alternative strategy, we immunostained neurons grown in primary culture (3 days after plating) for msAPPL and Goα. As shown in figure 10B–C, the distribution of the two proteins largely overlapped within the major neurites and finer filopodial extensions of cultured neurons undergoing active outgrowth. These results indicate that msAPPL and Goα are co-localized in the motile regions of both central and peripheral neurons in Manduca.
To test whether msAPPL and Goα interact within the EP cells, we immunoprecipitated msAPPL from extracts of embryonic midgut and probed the immunoprecipitate with the anti-Goα antibody. As shown in figure 11A, the initial midgut extracts (“input”) contained a substantial amount of Goα. When we immunoprecipitated this extract with pre-immune sera, no detectable Goα was seen in precipitated protein pellet (“control IP”). In contrast, when we immunoprecipitated midgut extract with either of our anti-msAPPL antibodies, approximately 10% of the total amount of Goα that was detectable in the original lysate co-immunoprecipitated with msAPPL (“anti-APPL IP”). Pre-incubation of the anti-msAPPL antibodies with their peptide epitopes completely abolished their ability to co-immunoprecipitate Goα (“pre-adsorbed anti-APPL IP”). In a parallel study, we also co-immunoprecipitated msAPPL and Goα from lysates of desheathed CNS ganglia (isolated by the same protocol used for preparing our primary neuronal cultures; see Fig. 10). As with the embryonic midgut extracts, extracts of CNS ganglia contained robust levels of Goα (Fig. 11B, lane 1). Pre-immune serum failed to co-immunoprecipitate any detectable Goα (Fig. 11B, lane 2), whereas our anti-msAPPL antisera immunoprecipitated approximately 10% of the total Goα (Fig. 11B, lane 4) detectable in the initial ganglia lysate. Neither of our anti-msAPPL antibodies co-precipitated the related G protein Gsα (not shown), providing further evidence that the interaction between msAPPL and Goα is specific.
These results show that msAPPL and Goα are closely associated in Manduca neurons. Moreover, since the EP cells are the only cells on the midgut that express detectable levels of both msAPPL (Figs. 7–8) and Goα (Horgan et al., 1995), these experiments indicate that msAPPL interacts with Goα specifically in these migratory neurons. In combination with our previous demonstration that Goα-mediated signaling inhibits EP cell migration in vivo (Horgan and Copenhaver, 1998), our data support a model whereby interactions between msAPPL and Goα may regulate specific aspects of neuronal migration, as discussed below.
In previous work, we presented evidence that activation of Goα in the EP cells negatively regulates their motile behavior: Goα is expressed in the EP cells coincident with the onset of their migration (Horgan et al., 1994), and activation of Goα within individual EP cells leads to an inhibition of their migration and outgrowth (Horgan and Copenhaver, 1998; Horgan et al., 1994). This inhibitory effect of Goα was shown to mediated by Ca2+ influx via voltage-independent channels; moreover, G protein activation induces the onset of intrinsic Ca2+ spiking activity in the EP cells and results in a down-regulation of their motility (Horgan and Copenhaver, 1998). Spontaneous Ca2+ spikes of similar amplitude can be detected in the EP cells at the end of their migration, supporting the conclusion that Goα-dependent activation of Ca2+ influx normally regulates the guidance or extent of EP cell migration and outgrowth along the midgut muscle bands (Horgan and Copenhaver, 1998). Based on these results, we proposed a model whereby Goα-coupled receptors expressed by the EP cells would be activated by ligands encountered within the local cellular environment, resulting in the stimulation of Goα-mediated Ca2+ spiking activity in the EP cells and a concomitant regulation of their migratory behavior (Horgan and Copenhaver, 1998). However, the identity of candidate receptors capable of activating Goα in these neurons remained unknown.
The results described in current study suggest that msAPPL, a member of the APP family of transmembrane proteins, may serve this function. Our data show that msAPPL and Goα are co-localized within the EP cells during their migration and within the growing processes of neurons in primary culture (Fig. 10). We have also shown that Goα can be co-immunoprecipitated with anti-msAPPL antibodies from extracts of both the ENS and CNS (Fig. 11). Given the high degree of sequence similarity between the Goα binding domain of human APP695 (Nishimoto et al., 1993) and the equivalent domain in msAPPL (Fig. 2), these results are consistent the proposal that msAPPL and Goα directly interact in the EP cells.
A hypothetical model supported by our data is that msAPPL acts as an atypical Goα-coupled receptor in the EP cells that regulates their response to inhibitory guidance cues in a Goα-dependent manner (Fig. 12). In previous work, we showed that the EP cells extend filopodial processes onto both the muscle band pathways (B) and the adjacent interband regions (IB), areas that are inhibitory to migration (Copenhaver et al., 1996). Filopodia that extend onto the interband regions are significantly shorter and have briefer lifespans than those that grow along the muscle bands (unpublished observations), supporting the notion that ligands (Fig. 12, red) encountered within the interband regions may induce a local retraction response. By the proposed model, binding of these ligands by msAPPL would induce the local activation of Goα and Ca2+ influx in filopodia growing onto the interbands, leading to their retraction; msAPPL-dependent activation of Goα would therefore serve to restrict the migration and outgrowth of the EP cells to their permissive muscle band pathways. Alternatively, more global activation of this signaling cascade within an EP cell would inhibit its overall migration, a response that might help regulate the normal distribution of these neurons to the anterior segment of the midgut (Copenhaver and Taghert, 1989a; Copenhaver and Taghert, 1989b). As with APP695, authentic ligands for msAPPL remain unknown, but they might include either membrane-associated molecules or secreted factors derived from interband cells, both of which would be readily accessible to msAPPL in the leading processes of the EP cells.
Support for this model may be inferred from a variety of studies in other systems. APP-related proteins are expressed in high levels within in the leading processes and growth cones of motile neurons in all species examined (Moya et al., 1994; Lahiri et al., 2002; Clarris et al., 1995; Ohta et al., 1993), and experiments in cell culture have shown that APP (or its proteolytic derivatives) can modulate neuronal motility and outgrowth in a variety of contexts (Kibbey et al., 1993; Beher et al., 1996; Perez et al., 1997; Qiu et al., 1995). Although an authentic role for human APP695 in the developing nervous system remains controversial, a recent study on triple-knockout mice lacking APP and the related genes APLP1 and APLP2 revealed a dramatic defect in neuronal migratory behavior: subsets of cortical neurons in these mice failed to recognize their normal cues to terminate migration, resulting in aberrant migration through the overlying pial layer (Herms et al., 2004). Similarly, a recent examination of developing neuromuscular junctions in mice lacking both APP and APLP2 revealed that these animals exhibit defective synaptic structures associated with excessive nerve terminal sprouting (Wang et al., 2005). In preliminary studies, we recently found that inhibition of msAPPL expression (using msAPPL-specific morpholino probes) resulted in excessive, inappropriate migration by the EP cells onto the normally inhibitory interband regions of the midgut, consistent with the model shown in figure 12 (Copenhaver et al., in preparation). Although in both mice and Manduca, the cues that are recognized by APP-related molecules remain to be determined, these results support the hypothesis that members of the APP family may act as transmembrane receptors involved in the regulation of neuronal migratory behavior and outgrowth.
Precedent for our model that msAPPL may act as a novel Goα-coupled receptor can also be found in studies on human APP695. Nishimoto et al. showed that Goα (but not other G proteins) binds directly to a cytoplasmic domain of APP695 containing a TPEERH motif (Nishimoto et al., 1993), an amino acid sequence that is 100% conserved in Manduca msAPPL (Fig. 2). Moreover, activation of APP695 with an antibody against its ectodomain induced a significant increase in GTPγS turnover by Goα, analogous to G protein activation by conventional G protein-coupled receptors (Nishimoto et al., 1993; Okamoto et al., 1995). This same group showed that one class of APP695 mutations linked to Familial AD caused these proteins to act as constitutively active Goα-linked receptors when incorporated into vesicles (Okamoto et al., 1996) or in cultured cells (Ikezu et al., 1996). The direct interaction between APP695 and Goα was verified in other experiments using neuronal membranes (Brouillet et al., 1999), although in this study, activation of APP695 caused an inhibition of Goα GTPase activity, underscoring the challenges of working with isolated membrane preparations. Nevertheless, these studies advanced the provocative model that APP may act as a novel type of G protein-coupled receptor that regulates Goα activity, although the downstream signaling events that may be modulated by APP proteins remain undefined (Turner et al., 2003). Our demonstration that msAPPL interacts with Goα in the EP cells provides a model system that can now be used to test the role of this APP family protein in regulating neuronal motility, specifically via the modulation of Ca2+ spiking activity in vivo; these experiments are currently underway.
As noted above, msAPPL shares most of the major structural features found in other APP family proteins, including conserved motifs in both the extracellular and intracellular domains that have been implicated in a variety of potential interactions with other proteins (Fig. 2; Suppl. Fig. 2; De Strooper and Annaert, 2000; Turner et al., 2003). Like “neuronal” APP695 and fly dAPPL (Rosen et al., 1989), msAPPL lacks the KPI domain found in the larger APP isoforms that are expressed in non-neuronal cells (Turner et al., 2003). Both Manduca and Drosophila APPLs are expressed exclusively in neurons (Figs. 4–6; Martin-Morris and White, 1990), although they lack obvious homology with the Aβ region of human APP695. In this regard, the insect versions of this protein more closely resemble the vertebrate APLPs, which have been postulated to have arisen along with APP via gene duplication from a single ancestral gene (Coulson et al., 2000).
Our data also indicate that like human APP695 (Walter et al., 2001) and fly dAPPL (Luo et al., 1990), msAPPL is expressed as a glycosylated transmembrane protein that is rapidly processed to produce a smaller, secreted ectodomain fragment. In the CNS, the holoprotein appears primarily as a 135 kDa protein that can be processed by secretase-like activity to produce a soluble 120 kDa fragment. As shown in figure 3, we were able to detect this 120 kDa fragment but not the 135 kDa protein in the medium of cultured primary neurons, using our antibody against the N-terminal domain of msAPPL but not with our C-terminal-specific antibody. These data support our conclusion that the 120 kDa protein represents the cleaved ectodomain fragment of the holoprotein. Notably, we only detected the 120 kDa protein in the medium surrounding the neurons after several days in culture, suggesting that continued expression and cleavage of msAPPL leads to the gradual accumulation of this fragment as the neurons grow. Other bands recognized by our C-terminal-specific antibody (including the 135 kDa isoform) were not detected in the medium, nor was the 120 kDa band seen in control samples of the culture medium (not shown). We also obtained analogous results when we expressed msAPPL in Drosophila S2 cells, similar to previous published studies by Luo et al. on dAPPL (Luo et al., 1990), albeit with somewhat altered apparent molecular weights (unpublished data). These results support the conclusion that msAPPL undergoes secretase-dependent processing in a manner similar to that of other APP family members. Although the specific identity of the secretases responsible for msAPPL processing in Manduca have yet to be verified, we have now isolated cDNAs encoding evolutionarily conserved members of all of the α-, β-, and γ-secretase families implicated in APP695 processing (De Strooper and Annaert, 2000). In future work, we intend to investigate how secretase-dependent cleavage of msAPPL affects different aspects of EP cell motility in vivo.
Several aspects of msAPPL expression in the EP cells may pertain to its potential role as a guidance receptor involved in modulating neuronal migration. First, expression of msAPPL-specific mRNA (Fig. 4) and protein (Fig. 7) was initially detected in the EP cells just prior to the onset of their slow spreading phase of development, several hours after the completion of their terminal mitoses (Copenhaver and Taghert, 1990) and after the onset of ELAV expression (Fig. 7). These results are consistent with the delayed expression of dAPPL in the Drosophila CNS, which is detectable only after the expression of other neuronal-specific markers (Martin-Morris and White, 1990). Because the EP cells are the only neurons on the midgut during embryogenesis, immunoblots of this tissue provided an accurate view of msAPPL processing in the EP cells at different stages of their development (Fig. 6B). Notably, during periods of active migration and outgrowth, much of the msAPPL associated with the EP cells appeared to undergo extensive post-translational modification, resulting in the three prominent bands at 155–180 kDa. By the completion of embryogenesis, these larger isoforms were no longer apparent, so that virtually all of the full-length protein migrated at 135 kDa (as seen in extracts of the CNS). These results suggest that stage-specific regulation of post-translational processing events may contribute to the developmental role of msAPPL in actively motile neurons.
The nature of the modifications that contribute to the apparent size of these larger isoforms remains incompletely understood. As shown in figure 6C, PNGase F treatment of lysates prepared with hot extraction buffer suggested that the trio of bands at 155–180 kDa may be hyperglycosylated forms of full-length APPL; however, this protocol failed to shift the apparent size of the 135 and 120 kDa proteins, which is inconsistent with published reports that both the holoprotein and secreted ectodomains of APP proteins are glycosylated (Luo et al., 1990; Oltersdorf et al., 1990). In contrast, PNGase treatment of lysates prepared in RIPA buffer did shift these bands by 3–4 kDa but only reduced the larger trio of bands by 10–15 kDa. Our hot extraction protocol might possibly sensitize carbohydrate side chains on msAPPL that are normally resistant to PNGase treatment, although treatments with O-glycanase and Fucosidase had no effect on any of the msAPPL-associated bands. APP family proteins are also subject to phosphorylation (Russo et al., 2005) and sumoylation (Li et al., 2003), but such modifications should not be affected by the protocols used in this study. Alternatively, the higher molecular weight bands might be due in part to homodimerization of msAPPL isoforms, as has been suggested for human APP695 (Scheuermann et al., 2001). In such a case, the combination of hot extraction buffer and PNGase treatment might have serendipitously disrupted potential homophilic interactions, resulting in the shift in apparent molecular weights that we observed.
Although the significance of these different aspects of msAPPL processing remain to be explored in terms of the control of neuronal migration, these results suggest that the post-translational modification and rapid turnover of transmembrane msAPPL may play important roles in regulating the distribution and bioavailability of the holomolecule or its proteolytic fragments. For example, studies performed with rat neuronal membranes showed that only a small fraction of intact APP was accessible to the extracellular space at any given time, but that this same fraction co-localized with Goα and was responsible for modulating GTPγS turnover when activated by antibody cross-linking (Brouillet et al., 1999). Consistent with this view, immunostaining unpermeabilized preparations of the ENS with our anti-msAPPL antibodies resulted in a greatly diminished signal compared to that seen in fixed, permeabilized embryos (unpublished observations), indicating that the holoprotein of msAPPL may reside on the surface of growing neurons only briefly before being proteolytically cleaved. This pattern of rapid expression and processing of a potential guidance receptor might provide a means of continually refreshing a neuron’s response to dynamic changes in the extracellular cues that it encounters during its migration into the ENS. In this regard, the distribution of msAPPL on the leading processes of the migrating EP cells is intriguing, in that we observed that msAPPL extended into the distal tips of some filopodia but not others (Fig. 9). In the context of our proposed model, msAPPL may preferentially traffic to filopodia that are undergoing active elongation and are therefore most likely to encounter potential ligands in the local environment; alternatively, the protein might accumulate in filopodia that have become stabilized and are forming focal contacts, as has been suggested from studies of mouse APP in transfected cells and primary neuronal cultures (Sabo et al., 2001; Sabo et al., 2003).
Lastly, while msAPPL expression remained strong in the EP cells throughout their differentiation, we noted a marked redistribution of the protein as the neurons matured. During their active phase of migration, msAPPL was distributed in a punctate manner throughout the EP cells, consistent with the vesicular trafficking of APP by fast anterograde transport (Lazarov et al., 2005; Stokin et al., 2005). In addition, a fraction of msAPPL also co-localized with mFas II (Fig. 9) and Goα (Fig. 10) in the leading processes and axons of the EP cells in vivo and CNS neurons grown in primary culture, lending credence to our model that msAPPL- dependent activation of Goα plays a role in migratory guidance. By the end of embryogenesis, however, most msAPPL within the somata of the EP cells was confined to a ring of large, perinuclear vesicles (Fig. 9C), although some of the protein remained associated with the axonal fascicles and terminal branches of these neurons. While the identity of these perinuclear vesicles is still unknown, they resemble the intracellular compartments in which vertebrate APPs (or their derivatives) have been shown to accumulate (Lee et al., 2003; Nixon et al., 2000; Takahashi et al., 2004), which may be endosomal or lysosomal in nature.
These results suggest that msAPPL may be differentially trafficked during EP cell development, so that membrane-associated msAPPL is concentrated in regions of active motility and target innervation, while much of the remaining protein or its processed fragments becomes localized within this perinuclear compartment. Although cytoplasmic fragments of human APP695 have been shown to be capable of trafficking to the nucleus and regulating gene transcription (Kinoshita et al., 2002; but see Cao and Sudhof, 2004), the pattern of sequestration that we observed at the completion of migration and outgrowth would also provide an effective means of limiting the capacity of msAPPL to modulate Goα activity. In summary, our data indicate that in actively migrating EP cells, msAPPL undergoes a dynamic sequence of post-translational modification, trafficking, and processing, consistent with the view that this protein may play an important role in regulating one or more aspects of directed neuronal migration in this system.
Nucleotide and deduced amino acid sequences of the cDNA clone encoding msAPPL. The predicted signal sequence and transmembrane domains are indicated by dotted overlines. The conserved extracellular (E1 and E2) and cytoplasmic (CYTO) regions that share sequence similarity with homologous domains other APP family proteins are labeled and indicated by solid overlines (see also text Fig. 2). Predicted N-linked glycosylation sites are indicated by “*”; potential O-linked glycosylation sites by “#”; and a potential fucosylation site by “@”. The C-terminal consensus site for ER retention/internalization (FEVK) is indicated by “x x x x “.
Schematic diagram of structural similarities between dAPPL (from Drosophila); msAPPL (from Manduca); human APP695, human APLP1, and human APLP2. The conserved extracellular domains are indicated as E1 and E2. Within the cytoplasmic domain, the conserved domain associated with Goα binding is designated as “Go”, and the putative internalization motif (which also contains the binding sites for X11/Mint, Fe65, and DAB1) is designated as “Int”. The Aβ motif of APP695 is not obviously conserved at the amino acid level in other members of this protein family.
Schematic diagram of msAPPL indicating the location of the epitopes recognized by our anti-msAPPL-ect antibody (a peptide within the E2 extracellular domain) and our anti-msAPPL-cyt antibody (a peptide within the C-terminal cytoplasmic domain; see methods for peptide sequences). Like other APP family proteins, msAPPL appears to be cleaved by one or more juxtamembrane secretases, whereby the soluble ectodomain fragment is released from the neuronal membrane. The anti-msAPPL-ect antibodies should detect both the holoprotein and secreted ectodomains, while the anti-msAPPL-cyt antibody should only detect full-length isoforms of the protein.
We wish to thank Drs. Vivian Budnik, Mary Packard, and Kalpana White for generously sharing their anti-dAPPL antibodies, and Dr. Budnik for many fruitful discussions on the biology of APPL. We also wish to thank Drs. Michael Forte and Doris Kretzschmar for critical evaluations of this work. Dr. Stefani Kaech-Petrie (of the Live Cell Imaging Facility in the Center for Research in Occupational and Environmental Toxicology) and Aurelie Snyder (of the OHSU-MMI Research Core Facility; (http://www.ohsu.edu/core) provided invaluable assistance in the generation of several of our confocal images. Dr. Tiffani Howard (moc.knidrawoh@inaffit) produced the excellent graphics shown in Figure 12 and supplemental figures 2 and 3. This work was supported in part by NIH # NS34439 to PFC, by NIH P30 AG08017, and a grant from the Oregon Partnership for Alzheimer’s Research. The GenBank accession number for msAPPL is DQ008058.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.