The Drosophila beta-amyloid precursor protein homolog promotes synapse differentiation at the neuromuscular junction

Abstract

Although abnormal processing of beta-amyloid precursor protein (APP) has been implicated in the pathogenic cascade leading to Alzheimer's disease, the normal function of this protein is poorly understood. To gain insight into APP function, we used a molecular-genetic approach to manipulate the structure and levels of the Drosophila APP homolog APPL. Wild-type and mutant forms of APPL were expressed in motoneurons to determine the effect of APPL at the neuromuscular junction (NMJ). We show that APPL was transported to motor axons and that its overexpression caused a dramatic increase in synaptic bouton number and changes in synapse structure. In an Appl null mutant, a decrease in the number of boutons was found. Examination of NMJs in larvae overexpressing APPL revealed that the extra boutons had normal synaptic components and thus were likely to form functional synaptic contacts. Deletion analysis demonstrated that APPL sequences responsible for synaptic alteration reside in the cytoplasmic domain, at the internalization sequence GYENPTY and a putative G(o)-protein binding site. To determine the likely mechanisms underlying APPL-dependent synapse formation, hyperexcitable mutants, which also alter synaptic growth at the NMJ, were examined. These mutants with elevated neuronal activity changed the distribution of APPL at synapses and partially suppressed APPL-dependent synapse formation. We propose a model by which APPL, in conjunction with activity-dependent mechanisms, regulates synaptic structure and number.

Fig. 1.

Transport of APPL to motor axons and nerve terminals, and synapse morphology at NMJs with altered APPL levels.A1–D1, Anti-HRP staining of larval NMJs in wild-type muscles 12 and 13 (A1),Appl^d mutant muscles 12 and 13 (B1), a larva overexpressing APPL⁺(muscles 6/7) (C1), and a larva overexpressing APPL^sd (muscles 6/7) (D1).A2–D2, Anti-APPL immunoreactivity in the same samples. Note the presence of bright immunoreactivity at the nerve of the wild type and the weak immunoreactivity at synaptic boutons. This anti-APPL signal is eliminated in the Appl^d mutant. NMJs from larvae overexpressing either APPL⁺or APPL^sd show strong anti-APPL immunoreactivity. In Appl^sd, immunoreactivity is more restricted to boutons, whereas in APPL⁺ immunoreactivity is not restricted to boutons but is also found well outside boutons (not colocalized with anti-HRP). Arrow points to a satellite budding off from an NMJ process. Scale bar: A,B, 25 μm; C, D, 12 μm.

Fig. 2.

Light microscopic analysis of satellite boutons.A, B, Anti-HRP immunoreactivity in a string of synaptic boutons at muscles 6/7 of wild type (A) and a larva overexpressing APPL⁺(B). Arrowheads point to a satellite bouton. C, D, Anti-synapsin immunoreactivity in wild type (C) (same preparation shown inA) and a larva overexpressing APPL⁺(D) (same preparation shown in B). Note that each satellite bouton contains synapsin immunoreactivity.E, F, Anti-HRP (red) and anti-DLG (green) double labeling of NMJs in wild type (E) and a larva overexpressing APPL⁺(F). Note that virtually all satellite boutons are surrounded by DLG immunoreactivity. Scale bar, 13 μm.

Fig. 3.

Quantitative analysis of NMJs at muscles 6 and 7 (abdominal segment 3) of wild type and APPL variants expressing normal and altered forms of APPL. A, Total number of boutons.B, Percentage of satellite boutons relative to total number of boutons (see Results for a definition of satellites).C, Number of parent (nonsatellite) boutons. The number of preparations quantified are the same as shown above thebars in B.

Fig. 4.

Ultrastructural organization of type I synaptic boutons in a larva overexpressing APPL⁺.A, Cross-section through a type I synaptic bouton in wild type. B, Cross-section through type I boutons and satellite boutons in a larva overexpressing APPL⁺. Note that, in wild type, a single bouton is completely surrounded by the elaborate SSR. In larvae overexpressing APPL⁺, parent boutons (b) and their satellite boutons (asterisks) appear surrounded by a common SSR. Note a satellite bouton budding off from a parent bouton (arrow). Inset shows a view of a satellite bouton containing a presynaptic density.Arrowheads indicate presynaptic densities. Scale bar, 1.2 μm.

Fig. 5.

Mutant APPL proteins. A, Schematic representation of the mutant APPL proteins used in this study. E1 and E2 are the two most conserved regions in the extracellular domain. C is the cleavage site that is deleted in all APPL^sd forms.TM represents the transmembrane domain. In the cytoplasmic domain, g refers to the putative G_o-binding domain, and I is the internalization signal (see Results). B, Expression of mutant APPL proteins detected in Western blots of adult head total protein extracts. Blots were probed with a polyclonal antibody against the extracellular region of APPL that detects both transmembrane and secreted forms (lanes 1–9), or with a polyclonal antibody against the cytoplasmic internalization signal that detects only the transmembrane form (lanes 10–16). In either case, no protein is observed inAppl^d heads (lanes 1,10), and the expected bands are present in wild-type Canton special heads (lanes 2, 11). Inlane 2, the intense top band contains transmembrane APPL forms with different post-translational modifications. Directly below, a broad weak bandincludes soluble APPL proteins with different post-translational modifications. As expected, this soluble form is not recognized by the antibody against the cytoplasmic domain (lane 11).Lanes 3–9 and 12–16 are extracts from heads of the genotype Appl^dAppl-Gal4; UAS-Appl*/+ (asterisk indicates any of the deletion constructs) raised at 30°C (lanes 3, 12) or at 20°C (lanes 4–9, 13–16) in which APPL protein is derived exclusively from the transgene encoding wild-type (APPL⁺) or mutant (APPL^s and APPL^sdΔ*) APPL forms. Appl⁺ generates both transmembrane and soluble forms (lanes 3, 4, 12,13), whereas Appl^sproduces only the soluble forms detected as a weak broad band (lane 5). All Appl^sd-derived transgenes generate only transmembrane forms that contain the cytoplasmic domain (compare lane 6with 14); the bottom band at ∼100 kDa most likely corresponds to unglycosylated forms. Molecular weights are expressed in kilodaltons.

Fig. 6.

APPL domains required for satellite bouton formation and effect of elevated neuronal activity in APPL expression and satellite bouton formation. A–D, Anti-HRP immunoreactivity in wild type (A), an APPL^sd larva (B), an APPL^sdΔC larva (C), and an APPL^sdΔCI larva (D). Note that normal boutons are present in larvae overexpressing the APPL variant lacking the cytoplasmic domain or the conserved internalization sequence. E, Expression of APPL in eag Shmutants. Note that, unlike wild type (Fig. 1A2), APPL immunoreactivity at the NMJ of eag Sh mutants is enhanced, and it is associated with the borders of the most distal boutons. F1, NMJs in eag Sh mutants expressing APPL^sd, showing a decrease in satellite boutons (but see histogram in Fig. 2B).F2, Anti-APPL immunoreactivity in the same preparation asF1. Scale bar: A–D, F, 12 μm; E, 18 μm.

Fig. 7.

Role of APPL on NMJ expansion.A, Stages of NMJ expansion during larval development. According to our model, new boutons at expanding NMJs are formed by sprouting, followed by the consolidation of some of the sprouts into undifferentiated boutons, which subsequently differentiate to form a new mature bouton (shaded boutons). B, Model of APPL function as a G_o-coupled receptor. According to this model, the binding of a ligand to APPL activates a transduction cascade that involves G_o-protein. Ligand-bound or unbound APPL is internalized and subsequently recycled to the cell membrane. We propose that surface APPL is involved in sprouting, whereas activation of the transduction cascade is involved in the regulation of bouton differentiation. Deletion of the G_o-protein binding site (ΔCg) or the conserved extracellular domains (ΔE1, ΔE2) results in oversprouting with diminished bouton differentiation, leading to satellite boutons. Deletion of the internalization signal (ΔCI) or elevated levels of electrical activity (eag Sh mutants) lead to an increase in surface APPL, resulting in the differentiation of a larger than normal number of sprouts and therefore an increase in the number of parent boutons.