The pesticide chlorpyrifos increases the risk of Parkinson’s disease

To assess CPF and PD associations, we used the Parkinson’s Environment and Genes (PEG) study (n = 829 PD patients; n = 824 controls). The PEG study is a population-based case-control studies conducted in three agricultural counties in central California (Kern, Fresno, and Tulare) [14, 15]. Participants were recruited in two waves: Wave 1 (PEG1): 2000–07, n = 357 patients, n = 400 population-based controls; and Wave 2 (PEG2): 2009–15, n = 472 patients, n = 424 population-based controls. Patients were enrolled early in the disease course [mean PD duration at baseline, 3.0 years (SD = 2.6)] and all were seen by University of California, Los Angeles (UCLA) movement disorder specialists for in-person neurological examinations, many on multiple occasions, and confirmed as having probable idiopathic PD [16, 17]. As shown in Table 1, PD patients were on average slightly older than controls and a higher proportion were men, had European ancestry, and were never smokers.

We estimated ambient exposure due to living or working near agricultural CPF application, using pesticide use report (PUR) pesticide application data within a geographical information system (GIS)-based model [18]. Since 1972, California law mandates the recording of commercial pesticide use in a database maintained by the California Department of Pesticide Regulation (CA-DPR) that includes all commercial agricultural pesticide use by pest control operators and all restricted pesticide use until 1989, and afterwards (1990–current), all commercial agricultural pesticide use. This database records the location of applications, which was linked to the Public Land Survey System (PLSS), and the poundage, type of crop, and acreage a pesticide has been applied on, as well as the method and date of application. We combined the PUR with maps of land use and crop cover, providing a digital representation of historical land use, to determine pesticide applications at specific agricultural sites [19]. PEG participants provided lifetime residential and workplace address information, which we geocoded in a multi-step process [20]. As shown in Fig. 1, the scatter plot of the data is accompanied by smoothed trend lines based on local regressions (Fig. 1B-C). We determined the pounds of CPF applied per acre within a 500-meter buffer of the latitude and longitude representing each residential and workplace address per year since 1974, weighting the total poundage by the proportion of acreage treated (lbs/acre). For our study participants, this resulted in 12,904 annual records for residential and 8,968 for occupational site CPF exposure. After we identified and removed several extreme outliers (values >99th percentile of the distribution), the resulting data ranged from 0 to 108.44 lbs/acre. We also estimated exposure to paraquat, diazinon, and glyphosate in a similar manner for co-exposure adjustment.

Six-month-old, 24 male wild-type mice on C57BL/6 background weighing 29–33 g were purchased from the Jackson Laboratory, Sacramento, California, USA. The mice were housed in a vivarium with 12/12 hours of light and dark cycles. UCLA Animal Research Committee approved all experiments using animals in accordance with the US National Institutes of Health guidelines.

Sixteen 26-week-old male mice were exposed to aerosolized CPF (Sigma #45395) and another 16 were exposed to room air passed through the control vehicle containing 2% ethanol in identical chambers using custom-built whole-body chamber systems (CH Technologies, Westwood, NJ). The chamber holds up to 20 mice and is connected to a Blaustein Jet Atomizer, which produces aerosols that are tightly controlled with a Lab Flow C Control box. Aerosols were generated from solutions containing CPF (0.2 mg/mL) in 75% ETOH at a flow rate into the atomizer of 16 mL/hr while aerosol generator air flow was 1.5 L/min and dilution air flow was 4.5 L/min. The concentration of CPF initially in the chamber was 76.63 µg/m³ and was gradually increased to 300 µg/m³. Mice were exposed 5 days/wk. for 6 h./day for 11 weeks with total exposures of 0.65–2.9 mg/m³/day.

Behavioral testing was performed prior to CPF exposures and 3 days after the final day of exposure to wash out any residual effects of the pesticide and vehicle.

Rotarod Test: Rotarod testing was carried out as previously described [20, 21]. Mice were trained to run on the rotarod (Bioseb, Park, FL USA) with four 5-minute trials at 12 rotations per minute (rpm) separated by five minutes. One hour later, mice were placed on the rotarod at 4 (rpm) and the speed was linearly accelerated to 40 rpm over 300 s. The latency to fall and speed at which the mice fell were recorded. The mean latency to fall was calculated from two consecutive trials in each mouse.

Wire Hang Test: Wire hang testing was carried out according to a modified protocol as previously described [20, 22]. Briefly, the mice were placed on the top of a wire mesh (wire diameter 0.047” with 0.45” openings) connected to a stepper motor controlled by an Arduino microcontroller, with speed set to 15 rpm. At the start of the trial, the wire mesh was sequentially rotated 36 degrees in one direction then the other, causing the mice to grip the wires, and then flipped 180 degrees so mice were suspended upside down. Latency to fall was measured using an infrared sensor placed below the wire mesh. If mice did not fall within 15 min, the trial was terminated.

Open Field Test: Open field testing was performed using a custom-made acrylic square arena (16” x 16” x 12”) with a grey floor that was evenly illuminated to 100–200 lux [20, 23]. During the trial, the mice were allowed to explore freely for 10 minutes while being videotaped. A deep neural network was trained on behavior videos to identify arena boundaries and anatomical markers on mice (nose, ears, and tailbase) using DeepLabCut [20, 24]. The trained model was then used to automatically extract marker positions during each video frame. Mouse position was calculated as the centroid of the polygon between the mouse ear and tailbase markers, and custom-written code in MATLAB was used to calculate mean velocity, distance traveled, and time in the center of the arena (defined as a square 20% of the arena diameter away from each wall).

Commercially available CPF was acquired from Chem Service Inc. Diethyl-d10 (CPF-d10) is obtained from CDN Isotopes. Methanol and acetonitrile optima LC/MS grade were purchased from Thermo Fisher Scientific. Standards, controls, and brain tissue samples were prepared in 2 ml bead-beating tubes in duplicates. A five-concentration point (100, 200, 400, 800, 1600 ng/50 mg) standard curve and two control samples at low (200 ng/50 mg) and high (1600 ng/50 mg) concentrations were utilized for compound quantification. Standards, controls, and brain tissue samples stored at -80 °C were left to thaw at ambient temperature. To each tube were added six 1.4 mm and three 2.8 mm ceramic beads, 400 ng diethyl-d10 stock solution as an internal standard (IS) prepared in acetonitrile, and 500 mL of a 4:1 mixture of acetonitrile/ deionized water for extraction. The samples were homogenized in the bead-beater for 30 s and centrifuged at 16,000 x g for five minutes. The supernatant was transferred to new, respectively labeled 1.5 ml microcentrifuge tubes, and samples were dried down in a low-speed vacuum concentrator. The samples were then reconstituted in 100 µl of a methanol/water (v/v 70/30) mixture, vortexed thoroughly, and centrifuged at 16,000 x g for five minutes. The supernatant was transferred to HPLC vials, from which 15 µl was injected into the mass spectrometer system for analysis. Analysis of the analyte concentration was done at the UCLA Pasarow Mass Spectrometry Lab. A targeted LC-MS/MS multiple reaction monitoring (MRM) acquisition method was developed and optimized on a 6460 Agilent Technologies triple quadrupole mass spectrometer. The mass spectrometer was coupled to a 1290 Infinity HPLC system (Agilent Technologies) through a Scherzo SM C-18 analytical column (3 μm 50 × 2 mm UP). For compound elution, a mixture of solvent A (water/formic acid v/v 100/0.1) and solvent B (acetonitrile/formic acid v/v 100/0.1) was used as a mobile phase combined with a linear gradient (min/%B: 0/0, 2/5, 6/100, 7/100, 8/5, 10/5). For the MRM method, the transition from the precursor ion m/z 349.9 to the production m/z 97 isolated for CPF was monitored in positive ion mode at a specific LC retention time. The standard curve was made by plotting the known amount of analyte per standard vs. the ratio of measured chromatographic peak areas corresponding to the analyte over that of the IS (analyte/IS). The trendline equation was used to calculate the absolute concentrations of the analyte in brain tissue.

After perfusion and fixation of six CPF-exposed and six control mice, brains were embedded in paraffin blocks, cut into 6 mm sections and mounted on glass slides as previously described with some modifications [20, 25]. Briefly, samples containing slides were deparaffinized and rehydrated, followed by immersion in 88% formic acid for 5 min to enhance antigen detection. Then the sections were treated with 5% hydrogen peroxide (Sigma#H1009) in methanol (Fisher Scientific#A454-4) for 30 min to quench endogenous peroxidases. The sections were then blocked in 0.1 M Tris with 2% fetal bovine serum (Tris/FBS) for 5 min. Both TH primary and secondary antibodies were diluted in Tris/FBS. Samples were then incubated with primary antibody (Abcam#AB76442) overnight at 4 °C. Following washing, sections were incubated for 1 h with species-specific biotinylated secondary antibody (VectorLabs#BA-9010) and then for 1 h with VECTASTAIN^® Elite^® ABC-HRP Kits (VectorLabs#PK-6100). After rinsing with 0.1 M Tris, the slides were incubated for 4 min with DAB reagent (VectorLabs#SK-4105). In subsequent steps, slides were rinsed and washed, counterstained in Hematoxylin (Fisher Scientific#6765001), then dehydrated and cover slipped. Slides were scanned into digital format on a Lamina scanner (Perkin Elmer) at 20x magnification. Digitized slides were then used for quantitative pathology. All TH positive neurons in the midbrain (SNpc and VTA) were counted in a blinded manner.

In a separate group of 6 pairs of mice, brains were quickly removed after perfusion with PBS and euthanasia and cut in half sagittally. Half of the brains were quickly frozen for biochemical studies, and the other half were fixed in 4% paraformaldehyde for at least 24 h. Following dehydration in 30% sucrose, they were embedded in OCT (Sakura Finetek#4583) and 40 µM sections were collected. The sections were blocked with 3% FBS and 3% bovine serum albumin (Sigma#B-4287) and incubated overnight at 4 °C in primary antibodies, washed with PBS and incubated in secondary antibodies overnight at 4 °C before cover slipping. Primary antibodies used were anti-TH (Abcam#ab76442; 1:500 dilution), anti-pS129 (Abcam#ab51253; 1:1000 dilution), anti-ubiquitin (Santa Cruz Biotechnology#sc-271289; 1:500 dilution), anti-IL-1β (Thermo Fisher Scientific#P420B; 1:250 dilution), anti-Lc3 (Cell Signaling Technology#83506; 1:500 dilution), anti-Lamp2a (Abcam#ab18528; 1:250 dilution) and anti-Iba1 (Synaptic Systems#234 308 Gp311H9; 1:500 dilution). The secondary antibodies used were Goat anti-Mouse IgG (H + L) Cross-Adsorbed, Alexa Fluor™ 488 (Thermo Fisher Scientific#A-11001); 1:1000 dilution, Goat anti-Rabbit IgG (H + L) Cross-Adsorbed, Alexa Fluor™ 488 (Thermo Fisher Scientific#A-11008); 1:1000 dilution, Donkey anti-Chicken IgY (H + L) Cross-Adsorbed, Alexa Fluor™ 568 (Thermo Fisher Scientific#A78950); 1:1000 dilution, Goat anti-Mouse IgG (H + L) Cross-Adsorbed, Alexa Fluor™ 405 (Thermo Fisher Scientific#A-31553); 1:1000 dilution, Goat anti-Rabbit IgG (H + L) Cross-Adsorbed, Alexa Fluor™ 647; 1:1000 dilution, and Goat anti-Guinea Pig IgG (H + L) Highly Cross-Adsorbed, Alexa Fluor™ 647, 1:1000 dilution]. The sections were imaged using a Zeiss Z.1 Light dual-sided illumination sheet fluorescence microscope, equipped with ZEN software. Quantification of pS129-syn, ubiquitin, LC3, and Lamp2a were performed blindly by identifying TH neurons in the red channel and measuring fluorescence in the green channel.

Images of mouse microglia were obtained using a 40x Z-stack confocal microscopy were processed and refined using ImageJ for further analysis. The “far red channel” images were used to define regions of interest (ROIs) for the corresponding microglia channel. Multilayer confocal images were compressed to generate projected microglia images. These images were then converted to binary and further processed with a median filter (radius = 2.0 pixels). To restore continuity in microglia structures fragmented by binarization, dilate and close operations were applied. ROIs from the far-red channel were applied to the microglia channel to establish brain region boundaries. Finally, individual microglia cells were analyzed using ImageJ’s “wand tracing tool,” measuring area, perimeter, and roundness within the ROIs.

Extractions were performed using a modified procedure of Henderson et al. [20, 26]. Briefly, mouse brains (half) were homogenized in 9 volumes of high salt (HS) 1% Triton X buffer with protease and phosphatase inhibitors using Dounce homogenizers for 20 strokes. The samples were centrifuged at 100,000 x g for 30 min and supernatants were stored at -80. The pellets were homogenized in 9 volumes HS-Triton X buffer with 30% sucrose and centrifuged at 100,000 x g for 30 min at 4 °C. Supernatants were again stored at -80 °C. The pellets were homogenized in 9 volumes 1% Sarkosyl HS buffer, shaken for 30 min at room temperature (RT), and spun at 100,000 x g for 30 min. The resultant pellet was washed in 1.4 mL PBS without Ca⁺⁺ or Mg⁺⁺, resuspended in DPBS, and sonicated. An equal volume of supernatants from each extraction was combined and run on a 12% Bolt Bis Tris gel (Thermo Fisher Scientific#NP0321BOX) for 50 min at 150 volts, transferred to a nitrocellulose membrane, and blocked in 5% non-fat milk. Primary antibodies were diluted in 1% non-fat milk and are as follows: for pS129 (Abcam#ab51253; 1:2000 dilution), SQSTM1/p62 (Cell Signaling Technology#5114; 1:2500 dilution), LC3B (Cell Signaling Technology##3868; 1:2500 dilution), beclin1 (proteintech#11306-1-AP; 1:2500 dilution) and tubulin (Abcam#ab52866; 1:2500 dilution). Secondary donkey anti-rabbit horseradish peroxidase (HRP) (Santa Cruz Biotechnology#sc-2004; 1:2500 dilution) antibody diluted in 1% non-fat milk. Blots were developed in ECL Plus substrate for 5 min before imaging under Azure 300 - Chemiluminescent Imaging System and band intensity was quantified using ImageJ. The pSer129 blots were developed with SuperSignal Atto substrate.

Enzyme-linked immunosorbent assays (ELISA) were used to quantify the following cytokines: IL-1α, IL-1β, TNF-α, IL-6, IL-10 and IFN-Ƴ levels in the soluble fractions from control and CPF-exposed mouse brains. The multiplex ELISAs were performed in triplicate by Quansys Biosciences (Q-Plex Mouse Cytokine Panel 1 HS, Logan, UT).

Transgenic lines and wild-type (AB) ZF were maintained at 28 °C, fed brine shrimp twice a day, and kept on regulated by 14/10 hours light/dark cycles. Fish eggs were obtained from natural mating, and embryos were collected and staged based on post-fertilization days. Transgenic ZF Tg(vmat2:GFP) [20, 27] expressing green fluorescent protein (GFP) driven by the vesicular monoamine transporter promoter (VMAT2) were used to identify aminergic neurons including DA neurons. Tg(isl1[ss]: Gal4-VP16, UAS: EGFP)zf154 [20, 28] transgenic line was used to visualize trigeminal and Rohon-Beard peripheral sensory neurons. To study microglial activation and autophagy, Tg(mpeg1:mCherry) [20, 29] and Tg(elavl3:eGFP:map1lc3b) [30] lines were utilized, respectively. All experiments were conducted following protocols approved by the Animal Research Committee at the University of California, Los Angeles. ZF embryos were manually dechorionated and were exposed to 250 nM CPF at 24 hpf at 28 °C for 4–6 days.

ZF larvae were anesthetized with 0.01% tricaine methane sulfonate (tricaine-S; Sigma-Aldrich#E10521), and fixed in 4% paraformaldehyde (PFA) overnight at 4 °C. The larvae were washed in 1X Dulbecco’s Phosphate Buffered Saline (DPBS, ThermoFisher Scientific #J67670.K2) and treated with 1 mL peroxide mix (3% H²O² + 0.8% KOH in DPBS) for 8–10 min to bleach pigment. The larvae were washed with DPBS three times at RT immediately after the peroxide treatment and permeabilized for 5–7 min with 10 µg/mL Proteinase K (ThermoFisher Scientific #EO0491). Larvae were then washed in ddH2O and placed in a 10% blocking solution of 5% BSA + 5% goat serum for 1 h and incubated with primary antibody (anti-GFP; ThermoFisher Scientific #A11120, 1:500 dilution; anti-mCherry, Abcam #ab167453, 1:500 dilution; anti-TH, Sigma-Aldrich #MAB318, 1:250 dilution; or anti-Active Caspase-3, BD Pharmingen#559565, 1:500 dilution) in 1% BSA + 1% goat serum overnight at 4 °C. After five washes at 15-minute intervals, the larvae were incubated in secondary antibody overnight at 4 °C. After the secondary antibody washes, the samples were placed in 50% glycerol for 30 min and cleared in 100% glycerol. Larvae were dissected and mounted in 100% glycerol for imaging on the Leica SPE (Leica Microsystems Inc, Buffalo Grove, IL) confocal microscope. The vmat2:GFP positive neurons, TH neurons, microglia, and caspase-positive cells in the telencephalon and diencephalon regions were imaged in Z-stacks and analyzed. Peripheral sensory neurons were imaged and quantified by anesthetizing the transgenic isl1:GFP positive larvae and imaged at 10× magnification in the tail region and analyzed.

Exon 1 of the ZF sncg1 gene encoding γ1-syn [20, 31] (also called sncg1) was targeted using a pair of custom transcription activator-like effector nucleases (TALENs; Supplemental Fig. 1) [32]. Possible target sequences were found using ZiFiT [33] and selected to: (i) encompass a unique restriction site that could be used to identify mutants and (ii) avoid off-target homology by BLAST search. DNA constructs encoding custom TALENs were generated by iterative assembly using the Joung Lab TAL plasmid set [20, 32]. The resulting TALEN plasmids were linearized and transcribed in vitro to generate mRNA (mMessage Machine T7, Invitrogen). Single cell ZF embryos were microinjected with 300 pg mRNA for each TALEN (total 600 pg) in 0.5–1.0 nL microinjection buffer. Surviving embryos were raised to sexual maturity and screened for the germline transmission of sncg1 deletions by pairwise mating followed by PCR amplification and restriction digest of DNA from progeny embryos. Mating pairs with germline transmission were then outcrossed and their progeny were raised to adulthood. F1 founders were identified by fin clip PCR and the mutant allele was cloned and sequenced. We identified 3 different mutant alleles with deletions between 8-14 bp (Supplemental Fig. 1A). We selected a 13 bp deletion (allele designation Pt131) that is predicted to disrupt the sncg1 open reading frame (Supplemental Fig. 1B) for further analysis. The line was expanded from a single F1 founder and outcrossed through four generations prior to analysis to remove any off-target mutations. Heterozygous in-crosses resulted in progeny genotypes at the predicted Mendelian frequency and the sncg1^Pt131 mutation was stable over multiple generations. Homozygous sncg1^{Pt131/ Pt131} ZF were viable, fertile and showed no morphological or behavioral deficits during larval development. Rapid genotyping of the Pt131 allele exploited the deletion of a unique AfeI restriction site in exon 1 of sncg1 in the mutant (Supplemental Fig. 6C). Genomic DNA was extracted from adult fin clips or whole embryos and amplified using primers sncg1F 5’-CCTCTCTCTGTCATTGAAAC-3’ and sncg1R 5’-TAGGAAACACATGCACACAC-3’. The resulting 256 bp amplicon was then digested with AfeI (New England Biolabs), yielding bands of 164 bp and 92 bp for the WT allele, and a single band of 243 bp for the deletion allele (Supplemental Fig. 1C). Western blot analysis of adult brain homogenates showed the expected 14 kDa γ1-syn band in WT siblings but reduced expression in heterozygous sncg1^{+/ Pt131} ZF and complete loss of γ1-syn expression in homozygous sncg1^{Pt131/ Pt131} ZF (Supplemental Fig. 1D). These data confirm that Pt131 is a null allele. The homozygous mutant is correspondingly abbreviated to sncg1^−/− throughout the remainder of the paper.

ZF larvae were treated from 1 to 7 dpf with 250 nM CPF and washed with E3 egg water for 6 h prior to testing. Morphologically normal appearing ZF larvae (7 dpf) were transferred to 96-well plates with 16 larvae per condition, acclimated in the dark for 30 min, and their movements were monitored using a ZebraBox (ViewPoint ZebraLab, Civrieux, France). A 10-minute light/10-minute dark cycle was used for 3 cycles for a total of 1 h of recording. The distances moved by larvae during each increment of 10 min were averaged.

We used LysoTracker staining to image lysosomes in living ZF. Tg(mpeg1:mCherry) ZF (5 dpf) that have microglia labeled with mCherry, were incubated in the dark for 45 min using 10µM LysoTracker Green DND-26 (ThermoFisher Scientific#L7526) diluted in the treatment solution and then washed three times in E3 medium before mounting and imaging. Microglial structure was analyzed using ImageJ software as previously described [34, 35]. Briefly, Z-projected images were converted into a 16-bit grayscale file. The images were then made binary, and the skeletonization algorithm at FIJI was used for quantification.

To measure autophagic flux in ZF neurons, we used transgenic [Tg(elavl3:eGFP:map1lc3b)] ZF larvae as previously described [30, 34]. Briefly, ZF larvae were treated with 250nM CPF between 24 and 120 hpf. The larvae were anesthetized with 0.01% tricaine and mounted with 1% low-melting agarose in a glass-bottom culture plate. eGFP-Lc3 punctate were counted with confocal imaging using a 40x oil immersion objective (numerical aperture = 1.15) and an excitation laser line of 488 nm. Z-stacks comprised 13 sections, and each section of punctate eGFP-Lc3 positive tectum regions was acquired with a slice thickness of 1.5 μm and 1024 × 1024 pixel resolution. For the flux assay, ZF larvae were incubated with 2µM BafA1 (Cayman Chemical, 11038) for 55 min and then immediately washed three times with E3 water prior to live imaging. For restoring autophagic flux, larvae were treated with 7.5µM calpeptin (Sigma-Aldrich#C8999) from 48 hpf to 7 dpf with solution refreshed daily.

For analysis of ZF proteins, approximately 40 brains were dissected and homogenized with Radio-Immunoprecipitation Assay (RIPA) lysis buffer with protease inhibitor. After homogenization on ice, samples were centrifuged and protein concentrations in the supernatant were determined using the Bicinchoninic acid (BCA) assay. A final volume of 25 µl was used to load and run 20–30 µg of protein on a 12% SDS-PAGE gel with 1-mercaptoethanol and 1x loading dye. Using XCell-II Blot Module (ThermoFisher Scientific#EI9051), proteins were transferred to the nitrocellulose membrane. Afterward, the transferred membrane was blocked in 5% non-fat milk for 2 h at RT and probed with anti-sncg1 (generated as described previously [34, 36]), anti-Sqstm1/p62 (Cell Signaling Technology#5114; 1:1000 dilution), anti-tubulin (Sigma Aldrich#T9026; 1:2500 dilution) primary antibodies at 4 °C overnight. Then the membranes were washed with Tris-buffered saline (50 mM Tris base, 150 mM NaCl, pH 7.5) with Tween-20 (Bio-Rad Laboratories, 1610781) (TBST) followed by donkey-anti-rabbit HRP (Santa Cruz Biotechnology#sc-2313; 1:2500 dilution), goat-anti-rabbit (Vector Laboratories Burlingame#BA-1000; 1:2500 dilution), and goat-anti-mouse HRP (ThermoFisher Scientific#62-6520; 1:2500 dilution) secondary antibodies. Bands were visualized by exposing the blots using chemiluminescent substrate ECL Plus (ThermoFisher Scientific#32132) and then quantified using ImageJ’s gel analysis feature.

RNA was extracted from ZF larvae at 5 days post fertilization using TRIzol reagent (Invitrogen, 15596026) according to the manufacturer’s instructions and converted into cDNA using iScript™ cDNA Synthesis Kit (Bio-Rad Laboratories#1708890). Gene-specific primer sets (β-actinF 5’-TTCCTTCCTGGGTATGGAATC-3’, β-actinR 5’-GCACTGTGTTGGCATACAG G-3’; sncg1F 5’-ATGGTGGTATGGAAGGAGGA-3’, sncg1R 5’-GGGCTCAGGGAAAGTCT TTT-3’; p62F 5’-GTCATATGGGTCCATCTCCAAT-3’, p62R 5’-AGGTGGGGCACAAGTCA TAA-3’) were used to conduct the real-time (rt) PCR reaction using SsoAdvanced Universal SYBR Green Supermix (Bio-Rad Laboratories#1725270) and cDNA. The 2^−ΔΔcT method was used to represent fold change values from cycle number data.

To determine if microglia were contributing to aminergic neuron loss after CPF exposure, we used PU.1 targeted morpholino oligonucleotides (MOs) to inhibit microglial development. We targeted LC3 expression with MOs to determine if reduced autophagic flux can lead to aminergic neuron loss. Both MOs targeted the translational start codon (ATG) and were as follows: LC3, 5′-AGATCTGCCTAATTACTATCGTTTT-3′; PU.1 5′-GATATACTGATACTCCATTGGTGGT-3′; and scramble MO as a control 5′-CCTCTTACCTCAGTTACAATTTATA-3′ (Gene Tools, LLC). ZF embryos from 1- to 4-cell stages were injected into the yolk with MO stocks (1 mM) diluted 1:1 with 2% phenol red dye as previously described [34].

To assess the impact of CPF exposure on PD risk, we utilized the Parkinson’s Environment and Genes (PEG) study and performed an updated analysis of a previous report that included more subjects and controlled for additional exposures (n = 829 PD patients; n = 824 controls) [34, 37]. PD patients were, on average, slightly older than controls, a higher proportion were men, had European ancestry, and were never smokers (Fig. 1A; Table 1). We estimated ambient exposure to CPF due to living or working near agricultural facilities applying CPF over a 30-plus year period. We observed positive associations between CPF and PD with exposure estimated at residential and workplace addresses and over different exposure time windows (Fig. 1D; Table 2). The strongest association was with the longest duration of exposure at the workplace, with an OR of 2.74 (CI 1.55, 4.89). Importantly, CPF exposures that occurred 10–20 years prior to disease onset were more strongly associated with PD than the 10-year period before PD onset. These associations were similar when adjusted for occupational pesticide exposures and other common ambient pesticide exposures (paraquat, glyphosate and diazinon; Supplemental Table 1).

Most human pesticide exposure is through inhalation, which escapes the 1st pass circulation to the liver with oral ingestion and therefore reduces its metabolism. To model human exposures, mice were exposed to aerosolized CPF or ethanol vehicle in closed chambers five days a week for 11 weeks. We found in preliminary experiments that female mice were much more resistant to CPF than males (data not shown), therefore, only male mice were tested in the current study. The concentrations of CPF used were determined empirically (the highest dose that was well tolerated), and it was increased over time as the mice adapted to the exposures. CPF-exposed mice maintained their body weight as did the ethanol vehicle controls until the last week of the experiment (Fig. 2A). The concentration of CPF in the chamber reached 650 µg/m³/day initially and peaked at 2900 µg/m³/day by week 11 (Fig. 2B). Concentration of CPF in the brains of exposed mice was measured by liquid chromatography-tandem mass spectrometry and reached 1.44 to 2.42 ng/mg tissue (Fig. 2C).

Mouse behavior was analyzed before the initiation of CPF exposures and 3 days following the last day of exposure using rotarod, wire hanging, and open field testing. The 3-day washout was used to ensure that the behavior was not altered by CPF or vehicle and likely reflected underlying pathology. The baseline measurements for the open field tests for the CPF group were different than the controls but both the exposed and control mice performed worse after 11 weeks (Supplemental Fig. 2). CPF-exposed mice deteriorated more than controls in both rotarod, and wire hang tests, but not in the open field test (Fig. 2D-F).

A core pathological feature of PD is the relative selective loss of DA neurons, which results in impaired motor behavior. We determined the total number of DA neurons in control and CPF-exposed mice using stereological methods. CPF inhalation resulted in a 26% loss of tyrosine hydroxylase (TH) positive dopaminergic neurons in the SN compared to control mice. This DA neuron loss in the SN was selective, as CPF exposure did not affect the ventral tegmental area (VTA) DA neurons in comparison to control animals (Fig. 3A). Staining for TH in the striatum was reduced in the CPF-exposed brains relative to controls, consistent with DA neuron loss in the SNpc, but the difference did not reach statistical significance (p = 0.06, Fig. 3B). Phosphorylated α-syn at serine 129 (pS129) is a biomarker for pathological forms of α-syn and is elevated in PD brains. Western blot analysis confirmed that pS129 was significantly elevated 1.66-fold in the insoluble fractions of the CPF brains compared to controls (Fig. 3C, Supplemental Fig. 3) and a trend for increased pS129 protein level in the Sarkosyl soluble fractions (Supplemental Fig. 3). We found elevated levels of pS129 staining and ubiquitin in DA neurons of the SNpc compared to controls, whereas staining was similar in VTA neurons (Fig. 3D-G).

The increase in phosphorylated α-syn and ubiquitin in DA neurons in CPF-treated mice suggested that the accumulation might be due to dysfunction in protein degradation. Indeed, LC3-II and lamp 2a levels were reduced in TH-positive neurons in both the SNpc and VTA suggesting a defect in autophagy (Fig. 4). This defect does not appear to be widespread since Western blot analysis of LC3-II, p62 and Beclin1 in the soluble fraction from the CPF-treated whole brain homogenates did not show any differences from control fractions (Supplemental Fig. 4).

CNS inflammation is a common feature in PD and may contribute to its pathogenesis [34, 38]. Microglia change their shape and become more rounded in response to various stimuli such as environmental toxicants and infections [34, 38‐40]. To investigate whether microglia take on an activated morphology after CPF exposure, we determined the mean size and perimeter of Iba1-stained cells using the method of Fernandez-Arjona et al. [41]. We found that CPF-exposed microglia were more rounded and had shorter projections in the SNpc and VTA but not in the SNpr (Fig. 5A-C). These morphological changes are consistent with activated microglia similar to those seen in PD brains. We tested the brain soluble fractions for the proinflammatory cytokines IFNү, IL-1α, IL-1ß, IL-6, and TNFα and the anti-inflammatory cytokine IL-10 by ELISA. All cytokines except IL-1α and IL-1ß were at the lower end of detectability and there were no statistical differences between the CPF-exposed brains and controls (Supplemental Figure 5). We did find a non-significant increase in IL-1ß in the SNpc region but not in the soluble fractions of whole brain homogenates (Fig. 5).

We have found that exposure to CPF is associated with an increased risk of developing PD and that mice exposed to CPF through inhalation develop many PD features, including motor dysfunction, loss of DA neurons, α-syn pathology and activated microglia. To determine the mechanism of CPF neurotoxicity, we utilized transgenic ZF that are transparent and contain a well-developed DA system in the larval stage. Dechorionated ZF, 24 h post fertilization (hpf), were exposed to 0 to 100µM CPF for 6 days to determine its overall toxicity and found that all fish died at 30µM or higher by day 7 (Supplemental Fig. 6A, B). To determine whether CPF is toxic to aminergic neurons, including DA neurons, we have utilized the transgenic (Tg) vmat2:GFP line that expresses GFP under the control of vesicular monoamine transporter 2 (vmat2) promoter to monitor aminergic (DA, noradrenergic, and serotonergic) neuronal integrity as previously described [34, 36, 42]. We used the lowest concentration at which we observed decreased locomotor activity (250 nM) for all subsequent experiments. A stereotypical pattern of swimming response to light was observed in vehicle-treated embryos and CPF-treated fish at 5 dpf at similar speeds but the CPF-exposed ZF swam slower than controls in the dark at 7 dpf (Supplemental Fig. 6C-F). This pattern of ZF behavior is consistent with DA neuron loss [34, 36].

The integrity of the aminergic system in ZF was determined at 5 and 7 dpf after CPF (250 nM) exposure. The number of aminergic neurons in the telencephalic (Tc), diencephalic (Dc), and TH positive neurons in the Dc clusters at 5 and 7 dpf ZF embryos were significantly reduced (Fig. 6A-C). Interestingly, increased apoptosis was observed in the telencephalic and diencephalic regions of 5 dpf embryos but not in the larvae at 7 dpf (Supplemental Fig. 7), suggesting that this is the time of active neuronal death. In order to test for selective DA neuron loss as seen in PD brains, we exposed 7 dpf Tg(isl1[ss]: Gal4-VP16,UAS: eGFP)^zf154 larvae expressing GFP in Rohon-Beard neurons to 250 nM CPF and found no differences compared to controls at 7 dpf (Fig. 6D).

Another pathological feature of PD is CNS inflammation, which we also found in the CPF-exposed mice. We utilized MPEG-mCherry transgenic ZF larvae that have red microglia and exposed them to CPF and found a higher number of microglia that were more rounded and had shorter processes at 7 dpf compared to controls (Fig. 7A). Furthermore, the maximum branch length, the number of branches, and the number of junctions of microglia decreased. These structural changes in microglia are indicative of a more activated state, similar to that observed in mice. Activated microglia have more and/or larger lysosomes [43] so we used lysotracker labeling and live imaging to determine if the microglia were functionally activated. We found that the fluorescence intensity (MFI) of lysotracker in mCherry-labeled microglia was significantly increased following exposure to CPF consistent with microglia activation (Fig. 7B, Supplemental Fig. 8).

To evaluate whether microglia were involved in CPF-induced neurotoxicity, we inhibited microglial development using PU1 morpholinos (MO) before exposing the larvae to CPF. As shown in Fig. 7C and D, injection of PU1 MO resulted in marked reductions (87%-91%) of mpeg1:mCherry-positive microglia up to 7 dpf in both Tc and Dc regions of the brain. Neither scrambled control nor PU1 MO injections caused significant toxicity, nor did the injection significantly affect the number of surviving neurons in the embryos. Reduction of microglia had no significant effect on CPF neurotoxicity. Aminergic neurons were reduced by 22% in the telencephalon with no injection, 21% loss with control MO (scrambled) injection, and 14% loss with PU1 MO injection following CPF treatment (Fig. 7C). Similar results were found in the diencephalon (Fig. 7D).

Synucleins are neuronal proteins that comprise α-, β-, and γ-synucleins in mammals and α-syn aggregation and accumulation play a central role in PD pathophysiology [34, 44]. ZF do not contain an ortholog of human α-syn, although ZF γ1-synuclein (γ1-syn) has a similar function as α-syn [34, 45]. In order to determine if γ1-syn accumulates in CPF-exposed ZF, we performed Western blot analysis on dissected brain homogenates from 7 dpf larvae and found that γ1-syn protein levels were increased compared to controls (Fig. 8A), but mRNA levels were not, (data not shown). We then utilized γ1-syn knockout ZF to determine if the increase in γ1-syn contributed to DA neuron loss. CPF exposure did not result in DA neuron loss in γ1-syn knockout ZF suggesting that γ1-syn was required for CPF neurotoxicity in ZF (Fig. 8C).

CPF exposure led to an increase in γ1-syn protein levels but not its mRNA suggesting that the increase was due to decreased degradation. α-Syn is degraded by both the ubiquitin proteasome system (UPS) and autophagy but we focused on autophagy since aggregated α-syn is primarily degraded by this system [34, 46]. We utilized the Tg(elavl3:eGFP:map1lc3b) transgenic line to measure autophagic flux in neurons in vivo after CPF exposure and found a significant reduction in GFP-positive autophagosome (Lc3-II) puncta after treatment with CPF (Fig. 8D). We then measured the number of autophagosomes after 1 h of treatment with saturating concentrations of BafA1 and we found that CPF autophagic reduced flux (Fig. 8D). A reduction in autophagic activity by CPF exposure was further supported by the finding that the autophagy substrate p62 protein levels were increased while p62 expression was unchanged (Fig. 8B and Supplemental Fig. 9). Furthermore, knocking down LC3 using morpholinos reduced LC3 levels by approximately 60% and resulted in a significant decrease in DA neurons measured by both VMAT-GFP and TH IH (Fig. 9A-D). If disruption of autophagy contributed to DA injury, stimulating autophagy should be neuroprotective. We utilized calpeptin to induce autophagic flux (Fig. 9E-G and Supplemental Fig. 10), and we found that it rescued the neuron loss caused by CPF measured both by counting vmat2-GFP neurons and TH-positive neurons (Fig. 9E-G). Thus, CPF exposure led to reduced autophagic flux that contributes to DA neuron loss.