Abstract
Phosphoinositides and their metabolizing enzymes are involved in Aβ42 metabolism and Alzheimer's disease pathogenesis. In yeast and mammals, Eighty-five requiring 3 (EFR3), whose Drosophila homolog is Rolling Blackout (RBO), forms a plasma membrane-localized protein complex with phosphatidylinositol-4-kinase Type IIIα (PI4KIIIα) and a scaffold protein to tightly control the level of plasmalemmal phosphatidylinositol-4-phosphate (PI4P). Here, we report that RBO binds to Drosophila PI4KIIIα, and that in an Aβ42-expressing Drosophila model, separate genetic reduction of PI4KIIIα and RBO, or pharmacological inhibition of PI4KIIIα ameliorated synaptic transmission deficit, climbing ability decline, premature death, and reduced neuronal accumulation of Aβ42. Moreover, we found that RBO-PI4KIIIa downregulation increased neuronal Aβ42 release and that PI4P facilitated the assembly or oligomerization of Aβ42 in/on liposomes. These results indicate that RBO-PI4KIIIa downregulation facilitates neuronal Aβ42 release and consequently reduces neuronal Aβ42 accumulation likely via decreasing Aβ42 assembly in/on plasma membrane. This study suggests the RBO-PI4KIIIα complex as a potential therapeutic target and PI4KIIIα inhibitors as drug candidates for Alzheimer's disease treatment.
SIGNIFICANCE STATEMENT Phosphoinositides and their metabolizing enzymes are involved in Aβ42 metabolism and Alzheimer's disease pathogenesis. Here, in an Aβ42-expressing Drosophila model, we discovered and studied the beneficial role of downregulating RBO or its interacting protein PI4KIIIα—a protein that tightly controls the plasmalemmal level of PI4P—against the defects caused by Aβ42 expression. Mechanistically, RBO-PI4KIIIα downregulation reduced neuronal Aβ42 accumulation, and interestingly increased neuronal Aβ42 release. This study suggests the RBO-PI4KIIIα complex as a novel therapeutic target, and PI4KIIIα inhibitors as new drug candidates.
Introduction
Cerebral accumulation of Aβ plays an important role in Alzheimer's disease (AD) pathogenesis (Hardy and Selkoe, 2002). In brains of AD patients, Aβ accumulates both extracellularly and intracellularly. Extracellular Aβ accumulates as diffuse deposits and fibrillary plaques. Although its toxicity has been widely studied, no clinical trial on the removal of extracellular Aβ or inhibition of Aβ production has so far succeeded. In recent years, neuronal accumulation of Aβ has been found to be involved in synaptic deficits, amyloid plaque formation, and cell death (Wirths et al., 2004; LaFerla et al., 2007; Gouras et al., 2010). In the brains of AD patients, oligomeric Aβ, possibly the most deleterious form of Aβ to synaptic and cognitive functions (Lambert et al., 1998; Walsh and Selkoe, 2007; Selkoe, 2008), is found to be associated with lipid membranes but is rarely detectable in the CSF and interstitial fluid (Yang et al., 2013; Hong et al., 2014; Savage et al., 2014), indicative of predominant membrane and intraneuronal localization. Consistently, presenilin-2-containing γ-secretase is found to be mainly localized at late endosomes/lysosomes and generates a prominent pool of intracellular Aβ42; disease-causing mutations in presenilin-2 could further increase the Aβ42 level in this pool. Moreover, a subset of FAD mutants in presenilin-1, whose wild-type form is more broadly distributed in the cell, phenocopies presenilin-2 and shifts their localization to late endosomes/lysosomes. This study clearly demonstrates the intracellular production and accumulation of Aβ42 (Sannerud et al., 2016). Thus, reducing the oligomerization and neuronal accumulation of Aβ is of therapeutic potential.
Phosphoinositides, the well-known second messengers, have been widely reported to be involved in AD pathogenesis (Arancio, 2008; Di Paolo and Kim, 2011). In APP (β amyloid precursor protein) transgenic mice, genetic reduction of synaptojanin-1, which converts PI4,5P into PI4P, suppresses synaptic and behavioral impairments (McIntire et al., 2012; Zhu et al., 2013). In patients, changes in the plasma levels of certain phospholipids have been found to correlate with the conversion from normal aging to either amnestic mild cognitive impairment or early AD (Mapstone et al., 2014); moreover, depletion of phosphoinositol (PI) and PI4,5P, as well as a reciprocal elevation of PIP was observed along with disease progression in the parietal cortex of AD patients (Zhu et al., 2015). Mechanistically, the relationship between certain phosphoinositides and Aβ has been studied. PI and PI4,5P facilitate Aβ assembly in/on lipid membrane, and Aβ42 disrupts PI4,5P metabolism and PI3P kinase signaling (McLaurin and Chakrabartty, 1997; Berman et al., 2008; Chiang et al., 2010). However, whether phosphoinositides and their metabolizing enzymes can affect AD pathogenesis via influencing neuronal accumulation of Aβ has yet to be explored.
Drosophila has been used to model AD by pan-neuronal expression of Aβ42. These flies show intraneuronal Aβ accumulation and behavioral deficits (Iijima et al., 2004; Crowther et al., 2005; Iijima-Ando et al., 2008). By expressing Aβ42 in the Drosophila giant fiber (GF) pathway, we developed another model, in which in addition to neuronal accumulation of Aβ and motor defects, age-dependent synaptic transmission failure is evident and easy to record (Zhao et al., 2010; Huang et al., 2013; Lin et al., 2014; Han et al., 2015; Liu et al., 2015), providing a convenient platform for testing the role of candidate genes in neuronal accumulation of Aβ and associated synaptic deficits, and for performing genetic screening for modifiers of the those neural deficits. Using this model, we found the potential role of RBO and PI4KIIIα in AD pathogenesis. RBO has homologs in yeast and mammals, named as EFR3, which forms a plasma membrane localized protein complex with PI4KIIIα and a scaffold protein to control the levels of plasmalemmal PI4P and PI4,5P, particularly PI4P (Faulkner et al., 1998; Huang et al., 2004; Baird et al., 2008; Hammond et al., 2012; Nakatsu et al., 2012). Here we present the effects of genetic reduction of RBO-PI4KIIIα, or pharmacological inhibition of PI4KIIIα on the defects in our fly model, and the potential mechanism.
Materials and Methods
Animal strains and genetics.
P{UAS-Aβarc}, P{UAS-Aβ42} (a gift from Dr. Crowther, Cambridge University), P{UAS-dTAU} (Mershin et al., 2004), P{UAS-shibirets1}3 (Kitamoto, 2001), and P{UAS-mCD8-GFP} (lab stock) transgenes lines were crossed to [GAL4]A307 or P{GAL4-ninaE.GMR}12 (lab stock) line to generate flies expressing arctic mutant Aβ42, wild-type Aβ42, Drosophila tau, a temperature-sensitive dominant negative form of Dynamin, a RBO protein with mutation in the putative lipase catalytic site, and mCD8-GFP in the GF system, respectively. Other lines used are rbots1 (missense mutation, lab stock), rbo2 (null allele, lab stock), P{RBOS358A} (a transgenic line containing a rbo transgene generated by subcloning the genomic DNA of endogenous rbo gene with PCR-mediated point mutation to express the mutant RBO(S358A) under the control of rbo's own premotor) (Vijayakrishnan et al., 2010), itpr-83Asv35 (nonsense mutation of IP3 receptor, also known as itprsv35), a chromosomal deficiency line covering the PI4KIIIα gene, pi4kdf/+(Df(1)ED6579, P{3′.RS5 + 3.3′}ED6579), and PI4KIIIα nonsense mutants (PI4KIIIαGS27 and PI4KIIIαGJ86, gift from Dr. Schupbach, Princeton University). Before use, all these transgenic and mutant flies were separately backcrossed to an isogenic control line w1118 for at least 5 generations. The fly strains used in this study are listed as follows: wild-type: w1118 (isogenic line); ctrl: [GAL4]A307/+; rbots1/+: [GAL4]A307/rbots1; rbo2/+: [GAL4]A307/rbo2; Aβarc: [GAL4]A307/P{UAS-Aβarc}; Aβarc-rbots1/+: [GAL4]A307,rbots1/P{UAS-Aβarc}; Aβarc-rbo2/+: [GAL4]A307,rbo2/P{UAS-Aβarc}; Aβ42: [GAL4]A307/P{UAS-Aβ42}; P{UAS-Aβ42}/+; Aβ42-rbots1/+: [GAL4]A307,rbots1/P{UAS-Aβ42}; P{UAS-Aβ42}/+; Aβ42-rbo2/+: [GAL4]A307,rbo2/P{UAS-Aβ42}; P{UAS-Aβ42}/+; GMR: P{GAL4-ninaE.GMR}12/+; GMR; Aβarc: P{UAS-Aβarc}/ P{GAL4-ninaE.GMR}12; GMR; Aβarc-rbo2/+: P{UAS-Aβarc}, rbo2/ P{GAL4-ninaE.GMR}12; GMR; Aβarc-rbots1/-: P{GAL4-ninaE.GMR}12rbots1/P{UAS-Aβarc}, rbo2; dTAU: [GAL4]A307/P{UAS-dTAU}; dTAU-rbots1/+: [GAL4]A307,rbots1/P{UAS-dTAU}; dTAU-rbo2/+: [GAL4]A307,rbo2/P{UAS-dTAU}; SHIBIREts1: [GAL4]A307/+;P{UAS-shits1}/+; Aβarc-SHIBIREts1: [GAL4]A307/P{UAS-Aβarc}; P{UAS-shits1}/+; Aβarc-RBOS358A: [GAL4]A307/P{UAS-Aβarc}; P{RBOS358A}/+; Aβarc-itprsv35: [GAL4]A307/P{UAS-Aβarc}; itp-r83Asv35/+; Aβarc-rbots1/−: [GAL4]A307,rbots1/P{UAS-Aβarc},rbo2; pi4kdef/+: Df(1)ED6579, P{3′.RS5 + 3.3′}ED6579/+; [GAL4]A307/+; pi4kGS27/+: PI4KIIIαGS27/+; [GAL4]A307/+; Aβarc-pi4kdef/+: Df(1)ED6579, P{3′.RS5 + 3.3′}ED6579/+; [GAL4]A307/P{UAS-Aβarc}; Aβarc-pi4kGS27/+: PI4KIIIαGS27/+; [GAL4]A307/P{UAS-Aβarc}; Aβarc-pi4kGJ86/+: PI4KIIIαGJ86/+; [GAL4]A307/P{UAS-Aβarc}.
Fly climbing and longevity assays.
We used an automatic system for a highly replicable examination of the fly climbing ability (Liu et al., 2015). Briefly, for each group of flies, 30 flies were equally divided and separately transferred into three testing vials. After being tapped down to the bottom of the testing vials, flies were allowed to climb up along the walls of vertically placed testing vials, and were videotaped. The height of each fly in each vial at the fifth second was measured for evaluating the fly's climbing ability.
For longevity analysis, 200 flies (100 female and 100 male) of each genotype were equally separated into 10 vials containing standard fly food and dry yeast, and cultured at 25°C. Food vials were changed every 3 d, and dead flies were counted at that time. Survival rates were analyzed with the SPSS 11 Kaplan–Meier software (IBM).
Electrophysiology test of synaptic transmission failure in flies.
Evoked excitatory junctional potential (EJP) in the GF system was recorded intracellularly as described previously (Zhao et al., 2010). Briefly, an adult female fly at certain age was mounted ventral side down on a glass slide with tackiwax (Boekel Scientific) under a dissection microscope. Before the recording, a reference electrode was inserted into the abdomen, two stimulating electrodes into the two eyes, and a recording electrode into one dorsal longitudinal muscular cell. Intracellular penetration of the recording electrode into the muscle was monitored by a sudden potential drop of 40–70 mV. Square wave electrical stimulation (0.2 ms pulse width, 100 Hz, 50 pulses) was applied to both eyes at the intensity of 5–20 V (∼150% of the threshold stimulation intensity). The signal of the elicited EJPs, whose success rate reflects the well-being of synaptic function, was recorded and amplified by Axonal clamp 900A (Molecular Devices) and digitized at 10 kHz by Digidata 1440A (Molecular Devices). Data were collected and analyzed with the pClamp software (version 10.0; Molecular Devices). All electrodes were glass electrodes and filled with 0.1 m KCl. The recording environment temperature was 25°C.
Toluidine blue staining of Drosophila eye.
Fly heads, with their mouthparts removed, were fixed in 4% glutaraldehyde in 0.1 m phosphate buffer, pH 7.4, overnight at 4°C, after which the preparations were postfixed in 2% OsO4 in the same buffer for 2 h at 4°C, dehydrated with increasing concentrations of ethyl alcohol, and embedded in Epon 812. Leica UC6 ultramicrotone was used to cut fly eyes transversely into 2-μm-thick sections. Sections were stained with toluidine blue and visualized with NIKON E600FN Neurolucida light microscope. Rhabdomere number per ommatidium was analyzed with Stereo Investigator software (MBF Bioscience, RRID: SCR_002526).
Immunoprecipitation and immunoblot.
A total of 300 fly heads of the designated genotype were collected and homogenized with Tris buffer containing 1% NP40 and protease inhibitors. The lysates were subjected to immunoprecipitation or immunoblot with a custom mouse monoclonal antibody against RBO and a custom rabbit polyclonal antibody against Drosophila PI4KIIIα (anti-RBO was generated in collaboration with Abmart (Shanghai) using the RBO fragment 251a.a.-500a.a. as the epitope; anti-PI4KIIIα was generated in collaboration with Abgent (Suzhou) using the peptide NH2- KRSNRSKRLQYQKDSYC-CONH2 as the epitope).
To compare the binding affinity of wild-type RBO versus temperature-sensitive mutant RBO to PI4KIIIα using immunoprecipitation, the lysates were first immunoprecipitated using anti-RBO. The precipitates of each experiment group were then subjected to immunoblot to determine their relative RBO protein levels via quantifying gray value of RBO's bands. Subsequently, another immunoblot was performed, in which the input of each precipitate was adjusted accordingly to guarantee similar RBO protein levels. Finally, the relative PI4KIIIα protein levels normalized against RBO protein levels in each precipitate were compared. The primary antibody used in the immunoblot of APP was 6E10 (RRID: AB_2565327).
Staining and imaging in flies.
Aβ staining in fly CNS was performed as described previously (Zhao et al., 2010). Briefly, the whole CNS of flies, including the brain and ventral ganglion, was dissected out in cold PBS and fixed with 4% PFA in PBS for ∼45 min. Preparations were washed with PBS for 30 min, treated with formic acid (70% in water) for 45 min to reexpose the epitope, washed repetitively with 5% BSA in PBS solution supplemented with 0.25% Triton, incubated with primary antibody (6E10, 1:100, RRID: AB_2565327) at 4°C for 10–12 h, washed with PBS again, and finally incubated with cy3-conjugated secondary anti-mouse antibody (Jackson ImmunoResearch Laboratories, 1:200, RRID: AB_2340813) at room temperature for 2 h. Images were taken under Nikon A1R-A1 confocal microscope; the genotypes of fly CNS were blind to the imager.
ELISA of Aβ42 in fly CNS.
To analyze Aβ42 level in fly CNS, intact brains of 25 flies per strain were dissected out in cold PBS and placed immediately into 50 μl cold ELISA sample buffer supplemented with cocktail protease inhibitor (Calbiochem). Brains were homogenized thoroughly, incubated at room temperature for 4 h. After centrifuge at 4000 × g and at 4°C for 10 min, the supernant was collected. ELISA for Aβ42 level in fly CNS was performed using the Aβ42 Human ELISA Kit (Invitrogen, catalog #KHB3441) according to the manufacturer's instructions. The sensitivity of the kit is <10 pg/ml. A total of 3 μl of the supernatant was diluted to 60 μl with dilution buffer, and 50 μl of the diluent was used for ELISA.
RT-PCR analysis.
Trizol was used to extract total RNA, and a conventional two-step RT-PCR method was used to quantify the transcription level of genes. The primers used are listed as follows: rbo: 5′-GCGCCGTGCGTCCACTATCT-3′ and 5′-ATGGCACGCGCTCCACACAA-3′; pi4kIIIa: 5′-AAACAGAACCGTCAGGTGTC-3′ and 5′-AGCTCAAGGATCAGATTGCG-3′; Efr3a: 5′-TGGAATCGGGGGAACCAAAG-3′ and 5′-AAGACTTCCAGCACTGTCGG-3′; PI4KA: 5′-TTTCAACACGGTCCTGTCACT-3′ and 5′-ACGCCTTCGAACACCTTCAA 3′; and Aβ42: 5′-ATGGCGAGCAAAGTCTCGATC-3′ and 5′-CGCAATCACCACGCCGCCCAC-3′.
Culture of dissected Aβ-expressing larvae or APP-expressing HEK293T cells.
Cleaned and sterilized wandering third-instar larvae were dissected along the dorsal middle line in Schneider's culture medium (Sigma-Aldrich, CAS #S9895). The tracheal, gut, and fat body were removed with caution. The dissected larvae were washed with medium and transferred into a 2 ml centrifuge tube containing 150 μl Schneider's culture medium supplemented with gentamycin (20 mg/ml) and 1.2 mm Ca2+. The tubes, each containing 5 dissected larvae, were placed in a humid and dark environment at 25°C for 8 h. Then 100 μl culture medium from each tube was used for ELISA quantification of Aβ42.
HEK293T cells stably transfected with human APP were cultured in DMEM (Hyclone) supplemented with 10% FBS (Invitrogen), penicillin-streptomycin, and 100 μg/ml G418. For RNA interference, cells were transfected with plasmid pSUPER. basic encoding shRNA for targeted genes using lipofectamine 2000 (Invitrogen). The targeting sequences of shRNA were 5′-GGTTATTGAAATTCGAACT-3′ and 5′-TGCTCATT AGCAGTAAAGA-3′ for Efr3a and PI4KA, respectively. The knocking down efficacy of the shRNAs was ∼20% for Efr3a and 30% for PI4KA (and data not shown). Experiments were performed 2 d after transfection. For quantification of secreted Aβ42, fresh-changed culture medium was examined for Aβ42 concentration after culturing cells for 12 h using ELISA. ELISA for Aβ42 level in culture mediums was performed using the Aβ42 Human ELISA Kit (Invitrogen, catalog #KHB3441) according to the manufacturer's instructions.
Assay of α-, β-, and γ-secretases' activities.
The enzymatic activities of α-, β-, and γ-secretases were measured as previously described (Wang et al., 2015). Briefly, cells were first homogenized and centrifuged, and secretases in the pellet were resolubilized. The α-secretase activity was assayed in 100 mm sodium acetate, pH 7.0, with 2 μg fluorogenic substrate (Merck, 565767), whereas β-secretase activity assayed in 100 mm sodium acetate, pH 4.5, with 2 μg fluorogenic substrate (Merck, 565758) and γ-secretase activity assayed in 50 mm Tris-HCl, pH 6.8, 2 mm EDTA, and 0.25% CHAPSO (w/v) with 8 μg fluorogenic substrate (Merck, 565764). After incubation at 37°C, reactions were centrifuged at 13,200 rpm for 15 min at 4°C and then placed on ice. Supernatants were transferred to a transparent 96-well plate (Nunclon, Nunc), the fluorescence intensities were measured with an excitation wavelength at 340 nm and an emission wavelength at 490 nm for α-secretase, 350 and 490 nm for β-secretase, and 355 and 440 nm for γ-secretase.
Phenylarsine oxide (PAO) application and toxicity examination.
For PAO application to HEK293 cells, dissected Drosophila larva, living larva, and adult flies, a stock solution of 10 mm PAO was made by resolving PAO powder (Sigma-Aldrich, catalog #637-03-6) in DMSO and then diluted with the corresponding culture medium to the wanted concentrations. Although PAO at different concentrations was used to treat HEK293 cells or Drosophila, the final concentration of DMSO was adjusted to identical level to ensure the experimental results were not influenced by DMSO variation. PAO application to living Drosophila started from embryonic stage; PAO was mixed into the culture medium and stored at 4°C for no more than a week before use; when flies consumed food, they took in PAO in the meantime.
To test the toxicity of PAO in adult flies, we cultured control flies with fly food containing 50, 100, 200, 300, 400, and 600 μm PAO and found that PAO treatment at ≤200 μm neither changed the eclosion ratio of pupa nor altered the climbing ability of control flies. Then 25, 50, 100, and 150 μm PAO were chosen for formal experiments.
To test the toxicity of PAO on dissected Drosophila larvae, 50, 100, 200, 300, 400, and 500 nm PAO were applied to the culture medium overnight. We found that CNS neurons and garland cells in the larvae treated with PAO at 300 nm and above turned white, reflecting damage. Then 50, 100, and 150 nm PAO were chosen for formal experiments.
To test the toxicity of PAO on cultured HEK293T cells, 50, 100, 200, 300, 400, and 500 nm PAO were applied to the culture medium of confluent cells. We found that PAO at ≥250 nm killed most of the cells after 12 h incubation, whereas PAO at ≤150 nm did not show evident toxicity after 12 h treatment, according to MTT tests. Then 25, 50, 100, and 150 nm PAO were used to treat HEK293T cells.
Studying the regulation of Aβ42 aggregation in liposome by PI4P.
Based on the procedures described previously (Tashima et al., 2004), we prepared liposome solutions containing synthetic monomeric Aβ42 and PI, PI4P, or PIP2 at certain concentrations. The lipid composition of the liposomes is similar to that found in the cerebral cortex membranes (CCM-lipid) (Tashima et al., 2004) containing phosphatidylcholine (PC), phosphatidylserine (PS), phosphatidylethanolamine (PE), sphingomyelin 1 (SM1), and cholesterol (Ch), but without monosialoganglioside (GM1), cereoside, and plasmalogen (see Table 2).
To prepare the monomeric Aβ42 we used in this experiment, we disaggregated synthetic Aβ42 by following the protocol described by Zagorski et al. (1999) as follows:
Add distilled trifluoroacetic acid (TFA) into a glass tube containing synthetic Aβ42 powder (GL Biochem) at an ∼1:1 ratio (mg:ml), followed by sonication with a KUDOS unltrasonic instrument (model SK250HP) at room temperature for ∼15 min, during which additional portions (1 ml) of TFA are added until the peptide dissolves completely (complete solubilization is judged by the absence of precipitate after centrifugation).
Remove TFA by N2 gas (after which the peptide coats the glass walls of the tube as a waxy layer), the residule trace amount of TFA is removed by addition of distilled hexafluoro-2-propranol (HFIP), sonication, and removal of TFA with N2 gas. The latter is repeated 3 times.
The residual HFIP is removed by vacuum (<1 mmHg, 2 h). After removal of TFA, a concentrated stock solution (1.0 mg/ml) prepared in HFIP is stable for several months.
To prepare the liposomes used in this experiment, we used the following protocol:
Dissolve each lipid in chloroform at the concentration of 1 mg/ml to make a stock solution for each lipid. PC, PS, PE, SM1, and Ch were purchased from Sigma. PI, PI4P, and PI4,5P were purchased from Echelon Bioscience.
Mix the lipids according to Table 2. For example, to make the control liposome containing only Aβ42, mix 113 μl PC, 68 μl PS, 198 μl PE, 87 μl SM1, 57 μl Ch, and 45 μl Aβ42 together in a glass tube; to make liposomes containing Aβ42 and PI (20 μm in final liposome solution), an additional 17 μl PI is mixed in.
Remove the HFIP and chloroform by lyophilization (<1mPar, 2 h) using FreeZone2.5 (Labconco).
Resuspend the precipitate by adding 1.0 ml buffer (5 mm Tris, 100 mm NaCl, pH 7.0) followed by sonication at room temperature for 30 min.
Stay at 4°C for 48 h.
Immunoblot analysis of the liposome solutions.
Data analysis and statistics.
SPSS software (IBM, RRID: SCR_002865) was used for data analysis. The data were presented as mean ± SEM if not specified elsewhere.
Results
Genetic method identifies RBO as a modifier of Aβ's effects
Adult fruit flies expressing wild-type or arctic-mutant Aβ (Aβ42 and Aβarc flies) in neurons of the GF pathway exhibit neuronal Aβ accumulation, age-dependent increase in synaptic transmission failure, decline of flight and climbing ability, as well as premature death (Zhao et al., 2010; Liu et al., 2015). To identify modifiers of Aβ's effects, heterozygous mutations, transgenes, and chromosome deficiencies (Bloomington deficiency kits) were individually introduced into Aβarc flies. The modifying effect was first examined by the climbing assay; then those can influence climbing ability were further tested for age-dependent synaptic transmission failure. A candidate suppressor of Aβ's effects, for example, was isolated if its loss-of-function mutation can ameliorate the age-dependent decline in climbing ability, and can slow down the age-dependent increase in synaptic transmission failure. Modifiers isolated from deficiencies of the third chromosome were reported previously (Liu et al., 2015). Loss-of-function mutations of rolling blackout (rbots1 and rbo2), which were isolated as suppressors (data not shown), are investigated in this study.
RBO insufficiency suppresses neural deficits in flies expressing Aβ42
The effect of rbo mutations on the defects induced by the expression of arctic mutant Aβ42 were further studied with four groups of flies cultured at 25°C: control (ctrl), rbots1/+ or rbo2/+ heterozygotes (rbo), Aβarc-expressing flies (Aβarc), rbots1/+ or rbo2/+ heterozygotes expressing Aβarc (Aβarc-rbo) (for genotypes of these strains, see the list of fly strains in Materials and Methods).
Synaptic transmission was examined by intracellular recording of EJPs in the dorsal longitudinal muscle fibers under high-frequency brain stimulation (100 Hz, 50 pulses). The success rates of elicited EJPs in all groups were indistinguishable on the 3–7th and 15–17th days after eclosion (Fig. 1A,B). On the 25–27th and 31–35th days, however, success rate in Aβarc flies became lower than that in control flies (Fig. 1A,B), confirming our previous results that Aβarc expression causes age-dependent synaptic transmission impairment (Zhao et al., 2010; Huang et al., 2013). The success rates in Aβarc-rbo flies were higher than those in Aβarc flies, although lower than those in control and rbo flies, which did not differ much from each other (Fig. 1A,B). This result demonstrates that rbo insufficiency ameliorated the age-dependent synaptic failure. The transgenic flies and rbo mutants used for generating the 4 groups of flies had been backcrossed to an isogenic wild-type line for at least 5 generations. Therefore, it is unlikely that the difference in genetic background had contributed to the improved synaptic transmission in Aβarc-rbo flies. Because homozygous null mutation of rbo causes embryonic lethality (Faulkner et al., 1998), its effect on Aβarc-induced synaptic failure could not be examined.
rbo mutations suppress the neural deficits in Aβarc-expressing flies. Representative traces (A) and quantification (B) of elicited EJPs in four groups of flies at different ages. The first two groups of flies did not express Aβarc and were classified as “non-Aβ flies,” whereas the other two groups expressed Aβarc and were classified as “Aβ flies.” Notably, insufficiency of rbo suppressed the age-dependent synaptic failure in Aβ flies. Minimum n = 6 flies per bar, one-way ANOVA using LSD test for day 25–27, and Dunnett's test for day 31–35. C, Insufficiency of rbo ameliorated the age-dependent decline of the climbing ability in Aβarc-expressing flies. Minimum n = 30 flies per bar, one-way ANOVA using LSD test for day 26, and Dunnett's test for day 31. D, Insufficiency of rbo significantly prolonged the lifespan of Aβarc flies. n = 200 flies for each line, log-rank test, compared with Aβarc, p for Aβarc-rbots1/+ was 2.3727E-12, p for Aβarc-rbo2/+ was 1.1833E-30; compared with ctrl, p for Aβarc-rbots1/+ was 2.3727E-12, p for Aβarc-rbo2/+ was 1.1833E-30. *p < 0.05. **p < 0.01. ***p < 0.001.
The motor ability was evaluated with climbing assay. Climbing ability was examined on the 3rd, 16th, 26th, and 31st day after eclosion. On the 3rd and 16th day, the climbing abilities of the 4 groups were similar (Fig. 1C). On the 26th and 31st day, Aβarc-rbo flies climbed significantly higher than Aβarc flies, although not as high as control and rbo flies (Fig. 1C). Lifespan study showed that Aβarc-rbo flies lived longer than Aβarc flies, although shorter than control and rbo flies (Fig. 1D). The same conclusion can be made by comparing the mean lifespan values among the 4 groups (Table 1). These results are consistent with the findings on synaptic transmission. The lifespan of Aβarc flies in this study was longer than that in our previous study (Zhao et al., 2010). The major reason is that the flies in this study were cultured at 25°C, rather than at 29°C. Higher temperature may facilitate Aβarc expression.
Mean lifespan of flies and comparison of meansa
We further investigated the effects of rbo insufficiency on the synaptic transmission, climbing ability, and lifespan of flies expressing wild-type Aβ42 and found even better improvement (Fig. 2A–D). We also investigated the effect of rbo insufficiency on the loss of photoreceptor cells caused by the expression of Aβarc in the retina, and found a beneficial effect too (Fig. 2E,F), showing that the rescue effect of rbo insufficiency is not limited to the GF pathway.
Insufficiency of rbo suppresses neural deficits in Aβ42-expressing flies, and the death of photoreceptor cells caused by the expression of Aβarc in the retina. The representative EJP recordings (A) and quantification of success rates of elicited EJPs (B) in ctrl, rbots1/rbo2, Aβ42,Aβ42-rbots1/+, and Aβ42-rbo2/+ flies at different stages of age. Minimum n = 6 flies per bar, one-way ANOVA using LSD test for day 37–39. C, Insufficiency of rbo ameliorated the age-dependent decline of the climbing ability in Aβ42-expressing flies. n = 30 flies per arm, one-way ANOVA using Dunnett's test for day 37. D, Insufficiency of rbo significantly prolonged the lifespan of Aβ42 flies (log-rank test, p < 0.01), but the lifespans of Aβ42-rbots1/+ and Aβ42-rbo2/+ flies were not different from that of ctrl flies. n = 200 flies per line, compared with Aβ42, p for Aβ42-rbots1/+ was 2.8220E-18, p for Aβ42-rbo2/+ was 3.0199E-28; compared with ctrl, p for Aβ42-rbots1/+ was 0.2173, p for Aβ42-rbo2/+ was 0.0569. E, Representative pictures of transverse sections from the eyes of in GMR, GMR;Aβarc, GMR;Aβarc-rbo2/+, and GMR;Aβarc-rbots1/- flies at the age of 30 d. F, Quantification of survival rhabdomeres per ommatidium section, n > 200 ommatidia from 2 or 3 animals for each strain. One-way ANOVA using LSD test. *p < 0.05. **p < 0.01.
The protection of rbo mutations against the Aβ42 toxicity could not be ascribed to a general effect against neuronal accumulation of toxic proteins becasue rbo insufficiency could not ameliorate the motor defect and premature death of flies overexpressing Drosophila tau (Fig. 3A,B).
The effects of rbo, shibire, and itpr mutations as well as rboS358A expression on the motor defect or premature death induced by dTau or Aβarc. Expression of Drosophila tau in the GF pathway produced aged-dependent decline of climbing ability (A) and premature death (B). rbo insufficiency worsened these deficits. Minimum n = 30 per strain for climbing assay, one-way ANOVA using LSD test; n = 200 per strain for longevity assay, log-rank test; compared with ctrl, p for dTAU is 1.7072E-5; compared with dTAU, p for dTAU-rbots1/+ is 1.1173E-4, p for dTAU-rbo2/+ is 0.5831. Expression of dominant SHIBIREts1 mutation neither improved the climbing ability (C) nor expanded the lifespan (D) of Aβarc flies. Minimum n = 30 per strain for climbing assay, one-way ANOVA using LSD test; n = 200 per strain for longevity assay, log-rank test; compared with ctrl, p for SHIBIREts1 is 0.0004, p for Aβarc is 0.0001; compared with Aβarc, p for Aβarc-SHIBIREts1 is 0.4895. Neither the expression of RBOS358A (E) nor the mutation of itpr (F) expanded the lifespan of Aβarc flies. n = 200 per strain for longevity assay, log-rank test; compared with Aβarc, p for Aβarc-RBOS358A is 0.4996, p for Aβarc-itprsv35 is 0.0470. **p < 0.01.
RBO is enriched in synapses and functions in endocytosis (Huang et al., 2004, 2006; Vijayakrishnan et al., 2009), and the rbots1 hemizygotes exhibit similar defects in synaptic transmission, bulk endocytosis, and motor function as shibirets1 homzygotes at 37°C (restrictive temperature) (Huang et al., 2006; Vijayakrishnan et al., 2009, 2010). Nevertheless, the protective effect of rbo insufficiency in Aβarc-expressing flies could not be ascribed to a general effect of mutations in synaptic proteins or endocytosis-related process because introducing dominant shibirets1 mutation into Aβarc-expressing flies could not attenuate the motor defect and premature death (Fig. 3C,D).
Although RBO might be a putative DAG lipase (Huang et al., 2004), and the activity of DAG lipase was reported to be increased in the hippocampus of AD patients and an animal model (Farooqui et al., 1988, 1991), RBO might not regulate Aβarc toxicity by acting as a lipase because introduction of a genomic DNA transgene rboS358A containing the promotor of rbo gene to express a RBO protein with mutation at the putative catalytic site, which could rescue the lethality caused by rbo depletion (Vijayakrishnan et al., 2010), into Aβarc-expressing flies could not change the premature death (Fig. 3E).
PI4KIIIα insufficiency suppresses the deficits in Aβarc flies
The RBO homologs in yeast and mouse recruit PI4KIIIα and form a complex with it on the plasma membrane to control the plasmalemmal level of PI4P (Baird et al., 2008; Nakatsu et al., 2012). Consistent with this, we detected the coimmunoprecipitation between RBO and Drosophila PI4KIIIα (encoded by PI4KIIIα) (Fig. 4A). In addition, we found that removing one copy of rbo gene (rbo2/+) significantly reduced the levels of both RBO and PI4KIIIα protein in ctrl and Aβarc-expressing flies (Fig. 4B). rbots1 mutation did not significantly reduce PI4KIIIα's level but weakened their interaction (Fig. 4B,C). Notably, neither rbo mutations changed the PI4KIIIα transcription level (Fig. 4D). These data indicate that rbo mutations reduced the amount of PI4KIIIα or membrane-localized PI4KIIIα and presumably reduced the plasmalemmal level of PI4P.
Physical interactions between RBO and PI4KIIIα. A, Representative immunoblot showing the coimmunoprecipitation of RBO and PI4KIIIα. w1118 flies were used. n = 3 replicates. B, Representative immunoblot (left) and normalized quantification of RBO and PI4KIIIα protein levels (middle and right) in rbo heterozygotes and control flies without or with the expression of Aβarc. n = 7 per bar, one-way ANOVA using Dunnett's test. C, Representative immunoblot (left) and normalized quantification (right) of PI4KIIIα protein coimmunoprecipitated with the mutant RBO protein (mRBO, encoded by rbots1), compared with that with the wild-type RBO (wtRBO). n = 4 per bar, paired t test. D, Normalized quantification of the PI4KIIIα mRNA levels in Aβarc, Aβarc-rbots1/+, and Aβarc-rbo2/+ larva and adult flies. Minimum n = 5 per bar, one-way ANOVA using Dunnett's test. *p < 0.05.
To test whether PI4KIIIα plays a similar role in Aβarc expression-induced deficits, we separately introduced a chromosomal deficiency (deletion of a PI4KIIIα-containing DNA segment of a chromosome, pi4kdef/+) and a nonsense mutation of PI4KIIIα (pi4kGS27/+) (Yan et al., 2011) into Aβarc-expressing flies. Indeed, both PI4KIIIα mutations suppressed the Aβarc-induced defects in synaptic transmission, motor function, and lifespan (Fig. 5A,C,D). Consistently, feeding Aβarc flies with PAO, a cell well-known membrane-permeable inhibitor of PI4KIIIα (Balla et al., 2005, 2008; Hammond et al., 2012), also markedly ameliorated these defects in a dose-dependent manner (Fig. 5B,E).
Genetic reduction or pharmaceutical inhibition of PI4KIIIα suppresses neural deficits in Aβarc-expressing flies. The representative EJP recordings in ctrl, pi4kdef/+, pi4kGS27/+, Aβarc, Aβarc-pi4kdef/+, Aβarcpi4kGS27/+ flies of 31–35 d (A) and in Aβarc flies of 31–35 d treated at different doses of PAO (B). The synaptic failure, motor defect, and premature death in Aβarc-expressing flies were attenuated by introducing a chromosome deficiency covering PI4KIIIα (A, C), and a nonsense mutation of PI4KIIIα (A, D), and by feeding with PAO (B, E). For clarity, the lifespan curves of control and Aβarc flies were presented separately in E (bottom). Minimum n = 6 flies per bar for synaptic failure test, one-way ANOVA using Dunnett's test. Minimum n = 30 flies per bar for climbing assay, t test; n = 100 flies per line for longevity assays in C and D. Compared with Aβarc, p for PI4KIIIαdef/+ is 7.5337E-10, p for PI4KIIIαGS27/+ is 5.5708E-22; n = 200 flies per line for longevity assays in E, log-rank test, compared with Aβarc 0 μm, p for Aβarc 50 μm is 3.3300E-8, p for Aβarc 100 μm is 1.9794E-10, p for Aβarc 150 μm is 1.7064E-22. *p < 0.05. **p < 0.01. ***p < 0.001.
The suppression of the neural deficits by downregulating RBO/PI4KIIIα could not attribute to an attenuation of calcium release mediated by the receptor of inositol triphosphate (IP3R) because introducing a nonsense mutation of the single IP3R-encoding gene into Aβarc flies could not attenuate the synaptic failure (data not shown) and premature death (Fig. 3F).
Downregulation of RBO-PI4KIIIα reduces neuronal Aβ accumulation
Previously, we determined that wild-type and mutant Aβ42 expressed in the GF pathway accumulates in neurons in Aβ42-expressing flies in an age-dependent fashion based on the localization of Aβ deposits along the GF pathway (Zhao et al., 2010; Huang et al., 2013; Lin et al., 2014; Han et al., 2015; Liu et al., 2015). Here, we further confirmed Aβ's neuronal accumulation in this model by introducing another transgene uas-mCD8-gfp. The expression of mCD8-GFP, which targets the plasma membrane of the cell body and processes of fly neurons, was driven by the same driver as that of Aβarc, so that the Aβarc-expressing neurons could be labeled with GFP. Confocal imaging revealed that Aβ immunostaining signal in aged flies generally colocalized with neuronal GFP (Fig. 6A; the GFP imaging is not ideal because of formic acid treatment), demonstrating that Aβarc predominantly accumulates in neurons in this fly model.
Downregulation of RBO/PI4KIIIα reduces intraneuronal Aβ accumulation. A, Whole-mount Aβ staining of the ventral ganglion in control (top row) and Aβarc flies (middle row). In both flies, the expression of mCD8-GFP was driven by the same driver [Gal4]A307. Bottom row represents an enlarged view of the region defined by the square in the middle row. Notably, the Aβ immunosignal in the ventral ganglion of Aβarc flies generally colocalized with mCD8-GFP. Representative confocal images of whole-mount Aβ staining of the ventral ganglion (B), and normalized ELISA quantification of CNS Aβ42 level in Aβarc, Aβarc-rbots1/+, Aβarc-rbo2/+, Aβarc-PI4KIIIαdef/+, and Aβarc-PI4KIIIαGS27/+ flies (C). Minimum n = 3 replicates per bar, unpaired t test. D, Normalized ELISA quantification of CNS Aβ levels in Aβarc flies with and without PAO treatment. n = 3 replicates per bar, paired t test. E, Quantification of the Aβarc mRNA level in Aβarc-expressing flies at different ages. Minimum n = 3, paired t test. *p < 0.05. **p < 0.01. ***p < 0.001.
To investigate whether RBO-PI4KIIIα insufficiency affects neuronal Aβarc accumulation, we immunostained Aβ in Aβarc, Aβarc-rbo, and Aβarc-PI4KIIIα flies at the age of 25 d when the neural defects in Aβarc flies became prominent. The Aβ-immunoreactive signals in Aβarc-rbo and Aβarc-PI4KIIIα flies were weaker than that in Aβarc flies (Fig. 6B). Consistently, ELISA quantification revealed a remarkable reduction of Aβ42 in the brains of Aβarc-rbo, Aβarc-PI4KIIIα, and PAO-treated flies (Fig. 6C,D). Thus, RBO/PI4KIIIα insufficiency and PI4KIIIα inhibition reduced neuronal Aβ accumulation in flies expressing Aβarc. Notably, neither PI4KIIIα nor rbo mutations reduced the transcription of Aβarc in Aβarc-expressing flies at different ages (Fig. 6E).
Downregulation of RBO/PI4KIIIα facilitates Aβ42 secretion
Downregulation of RBO-PI4KIIIα presumably reduced the plasmalemmal level of PI4P, and the plasmalemmal level of PI4,5P was reported to inversely correlate with the cellular secretion of Aβ42 (Landman et al., 2006). It is conceivable that downregulation of RBO-PI4KIIIα may facilitate Aβ42 secretion and thereafter help reduce neuronal Aβ42 accumulation. To test whether a change in Aβ42 release contributes to the reduced neuronal accumulation of Aβ in PAO-treated and RBO/PI4KIIIα insufficiency flies, we developed a semi-in vivo analysis method to test Aβ42 release. The Aβ42 expressed in our flies was fused to a secretion signal peptide and could be released into extracellular spaces (Crowther et al., 2005), or further released out of the CNS. Aβarc-expressing third-instar larvae were dissected and cultured in Schneider's culture medium, and the Aβ42 released into the medium was quantified by ELISA. Indeed, PAO treatment increased Aβ42 concentration in the medium in a dose-dependent manner (Fig. 7A), indicating that PI4KIIIα inhibition facilitates the Aβ42 release. Similarly, relatively higher Aβ42 concentrations were found in the medium culturing the dissected Aβarc-rbo and Aβarc-PI4KIIIα larvae (Fig. 7B,C).
Downregulation of RBO/Efr3a-PI4KIIIα facilitates cellular Aβ42 release. Normalized ELISA quantification of Aβ42 concentrations in the media culturing dissected Aβarc-expressing larvae with and without PAO treatment (A), rbo mutation (B), PI4KIIIα mutation (C), and culturing APP-HEK293T cells with PAO (D), A1 (E), and Efr3a RNAi /PI4KA RNAi (F) treatment. A–C, E, Minimum n = 6 replicates per bar. D, F, Minimum n = 4 replicates per bar, t test. G, Normalized activities of α, β, and γ secretases with and without PAO treatment at different concentrations. n = 5 for each bar, one-way ANOVA with Dunnett's test. H, Representative immunoblot showing the APP level in HEK293T cells with and without PAO treatment at different concentrations, repeated at least 3 times. *p < 0.05. **p < 0.01. ***p < 0.001.
We also tested whether this facilitation applies to APP-derived Aβ42 by examining Aβ42 release from cultured HEK293T cells stably overexpressing human APP (APP-HEK293T cells). Indeed, PAO similarly increased Aβ42 concentration in the medium (Fig. 7D). A1, another specific inhibitor of PI4KIIIα (Bojjireddy et al., 2014), which is structurally different from PAO, also facilitated Aβ42 secretion from APP-HEK293T cells (Fig. 7E). Knockdown of Efr3a (encoding a human homolog of RBO) and PI4KA (encoding PI4KIIIα) significantly increased the medium Aβ42 as well, but to a lesser extent (Fig. 7F). Notably, PAO neither changed the enzymatic activities of α-, β-, and γ-secretase (Fig. 7G) nor altered the protein level of APP (Fig. 7H). PAO at the concentration of ≤150 nm unlikely inhibited the activities of tyrosine phosphatase, receptor endocytosis, and glucose transport (Le Cabec and Maridonneau-Parini, 1995). We have tried to investigate the effect of PAO on the intracellular level of Aβ42 in APP-HEK293T cells by ELISA, yet intracellular Aβ42 was undetectable, likely due to a rapid degradation mediated by N-end rule pathway (Brower et al., 2013).
PI4P facilitates Aβ42 assembly or oligomerization in/on liposomes
Previously, PI and PI4,5P were reported to facilitate Aβ42 assembly or aggregation in phospholipid membranes at neutral pH condition (McLaurin and Chakrabartty, 1997; McLaurin et al., 1998). Given the role of RBO/Efr3a-PI4KIIIα in the metabolism of phosphoinositides, especially PI4P, it may affect Aβ assembly or oligomerization in/on the plasma membrane. To test this possibility, we examined the effect of PI, PI4P, and PI4,5P on the oligomerization of monomeric Aβ42 in liposomes with lipid composition similar to that found in the cerebral cortex membranes at neutral pH condition (Table 2). PI, PI4P, and PI4,5P each only constitute a small percentage of total lipids in the liposomes (Table 2). As shown in Figure 8A, PI4P in liposomes enhanced the assembly or oligomerization of Aβ42 in a dose-dependent manner. Moreover, PI4P appeared to be the most efficient enhancer compared with PI and PI4,5P (Fig. 8B). This is consistent with a previous finding that among the three forms of the headgroups of phosphoinositides (i.e., the monophosphorylated, diphosphorylated, and triphosphorylated inositol ring, the inositol-1,4-phosphate form appears to be the strongest inducer of β-structure in Aβ42 at neutral pH condition (McLaurin et al., 1998).
Composition of liposome preparations
PI4P facilitates the assembly or oligomerization of Aβ42 in liposomes. A, Representative immunoblot showing the Aβ42 oligomerization in the liposomes containing PI4P at different concentrations, repeated three times. B, Comparison between the effects of PI, PI4P, and PI4,5P on the Aβ42 oligomerization in liposomes, repeated at least three times. A, B, Top and middle rows represent the films exposed to the immune blot for a relatively short and long time, respectively. Bottom row represents the concentrations of PI, PI4P, and PI4,5P in the liposome solutions. Although the concentrations of phosphoinositides were in micromolar, most of them constitute only ∼3% (w/w) of total lipids in the liposomes.
Discussion
Consistent with previous studies (Zhao et al., 2010; Han et al., 2015), this study demonstrates that expression of Aβ42 in the GF pathway of Drosophila leads to age-dependent neuronal Aβ42 accumulation, synaptic failure, and other neural deficits. Genetic reduction of RBO and PI4IIIα, or pharmaceutical inhibition of PI4KIIIα, can reduce neuronal Aβ42 accumulation and suppress the associated phenotypic changes. The yeast and mammal homologs of RBO are membrane-localized proteins that recruit cytosolic PI4KIIIα to form a complex that tightly controls the plasmalemmal level of PI4P (Faulkner et al., 1998; Huang et al., 2004; Baird et al., 2008; Nakatsu et al., 2012). In Drosophila, RBO is also a plasma membrane protein (Faulkner et al., 1998; Huang et al., 2004), which forms a complex with PI4KIIIα and presumably controls the plasmalemmal level of PI4P. PI4P can facilitate Aβ42 assembly or oligomerization in/on liposomes with lipid composition similar to that of human cerebral cortex membranes. Aβ42 oligomers, compared with Aβ42 monomers, have a much higher affinity to plasma membrane and are much more difficult to leave the cells (Narayan et al., 2013; Sarkar et al., 2013). Therefore, downregulation of RBO/Efr3-PI4KIIIα presumably decreased the level of PI4P and oligomeric Aβ42 in/on the plasma membrane, hence allowed more Aβ42 release into the extracellular space. Consistently, genetic downregulation of RBO-PI4KIIIα and chemical inhibition of PI4KIIIα's enzymatic activity facilitated cellular Aβ42 release.
Downregulation of RBO-PI4KIIIα facilitated cellular Aβ42 release, but both immunostaining and ELISA revealed a reduction of Aβ42 level in the CNS of Aβ42-expressing flies. The reason might be that the Aβ42 released from neurons was at least partially further released out of the CNS, as was reflected by the ELISA results of Aβ42 in the medium culturing the dissected Aβ42-expressing larvae.
In the Aβ42-expressing transgenic flies, Aβ42 is released by γ-secretase-independent processing of a secretion signal peptide (Crowther et al., 2005). In contrast, Aβ42 from HEK293-APP cells is produced through sequential cleavage of APP by β- and γ-secretase. Despite all the differences, the Aβ42 release from HEK293-APP cells was also facilitated by inhibiting PI4KIIIα's enzymatic activity and by knocking down PI4KIIIα or a human RBO homolog Efr3a, without changing APP level or APP-processing secretases' activities. These results suggest that plasmalemmal PI4P acts on the APP-derived Aβ42 similarly as it does on the Aβ42 in the transgenic flies. That is to say, HEK293-APP cells may also undergo PI4P-faciliated oligomerization or assembly of Aβ42 on the plasma membrane. There are two origins of plasmalemmal Aβ42: (1) the production and release of a portion of APP-derived Aβ42 occur at the plasma membrane (Haass et al., 2012; Sannerud et al., 2016); and (2) some APP-derived Aβ42 generated intracellularly might also be transferred to the plasma membrane by membrane trafficking (Rajendran et al., 2006; Annunziata et al., 2013). On the plasma membrane, the conformation of all Aβ42, no matter produced locally or secreted by other cells, could be affected similarly by PI4P. Therefore, downregulation of RBO-PI4KIIIa produced similar effect on Aβ42 release from HEK293-APP cells. Nevertheless, we could not exclude other possible ways for downregulation of RBO-PI4KIII to regulate cellular release of Aβ42 in both Aβ42-expressing fly neurons and HEK293-APP cells.
Aβ accumulates both extracellularly and intracellularly in AD brains. For many years, there has been much focus on the prevention and removal of extracellular amyloid deposits in both patients and animal models. However, neither the inhibition of Aβ production by using γ- and β-secretase inhibitors (Doody et al., 2013) (www.alzforum.org) nor the removal of extracellular Aβ by immunotherapy against Aβ (St-Amour et al., 2016) managed to slow down the cognitive decline in AD patients. Given these repetitive failures in efforts targeting extracellular Aβ, both the academic and industrial area of AD research should pay attention to intraneuronal Aβ, even though there is some controversy about intraneuronal Aβ due to technical difficulties and cross reactions in the immunostaining for Aβ in neurons (Gouras et al., 2010).
Our study suggests that inhibiting RBO-PI4KIIIα may serve as an alternative strategy to treat AD via the following ways: (1) inhibiting RBO-PI4KIIIα lowers plasmalemmal PI4P, which can promote the oligomerization of Aβ42 produced either on the plasma membrane or intracellularly, and thus helps reduce oligomeric Aβ42, the Aβ form that is believed to be the culprit in AD pathogenesis (Selkoe and Hardy, 2016); (2) inhibiting RBO-PI4KIIIα facilitates cellular release of Aβ42 and contributes to reducing intracellular Aβ42 accumulation, which can disrupt neuronal and synaptic functions, lead to plaque formation, and even cause cell death; (3) lower PI4P, which is increased in the brains of AD patients, reduces the possibility for extracellular monomeric and oligomeric Aβ42 to undergo oligomerization or further assembly after binding to the plasma membrane; and (4) Aβ-degrading enzymes, such as insulin-degrading enzyme, digest monomeric but not oligomeric Aβ42 (Qiu et al., 1998); therefore, decreased oligomerization of Aβ42 is good for Aβ clearance.
Synthetic Aβ applied to the extracellular space of cultured brain tissue and cells has been reported to produce toxicity (Lambert et al., 1998; Walsh et al., 2002; Lacor et al., 2004). One concern about treating AD via inhibiting RBO-PI4KIIIα is that the facilitation of cellular Aβ42 release may produce similar toxicity as the extracellular application of synthetic Aβ does. However, this concern is unsubstantiated. For one thing, in the above-mentioned studies, the extracellular synthetic Aβ was applied at nonphysiological micromolar level, and the neural toxicity of extracellular synthetic Aβ was ascribed to oligomeric Aβ rather than monomeric Aβ (Walsh et al., 2002) and to the binding of extracellular oligomeric Aβ to the plasma membrane (Lambert et al., 1998; Lacor et al., 2004), but whether there is extracellular oligomeric Aβ42 is not clear in AD patients and animals (Yang et al., 2013; Hong et al., 2014), even though people argue that extracellular oligomeric Aβ42 might be rapidly sequestered by cell membrane (Hong et al., 2014). For another thing, the extracellular Aβ42 is reduced by approximately half in AD and AD incipient patients (Mattsson et al., 2009), and reduction of extracellular Aβ42 may occur decades before the onset of clinical symptoms (Bateman et al., 2012). These studies argue against a causal role of extracellular Aβ42, especially of soluble monomeric Aβ42, in AD pathogenesis.
PAO is a very potent inhibitor of PI4KIIIα. It binds to vicinal thiols and may influence other targets or cellular events, such as the activities of tyrosine phosphatase, receptor endocytosis, and glucose transport, but only at concentrations much higher than those used in the present study (Le Cabec and Maridonneau-Parini, 1995). Importantly, PAO's effect on cellular Aβ42 release via inhibiting PI4KIIIα's enzymatic activity was confirmed by using another structurally different PI4KIIIα-specific inhibitor A1, and by genetic downregulation of PI4KIIIa and Efr3a.
The less robust effect of RBO-PI4KIIIα downregulation on HEK293 cells than fly CNS neurons could be because Aβ42 was mostly produced on the plasma membrane in fly neurons expressing Aβ42, while Aβ42 was produced both on the plasma membrane and intracellular endosomes/lysomes in HEK293 cells, and that the examination of Aβ42 release from fly CNS was a semi-in vivo assay whereas the former was a pure in vitro assay.
In control flies, reduction of RBO or PI4KIIIα level, weakening of the interaction between RBO and PI4KIIIα, or inhibition PI4KIIIα's enzymatic activity by PAO did not significantly change the synaptic transmission, mobility, and lifespan. These results are consistent with the discovery that a significant number of normal humans have only one copy of the PI4KA gene (Conrad et al., 2010). One possible explanation is that insufficiency of RBO/PI4KIIIα may not significantly change the plasmalemmal level of PI4,5P, whose alternation may remarkably influence synaptic transmission and cellular function. This explanation is supported by the reports that complete deletion of PI4KA gene or inhibition of PI4KIIIα with much higher concentration of PAO produces minor or moderate effect on the level of plasmalemma PI4,5P, despite that the plasmalemma PI4P is markedly reduced (Hammond et al., 2012; Nakatsu et al., 2012). Interestingly, in the presence of Aβ42, downregulation of PI4KIIIα may actually attenuate the depletion of plasmalemmal PI4,5P caused by oligomeric Aβ42 (Berman et al., 2008) via reducing PI4P-facilitated formation of oligomeric Aβ42.
Footnotes
F.-D.H. conceived the project, designed, and participated in all experiments except the cell culture studies; X.Z. designed and participated in experiments of flies, cells, WB, co-IP, and liposome; W.-A.W. firstly obtained the quantitative data of fly behavior; L.-X.J. studied the effect of compounds on Aβ42 secretion, secretases' activity, and APP expression; B.-Z.Z for WB and co-IP, B.-Z.Z and Q.-Y.L for purification of genetic background; F.-D.H., X.Z., and N.L. for paper preparation.
This work was supported by National Key Basic Research Program of China Grant 2013CB530900 and National Natural Science Foundation of China Grants 81371400, 81571101, and 81071026. We thank Dr. M. Ho, Dr. J. Li, and Dr. Y. Hu for critical reading; J. Peng, A.Z. Chen, Dr. W.X. Li, and Q.Y. Li for technical support.
F.-D.H. has shareholding of a company possessing part of the intellectual property raised in this study. The remaining authors declare no competing financial interests.
- Correspondence should be addressed to Dr. Fu-De Huang, Shanghai Advanced Research Institute, University of Chinese Academy of Sciences, Chinese Academy of Sciences, Shanghai 201210, China. huangfd{at}sari.ac.cn
References
In this issue
