Natural variation in age-related dopamine neuron degeneration is glutathione dependent and linked to life span

Abstract

Aging is the biggest risk factor for Parkinson's disease (PD), suggesting that age-related changes in the brain promote dopamine neuron vulnerability. It is unclear, however, whether aging alone is sufficient to cause significant dopamine neuron loss, and if so, how this intersects with PD-related neurodegeneration. Here, through examining a large collection of naturally varying Drosophila strains, we find a strong relationship between life span and age-related dopamine neuron loss. Strains with naturally short-lived animals exhibit a loss of dopamine neurons without generalized neurodegeneration, while animals from long-lived strains retain dopamine neurons across age. Metabolomic profiling reveals lower glutathione levels in short-lived strains which is associated with elevated levels of reactive oxygen species (ROS), sensitivity to oxidative stress, and vulnerability to silencing the familial PD gene parkin. Strikingly, boosting neuronal glutathione levels via glutamate-cysteine ligase (Gcl) overexpression is sufficient to normalize ROS levels, extend life span, and block dopamine neurons loss in short-lived backgrounds, demonstrating that glutathione deficiencies are central to neurodegenerative phenotypes associated with short longevity. These findings may be relevant to human PD pathogenesis, where glutathione depletion is reported to occur in the idiopathic PD patient brain through unknown mechanisms. Building on this, we find reduced expression of the Gcl catalytic subunit in both Drosophila strains vulnerable to age-related dopamine neuron loss and in the human brain from familial PD patients harboring the common LRRK2 G2019S mutation. Our study across Drosophila and human PD systems suggests that glutathione synthesis and levels play a conserved role in regulating age-related dopamine neuron health.

Keywords: Parkinson’s disease; aging; glutathione; neurodegeneration.

Fig. 1.

Age-related dopamine neuron degeneration correlates with life span and locomotor dysfunction. (A), Kaplan–Meier survival curves for 27 DGRP strains (see SI Appendix, Table S1 for details and strain IDs). Each line represents a single genotype. (B), mean life span ± SE for the 27 strains in (A). SEM life span ≤ 2.0 d; log-rank across 27 strains (χ² = 2,866, P < 1e⁻¹⁵). Strain IDs in SI Appendix, Table S1. (C), TH-positive dopamine neuron number in the PPL1 cluster (n = 7 to 12 brains/genotype), plotted against strain mean life span with six DGRP strains annotated, Spearman r and P values annotated. See SI Appendix, Table S1 for strain IDs. (D), Confocal projection images of PPL1 dopamine neurons in a single hemisphere for 30-d-old flies belonging to the six DGRP strains annotated in (C), mean lifespans ± SEM annotated. (Scale bar, 10 µM.) (E), PPL1 cell counts in 1-d-old and 30-d-old flies for all DGRP strains with ≥1 neuron less than the average cell counts across all 27 strains at 30 d (dotted line). Two-way ANOVA for effect of age (P < 0.0001) and DGRP strain (ns), Šídák’s multiple comparison test, ***P < 0.001 (n = 5 to 12 brains/genotype/age). (F) Long-lived strain PPL1 counts at 30 d. (G) locomotor function in negative geotaxis assays correlates with PPL1 dopamine neuron counts in 30-d-old flies from the same 27 DGRP strains as above (n = 4 groups of 25 flies/genotype for geotaxis and 7 to 12 brains/genotype, Spearman r and P values annotated). Strain IDs in SI Appendix, Table S1 and Methods. Data in (B), (E) and (F) are mean ± SEM.

Fig. 2.

No generalized neurodegeneration in short-lived DGRP strains. (A) Brain vacuoles (Methods) assessed in short-lived and long-lived DGRP strains aged for 5 wk (Inset magnified images show vacuoles adjacent to asterisks). Occasional vacuoles in DGRP brains were minor in number (B) and area (C) compared to sws⁴ heterozygous mutant females aged 2 wk. Individual ANOVAs for effect of genotype: vacuole number (P < 0.01 without sws; P < 0.0001 with sws, n = 6 to 13 brains/genotype); vacuole area (P < 0.01 without sws; P < 0.0001 with sws), Dunnett’s post hoc analysis (****P < 0.0001, n = 6 to 13 brains/genotype).

Fig. 3.

Metabolomic profiles of short- and long-lived strains. (A) Principal components 1 and 2 of head tissue samples, colored by DGRP strain (strain numbers annotated). Ellipses are 95% CI regions. Sample replicates cluster well within a genotype and are not readily separable by life span group. (B) Pathway analysis of significantly differing metabolites, integrating enrichment analysis (y axis) and pathway topology analyses (x axis). (C) Volcano plot of 147 metabolites assessed, showing the most down-regulated and up-regulated metabolites based on log₂(fold-change) at a FDR < 0.1 threshold (dotted line). (D) Head total glutathione content in 1-wk-old females from DGRP strains (Methods) correlated to PPL1 dopamine neuron counts or mean life span (n = 4 replicates of 25 heads/replicate). Spearman r and P values annotated. (E) glutathione biosynthesis pathway with enzymes annotated. Glutathione exported from astrocytes can supply neurons with glycine and cysteine substrates following its breakdown to cysteinylglycine by Ggt-1. (F) Correlation of 1-wk-old glutathione levels with expression of gclc, gclm, Gss, and Ggt-1 at the same age across the same 27 DGRP strains (Spearman r with P values in parenthesis).

Fig. 4.

Glutathione dependency of age-related dopamine neuron loss (A) PPL1 dopamine neuron rescue in short-lived DGRP backgrounds following pan-neuronal gclc overexpression (Scale bar, 20 μM) quantified from both hemispheres in (B), two-way ANOVA for effect of gclc overexpression (P < 0.0001), DGRP strain (P < 0.0001), and interaction (P < 0.0001) with Bonferroni’s posttest (**P < 0.01, ****P < 0.0001, n = 10 to 12 brains/genotype). There was no significant effect of gclc overexpression in the long-lived DGRP 821 strain (Bonferroni’s posttest, ns, n = 10 to 12 brains/genotype). (C) Survival is significantly enhanced in three short-lived strains following pan-neuronal gclc overexpression, but not in a long-lived strain (P-values indicate log-rank tests for each control genotype relative to elav-GAL4/UAS-gclc, see also SI Appendix, Table S2 for details). Genotypes in (A–C) are the DGRP strain indicated crossed to elav^C155-GAL4 (elav-GAL4/+), crossed to UAS-gclc (UAS-gclc/+), or to elav^C155-GAL4/UAS-gclc (elav-GAL4/UAS-gclc).

Fig. 5.

ROS levels and oxidative stress sensitivity in short-lived backgrounds (A) Head H₂O₂ levels correlate inversely with strain mean life span at 10 d of age, but not at 2 d, for three short-lived (RAL-727, RAL-911, and RAL-356) and three long-lived (RAL-335, RAL-379, and RAL-821) DGRP backgrounds (n = 3 groups of 10 heads/genotype). (B) Head H₂O₂ levels are reduced following Gclc overexpression in 10-d-old flies (two-way ANOVA for effect of Gclc (P < 0.0001) and DGRP background (P < 0.0001), Šídák’s multiple comparison test, *P < 0.05, **P < 0.01, ****P < 0.0001, n = 3 groups of 10 heads/genotype). Genotypes are elav^C155-GAL4/+, UAS-gclc, or elav^C155-GAL4/UAS-gclc in the DGRP background indicated. (C) Correlation of mean life span and survival on H₂O₂-containing food for the same six DGRP backgrounds as in (A). (D) PPM neurons in three short-lived and three long-lived DGRP backgrounds outcrossed to TH-GAL4 (ctrl) or to TH-GAL4/UAS-parkin-IR (dopamine neuron-specific parkin RNAi line 37509). All three short-lived backgrounds and one long-lived background (RAL-335) exhibit dopamine neuron loss upon parkin knock-down, quantified in (E), two-way ANOVA for effect of parkin knock-down, P < 0.0001, and DGRP background, P < 0.01, interaction P < 0.0001, Šídák’s multiple comparison test, *P < 0.05, **P < 0.01, ***P < 0.001, n = 10 to 13 brains/genotype.

Fig. 6.

Phenotype relationships are upheld in virgin females, low-density-cultured mated females, and mated males (A) Mean lifespan of 27 DGRP strains positively correlates between mated females, virgin females (data from ref. 26), mated males and mated females cultured under low density (8/vial) with more frequent food changes (Methods). F, females; M, males. (B) Mean lifespan (n = 100 to 115 flies), PPL1 dopamine neuron counts (n = 9 to 10 brains), head total glutathione (n = 4 groups of 25 flies), and H₂O₂ levels (n = 3 groups of 10 flies) correlate in virgin females, mated females cultured under low density with more frequent food changes and mated males. Coefficient r and P values (parenthesis) annotated. See also SI Appendix, Tables S1 and S3–S7.

Fig. 7.

GCL expression is reduced in vulnerable fly strains and the familial LRRK2 G2019S PD brain. (A) gclc transcript levels in fly heads from DGRP strains with relative loss or retention of dopamine neurons. “DA loss” are strains with less than the average PPL1 dopamine neuron counts at 30d of age (Fig. 1), “DA retention” are strains with above average PPL1 dopamine neuron counts at 30 d. See Methods for strain IDs. (B) Levels of GCLc and GCLm in human brain cortical extracts from PD patients harboring LRRK2 G2019S are different from age-matched healthy controls, quantified in (C) relative to β-actin and normalized, analyzed by multiple unpaired Student’s t tests with Benjamini, Krieger, and Yekutieli correction for multiple comparisons, *q < 0.05, n = 5 per group.