Pyroglutamate–amyloid-β peptide expression in Drosophila leads to caspase-dependent and endoplasmic reticulum stress–related progressive neurodegeneration

Abstract

Alzheimer‘s disease (AD) is the most common neurodegenerative disorder among the elderly. During the progression of AD, massive neuronal degeneration occurs in the late stage of the disease; however, the molecular mechanisms responsible for this neuronal loss remain unknown. Aβ_pE3–42 (an N-terminal–truncated amyloid-β peptide that begins with pyroglutamate at the third position) is produced during late-stage AD. It also aggregates more rapidly in vitro and exhibits greater toxicity in neurons than full-length Aβ_1–42. In the present study, we established a Drosophila melanogaster model that expresses Aβ_3–42^E3Q, which effectively produces Aβ_pE3–42, and investigated the function of Aβ_pE3–42 using the photoreceptor neurons of Drosophila. Aβ_pE3–42 induced caspase-dependent apoptosis and caused progressive degeneration in photoreceptor neurons. Mutations in ER stress response genes or the administration of an inhibitor of the ER stress response markedly suppressed the degeneration phenotype, suggesting that the ER stress response plays an important role in neurodegeneration caused by Aβ_pE3–42. We also confirmed that human Tau-dependent apoptotic induction was strongly enhanced by Aβ_pE3–42. Thus, Aβ_pE3–42 expression system in the fly may be a promising new tool for studying late-onset neurodegeneration in AD.

Introduction

Alzheimer‘s disease (AD), also known as Alzheimer disease, is a chronic neurodegenerative disease that is the most common neuronal disorder among older individuals (1). The typical pathological characteristics of AD are the presence of intracellular neurofibrillary tangles (NFTs) and extracellular amyloid plaques, called senile plaques (SPs) (2). Massive neurodegeneration has also been reported in the cerebral cortex and hippocampus of AD patients (3,4). According to accurate volumetric measurements of brain size using magnetic resonance imaging protocols, a strong correlation exists between atrophy of the hippocampus and neuronal loss (5). Apoptotic cell death in neural cells during the later stage of the disease may be involved in brain atrophy in AD; however, the precise mechanisms responsible for this slow, progressive neurodegeneration have yet to be clarified (6).

NFTs consist of a hyper-phosphorylated microtubule-associating protein, called tau (7). Tau has been implicated not only in AD, but also in many other forms of dementia, such as frontotemporal dementia and Parkinsonism linked to chromosome 17 (FTDP-17) (8). A cross-species analysis revealed that the functions of these FTDP-17 mutations in tau appeared to be evolutionarily conserved because tau with FTDP-17 mutations induced neurodegeneration in several species (9–11). Since the formation of NFT appears to correlate with the region of neurodegeneration in the brains of AD patients, tau may play an important role in AD neurodegeneration (3); however, difficulties are associated with explaining all neurodegeneration-related phenomena based on the formation of NFT.

The study of rare families of AD led to the identification of three familial mutations: amyloid precursor protein (APP) and presenilins 1 and 2 (PS1 and PS2) (12). A small peptide, called amyloid beta (Aβ), is cleaved from APP and causes neuronal toxicity. Hardy and Allsop proposed a theory, called the amyloid hypothesis, in which the mis-regulation of APP, Aβ deposition and tau phosphorylation are the main events in the development of AD (13). Aβ is a major component of SPs and is produced from APP by sequential proteolysis with the enzyme complexes β-secretase (cysteine proteases and a β-site APP-cleaving enzyme) and γ-secretase (a multimeric protein complex composed of presenilin, nicastrin, Aph-1, and Pen-2) (14). The non-toxic form of Aβ, Aβ_1–40, is mainly generated during the early stage of Aβ production; however, the production of toxic Aβ_1–42 increases as AD progresses (2). Aβ_1–42 is suggested to be highly aggregated in vivo and exerts toxic effects on synaptic functions (1). Therefore, Aβ_1–42 is regarded as a causative element of cognitive defects in AD; however, recent findings obtained using mouse models for AD indicated that the simple overproduction of Aβ_1–42 does not induce prominent neuronal loss during AD (15). Thus, a factor other than full-length Aβ_1–42 may be involved in the induction of neural degeneration in AD.

In AD patients, the N-terminal region of Aβ_1–42 is truncated by aminopeptidase activity at either the second or tenth amino acid (16). These truncated peptide forms predominantly exist in AD plaques and accumulate with disease progression (17,18). They also form highly amyloidogenic pyroglutamate-Aβ (pE-Aβ) peptides as a result of the enzymatic activity of glutaminyl cyclase (QC), and produce Aβ_3pE–40, Aβ_3pE–42, Aβ_11pE–40, and Aβ_11pE–42 (19). Pyroglutamate formation increases the hydrophobicity of Aβ and β-sheet (aggregate form) stability, which promotes fibril ultrastructures (20,21). Thus, pE-Aβ exerts stronger toxic effects on neuronal activity than full-length Aβ (22–24). A recent study indicated that neurotoxicity induced by pE-Aβ is caused by lipid peroxidation, membrane permeabilization, and calcium influx, which is mediated by the NMDA receptor, in neurons (25).

The ER-mediated stress response, referred to the ER stress response, has recently gained attention as a causative factor for neuronal diseases, such as Parkinson’s and Huntington‘s diseases, familial and sporadic amyotrophic lateral sclerosis, and AD (26–29). The ER is a cellular organelle that has a number of functions, including the synthesis of carbohydrates, lipids, and proteins (29). Growing evidence suggests that the ER stress response could induce the inflammatory response (30). An inflammatory response begins when cells of the immune system and/or cells involved in metabolic processes sense pathogens or cellular damage, those of which trigger the release of inflammatory substances, including cytokines and/or other small molecules (31). It has been shown that inflammation and ER stress response are linked at any levels: both are necessary for the function and survival of the organism, and the ER stress signalling networks targets many types of inflammatory substrates (31). There are three ER-mediated responses that may lead to ER stress-mediated inflammation in cells: the cellular response against misfolded proteins, called the unfolded protein response (UPR); an abnormally high calcium concentration followed by the increasing activation of ion channel–coupled neurotransmitter receptors, called excitotoxicity; and higher levels of reactive oxygen species (ROS) (31–33). There are three ER transmembrane proteins that sense and mediate stress signaling to induce the ER stress response: inositol-requiring protein-1 (IRE-1), activating transcription factor-6 (ATF6), protein kinase RNA-like ER kinase (PERK), and activating transcription factor-4 (ATF4), a downstream factor of PERK (34). IRE-1 is a Ser/Thr kinase and catalyzes the splicing of X-box-binding protein 1 (Xbp1) mRNA to produce the active transcription factor XBP1s (35). XBP1s up-regulates a number of genes that regulate protein misfolding or protein quality control (35). ATF6 is cleaved by two different proteases after ER stress activation to produce the cytoplasmic domain of ATF6 (ATF6 50 kDa) (36). ATF6 50 kDa migrates into the nucleus and activates the transcription of ER protein-folding chaperones such as glucose-regulated protein-78 (GRP78, also known as BiP) and Gadd34 (also known as CHOP) (37).

D. melanogaster is a useful model for analysing the progression of AD (38). There is increasing evidence that endogenous signaling and cellular processes disrupted by Aβ42 induces apoptosis in Drosophila. Macro-autophagy has been implicated in cellular toxicity caused by Aβ42 in Drosophila; the overproduction of Aβ42 in the brain induced the activation of atg8 and caused age-dependent autophagic-lysosomal injury (39). ER stress is also known to be involved in neuronal degeneration caused by Aβ42 in Drosophila. Aβ42 was previously suggested to induce ER stress in the Drosophila eye through the activation of the unconventional splicing of XBP1 (40). Another cellular signaling pathway, such as the JNK/dFOXO or Toll/NF-kappaB pathway, has also been shown to play a role in Aβ42-induced photoreceptor degeneration (41,42). However, most cases were studied under severe developmental defects in the eye.

In order to elucidate the function of Aβ_pE3–42, a Drosophila model that expresses the pyroglutamated form of Aβ has been established and used to show that Aβ_pE3–42 and Aβ_pE11–42 cause severe neuronal toxicity when they are expressed in neural cells (43,44). In the present study, we investigated the mechanisms by which Aβ_pE3–42 leads to neuronal loss by comparing it to the activity of full-length Aβ, Aβ_1–42. Our results indicated that Aβ_pE3–42 induced progressive neuronal degeneration through the caspase-dependent and ER stress response-dependent induction of apoptosis. Although an endogenous mutation in the Drosophila homolog of tau (dtau) did not suppress the degeneration phenotype caused by Aβ_pE3–42, we observed a synergistic interaction between Aβ_pE3–42 and human tau that induces the expression of pro-apoptotic genes. These results suggest that Aβ_pE3–42 leads to neurodegeneration by two pathways: tau-dependent pro-apoptotic induction and tau-independent activation of the ER stress response. Thus, the study of Aβ_pE3–42 in Drosophila may be a useful model system for investigating the mechanisms underlying neurodegeneration during the development of AD.

Results

Production of Aβ_pE3–42 in Drosophila

Aβ_pE3–42 is derived from N-terminal modifications in Aβ_1–42 and is considered to play an important role in the progression of AD (19). In order to investigate the function of this N-terminal-modified small peptide, we established a model system for the study of Aβ_pE3–42 using D. melanogaster. Aβ_3–42^E3Q, in which glutamate is replaced by glutamine at the N-terminal region of Aβ_3–42 to rapidly produce the highly aggregative form, Aβ_pE3–42 (Fig. 1A) (43). Thus, we constructed transgenic lines that express Aβ_1–42 or Aβ_3–42^E3Q along with the rat pre-proenkephalin signal peptide sequence, which allowed the construct to be expressed outside of neural cells in Drosophila (Fig. 1A) (45). In order to compare the activity of Aβ_1–42 with that of Aβ_3–42^E3Q, we selected transgenic lines that showed similar mRNA expression levels in the organ: both chosen lines of Aβ_3–42^E3Q, Aβ_3–42^E3Q[#4-9], and two copies of the respective constructs of Aβ_1–42 (data not shown). Using the eye-specific Gal4-expressing line, GMR-Gal4, we expressed these molecules in the eye by UAS-based method and confirmed their expression using 6E10, a monoclonal antibody against Aβ, which recognizes the first 12 N-terminal amino acids of Aβ (Fig. 1B). Soluble and insoluble forms of Aβ were present at similar levels in the Aβ_1–42 and Aβ_3–42^E3Q lines (Fig. 1B). A previous study indicated that pE-Aβ acts as a scaffold molecule for Aβ_1–42 aggregation in vitro, and, thus, the expression of both pE-Aβ and Aβ_1–42 increases the aggregation of Aβ_1–42 (46). In order to confirm this in our system, we crossed the two transgenic lines to express Aβ_1–42 (42) and Aβ_3–42^E3Q (3Q) in the eye at the same time (42/3Q). In 42/3Q flies, the amount of Aβ in the insoluble fraction was markedly higher than the amount of Aβ, and was even greater than the calculated sum from the individual Aβ_1–42 and Aβ_3–42^E3Q fractions (42 + 3Q; Fig. 1B, right panel). However, we were unable to conclude that pE-Aβ acts as a scaffold molecule for Aβ_1–42 aggregation from this experiment because it was not possible to completely distinguish Aβ_1–42 from Aβ_3–42^E3Q under these conditions. During this experiment, we noted that 82E1, another antibody against Aβ, which is specific for the N-terminal end of the molecule, did not recognize Aβ_3–42^E3Q (Fig. 1C) (47). This may be because the epitope recognized by 82E1 includes the very N-terminal region of Aβ_1–42, which is deleted in Aβ_3–42^E3Q (Fig. 1A). Using 82E1, we confirmed that the amount of Aβ, particularly in the insoluble fraction, was increased in 42/3Q flies (Fig. 1C). These results suggest that pE-Aβ acts as a scaffold for the aggregation of Aβ_1–42in vivo (48).

Figure 1.

Establishment of flies that express Aβ_1–42 and Aβ_3–42^E3Q. (A) The Aβ_1–42 (42) and Aβ_3–42^E3Q (3Q) constructs with the rat pre-proenkephalin secretion signal (Sec). Pyroglutamated Aβ, Aβ_pE3–42 (pE-Aβ) is shown for comparison. These constructs were expressed in the eye by UAS-based expression system. (B) (Left) Western blot analysis using pooled head extracts from transgenic flies of each genotype 1 week after eclosion and anti-Aβ_1–42 (6E10) or anti-tubulin (tub). 42 (42^WT): GMR-Gal4/Y; UAS-Aβ_1–42/+; UAS-Aβ_1–42/+. 3Q: GMR-Gal4/Y; UAS-Aβ_3–42^E3Q/+. 42/3Q: GMR-Gal4/Y; UAS-Aβ_1–42/+; UAS-Aβ_1–42/UAS-Aβ_3–42^E3Q. Results from three independent replicates are shown. (Right panel) Quantitative analysis of the expression of Aβ based on the Western blot analysis using 6E10. Soluble (TBS-soluble) and insoluble (TBS-insoluble and FA-soluble) forms of Aβ were quantified. The light gray bar represents the expected summation of 42 (42^WT) and 3Q (42 + 3Q) proteins. (C) (Left panel) Western blot analysis as in B using 82E1. (Right panel) Quantitative analysis of Aβ expression based on the Western blot analysis using 82E1. Soluble and insoluble forms of Aβ were measured. ND: not detectable. (D) Time-course analysis of Aβ_3–42^E3Q expression in the heads of 3Q flies. Soluble or insoluble Aβ or pE-Aβ was observed using 6E10 or anti-pE (N3pE, rehybridized) (also see Supplementary Material, Fig. S1) (E) Cellular fractions of adult heads expressing 42, 3Q, or 42/3Q using TBS, 1% Triton-X100 (Tx), 1% Sarkosyl (Sar), 2% SDS (SDS), and 100% FA (FA). Western blot analysis using fractionated samples from transgenic flies of each genotype 1 week after eclosion and anti-Aβ (6E10 or 82E1) for detection. (F) Quantitative analysis of the amount of Aβ42 or 3Q in each fraction using anti-6E10 (based on the data of Supplementary Material, Fig. S1). TBS fractions of 42 or 3Q were normalized into 1 relative value, respectively. (G) Quantitative analysis of the amount of Aβ42 in each fraction using anti-82E1 (based on the data of Supplementary Material, Fig. S1). TBS fraction of 42 was normalized into 1 relative value. Student‘s t-test was performed to examine the significance of differences.

Aβ_3–42^E3Q is rapidly modified into its pyroglutamated form (43,44); however, no time-course analysis has been performed in the Drosophila system. We overexpressed Aβ_3–42^E3Q under the eye-specific Gal4 line, GMR-Gal4, and monitored extracted head proteins from multiple time points after eclosion. The soluble and insoluble forms of Aβ were both increased during this 16-day period (Fig. 1D). In order to confirm when pyroglutamation was initiated, we used an antibody against pE. The emergence of pE expression was slightly more abundant than Aβ, and other stages were similar to that of Aβ, suggesting that Aβ_3–42^E3Q was transformed into Aβ_pE3–42 soon after translation (Fig. 1D;Supplementary Material, Fig. S1). Our results are slightly different from previous report done by Jonson et al. (2015); they reported that there was no soluble fraction from flies, which are expressing Aβ_3–42^E3Q (44). We found that the soluble form of Aβ_3–42^E3Q was hardly detected at the early stage of adult stage (within several days after eclosion), however, it increased expression level there after (Fig. 1D). Thus, we believe that the difference about the soluble fraction from the previous report is the timing of the fractionation.

A previous study indicated that the detergent solubility of the TAR DNA-binding protein (TDP-43) correlated with neuronal toxicity in Drosophila (49). We performed cellular fraction experiments using adult heads expressing 42, 3Q, or 42/3Q and quantified the amount of Aβ_1–42 or Aβ_3–42^E3Q present in the serial treatment with TBS, 1% Triton-X-100 (Tx), 1% Sarkosyl (Sar), 2% SDS (SDS), and 100% FA (FA) (See the Materials and Methods). We found that insoluble forms of Aβ_3–42^E3Q were more resistant to the 2% SDS treatment than Aβ_1–42 (Fig. 1E and F;Supplementary Material, Fig. S2). This result supports previous mammalian findings showing that Aβ_pE3–42 is more aggregative than Aβ_1–42 (17). Aβ_1–42 appeared to become more resistant to 2% SDS than Aβ_1–42 when Aβ_3–42^E3Q was co-expressed (Fig. 1E and G). This result supports previous findings showing that pE-Aβ acts as a scaffold for the aggregation of Aβ_1–42, which is known as the seed hypothesis (48).

Aβ_3–42^E3Q induces progressive retinal degeneration

The strong expression of Aβ_1–42 in neural cells causes locomotive defects in Drosophila (35,38). We confirmed previous findings that pE-Aβ shows a more severe locomotive defect than Aβ_1–42 by overexpressing Aβ_3–42^E3Q or Aβ_1–42 using the pan-neuronal Gal4 line, elav-Gal4, as reported previously (Supplementary Material, Fig. S3) (50). The compound eyes of Drosophila have been used to analyse human neurodegenerative disorders (51). In order to identify late-onset neurodegeneration associated with the expression of Aβ, we observed the retinal phenotype 6 weeks after eclosion when Aβ_1–42 or Aβ_3–42^E3Q was overexpressed in the retina. Morphologically, Aβ_1–42- and Aβ_3–42^E3Q-expressing flies showed similar mild defects (Fig. 2A–C). An analysis of retinal tissue sections showed that only Aβ_3–42^E3Q-expressing flies had severe retinal degeneration by 6 weeks after eclosion (Fig. 2D–F).

Figure 2.

Aβ_pE3–42 leads to progressive retinal degeneration. (A–C) Eye morphology of adult flies 6 weeks after eclosion. (A) Control Oregon-R (wild type) flies showed the normal array of ommatidia. Expression of (B) Aβ_1–42 (42^WT; GMR-Gal4/Y; UAS-Aβ_1–42/+; UAS-Aβ_1–42/+), or (C) Aβ_3–42^E3Q (3Q; GMR-Gal4/Y; UAS-Aβ_3–42^E3Q/+) in the eye led to a slightly abnormal eye morphology. (D–F) Retinal sections from (D) wild-type, (E) 42^WT, and (F) 3Q flies 6 weeks after eclosion. (G–J) Time-course analysis of Aβ_3–42^E3Q-expressing flies (G) 5 days, (H) 2 weeks, (I) 4 weeks, and (J) 6 weeks after eclosion. Arrows indicated ommatidia with abnormal structures (arrows indicate degenerative figures) (K) Summary of ommatidium defects at different time points after eclosion. We assessed the defect of ommatidia in two ways; the percentage of normal ommatidia (white bars) or total ommatidia (closed bars). Cont (Oregon-R): n = 8 flies and n = 116 ommatidia were counted (8 flies, 116 ommatidia). 42^WT: (7 flies, 210 ommatidia). 3Q (5 days): (9 flies, 262 ommatidia). 3Q (2 weeks): (7 flies, 190 ommatidia). 3Q (4 weeks): (9 flies, 289 ommatidia). 3Q (6 weeks): (8 flies, 253 ommatidia). Data are expressed as the mean ± SEM. Data sets of each time point were compared with 3Q (5 days). The Student‘s t-test was performed to examine the significance of differences. *P < 0.01; **P < 0.05.

In an attempt to clarify the timing of retinal degeneration, we analysed retinal sections at different time points in Aβ_3–42^E3Q-expressing flies (Fig. 2G–J). At the initial stage (5 days after emergence), most ommatidia contained the correct number and configuration of rhabdomeres (a dark structure consisting of multiple microvilli near the center of ommatidia) of photoreceptor cells (Fig. 2G). Two weeks after eclosion, many abnormal ommatidia were present; the number of photoreceptor cells in each ommatidium appeared to be reduced, and the array of photoreceptors within each ommatidium was disorganized (Fig. 2H, arrows). By 4 weeks after eclosion, the % of abnormal ommatidia had increased further (Fig. 2I). Most ommatidia had degenerated 6 weeks after eclosion (Fig. 2J, arrow heads). We counted the % of ommatidia in two ways, normal ommatidium (containing seven rhabdomeres surrounded by pigment cells) or total ommatidia (containing rhabdomere(s) including normal ommatidia), and found that the expression of Aβ_3–42^E3Q progressively reduced the % of ommatidia (Fig. 2K). These results suggest that pE-Aβ causes progressive retinal degeneration during the adult stage.

Aβ_3–42^E3Q induces the expression of ER stress response markers

In many cases of neurodegeneration, caspase-dependent apoptotic induction is responsible for the resulting cell death (52). Therefore, we introduced p35, which encodes a baculovirus-derived caspase inhibitor, into Aβ_3–42^E3Q-expressing flies (53). The neurodegeneration phenotype caused by Aβ_3–42^E3Q was strongly suppressed by p35 (Fig. 3A and B). We counted the numbers of normal ommatidia or total ommatidia (Fig. 3C). Although many ommatidia did not appear to have normal numbers of rhabdomeres, most seemed to be restored by the treatment (Fig. 3C). We also stained 4 weeks old adult eyes with anti-Death caspase-1 (Dcp1) antibody, which is a marker for the caspase activation. While Aβ_3–42^E3Q-expression caused Dcp1 positive photoreceptor cells at cell bodies, it was not observed Dcp1 expression in control flies or Aβ_3–42^E3Q and p35 expressing flies (Supplementary Material, Fig. S4) (54). These results suggest that Aβ_3–42^E3Q induces caspase activation in fly eyes, and caused degeneration.

Figure 3.

Aβ_pE3–42 causes caspase-dependent apoptosis and induces ER stress marker expression. (A,B) Retinal sections from a (A) 3Q (GMR-Gal4/Y; UAS-Aβ_3–42^E3Q/+) and (B) 3Q + p35 (GMR-Gal4/Y; UAS-Aβ_3–42^E3Q/UAS-p35) fly 6 weeks after eclosion. p35 inhibited Aβ_pE3–42-induced retinal degeneration. (C) We counted the % of normal ommatidia (white bars) or total ommatidia (closed bars) of cont (Oregon-R), 3Q, and 3Q + p35 flies 6 weeks after eclosion. cont: (8 flies, 116 ommatidia), 3Q: (8 flies, 253 ommatidia), and 3Q + p35: (7 flies, 210 ommatidia) were counted. Data are expressed as the mean ± SEM. The Student‘s t-test was performed to examine the significance of differences. *P < 0.01. (D) Expression of a hid-lacZ reporter was not affected by Aβ_3–42^E3Q expression. Relative mRNA expression of lacZ or rp49 as assessed by qPCR in +/+ (+/Y; hid-lacZ/+) and 3Q (GMR-Gal4/Y; UAS-Aβ_3–42^E3Q/hid-lacZ) flies at the indicated times after eclosion. Data were normalized relative to rp49. (E) Western blot analysis using anti-Hsc3. Aβ_3–42^E3Q-expressing flies showed higher HSC3 protein levels. (F,G) qPCR analysis of (F) HSC3 and (G) XBP1s using mRNAs from the heads of different fly strains as indicated. Three time points were examined: 1 week, 2 weeks, and 4 weeks after eclosion. GFP: GMR-Gal4/Y; UAS-EGFP/+, 42^WT: GMR-Gal4/Y; UAS-Aβ_1–42/+; UAS-Aβ_1–42/+, 3Q: GMR-Gal4/Y; UAS-Aβ_3–42^E3Q/+. Data are expressed as the mean ± SEM. *P < 0.01; **P < 0.05, ^#P > 0.05. n = 4.

One of the major pathways that activate caspase-dependent apoptosis is the induction of pro-apoptotic genes that are regulated by stress signaling pathways, such as c-jun-N-terminal kinase (JNK) (55). We investigated whether the JNK pathway is activated along with Aβ_3–42^E3Q expression using an antibody against phosphorylated (p)-JNK and found that there was no JNK activation in the presence of Aβ_1–42 or Aβ_3–42^E3Q expression (data not shown). We then observed the expression of hid, a pro-apoptotic gene that is expressed as part of the JNK signaling pathway, by monitoring the expression of lacZ mRNA in a hid-lacZ reporter strain (56). However, the induction of lacZ under Aβ_3–42^E3Q expression was not observed, even 4 weeks after eclosion (Fig. 3D). We also detected the expression of reaper, another pro-apoptotic gene, using real-time quantitative PCR (qPCR); however, no changes were observed in its expression (Supplementary Material, Fig. S5). These results suggest that the activation of JNK and subsequent induction of pro-apoptotic genes do not contribute to retinal degeneration induced by pE-Aβ.

The ER stress response has recently emerged as another inducer of caspase-dependent apoptosis and may be involved in the development of AD (57). In order to monitor the ER stress response quantitatively, we analysed the expression of HSC3, the Drosophila homolog of GRP78/BiP, with a HSC3-specific antibody (58). In wild-type flies, the HSC3 protein was present 1 week after eclosion and it expression was slightly increased at 4 weeks (Fig. 3E). The expression of HSC3 did not increase in the presence of Aβ_1–42, but was stronger in the presence of Aβ_3–42^E3Q (Fig. 3E). A quantitative analysis of mRNA levels using qPCR supported Aβ_3–42^E3Q expression inducing HSC3 transcription (Fig. 3F). Aβ_1–42 expression also appeared to induce Grp78 expression; however, the induction level was not as high as that with Aβ_3–42^E3Q.

We then analysed the level of the small splicing variant of XBP1, XBP1s, which is another marker of the ER stress response because it is produced under UPR stress (35). Although the background level was high, the qPCR analysis of XBP1s mRNA using primer sets specific for this splice variant showed that XBP1 splicing correlated with Aβ_3–42^E3Q expression (Fig. 3G). Collectively, these results suggest that pE-Aβ expression leads to the activation of the ER stress response.

ER stress activation is involved in neurodegeneration caused by Aβ_3–42^E3Q

In order to clarify whether the ER stress response is involved in retinal degeneration induced by Aβ_3–42^E3Q, we used pharmacological and genetic approaches. We administrated the ER stress inhibitor, tauroursodeoxycholic acid (TUDCA) to Aβ_3–42^E3Q-expressing flies and found that this treatment suppressed the degeneration phenotype caused by Aβ_3–42^E3Q (Fig. 4A, B and G) (59). We then searched for genetic interactions between Aβ_3–42^E3Q and mutations in several mediators of the ER stress response. Reductions in the gene dosage of Ire1 strongly suppressed the degeneration phenotype, suggesting that Ire1 mediates the degeneration caused by Aβ_3–42^E3Q (Fig. 4C and G). Although suppression was relatively weak, a mutation in dXBP1, ATF6, or ATF4 partially suppressed the degeneration phenotype induced by Aβ_3–42^E3Q at 6 weeks after eclosion (Fig. 4D–G). These results strongly suggest that the ER stress response is a mediator of the neurodegeneration caused by pE-Aβ.

Figure 4.

The ER stress response is involved in the retinal degeneration caused by Aβ_pE3–42. (A–F) Retinal sections of 3Q flies in combination with heterozygote mutations of the ER stress response genes at 6 weeks after eclosion. 3Q: GMR-Gal4/Y; UAS-Aβ_3–42^E3Q/+. (A) 3Q + DMSO: 3Q flies were fed 1% DMSO (vehicle control) every other day for 6 weeks. (B) 3Q + TUDCA: 3Q flies were fed 100 µM TUDCA in 1% DMSO every other day for 6 weeks. (C) 3Q; Ire1^–/+ (GMR-Gal4/Y; UAS-Aβ_3–42^E3Q/Ire1[f02170]). (D) 3Q; Atf4^+/− (GMR-Gal4/Y; crc[1]/; UAS-Aβ_3–42^E3Q/+). (E) 3Q; Atf6^–/+ (GMR-Gal4/Y; Atf6[c05057]/+; UAS-Aβ_3–42^E3Q/+). (F) 3Q; XBP1^+/– (GMR-Gal4/Y; XBP1[k13803]/+; UAS-Aβ_3–42^E3Q/+). Regarding C–F, 3Q flies were crossed into a mutant line to generate flies with only one functional copy of the gene of interest. (G) Summary of the % of normal ommatidia (white bars) or total ommatidia (closed bars) at each genotype. Cont (Oregon-R) +DMSO: (6 flies, 157 ommatidia), 3Q + DMSO: (8 flies, 232 ommatidia), 3Q + TUDCA: (7 flies, 218 ommatidia), 3Q; Ire1^–/+: (8 flies, 257 ommatidia), 3Q; Atf4^+/–: (9 flies, 288 ommatidia), 3Q; Atf6^–/+: (6 flies, 152 ommatidia), and 3Q; XBP1^+/–: (7 flies, 238 ommatidia) were counted. Data are expressed as the mean ± SEM. Data sets of each genotype were compared with 3Q + DMSO.

Aβ_3–42^E3Q synergistically enhances tau-dependent pro-apoptotic gene induction

Previous studies suggested that a genetic interaction exists between Aβ and tau (60,61). In order to investigate whether the endogenous Drosophila homolog of tau (dtau) mediates the retinal degeneration induced by Aβ_3–42^E3Q, we examined the degeneration phenotype in the background of a dtau mutation (dtau^EP3203/dtau^MR22). Although we confirmed that this background showed a low level of dtau mRNA, the degeneration phenotype did not appear to be affected (Fig. 5A–C). These results suggest that dtau is not a mediator for pE-Aβ–induced neuronal degeneration.

Figure 5.

Genetic interaction between Aβ_pE3–42 and tau. (A,B) Retinal sections of each genotype 6 weeks after eclosion. (A) GMR > 3Q; +/+ (GMR-Gal4/Y; UAS-Aβ_3–42^E3Q/+; +/+). (B) GMR > 3Q; dtau^–/– (GMR-Gal4/Y; UAS-Aβ_3–42^E3Q/+; dtau^EP3203/dtau^MR22). (C) dtau mRNA expression was monitored by a qPCR analysis using mRNA from the heads of GMR > 3Q; +/+ (3Q) and GMR > 3Q; dtau^–/– (3Q; dtau^-) flies. Data were normalized relative to rp49. Data are expressed as the mean ± SEM. *P < 0.01. (D–I) Scanning electron micrographs of the adult compound eye. (D) Oregon-R. (E) GMR-Gal4/Y; UAS-tau^WT/+ (tau^WT/+). (F) GMR-Gal4/Y; UAS-Aβ_3–42^E3Q/+ (3Q/+). (G) GMR-Gal4/Y; UAS-EGFP/+ (GFP). (H) GMR-Gal4/Y; UAS-tau^WT/UAS-Aβ_3–42^E3Q (tau^WT/3Q). (I) GMR-Gal4/Y; UAS-tau^R406W/+ (tau^R406W/+).

We then investigated the synergistic interaction between Aβ_3–42^E3Q and tau. We established transgenic flies that express the wild-type human tau full-length isoform (2 N/4 R) (tau^WT: MAPT) in the eye. Although it has been reported that tau^WT expression leads to severe morphological defects in the Drosophila eye (62), our strain only had a mild morphological defect, which was similar to the effect of Aβ_3–42^E3Q expression (Fig. 5D–G). This discrepancy may be due to the different expression levels between the two strains. In an additional strain in which tau^WT and Aβ_3–42^E3Q were both overexpressed in the eye, this co-expression induced severe morphological defects (Fig. 5H). The morphological defects caused by this co-expression were similar to those caused by the overexpression of tau^R406W, which has a causative mutation for FTDP-17 (Fig. 5I) (63,64). The severe toxicity observed with the expression of tau^R406W is considered to result from this mutation generating an active form of tau (11). These results suggest that the effects of tau^WT were enhanced by the presence of pE-Aβ.

In order to clarify whether this synergistic interaction between tau^WT and Aβ_3–42^E3Q is due to the activation of the ER stress response pathway, we investigated the expression of HSC3 under these conditions. However, the activation of HSC3 expression was not detected when Aβ_3–42^E3Q and tau^WT were overexpressed (Fig. 6A). In addition, we found that tau^R406W did not induce the expression of this ER stress marker (Fig. 6A), as was reported previously (65). Thus, morphological defects in Aβ_3–42^E3Q/tau^WT double-transgenic flies may be mediated by pathways other than the ER stress response.

Figure 6.

Aß_pE3–42 and tau synergistically induce the expression of pro-apoptotic genes. (A–D) qPCR analysis of ER stress response genes and pro-apoptotic genes using mRNA from the heads of 1-week-old adult flies with the genotype defined in Figure 5. Data were normalized relative to rp49. Primer sets were used to amplify (A) HSC3, (B) reaper, (C) hid, and (D) grim mRNA. (E) Post-developmental expression of Aβ_pE3–42 with htau induced pro-apoptotic gene expression. qPCR analysis of ER stress response genes and pro-apoptotic genes using mRNA from the heads of flies 1 week after the induction (temperature shift to 29 °C) of 3Q and dtau using the elav-Gal4 and tub-Gal80ts system. Data were normalized relative to rp49. The following primer sets were used for amplification: HSC3, reaper, hid, grim, and rp49 mRNA. Data are expressed as the mean ± SEM. *P < 0.01. Control: elav-Gal4/Y; tub-Gal80ts/+; +/+. 3Q/htau: elav-Gal4/Y; tub-Gal80ts/UAS- Aβ_pE3–42; UAS-htau/+. Data are expressed as the mean ± SEM. *P < 0.01; **P < 0.05; ^#P > 0.05.

The activation of stress-induced signaling pathways strongly correlated with the phosphorylation of tau in brains from individuals with AD (66–68). Therefore, we investigated the expression of the pro-apoptotic genes reaper, hid, and grim, which are inducers of apoptosis and targets of stress signaling pathways. Although tau^WT did not activate the expression of these pro-apoptotic genes, tau^R406W expression led to a significant increase (Fig. 6B–D). The combined expression of Aβ_3–42^E3Q and tau^WT also induced a significant increase in the expression of reaper, hid, and grim (Fig. 6B–D). These results suggest that tau activation induces stress signaling, and pE-Aβ enhances the expression of pro-apoptotic genes when there is a high level of tau expression. However, severe morphological defects may contribute to the induction of pro-apoptotic genes. In order to avoid the contribution of developmental defects, we induced Aβ_3–42^E3Q and tau^WT at the adult stage using the elav-Gal4 and tub-Gal80ts system (69). Although the induction of Aβ_3–42^E3Q and tau^WT decreased the expression of HSC3, pro-apoptotic genes (reaper, hid, and grim) were induced (Fig. 6E). This result supports Aβ_3–42^E3Q and tau^WT synergistically inducing pro-apoptotic gene expression.

Discussion

Drosophila compound eyes expressing pE-Aβ may be a model for studying AD-related neurodegeneration

An ideal in vivo model system for studying apoptotic induction during AD currently does not exist. In order to analyse the molecular mechanisms of AD, laboratories have attempted to establish a transgenic mouse model in which mutated forms of APP are expressed in the central nervous system (15,70). These APP mouse models successfully reflect the deficits in learning and memory formation that correlate with the pathology of AD (15,70); however, neurodegeneration is rarely observed (71).

D. melanogaster provides another tool for the study of AD. The overexpression of human Aβ_1–42 or its derivatives in the fly nervous system leads to decreased neuronal activity, including learning and memory formation (72,73). In terms of neurodegeneration, there have been conflicting observations. Previous studies indicated that Aβ_1–42 leads to retinal degeneration when it is expressed in the eye (74,75). These flies also showed severe morphological defects in the eye. In contrast, Sofola et al. reported that the induction of Aβ_1–42 during the adult stage did not induce retinal degeneration (76). This is consistent with our result that the expression of Aβ_1–42 in the eye resulted in very mild morphological defects and did not cause retinal degeneration. Collectively, these findings suggest that the moderate expression of Aβ_1–42 does not induce retinal degeneration.

In the brains of AD patients, most of the Aβ found in SPs is pyroglutamated (17,20). In contrast, the expression of pE-Aβ is extremely weak in APP mouse models (77,78). Since neurodegeneration is rare in APP mouse models, pE-Aβ may play an important role in neurodegeneration in AD (15,23,25). In the present study, we demonstrated that pE-Aβ expression in the retina of Drosophila caused caspase- and ER stress response–dependent progressive retinal degeneration. These results suggest that the expression of pE-Aβ in the Drosophila compound eye has potential as a model for studying neurodegeneration in AD. These experiments have been done by UAS-based expression system, therefore, we do not know if the expression levels of our constructs are comparable levels to the endogenous events. However, our data clearly show that pGlu-Aβ42 is more toxic than Aβ42, when they were expressed under the same condition.

ER stress induction and AD formation

In the present study, we found that Aβ_3–42^E3Q expression activated the ER stress response and caused retinal degeneration as flies aged, whereas Aβ_1–42 expression did not induce retinal degeneration. This is distinct from the previous findings by Ryoo et al. showing that Aβ_1–42 expression may induce a strong ER stress response and cause retinal degeneration (64). As we discussed above, Aβ_1–42 expression levels may have been markedly higher in their fly strain because severe morphological defects were present in the eye (64). In support of this, we noted that Aβ_1–42 expression also caused a slight increase in the expression of an ER stress marker; however, its induction was not as strong as in the presence of Aβ_3–42^E3Q (Fig. 3F and G). Thus, we conclude that pE-Aβ is a stronger inducer of the ER stress response than Aβ_1–42 and causes retinal degeneration at later stages.

Chronic inflammation has been implicated as a prominent component in the development of AD, and, thus, ER stress may trigger chronic inflammation that leads to neurodegeneration in the brains of AD patients (79). A recent genetic linkage study using 276 patients with AD revealed that a polymorphism in the XBP1 promoter region was strongly associated with an increased risk of AD (80). Additionally, GRP78 was shown to be increased in the brain of an AD patient (81), as we observed in the retina of flies that express Aβ_3–42^E3Q. These findings suggest that the pathway that mediates ER stress activation in response to pE-Aβ is evolutionarily conserved and plays an important role in the progression of AD. Although we did not investigate whether immune cells, such as macrophages, are involved in the degeneration phenotype of Aβ_3–42^E3Q, our results suggest that Aβ_3–42^E3Q-induced retinal degeneration provides a useful tool for analysing the link between the ER stress response and development of AD. Since the administration of an ER stress inhibitor into fly food blunted the neurodegeneration phenotype (Fig. 4B), the pE-Aβ system in flies may be an efficient tool for the study of therapeutic drugs for AD that protect against neurodegeneration.

The relationship between pE-Aβ and tau in the development of AD

Evidence has accumulated to show that Aβ and tau are causative elements of AD, and there appears to be cross-talk between these two molecules (60). A previous study suggested that ER stress–positive (i.e., Grp78-expressing) neurons are protected against tau-induced neuronal toxicity in the brain because most neurons that contained phosphorylated tau were negative for Grp78 in the brains of AD patients (81). At least two models have been developed in order to explain the relationship between Aβ and tau: the accumulation of Aβ leads to the hyperphosphorylation of tau, which results in synaptic and neuronal loss, or Aβ accumulation and the hyper-phosphorylation of tau are parallel events that both induce apoptosis (61). In the present study, we confirmed that a synergistic interaction exists between pE-Aβ and tau. In the parallel model, at least two possibilities may be present in terms of the dependency of the tau-related induction of apoptosis on the ER stress response (Supplementary Material, Fig. S6). The synergistic interaction between Aβ_3–42^E3Q and tau did not activate the ER stress response, but did induce the expression of pro-apoptotic genes, which appeared to be tau-dependent. Thus, our results suggest that one of the parallel models clearly describes the relationship between pE-Aβ and tau, and that pE-Aβ induces apoptosis through the ER stress response and tau-dependent pro-apoptotic pathways (Supplementary Material, Fig. S6).

Materials and Methods

Drosophila stocks

Flies were maintained in vials with SY medium at 25 °C, with a 12 h light/12 h dark cycle. Aβ_1–42 and Aβ_3–42^E3Q, both of which were preceded in-frame by the rat pre-proenkephalin signal peptide (Sec), and human Tau (2 N/4 R) were cloned into pUAST vectors. BestGene Inc. carried out the injection of these vectors to generate the initial strains. We used Oregon-R as a wild type. In this study, we used the following transgenic strains: w[1118]; UAS-EGFP[S65T]/CyO, w[1118]; UAS-Aβ_3–42^E3Q/CyO, w[1118]; UAS-Aβ_3–42^E3Q[#4-9]/TM3, w[1118]; UAS-Aβ_1–42; UAS-Aβ_1–42/S: T.

The transgenic flies used in this experiment were crossed with GMR-GAL4 or elav-GAL4 lines (elav-GAL4c155), w[1118]; Ire1[f02170]/TM6B, w[1118]; Atf6[c05057], cn[1]crc[1]/SM5 (Fly ATF4 line), w[67c23]; XBP1[k13803]/CyO, w[1118]; tau^EP3203, w[1118]; tau^MR22/TM3 and w[1118]; tub-Gal80ts (II) (Bloomington Drosophila Stock Center). UAS-p35 (53), hid-lacZ (82), and UAS-tau^R406W (83) were also used.

GAL4/UAS/Gal-80ts system

Gal80ts is a temperature sensitive mutant of Gal4 inhibitor, and has been shown to be inactivated at 29 °C (84). To induce both Aβ_pE3–42 and htau in neurons at the adult stage, we have established the flies, elav-Gal4/Y; tub-Gal80ts/UAS-Aβ_pE3–42; UAS-htau/+, and shifted the temperature from 18 to 29 °C at 1 week after eclosion (69).

Cellular fractions and western blot analysis

Heads were collected from 20 flies per strain at the indicated age and genotype, and were homogenized in TBS-buffer containing protease inhibitors. The resulted homogenate was sonicated using a Bioruptor (CosmoBio) before ultracentrifugation at 100, 000 × g at 4 °C for 60 min. The supernatant was collected as the TBS-soluble fraction. The pellet was dissolved in 100% formic acid (FA) and then centrifuged at 17,800 × g at 4 °C for 20 min. This supernatant was collected as the TBS-insoluble fraction and was dried at room temperature for 1 hr in SpeedVac. The pellet was resuspended in Laemmli buffer. In order to perform the detergent solubility assay, the pellet of the TBS-soluble fraction was further extracted with different detergents. The pellet was extracted with Triton-X100-containing buffer (TBS, 1% Triton-X100, pH 7.6) and ultracentrifuged at 100, 000 × g at 4 °C for 60 min. The supernatant from this step represented as the 1% Triton-X fraction. The pellet was further extracted with Sarkosyl-containing buffer [TBS, 1% Sarkosyl (v/v)] and ultracentrifuged at 100, 000 × g at 4 °C for 60 min. The supernatant from this step represented as the 1% Sarkosyl fraction. The pellet was further extracted with SDS-containing buffer (TBS, 2% SDS, pH 7.6) and ultracentrifuged at 100, 000 × g at 4 °C for 60 min. The supernatant from this step represented as the 2% SDS fraction. The pellet was then dissolved in 100% FA and prepared as above. The sample from the final step represents as 100% FA fraction. In each step, we performed sonication using a Bioruptor before the ultracentrifuge treatment.

Each fraction was separated on a 15–20% Tris-tricine gel (Wako) at room temperature for 2 hr, and proteins were transferred to a 0.22-μm nitrocellulose membrane (Whatman) at 4 °C for 1 hr. The membrane was blocked with 5% skim milk (Difco). The following antibodies were used for immunoblotting: 6E10 (Covance), 82E1 (IBL), N3pE (IBL), α-tubulin (Sigma), and HSC3 (Bisciences). Protein signals were visualized with Immunostar Zeta (Wako). Synthetic Aβ42 (Syn.Aβ42) (Peptide Institute, Inc.) was used as a control.

Climbing assay

In the climbing assay [also referred to as a negative geotaxis assay (45)], 20 flies were placed in an empty plastic vial (diameter, 3 cm; height, 20 cm) that had been marked to divide it into three equal areas along its height: bottom, middle, and top.

The vial was tapped gently so that all flies fell into the bottom area and was then placed on a tabletop. After 1 min, the number of flies in the top area was scored. Each assay was repeated three times, and average values (%) of the top area were calculated for each line.

qPCR analysis

qPCR was performed as described previously (56). Briefly, total RNA was extracted from 30 fly heads using an RNA purification kit (QIAGEN). cDNA was synthesized with a PrimeScript RT reagent kit (TaKaRa). SYBR Premix Ex-TaqII (TaKaRa) was used for amplification with 40 cycles of 95 °C for 10 s and 60 °C for 30 s. The thermal Cycler Dice Real Time System (TaKaRa) was used for amplification and data analyses; the relative value program was used to show expression levels. Data were normalized relative to rp49.

The following primers were used:

Grp78, Forward: 5’-GAATCAGTTGACCACCAATCCC -3’ and Reverse: 5’-AACTTGATGTCGTGTTGCACA -3’;

dtau, Forward: 5’- AACAACCACCACCCCATCAG -3’ and Reverse: 5’-CTCGCATTTGCTGGCTTCTG-3’.

A primer set for the splice variant form of XBP1 (XBP1s) was described in a previous study (65). Primer sets for hid, grim, reaper, rp49, and lacZ are described in a previous study (56).

Histochemistry

The sectioning of plastic-embedded fly eyes was performed as described previously (85). Briefly, dissected eyes were fixed in 2.5% glutaraldehyde, dehydrated, and embedded in Epon plastic. Thin sections were stained with toluidine blue for light microscopy. Slide images were captured by microscopy (NIKON).

Whole mount staining of adult eyes with Rabbit anti-Dcp1 (Cell Signaling Technology), Rat Elav (DSHB) and mouse anti-Arm (DSHB) was performed as previously (86). Briefly, dissected adult eyes were fixed with 4% PFA for 15 min. The samples were mounted with Vectashield with DAPI (Vector Laboratories), and then confocal microscopic analysis was performed by Zeiss LSM700 (Zeiss).

Pharmacological approach

The feeding of drugs to adult flies was performed as described previously (87). Newly eclosed flies of GMR > 3Q were maintained at 10 flies/vial on standard laboratory food (0.7% agar, 10% glucose, 4.5% corn powder, and 4% dry yeast) with 0.1% DMSO or 100 µM TUDCA (Wako)/0.1% DMSO. Flies were transferred to fresh vials every 2 days.

Statistical analysis

The Microsoft Excel macro program (Microsoft) and the biostatistics software GraphPad Prism 5 were used for statistical analyses. The Student‘s t-test was performed unpaired, one-sided way to examine the significance of differences.

*P < 0.01; **P < 0.05.

Supplementary Material

Supplementary Material is available at HMG online.

Acknowledgements

We would like to acknowledge Drs. Mel B. Feany (Harvard Medical School, USA) and Takashi Adachi-Yamada (Gakushuin University, Japan) for providing tau^R406W and hid-lacZ stocks, respectively. We further thank Dr. Masato Hasegawa (Tokyo Metropolitan Institute of Medical Science, Japan) for giving us the cDNA of the human full-length form of tau (2 N/4 R) in a pRK172 plasmid from Dr. Michel Goedert (Medical Research Council, UK). We would like to thank Drs. Katsuhiko Yanagisawa and Mitsuo Maruyama for their useful discussions.

Conflict of Interest statement. None declared.

Funding

Ministry of Education and Scientific Research for Priority Areas, Japan grant numbers [15K07092] to Y-M.L., L.T.

Author notes

†

Present address: National Agriculture and Food Research Organization, 3 Chome-1-1 Kannondai, Tsukuba, Ibaraki Prefecture 305-8517, Japan.