Research ArticleParkinson’s Disease

Mitochondrial pyruvate carrier regulates autophagy, inflammation, and neurodegeneration in experimental models of Parkinson’s disease

See allHide authors and affiliations

Science Translational Medicine  07 Dec 2016:
Vol. 8, Issue 368, pp. 368ra174
DOI: 10.1126/scitranslmed.aag2210

A mitochondrial target for slowing Parkinson's disease

Currently, there are no disease-modifying treatments to stall progression of Parkinson’s disease (PD). A drug in development to treat diabetes might provide a new way to slow the progression of PD according to new work by Ghosh and colleagues. The drug, MSDC-0160, targets a recently identified carrier of pyruvate (a major substrate for energy production) into the mitochondria. Ghosh et al. now show that this drug, which attenuates the mitochondrial pyruvate carrier, blocks neurodegeneration in several different cellular and animal models of PD. Furthermore, cellular autophagy was restored, and neuroinflammation was reduced in two mouse models of PD. These results support continued investigations into whether the mitochondrial pyruvate carrier will be a useful therapeutic target in PD.

Abstract

Mitochondrial and autophagic dysfunction as well as neuroinflammation are involved in the pathophysiology of Parkinson’s disease (PD). We hypothesized that targeting the mitochondrial pyruvate carrier (MPC), a key controller of cellular metabolism that influences mTOR (mammalian target of rapamycin) activation, might attenuate neurodegeneration of nigral dopaminergic neurons in animal models of PD. To test this, we used MSDC-0160, a compound that specifically targets MPC, to reduce its activity. MSDC-0160 protected against 1-methyl-4-phenylpyridinium (MPP+) insult in murine and cultured human midbrain dopamine neurons and in an α-synuclein–based Caenorhabditis elegans model. In 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine (MPTP)–treated mice, MSDC-0160 improved locomotor behavior, increased survival of nigral dopaminergic neurons, boosted striatal dopamine levels, and reduced neuroinflammation. Long-term targeting of MPC preserved motor function, rescued the nigrostriatal pathway, and reduced neuroinflammation in the slowly progressive Engrailed1 (En1+/−) genetic mouse model of PD. Targeting MPC in multiple models resulted in modulation of mitochondrial function and mTOR signaling, with normalization of autophagy and a reduction in glial cell activation. Our work demonstrates that changes in metabolic signaling resulting from targeting MPC were neuroprotective and anti-inflammatory in several PD models, suggesting that MPC may be a useful therapeutic target in PD.

INTRODUCTION

People with Parkinson’s disease (PD) exhibit a range of nonmotor and motor symptoms (1, 2), with the latter being strongly linked to the degeneration of nigral dopamine neurons. Whereas the disease mechanisms are incompletely understood, evidence suggests that mitochondrial deficits, failure of autophagy, and neuroinflammation each plays a role (35). Disease-modifying therapies are not available for PD, and several trials targeting individual pathways implicated in PD pathogenesis have failed (614). Therefore, a potentially more powerful strategy is the targeting of molecules that are upstream in the signaling cascades that modulate altered cellular functions and that affect both neurons and glia in the brain.

The metabolism of all nutrients flows through various molecular pathways, but ultimately, all are linked to the metabolism of pyruvate (15). Recently, a protein complex for transporting pyruvate into the mitochondria, mitochondrial pyruvate carrier (MPC), was discovered in the internal mitochondrial membrane (1618). MPC contains the proteins MPC-1 and MPC-2, is conserved in yeast, flies, worms, and mammals (1618), and is essential for cellular function (1721).

The MPC complex is the target of a group of compounds known as the thiazolidinedione (TZD) insulin sensitizers (2225). TZDs slow the entry of pyruvate across the mitochondrial membrane (24), and the effects of TZDs on gluconeogenesis require the MPC complex (21). Knockout of the MPC complex in Drosophila eliminated the positive effects of TZD treatment in flies on a high-sucrose diet (22). Slowing the entry of pyruvate at the MPC—for example, by administration of TZDs or genetically—results in a compensatory increase in utilization of other substrates, that is, amino acids and fatty acids, resulting in alterations in cellular metabolism (20, 21, 26, 27).

For the past two decades, the general consensus has been that TZDs activate the transcription factor peroxisome proliferator–activated receptor-γ (PPARγ) and have multiple effects on mitochondrial function, including partial inhibition of complex I [for review, see (28)]. The identification of MPC, however, suggests that some of the antidiabetic effects and other actions of TZDs might be mediated through this protein complex (29). The use of TZDs in diabetes has declined over recent years because of several side effects driven by the activation of nuclear receptor PPARγ, including fluid retention, weight gain, and concern over a potential increased risk for urinary bladder cancer (30) and other cancers (31). Notwithstanding the risk for dose-limiting and sometimes serious side effects, TZDs have recently garnered great attention in PD. A large retrospective study demonstrated that diabetic subjects prescribed with TZDs have a 29% reduced risk of developing PD within 14 years (32), suggesting that these compounds might affect molecular mechanisms involved in PD. Furthermore, TZDs are protective in PD animal models (33, 34). Recent research has shown that the disease processes in diabetes and PD share several features, including metabolic perturbations and inflammation (3538).

MSDC-0160 is a member of a new class of compounds that modulate MPC and act as insulin sensitizers without activating PPARγ (2224, 39). Therefore, MSDC-0160 lacks the negative side effects of the first-generation insulin sensitizers (40). In a phase 2b trial in diabetes, MSDC-0160 effectively lowered glucose and, importantly, caused minimal fluid retention and weight gain (41). Furthermore, MSDC-0160 preserved cerebral 2-deoxyglucose uptake after 3 months of use in Alzheimer’s disease patients, suggesting that it engaged the target MPC in the brain after oral administration (42). Given this background, we hypothesized that modulating MPC function could normalize molecular pathways that are perturbed in PD and reduce neurodegeneration in cellular and animal models of PD.

RESULTS

Protective role of MPC modulation in cellular and nematode models

Previous studies have shown that micromolar concentrations of MSDC-0160 acutely modulate the metabolism of pyruvate in multiple cells types, including neurons, through a direct interaction with MPC (24). We hypothesized that this modulation of MPC function by MSDC-0160 would protect compromised dopaminergic neurons in models of PD both in vitro and in vivo. We used several model systems to evaluate the potential pathways involved.

We found that modulation of MPC shielded cultured human, murine, and invertebrate dopaminergic neurons from 1-methyl-4-phenylpyridinium (MPP+) toxicity (Fig. 1, B, G, and K, and fig. S1, A and B). Specifically, MSDC-0160 (10 μM) pretreatment (1 hour) prevented the MPP+ (10 μM)–induced loss of both tyrosine hydroxylase (TH)–immunoreactive differentiated Lund human mesencephalic (LUHMES) cells (Fig. 1, A and B, and fig. S1A) (43) and TH-immunoreactive mouse primary mesencephalic neurons in culture (Fig. 1, F and G) (44). We determined that about 11% of the neurons in differentiated LUHMES cultures and 6% of the neurons in the primary murine mesencephalic culture expressed TH. Further, we found that MSDC-0160 protected only TH-immunoreactive neurons (no change in the number of TH-negative cells was observed), which is consistent with the selected concentration of MPP+ primarily being toxic to dopamine neurons. In addition, MSDC-0160 counteracted both MPP+-induced shortening of neurite length and reduced branching in both LUHMES cells (Fig. 1, C to E) and primary dopaminergic cultures of mouse midbrain tissue (Fig. 1, H to J). To observe the effects of MPC modulation in a whole organism, we exclusively treated Caenorhabditis elegans nematodes expressing green fluorescent protein (GFP) in dopaminergic neurons with 0.75 mM MPP+ and quantified the resulting neuron loss according to established protocols (45, 46). Treatment with MSDC-0160 (10 or 100 μM) prevented the loss of GFP-fluorescent dopaminergic neurons induced by MPP+ (0.75 mM) in nematodes (Fig. 1K and fig. S1B) (P = 0.0001), whereas 1 μM MSDC-0160 did not. These results indicate that dopaminergic neurons were protected by MSDC-0160 treatment.

Fig. 1. MSDC-0160 protects dopaminergic neurons against MPP+-induced toxicity in cell cultures and C. elegans.

LUHMES cells were pretreated with 10 μM MSDC-0160 (0160) for 1 hour followed by 10 μM MPP+ treatment for 24 hours. (A) Double-label immunocytochemistry for TH and Tuj1 in LUHMES cells. (B) Number of TH-positive dopaminergic neurons, (C) mean neurite length (μm), (D) neurite branching points in each neuron, and (E) longest neurite length (μm). Primary mesencephalic mouse neurons were pretreated with 10 μM MSDC-0160 for 1 hour followed by 10 μM MPP+ treatment for 24 hours. (F) Double-label immunocytochemistry for TH and Tuj1 in primary mesencephalic culture. (G) Number of TH-positive dopaminergic neurons, (H) mean neurite length (μm), (I) neurite branching points in each neuron, and (J) longest neurite length (μm). C. elegans were synchronized, and L1 larvae were placed in 96-well plates containing OP50 bacteria in liquid culture (6 mg/ml) and MPP+ (0.75 mM) with MSDC-0160 (1, 10, and 100 μM) for 48 hours. (K) Quantification of percent dopaminergic neuron loss in C. elegans. Scale bars, 50 μm. Data are means ± SEM of three independent experiments. ****P < 0.0001, ***P < 0.001, **P < 0.01, *P < 0.05, analyzed by one-way analysis of variance (ANOVA) with Tukey’s multiple comparison test.

Protective role of MPC modulation in mouse models of PD

To determine whether MPC targeting was neuroprotective in mammals, we examined the effects of MSDC-0160 on motor behavior and neurodegeneration in mouse models of PD. To confirm that MSDC-0160 effectively enters the brain, we pretreated C57BL/6J mice with MSDC-0160 (30 mg/kg per day via oral gavage) and euthanized the mice 2 or 4 hours later. Our measurements revealed MSDC-0160 and its hydroxymetabolite MSDC-0037 both in plasma and in brain tissue (fig. S2, A and B), confirming that orally administered MSDC-0160 was able to gain access to the brain. When expressed as nanogram per gram in the brain and nanogram per milliliter in the plasma, at 2 hours after an oral dose, there was parent drug (278 ng/g) and hydroxymetabolite (21,100 ng/ml) in the brain, and at the same time point, the plasma concentrations were 164 ng/ml for the parent drug and 51,600 ng/ml for the metabolite. This suggested that oral administration was sufficient to provide direct targeting of MPC in brain cells.

We treated a separate cohort of mice with the neurotoxin 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine (MPTP) as a model of PD (47). Some mice were pretreated with MSDC-0160, whereas controls received vehicle. Twenty-four hours later, the mice were subjected to a subacute MPTP regimen, that is, five daily injections of MPTP (25 mg/kg per day intraperitoneally) (48, 49). During this time, mice received daily oral gavage of MSDC-0160 or vehicle for a total of 11 days (fig. S4A). Four days after the last MPTP injection, we tested the mice for spontaneous locomotion in an open-field arena and for motor coordination on the rotarod test. Consistent with previous reports (48, 50, 51), MPTP injections caused significant impairment on both tests, with decreases in distance traveled, speed, and mobility time in the open field, as well as time spent on the rotating rod. However, MSDC-0160 pretreatment attenuated or prevented these deficits (Fig. 2, A and B, and fig. S4B). The brains from treated mice were subjected to immunohistochemistry, stereological counting of neurons, and high-performance liquid chromatography (HPLC) and Western blot analyses. As expected (47, 48), MPTP treatment induced a loss of dopaminergic neurons and terminals in the substantia nigra and striatum, respectively (Fig. 2, C to E), a reduction of cresyl violet–positive nigral neurons (fig. S4C), depletion of striatal dopamine and its metabolite 3,4-dihydroxyphenylacetic acid (DOPAC) (Fig. 2, F and G), and a loss of TH expression in the nigrostriatal pathway (Fig. 2, H to J). These toxic effects of MPTP were attenuated or blocked by modulation of MPC through pretreatment with MSDC-0160 (Fig. 2, C to J).

Fig. 2. MSDC-0160 improves motor behavior, protects nigrostriatal neurons, and suppresses disease progression in the MPTP mouse model of PD.

(A) Open-field parameters: distance traveled (cm), mean speed (cm/s), and time spent mobile (s). (B) Time spent on rotarod (s). (C) Diaminobenzidine immunohistochemistry for TH in substantia nigra (SN; upper panel) and striatum (ST; lower panel) of MPTP mice. (D) Stereological counting of TH-positive neurons from the SN. (E) Relative density of TH-positive neuronal fibers in the ST. Quantification of (F) dopamine (ng/mg) and (G) the dopamine metabolite DOPAC (ng/mg) by HPLC in the ST. (H) Representative Western blots illustrating the expression of TH in SN and ST. Bar graph showing mean Western blot TH/β-actin in SN (I) and in ST (J). For the disease progression studies, mice were administered MSDC-0160 (30 mg/kg per day) starting 3 days after MPTP treatment, with continuing treatment for another 7 days. (K) Stereological counting of TH-positive neurons from SN. (L) Relative density of TH-positive neuronal fibers in ST. Quantification of striatal (M) dopamine and (N) DOPAC by HPLC. Scale bars, 50 μm (SN) and 200 μm (ST). Data are means ± SEM of 7 to 10 mice per group. ****P < 0.0001, ***P < 0.001, **P < 0.01, *P < 0.05. Data were analyzed by one-way ANOVA with Tukey’s multiple comparison test.

To address the possibility that the neuroprotection induced by MSDC-0160 was due to inhibition of the conversion of MPTP to the neurotoxic metabolite MPP+ by glia, we measured MPP+ concentrations in striatum 3 hours after the fourth MPTP injection, with or without MSDC-0160 treatment. We found that MSDC-0160 did not affect striatal MPP+ concentrations (fig. S4D).

Next, we tested whether modulation of MPC could halt neurodegeneration in the MPTP mouse model after the insult has already been triggered. We found that starting MSDC-0160 administration 2 days after the initial dose of MPTP (in the subacute MPTP regimen) still mitigated motor behavioral deficits (fig. S4G), rescued the loss of dopaminergic neurons and terminals (Fig. 2, K and L), increased the numbers of Nissl-stained neurons (fig. S4F), and restored neurotransmitter concentrations (Fig. 2, M and N). Collectively, these results demonstrate that modulation of MPC both protected against and could rescue MPTP-induced neurodegeneration in the subacute MPTP mouse model of PD.

To validate the effects of modulating MPC in a progressive PD model, we turned to the Engrailed1 heterozygous (En1+/−) mouse. En1+/− mice represent a chronic degeneration genetic model with a PD-like pattern of dopaminergic neurodegeneration that exhibits decreased autophagy and increased neuroinflammation. The En1+/− mice have a normal complement of dopaminergic neurons at birth, but at 8 weeks after birth, the mice exhibit the first signs of a progressive loss of nigral dopaminergic neurons, reaching a plateau at 24 weeks (52). Furthermore, En1+/− mice display motor impairment at 24 to 27 weeks of age in the open-field and rotarod tests. Here, we chose to initiate the modulation of MPC function at two different ages to determine whether the drug would be effective when pathology is mild or modest. Treatment starting at 3 weeks after birth served as a “mild pathology stage,” at which there is axonal pathology but no cell death (Fig. 3A) (53). Treatment starting at 8 weeks after birth served as a “modest pathology stage,” at which 20% of TH-immunoreactive neurons have been lost in En1+/− mice (Fig. 3B).

Fig. 3. MSDC-0160 improves motor behavior in the open-field and rotarod tests in the En1+/− genetic mouse model of PD.

(A) Schedule for MSDC-0160 treatment of En1+/− mice in the mild pathology stage. (B) Schedule for MSDC-0160 treatment of En1+/− mice in the modest pathology stage. (C) Distance traveled (cm), (D) mean speed (cm/s), (E) maximum speed (cm/s), and (F) time on rotarod (s) in the mild pathology stage. WT, wild type. (G) Distance traveled (cm), (H) mean speed (cm/s), (I) maximum speed (cm/s), and (J) time on rotarod (s) in the modest pathology stage. Data are means ± SEM of 8 to 12 mice per group. ***P < 0.001, **P < 0.01, *P < 0.05, oP < 0.05 to 0.1. Data were analyzed using linear mixed-effects regression analysis with false discovery rate (FDR)–corrected P values.

At the mild pathology stage, En1+/− mice were fed a diet formulated to deliver MSDC-0160 (30 mg/kg) starting at week 3 after birth until euthanasia at week 28 or week 48. At the modest pathology stage, En1+/− mice were fed the same diet starting at week 8 until euthanasia at week 16 or week 28. At each time point, we assayed motor behavior, and after euthanasia, we performed immunohistochemistry, Western blotting, and HPLC analysis to examine dopaminergic markers and neurotransmitter concentrations. For motor behaviors, including spontaneous activity in the open-field and coordination testing on the rotarod, the MSDC-0160–treated group exhibited improvements on all measures (Fig. 3, C to G, fig. S5, and table S1).

For all outcomes, the En1+/− non–MSDC-0160–treated control group of mice performed significantly worse than did the wild-type control mice (see the En1 genotype variable in table S1 for adjusted and unadjusted P values). Treatment with MSDC-0160 did not significantly affect the behavior of wild-type mice. However, the En1+/− group treated with MSDC-0160 showed evidence of improvement (after false discovery multiple testing adjustments) in mean maze speed for the groups that started on MSDC-0160 at 3 and 8 weeks of age (3 weeks, P = 0.032; 8 weeks, P = 0.027), mean distance traveled (group that started on treatment at 8 weeks of age; P = 0.021), and time spent mobile in the maze (group that started on treatment at 8 weeks of age; P = 0.003). There was little evidence that treatment of En1+/− mice with MSDC-0160 reduced the amount of time spent frozen in either of the groups that started MSDC-0160 treatment at 3 or 8 weeks of age. All other behavioral measures for the En1+/− groups treated with MSDC-0160 exhibited significant improvement (P < 0.05) before the multiple testing adjustments, but these apparent improvements did not remain significant after adjustment (see the En1 genotype * MSDC treatment variable in table S1 for adjusted and unadjusted P values).

Consistent with the results of the behavioral testing, stereological analyses of immunohistochemically stained sections of the brains of the En1+/− mice revealed that treatment with this MPC modulator protected the TH-immunoreactive (Fig. 4, A, B, and G to I) and Nissl-stained (fig. S6, A to C) dopaminergic neurons of the substantia nigra. In addition, MPC modulation with MSDC-0160 increased the expression of TH and restored striatal dopamine and DOPAC content in all groups at each studied time point in the En1+/− mice [Fig. 4, C and D (P = 0.0018), E (P = 0.0018), F (P = 0.0168), J to L (P = 0.0001), M (P = 0.0001), N (P = 0.0014), O (P = 0.0001), and P (P = 0.0009), and fig. S6, E (P = 0.0002) and F (P = 0.0001)]. Collectively, these data demonstrate that modulation of MPC was neuroprotective in the slowly progressing En1+/− mouse model of PD.

Fig. 4. MSDC-0160 prevents dopaminergic neurodegeneration in the En1+/− genetic mouse model of PD.

(A) TH immunostaining in the SN. (B) Stereological counting of TH-immunoreactive (TH+) neurons in the SN. (C) Representative Western blot illustrating the expression of TH in the SN. (D) Bar graph showing mean Western blot TH/β-actin ratios in SN, quantification of striatal (E) dopamine (ng/mg), and (F) DOPAC (ng/mg) by HPLC in the mild pathology stage. (G) TH immunostaining in SN (upper panel pictures are from 16 weeks of age, and lower panel pictures are from 28 weeks of age). Scale bars, 50 μm. (H) Stereological counting of TH-positive neurons in SN from 16 weeks of age. (I) Stereological counting of TH-positive neurons in SN from 28 weeks of age. (J) Representative Western blot illustrating the expression of TH in SN from 16 weeks of age (top) and 28 weeks of age (bottom), and bar graphs showing mean Western blot TH/β-actin ratios in SN at 16 weeks (K) and at 28 weeks of age. (L) Quantification of striatal (M) dopamine and (N) DOPAC by HPLC at 16 weeks of age. Quantification of striatal (O) dopamine and (P) DOPAC by HPLC at 28 weeks of age in the modest pathology stage. Data are means ± SEM of six to eight mice per group. ****P < 0.0001, ***P < 0.001, **P < 0.01, *P < 0.05. Data were analyzed by two-way ANOVA with Bonferroni’s multiple comparison test.

The effects of MPC modulation on degeneration of dopaminergic neurons

MPC, the target of MSDC-0160, is present in nigral dopaminergic neurons (fig. S7, A and B) (17). Thus, we set out to dissect the molecular mechanisms underpinning the neuroprotective effect of MPC modulation using C. elegans and cultured cells, as well as the MPTP and En1+/− mouse models of PD. C. elegans expresses brain protein 44–like (BRP44L), which is an ortholog of MPC-1. In the C. elegans model, we used worms expressing both GFP and A53T mutant α-synuclein in dopaminergic neurons (a point mutation that causes autosomal dominant PD in humans). Aggregated α-synuclein is a major component of Lewy bodies in PD (54), and α-synuclein is genetically linked to both inherited and sporadic forms of PD (55). We visualized dopaminergic neurons using GFP in worms expressing A53T α-synuclein. These worms showed marked degeneration of dopaminergic neurons after 8 days (P = 0.05), which was rescued when BRP44L function was modulated by treatment with 100 μM MSDC-0160 (P = 0.0001) (Fig. 5, A and B).

Fig. 5. MSDC-0160 modulates mTOR signaling in C. elegans and the MPTP mouse model of PD.

(A) Images show dopaminergic neuron loss in worm strains BY250 and JVR203 at 12 days of age. Scale bar, 10 μm. (B) Quantification of dopaminergic neuron loss (percentage of total number of dopamine neurons). (C) Quantification of dopaminergic neuron loss when C. elegans were fed bacteria expressing the empty vector L4440 as control. Quantification of dopaminergic neuron loss in C. elegans after knocking down BRP44L/MPC-1 (D), AKT-1 (E), RHEB-1 (F), LET-363/mTOR (G) and AAK-1/AMPK (H). Mice were coadministered MPTP and MSDC-0160. After 1 day of pretreatment with MSDC-0160 (30 mg/kg, by oral gavage) and 5 days of cotreatment with MSDC-0160 (30 mg/kg, by oral gavage) and MPTP (25 mg/kg per day intraperitoneally), mice were euthanized, and SN tissue was dissected and analyzed by Western blot. (I) Representative Western blot illustrating the expression of p-mTOR (Ser2448), total mTOR, pS6 (Ser235/Ser236), and total S6 in the SN. Bar graphs showing mean Western blot p-mTOR/mTOR ratios relative to β-actin (J) and mean Western blot p-S6/S6 ratios relative to β-actin (K). (A to H) Data are means ± SEM (n = 3 experiments) and were analyzed by unpaired Student’s t test except (B), which was analyzed by two-way ANOVA with Bonferroni’s multiple comparison test. (I and J) Data are means ± SEM of six to eight mice per group and were analyzed by two-way ANOVA with Bonferroni’s multiple comparison test. ***P < 0.001, **P < 0.01, *P < 0.05.

On the basis of the literature, we hypothesized that the mammalian target of rapamycin (mTOR) pathway is involved downstream of MPC (29, 5659). In nematodes, we knocked down genes of interest related to MPC and to the mTOR pathway by RNA interference and evaluated whether the knockdown of these genes affected the protection of A53T α-synuclein–induced dopaminergic neurons from degeneration by MSDC-0160. MSDC-0160 treatment prevented neuron loss due to the A53T mutation in α-synuclein in control worms, which were fed bacteria expressing the empty vector L4440 (Fig. 5C). Knockdown of BRP44L (the ortholog of MPC-1) prevented neuroprotection by MSDC-0160 (Fig. 5D). Furthermore, knockdown of AKT-1 (a serine/threonine kinase that functions upstream of RHEB-1; Fig. 5E), RHEB-1 (a guanosine triphosphatase that is a stimulator of the mTOR pathway; Fig. 5F), or LET-363 (the C. elegans ortholog of the human mTOR protein; Fig. 5G) also prevented the neuroprotection exerted by MSDC-0160 in this dopaminergic neuronal loss assay. By contrast, knockdown of AAK-1, which is a homolog of adenosine monophosphate (AMP)–activated protein kinase, mimicked the control condition (Fig. 5H), suggesting that BRP44L does not control this pathway. RNA interference knockdown of the genes tested above was confirmed by quantitative polymerase chain reaction (fig. S1C). The conclusion of these experiments is that engaging MPC and the mTOR pathway is necessary for the neuroprotective effect of MSDC-0160 in C. elegans.

In a second set of experiments, we used the mitochondria isolated from rat brain and intact cultured LUHMES cells to define the time course of the effects of MSDC-0160 on the different components of the mTOR pathway. Initially, we demonstrated that MSDC-0160 decreased pyruvate oxidation in isolated brain mitochondria in a direct and immediate fashion (fig. S2, C and D). We then measured the effect of MSDC-0160 on cellular respiration in LUHMES neuronal cells treated with the mitochondrial toxin MPP+ and found that when MSDC-0160 was added 1 hour before or after MPP+ addition, MSDC-0160 normalized oxygen consumption (fig. S2E). These data show that modulation of MPC has immediate effects on mitochondrial function. On the other hand, when LUHMES cells were treated for 2, 6, or 24 hours with MSDC-0160, no changes in mTOR activity [monitored as the ratio of phosphorylated mTOR (p-mTOR)/mTOR] could be detected (fig. S3, A and B). These data indicated that the changes in nutrient-sensing pathways were downstream of the metabolic action of MPC and upstream of the demonstrated neuroprotective effect (fig. S2C). Together, the data obtained in nematodes and cultured neurons showed that when modulating MPC, functional changes in the mTOR pathway (described in detail below) appear with a delay after the immediate effects seen on mitochondrial function.

Next, we queried the role of the mTOR pathway on the neuroprotection observed in vivo after MPC modulation in the MPTP and En1+/− mouse models of PD. In the MPTP model, using Western blotting of substantia nigra samples, we measured mTOR, p-mTOR (Ser2448), S6, and phosphorylated S6 (pS6) (Ser235/Ser236), normalized to β-actin (Fig. 5I). As expected, MPTP injections increased p-mTOR measured as the p-mTOR/mTOR ratio relative to control conditions (P = 0.04) (Fig. 5J). MPTP treatment also increased the ratio of pS6/S6 (P = 0.03) (Fig. 5K). Although there was little to no effect under control conditions, MPC modulation using MSDC-0160 pretreatment reduced the activation of both mTOR and phosphorylation of its downstream substrate, indicating that this regulation was a component of neuroprotection (P = 0.032) (Fig. 5, I to K).

We also returned to the En1+/− mouse model to validate further the involvement of the mTOR pathway downstream of MPC. Using Western blotting of substantia nigra samples, we quantitated mTOR, p-mTOR (Ser2448 and Ser2481), p70S6kinase, phosphorylated p70S6kinase (Thr389), pS6 (Ser235/Ser236), S6, AKT, phosphorylated AKT (Thr308), regulated in development and DNA damage responses 1 (REDD1), LC3b, and p62 (Fig. 6, A to E, and fig. S7). These data show that all of these pathways were perturbed in the En1+/− mice versus wild-type animals (P value between 0.01 and 0.0001). Treatment with MSDC-0160 moved all of these changes toward the values observed in wild-type mice. The changes included those in mTOR signaling—increases in p-mTOR Ser2448/mTOR/β-actin, phosphorylated p70S6kinase/p70S6kinase/β-actin, pS6/S6/β-actin (Fig. 6, B to D, F, G, I, and J)—and also REDD1/β-actin ratios (fig. S7, C and F). As in the MPTP experiments described above, MSDC-0160 treatment of the wild-type mice did not affect mTOR signaling proteins (fig. S7). Notably, En1+/− mice also had reduced LC3b/β-actin (Fig. 6, E, H, and K) and p62/β-actin ratios (fig. S7G) relative to the wild-type mice, suggesting that the genetic model has a disturbance in autophagy. Treatment of the En1+/− mice with MSDC-0160 returned the amounts of LC3b and p62 proteins to those seen in wild-type mice. Again, there was no significant effect of MSDC-0160 treatment on the expression of these autophagy markers in wild-type mice (Fig. 6, E and K, and fig. S7, C to G).

Fig. 6. MSDC-0160 down-regulates mTOR signaling and restores autophagy in the En1+/− genetic mouse model of PD.

(A) Representative Western blot illustrating the expression of p-mTOR (Ser2448), mTOR, phosphorylated p70S6kinase (p-p70S6K) (Thr389), p70S6kinase, pS6 (Ser235/Ser236), S6, and LC3b in the SN. Bar graphs showing mean Western blot p-mTOR/mTOR ratios relative to β-actin (B), mean Western blot p-p70S6K/p70S6K ratios relative to β-actin (C), mean Western blot pS6/S6 ratios relative to β-actin (D), and mean Western blot LC3b/β-actin ratio (E). (F) Immunostaining for TH and p-mTOR (Ser2448) in SN; insets demonstrate overlap of TH and p-mTOR. (G) Immunostaining for TH and pS6 (Ser235/Ser236) in the SN. (H) Immunostaining for TH and LC3b. DAPI, 4′,6-diamidino-2-phenylindole. Intensity of p-mTOR (Ser2448) (I), pS6 (Ser235/Ser236) (J), and LC3b (K) in the TH-positive neurons [a.u. (arbitrary units)]. Scale bars, 10 μm. Data are means ± SEM of six to eight mice per group. ****P < 0.0001, ***P < 0.001, **P < 0.01, *P < 0.05. Data were analyzed by two-way ANOVA with Bonferroni’s multiple comparison test.

Together, the results indicate that modulation of MPC leads to an immediate effect on mitochondrial metabolism associated with a later reduction in the overactivation of the mTOR pathway. The data suggest that MPC modulation induces autophagy as part of the response that prevents neurodegeneration.

Role of inflammation in the effects of MPC modulation

Given that neuroinflammation is also considered a contributory factor to PD pathogenesis and occurs in several PD animal models (60), we were curious whether modulation of MPC with MSDC-0160 treatment would directly reduce inflammation. We found that MSDC-0160 reduced glial fibrillary acidic protein (GFAP; an astrocyte marker), ionized calcium-binding adapter molecule 1 (Iba-1; a microglial marker), and inducible nitric oxide synthase (iNOS) expression in the substantia nigra of MPTP-treated mice (Fig. 7, A to D, and fig. S8D). Similarly, in En1+/− mice, treatment with MSDC-0160 at the modest pathology stage attenuated expression of Iba-1, GFAP, and iNOS in the ventral midbrain, as assayed by immunoblot (Fig. 7, E to G, and fig. S8F). In addition, immunohistochemical analysis for Iba-1 and GFAP in sections through the substantia nigra of En1+/− mice showed that MSDC-0160 reduced elevated astrogliosis and microgliosis (Fig. 7H and fig. S8G).

Fig. 7. MSDC-0160 attenuates inflammation in animal models and in mouse microglial cells.

(A) Representative Western blot of Iba-1 and iNOS in the SN. Bar graphs show mean Iba-1/β-actin (B) and mean iNOS/β-actin in SN (C). (D) Iba-1 immunostaining in SN. Bar graph shows number of Iba-1–positive cells in SN. (E) Representative blot of Iba-1 and iNOS in SN. Bar graph shows mean Iba-1/β-actin (F) and mean iNOS/β-actin in SN (G). (H) Iba-1 immunostaining in SN. Bar graph shows number of Iba-1–positive cells in SN. (I) Immunocytochemistry of Iba-1 and iNOS in primary mouse microglial cells. (J) Nitrite measurement and (K) immunocytochemistry of Iba-1 and NF-κB–p65 in BV2 mouse microglial cells. (L) Representative Western blot of NF-κB–p65 from cytosol and nucleus in BV2 cells. Concentration of interleukin-1β (IL-1β) (pg/ml) (M), tumor necrosis factor–α (TNF-α) (pg/ml) (N), and IL-6 (pg/ml) (O) in BV2 cells. Immunocytochemistry of Iba-1 and p-mTOR (P) and Iba-1 and pS6 (Q) in BV2 cells. Scale bars, 50 μm (D, H, and I), 5 μm (K, P, and Q). (A to H) Data are means ± SEM of six to eight mice per group. (I to Q) Data are means ± SEM (n = 3 experiments). ****P < 0.0001, ***P < 0.001, **P < 0.01, *P < 0.05. Data were analyzed by two-way ANOVA with Bonferroni’s multiple comparison test.

To determine in a less complex model whether modulation of MPC directly affected inflammatory cells, we moved to a microglia cell line. As expected (see above), we observed expression of MPC-1 and MPC-2 in mouse BV2 microglia cells (fig. S8, A and B) (61). To induce inflammation in these cells, we used pretreatment (1 hour) with lipopolysaccharide (LPS; 1 μg/ml), which is an endotoxin that induces inflammation in vitro and in vivo. We found that in BV2 cell lysates, 10 μM MSDC-0160 blocked LPS-induced increases in iNOS expression (fig. S8C). We also confirmed our findings in mouse primary microglia cells. After harvesting primary microglia (62), we used immunofluorescence to determine that MSDC-0160 pretreatment (1 hour) blocked LPS-induced iNOS expression (Fig. 7I). Next, we performed a functional Griess assay on supernatant from the treated BV2 cells to detect the concentration of nitrite and found that MSDC-0160 blocked the LPS-induced increase in nitrite (Fig. 7J). Promoter regions of proinflammatory molecules contain the DNA binding site for nuclear factor κB (NF-κB), a transcription factor involved in inflammatory responses. The inhibition of NF-κB activation reduces the induction of proinflammatory molecules in PD animal models (63). Therefore, we monitored whether MPC modulation blocked the entry of p65 into the nucleus. Immunofluorescence of BV2 cells stained with antibodies against Iba-1 and phosphorylated (Ser276) p65 showed that pretreatment with MSDC-0160 prevented nuclear accumulation of phosphorylated p65 (Fig. 7K). Consistent with the immunofluorescence data, we found increased expression of phosphorylated (Ser276) p65 in the nuclear fraction of BV2 microglial cells (Fig. 7L). Next, we analyzed supernatants from BV2 cells on a Meso Scale platform to detect multiple cytokines. We determined that MSDC-0160 pretreatment lessened the LPS-related induction of expression of IL-1β (Fig. 7M), TNF-α (Fig. 7N), and IL-6 (Fig. 7O). Together, our results from BV2 cell cultures indicated that modulating MPC by MSDC-0160 directly in microglia prevented inflammation induced by LPS.

Consistent with the direct effect of MSDC-0160 on mitochondrial function in LUHMES cells subjected to MPP+ (fig. S2), MSDC-0160 normalized oxygen consumption in LPS-activated BV2 cells when added 1 hour before or after LPS addition, demonstrating its immediate action on mitochondrial function (fig. S9A). Also consistent with the effects in the LUHMES neuronal cultures (fig. S3, A and B), no early changes (2 and 6 hours) in the p-mTOR/mTOR ratio could be detected in LPS-activated BV2 cells treated with MSDC-0160. A reduction in mTOR activation caused by LPS was only observed after 24 hours of incubation with MSDC-0160 (Fig. 7P and fig. S9, B and C). Subsequently, LPS-induced activation of S6 was also attenuated by MSDC-0160 in BV2 cells (Fig. 7Q). However, no changes in the AKT pathway (pAKT/AKT ratio) were detected at any time point (fig. S9, B and C). This supports the notion that the anti-inflammatory effects of modulating MPC are downstream of the immediate effects on metabolism. Modulation of MPC had anti-inflammatory consequences in cellular and mouse models of inflammation and PD. The earliest measured effects involved direct modulation of pyruvate metabolism and protection of oxidative metabolism followed by changes in the mTOR/AKT pathways (increases in mTOR phosphorylation while AKT activation was reduced).

DISCUSSION

We show that modulating the MPC consistently normalized metabolism and downstream molecular pathways that are dysregulated in a range of cell, nematode, and mouse models of PD (both toxin-induced and genetic). As a consequence, autophagy and neuroinflammatory changes were normalized, and survival of substantia nigra dopaminergic neurons was promoted.

Modulation of MPC function has been tested as a therapeutic strategy in diabetes. The disease processes in diabetes and PD share several features, with metabolic alterations and inflammation being common to both diseases (3538, 64) and specific metabolic genes being linked to PD (64). Attempts to modify the course of PD with an antidiabetic agent of the TZD class (pioglitazone) have yielded conflicting results, possibly because of inadequate engagement of the target in the brain (8, 32). A safety and efficacy trial of the antidiabetic agent, exendin-4, a glucagon-like peptide-1 agonist, suggested a reduction in motor and cognitive decline in PD patients, which was maintained even 12 months after cessation of treatment (65, 66). These findings tentatively support the idea that disease mechanisms in diabetes and PD have common features, warranting further exploration of molecular therapeutic targets that could be common to the two diseases.

Here, modulation of MPC by MSDC-0160 normalized mTOR activity in response to multiple insults (Figs. 5 and 6). mTOR signaling is known to be activated in clinical samples from PD patients (67, 68). Aberrant mTOR signaling is also found in several models of PD, including the MPTP toxin rodent model (6971) and recently in the En1+/− genetic mouse model (53). The mTOR pathway interacts with a stress response protein, RTP801/REDD1, and has been suggested to control neuronal death in PD (72). Rapamycin, which is an inhibitor of mTOR signaling, promotes longevity, protects dopaminergic and other neurons, and stimulates autophagy in the MPTP mouse model of PD (72). In the mTOR signaling complex 1 (mTORC1), mTOR is predominantly phosphorylated on Ser2448, whereas in the mTORC2, it is predominantly autophosphorylated on Ser2481. In our PD models, we observed increased phosphorylation of mTOR at Ser2448 and Ser2481. We found that MSDC-0160 reduced phosphorylation at Ser2448, but not at Ser2481, mimicking the effects of rapamycin treatment, which inhibits signaling by mTORC1 but not mTORC2. The mTORC1 complex also phosphorylates the ribosomal protein p70S6kinase, an AGC kinase family member, on its hydrophobic motif site, Thr389 (73).

Our studies showed that the En1+/− mouse exhibits greater phosphorylation of p70S6kinase and its downstream target molecule ribosomal S6 and that modulation of MPC reduced the phosphorylation to wild-type levels. We also observed increased expression of REDD1 protein and inhibition of AKT phosphorylation at Thr308 in the En1+/− mice. These changes are all relevant to the pathophysiology of PD because dopaminergic neurons in PD exhibit down-regulation of the Ser473- and Thr308-phosphorylated forms of AKT (67). REDD1, which functions upstream of AKT, is induced during the neurodegenerative process (72). Modulation of MPC by MSDC-0160 reduced REDD1 (fig. S7, C and F). Although this could be due to attenuation of p70S6kinase activation, REDD1 was increased in response to multiple stresses, and it is likely that the upstream modulation of stress pathways is a component of the metabolic modulation.

Part of the metabolic modulation by MSDC-0160 in all of the models we examined included increased phosphorylation of AKT at Thr308. Although the precise mechanism underlying this change is not clear, reduction in endoplasmic reticulum stress might play a role (74). Some of the effects we observed after modulation of MPC function were consistent with previous work using mTOR inhibitors (72). Notably, whereas those inhibitors directly inhibit mTOR, the action of MSDC-0160 is on upstream metabolic changes deriving from the change in the mitochondrial handling of pyruvate (20, 21, 29). Thus, the effects of MSDC-0160 are mainly to prevent the activation of mTOR produced by the metabolic changes rather than to directly inhibit mTOR kinase activity. This is consistent with our observations of an immediate effect of MPC modulation on pyruvate utilization and oxygen consumption, followed by a delayed inhibition of the mTOR pathway.

It is well established that mTOR signaling is intricately tied to autophagy, which is a conserved homeostatic process by which unwanted cellular components are degraded by the lysosome (75). Several studies have implicated changes in the lysosome autophagy pathway in both idiopathic and certain rare genetic forms of PD (76, 77). By boosting lysosomal biogenesis and inducing autophagy, rapamycin restores normal lysosomal activity and mitigates dopaminergic neurodegeneration in the MPTP mouse model of PD (78). We show here that MSDC-0160 attenuated the activation of mTOR and its downstream signaling pathway in the subacute MPTP mouse model (Fig. 5, I to K). Similarly, we also found down-regulation of mTOR signaling along with reduced LC3b and p62 expression in En1+/− mice (Fig. 6 and fig. S6) and that modulation of MPC in this slowly progressing mouse model of PD reversed these changes. Together, we found that modulation of MPC normalized autophagy in multiple models of PD, which was likely to be key for preserving the nigral dopaminergic neurons.

Autophagic signals modulate inflammatory pathways (7981), and neuroinflammation is an integral part of the pathogenesis of PD (8284). Neuroinflammation was not only evident after LPS or MPTP treatment but was also found in the substantia nigra of En1+/− mice. We hypothesized that modulation of MPC influences inflammatory pathways. We found that modulation of MPC mitigated inflammation in all of the models we examined. In both MPTP-treated mice and En1-deficient mice, modulation of MPC attenuated the activation of GFAP-positive astrocytes and Iba-1–positive microglia. The reduced inflammatory response was likely not just the result of reduced neurodegeneration but also the result of direct effects of MSDC-0160 on the glial cells. We deduced this from our observations in cultured mouse BV2 cells, in which we found that modulation of MPC by MSDC-0160 blocked nuclear transport of p65 and reduced cytokine release in response to LPS administration. Given that mTOR also plays a crucial role in the regulation of the innate immune system (85, 86), and because mTOR inhibitors can inhibit release of inflammatory cytokines from activated macrophages (85, 86), we speculated that MPC modulation would also mitigate mTOR activation in stimulated microglia. In LPS-stimulated BV2 cells, exposure to MSDC-0160 caused a down-regulation of p-mTOR (Ser2448) and its target molecule, pS6 (Fig. 7). As was the case for neurons, these effects were temporally downstream from the direct effects of mitochondrial metabolism in the microglia. These results are key because they showed that MPC was engaged in both neurons and glial cells and that down-regulating the mTOR signaling pathway was a consequence in both cell types. On the basis of our observations, we propose that strategies that modulate MPC can have dual beneficial effects in PD, by both improving autophagy in neurons and reducing microglia activation (fig. S10). We suggest that modulation of MPC by MSDC-0160 reduced the effects of PD-causing insults, whether a toxin, a misfolded protein (α-synuclein), or an En1+/− mutation. The direct interaction of MSDC-0160 with the MPC complex has recently been modeled, showing a potentially unique pattern of interaction for both components of the complex (87). Downstream events may involve specific substrates, posttranslational modifications, and redox signals.

There are limitations to our study. It is not yet clear exactly how modulation of MPC results in the downstream changes that we have documented. In the rodent PD models, we cannot determine whether the neuroprotection was primarily the consequence of modulation of MPC in the neurons or the microglia, or both, because our cell culture studies clearly demonstrated that the MPC modulator MSDC-0160 could positively affect either cell population when they are grown in isolation. Similar to what has been published in flies (22), we found that knockdown of MPC-1 in the worm model blocked the response to MSDC-0160, although the knockdown itself did not protect against neurodegeneration. This illustrates a difference between modulating MPC by reversible inhibition and completely diminishing its activity by genetic manipulation. Notably, although the protective effects produced by MSDC-0160 treatment were consistent in all of the PD models we tested, some dopaminergic neurons still died in the different PD models, suggesting that it may not be possible to fully prevent neurodegeneration by MPC modulation alone.

Antidiabetic agents have recently been proposed as possible disease-modifying therapies for PD (36, 88). Diabetes, insulin resistance, inflammation, and PD are all linked (89), and PD patients exhibit an increased prevalence of diabetes (90). A recent phase 2, multicenter study using pioglitazone, a member of the first generation of TZDs, failed to show a disease-modifying effect in PD (8). A retrospective epidemiological study, however, determined that persons with diabetes with a prescription for a TZD insulin sensitizer drug were less likely to develop PD (32), suggesting that intervention with TZD insulin sensitizers early in the PD disease process, or its prodromal phase, would be more effective. First-generation TZDs, such as pioglitazone, not only modulate the MPC complex. In contrast to MSDC-0160, they are also direct activators of the nuclear transcription factor PPARγ, which drives dose-limiting side effects (29, 91). Consequently, the safety profile of MSDC-0160 is more favorable, which allows dosing to higher exposures than achieved with pioglitazone and thus makes it more likely to significantly engage the target MPC in the brain. The safety of MSDC-0160 has been demonstrated in people with diabetes and Alzheimer’s disease during 3 months of exposure (41, 42), with tentative evidence of target engagement in the brain from [18F]deoxyglucose imaging scans. In these trials, exposure to circulating MSDC-0160 and its metabolite was 50% higher than could be achieved with the highest approved dose of pioglitazone (45 mg, once daily). In response to MSDC-0160, hemoglobin A1c and fasting blood glucose in diabetic patients fell to levels similar to those of people taking pioglitazone, but fluid retention and other side effects were less prominent with MSDC-0160 treatment (41).

Given the effects of MPC modulation by MSDC-0160 in several different PD models in this study and its favorable safety profile in humans, we believe that targeting MPC in PD is warranted. MPC as a target is attractive because it affects multiple processes that are implicated in PD pathogenesis, including autophagy and neuroinflammation, through actions on both neurons and glia. These findings should stimulate the development of other MPC modulators as potential disease-modifying therapeutic agents in PD and related neurodegenerative disorders.

MATERIALS AND METHODS

Study design

The goal of this study was to establish evidence, using multiple PD models, that modulation of the mitochondrial pyruvate carrier complex via MSDC-0160 is a viable approach to modify disease progression in PD. All experiments were blinded. Typically, one individual would randomize the animals, plates, and slides, and another would analyze them. The minimum sample size for all experiments was held at six mice per group based on the design of previous studies (51, 92). To improve our power, and thus our ability to statistically detect smaller effects, many of our analyses included more rodents per group and/or repeated measures. Further experimental details and protocols of each model, including animal care/handling and the number of biological/technical replicates, are in this section, in the Supplementary Materials, and in the figure legends.

C. elegans strains and protocols

C. elegans strains were cultured and maintained on nematode growth medium containing a lawn of OP50 bacteria at 20° or 16°C using established techniques (93). The transgenic strain BY250 (vtIs7[pDAT::GFP(pRB490)]) was provided by the laboratory of R. Blakely (Florida Atlantic University, Jupiter, FL). The strain TU3401 (sid-1(pk3321) V; uIs69[pCFJ90(myo-2p::mCherry) + unc-119p::sid-1] V) was obtained from the Caenorhabditis Genetics Center, which is funded by the National Institutes of Health Office of Research Infrastructure Programs (P40 OD010440). The α-synuclein–expressing strain JVR107 was produced by transforming a Pdat-1::α-syn (A53T) construct into wild-type C. elegans (N2 Bristol strain), generated by T. Iwatsubo (University of Tokyo, Japan). Additional strains were generated by crossing these strains to generate JVR203 (Pdat-1::a-syn(A53T), Pges-1::RFP; vtIs7[Pdat-1::GFP(pRB490)]), JVR325 (Pdat-1::a-syn(A53T), Pges-1::RFP; vtIs7[Pdat-1::GFP(pRB490)]; sid-1(pk3321) V; uIs69[pCFJ90(myo-2p::mCherry) + unc-119p::sid-1] V) and JVR326 (vtIs7[Pdat-1::GFP(pRB490)]; sid-1(pk3321) V; uIs69[pCFJ90(myo-2p::mCherry) + unc-119p::sid-1] V).

C. elegans MPP+ neurodegeneration assay

Worms were synchronized, and L1 larvae (BY250 strain, GFP expression in dopaminergic neurons) were placed in 96-well plates containing OP50 bacteria in liquid culture (6 mg/ml) and MPP+ (0.75 mM) with MSDC-0160 (1, 10, or 100 μM). Wells containing no MPP+ [substituted with water or vehicle (methylcellulose)] and with MSDC-0160 alone were also included. Because in these experiments worms were soaked in solution containing MSDC-0160 of varying concentrations, the exact dose ingested by the worm cannot be accurately calculated. Therefore, the maximum concentration that had a neuroprotective effect was used in subsequent experiments to ensure efficacy. About 50 worms were added per well, and the plate was incubated at 20°C for 48 hours. After 48 hours, worms were moved and allowed to recover on unseeded nematode growth medium plates for several minutes before being mounted on an agarose pad with 3 mM levamisole and analyzed microscopically. Each worm was scored for the presence of GFP-fluorescent dopaminergic neurons in the anterior of the worm (CEP and ADE neurons). A minimum of 25 worms were analyzed for each condition in three independent trials.

Measurement of brain and plasma drug concentrations

C57BL/6J mice were orally dosed with MSDC-0160 (30 mg/kg) suspended in 1% low-viscosity methylcellulose with 0.01% Tween 80 in distilled water (10 ml/kg) and then euthanized 2 and 4 hours later. Plasma and whole-brain homogenates were extracted and subjected to LC/mass spectrometry measurement of both active drug and active alcohol metabolite (MSDC-0037), as previously described (41).

MPTP mouse model

Ten- to 12-week-old male C57BL/6J mice weighing 24 to 28 g were housed under standard conditions: constant temperature (22°C), humidity (relative, 30%), and a 12-hour light/dark cycle, with free access to food and water. Procedures were performed during daylight hours and were approved and supervised by the Institutional Animal Care and Use Committee at the Van Andel Research Institute. Mice were administered MSDC-0160 (30 mg/kg) by oral gavage beginning 24 hours before MPTP (Sigma-Aldrich) treatment. Next, mice received five consecutive doses of MPTP via intraperitoneal injection at 25 mg/kg per day (subacute regimen) along with coadministration of MSDC-0160, followed by 6 days of MSDC-0160 treatment. MSDC-0160 was dissolved in 1% methylcellulose with 0.01% Tween 80. Control mice received vehicle treatment (1% methylcellulose with 0.01% Tween 80). In the modest pathology stage, mice were administered MSDC-0160 (30 mg/kg per day) by oral gavage for 7 days starting 3 days after MPTP (25 mg/kg per day) treatment. Seven days after MPTP treatment, mice were euthanized, and tissues were processed for further evaluation. Mice were randomized to the experimental groups.

En1+/− mouse model

The En1+/− heterozygous mice were maintained on an OF1 genetic background (52) and under the same conditions as the C57BL/6J mice. In the mild pathology stage, En1+/− mice were fed a diet of chow formulated to deliver MSDC-0160 (30 mg/kg) starting at 3 weeks of age (at this point, no nigral cell death is detectable) and were euthanized at two different time points: 28 and 48 weeks. In the modest pathology stage, En1+/− mice were fed a diet of chow formulated to deliver MSDC-0160 (30 mg/kg) starting at 8 weeks of age (at this point, 15 to 20% nigral cell death is detectable) and were euthanized at either 16 or 28 weeks. Mice were randomized to the groups, and control mice were fed regular chow.

Behavioral experiments

Spontaneous activity data were collected and analyzed by an ANY-maze analyzer (Stoelting). The dimensions of the activity chamber (San Diego Instruments) were 109 cm by 109 cm by 38 cm. Before any treatment, mice were placed daily inside the chamber for 10 min for two consecutive days. Open-field activities were recorded for 10-min test sessions. We evaluated a total of seven parameters: total distance traveled, mean speed, maximum speed, time frozen, time mobile, time immobile, and head distance traveled. For the rotarod experiment, a speed of 18 rpm for C57BL/6J mice and 10 rpm for En1+/− mice was used. Before testing, each mouse was trained on the rotarod for 5 min on three consecutive days. Mice were given a 7- to 10-min rest between rotarod recording sessions. All behavior experiments were conducted in a blinded fashion.

Measurement of fluorescence intensity and densitometry

A total of three to four sections per brain containing the substantia nigra and three mice per group were stained with antibodies directed against TH, p-mTOR, pS6, and LC3b. After immunofluorescence staining, we took five 40× images from each section, blind-coded them, and used the NIS-Elements AR 4.00.08 software (Nikon) to quantify mean intensity of fluorescence of p-mTOR, pS6, LC3b, and TH in TH-immunoreactive neurons, as well as the number of LC3b-immunoreactive punctae in TH-positive nigral neurons (53).

HPLC analysis for striatal monoamine detection

Striatal levels of dopamine and DOPAC were quantified as described previously (48). Briefly, striata tissues were collected, immediately frozen on dry ice, and stored at −80°C until analysis. On the day of analysis, tissues were sonicated in 0.2 M perchloric acid containing isoproterenol (internal standard), and the homogenates were centrifuged at 20,000g for 15 min at 4°C. Dopamine and DOPAC were separated isocratically in a C18 reversed-phase column using an HPLC system with an automatic sampler equipped with a refrigerated temperature control and electrochemical detector (Thermo Fisher Scientific). Data acquisition and analysis were performed using the Chromeleon HPLC Software.

Pyruvate oxidation assay

The effect of MSDC-0160 on oxygen consumption rates in brain mitochondria was analyzed using MitoXpress Xtra (Luxcel). This fluorescence reagent, which is quenched by oxygen, allows for the direct, real-time analysis of oxygen consumption. Mitochondria were prepared by differential centrifugation of freshly prepared Sprague-Dawley rat brain homogenates and resuspended in respiration buffer containing 250 mM sucrose, 15 mM KCl, 1 mM EGTA, 5 mM MgCl2, and 30 mM K2HPO4 (pH 7.4). MSDC-0160 concentration curves (0, 0.1, 0.3, 1.0, 3.0, and 10 μM) were prepared in dimethyl sulfoxide (0.2% final) and assayed in duplicate with 50 μg of mitochondrial protein per well in a 96-well format (Corning). Malate (5 mM)/pyruvate (5 mM)/adenosine 5′-diphosphate (1 mM), determined empirically to be optimal substrate concentrations, were included in the reaction. Time-resolved fluorescence (excitation/emission, 380/645; time-resolved 30-μs delay; 100-μs read) was measured over a 2-hour time course on a BioTek Synergy 2 plate reader. Respiration curves and kinetic analysis were conducted using R version 3.2.2 and GraphPad Prism software, respectively.

Seahorse assay for cellular oxygen consumption

Basal cellular oxygen consumption was measured using the Seahorse XFe96 Analyzer. Two days before experiments, BV2 cells were plated at 1.5 × 105 cells/ml (seeding density) in Seahorse 96-well utility plates and supplemented with RMPI 1640 medium. LUHMES cells were plated at 5 × 105 cells/ml (seeding density) in Seahorse 96-well utility plates and supplemented with Advanced Dulbecco’s modified Eagle’s medium/F12 medium. At 24 hours before the assay, 10 μM MPP+ or 10 μM LPS was added to appropriate wells, and Seahorse probes were hydrated with Seahorse calibrant solution (pH 7.4). The day of the assay, medium was removed and replaced with Seahorse media (pH 7.4). One hour before the assay, cells were treated with 10 μM MSDC-0160 (posttreatment conditions). Cells were then washed twice in Seahorse medium and calibrated in a 37°C non-CO2 incubator for 1 hour. Oxygen consumption was measured twice basally. Injections of vehicle, 10 μM MSDC, or 10 μM MPP+/LPS (pretreatment conditions) were then added, and six oxygen consumption rate measurements of 3 min each were taken in a blinded manner. Measurements were normalized to 50,000 cells per well. A total of 15 wells per condition were measured, and three independent experiments for each cell line were performed.

Statistical analysis

Except for the behavioral data from En1+/− mice, data were analyzed with Prism 3.0 software (GraphPad Software). Raw data were first analyzed using either one-way or two-way ANOVA, and then either Tukey’s or Bonferroni’s multiple comparison test was performed to compare all treatment groups. Differences with P < 0.05 were considered significant. Linear mixed-effects models (LMMs) (94) were used to determine whether the interaction between the En1+/− genotype and the MSDC drug treatment was significant.

There are two major advantages to using LMMs: first, they appropriately adjust for repeated measures but allow for the inclusion of mice without repeated measures, unlike repeated-measures ANOVA; second, LMMs adjust for a portion of the within-subject variation via random effects. Both of these features should result in superior model performance and accuracy (95). Each LMM fit included independent intercepts to adjust for some of the within-subject variation observed between mice at the initial time points. Random slopes were also considered, but there were not enough data to include this effect.

Analyses for each outcome were stratified by the week in which chow treatment was initiated (that is, 3 or 8 weeks). For each outcome, the LMMs were first fit with a three-way interaction between genotype (wild-type/En1+/−), time point (28 and 44 weeks or 16 and 28 weeks for 3- and 8-week treatment start times, respectively), and treatment (control/MSDC). If this interaction was not found to be significant, then it was removed from the model, and the model was then refit, including all two-way interactions (that is, genotype × treatment, genotype × time, and treatment × time). If the genotype × time interaction was not significant, then it was removed; likewise, for the treatment × time interaction. The genotype × treatment interaction was retained regardless of significance because this interaction pertained to one of the primary questions of interest, namely, does MSDC-0160 improve the En1+/− measures. Thus, the minimal final model included at least the following variables: genotype, treatment, time, and the genotype × treatment interaction.

An FDR correction was used to adjust all of the P values calculated by the 18 LMMs. FDR corrections are less reliant upon the need for all tests to be independent (96), which is particularly important because many of our outcome measurements are highly correlated and, thus, not independent.

Observations were determined to be outliers if they were more than 3 SDs away from the mean of the residuals. All observations that met this criterion were investigated to ensure that the data were recorded accurately (for example, not miskeyed) and that the measurements were within the realm of plausibility. Measurements found to have recording errors were corrected. No observations were removed. All LMM analyses were performed using R version 3.2.0 (97).

SUPPLEMENTARY MATERIALS

www.sciencetranslationalmedicine.org/cgi/content/full/8/368/368ra174/DC1

Materials and Methods

Fig. S1. MSDC-0160 protects dopaminergic neurons against MPP+-induced toxicity in LUHMES cells and C. elegans.

Fig. S2. Modulation of MPC improves mitochondrial function.

Fig. S3. Time-course treatment of MSDC-0160 in LUHMES cells.

Fig. S4. MSDC-0160 improves behavioral parameters in the MPTP mouse model.

Fig. S5. MSDC-0160 improves behavioral activities in the En1+/− mice.

Fig. S6. MSDC-0160 prevents dopaminergic neurodegeneration in the En1+/− mice.

Fig. S7. MSDC-0160 down-regulates mTOR signaling pathway and restores autophagy in the En1+/− mice.

Fig. S8. MSDC-0160 attenuates inflammation in BV2 cells and in vivo animal models.

Fig. S9. MSDC-0160 improves mitochondrial function and blocks mTOR activation in BV2 cells.

Fig. S10. Schematic diagram of proposed mechanism of action of MSDC-0160 in glial cells and dopaminergic neurons.

Table S1. LMM results.

Reference (98)

REFERENCES AND NOTES

  1. Acknowledgments: We thank R. Wyse, Cure Parkinson’s Trust, for helpful discussions. We acknowledge the technical expertise of C. Cole, D. Dues, L. Kefene, and D. Marckini. We also acknowledge excellent service from the Vivarium Core and the Biostatistics and Bioinformatics Core of the Van Andel Research Institute. We thank D. Nadziejka for manuscript editing assistance. We also thank J. Das and J. L. Leasure from the University of Houston for assistance with the StereoInvestigator software and Western blot settings in their laboratories. Funding: The work presented here was supported by Cure Parkinson’s Trust (to P.B.), Campbell Foundation (to P.B.), Spica Foundation (to P.B.), and Van Andel Research Institute (to P.B. and J.M.V.R.). Author contributions: A.G., J.R.C., J.M.V.R., and P.B. designed the study. A.G., T.T., S.G., E.N.H., E.M., E.S., and W.G.M. carried out the experiments: A.G. did all of the experiments except the following: C. elegans studies (T.T.), microglia cultures (E.N.H.), Seahorse assay (E.M.), drug pharmacodynamics (W.G.M.), and stereological cell counts in some experiments (E.S.). A.G., T.T., S.G., E.N.H., Z.M., W.G.M., M.L.E.G., J.A.S., J.R.C., and J.H.K. performed analyses and figure preparation. A.G., J.A.S., T.T., Z.M., M.L.E.G., J.H.K., J.R.C., and P.B. wrote the manuscript. Competing interests: JRC is the cofounder and significant owner of Metabolic Solutions Development Company (MSDC), which is currently developing MSDC-0160 as a potential treatment for Alzheimer’s disease. MSDC-0160 is protected by several patents, including US 8389556 B2 and EP 2001468 B1, which are assigned to MSDC. P.B. has received paid support as a consultant from Renovo Neural Inc., Roche, Teva Pharmaceutical Industries, Lundbeck A/S, AbbVie Inc., IOS Press Partners, and Versant Ventures. In addition, P.B. has received support for his research from Renovo, Teva, and Lundbeck. P.B. has ownership interests in Acousort AB and ParkCell AB. J.H.K. has received paid commercial support as a consultant from Cellular Dynamics International Inc., Michael J. Fox Foundation, nLife Therapeutics S.L., NsGene, Clintrex, NeuroDerm, and BrainEver. Data and materials availability: All enquiries regarding MSDC-0160 should be directed to J.R.C.; MSDC-0160 will be made available through a material transfer agreement.
View Abstract

Navigate This Article