CHIP modulates APP‐induced autophagy‐dependent pathological symptoms in Drosophila

Abstract Dysregulation of autophagy is associated with the neurodegenerative processes in Alzheimer's disease (AD), yet it remains controversial whether autophagy is a cause or consequence of AD. We have previously expressed the full‐length human APP in Drosophila and established a fly AD model that exhibits multiple AD‐like symptoms. Here we report that depletion of CHIP effectively palliated APP‐induced pathological symptoms, including morphological, behavioral, and cognitive defects. Mechanistically, CHIP is required for APP‐induced autophagy dysfunction, which promotes Aβ production via increased expression of BACE and Psn. Our findings suggest that aberrant autophagy is not only a consequence of abnormal APP activity, but also contributes to dysregulated APP metabolism and subsequent AD pathogenesis.

while many risk factors or genes associated with the late-onset AD also affect APP metabolism and Aβ production (Wen et al., 2013).
Autophagy is an evolutionarily conserved catabolic process leading to clearance of proteins and dysfunctional organelles, and is induced mostly by starvation, oxidative stress, and neuronal toxicity (Levine & Kroemer, 2008). In the brain, autophagic activity is maintained at low levels even with nutrient starvation (Kuma et al., 2004). Yet, overactivated autophagy is observed in AD patients, with the marked accumulation of autophagosomes and the aggregation of autophagic vacuoles in dystrophic neurites of AD brains (Yu et al., 2004). The same phenotype is also observed in AD-mouse models and neuronal cells treated with Aβ peptide (Yang et al., 2011). There is an emerging agreement that autophagy defects likely contribute to the neurodegenerative processes in AD (Zare-Shahabadi, Masliah, Masliah, Johnson, & Rezaei, 2015). However, little is known about the mechanism by which autophagy deficits are regulated in AD.
The carboxyl-terminus of Hsc70-interacting protein (CHIP) is a Ubox type chaperone associated E3 ligase that contains the N-terminal tetratricopeptide repeats domain involved in the protein-protein interaction and the C-terminal U-box domain responsible for catalyzing the transfer of ubiquitin to substrate proteins (Paul & Ghosh, 2014).
In this study, we found that CHIP is indispensable for APPinduced AD-like symptoms in Drosophila, which include wing expansion defect, photoreceptor degeneration, DA neuron loss, locomotor disability, lifespan shortening, choice, and learning deficits.
Mechanistically, CHIP promotes APP-induced aberrant autophagy, which contributes to Aβ production through transcriptional upregulation of BACE1 and Psn, and subsequent AD-like neurodegeneration.

| CHIP depletion palliates APP-induced wing expansion defect
Drosophila wing expansion is governed by the neurohormone bursicon (burs), and dysfunction of bursicon-expressing neurons in the CNS results in wing expansion defect and produces an infant wing phenotype (Peabody et al., 2008). Pan-neural overexpression of APPL, the Drosophila homologue of APP, driven by Appl-Gal4 (Appl > APPL), disrupts neuroendocrine function and generates a similar infantile phenotype (Torroja, Chu, Chu, Kotovsky, & White, 1999), which is recapitulated by expressing human APP (Appl > APP; Figure 1b; Peng et al., 2015). Thus, this pathological function of APP/APPL has been evolutionary conserved from fly to human, and factors that regulate this function are likely conserved as well. To identify additional genes that modulate APP's pathological functions, we have previously performed a genetic screen for dominant modifiers of APP-induced wing expansion defect (Peng et al., 2015). We found that Appl > APP-induced wing defect was effectively suppressed in heterozygous CHIP 1 mutants (28% partial expansion and 8% full expansion, Figure 1c-e). Consistently, APP-induced infantile wing phenotype was suppressed by expressing two independent CHIP RNAi (CHIP-IR-1, 34% partial expansion and 15% full expansion; CHIP-IR-2, 42% partial expansion and 11% full expansion, Figure 1e), but remained unaffected by the expression of Dcr2 serving as a negative control ( Figure 1e). Depletion of CHIP by itself did not produce any discernible wing phenotype (Figure S1a-d). The knockdown efficiencies of the two CHIP-RNAi lines were verified by quantitative reverse transcription-polymerase chain reaction (qRT-PCR; Figure 1f). On the other hand, CHIP transcription in 3rd instar larval brains and adult heads was not significantly altered by APP overexpression ( Figure S2). Since Appl is located on the X chromosome, Appl > APP males are lethal due to the dosage compensation effect, only females are used in this and following experiments, except in the courtship choice and learning assay where fru > APP and elav > APP males were utilized, respectively. Overall, these results suggest that CHIP reduction palliates APP-induced wing expansion defect.

| CHIP depletion alleviates APP-induced locomotor deficit
To verify the physiological benefits of CHIP depletion on APP's pathological functions, we next examined the locomotor performance of adult flies as a surrogate functional assay for APP-induced neurotoxicity (Iijima et al., 2004). For AD is an age-related disease, to overcome the developmental defects and investigate the pathological F I G U R E 1 CHIP depletion palliates APP-induced wing expansion defect and adult-specific locomotor deficits. (a-d) Images showing varying degree of wing expansion phenotypes in adult female flies raised at 25℃ (abbreviation: N.E., no expansion; P.E., partial expansion; F.E., full expansion). Compared with the Appl-Gal4 controls (a), overexpression of APP driven by Appl-Gal4 produces an infant wing phenotype (no expansion, b), which is suppressed partially (c) or fully (d) in heterozygous CHIP 1 mutants, or by RNAi-mediated downregulation of CHIP, but remained unaffected by expressing UAS-Dcr2. (e) Histogram showing the percentage of adult wing phenotypes in different genotypes. The number of female flies tested for each genotype is n > 100. (f) Histogram showing the knockdown efficiencies of two independent UAS-CHIP-RNAi lines measured by qRT-PCR. Data were obtained from three independent biological replicates. (g, h) Histograms and line charts showing longitudinal activity of the indicated genotypes at different time points. (g) Compared with the controls, adult-specific expression of APP has no discernable effect on the climbing ability of 2-day-old flies, but displays an age-dependent decline of climbing ability after day 6. Decrease of CHIP significantly alleviates APP-induced locomotor deficits. (h) Appl-Gal4 controls reach 50% climbing index after day 16. Appl > APP flies reach 50% climbing index after day 8, which is partially restored to days 12-13 by depletion of CHIP, but remained unaffected by expressing UAS-Dcr2. All values are shown as mean ± SEM. *p < .05, ***p < .001, ****p < .0001, ns, not significant. More than 200 female flies were tested per genotype ns functions of APP in aging adults, we took advantage of the temperature dependence of Gal4 activity (Duffy, 2002). To this end, APP expression was restricted throughout development at 17℃ due to the minimal Gal4 activity and was activated specifically in adulthood by shifting to 29℃ after eclosion ( Figure S3). As expected, such flies displayed normal wings and climbing ability that was indistinguishable from that of Appl-Gal4 controls at 2 days after eclosion ( Figure 1g).
However, at day 6, Appl > APP flies showed a dramatically reduced climbing ability (0.83 cm/s), as compared with age-matched controls (1.33 cm/s; Figure 1g). APP-induced locomotor defect was significantly suppressed in heterozygous CHIP 1 mutants (1.13 cm/s), or by expressing two CHIP-IR lines (1.19 and 1.02 cm/s). At days 10 and 14, the control flies displayed a gradual reduction of climbing ability, indicating an age-dependent locomotor decline, which was accelerated by APP expression (Figure 1g). Again, decrease of CHIP was able to suppress APP-induced locomotor deterioration ( Figure 1g).
As a control, decrease of CHIP alone did not alter the climbing ability ( Figure S1e).
To directly evaluate the locomotor decline, we defined a performance index (PI) by comparing the climbing velocity between aged and young (2-day-old) flies: PI = velocity aged /velocity young (Peng et al., 2015). We found that Appl-Gal4 controls displayed a gentle decline with aging ( Figure 1h, reached 50% PI after day 16). APP-expressing flies revealed a more obvious and drastic decline (Figure 1h, reached 50% PI by day 8), which was strikingly suppressed by down-regulation of CHIP (Figure 1h, reached 50% PI between day 12 and day 13).
Together, these observations indicate that decrease of CHIP alleviates APP-induced, age-dependent locomotor deficit of adult flies.

| Down-regulation of CHIP suppresses APPinduced toxicity in eye development
Drosophila eyes have been widely used to express human neurotoxic proteins to approximate neurodegenerative diseases (Burr, Tsou, Tsou, Ristic, & Todi, 2014). Co-expression of human APP and BACE1 in fly compound eyes is able to induce age-dependent neurodegeneration of the photoreceptor cells (Greeve et al., 2004). Since Drosophila also encodes a protein with β-secretase activity, we wonder whether APP expression is sufficient to trigger neurodegeneration in fly eyes. To this end, we mobilized the UAS-APP transgene by the Δ2-3 transposase located at 99B with standard fly genetics and obtained a stronger line located on the F I G U R E 2 Down-regulation of CHIP suppresses APP-induced toxicity in eye development. (a-j) Optical microscopic images showing eye phenotypes in 2-and 20-day-old female flies. GMR-Gal4 controls display a highly ordered ommatidial lattice (a, b). Overexpression of APP driven by GMR-Gal4 causes roughened and depigmented eyes at day 2 (c), which becomes increasingly obvious in 20-day-old females (d). APPinduced eye phenotypes are suppressed by depletion of CHIP (g-j), but not by expression of Dcr2 (e, f). (k-t) Images showing SEM of eyes from 2-and 25day-old females. Compared with the controls (k, l), APP expression induces the loss of interommatidial bristles at day 2 (m), which is exacerbated at day 25 (n). The defect is suppressed by depletion of CHIP (q-t), but remains unaffected by expressing Dcr2 (o, p). Female flies were raised at 25°C and shifted to 29°C after eclosion, n > 15 per genotype

Day 25
Day 20 Day 2 Day 2 2nd chromosome. We found the GMR-Gal4 controls showed a normal, well-organized, and smooth external eye surface on day  Figure S4). Together, these results suggest that CHIP is essential for APP-induced age-dependent photoreceptor degeneration.

| Decrease of CHIP ameliorates APP-induced age-dependent DA neuron loss
For the progressive loss of neurons in the brain is a hallmark of AD (Hardy, 2006), we next sought to investigate the genetic interaction between CHIP and APP in this context. It has been reported that neurotransmitter dopamine (DA, released from DA neurons) modulates movement and cognition, and dopaminergic dysfunction plays a pathogenic role in cognitive decline symptoms of AD (Martorana & Koch, 2014). Consistently, progressive elimination of DA neurons has been observed in fly brains expressing APP (Bolshakova, Zhuk, Zhuk, Rodin, Kislik, & Sarantseva, 2014). To confirm whether F I G U R E 3 Decrease of CHIP ameliorates APP-induced age-dependent DA neuron loss. (a-j) Confocal images of DA neuron clusters in adult posterior brains. 2-and 7-day-old females' brains were dissected. DA neurons are labeled with mCD8GFP (green) driven by DDC-Gal4. Compared with the controls (a, b), overexpression of APP causes a reduction of cell number in most DA neuron clusters of 2-day-old adult brains (c), which becomes more obvious in 7-day-old brains (d). This age-dependent DA neuron loss is ameliorated by depletion of CHIP (g-j), but not by expressing Dcr2 (e, f). (k, l) Statistical analysis showing the number of GFP-positive cells in different DA neuron clusters. Values are shown as mean ± SEM. * < .05, **p < .01, ***p < .001, ****p < .0001, ns, not significant. n = 15 brains per genotype. Scale bars, 50 μm DA neuron cell number (Day 2) P P M 1 / 2 P P L 1 P P M Together, these data suggest that CHIP is necessary for APP-induced age-dependent DA neuron loss.

| Down-regulation of CHIP suppresses APPinduced lifespan shortening
We also monitored the longevity of the above flies, since AD seriously affects neurological integrity resulting in an increased risk of earlier death (Mhatre et al., 2014). The Appl-Gal4 control flies lived up to ~80 days with a median lifespan of 60 days, whereas APP-expressing flies exhibited a drastically reduced longevity with a median lifespan of 8 days, and a maximal lifespan of 18 days (Figure 4a,b).
APP-induced lifespan shortening was effectively suppressed by CHIP depletion, which by itself had no effect on the lifespan (Figure 4a,b).
These results suggest that CHIP is indispensable for APP-induced lifespan shortening.

| Decrease of CHIP restores APP-induced choice and learning defects
The above data indicate that decrease of CHIP salvages a diverse array of morphological and behavioral defects caused by APP overexpression. Mounting evidence suggests, despite its relative simplicity, the fly brain is able to drive sophisticated behaviors like decision-making, learning, and memory. Since cognitive impairment is one of the major clinical characteristics of AD patients, we attempted to determine the role of CHIP in APP-induced cognitive deficits. To this end, we first carried out the male courtship choice assays, in which male flies were provided with both younger and older virgin females simultaneously, and the courtship index (CI) was checked as previously described (Hu, Han, Han, Wang, & Xue, 2014).
To accurately quantify the extent of males' preference for younger or older females in courtship choice assay, we measured the preference index (PI) indicating a relative difference between males' CI toward younger or older females (Hu et al., 2014). Consistent with the previous report (Hu et al., 2014), 3-day-old control males courted vigorously toward the younger females rather than the older ones (Figure 4c), indicating a courtship preference behavior for younger mates (Figure 4d). In contrast, 3-day-old males expressing APP driven by the courtship neuron-specific fruitless-Gal4 (Demir & Dickson, 2005) Figure S7a). To rule out the possibility that APP-induced choice impairment is a result of mobility decline, we checked the total CI and climbing ability of these males, and found no significant difference between control males and APP-expressing males ( Figure S7c,d).
These results indicate that CHIP is involved in APP-induced choice dysfunction.
Since AD is associated with impaired learning and memory, we wonder whether CHIP plays a role in APP-induced learning deficit.
To this end, we performed the courtship conditioning procedure (also termed as courtship suppression assay) wherein a male's courtship behavior is affected by experience with an unreceptive female (Siegel & Hall, 1979). Wild-type naive males court virgin females persistently, displaying all defined courtship steps including orientation, tapping, wing extension, licking, and attempted copulation (Hall, 1994). However, after being paired with an unreceptive mated female for a certain period, a wild-type naive male shows significantly decreased courtship activity, as the mated female displays an altered behavior and pheromone profile to reject the naive male constantly (Burnet, Connolly, Connolly, Kearney, & Cook, 1973).
To evaluate the learning ability, we performed the one-hour courtship suppression training, measured the CI of the initial f r u > A P P the males' learning ability ( Figure S7b). Noticeably, all males showed similar CI initial -courtship index in the initial 10 min of the training phase (Figure 4f), indicating that the normal courtship ability was not affected by APP expression. Overall, these results indicate that down-regulation of CHIP restores APP-induced learning defect.

| Down-regulation of CHIP suppresses APPinduced autophagy
Marked accumulation of autophagosomes and late autophagic vacuoles is found in postmortem brain samples of AD patients (Nixon et al., 2005;Yu et al., 2004), suggesting aberrant autophagy is associated with neurodegeneration (Nixon, 2007;Zare-Shahabadi et al., 2015), yet it remains elusive whether abnormal autophagy is a cause or consequence of degenerative process in AD. To address this issue, On the other hand, CHIP has been reported to promote or inhibit autophagy (Arndt et al., 2010;Ferreira et al., 2013;Guo et al., 2015), yet its role in APP-induced autophagy remains unknown. We found that APP-induced autophagic vacuole accumulation in the larval VNC, wing, or eye disks was significantly impeded by depletion of CHIP, but not expression of Dcr2 (Figure 5c-e,i-k,o-q,u-w and i-iv).
As a positive control, APP-induced autophagy defect was effectively eliminated by knocking down Atg7 (Figure 5f,l,r,x). Taken together, these results demonstrate that CHIP plays an essential role in APPinduced autophagy abnormality in both neuronal and non-neuronal tissues.

| Blocking autophagy suppresses APP-induced morphological and behavioral defects
To investigate whether disordered autophagy contributes to APP-induced pathological symptoms, we blocked autophagy in APP-expressing flies by knocking down Atg7 or Atg12, which encodes an essential regulator of autophagosome assembly and is necessary to promote autophagy initiation. As APP overexpression in Drosophila triggers pathological defects in morphology, behavior, and cognition, we examined three representative phenotypes-wing expansion, locomotor activity, and learning ability, respectively. We found that APP-induced wing expansion defect (Figure 6a), climbing disability (Figure 6b), and learning deficit (Figure 6c) were significantly suppressed by blocking autophagy initiation. As a control, blocking autophagy did not affect the normal courtship behavior of APP-expressing flies (Figure 6d). Thus, aberrant autophagy contributes to APP-induced morphological, behavioral, and cognitive defects in Drosophila.

| CHIP depletion impedes Aβ production and accumulation in APP-expressing brains
Autophagy plays a pivotal role in the clearance of cellular waste, including toxic protein aggregates. However, recent studies reveal that blocking autophagy impedes Aβ plaque load in AD mice (Nilsson et al., 2013). Thus, decrease of CHIP may alleviate APP-induced Atg ( Figure S9). Thus, the above results suggest that CHIP promotes autophagy-mediated Aβ production and accumulation.

| D ISCUSS I ON
In this study, we exploited the powerful strength of Drosophila genetics by expressing human APP to recapitulate a number of AD features, including locomotor dysfunction, eye neurodegeneration, longevity shortening, DA neuron loss, choice, and learning deficits. Although AD is characterized by extracellular Aβ aggregates and intraneuronal accumulations of hyperphosphorylated Tau, autophagy deficits likely precede the appearance of these pathological hallmarks (Nixon, 2007;Nixon et al., 2005). Despite a strong correlation between impaired autophagy and AD, the role of autophagy in AD pathogenesis has been controversial. For example, induction of autophagy by rapamycin promotes Aβ clearance, lowers intracellular Aβ level, reduces plaque load, and improves cognition (Caccamo, Majumder, Majumder, Richardson, Strong, & Oddo, 2010). Consistent with these results, loss of autophagy-initiating Beclin1 increases both intracellular and extracellular Aβ load (Pickford et al., 2008). On the other hand, autophagy deficiency impedes Aβ secretion and reduces extracellular Aβ accumulation and plaque burden in AD mice (Nilsson et al., 2013). Consistently, high brain cholesterol impairs Aβ degradation but stimulates autophagy-dependent Aβ secretion and amyloid deposit formation (Barbero-Camps et al., 2018). In this study, we found that blocking autophagy not only reduces Aβ production and accumulation, but also suppresses APP-induced morphological, behavioral, and cognitive defects. Thus, aberrant autophagy is a consequence of enhanced APP activity and also contributes to APP-induced pathological symptoms by facilitating Aβ production.
CHIP, a well-conserved protein with ∼60% amino acid sequence similarity between human and fruit fly (Paul & Ghosh, 2014), is highly expressed in tissues with high metabolic activity and vibrant protein turnover. CHIP is known to function as a protein quality control or modifier of apoptosis. In addition, recent studies have pointed out an alternative function of CHIP modulation of autophagy (Arndt et al., 2010;Ferreira et al., 2013;Guo et al., 2015). However, contradictory results have been reported about the role of CHIP in autophagy. On one hand, CHIP is reported to initiate chaperone-assisted selective autophagy (CASA) by cooperating with HSP70 and BAG3 to degrade the damaged Z-disk proteins (Arndt et al., 2010). On the other hand, knockdown CHIP induces autophagosome formation by increasing the PTEN protein level and decreasing the AKT/mTOR activity, suggesting a suppressive effect of CHIP on autophagy (Guo et al., 2015).
Additionally, there have been studies focusing on CHIP-mediated clearance of tau in AD. Petrucelli et al. showed that CHIP interacted directly with Hsp70 to induce tau ubiquitination, which may protect against tau aggregation and neurofibrillary degeneration (Petrucelli et al., 2004). However, they also found excess CHIP dramatically increased the accumulation of tau, suggesting that the balance between CHIP and substrate levels may be critical.
In this study, we found that decrease of CHIP strikingly suppressed APP-induced aberrant autophagy, which was indispensable for APP-induced pathological symptoms. As CHIP is known to regulate other biological processes, for example, apoptosis and protein quality control, it is intriguing to test whether these roles of CHIP contribute to APP's pathological functions. We found We are also curious about how autophagy contributes to APPinduced Aβ production, which results from a sequential cleavage of APP by β-and γ-secretases. To this end, we examined the transcriptional level of β-secretase (BACE1) and components of γ-secretase (Psn, Aph-1, nicastrin). Intriguingly, we found blocking autophagy decreased the mRNA levels of BACE1 ( Figure S11a) and Psn ( Figure   S11b), but not that of Aph-1 and Nicastrin ( Figure S11c,d). These data imply that autophagy may regulate APP cleavage and Aβ production via affecting BACE1 and Psn expression.
Finally, since Drosophila encodes an APP-like protein, APPL, which is also subjected to cleavage by the β-and γ-secretases, we wonder whether the cleavage of APPL is regulated by autophagy. To this end, we performed Western blot assay and found that α-CTF, produced from the α-cleavage of APPL, was significantly increased upon blocking autophagy ( Figure S12), which is consistent with our observation that BACE1 and Psn expression depends on autophagy ( Figure S11a,b).

| Statistical analysis
Unless otherwise noted, one-way ANOVA followed by Tukey's multiple comparison test was used to determine the statistical F I G U R E 6 Blocking autophagy impedes APP-induced pathological functions and Aβ production. (a) Histogram showing the percentage of adult wings with morphological defects in different genotype flies (abbreviation: N.E., no expansion; P.E., partial expansion; F.E., full expansion). Overexpression of APP using Appl-Gal4 produces an infant wing phenotype, which is suppressed partially or fully by knocking down Atg7 or Atg12. n > 100 females for each genotype. (b) Histogram showing the climbing abilities of 2-day-old female flies with indicated genotypes. Depletion of Atg7 or Atg12 suppresses APP-induced locomotion defect. n > 200 females for each genotype. (c) Learning index of 3-day-old naive males with indicated genotypes in the courtship suppression assay. Compared with elav-Gal4 controls, overexpression of APP driven by elav-Gal4 displays a learning impairment, which is improved by inhibiting autophagy. (d) Courtship index of matched 3-dayold naive males in the initial 10 min of the training phase in the courtship suppression assay. n > 30 males for each genotype. (e-j) Confocal images of 2-day-old females' brains stained with Aβ-specific antibody 6E10. Compared with the controls (e), flies co-expressing APP with Dcr2 accumulate intensive fluorescence in the central brains (f), which is effectively reduced by knocking down CHIP (g, h), Atg7 (i), or Atg12 (j). (k-p) Confocal images of 7-day-old females' brains stained with thioflavin S to detect the amyloid deposits. Compared with the controls (k), APP expression results in increased amyloid deposits (l), which are effectively reduced by knocking down CHIP (m, n), Atg7 (o), or Atg12 (p). Scale bars, 50 μm significance between multiple genotypes. Specifically, for Figures 1e, 6a, and S10m, Kruskal-Wallis rank sum test followed by Nemenyi multiple comparisons test with Benjamini and Yekutieli's p value adjustment was applied. For Figure 4b, Mantel-Cox log-rank statistical analysis was used to determine the p values of median lifespan between different genotypes. For Figure 4c, Wilcoxon matched-pairs test was used to determine significance between intragroup courtships toward younger females and older ones. All values are shown as mean ± SEM. *p < .05, **p < .01, ***p < .001, ****p < .0001, ns, not significant.
Additional experimental procedures are available in Supporting Information.

CO N FLI C T O F I NTE R E S T
The authors declare no conflict of interest.