BIOLOGICAL OVERVIEW

The effects of amyloid precursor protein (APP) on neuronal development and function have been extensively studied because of its implication in Alzheimer's disease (for review, see Mattson, 1997; Small, 1998; Sinha, 1999). The Drosophila ß-amyloid precursor protein homolog ß amyloid protein precursor-like (Appl) is a pan-neural protein belonging to the conserved APP family. This family includes APP and amyloid precursor-like protein (APLP)1/2 in mammals and apl-1 in C. elegans. APP is synthesized as a transmembrane protein composed of extracellular, transmembrane, and cytoplasmic domains. In the normal process of APP function, APP undergoes a regulated proteolytic cleavage, releasing the long ectodomain (Torroja, 1999 and references therein).

A principal component of amyloid plaques in Alzheimer's disease (AD) is a 38-43-amino acid ß-amyloid peptide (A ß), which is derived from the C-terminal portion of the extracellular domain of APP and from the adjacent transmembrane domain. Production and accumulation of A ß is involved in the etiology of AD. APP is an integral membrane protein with a short intracellular C-terminus. Normal proteolytic processing of APP generates two fragments, a large soluble secreted amino-terminal domain (sAPPalpha) and a truncated carboxyl-terminal fragment (CTFalpha), both of which are products of cleavage within the A ß domain at approximately amino acid number 17: this site is termed the alpha cleavage site. Another form of proteolytic processing occurs at the ß-site, located at amino acid number 1 of the A ß peptide. This gives rise to production of carboxyl-terminal fragments (CTFß) and A ß in unidentified intracellular compartments of the protein secretory pathway. The protein Presenilin (see Drosophila Presenilin) directs the cleavage of APP at even a third site, gamma, located between amino acids 40 and 42, the C-terminus of the A ß peptide. The A ß peptide is generated by cleavages at both the beta site and the gamma site; cleavage at these two sites generate the 40 to 42 amino acid A ß peptide (Sinah, 1999 and references therein).

In vivo evidence supports a role for APP in synapse formation, maintenance, and plasticity. APP infusion into rat brains increases synaptic density and memory retention (Roch, 1994), and exposure of hippocampal slices to secreted APP enhances long-term potentiation and modifies the induction of long-term depression (Ishida, 1997). APP knock-out mice show impaired learning and memory (Müller; 1994; Zheng, 1995). Alternative mRNA splicing can generate isoforms of APP that contain a Kunitz protease inhibitor (KPI) domain. The KPI forms of APP are implicated in the regulation of extracellular cleavage of secreted APP by acting to inhibit the activity of a secreted APP-degrading protease (Caswell, 1999). A second extracellular target of the KPI domain is the proneuropeptide processing enzyme prohormone thiol protease (PTP), involved in the processing of active neurotransmitters and peptide hormones; for example the prohormone proenkephalin (Hook, 1999).

To investigate the role of APP at single, identified synapses, the Drosophila neuromuscular junction (NMJ) has been used as a model system. Appl overexpression in motoneurons results in a dramatic increase in the number of synaptic boutons. Conversely, NMJs of Appl null larvae exhibit a significant decrease in synaptic bouton number. The synapse-promoting function requires a conserved internalization sequence and a putative G_o binding site at the cytoplasmic domain. This function is partially suppressed by mutations with increased excitability, which by themselves regulate synapse growth (Budnik, 1990) and Appl expression at the NMJ (Torroja, 1999).

To determine whether Appl is transported to motoneuron synapses, antibodies directed against the Appl ectodomain were used to label larval body wall muscle preparations. Appl is found to be consistently distributed in a punctate pattern within the abdominal nerves, down to the point of contact with body wall muscles, consistent with axonal transport of Appl. NMJs show very weak Appl signal. No signal is detected in nerves or NMJs of the Appl null mutant (Appl^d) larvae, demonstrating that Appl immunoreactivity at nerves and NMJs is specific (Torroja, 1999).

Appl signal at synaptic terminals is enhanced when Appl is overexpressed in the nervous system. Neural Appl overexpression was achieved using the UAS/Gal4 system for targeted gene expression. The UAS-Appl⁺ transgene was driven by pan-neural Gal4 drivers c155 or Appl-Gal4, or the motoneuron-specific Gal4 driver P[Gal4w⁺]C164. Expression of Appl and its influence on synapse structure was examined at muscles 6 and 7. These muscles are innervated by two glutamatergic motoneurons that give rise to morphologically and physiologically distinct synaptic boutons [type Ib (big) and type Is (small)]. At the NMJs of larvae overexpressing APPL⁺, Appl immunoreactivity is readily evident. Appl signal appears in a nonhomogenous punctate pattern, primarily inside synaptic boutons (colocalized with anti-HRP as a presynaptic marker) but also outside in the immediate vicinity of synaptic boutons. The Appl immunoreactivity outside the boutons may represent secreted Appl, because overexpressing a secretion-deficient Appl form (APPL^sd) results in Appl immunoreactivity that is more restricted to synaptic boutons (Torroja, 1999).

The presence of Appl in motor axons and their synaptic terminals suggested that it might be involved in synapse development and/or function. NMJs from Appl^d larvae were examined to ascertain the effects of the lack of Appl on synapse structure. NMJs from mutant larvae, examined in anti-HRP stained preparations, have a generally normal appearance but contain fewer boutons. Appl^d shows a significant decrease in the number of synaptic boutons (34%), as compared with wild type. When Appl is overexpressed in motoneurons, the increase in Appl signal is accompanied by a change in overall morphology of the NMJ. These profound structural changes include a dramatic increase in the number of synaptic boutons that result largely from numerous small synaptic boutons of abnormal appearance. In wild-type NMJs, type I synaptic boutons within a branch resemble a string of beads, with boutons connected to one another by a short neuritic process. In contrast, in larvae overexpressing Appl, many abnormal small 'satellite' boutons appear to bud off from a central bouton of normal appearance. This phenotype is associated with both type Ib and type Is endings. A fraction of satellite boutons (~24%) are also observed budding off from neuronal processes connecting two boutons (Torroja, 1999).

To determine whether the satellite boutons have normal features of synaptic boutons, presynaptic and postsynaptic markers were used. Analysis of the presynaptic markers synapsin, synaptotagmin, and CSP reveals that each satellite bouton is immunopositive for these markers. Furthermore, analysis of Discs large immunoreactivity, a marker for postsynaptic subsynaptic reticulum (SSR), clearly demonstrates that virtually all satellite boutons are surrounded by SSR as is the case with normal boutons. Thus, at least some elements of the presynaptic apparatus and postsynaptic SSR are likely to be normal at the satellite synapses. The fine structure of the satellite boutons was analyzed at the EM level. As expected from the light microscopic appearance of these NMJs, clusters of boutons are common in APPL⁺ overexpressors, and larger boutons are often surrounded by many small satellite boutons. These satellite boutons appear to share a common SSR with the central boutons and are frequently observed budding off from a parent bouton. At the presynaptic site, the satellite boutons contain 40 nm clear synaptic vesicles and mitochondria and display one or more T-shaped presynaptic densities (active zones) of normal appearance. Quantification of the number of active zones per bouton midline reveals that boutons from Appl overexpressors and from wild type have a similar number of active zones. These observations suggest that, although small, satellite boutons have at least several normal type I bouton presynaptic and postsynaptic elements and therefore may form functional synaptic connections (Torroja, 1999).

To quantify the increase of synaptic boutons observed in APPL⁺ overexpressing NMJs, the total number of synaptic boutons, the percentage of satellite boutons, and the number of normal boutons (parent boutons) at muscles 6 and 7 of abdominal segment 3 were counted at the light microscopic level. For this quantification, boutons were considered to be satellites if they were substantially smaller than normal type I boutons and were connected to a central (parent) bouton. Boutons that have a size and morphology similar to typical satellites but emerge directly from an NMJ process were also considered to be satellites. In larvae overexpressing APPL⁺, there was an ~2.5-fold increase in the total number of boutons, and an approximate threefold increase in the percentage of satellite boutons compared with wild type. No significant change in the number of synaptic boutons compared with wild type was observed in the control progeny. Thus, whereas the lack of Appl in Appl^d mutants leads to a relatively small decrease in the number of synaptic boutons, increasing Appl levels results in a dramatic rise in bouton number. This increase in bouton number resulting from both an increase in satellite boutons and an elevated number of nonsatellite boutons of normal appearance (parent boutons). Interestingly, expression of a human APP isoform, APP₆₉₅, also results in a significant but comparatively smaller increase in the total number of synaptic boutons and, particularly, an increase in the percentage of satellite boutons (approximately twofold) (Torroja, 1999).

All APP family members are proteolytically processed and can be found as transmembrane and soluble protein forms. Studies demonstrating extensive regulation of the cleavage and release of soluble APP suggest that the two protein forms may play specific roles. To determine whether the transmembrane or soluble Appl isoform can promote satellite formation, transgenes were used that express mutant Appl proteins upon Gal4 activation. UAS-Appl^sd transgene encodes a secretion-defective transmembrane form in which Appl lacks the proteolytic cleavage site and consequently is expressed only as a membrane-bound protein. UAS-Appl^s transgene encodes a constitutively secreted form because it lacks the transmembrane and cytoplasmic domains. Analysis of NMJs in larvae expressing APPL^sd or APPL^s demonstrates that APPL^sd expression results in a phenotype similar to the phenotype observed in larvae overexpressing APPL⁺, increasing the percentage of satellite boutons (approximately fourfold) and, to a lesser extent, the number of parent boutons (~1.7-fold). In contrast, APPL^s does not significantly differ from wild type with regard to the total number of synaptic boutons or satellites (Torroja, 1999).

The lack of effect seen in APPL^s may be because of differences in the amount of protein expressed. Therefore, the levels of expression of the different transgenic Appl forms were assessed in immunoblots probed with anti-APPL antibodies. Immunoblots of adult head extracts of wild type, Appl^d, and transgenic flies expressing different Appl constructs in an Appl^d background reveal that Gal4-driven expression of both APPL⁺ and APPL^sd results in high levels of protein relative to the endogenous level of Appl. In contrast, the amount of APPL^s generated from the transgene is only comparable with the amount of endogenous soluble Appl. It is suspected that this is attributable to the lability of the expressed soluble protein, because other Appl mutant constructs expressed with this system all result in robust expression. Attempts were made to increase the amount of APPL^s by using two doses of the transgene. This manipulation does not alter the total number of boutons but significantly reduces the number of satellites compared with wild type. In conclusion, both overexpression of APPL⁺ holoprotein or transmembrane APPL^sd can strongly promote satellite bouton formation, but APPL^s does not affect the number of total boutons, even after doubling the number of copies of the transgene. However, two doses of APPL^s decreases the number of satellite boutons (Torroja, 1999).

These results suggest that the synaptic bouton-promoting activity associated with Appl overexpression is sensitive to Appl concentration. To test this possibility, the phenotypes of NMJs were examined in Appl^d larvae that overexpress APPL^sd to eliminate wild-type contribution. The increase in the number of synaptic boutons, percentage of satellites, and number of parent boutons is intermediate but significantly different from wild-type and APPL^sd overexpressing larvae, supporting a concentration effect of APPL on this activity (Torroja, 1999).

Which domains of Appl are important for satellite and parent bouton-forming activity? APP family proteins have several conserved domains that have been initially recognized from APP and Appl comparison (Rosen, 1989). Among these are a perfectly conserved internalization sequence in the cytoplasmic domain (GYENPTY) and extracellular regions of high homology E1 and E2. Additionally, in the cytoplasmic domain of APP, a G_o-protein binding site, which is conserved in Appl, has been demonstrated (Nishimoto, 1993; Okamoto, 1995). These defined domains have been used as a guide to construct a series of deletion mutants: DeltaE1, lacking the distal half of E1 and those amino acids (aa) c-terminal to it (85-321 aa); DeltaE2, lacking 75% of the distal E2 and amino acids c-terminal to it (449-740 aa); C100, lacking the entire ectodomain (21-787 aa); DeltaC, lacking the entire cytoplasmic domain (836-886 aa); DeltaCI, lacking the internalization sequence (872-883 aa); and DeltaCg, lacking the putative G_o-protein binding site (845-855 aa). Appl^sd was chosen rather than Appl⁺ as a parent construct to avoid complications in interpretation that could arise if generated APPL^s had an independent activity. Because several transformants were obtained with each construct, those lines were selected that showed strong and similar levels of protein expression in immunoblot analysis (Torroja, 1999).

NMJs from larvae expressing the different constructs were double stained with anti-HRP and anti-APPL, and the total number of boutons, the percentage of satellite boutons, and the number of parent boutons were determined. Labeling with anti-APPL antibodies determined that, with the exception of C100, all Appl mutant proteins are transported at similar levels to the NMJs in which they are present. The salient observations from quantification of synaptic bouton number in larvae expressing APPL^sd -deletion proteins, DeltaE1, DeltaE2, and DeltaC, are as follows. (1) APPL^sd, further deleted for either of the two extracellular domains (DeltaE1 or DeltaE2), prevents an increase in the number of parent boutons, yet it still retains satellite bouton-forming activity. (2) In contrast, deletion of the entire cytoplasmic domain of APPL^sd (DeltaC) completely abolishes both the satellite bouton-promoting activity and the increase in the number of parent boutons. Thus, the cytoplasmic domain is essential for both the satellite bouton-promoting activity and the formation of extra parent boutons (Torroja, 1999).

To further dissect the likely cytoplasmic domain regions responsible for promoting synapse formation, smaller deletions in the cytoplasmic domain were overexpressed. Deletion of the putative G_o-protein binding site (DeltaCg) retains the satellite bouton-promoting capacity, but the formation of extra parent boutons is suppressed. Conversely, larvae expressing APPL variants that lack the conserved internalization signal (DeltaCI) have NMJs with a wild-type appearance with regard to the number of satellite boutons. Interestingly, however, these larvae still show a dramatic increase in the number of parent boutons, which translates into an increase in the total number of boutons. Thus, the satellite bouton-promoting activity depends on the presence of the internalization signal, whereas the formation of extra parent boutons depends on the presence of the G_o-protein binding site and the presence of an intact extracellular domain (Torroja, 1999).

A role for neuronal activity in the development and maintenance of synapses has been documented in many systems. Neuronal activity has also been shown to regulate APP metabolism. It was therefore of interest to enquire whether Appl-mediated regulation of the NMJ growth is induced through neuronal activity. To explore this question, the effects of Appl overexpression in the hyperexcitable double mutant eag Sh was examined. Intriguingly, Appl immunoreactivity in eag Sh differs from wild type. In wild type, Appl signal at synaptic boutons is very low but is observed throughout the NMJ. In contrast, in eag Sh mutants, Appl signal is consistently brighter and more concentrated at the most distal bouton(s) of a given NMJ branch. In addition, Appl signal appears more concentrated at the bouton border, suggesting that there might be an increase in plasma membrane-associated Appl. Western analysis of adult heads has revealed no significant difference in the amount of total Appl between wild type and eag Sh (Torroja, 1999).

In eag Sh mutants, both the number of boutons and NMJ branches are increased. However, the increase in synaptic boutons in eag Sh double mutants differs from that observed in NMJs with Appl overexpression. In NMJs with enhanced Appl levels, the increase in the number of synaptic boutons derives primarily from satellite boutons, whereas in eag Sh mutants the increase in the number of synaptic boutons results from an increased elongation of the presynaptic terminals and the formation of new secondary branches. In eag Sh larvae overexpressing APPL^sd, a reduction in the percentage of satellite boutons is observed when compared with APPLl^sd overexpressed in a wild-type background. However, the formation of extra parent boutons is still observed. Thus, eag Sh changes the pattern of endogenous Appl expression and partially attenuates the satellite bouton-promoting activity elicited by overexpression of Appl (Torroja, 1999).

It is suggested that Appl is involved in synaptic plasticity, because Appl is nonessential for formation and maintenance of synapses but can promote synapse formation and appears to be affected by neuronal activity. NMJ phenotypes resulting from the expression of Appl proteins lacking specific domains suggest that Appl trafficking and Appl-dependent signal transduction are two processes that regulate Appl-induced synaptic growth. The evidence indicates that (1) APP can be rapidly internalized (Koo, 1996); (2) processing and trafficking of Appl and APP is affected by activity (Allinquant, 1994); and (3) plasma membrane APP, and possibly Appl, behave as G_o-protein linked receptors (Nishimoto, 1993 and Torroja, 1999).

It is proposed that NMJ expansion occurs by two consecutive steps: the formation of sprouts and the consolidation of some of these sprouts into differentiated boutons. Bouton differentiation entails the proper arrangement of presynaptic and postsynaptic components, as well as the enlargement of the sprout to accommodate all the elements required for synaptic transmission. Both sprouting and differentiation are modulated by Appl. It is suggested that plasma membrane Appl induces sprouting and that this response is independent of Appl signal transduction. In contrast, bouton differentiation is regulated by Appl signal transduction, which may involve G_o. Internalization of Appl stops both activities. It is proposed that a satellite bouton is formed when Appl-induced sprouting is initiated, followed by rapid internalization of Appl, thus reducing Appl-dependent signal transduction and, therefore, bouton differentiation. As a result, some degree of differentiation, such as the formation of active zones and transport of vesicles, does occur, but other aspects, such as bouton enlargement, do not (Torroja, 1999).

Several predictions from this model are in line with these findings. For instance, decreasing Appl internalization (DeltaCI) would reduce formation of satellites. A persistent increase in Appl activation, as a result of decreased internalization, is predicted to promote the differentiation of sprouts, thus effectively increasing the number of parent boutons. In contrast, overexpression of Appl variants that are unable to undergo ligand-dependent receptor activation (DeltaE1, DeltaE2, DeltaCg) would reduce bouton differentiation, preventing the increase of parents but not of satellites. Full-length Appl (Appl⁺, Appl^sd) would stimulate both sprouting and differentiation and could be rapidly internalized as in wild type, thus promoting both parent and satellite bouton formation (Torroja, 1999).

Rapid Appl internalization appears to be key to the normally low level of plasma membrane Appl. Increases in plasma membrane-associated Appl result from increased neuronal activity, a factor shown to increase bouton number (Budnik, 1990). Strikingly, overexpression of Appl^sd in eag Sh results in reduction of satellites, with a concomitant increase in the number of parent boutons. This further supports the idea that neuronal activity can drive Appl-mediated bouton differentiation. Based on these observations, it is concluded that Appl has functional significance for the regulation of synapse formation. Moreover, this process involves the Appl domains that are likely to affect Appl signal transduction, suggesting a novel mechanism for the regulation of the size of synaptic arbors (Torroja, 1999).

Presenilin-dependent transcriptional control of the Aß-degrading enzyme Neprilysin by intracellular domains of ßAPP and APLP

Amyloid β-peptide (Aβ), which plays a central role in Alzheimer's disease, is generated by presenilin-dependent γ-secretase cleavage of β-amyloid precursor protein (βAPP). The presenilins (PS1 and PS2) also regulate Aβ degradation. Presenilin-deficient cells fail to degrade Aβ and have drastic reductions in the transcription, expression, and activity of neprilysin, a key Aβ-degrading enzyme. Neprilysin activity and expression are also lowered by γ-secretase inhibitors and by PS1/PS2 deficiency in mouse brain. Neprilysin activity is restored by transient expression of PS1 or PS2 and by expression of the amyloid intracellular domain (AICD), which is cogenerated with Aβ, during γ-secretase cleavage of βAPP. Neprilysin gene promoters are transactivated by AICDs from APP-like proteins (APP, APLP1, and APLP2), but not by Aβ or by the γ-secretase cleavage products of Notch, N- or E- cadherins. The presenilin-dependent regulation of neprilysin, mediated by AICDs, provides a physiological means to modulate Aβ levels with varying levels of γ-secretase activity (Pardossi-Piquard, 2005).

If Aβ production and degradation are tightly linked, this raises the question of why Aβ accumulates in AD. The net accumulation of Aβ in AD pathology likely reflects the cumulative effect of multiple events acting on production, fibrillogenesis, and degradation. In many forms of AD, especially the late-onset sporadic forms, it has not been shown that there is increased β- and γ-secretase activity. In fact, some have suggested that these forms may reflect defective degradation of Aβ. Therefore, AICD levels are likely to be unchanged in these late-onset forms of AD, and as a result, the AICD-mediated ability to upregulate neprilysin activity would not be efficiently brought into play to protect the brain. In contrast, in those cases of AD arising from mutations in APP and PS1, which activate γ-secretase and AICD production, the principal effect is to produce longer Aβ isoforms such as Aβ42. However, although Aβ40 is efficiently degraded by NEP, Aβ42 is degraded by NEP both in vitro and in vivo at a 6-fold lower rate. As a result, the upregulation of AICD (and thus, NEP expression), which would be anticipated in subjects with presenilin mutations, would not completely abolish the accumulation of Aβ42 in these cases. It should be noted that, in agreement with the above hypothesis, neprilysin expression and activity were higher only in brain tissues with familial Alzheimer’s disease linked to various presenilin-1 mutations, while sporadic AD cases displayed neprilysin levels similar to those exhibited by normal brain tissues. Interestingly, PS1 mutations selectively affect neprilysin and do not alter insulin-degrading enzyme expression (Pardossi-Piquard, 2005).

The above observations are also of direct practical interest because they indicate the possibility of new avenues for controlling Aβ levels without directly affecting γ-secretase. This latter concept is important because of the various developmental and postnatal side-effects associated with the inhibition of γ-secretase-mediated cleavage of other signaling molecules, including Notch. This work now suggests that Aβ levels might be modulated by directly increasing neprilysin expression, using AICD or small molecule mimics of AICD. Upregulation of neprilysin by transgenic overexpression, at least to modest levels, appears to be sufficient to reduce brain Aβ levels and to pose few toxic side effects. This strategy would also circumvent the other side effects of γ-secretase inhibitors, including the potentially self-defeating effect of reducing AICD and thus preventing NEP-mediated degradation of Aβ (Pardossi-Piquard, 2005).

GENE STRUCTURE

cDNA clone length - 5397

Exons - 7

Bases in 3' UTR - 2577

PROTEIN STRUCTURE

Amino Acids - 887

Structural Domains

Genomic and cDNA clones for a Drosophila gene resembling the human beta-amyloid precursor protein (APP) have been isolated. This gene produces a nervous system-enriched 6.5-kilobase transcript. Sequencing of cDNAs derived from the 6.5-kilobase transcript predicts an 886-amino acid polypeptide. This polypeptide contains a putative transmembrane domain and exhibits strong sequence similarity to cytoplasmic and extracellular regions of the human beta-amyloid precursor protein. The Appl protein contains multiple glycosylation sites. The transmembrane domain is located between amino acids 811 to 833, and the intracellular region possesses a potential clathrin-binding sequence (Rosen, 1989)

A Drosophila beta-amyloid protein precursor-like (Appl) gene has been molecularly delineated and its pattern of expression analyzed. Appl defines a new locus within the 1B division of the X-chromosome, a region previously shown to be important for neural development. The genomic limits of the Appl gene have been defined by mapping of the Appl cDNAs. The Appl transcript spans approximately 38 kb of genomic DNA. Genomic regions surrounding the first two exons have been sequenced. The first exon contains 78 nucleotides of the coding sequence and is separated from the second exon by a approximately 21 kb intron. The second exon is 171 nucleotides long and is separated from the third exon by a approximately 7 kb intron (Martin-Morris, 1990).

date revised: 19 February 2000

Home page: The Interactive Fly © 1995, 1996 Thomas B. Brody, Ph.D.

The Interactive Fly resides on the
Society for Developmental Biology's Web server.