ATP Maintenance via Two Types of ATP Regulators Mitigates Pathological Phenotypes in Mouse Models of Parkinson's Disease

Parkinson's disease is assumed to be caused by mitochondrial dysfunction in the affected dopaminergic neurons in the brain. We have recently created small chemicals, KUSs (Kyoto University Substances), which can reduce cellular ATP consumption. By contrast, agonistic ligands of ERRs (estrogen receptor-related receptors) are expected to raise cellular ATP levels via enhancing ATP production. Here, we show that esculetin functions as an ERR agonist, and its addition to culture media enhances glycolysis and mitochondrial respiration, leading to elevated cellular ATP levels. Subsequently, we show the neuroprotective efficacies of KUSs, esculetin, and GSK4716 (an ERRγ agonist) against cell death in Parkinson's disease models. In the surviving neurons, ATP levels and expression levels of α-synuclein and CHOP (an ER stress-mediated cell death executor) were all rectified. We propose that maintenance of ATP levels, by inhibiting ATP consumption or enhancing ATP production, or both, would be a promising therapeutic strategy for Parkinson's disease.


Introduction
The human brain constitutes only 2-3% of the total body weight but monopolizes about 15% of the total blood flow, and it consumes about 20% of the total oxygen inhaled by the lungs. Simply put, the brain needs a lot of energy, namely ATP, for its functions and survival. Indeed, a decrease in ATP production in certain area(s) of the brain or the central nervous system can lead to neuronal cell death in the affected area(s) and thus cause ischemic or degenerative neuronal diseases.
Parkinson's disease is the second-most frequent neurodegenerative disease, with an incidence of approximately one per 1000 individuals. Parkinson's disease is caused by gradual cell death of dopaminergic neurons in the substantia nigra, and therapeutic chemicals or drugs that prevent this cell death are currently not available. In the early 1980s, certain numbers of drug users near San Francisco were observed manifesting Parkinson's disease phenotypes (Langston et al., 1983). The examination clarified that they suffered from accidental intoxication by 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine (MPTP), which was a contaminant in the drug they used. Later, it was found that 1-methyl-4-phenylpyridinium (MPP +), a metabolite of MPTP, but not MPTP itself, is specifically incorporated into dopaminergic neurons in the substantia nigra, where it inhibits mitochondrial complex I, suppresses ATP production, and ultimately kills the dopaminergic neurons in vivo, not only in humans but also in other animals, e.g. mice and rats (Dauer and Przedborski, 2003;Davis et al., 1979;Langston et al., 1983;Ramsay et al., 1991). These results have strongly implicated mitochondrial dysfunction or ATP decrease as a pathological mechanism in Parkinson's disease. Since then, MPTP has been widely used to experimentally create animal models of Parkinson's disease.
In addition to Parkinson's disease, abnormal protein aggregates are also observed in several other neurodegenerative disorders, e.g. Alzheimer's disease, amyotrophic lateral sclerosis (ALS), polyglutamine diseases, etc., suggesting that common mechanisms underlie these disorders (Kakizuka, 1998). During long-term analyses of the molecular bases of neuronal cell death in neurodegenerative disorders (Ikeda et al., 1996;Kakizuka, 1998;Kawaguchi et al., 1994), we found that ATP consumption by valosin-containing protein (VCP) was profoundly involved in neurodegenerative phenotypes (Higashiyama et al., 2002;Manno et al., 2010). It is notable that VCP mutations have been implicated in IBMPFD (inclusion body myopathy with Paget disease of bone and frontotemporal dementia) (Watts et al., 2004) and rare cases of familial ALS (Johnson et al., 2010). In our analyses, all examined mutated VCP proteins had elevated ATPase activities, and the relative increase in activity levels appeared to be correlated with the severity of the clinical phenotypes (Manno et al., 2010). From these results, we hypothesized that specific inhibitors of VCP ATPase activity might ameliorate the disease phenotypes of familial VCP diseases as well as neuronal cell death in other diseases.
Previously, we showed that PGC1β, a member of the PGC1 family, functions as a protein ligand specific for estrogen receptor-related receptors (ERRs) (Kamei et al., 2003). ERRs belong to the nuclear receptor superfamily and mediate transcription for mitochondrial biogenesis or enhancement of energy expenditure Audet-walsh and Giguère, 2015;Dufour et al., 2007;Huss et al., 2004). Indeed, transgenic expression of PGC1β produced phenotypes of high-energy expenditure: PGC1β mice are big eaters but thin (Kamei et al., 2003). These results indicate that natural or synthetic ERR ligands could enhance ATP production.
Recently, we successfully created small chemicals that are able to suppress the ATPase activity of VCP (Ikeda et al., 2014). Under various stress conditions in cultured cells, the chemicals (KUSs) were able to significantly maintain cellular ATP levels, and consequently suppress ER stress and cell death (Ikeda et al., 2014;Nakano et al., 2016). In addition, KUSs showed significant efficacies in preventing retinal neuronal cell death in in vivo mouse models of retinitis pigmentosa, glaucoma, and ischemic retinal disease (Hasegawa et al., 2016b;Hata et al., 2017;Ikeda et al., 2014;Nakano et al., 2016). In this manuscript, we show that two classes of small chemicals, one for limiting ATP consumption, and the other for enhancing ATP production via ERRs, possess strong efficacies in maintaining ATP levels and in protecting neuronal cells from death in both in vitro cell culture and in vivo mouse models of Parkinson's disease.

Experimental Design
Our objectives were to evaluate the efficacy of ATP maintenance via two types of ATP regulators in MPTP-induced and rotenone-induced Parkinson's disease models. We investigated two classes of small chemicals, one for preventing ATP consumption by VCP (KUSs), and the other for enhancing ATP production via ERRs (esculetin). Their efficacies were first determined by assessing the inhibition of neuronal cell death, ATP levels, and ER stress (CHOP expression) in in vitro cell culture models of Parkinson's disease. We also examined their efficacies in in vivo mouse models of Parkinson's disease by monitoring behavior, ATP levels, and histology. Sample numbers and sizes, the composition of replicates, and doses of compounds were based on pilot experiments and prior knowledge of cultured cells or mouse experiments (Hasegawa et al., 2016b;Ikeda et al., 2014;Nakano et al., 2016). Randomization of animals in the study was based on initial assessments of locomotor activity with a rotarod test and body weight to ensure an equal distribution in each group. Group sizes were set to enable statistical analyses in detection of locomotor activity (more than five per group), histological analyses (more than three per group), or measurement of ATP levels in mouse brain (fourteen or fifteen per group). ATP levels were evaluated using two methods (classic luciferase activity-based methods and imaging of ATeam (Imamura et al., 2009;Nakano et al., 2011;Tsuyama et al., 2013). All rules were predefined in advance. All data were included. Cell culture experiments were performed more than three times. The final end point before sacrifice in mouse experiments was when Parkinson's disease symptoms fully appeared. All animal studies containing the final end points before sacrifice were approved by the Animal Care and Use Committee of Kyoto University.

Glucose Consumption
In order to determine the levels of glucose, cell culture supernatants were collected and assayed for glucose levels with a glucose assay kit (Wako, Tokyo, Japan), following the manufacturer's protocol. Glucose concentrations in medium samples taken at the start and end of the experiments were used to determine cellular uptake. Glucose consumption was calculated, and normalized by the cell numbers.

ATP Assay
ATP levels in cell lysates or brain extracts were measured with luciferase chemiluminescence-based ATP assay reagent for cells (Toyo Bnet, Tokyo, Japan), following the manufacturer's protocol. Briefly, cell lysates were mixed with Glo-Lysis buffer (Promega Corporation, Madison, WI, USA) by shaking slightly and letting the mixture sit for 5 min. Brain sample was homogenized by beads in Homogenizing buffer (10 mM HEPES (pH 7.4), 0.25 M sucrose, protease inhibitor, 50% Glo-Lysis buffer) at 4°C. Trichloroacetic acid (final concentration 0.8%) was added to homogenized sample, and the homogenate was centrifuged at 12,000 rpm at 4°C for 15 min. Supernatants were neutralized with 1 M HEPES (pH 8.0). Chemiluminescence of the reaction mixtures was measured with an ARVO multi-label counter (Perkin Elmer, Inc.).

Extracellular Flux Analyses
Extracellular acidification rate (ECAR) and oxygen consumption rate (OCR) were determined in collagen-coated XF96-well plates using an XF96 extracellular flux analyzer (Seahorse Bioscience, Billerica, MA, USA). Briefly, 0.8 × 10 4 PC12 cells were cultured for 24 h in XF96-well plates in DMEM supplemented with 50 ng/ml NGF, 1% FBS, and 1% HS. Subsequent esculetin treatment was for 24 h. On the day of analysis, cells were washed once with assay medium (Seahorse Bioscience) (unbuffered DMEM with initial pH 7.4, based on the manufacturer's protocol, plus 5.6 mM glucose (Nacalai Tesque) and 1 mM Na pyruvate (Nacalai Tesque)), and then incubated with fresh assay medium for 1 h in a 37°C non-CO 2 incubator. ECAR and OCR were monitored using a Seahorse Bioscience XF96 extracellular flux analyzer. ECAR and OCR rates at each time point were averaged from 5 replicate wells (ECAR) and 7 replicate wells (OCR). ECAR and OCR were determined for each well and were normalized by cell number. The mitochondrial complex I inhibitor rotenone (3 μM) (Sigma), the mitochondrial complex III inhibitor antimycin (3 μM) (Sigma), the mitochondrial complex V inhibitor oligomycin (3 μM) (Sigma), the pharmaceutical uncoupler CCCP (1 μM) (Nacalai Tesque), and 2-deoxy-D-glucose (50 mM) (Nacalai Tesque) were used as inhibitors.

Imaging of ATP Concentrations in PC12 Cells
We first established PC12 cells continuously expressing cyt-ATeam (AT1.03 εD12N/εD109G ) or mit-ATeam (Imamura et al., 2009). In imaging, the cells were plated on a collagen-coated glass-bottom 35 mm dish (glass thickness 0.15-0.18 mm, Mat-tek, Ashland, MA, USA), and were cultured in phenol red-free DMEM (Nacalai Tesque). Wide-field observations of the cells were performed on a Nikon Ti-E inverted microscope using a 40× objective (Nikon, Tokyo, Japan; CFI Plan Apo λ 40× dry: NA 0.95) for cyt-ATeam, and 60× objective (Nikon, Tokyo, Japan; CFI Plan Apo λ 60× oil: NA 1.40) for mit-ATeam, and the following filter sets (Semrock, Rochester, NY, USA): for dual emission ratio imaging of cyt-ATeam and mit-ATeam, 438/24 excitation filter -dichroic mirror FF458-483/32 emission filter for CFP and 542/27 emission filter for YFP. Cells were illuminated using a xenon lamp through 12.5 and 25% neutral density filters. Fluorescence emissions from ATeam biosensors were captured with a Zyla4.2 sCMOS camera (ANDOR, Oxford instruments, USA). The exposure times were 200 ms for CFP and YFP images. The microscope system was controlled with NIS-Elements (Nikon). Cells were maintained on the microscope at 37°C with a continuous supply of a mixture of 95% air and 5% carbon dioxide using a stage-top incubator (Tokai Hit, Shizuoka, Japan). Imaging data were analyzed using MetaMorph (Molecular Devices, Sunnyvale, CA, USA). The YFP/CFP emission ratio was calculated by dividing YFP intensity by CFP intensity for each cell.

Transmission Electron Microscope (TEM)
Cultured cells were fixed in 4% paraformaldehyde with 2% glutaraldehyde overnight at 4°C. Sections were exposed to 1% osmium tetroxide in 0.1 M phosphate buffer (pH 7.4), for 60 min at room temperature, then dehydrated in a series of graded ethanol solutions, and embedded in epoxy-resin Luveak812 (Nacalai Tesque) according to the inverted beam capsule procedure and polymerized at 60°C for 3 days. Ultrathin sections were prepared on an ultramicrotome (EM UC6; Leica, Heidelberg, Germany), and stained with uranyl acetate and lead citrate, and then were observed with a Hitachi H-7650 transmission electron microscope (Hitachi, Tokyo, Japan).

Measurement of Mitochondrial Length
For imaging, the cells were plated on a collagen-coated glass-bottom 35 mm dish (glass thickness 0.15-0.18 mm, Mat-tek). For measurement of mitochondrial length, the cells were cultured in phenol red-free DMEM. Mitochondria were labeled with 50 nM of MitoTracker Green FM (MTG; Invitrogen, Carlsbad, CA, USA), and nuclear DNA was labeled with Hoechst33342 (Hoechst; Invitrogen). Wide field observations of the cells were performed on a Nikon Ti-E inverted microscope using a 60× objective (Nikon; CFI Plan Apo λ 60× oil: NA 1.40) and the following filter sets (Semrock): for imaging of MTG, 482/18 excitation filterdichroic mirror FF495-520/35 emission filter; and for imaging of Hoechst, 405/10 excitation filter -dichroic mirror FF495-520/35 emission filter. The exposure times were 200 ms for MTG and 400 ms for Hoechst. The microscope system was controlled with NIS-Elements (Nikon). Cells were illuminated using a xenon lamp through 12.5 and 25% neutral density filters. Fluorescence emissions from cells were captured with a Zyla4.2 sCMOS camera (ANDOR). Cells were maintained on the microscope at 37°C with a continuous supply of a mixture of 95% air and 5% carbon dioxide using a stage-top incubator (Tokai Hit).

Rotarod Test
The rotarod test was performed to measure forelimb and hindlimb motor coordination and balance. Mice were placed on the rotating rod using the following steps: for 1 week before MPTP injection, mice were trained to stay on the rotarod at constant speed (16 rpm) for 240 s. For each trial, the duration until the mice fell off the rotating rod (constant speed, 16 rpm) was recorded. Mice received three trials in one day without doing consecutive trials. Mean latencies and standard deviations were calculated and used for analysis.

Histochemical Analyses
Mouse brains were fixed with 4% paraformaldehyde and 1% glutaraldehyde overnight at 4°C and embedded in paraffin, and coronal sections were prepared with a thickness of 3 μm. The sections were treated with primary antibodies at room temperature overnight, and with secondary antibodies at room temperature for 2 h. In order to enhance immune signals, a Vectastain elite ABC standard kit (Vector Laboratories, Burlingame, CA, USA) was used, following the manufacturer's protocol. The stereomorphological analyses were performed on the substantia nigra of the mouse brain with a BZ-9000 Generation II microscopy and a BZ-X700 fluorescence microscope (KEYENCE), along with its analysis application programs. The primary antibodies used for immunocytochemistry were anti-tyrosine hydroxylase (1:2000, Millipore, AB152) and anti-αsynuclein (1:50, BD Transduction Laboratories, Lexington, KY, 610787). The secondary antibodies were biotinylated anti-rabbit IgG (H+L) (Vector Laboratories) and biotinylated anti-mouse IgG (H+L) (Vector Laboratories). Primary antibodies used for immunofluorescence staining were anti-tyrosine hydroxylase (1:1000, Millipore), anti-CHOP (1:100, Cell Signaling Technology, L63F7), and anti-α-synuclein (1:50, BD Transduction Laboratories). The secondary antibodies were Alexa488conjugated anti-rabbit (1:200, Invitrogen), and Alexa546-conjugated anti-mouse (1:200, Invitrogen). For immunofluorescence, stained brain sections were counterstained using VECTASHIELD Mounting Medium with DAPI (Vector Laboratories).

Imaging of ATP Concentrations in the Brain Sections of Go-ATeam2 Mice
Go-ATeam2 mice express a FRET-based ATP biosensor similar to the one used for cell culture, except the fluorescent protein reporters are OFP and GFP, rather than YFP and CFP, respectively (M.Y. et al., manuscript in preparation). Go-ATeam2 mice were anesthetized with 4% isoflurane inhalation with oxygen and the mouse brains were immediately dissected. Collected brains were immersed in Ringer's solution (119 mM NaCl, 2.5 mM KCl, 1 mM NaH 2 PO 4 ·2H 2 O, 26.2 mM NaHCO 3 , 11 mM D-glucose, 2.5 mM CaCl 2 , 1.3 mM MgSO 4 ) gassed with 95% O 2 and 5% CO 2 as soon as possible. During production of 300 μm brain slices with a vibratome (Leica Microsystems GmbH, Germany), the brain was kept in ice-cold Ringer's solution (gassed with 95% O 2 and 5% CO 2 ). In order to allow the slices to recover from the cutting procedure, slices were incubated at room temperature for approximately 1 h in Ringer's solution (gassed with 95% O 2 and 5% CO 2 ). Observations of brain sections were performed on a Leica M165 FC stereo microscope (Leica) using a 1× objective (Leica, Plan Apo 1.0) and the following DualView2 filter sets (INDEC Biosystems, Santa Clara, CA, USA): for dual emission ratio imaging of Go-ATeam2, 470/40 excitation filter -dichroic mirror 540 DCLP -515/30 for GFP and 575/40 for OFP. Fluorescence emission from Go-ATeam2 was captured using ORCA-Flash4.0 (Hamamatsu, Japan). The exposure times were 4 s for GFP and OFP images. Imaging data of OFP/GFP emission ratios were analyzed using MetaMorph (Molecular Devices, Sunnyvale, CA).

Statistical Analysis
For statistical analyses in experiments using cells (PC12 cells, HEK293A cells) or mice treated with or without KUSs or esculetin, ANOVA with Tukey (if variances were not unequal), Games-Howell (if variances were unequal) post-hoc test, or Dunnett's test were used. Statistical analyses were performed using PASW Statistics version 17.0 (SPSS Inc., Chicago, IL, USA). The level of statistical significance was set at P b 0.05. Specific statistical tests are indicated in figure legends.

Identification of Esculetin as an ERR Agonist, and Its Ability to Enhance ATP Production
In order to search for ERR agonists, we used our previous assay system for nuclear receptor transactivation (Kamei et al., 2003) (Fig. 1a). In this system, ligand binding domains of human ERRs were fused with the Gal4 DNA binding domain, and each of the resulting Gal4-ERR fusion proteins was expressed in HEK293A cells, together with a luciferase reporter plasmid, which contains 4 copies of GalRE (Kamei et al., 2003;Yamamoto et al., 2001), a Gal4-responsive element in the promoter region. We also co-transfected a plasmid expressing β-galactosidase for the normalization of transfection efficiencies. Twenty-four hours after the transfection, measurable levels of both luciferase and β-galactosidase were observed. Using this assay system, we screened a chemical library (approximately 100,000 chemicals) and noted that esculetin (CAS No. 305-01-1) (Fig. 1b), apart from 47 pseudo-positives, was able to specifically transactivate the luciferase reporters through all of the Gal4-ERRs, most strongly through Gal4-ERRγ, followed by Gal4-ERRβ, and weakly by Gal4-ERRα (Fig. 2a). The Gal4-ERRγ-mediated transactivation by esculetin or GSK4716 (CAS No. 101574-65-6) (Fig. 1c), a known ERRγ agonist, was cancelled by the addition of GSK5182 (CAS No. 877387-37-6) ( Fig. 1d), an ERRγ-specific inverse-agonist (Fig. 2a). Consistently, the Gal4-ERRα-or Gal4-ERRβ-mediated transactivation was not cancelled by the addition of GSK5182 (Fig. 2a). These results indicated that esculetin is able to function as an ERR agonist for all three ERRs.
Given that esculetin functions as an ERR agonist, it was expected that esculetin can alter the metabolic state of cells. We therefore examined this possibility. As expected, addition of esculetin dose-dependently enhanced glucose uptake as well as cellular ATP levels (Fig. 2b, c). Consistently, flux analyses revealed that addition of esculetin stimulated both glycolysis and oxygen consumption, also apparently in a dose-dependent manner (Fig. 2d, e, Fig. S1a, b). Note that treatment with CCCP (Carbonyl cyanide m-chlorophenyl hydrazine), a respiratory chain uncoupler, revealed that esculetin treatment raised not only basal respiration but also spare respiratory capacities and maximal respiration (Fig. 2e, Fig. S1).

KUSs and Esculetin Prevent MPP +-induced ATP Depletion, ER Stress, and Cell Death in PC12 Cells
We previously showed that KUSs, e.g. KUS121 (CAS No. 1357164-52-3) (Fig. 1e) and KUS187 (CAS No. 1357164-95-4) (Fig. 1f), were able to prevent ATP depletion, ER stress, and cell death in neuronally differentiated PC12 cells treated with antimycin and oligomycin, which are specific inhibitors of mitochondrial respiratory chain complex III and V, respectively (Ikeda et al., 2014;Nakano et al., 2016). We then examined whether KUS121 and esculetin could show similar effects in PC12 cells treated with several other types of mitochondrial respiratory chain inhibitors. In addition to previously tested antimycin and oligomycin, we used MPP+, rotenone, and metformin, all of which are inhibitors of mitochondrial respiratory chain complex I.
We treated neuronally differentiated PC12 cells with MPP+, rotenone, metformin, antimycin, and oligomycin in the presence or the absence of KUS121 or esculetin for 24 h; cellular ATP levels were then measured by the luciferase-based assay. All of the inhibitors clearly lowered ATP levels, and KUS121 and esculetin nearly equally suppressed this reduction in ATP levels (Fig. 3a). Under these conditions, KUS121 and esculetin inhibited AMPK activation (evidenced by the phosphorylation of AMPK Thr172), as well as CHOP induction, an ER stress marker with cell death-inducing activity (Zinszner et al., 1998) (Fig. 3b). Consistent with these results, KUS121 and esculetin were able to suppress cell death by these mitochondrial respiratory chain inhibitors (Fig. 3c, Fig. S2). Interestingly, KUS121 and esculetin appeared to more strongly prevent cell death induced by MPP+, rotenone, and metformin, which are inhibitors of mitochondrial respiratory chain complex I, than antimycin and oligomycin, which are inhibitors of mitochondrial respiratory chain complex III and V, respectively (Fig. 3c, Fig.  S2).
In order to examine the potential benefits of KUSs and esculetin for Parkinson's disease in more detail, we analyzed the neuroprotective efficacies of these compounds with MPP+ exposure. Neuronally differentiated PC12 cells were treated with 75 μM MPP+ for 28 h, and then live cell (or dead cell) numbers were estimated (Fig. 4a, b, Fig. S3). With this treatment, approximately 60% of the cells died. This cell death was not inhibited by KUS11 (CAS No. 146293-60-9) (Fig. 1g), which has no inhibitory activity on VCP ATPase (Ikeda et al., 2014). By contrast, addition of KUS121, KUS187, and esculetin significantly protected the cells from cell death; on average, N80% cells survived after MPP+ treatment (Fig. 4b). For KUS121 and esculetin, we further confirmed that the rescue was dose-dependent with these compounds (Fig. 4c, Fig. S3). Consistent with these results, MPP + treatment induced ATF4 and CHOP mRNAs, and addition of KUS121, KUS187, or esculetin completely inhibited these inductions (Fig. 4d). We confirmed the ATP-maintaining effects, the inhibitory effects on ER stress and on AMPK activation, and the neuroprotective effects of KUSs and esculetin in MPP +-treated fully differentiated PC12 cells (Fig. S4). It is also notable that the effects of KUS121 and esculetin were additive with regard to the prevention of all of the MPP +-induced features-ATP decrease, AMPK activation, CHOP induction (ER stress), and cell death-at doses of 25 μM each (Fig. S5).
We also examined the effects of KUSs and esculetin on cellular calcium dynamics. KUS121, KUS187, or esculetin alone did not affect calcium levels in neuronally differentiated PC12 cells (Fig. S6). Addition of MPP+ alone did not either. However, addition of both MPP+ and esculetin, but not MPP + and KUSs, significantly reduced cellular calcium levels (Fig. S6). These results suggest that calcium levels are not a critical determinant of MPP+-induced cell death.
We have shown that addition of methylpyruvate (membrane-permeable pyruvate, which is converted to ATP in mitochondria) as well as ATP itself has the ability to reduce tunicamycin-induced ER stress and ATP decrease (Ikeda et al., 2014). Consistent with these results, addition of 3 mM or 10 mM methylpyruvate was able to prevent the MPP +-induced ATP decrease, CHOP expression, and cell death (Fig.  5a, c, e, Fig. S7a). Likewise, addition of 1 mM or 3 mM ATP was able to prevent the MPP +-induced ATP decrease, CHOP expression, and cell death (Fig. 5b, d, f, Fig. S7b). Prevention of the MPP+-induced decrease of ATP levels by methylpyruvate or ATP was further evidenced by decreased levels of phosphorylated AMPK (Fig. 5c, d). These results demonstrated that ATP levels are a crucial determinant of ER stress and cell death.
Recently, we have developed several different FRET probes for ATP (ATeams), which allow us to estimate relative ATP levels in live cells (Imamura et al., 2009;Nakano et al., 2011;Tsuyama et al., 2013). In order to examine whether an ATeam probe could detect changes of ATP levels in the cytoplasm of MPP+-treated live PC12 cells, we created PC12 cells stably expressing ATeam1.03 in the cytoplasm (cyt-ATeam) and measured FRET signals. Twenty-four hours after treatment with MPP+, a decrease in the FRET ratio was clearly evident. Additions of KUS121, KUS187, or esculetin significantly prevented the decrease of FRET signals. In accord with these changes in FRET signals (Fig. 5g, h), changes in ATP levels were confirmed by classic luciferase activitybased methods (Fig. 5i, Fig. S8). We also created PC12 cells stably expressing ATeam1.03 in the mitochondria (mit-ATeam). Twenty-four hours after treatment with MPP+, a decrease in the FRET ratio was clearly evident in the mitochondria. Additions of KUS121 and KUS187 significantly mitigated the decrease (Fig. S9). By contrast, esculetin further enhanced the decrease (Fig. S9), indicating that KUSs and esculetin oppositely affect mitochondrial ATP levels. These results supported the notion that the maintenance of cytoplasmic ATP levels is more profoundly coupled with the mitigation of ER stress, and thereby with the therapeutic efficacies, than the maintenance of mitochondrial ATP levels.

Effects of KUSs and Esculetin on MPP+-induced Abnormal Features of Mitochondria in PC12 Cells
Given that MPP+ is an inhibitor of mitochondrial respiratory chain complex I, and it has been reported that ERR activation leads to mitochondrial biogenesis (Audet-walsh and Giguère, 2015;Dufour et al., 2007;Huss et al., 2004), we thus next examined mitochondrial morphology. Electron microscopic analyses revealed that treatment with esculetin, but not KUS121 or KUS187, appeared to enhance mitochondrial biogenesis, which was substantiated by the presence of mitochondria with high-density cristae, as compared with non-treated cells (Fig.  6a). Consistent with this, neither KUS121 nor KUS187 was capable of transactivating any of the ERRs (Fig. S10). Addition of MPP+, however, induced loss of cristae and vacuolation of mitochondria. These abnormal phenotypes were partly, if not completely, prevented by KUSs or esculetin (Fig. 6a). Immunochemical analyses showed that esculetin, but not KUS121 or KUS187, elongated the lengths of mitochondria in neurites. Addition of MPP+ reduced the lengths of the mitochondria, and KUSs or esculetin treatment was able to prevent the reduction (Fig. 6b, c).
We then examined the total mass of mitochondria by measuring intensities of MitoTracker Green FM (Cottet-Rousselle et al., 2011). Consistent with the notion that esculetin promotes mitogenesis (Audet-walsh and Giguère, 2015;Dufour et al., 2007;Huss et al., 2004), esculetin, but not KUS121 or KUS187, increased the total mitochondrial mass, regardless of the presence or absence of MPP + (Fig.  S11a). In addition, esculetin, but not KUS121 or KUS187, could mitigate MPP+-induced ROS production, which was consistent with previous reports (Kim et al., 2008;Ellis, 2013, 2016) (Fig.  S11b). By contrast, KUS121 and KUS187, but not esculetin, could prevent the decrease of MPP+-induced mitochondrial membrane potential (Fig. S11c). Thus, with respect to mitochondrial mass, inhibition of ROS production, and maintenance of mitochondrial membrane potential, KUSs and esculetin produced different effects, demonstrating that these effects are not commonly correlated to the efficacies of KUSs and esculetin.
We also examined the protein levels of mitofusin 2 (Mfn2) and dynamin-related protein 1 (Drp1), which play crucial roles in mitochondrial fusion and fission, respectively (Mishra and Chan, 2016;Westermann, 2010). We found that Mfn2 and Drp1 levels remained unchanged regardless of the treatment-even MPP+ alone did not alter the respective protein levels (Fig. S12a). Addition of MPP+ appeared to induce autophagy or mitophagy, based on the decrease of p62 levels and by the increase of LC3 II. Addition of MPP+ appeared to reduce parkin levels and to increase PINK1 levels; both proteins have been reported to be crucial in mitophagy (Nguyen et al., 2016;Pickrell and Youle, 2015;Youle and Narendra, 2011). KUSs treatment prevented all of these changes; esculetin treatment also prevented most of the changes, except the increase of PINK1 levels (Fig. S12b). These results indicate that the effects on mitochondrial dynamics and mitophagy are not likely to be common mechanisms underlying the efficacies of KUSs and esculetin.

KUSs and Esculetin Prevent MPP+-induced Cell Death in Primary Cultures of Dopaminergic Neurons
PC12 was originally established from a pheochromocytoma of adrenal origin, and neuronally differentiated PC12 cells are estimated to have characteristics of sympathetic neurons (Greene and Tischler, 1976), and possess some characteristics of dopaminergic neurons, as evidenced by their expression of tyrosine hydroxylase and their ability to incorporate MPP+, etc. (Grima et al., 1985;Jin et al., 2007;Kumer and Vrana, 1996). In order to further confirm the neuroprotective efficacies of KUSs and esculetin on dopaminergic neurons, we examined their effects on primary cultures of mesencephalic neurons with or without MPP + treatment. KUS121, KUS187, or esculetin treatment alone did not affect the survival of NeuN (Fox-3)-or tyrosine hydroxylase-positive neurons (Fig. 7a, b). Without KUSs or esculetin, MPP+ specifically killed dopaminergic neurons, as seen by the specific decrease of tyrosine hydroxylase-positive neurons (Fig. 7a, b). It is notable that MPP+ apparently reduced neurite lengths of remaining dopaminergic neurons (Fig. 7c). Addition of KUS121, KUS187, or esculetin significantly protected the dopaminergic neurons from the MPP +-induced cell death and from the MPP+-induced neurite shrinkage (Fig. 7).

Effects of KUS121 and Esculetin on Mouse Models of Parkinson's Disease
In order to evaluate the potential efficacies of KUS121 and esculetin on Parkinson's disease, we first set up a reproducible protocol for MPTPinduced Parkinson's disease in mice. We tested several different protocols and ultimately found that the following protocol could reproducibly create Parkinson's disease phenotypes in mice: 20 mg/kg MPTP together with 250 mg/kg probenecid  were injected intraperitoneally twice per week for 5 months. Probenecid has the ability to inhibit renal excretion of several chemicals, including MPTP, and thereby raise the blood concentrations of such chemicals (Lau et al., 1990;Masereeuw et al., 2000). Hereinafter, we refer to these mice as MPTP-treated mice (MPTP mice).
We then evaluated the efficacies of KUS121 and esculetin on MPTP mice. For the evaluation, we administrated vehicle or 50 mg/kg of KUS121 or esculetin, 4 times per week for 5 months concurrent with MPTP treatment (Fig. 8a). As controls, we also administrated GSK4716 and GSK5182, 4 times per week for the 5 months. As a result, MPTP treatments did not induce any difference in the grip strength test (Fig.  S13a), wire hang test (Fig. S13b), treadmill test (Fig. S13c), or open field test (Fig. S13d-h), as compared with no drug-treated control (NT) mice. We observed neither improvement nor exacerbation by KUS121, esculetin, GSK4716, or GSK5182 with these tests in MPTP mice. Importantly, however, MPTP mice stayed on rotor rods for significantly shorter durations, as compared with NT mice, and administration of KUS121, esculetin, or GSK4716 allowed the MPTP mice to stay on rotor rods for durations that were comparable to those of the NT mice (Fig. 8b).
These results implied that KUS121, esculetin, and GSK4716 had neuroprotective effects on dopaminergic neurons of the substantia nigra in MPTP mice. Histochemical examinations of midbrains from these mice revealed that MPTP treatments significantly reduced the numbers of tyrosine hydroxylase-positive neurons, namely dopaminergic neurons, in the midbrains, especially the substantia nigra pars compacta. Administrations of KUS121, esculetin, or GSK4716 significantly mitigated the loss of the dopaminergic neurons (Fig. 8c, d). Consistent with these results, immunohistochemical analyses of the striatum showed a reduction of tyrosine hydroxylase levels in MPTP mice, and the reduction was prevented by the administration of KUS121, esculetin, or GSK4716, but not GSK5182 (Fig. 8e, f). These results were further confirmed by western blot analyses of midbrain extracts from these mice (Fig. S14).
Very similar results were obtained from the analyses of rotenoneinduced Parkinson's disease mice (Dauer and Przedborski, 2003;Inden et al., 2011) (Fig. S15). One clear difference between the two models was the following: rotenone-induced Parkinson's disease mice, but not MPTP-induced Parkinson's disease mice, manifested less movement in the open field test (Figs. S13e, S15f). Importantly, the deficit was prevented by the addition of KUS121 or esculetin (Fig. S15f).

KUS121 and Esculetin Mitigate MPTP-induced α-Synuclein Accumulation, ATP Depletion, and ER Stress
Next, we examined the link between α-synuclein and ATP decrease. In the dopaminergic neurons of the substantia nigra of no drug-treated control (NT) mice and KUS121-or esculetin-treated mice, we were not able to detect the expression of α-synuclein proteins. However, MPTP treatment dramatically provoked elevated α-synuclein levels, and KUS121 or esculetin treatment almost completely abrogated the accumulation of α-synuclein protein (Fig. 9).
We then measured ATP levels in brain sections of control and MPTP mice. Consistent with the results from PC12 cells and the mouse phenotypes mentioned above, ATP levels in the midbrain extracts were significantly decreased in MPTP mice, as compared with NT mice, and the reduction was prevented by the administration of KUS121 and esculetin (Fig. 10a). These results were confirmed by the FRET signals in midbrain slices of Go-ATeam2 mice (Fig. 10b).
Finally, we analyzed ER stress levels in these mice by immunohistochemical analyses using an anti-CHOP antibody. In NT mice, approximately 10% of dopaminergic neurons (tyrosine hydroxylase-positive cells) in the substantia nigra showed CHOP expression. As expected, in the substantia nigra from MPTP mice, approximately 80% or more of the extant dopaminergic neurons were CHOP-positive, and the administration of KUS121 and esculetin reduced the relative abundance of CHOP-positive cells among dopaminergic neurons to approximately 30% or less (Fig. 10c, d, Fig. S16).

Discussion
In many incurable human diseases, including neurodegenerative diseases (e.g. Alzheimer's disease, Parkinson's disease, Amyotrophic lateral sclerosis (ALS), polyglutamine diseases, etc.), the most common pathology is early cell death in the affected organs, or specific brain areas in the case of neurodegenerative diseases. ATP levels would be one of the most important parameters representing overall health conditions of the individual cell, and an ATP decrease appears as the most common phenotype of dying cells. Nutrients are major sources of ATP, and among them glucose is known to be the most potent ATP-producing nutrient. However, high levels of glucose in serum, by themselves, do not raise or even maintain ATP levels in cells, especially in certain pathological conditions, e.g. in diabetes mellitus and in the Parkinson's disease model shown in this study. It is notable that diabetes mellitus is a risk factor for Alzheimer's disease, indicating that high serum glucose levels do not prevent or might even promote neuronal cell death in Alzheimer's disease (Cheng et al., 2012;Stewart and Liolitsa, 1999).
We have long been searching for chemicals that could prevent such cell death in vivo. For this purpose, we have hypothesized that such chemicals would maintain cellular ATP levels under these pathological conditions. In order to address this hypothesis, we developed KUSs as specific inhibitors of the ATPase activity of VCP, the most abundant soluble ATPase in essentially all types of cells. Indeed, KUSs could prevent ATP decreases provoked by several cell death-inducing insults, e.g. glucose or serum deprivation, mitochondrial respiratory chain inhibition, etc., leading to the suppression of ER stress and cell death in cultured cells (Ikeda et al., 2014;Nakano et al., 2016). Furthermore, KUS administration was able to inhibit in vivo retinal neuronal cell death and mitigate disease phenotypes in several mouse models of retinal diseases, e.g. retinitis pigmentosa, glaucoma, and ischemic retinal disease (Hasegawa et al., 2016b;Hata et al., 2017;Ikeda et al., 2014;Nakano et al., 2016). On the other hand, our previous results indicated that activation of nuclear receptors, specifically ERRs, would lead to increased ATP production in the targeted cells or organs (Kamei et al., 2003). We, thus, searched for synthetic agonistic ligands of ERRs and showed in this study that esculetin could function as an ERR agonist. To the best of our knowledge, this is the first small chemical that functions as an ERR agonist for all three ERRs. As expected, esculetin treatment dose-dependently induced an increase in glucose and oxygen consumption in neuronally differentiated PC12 cells, eventually leading to high ATP levels, up to 180% relative to the non-treated cells (Fig. 2b, c, e,  Fig. S1). We thus call these chemicals, as a whole, "ATP regulators".
Pathological hallmarks of Parkinson's disease are the loss of dopaminergic neurons in the substantia nigra and the presence of Lewy bodies, protein aggregates of α-synuclein, in the extant dopaminergic neurons. To date, 18 genetic loci have been linked to familial human Parkinson's disease, and are named PARK1 to PARK18 (Klein and Westenberger, 2012;Lin and Farrer, 2014). PARK1 is the α-synuclein gene itself. These causative genes have been overexpressed, in the case of dominant inheritance, or knocked out, in the case of recessive inheritance, to produce mouse models for Parkinson's disease. However, no such mice have yet successfully manifested either the loss of dopaminergic neurons or Parkinson's disease-like phenotypes. On the other hand, there are established chemically induced Parkinson's disease rodent models, using MPTP, rotenone, 6-hydroxydopamine, etc. (Dauer and Przedborski, 2003;Lindholm et al., 2016). As contrasted with genetically manipulated models, these models did manifest prominent loss of dopaminergic neurons. Since we are interested in the death of dopaminergic neurons and their protection, we used chemically induced Parkinson's disease models in this study.
Beyond our expectations, it was evident that KUSs and esculetin behave very similarly in preventing ATP decrease, ER stress, and cell death, in cell culture as well as in a mouse model of Parkinson's disease brought on by administration of MPTP. In neuronally differentiated PC12 cells and in primary cultures of dopaminergic neurons, MPP + treatment led to decreased ATP levels, induction of CHOP expression (ER stress), and eventually cell death; yet both KUSs and esculetin suppressed all of these 7,. Moreover, administration of KUS121 or esculetin concurrently with MPTP maintained ATP levels and prevented ER stress in dopaminergic neurons in the substantia nigra, and ultimately suppressed the manifestation of Parkinson's disease phenotypes, as seen with the rotarod test (Figs. 8,  10). We confirmed very similar efficacies of KUS121 and esculetin in protecting dopaminergic neurons in the substantia nigra of a rotenone-induced mouse Parkinson's disease model (Fig. S15). We also observed that administration of GSK4716, an ERRγ-specific agonist, could mitigate the Parkinson's disease phenotypes. On the other hand, with an ERRγ-specific inverse-agonist, GSK5182, Parkinson's disease features were either unchanged or exacerbated (Fig. 8). The most surprising data for us was the behavior of α-synuclein protein; it accumulated in the dopaminergic neurons of the substantia nigra in response to MPTP treatment, and it failed to accumulate during the treatment with KUSs and esculetin (Fig. 9). Related with this, very recently it has been reported that ATP itself has chaperone activities (Patel et al., 2017); ATP may help to prevent misfolding or accumulation of α-synuclein. It has also been reported that α-synuclein acts on mitochondria and stimulates ATP synthesis (Ludtmann et al., 2016). These lines of evidence emphasize a close link between α-synuclein protein levels and ATP levels. It is tempting to speculate that Lewy bodies might be a marker of a prior drop in ATP levels.
In addition to the activities shown in this study, esculetin has recently been shown to have cell-protective effects, and to mitigate MPTP-induced Parkinson's disease phenotypes in mice (Subramaniam and Ellis, 2013). These effects were proposed to be mediated by the reduction of oxidative stress caused by reactive oxygen species or nitrosylation (Kim et al., 2008;Ellis, 2013, 2016). These results further support a therapeutic potential of esculetin for the treatment of Parkinson's disease. Esculetin has also been shown to possess cell death-inducing activities in several cultured cancerous cells (Arora et al., 2016;Jeon et al., 2015;Wang et al., 2015), and thus long-term esculetin treatment, if approved, may have the added benefit of reducing cancer occurrence.
We have shown that KUSs can protect retinal neurons from death in mouse models of retinitis pigmentosa, glaucoma, and ischemic retinal disease, and we also have repeatedly discussed mechanisms that underlie the profound links between ATP decrease, ER stress, and cell death in these eye diseases (Hasegawa et al., 2016b;Hata et al., 2017;Ikeda et al., 2014;Nakano et al., 2016). In this study, we further showed that KUSs and esculetin can protect dopaminergic neurons from MPTP-or rotenone-induced toxicity in the respective mouse models of Parkinson's disease (Figs. 8, 10, Figs. S14, S15, S19). From these results, we propose a model for the pathological features of these disorders, in which ATP decrease or depletion is a common central phenomenon, even though each disease has a unique etiology. An ATP decrease in early stages would diminish the functions of affected cells or organs, and then in later stages it would result in cell death or organ failure, culminating in overt disease phenotypes. Furthermore, this model suggests potential therapeutic strategies commonly applicable to these disorders. Namely, maintenance of ATP levels, by either inhibiting ATP consumption or enhancing ATP production, or both, would slow or stop the disease progression. Functional recovery of the remaining cells might even ameliorate the disease phenotypes.
In conclusion, ATP maintenance by "ATP regulators", i.e., KUSs and esculetin (or ERR agonists), showed significant efficacies for neuroprotection in the MPTP-and rotenone-induced Parkinson's disease model mice. Given that many incurable human disorders, e.g. neurodegenerative diseases, ischemic diseases, etc., also manifest early cell death in the affected organs, ATP maintenance by ATP regulators could also provide a strategy for cell protection in these disorders.

Funding Sources
This research was supported in part by research grants from the Mitsubishi Foundation (28107), the Ministry of Education, Culture, Sports, Science and Technology, Japan (16H05151), and by Solution-Oriented Research for Science and Technology (SORST-H16-3) from the Japan Science and Technology Agency (JST). And this research was supported by the Platform Project for Supporting Drug Discovery and Life Science Research (Platform for Dynamic Approaches to Living Systems) from the Japan Agency for Medical Research and Development (AMED) (16am0101009j0005).

Conflicts of Interest
In relation to this manuscript, Kyoto University and Daito Chemix obtained patents (Japan No. 5828201, US No. 9573887, EU No. 2699771), and T. F., T. S., and A. K. were inventors of the patents. The other authors declare no competing interests.

Author Contributions
M. N. and A. K. conceived and designed the experiments and prepared the manuscript. M. N. conducted and analyzed the majority of both the cell and the animal experiments. N. S. performed RT-qPCR. H. I. helped with imaging of the ATeam stable PC12 cell line. M. Y. helped with imaging of brain sections of Go-ATeam2 mice. N. U. and R. T. helped with the experiments on the Parkinson's disease mouse models. T. F. and T. S. synthesized KUS compounds.
All authors discussed the results and commented on the manuscript.