Introduction

Parkinson’s disease (PD), the second most common neurodegenerative disease, is characterized by the substantial loss of dopaminergic neurons in the substantia nigra pars compacta (SNpc), accompanied by motor impairments, such as tremor, rigidity and bradykinesia, and nonmotor symptoms [1]. Although mitochondrial dysfunction is considered to be one of the molecular pathologies in PD, the mechanisms have still not been clearly elucidated.

Mitochondrial biogenesis plays an important role in neurodegeneration and is regulated by several cell signaling pathways, including the AMP-activated protein kinase (AMPK)/peroxisome proliferator–activated receptor gamma coactivator-1α (PGC-1α) and sirtuin 1 (SIRT1)/PGC-1α pathways [2, 3]. The suppressive activity of PGC-1α, a master mitochondrial biogenesis regulator, is considered the main cause of familial PD induced by PTEN induced putative kinase 1 (PINK1) and parkin mutations [3]. Therefore, the development of new drugs targeting mitochondrial biogenesis seems to have potential applications for PD treatment [4].

Rotenone is widely used to establish PD models in rodents [5, 6], as it reproduces the pathological hallmarks of PD as well as some parkinsonian motor deficits [7, 8]. Systemic administration of rotenone induced chronic progressive degeneration of the nigrostriatal pathway and α-synuclein pathology in rats, similar to observations in PD patients [5, 9]. Both human PD and chronic rotenone exposure lead to dopaminergic axon loss followed by cell death [10]. Our previous study indicated that pyrroloquinoline quinone (PQQ), a redox cofactor targeting mitochondrial respiratory chain complexes I and III, could antagonize rotenone-induced cell injury both in vitro and in vivo [11,12,13], suggesting that PQQ might be a candidate for PD treatment through its regulation of mitochondrial functions.

Posttranslational modifications (PTMs) are increasingly recognized as crucial mechanisms in PD pathogenesis, as many important proteins involved in PD, such as α-synuclein and leucine-rich repeat kinase 2 (LRRK2), are modulated by PTMs, which affect cellular functions by regulating protein stability, localization and function [14, 15]. Recently, some drugs and molecules were found to exert protective effects in PD models by regulating protein phosphorylation, acetylation, methylation or ubiquitination [1, 8, 16, 17]. Our previous studies also suggested that PQQ could phosphorylate extracellular regulated protein kinases (ERKs) and members of the phosphatidylinositol-3-kinase (PI3K)/protein serine-threonine kinase (AKT) signaling pathways to protect SH-SY5Y cells and dopaminergic neurons from rotenone injury [11, 12]. Further comprehensive investigation might be helpful to gain insight into the neuroprotective effect and mechanism of PQQ in PD models.

In the present study, a PD mouse model induced by intraperitoneal rotenone injections was used to observe the neuroprotective effect of PQQ. Moreover, an immunoaffinity-based LC-MS/MS method, PTMScan Direct analysis, was applied to identify signaling pathways that play specific roles in the effect of PQQ against rotenone injury. PTMScan Direct, an antibody-based proteomic method for the identification and quantification of hundreds of phosphorylated peptides and proteins in many different cell signaling pathways, has been employed to study mechanisms involved in neurodegenerative diseases [18, 19]. We aimed to investigate the neuroprotective effect of PQQ in a rotenone-induced PD model and to detect the changes in the protein phosphorylation profile of rotenone-injured SH-SY5Y cells induced by PQQ treatment. Our findings in a rotenone-induced PD model will provide a new understanding of the molecular mechanism involved in the neuroprotective effect of PQQ, indicating that PQQ might be a potential therapeutic target for PD treatment.

Materials and methods

Chemicals and antibodies

PQQ, rotenone, trypsin, monoclonal mouse anti-β-Actin antibody, and Epon 812 were purchased from Sigma-Aldrich (St. Louis, MO, USA). Polyclonal rabbit anti-p-AMPKα antibody, polyclonal rabbit anti-AMPKα antibody, the PTMScan® Multi-Pathway Enrichment Kit, HRP-conjugated goat anti-mouse IgG and HRP-conjugated donkey anti-rabbit IgG were obtained from Cell Signaling Technology (Danvers, MA, USA). Polyclonal rabbit anti-TH antibody, polyclonal rabbit anti-PGC-1α antibody, polyclonal rabbit anti-TFAM antibody and donkey anti-rabbit IgG-FITC were purchased from Abcam (Cambridge, MA, USA). Dulbecco’s modified Eagle’s medium (DMEM) and fetal bovine serum (FBS) were purchased from Gibco (Grand Island, NY, USA). Tissue lysis buffer, protease inhibitor cocktail, a BCA-based protein quantification kit and BeyoECL Plus were obtained from Beyotime (Shanghai, China). TRIzol reagent and MitoTracker® Green FM were purchased from Life Technologies (Carlsbad, CA, USA). Fast EvaGreen qPCR Master Mix was purchased from Biotium (Hayward, CA, USA). The Omniscript reverse transcription (RT) kit was provided by Qiagen (Valencia, CA, USA). Dorsomorphin (compound C) 2HCl was purchased from Selleck (Houston, TX, USA).

Animals

Adult ICR mice (25–30 g, male) were obtained from the Experimental Animal Center of Nantong University (China). The animal study was reviewed and approved by the Institutional Animal Care Committee of Nantong University (Nantong, China). The mice were randomly divided into five groups before drug administration and adapted to the environment for 3 days. A total of 18 mice were included in each group in the experiment. Rotenone (3 mg·kg−1·d−1 dissolved in normal saline with 2% DMSO) was intraperitoneally administered once daily for 3 weeks to establish a PD mouse model. PQQ at different dosages (0.8, 4, and 20 mg·kg−1·d−1) was intraperitoneally administered immediately after rotenone injection every day. The animals in the control group were injected with normal saline containing 2% DMSO. The body weight of each mouse was recorded every week.

For the detection of AMPK activity in PD mice, a time-course study was performed. Rotenone (3 mg·kg−1·d−1) was intraperitoneally administered once daily for 3 weeks. PQQ at a high dosage (20 mg·kg−1·d−1) was intraperitoneally administered immediately after rotenone injection. The control group was injected with normal saline containing 2% DMSO. Ten mice were used for each group. The midbrains were harvested at the end of weeks 1, 2, and 3 for protein analysis.

Behavioral tests

At the end of treatment, all mice were subjected to behavioral training for 3 consecutive days before the tests were performed. The rotarod test was used to evaluate motor coordination in PD mice using an accelerating rotarod apparatus (RWD, Shenzhen, China) as described previously [20]. The mice were trained in acceleration mode (4–40 rpm) over 5 min, and the latency before each mouse fell off the rod in the test was recorded. The pole test was applied with minimal modifications as described previously to assess bradykinesia [20]. Briefly, mice were placed with their head upward at the top of a rough wooden pole (16 mm in diameter and 60 mm in height), and the total time that each mouse took to orient downward and descend to the floor was recorded. The traction test was conducted as described previously [21]. Briefly, mice were suspended by the forepaws on a horizontal wire approximately 70 cm away from the ground. Each mouse was scored as follows: 3 points, the mouse grasped the wire with two hind limbs; 2 points, the mouse grasped the wire with one hind limb; 1 point, the mouse did not grasp the wire with either of the hind limbs. The open field test was performed, and the activity of each mouse was monitored for 30 min. The distance moved by each mouse was automatically calculated by a video tracking system. Eighteen animals in each group were used for the behavioral tests.

TH immunofluorescent staining

After the behavioral tests, the midbrains were dissected, sectioned, and incubated with anti-TH antibody (1:200) overnight at 4 °C and donkey anti-rabbit IgG-FITC (1:200) for 2 h at room temperature. The average immunostaining fluorescence intensity was measured with ImageJ software (version 1.51, NIH, Bethesda, MD, USA). Four animals in each group were used for TH staining.

Transmission electron microscopy (TEM)

The midbrains were dissected, fixed with precooled 2.5% glutaraldehyde, post fixed with a 1% osmium tetroxide solution, dehydrated stepwise with increasing concentrations of ethanol, and embedded in Epon 812 epoxy resin. Ultrathin sections were stained with lead citrate and uranyl acetate and then examined under a transmission electron microscope (JEOL Ltd., Tokyo, Japan). We classified the mitochondria into “type 1” and “type 2” based on organelle structure; “type 1” mitochondria were very packed and very electron-dense, and “type 2” mitochondria were less electron-dense and showed a reduced number of mitochondrial cristae and enlarged intracristal spaces. The “type 1” and “type 2” mitochondria were quantified. Four animals in each group were used for TEM observation. Cultured SH-SY5Y cells under different treatments were also subjected to TEM observation as described above.

Quantitative real-time reverse transcription PCR (qRT-PCR)

Total RNA was extracted from the midbrain and cortex tissues with TRIzol, and the RNA was reverse transcribed into cDNA with an Omniscript RT kit according to the manufacturer’s instructions. The reaction mixture consisted of 5 μL of 2×Fast EvaGreen Master Mix, 0.5 μL of forward primer, 0.5 μL of reverse primer (Table 1), and 1 μL of cDNA. PCR amplification was performed on the StepOne real-time PCR system (Applied Biosystems, Foster City, CA, USA). The relative mRNA expression level was calculated by the comparative 2−ΔΔCt method and normalized against 18S rRNA.

Table 1 Oligonucleotide sequences used in qRT-PCR analysis.

Western blotting

The mouse midbrain and cortex and cultured SH-SY5Y cells were individually collected in RIPA lysis buffer (20 mM Tris-HCl [pH 7.5], 150 mM NaCl, 1 mM Na2EDTA, 1 mM EGTA, 1% NP-40, 1% sodium deoxycholate, 2.5 mM sodium pyrophosphate, 1 mM β-glycerophosphate, 1 mM Na3VO4, 1 µg/mL leupeptin), and the protein concentration was quantified by BCA analysis, followed by electrophoretic separation by SDS-PAGE. After transfer to a PVDF membrane (Millipore, Bedford, MA), the membrane was blocked with 5% nonfat dry milk in Tris-buffered saline (TBS, pH 7.4) and incubated with anti-PGC-1α (1:500), anti-TFAM (1:500), anti-p-AMPKα (1:500), anti-AMPKα (1:500), or anti-β-Actin (1:4000) antibody at 4 °C overnight. After washing with TBS/T (TBS with 0.1% Tween 20), HRP-conjugated affinity-purified donkey anti-rabbit IgG (1:5000) or goat anti-mouse IgG (1:5000) was applied and incubated at room temperature for 1 h. The membrane was incubated with ECL substrate and scanned with an imaging system (Tanon, Shanghai, China). The data were analyzed with PDQuest 7.2.0 software (Bio-Rad, Hercules, CA, USA). β-Actin was used as an internal control to normalize protein loading.

Cell lysate preparation and PTMScan multipathway analysis

SH-SY5Y human neuroblastoma cells were cultured in DMEM supplemented with 10% FBS at 37 °C in a humidified atmosphere of 95% air and 5% CO2. The day before treatment, cells were plated in a 10-cm culture dish at a density of 1 × 106 cells/mL. The cells were divided into four groups: the control group (no rotenone or PQQ), PQQ group (100 μM PQQ), Rotenone group (100 μM rotenone), and PQQ + Rotenone group (100 μM PQQ + 100 μM rotenone). Thirty minutes after treatment, the cells were washed once with cold PBS and scraped in urea lysis buffer (9 M sequanal-grade urea, 20 mM HEPES [pH 8.0], 1 mM β-glycerophosphate, 1 mM sodium vanadate, 2.5 mM sodium pyrophosphate). The cells were sonicated in lysis buffer and centrifuged for 15 min at 4 °C and 20,000 × g. Supernatants were collected and reduced with 4.5 mM DTT for 30 min at 55 °C. Reduced lysates were alkylated with iodoacetamide (0.019 g/mL of H2O) for 15 min in the dark. The samples were diluted 1:4 with 20 mM HEPES [pH 8.0] and digested with 10 g/mL trypsin-tosylphenylalanyl chloromethyl ketone in 1 mM HCl overnight. Digested peptide lysates were acidified with 1% TFA, and the peptides were desalted over 360-mg SEP PAK Classic C18 columns (Waters, Milford, MA, USA). Peptides were eluted with 40% acetonitrile in 0.1% TFA, dried under a vacuum, and stored at −80 °C. PTMScan Direct was adapted from the PhosphoScan method developed at Cell Signaling Technology [18]. Differentially expressed phosphoproteins were identified and analyzed.

Cell viability measurements

SH-SY5Y cells were plated in a 96-well plate and treated with rotenone or rotenone and different concentrations of PQQ. The AMPK inhibitor dorsomorphin (4 μM) was applied to the cells 30 min before rotenone and PQQ treatment. Cell viability was assessed by the CCK-8 method after 24 h. The absorbance (optical density, OD) at 450 nm was measured by spectrophotometry with an ELx800 microplate reader (Bio-Tek Inc., Winooski, VT, USA).

MitoTracker® green staining

SH-SY5Y cells were plated in a 24-well plate and treated with rotenone or rotenone and different concentrations of PQQ. Dorsomorphin (4 μM) was applied to the cells 30 min before rotenone and PQQ treatment. Twenty-four hours after treatment, the cells were incubated with 100 nM MitoTracker® Green FM at 37 °C for 30 min, after which the cells were fixed in 3% glutaraldehyde for 30 min. Images were taken with a Leica confocal microscope, and the average fluorescence intensity was measured with ImageJ software (version 1.51, NIH, Bethesda, MD, USA).

Statistical analysis

Statistical analysis was performed by GraphPad Prism 5 software (GraphPad Software Inc., San Diego, CA, USA), and data are presented as the means ± SEMs. Differences between groups were analyzed by one-way or two-way analysis of variance (ANOVA) and subsequent Tukey’s multiple comparisons test. Statistical significance was set as P < 0.05.

Results

Effect of PQQ on locomotor activity in PD mice

PQQ at different concentrations was intraperitoneally injected for 21 consecutive days, and the effect of PQQ was observed in a rotenone-induced PD mouse model. Body weight increased steadily over the experiment with time, and no significant differences in body weight were observed between the groups (Fig. 1a). Motor functions were assessed by the rotarod test, pole test, traction test and open field at the end of injection. In the rotarod test, a significantly shorter latency to fall was observed in the rotenone (Rot) group (3 mg·kg−1·d−1) (38.56 ± 4.19 s) than in the control group (64.78 ± 3.76 s) (Fig. 1b). Treatment with 4 mg·kg−1·d−1 (62.89 ± 4.79 s) and 20 mg·kg−1·d−1 (57.56 ± 7.52 s) PQQ significantly prevented this decrease in latency (Fig. 1b). In the pole test, PQQ at a high dosage inhibited the increase in time spent descending from the pole, as shown by the following results (Fig. 1c): control, 12.44 ± 1.22 s; Rot, 21.44 ± 2.54 s; Rot + 0.8 mg·kg−1·d−1 PQQ, 16.28 ± 1.12 s; Rot + 4 mg·kg−1·d−1 PQQ, 15.83 ± 1.03 s; Rot + 20 mg·kg−1·d−1 PQQ, 14.06 ± 1.25 s. The rotenone-induced decrease of traction test score could also be rescued by 4 and 20 mg·kg−1·d−1 PQQ treatment, as shown by the following results (Fig. 1d): control, 2.50 ± 0.15; Rot, 1.06 ± 0.17; Rot + 0.8 mg·kg−1·d−1 PQQ, 1.67 ± 0.20; Rot + 4 mg·kg−1·d−1 PQQ, 1.83 ± 0.19; Rot + 20 mg·kg−1·d−1 PQQ, 1.83 ± 0.17. However, no significant differences in distance between groups were detected in the open field test (Fig. 1e): control, 8.56 ± 0.39 m; Rot, 8.30 ± 0.48 m; Rot + 0.8 mg·kg−1·d−1 PQQ, 8.48 ± 0.37 m; Rot + 4 mg·kg−1·d−1 PQQ, 7.66 ± 0.44 m; Rot + 20 mg·kg−1·d−1 PQQ, 8.21 ± 0.38 m. Together, the above data indicated that PQQ could improve locomotor activity in rotenone-induced PD mice.

Fig. 1: PQQ alleviated rotenone-induced locomotor deficits in mice.
figure 1

Rotenone (3 mg·kg−1·d−1 dissolved in normal saline with 2% DMSO) was intraperitoneally injected into ICR mice for 21 consecutive days, and PQQ at different doses was simultaneously injected. The mice were divided into the control group (Control), rotenone injury group (Rot), low-dosage PQQ group (Rot + 0.8 mg·kg−1·d−1 PQQ), medium-dosage PQQ group (Rot + 4 mg·kg−1·d−1 PQQ), and high-dosage PQQ group (Rot + 20 mg·kg−1·d−1 PQQ). Normal saline with 2% DMSO was used in the control group. a Body weight change after rotenone and PQQ treatment. b Average time of latency for the mice to fall in the rotarod test. c Total time for each mouse to orient downward and descend to the floor in the pole test. d Score in the traction test. e Distance measured in the open field test. Data are represented as the mean ± SEM (n = 18 for each group). *P < 0.05 and **P < 0.01 vs the Rot group. ##P < 0.01 vs the control group.

Effect of PQQ on the loss of TH-positive nigral neurons in PD mice

To observe the effect of PQQ on neuronal loss in the SNpc, TH staining was performed, and the images are shown in Fig. 2a. The average fluorescent intensity of TH staining in the SNpc was lower in the rotenone group than in the control group, but both 4 and 20 mg·kg−1·d−1 PQQ treatment significantly increased the number of TH-positive cells, as shown by the following data (Fig. 2b): control, 224.70 ± 7.77; Rot, 125.00 ± 10.92; Rot + 0.8 mg·kg−1·d−1 PQQ, 133.30 ± 8.47; Rot + 4 mg·kg−1·d−1 PQQ, 176.20 ± 13.38; Rot + 20 mg·kg−1·d−1 PQQ, 203.10 ± 11.35. These data suggested that PQQ could inhibit neuronal loss in the SNpc of PD mice, which might be responsible for the rescue of motor functions observed in behavioral tests.

Fig. 2: PQQ prevented rotenone-induced neuronal loss in the SNpc of mice.
figure 2

a Representative images of TH immunostaining (green) in the SNpc. Scale bar = 100 μm. b Quantitative analysis of the fluorescence intensity in the SNpc. *P < 0.05 and **P < 0.01 vs the Rot group. ##P < 0.01 vs the control group (n = 4).

Effect of PQQ on mitochondrial biogenesis in PD mice

To evaluate the impact of PQQ on mitochondrial biogenesis in rotenone-induced PD mice, the ultrastructure of mitochondria in the midbrain neurons was observed by electron microscopy. In the control group, most mitochondria showed a normal structure and, because they were packed and electron-dense, were classified as “type 1” mitochondria. After rotenone-induced injury, the total number of mitochondria decreased, and some mitochondria displayed a pathological morphology, such as incomplete membranes, vacuoles in the matrix, enlarged intercristal spaces and increased mitochondrial size with swelling, and were thus classified as “type 2” mitochondria (Fig. 3a). Although treatment with only 20 mg·kg−1·d−1 PQQ decreased the number of “type 2” mitochondria (6.23 ± 0.33 per 100 μm2) compared with that in the rotenone group (9.77 ± 0.70 per 100 μm2), treatment with PQQ at all doses (Rot + 0.8 mg·kg−1·d−1 PQQ, 27.86 ± 1.02; Rot + 4 mg·kg−1·d−1 PQQ, 32.68 ± 1.80; Rot + 20 mg·kg−1·d−1 PQQ, 35.55 ± 1.12 per 100 μm2) restored the number of “type 1” mitochondria in rotenone-induced PD mice (23.32 ± 1.60 per 100 μm2) and prevented morphological changes (Fig. 3b). The average numbers of “type 1” and “type 2” mitochondria per 100 μm2 in the control group were 42.68 ± 3.25 and 2.77 ± 0.18, respectively.

Fig. 3: PQQ affected the content and morphology of mitochondria in midbrains of PD mice.
figure 3

a Representative TEM images of mitochondria in the midbrains of PD mice with different treatments. Arrows indicate “type 1” mitochondria. Asterisks indicate “type 2” mitochondria. Scale bar = 1 μm. b Quantitative analysis of the average number of “type 1” and “type 2” mitochondria per 100 square micrometers (μm2). *P < 0.05 and **P < 0.01 vs the Rot group. ##P < 0.01 vs the control group (n = 20).

As PGC-1α and transcription factor A (TFAM) are key activators of mitochondrial gene transcription, the mRNA and protein levels of PGC-1α and TFAM in the midbrain and cortex were determined after PQQ treatment. Rotenone-induced injury decreased the mRNA level of TFAM in the cortex and midbrain to 66.18% ± 2.57% and 71.20% ± 4.70% of that in the control group, respectively (Fig. 4a, b). High-dose PQQ treatment upregulated the expression of TFAM in the cortex and midbrain to 128.5% ± 11.18% and 101.7% ± 10.79% of that in the control group, respectively (Fig. 4a, b). Although there was no significant change in the mRNA expression of PGC-1α in either the midbrain or cortex (data not shown), the protein level of PGC-1α was increased by medium and high dosages of PQQ to 153.50% ± 5.25% and 160.20% ± 8.57% compared with that in the control group in the cortex and midbrain, respectively (Fig. 4c–f). The downregulation of TFAM protein expression by rotenone-induced injury (cortex, 76.19% ± 2.27%; midbrain, 71.03% ± 3.14%) was also rescued by treatment with PQQ at a high dosage (cortex, 121.5% ± 10.91%; midbrain, 92.08% ± 4.18%) (Fig. 4c, d, g and h). These data indicated that PQQ might have promoted mitochondrial biogenesis in the brains of rotenone-injured PD mice by upregulating PGC-1α and TFAM expression.

Fig. 4: PQQ regulated the expression of mitochondrial biogenesis-related genes and proteins in PD mice.
figure 4

a, b The mRNA expression of TFAM in the cortex (a) and midbrain (b) was measured. All data are expressed as the ratio compared with the control group. **P < 0.01 vs the Rot group. #P < 0.05 and ##P < 0.01 vs the control group (n = 3). c, d Representative Western blotting images of PGC-1α and TFAM protein expression in the cortex and midbrain of PD mice with different treatments. β-Actin served as a loading control. e, f The protein expression of PGC-1α in the cortex and midbrain is expressed as the ratio compared with the control group. g, h The protein expression of TFAM in the cortex and midbrain is expressed as the ratio compared with the control group. *P < 0.05 and **P < 0.01 vs the Rot group. #P < 0.05 and ##P < 0.01 vs the control group (n = 3).

Effect of PQQ on mitochondrial number in rotenone-injured SH-SY5Y cells

Our previous research suggested that 100 μM PQQ pretreatment for 24 h could inhibit the apoptosis of rotenone-injured SH-SY5Y cells and increase mitochondrial DNA (mtDNA) content. Here, we evaluated the effect of PQQ on mitochondria number by quantifying the intensity of MitoTracker® Green staining and the number of mitochondria in TEM images. MitoTracker® Green is a cell-permeable green fluorescent probe for mitochondrial labeling in live cells that localizes to mitochondria regardless of the mitochondrial membrane potential. Images following MitoTracker® Green staining showed that rotenone-induced injury dramatically decreased the fluorescence intensity (control, 0.0391 ± 0.0039; Rot, 0.0154 ± 0.0032), while an increase in fluorescence intensity was observed by treatment with PQQ at a dose of 100 μM (0.0342 ± 0.0052) (Fig. 5a). Pathological mitochondria were observed with rotenone-induced injury in SH-SY5Y cells, but this effect was mostly eliminated by PQQ treatment (Fig. 5b). In addition, decreases in the numbers of total (control, 57.30 ± 6.16 per 100 μm2; Rot, 25.60 ± 1.33 per 100 μm2) and “type 1” mitochondria (control, 54 ± 5.97 per 100 μm2; Rot, 12.70 ± 1.56 per 100 μm2) were also inhibited by 10 μM PQQ (total, 39.90 ± 1.45 per 100 μm2; “type 1”, 31.90 ± 1.22 per 100 μm2) and 100 μM PQQ (total, 44.50 ± 2.36 per 100 μm2; “type 1”, 38.20 ± 2.19 per 100 μm2) (Fig. 5c). The above data indicated that PQQ could rescue the decreased mitochondrial number induced by rotenone injury.

Fig. 5: PQQ promoted mitochondrial biogenesis in rotenone-injured SH-SY5Y cells.
figure 5

SH-SY5Y cells were plated in a 24-well plate and treated with rotenone or rotenone with different concentrations of PQQ. a Average fluorescence intensity per pixel in cultured SH-SY5Y cells after different treatments as indicated was measured by MitoTracker® Green staining. *P < 0.05 vs the Rot group and ##P < 0.01 vs the control group (no treatment) (n = 10). b Representative TEM images of mitochondria in SH-SY5Y cells with different treatments. Arrows indicate “type 1” mitochondria. Asterisks indicate “type 2” mitochondria. Scale bar = 1 μm. c Quantitative analysis of the average numbers of “type 1” and “type 2” mitochondria per 100 μm2. **P < 0.01 vs the Rot group. #P < 0.05, ##P < 0.01 vs the control group (n = 20).

Effect of PQQ on protein phosphorylation in rotenone-injured SH-SY5Y cells

To further investigate the signaling pathways involved in the neuroprotective effect of PQQ, a rotenone-induced SH-SY5Y cell model was applied, and an antibody-based proteomic analysis was performed. Our previous studies suggested that 100 μM PQQ could restore mitochondrial dysfunction and inhibit cell apoptosis in rotenone-injured SH-SY5Y cells and midbrain dopaminergic neurons [11]. In this study, rotenone-injured SH-SY5Y cells were treated with or without 100 μM PQQ for 30 min, harvested, digested, and immunoprecipitated with PTMScan Direct reagents. Overall, 394 phosphosites and 260 phosphopeptides on 231 phosphoproteins were detected (Fig. 6a), covering 10 signaling pathways, including the AMPK signaling, adherens junction and Jak-STAT pathways (Fig. 6b). Thirty-three phosphosites displayed a statistically significant difference in expression (±2-fold, P < 0.05) between rotenone-injured cells and rotenone-injured cells treated with PQQ, and among these phosphosites, 4 phosphosites were upregulated, and 29 phosphosites were downregulated. A full list of the phosphoproteins identified as differentially expressed between the two groups is available in Table 2, and a Venn diagram is shown in Fig. 6c. Gene ontology (GO) analysis revealed that these proteins are mainly involved in the biological processes related to the regulation of neuronal cell death and apoptosis, wound healing, peptide biosynthesis, and metabolic processes (Fig. 6d). To further explore the impact of these differentially expressed phosphoproteins and discover internal connections between them, Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analysis was performed, and the data indicated that the neurotrophin pathway, axon guidance and cAMP pathway were affected (Fig. 6e). A protein–protein interaction (PPI) network was also generated. The network consisted of a number of interconnected differentially expressed phosphoproteins, including 5′-AMP-activated protein kinase catalytic subunit alpha-1 (PRKAA1, also known as AMPKα1), Histone H2B (HIST1H2BI), focal adhesion kinase 1 (PTK2), Abl interactor 2 (ABI2), vesicle-associated membrane protein 8 (VAMP8), phosphoinositide phospholipase C (PLCG1), Kin of IRRE-like protein 1 (KIRREL), putative uncharacterized protein (NCK2), Rho-related GTP-binding protein (RhoC), and mammalian target of rapamycin (mTOR) (Fig. 6f).

Fig. 6: PQQ changed the proteomic phosphorylation profile of rotenone-injured SH-SY5Y cells.
figure 6

SH-SY5Y cells were divided into four groups: the control group (no rotenone or PQQ), PQQ group (100 μM PQQ), Rotenone group (100 μM rotenone), PQQ + Rotenone group (100 μM PQQ + 100 μM rotenone). After 30 min of incubation, the cells were harvested for PTMScan multipathway analysis. a The total numbers of phosphosites, phosphopeptides and phosphoproteins detected in the experiment. b The number of proteins detected in different signal pathways. The AMPK signaling pathway was the most affected pathway. c Venn diagram showing the number of overlapping phosphopeptide sites in the control group (C), Rotenone group (R), PQQ group (P), and PQQ + Rotenone group (R+P). d GO enrichment analysis of phosphorylated proteins. The top 20 most differentially expressed phosphoproteins between the Rotenone and PQQ + Rotenone groups were classified depending on biological process (BP) and molecular function (MF). e KEGG enrichment analysis showing the pathways most significantly affected by PQQ treatment. f Protein–protein interaction (PPI) network analysis of differentially expressed phosphoproteins. Upregulated proteins are highlighted in red, and downregulated proteins are highlighted in green. PRKAA1, 5′-AMP-activated protein kinase catalytic subunit alpha-1; HIST1H2BI, Histone H2B; PTK2, focal adhesion kinase 1; ABI2, Abl interactor 2; VAMP8, vesicle-associated membrane protein 8; PLCG1, phosphoinositide phospholipase C; KIRREL, Kin of IRRE-like protein 1; NCK2, putative uncharacterized protein; RhoC, Rho-related GTP-binding protein; mTOR, mammalian target of rapamycin.

Table 2 Differentially expressed phosphoproteins between Rot and Rot + PQQ group.

AMPK phosphorylation was upregulated by PQQ in rotenone-injured SH-SY5Y cells and PD mice

PTMScan analysis showed that AMPK phosphorylation was significantly upregulated by PQQ in rotenone-injured cells. We used Western blotting to validate the results of the proteomic approach. The expression of phosphorylated AMPK was not affected by PQQ treatment in normal cells. AMPK phosphorylation was upregulated by 100 μM PQQ (1.80 ± 0.08) compared with that in the rotenone group (1.33 ± 0.05) (Fig. 7a, b), confirming that the AMPK pathway was activated by PQQ in rotenone-injured cells.

Fig. 7: Western blotting validation of PQQ-induced activation of AMPK signaling in rotenone-injured SH-SY5Y cells and rotenone-injured PD mice.
figure 7

a SH-SY5Y cells were treated with rotenone or rotenone with different concentrations of PQQ for 30 min, after which both phosphorylated AMPKα and total AMPKα were detected by Western blotting analysis. β-Actin served as the internal control. b Quantification of the relative expression of phosphorylated AMPKα compared with that in the control group. The ratio of phosphorylated AMPK (p-AMPKα) to total AMPKα in the control group was set as “1”. *P < 0.05 vs the Rot group (n = 3). c Rotenone (3 mg·kg−1·d−1) was intraperitoneally administered once daily for 3 weeks. PQQ at a high dosage (20 mg·kg−1·d−1) was intraperitoneally administered immediately after rotenone injection. The control group was injected with normal saline containing 2% DMSO. The midbrains were harvested at the end of weeks 1, 2, and 3 for Western blotting analysis. β-Actin served as the internal control. d Quantification of the relative expression of p-AMPKα. The ratio of p-AMPKα to total AMPKα in the control group before drug administration was set as “1”. *P < 0.05 vs the Rot group (n = 4).

Then, we observed AMPK activation in PD mice. In the control group, phosphorylated AMPK levels did not change much over the drug administration period (1.00 ± 0.21 before injection and 1.10 ± 0.16, 1.03 ± 0.20, and 0.89 ± 0.10 at the end of weeks 1, 2, and 3 after the first injection, respectively). A slight increase in phosphorylated AMPK in the rotenone group (1.46 ± 0.18) at 1 week was detected. However, the activation of AMPK decreased to 1.23 ± 0.15 and 0.95 ± 0.18 after 2 and 3 weeks, respectively. PQQ at a high dosage significantly promoted the activation of AMPK by phosphorylation compared with that in the rotenone group over the whole process (1.93 ± 0.28, 1.60 ± 0.20, and 1.43 ± 0.24 at weeks 1, 2, and 3, respectively) (Fig. 7c, d). In vitro and in vivo data suggested that PQQ could increase AMPK activation in both rotenone-injured cells and rotenone-injured PD mice.

AMPK activation is required for the protective effect of PQQ in rotenone-injured SH-SY5Y cells

Dorsomorphin (compound C) 2HCl is a potent, reversible, selective AMPK inhibitor. We first demonstrated by Western blotting that dorsomorphin at 4 μM significantly inhibited the expression of PGC-1α, a downstream molecule of AMPK, in SH-SY5Y cells after rotenone injury with or without PQQ treatment (control, 1.00 ± 0.09; dorsomorphin, 1.00 ± 0.05; Rot, 0.41 ± 0.03; Rot+dorsomorphin, 0.54 ± 0.04; Rot+PQQ, 0.80 ± 0.04; Rot+PQQ + dorsomorphin, 0.57 ± 0.04) (Fig. 8a, b). The increased cell viability in rotenone-injured SH-SY5Y cells treated with PQQ was inhibited by dorsomorphin pretreatment (control, 0.92 ± 0.03; dorsomorphin, 0.86 ± 0.02; Rot, 0.58 ± 0.02; Rot+dorsomorphin, 0.56 ± 0.01; Rot+PQQ, 0.76 ± 0.02; Rot+PQQ + dorsomorphin, 0.60 ± 0.02) (Fig. 8c), suggesting that the inactivation of AMPK could reduce the neuroprotective effect of PQQ.

Fig. 8: Inhibition of AMPK by dorsomorphin (compound C) 2HCl eliminated the protective effect of PQQ in rotenone-injured SH-SY5Y cells.
figure 8

Dorsomorphin at a final concentration of 4 µM was added to SH-SY5Y cells 30 min prior to rotenone and PQQ treatment. Cellular proteins were extracted after PQQ treatment for 30 min and subjected to Western blotting analysis, and cell viability was measured after PQQ treatment for 24 h. a Representative images of PGC-1α detected by Western blotting analysis. β-Actin served as the internal control. b Quantification of the relative expression of PGC-1α. **P < 0.01 vs the Rot group, ##P < 0.01 vs the control group, $P < 0.05 vs the Rot+PQQ group (n = 3). c Cell viability was measured by CCK-8 assay, and the absorbance (optical density, OD) at 450 nm was measured by spectrophotometry. **P < 0.01 vs the Rot group, ##P < 0.01 vs the control group, $$P < 0.01 vs the Rot + PQQ group (n = 10).

The average fluorescence intensity of MitoTracker® Green staining showed that inhibition of AMPK activation by dorsomorphin attenuated the increase in mitochondrial number due to PQQ pretreatment in rotenone-injured SH-SY5Y cells (control, 0.0400 ± 0.0037; dorsomorphin, 0.0360 ± 0.0025; Rot, 0.0180 ± 0.0023; Rot+dorsomorphin, 0.0148 ± 0.0015; Rot+PQQ, 0.0300 ± 0.0020; Rot+PQQ + dorsomorphin, 0.0192 ± 0.0203) (Fig. 9b, c). The mtDNA content quantified by qRT-PCR also indicated that AMPK inhibition reduced the mitochondrial mass (0.46 ± 0.04) compared with that in the PQQ treatment group (0.68 ± 0.04) (Fig. 9d).

Fig. 9: Inhibition of AMPK by dorsomorphin (compound C) 2HCl attenuated mitochondrial biogenesis induced by PQQ in rotenone-injured SH-SY5Y cells.
figure 9

Dorsomorphin at a final concentration of 4 µM was added to SH-SY5Y cells 30 min prior to rotenone and PQQ treatment. a MitoTracker® Green staining of SH-SY5Y cells was performed 24 h after PQQ treatment (scale bar = 25 μm). b Average fluorescence intensity per pixel in cultured SH-SY5Y cells after different treatments as indicated was measured by MitoTracker® Green staining. *P < 0.05 vs the Rot group, ##P < 0.01 vs the control group, $P < 0.05 vs the Rot + PQQ group (n = 10). c Mitochondrial DNA content was measured by real-time PCR of cultured SH-SY5Y cells after the indicated treatments. **P < 0.01 vs the Rot group. ##P < 0.01 vs the control group, $$P < 0.01 vs the Rot + PQQ group (n = 10). d Representative TEM images of SH-SY5Y cells treated with dorsomorphin. Arrows indicate “type 1” mitochondria. Asterisks indicate “type 2” mitochondria. Scale bar = 2 μm. e Quantitative analysis of the average number of “type 1” and “type 2” mitochondria per 100 μm2. **P < 0.01 vs the Rot group. #P < 0.05, ##P < 0.01 vs the control group, $$P < 0.01 vs the Rot + PQQ group (n = 20).

Moreover, TEM observation showed that the inactivation of AMPK significantly decreased the number of total and “type 1” mitochondria (total, 23.90 ± 1.06; “type 1”, 16.00 ± 0.82) in SH-SY5Y cells compared with those in the PQQ treatment group (total, 35.20 ± 2.12; “type 1”, 28.70 ± 1.88) (Fig. 9e, f). Together, these results demonstrated that activation of the AMPK signaling pathway might be involved in the regulation of mitochondrial biogenesis by PQQ in rotenone-injured SH-SY5Y cells.

Discussion

The pathogenesis of PD is multifactorial and associated with mitochondrial dysfunction, oxidative stress, apoptosis, and inflammation [2]. Although the exact mechanisms underlying PD progression have not been well elucidated, natural products and bioactive compounds with multiple targets in the central nervous system seem to be more effective and safer for PD treatment than single-target compounds [22]. PQQ is a redox cofactor found in soil, foods such as kiwifruit, and human breast milk [23]. Recent research indicates that PQQ protects against oxidative stress damage, apoptosis, and mitochondrial dysfunction [24, 25]. In the present study, we reported that PQQ protected rotenone-induced PD mice and rotenone-injured SH-SY5Y cells by promoting mitochondrial biogenesis. The AMPK signaling pathway was activated by PQQ, which was required for the protective effects of PQQ in PD models.

Bioactive compounds with the ability to stimulate mitochondrial biogenesis are thought to improve health by promoting energy utilization, protecting against reactive oxygen species and inhibiting apoptosis [26]. PQQ was reported to protect cells from autophagy-dependent apoptosis induced by doxorubicin via the lysosomal-mitochondrial axis [27]. PQQ could also exert protective effects against oxidative stress damage and apoptosis in a diabetic nephropathy model in vitro. These protective effects were associated with increased antioxidants and inhibited ROS production, which required activation of the PI3K/Akt signaling pathway [28]. Our previous study indicated that PQQ could prevent mitochondrial dysfunction in PD rats in which rotenone was injected into the medial forebrain bundle (MFB) [8]. However, this PD model established by the MFB injection of rotenone is not completely consistent with the slow progression of PD. In this study, we established a mouse PD model by intraperitoneal injection of rotenone and then investigated the neuroprotective effect and underlying mechanism of PQQ. As expected, intraperitoneal rotenone injection for 21 consecutive days induced motor function deficits and neuronal loss in the SNpc [29]. Bradykinesia and locomotor activity dysfunction were observed in a rat PD model induced by intraperitoneal injection of rotenone [30]. Here, we demonstrated that PQQ could prevent motor deficits and reduce the loss of dopaminergic neurons in the SNpc in a dose-dependent manner. As neurological dysfunctions in PD patients and in animal models are related to dopaminergic neuronal loss in the SNpc and the depletion of dopamine in the striatum, our data indicated that PQQ exerted a protective effect in PD mice by inhibiting neuronal loss, thus reducing locomotor dysfunctions.

Mitochondrial function is primarily maintained by mitochondrial quality control mechanisms, including mitochondrial biogenesis, mitochondrial fission and fusion, and mitochondrial autophagy [31]. Mitochondrial biogenesis is thought to be regulated by the PGC-1α network, while the Nrf2/ARE signaling cascade is involved in mitochondrial biogenesis [32, 33]. PQQ supplementation could upregulate the molecular signaling responses indicative of mitochondrial biogenesis within skeletal muscle [34]. In cultured mouse hepatocytes, PQQ exposure could promote the phosphorylation of CREB and the expression of PGC-1α, thus increasing the mtDNA content [26]. We also found that PQQ could maintain mitochondrial physiological functions and promote the expression and translocation of Nrf2 in glutamate-damaged hippocampal neurons [35], suggesting that PQQ exerts a neuroprotective effect by promoting mitochondrial biogenesis. In this study, we observed that the number of mitochondria with an integral structural was significantly increased with PQQ treatment both in vivo and in vitro, suggesting that PQQ might promote mitochondrial biogenesis in rotenone-induced PD models. As mitochondrial turnover by autophagy or proteolysis is important to eliminate defective mitochondria to maintain cellular homeostasis [36], mitochondrial degradation should also be considered in future studies.

As a transcriptional coactivator, PGC-1α plays a significant role in promoting mitochondrial biogenesis [37]. PGC-1α acts as a transcription factor for mitochondrial biogenesis by transactivating Nrf1/2 and increasing the expression of TFAM [38, 39]. TFAM, a mtDNA-binding protein that plays a central role in the mtDNA stress-induced inflammatory response, is necessary for mitochondrial genome maintenance [40]. Decreases in PGC-1α and its downstream genes were detected in the postmortem brains of PD patients [41]. PGC-1α-knockout mice were found to be more sensitive to MPTP damage than control mice [42], and deletion of the Drosophila PGC-1α ortholog spargel in flies caused PD-like symptoms [43]. In contrast, activation of PGC-1α was sufficient to rescue the disease phenotypes of Parkin and LRRK2 genetic fly models of PD [43]. Activation or overexpression of PGC-1α caused an increase in the expression of nuclear-encoded subunits of the mitochondrial respiratory chain and prevented the degeneration of dopaminergic neurons in cellular or animal models induced by mutant α-synuclein, rotenone, or MPTP [42]. Moreover, increased mitochondrial biogenesis caused by the stabilization of endogenous PGC-1α exerted a neuroprotective against mitochondrial insults [44]. This evidence suggests that targeting PGC-1α might be an effective strategy for neuroprotection in PD. Although we did not detect a significant change in PGC-1α mRNA in the cortex or midbrain of PD mice, which is not consistent with the data in SH-SY5Y cells [8], we found that PQQ had a significant inhibitory effect on the decrease in PGC-1α and TFAM protein expression caused by rotenone injury in vivo. The effect of PQQ on protein expression, combined with the increase in mitochondrial number, indicated that PQQ might promote mitochondrial biogenesis by upregulating PGC-1α and TFAM.

To further analyze the signaling mechanisms involved in the neuroprotective effect of PQQ, rotenone-injured SH-SY5Y cells and PTMScan analysis coupled with LC-MS/MS methods were applied [45]. PTMs have emerged as important modulators of some important proteins in PD pathogenesis [14, 46]. Activation of LRRK2 by phosphorylation is thought to be one of the causes of PD, and an LRRK2 kinase inhibitor could protect cells against rotenone-induced noxious effects [15]. Among the differentially phosphorylated proteins upon PQQ treatment in our cell injury model, AMPK is a serine/threonine kinase that plays an important role in cellular energy homeostasis, mitochondrial biogenesis, and autophagy [47]. Induction of adaptive responses via the AMPK signaling pathway was shown to be important for the survival of human cells under oxidative stress induced by mitochondrial dysfunction [48]. AMPK was expected to be activated by rotenone injury, as previously reported. However, AMPK activation was observed after incubation with rotenone for 24 h [49, 50]. In our experiment, rotenone was applied for only 30 min, and only weak activation of AMPK was observed in the rotenone group. In the PD mice, AMPK activation was observed after 1 week, and phosphorylated AMPK levels then decreased to normal as dopaminergic neuronal loss occurred. Consistent with the PTMScan data, we confirmed that PQQ could activate AMPK phosphorylation in both rotenone-injured cells and rotenone-injured PD mice.

Activation of the AMPK signaling pathway was reported to enhance mitochondrial biogenesis and energy bioavailability by regulating PGC-1α and TFAM [51]. Our previous study proved that the expression of PGC-1α and TFAM in rotenone-injured SH-SY5Y cells could be upregulated by PQQ [8]. Here, we also observed an increase in PGC-1α and TFAM in PD mice. Thus, we deduce that the upregulation of these transcription factors is related to activation of the AMPK signaling pathway. Activation of AMPK enhanced the expression of PGC-1α by p38MAPK signaling in cancer cells to induce mitochondrial biogenesis [52]. In skeletal muscle cells, AMPK upregulated PGC-1α expression by increasing PGC-1α promoter activity or the direct phosphorylation of PGC-1α [53, 54]. However, whether AMPK can upregulate the expression of PGC-1α in neuronal cells through the above pathways has not been confirmed.

Our data showed that inhibition of AMPK by dorsomorphin reduced the upregulation of PGC-1α induced by PQQ and eliminated the protective effect of PQQ in rotenone-injured SH-SY5Y cells. Mitochondria number and the mtDNA content were also regulated by AMPK inhibition, suggesting that the activation of AMPK was required for the regulation of mitochondrial biogenesis by PQQ. Dorsomorphin is a potent, reversible, selective AMPK inhibitor [55, 56]. Inhibition of AMPK activity by dorsomorphin was reported to almost completely inhibit autophagic proteolysis in HT-29 cells [57], while autophagy induction might have exerted protective effects in rotenone-injured cells [29, 58]. Whether PQQ can induce autophagy by AMPK activation and then display neuroprotective effects needs further investigation.

Conclusion

Taken together, the results of this study demonstrated that PQQ could improve motor functions, reduce the loss of midbrain dopaminergic neurons, and promote mitochondrial biogenesis in rotenone-induced PD mice, showing a neuroprotective effect. Activation of the AMPK signaling pathway by PQQ might be involved in the regulation of mitochondrial biogenesis and was responsible for the protective effect of PQQ in rotenone-injured SH-SY5Y cells.