Rotenone and paraquat perturb dopamine metabolism: A computational analysis of pesticide toxicity

Abstract

Pesticides, such as rotenone and paraquat, are suspected in the pathogenesis of Parkinson's disease (PD), whose hallmark is the progressive loss of dopaminergic neurons in the substantia nigra pars compacta. Thus, compounds expected to play a role in the pathogenesis of PD will likely impact the function of dopaminergic neurons. To explore the relationship between pesticide exposure and dopaminergic toxicity, we developed a custom-tailored mathematical model of dopamine metabolism and utilized it to infer potential mechanisms underlying the toxicity of rotenone and paraquat, asking how these pesticides perturb specific processes. We performed two types of analyses, which are conceptually different and complement each other. The first analysis, a purely algebraic reverse engineering approach, analytically and deterministically computes the altered profile of enzyme activities that characterize the effects of a pesticide. The second method consists of large-scale Monte Carlo simulations that statistically reveal possible mechanisms of pesticides. The results from the reverse engineering approach show that rotenone and paraquat exposures lead to distinctly different flux perturbations. Rotenone seems to affect all fluxes associated with dopamine compartmentalization, whereas paraquat exposure perturbs fluxes associated with dopamine and its breakdown metabolites. The statistical results of the Monte-Carlo analysis suggest several specific mechanisms. The findings are interesting, because no a priori assumptions are made regarding specific pesticide actions, and all parameters characterizing the processes in the dopamine model are treated in an unbiased manner. Our results show how approaches from computational systems biology can help identify mechanisms underlying the toxicity of pesticide exposure.

Keywords: 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine; 1-methyl-4-phenylpyridinium ion; 3,4-dihydroxyphenylacetate; Catsup; DAT; DOPAC; Dopamine; HVA; MPP+; MPTP; Mathematical model; Mode of action; ODEs; PD; Paraquat; Parkinson's disease; Rotenone; SNpc; catecholamines-up; dopamine transporter; homovanillic acid; ordinary differential equations; substantia nigra pars compacta.

Figure 1. Alterations of flux profiles under pesticide exposure, inferred through reverse engineering

The reverse engineering approach was applied to a dynamic model of dopamine metabolism in the so-called S-system representation. The results pinpoint different effects of rotenone and paraquat on the fluxes within the dopamine system. Fluxes are normalized to the corresponding fluxes under the control condition, i.e., without pesticide exposure. A. Rotenone affects fluxes associated with the compartmentalization of dopamine among cytosol, vesicles, and the synaptic cleft. In particular, the flux of dopamine into the synaptic cleft (eDA) which is the carrier of input signals to the postsynapse, is increased, whereas the flux of releasable dopamine stored in vesicles (IDA) shows a reduced overall magnitude. B. Paraquat exposure perturbs fluxes associated with dopamine and its breakdown metabolites, such as 3,4-dihydroxyphenylacetate (DOPAC) and homovanillate (HVA). Specifically, paraquat elevates fluxes into cytosolic and external DOPAC and suppresses the flux toward HVA. Interestingly, paraquat exposure does not seem to perturb the flux of dopamine through the synaptic cleft.

Figure 2. Schematic maps of pesticide action inferred through reverse engineering

Red arrows indicate an elevated overall flux through the corresponding metabolite due to pesticide exposure, while green arrows indicate reduced overall fluxes. A. Rotenone exposure; B. Paraquat exposure.

Figure 3. Distributions of parameters in dopaminergic neurons under rotenone exposure

Simulated distributions of parameter values for all processes in the dopamine system under rotenone exposure are shown. Out of 500,000 Monte Carlo simulations, only those results were retained that match experimental findings within a small error band. Several parameter values exhibit strongly skewed posterior distributions, indicating that they are candidate targets of rotenone. Sub-panel titles refer to the following kinetic parameters: P1: V_max-tyrosine; P2: K_{tyrosine-tyrosinebuffer}; P3: V_max-th; P4: V_max-aadc; P5: V_max-vmat2; P6: K_release; P7: V_max-dat; P8: V_{max-leakage-dat}; P9: V_{max-glialuptake}; P10: V_max-comt; P11: K_3mt-hva-mao; P12: K_{dopamine-dopac-maoaldh}; P13: K_{glialdopamine-glialdopac-maoaldh}; P14: K_{dopac-hva-comt}; P15: K_dopac-remove; P16: K_{glialdopac-hva-comt}; P17: K_{glialdopac-remove}; P18: K_hva-remove; P19: K_{dopamine-dopaminequinone}; P20: K_{dopaminequinone-dihydroxyindole-mif}; P21: K_{dihydroxyindole-melanin-tyr}; P22: V_max-tyr-dopa; P23: K_{dopaquinone-dopachrome}; P24: K_{dopachrome-dihydroxyindole-tyr}; P25: V_max-dct; P26: K_{melanin-remove}; P27: K_{extracellulardopamine-extracellulardopac-maoaldh}; P28: K_{extracellulardopac-hva-comt}; P29: K_{extracellulardopac-remove}. The corresponding results for paraquat are shown in Figure S1 of the Supplements.

Figure 4. Most significant mechanisms associated with toxicity of rotenone and paraquat in dopaminergic neurons

The distributions of three kinetic parameters that are most strongly affected by rotenone or paraquat (500,000 Monte Carlo simulations) are shown. The left column (maximum rate of the enzyme dopachrome isomerase V_max-dct) is used as a control, because it is unaffected by the pesticides, according to our results. The remaining three columns (from left to right) represent the maximum transport rate of VMAT2 (V_max-vmat2), V_max of tyrosine hydroxylase (V_max-th), and the rate constant for dopamine release (K_release). For rotenone (row A), the results indicate that the maximum rate for the transporter VMAT2 must be reduced. Furthermore, the rate constant for dopamine release into the cleft is elevated. In addition, the maximum turnover rate V_max-th for the rate-limiting enzyme TH seems to be mildly targeted by rotenone. For paraquat (row B), the same parameters are affected, but in a different manner. The maximum turnover rate V_max-th of the rate-limiting enzyme TH is inhibited. The rate constant for dopamine release is noticeably elevated. The maximum transport rate for the transporter VMAT2 could be targeted by paraquat, but the effect is not obvious.

Figure 5. Interdependency between parameters targeted by rotenone or paraquat in dopaminergic neurons

Locations of values of three parameters (V_max-vmat2, V_max-th, and K_release), which are strongly affected by rotenone and paraquat, are plotted in a 3D space. Parameter values are relative to their normal levels. Results are based on 500,000 Monte Carlo simulations. Each symbol represents an admissible scenario of toxicity. The results exhibit no obvious patterns of interdependency between these parameters. A: rotenone exposure; B: paraquat exposure.

Figure 6. Consistency of inferred rotenone mechanisms among different simulation scenarios

Different scenarios with 50,000, 100,000, 200,000, and 500,000 simulations were tested for the case of rotenone exposure. Rows A, B, C, and D represent 50,000, 100,000, 200,000, and 500,000 simulations, respectively. All four simulation scenarios led to the inference of the same mechanisms of toxicity, that is, the same sites and modes of action. As expected, simulations with higher numbers lead to smoother posterior distributions.

Figure 7. Quantification of pesticide action with respect to activation or inhibition

The ratios of the areas for activation to the corresponding areas for inhibition within the posterior distributions in bootstrap samples are shown for rotenone and paraquat exposure. For easier interpretation, the inverse ratios are used where the mean value of ratios is less than 1. Different mechanisms are associated with rotenone and paraquat not only qualitatively but also quantitatively. The activation of TH by rotenone is supported with statistical significance (p<5.0! 10⁻¹³⁹, paired t-test), although the mechanism is far less important than the other two mechanisms. The inhibition of the transporter VMAT2 by paraquat is statistically significant (p<9.0! 10⁻¹³⁹, paired t-test), but it is not as significant as the other two sites. From left to right, subplots are for the maximum transport rate of VMAT2 (V_max-vmat2), V_max of tyrosine hydroxylase (V_max-th), and the rate constant for dopamine release (K_release). Rotenone and paraquat likely inhibit VMAT2, while they activate dopamine release. However, these two pesticides show opposite actions with respect to TH. Within each subplot, two open bars are controls for rotenone (Control-1) and paraquat (Control-2), respectively; the blue bar represents rotenone exposure, while the yellow bar represents paraquat exposure. Significance is shown with *** indicating p <0.001 in a paired t-test.