Metabolic alterations in plasma from patients with familial and idiopathic Parkinson’s disease

The research of new biomarkers for Parkinson’s disease is essential for accurate and precocious diagnosis, as well as for the discovery of new potential disease mechanisms and drug targets. The main objective of this work was to identify metabolic changes that might serve as biomarkers for the diagnosis of this neurodegenerative disorder. For this, we profiled the plasma metabolome from mice with neurotoxin-induced Parkinson’s disease as well as from patients with familial or sporadic Parkinson’s disease. By using mass spectrometry technology, we analyzed the complete metabolome from healthy volunteers compared to patients with idiopathic or familial (carrying the G2019S or R1441G mutations in the LRRK2 gene) Parkinson’s disease, as well as, from mice treated with 6-hydroxydopamine to induce Parkinson disease. Both human and murine Parkinson was accompanied by an increase in plasma levels of unconjugated bile acids (cholic acid, deoxycholic acid and lithocholic acid) and purine base intermediary metabolites, in particular hypoxanthine. The comprehensive metabolomic analysis of plasma from Parkinsonian patients underscores the importance of bile acids and purine metabolism in the pathophysiology of this disease. Therefore, plasma measurements of certain metabolites related to these pathways might contribute to the diagnosis of Parkinson’s Disease.


INTRODUCTION
Parkinson's disease (PD) is the second most common neurodegenerative disease after Alzheimer's disease, affecting 7-10 million people worldwide. PD results from a complex interaction between genetic and environmental factors, appearing mostly as idiopathic cases, with no identifiable cause. Nevertheless, the discovery of several gene mutations associated with PD onset point to a genetic origin of the disease [1]. Mutations in the PARK8/LRRK2 gene are a common monogenic cause of PD. They are frequently found in early and late onset disease, in addition to this, familial or sporadic cases have been detected [2]. The substitution of glycine for serine in exon 41 of the protein kinase domain in LRRK2 (G2019S mutation), is the most common mutation, as was estimated by the international LRRK2 consortium, representing 1% of sporadic and 4% of familial PD cases worldwide [3]. Also, different substitutions in the conserved GTPase domain, in exon 31 of LRRK2 (R1441C, R1441G, and R1441H) have been identified as important genetic causes of familial PD [4].
Several therapies have been developed to relieve PDrelated symptoms, improving the patient's quality of life [5]. However, there are no efficient therapies available to stop this neurodegenerative process, and it is necessary to discover the mechanisms that trigger the onset of neurodegeneration in order to develop etiological therapies [6].
Metabolomic analysis offers an interesting tool to identify biochemical networks linked with the pathogenesis of this poorly understood disease [7]. Most of metabolomic studies are based on the analysis of metabolites in cerebrospinal fluid (CSF) [8] and blood samples [9], although there are also some studies that have examined other biological samples such as urine [10,11] or feces [12]. By using metabolomics, potential biomarkers have been discovered, including the biopyrrin, described as a new marker of idiopathic PD, after being found increased in the urine of these patients [11]. Moreover, dysregulated levels of polyamines, glutathione, kynurenine, pyruvate or cholesterol were all found in the plasma of PD patients compared to healthy individuals [13][14][15][16][17][18].
Cholesterol is a critical component of membrane bilayers and precursor of all steroid hormones and bile acids [19]. It plays key structural and functional roles in the general metabolism. Deregulations in cholesterol [9,17,18,[20][21][22] or in some of its products such as bile acids [23][24][25] have been linked to PD. However, studies on the association between serum cholesterol level and the risks of neurodegenerative diseases are currently under debate. High blood cholesterol, is a wellestablished risk factor for coronary disease and stroke [26,27], but its role in PD remains controversial. Blood concentrations of cholesterol have been associated with a higher prevalence of PD [28,29], and the use of cholesterol-lowering drugs, such as statins, have been associated with a decrease of PD [18]. However, another study failed to detect a significant association between serum cholesterol level and PD risk [30].
In addition, a correlation between changes in plasma levels of uric acid (UA) and the progression of neurodegenerative disorders has been described [14,31]. Most notably, decreased UA levels in blood (hypouricemia) were found in PD patients [32]. In contrast, high levels of UA in blood (hyperuricemia) were shown to lower risks for the disease [33][34][35], and to protect against clinical progression in PD [36]. However, the mechanisms of this neuroprotective role of UA needs to be investigated further, as there are some contradictory studies that have shown an association between gout (medical condition associated with high levels of blood uric acid) and a higher [37] or lower [38] risk of PD.
Here, we report an extensive mass spectrometry metabolomic analysis of plasma samples from patients with genetic and idiopathic PD and from a mouse model in which PD was induced by 6-hydroxydopamine (6-OHDA).

Convergent metabolic changes observed in plasma from PD patients and a mouse model of PD
We performed mass spectrometric metabolomics of plasma from: i, patients with idiopathic PD; ii, PD patients with the p.G2019S or p.R1441G mutations of the LRRK2 gene ( Figure 1A and Supplementary Table  1); and iii, from a PD mouse model (Supplementary Table 2). These mice were treated with the neurotoxin 6-OHDA, that triggers a rapid degeneration of midbrain dopaminergic neurons in the substantia nigra.
Our data revealed that both PD cases exhibit important metabolic changes compared to matched controls, both in humans ( Figure 1A) and in the mouse model (Supplementary Table 2). A deep analysis of the metabolites levels showed that PD patients groups (idiopathic and familial p.G2019S or p.R1441G) clustered together, underscoring that the differences observed in all metabolome ( Figure 1B) or in specific biological pathways are very similar between these groups (Supplementary Figure 1A). When splitting the correlation analysis into subtypes of metabolites, we found that bile acids and purine pathways were significantly modulated in all PD patients (R value > 0.86 in all comparisons) ( Figure 1C and Supplementary Figure  1), drawing our attention to these two pathways.

Increased level of unconjugated bile acids in PD patients' plasma
Bile acid synthesis takes place in liver and consists in the oxidation of cholesterol (Supplementary Figure 2A). By analyzing the metabolite level, a decrease in the levels of cholesterol in plasma of PD patients ( Figure  2A) (p = 0.006) and an important increase of the unconjugated bile acids, cholic acid (CA, p = 0.04) ( Figure 2B), deoxycholic acid (DCA, p < 0.001) ( Figure  2C) and lithocholic acid (LCA, p = 0.06) ( Figure 2D) were observed. However, except for the glycineconjugated DCA (G-DCA) metabolite, no general changes were found for other bile acids conjugated with glycine (G) ( Figure 2E) or taurine (T) ( Figure 2F). No significant differences in cholesterol plasma levels ( Figure 3A) were noticed in mice treated with 6-OHDA. Conversely, a general increase in unconjugated bile acid levels was observed in the plasma of these mice (CA, AGING p = 0.02; DCA, p = 0.14; LCA, p = 0.02) ( Figure 3B-3D), with no significant differences in conjugated bile acids ( Figure 3E, 3F), consistent with the results obtained in human PD.
Additionally, bile acids have been described to boost the synthesis and storage of glycogen in the liver, which leads to an FXR-dependent decrease in blood glucose levels [39]. In our models, both in PD patients (all patients, Glucose, p value = 0.035) (Supplementary Table 1) and in the murine model (All patients, Glucose, p value = 0.048) (Supplementary Table 2), we can remark a slight but significant hypoglycemia that could be an indirect consequence of the increase of bile acids.

The levels of uric acid and purine metabolic pathways are altered in plasma of all patients with PD
UA has been reported as a risk factor in PD [35]. Accordingly, we observed lower UA levels in the  AGING plasma of PD patients ( Figure 4A) (p = 0.13), however, this decrease was not spectacular. As UA is the end product of the metabolism of exogenous-and endogenous-derived purines (Supplementary Figure  3A), we decided to analyze in depth the rest of the metabolites of the purine pathway. Interestingly, the hypoxanthine levels in patients with PD are much higher than in healthy individuals ( Figure 4B) (p < 0.01), but no major changes in xanthine levels ( Figure  4C) were observed in these patients. In addition, analyzing the levels of a precursor of hypoxanthine, we can see a decrease in the levels of inosine ( Figure 4D) (p < 0.001). On the same line, we noticed a very significant increase in PD patients in the ratios between hypoxanthine and its precursor (hypoxanthine/inosine ratio) ( Figure 4E) (p < 0.001) or its product (hypoxanthine/UA ratio) ( Figure 4F) (p < 0.001), confirming the accumulation of hypoxanthine. In summary, we observed an increase in hypoxanthine levels and a decrease in both its product and its precursor, suggesting a blockade of this metabolic cascade.
As in the previous section, in order to verify these results, we also analyzed the modulation of the purine pathways in the murine PD model. Remarkably, an increase in the level of hypoxanthine was also observed in the serum of mice treated with 6-OHDA ( Figure 5A) (p < 0.01) similarly to the result obtained in patients. In this model, the change in hypoxanthine level was accompanied by an increase in xanthine levels as well ( Figure 5B) (p < 0.05). All these results together point to the robustness of the PD-associated increase in hypoxanthine.  AGING DISCUSSION PD is a progressive disorder caused by degeneration of neurons in the substantia nigra, the area of the brain that controls movement. The first clinical symptoms of the disease appear when 80% of these nerve cells have disappeared, reducing the effectiveness of dopaminergic neurotransmission in this area [40]. Different theories related to environmental toxins, genetic factors, and accelerated aging have been discussed as possible causes of this disease, but most patients diagnosed with PD (around 80-85 percent) have what is called primary parkinsonism or idiopathic PD, meaning that the cause of the disease is unknown. Only a small percentage of patients present a genetic cause of PD origin. Five Mendelian genes causing familial PD have been identified: PARK1/SNCA, PARK2/PARKN, PARK6/ PINK1, PARK7/DJ1, as well as the most prevalent one, PARK8/LRRK2 [1,41].
Our previous work has shown a basal autophagy impairment associated with an changes in intracellular protein acetylation levels in both genetic and idiopathic PD patients, leading to inefficient cellular responses to stress and increased susceptibility of cells to neurotoxins [41][42][43]. Indeed, several among the gene defects that cause human PD compromise the capacity of cells to destroy damaged mitochondria by autophagy (mitophagy), increasing cellular vulnerability to external and internal stress.
Extensive studies have been performed applying different "omic" technologies and particularly the study of the metabolome, the collection of small molecules (metabolites) contained in biological samples [7]. When performed in a high throughput manner [44], metabolomics can be considered as an emerging technology to explore PD-related biomarkers. Changes in metabolite concentrations have been detected in CSF samples from PD patients, as reported for oxidized glutathione, 3-hydroxykynurenine or homovanillate [8,[45][46][47]. However, the extraction of CSF carries some associated risks and is highly invasive [48]. Metabolomic studies performed on plasma samples, mainly focusing on idiopathic PD patients, identified new potential biomarkers in PD such as: polyamines [13], long-chain acylcarnitines [13], caffeine [49], tryptophan or bilirubin [50], glutathione or purine metabolism [14], cholesterol [9], kynurenine [15] or pyruvate [16]. However, in the past, only one study has been simultaneously performed on plasma from idiopathic and genetic (G2019S LRRK2 mutation) PD patients. That study described that an aberration of the purine pathway in PD would account for UA changes [51]. Furthermore, one metabolomic study has been carried out in parallel on both humans idiopathic PD and mice treated with the Parkinsonian toxin MPTP. That study reported convergent changes in L-DOPA and DOPAC levels in plasma, as well as an increase in DRD3 expression on lymphocytes in human and murine PD [52]. In the present study, we explored for the first time the general metabolome of plasma from idiopathic or familial (G2019S or p.R1441G mutations in LRRK2 gene) PD patients and also from a 6-OHDAtreated mice, providing evidence that bile acids and purine metabolic pathways play a role in the pathogenesis of PD.
We performed an in-depth analysis of the metabolic changes observed in plasma from all PD patients (idiopathic, p.G2019S or p.R1441G), observing that the subgroup of bile acids-pathways metabolites were convergently and significantly modulated in all PD patients and in 6-OHDA-treated mice. As previously shown [53], a significant decrease in the levels of cholesterol in the plasma from PD patients was observed. However, these alterations in cholesterol concentrations were relatively minor compared to the large increase in unconjugated bile acids in plasma from PD patients (Supplementary Figure 2B) or mice with PD (Supplementary Figure 2C). Considering that; i, there is still some uncertainty about the role of cholesterol in PD; ii, variations on bile acids occur in a murine model of prodromal PD in which mice received injections of α-SNCA fibrils [24], as well as in human idiopathic PD patients [54]; and iii, alterations of unconjugated bile acids in plasma occur in other neurodegenerative diseases, such as Alzheimer's disease [55]; these results highlight the importance of bile acids as a potential early biomarker of PD.
Bile containing bile acids, cholesterol and other organic molecules is secreted by hepatocytes into canaliculi, flowing into bile ducts. In the liver, the synthesis of bile acids represents the majority of cholesterol breakdown of the body and plays a critical role in the digestion and absorption of lipids in the small intestine. Multiple waste products are removed from the organism by their secretion into bile and finally discarded in feces [56]. However, bile acids can be deconjugated and/or dehydroxylated by the intestinal microbiota, return to the liver via the portal circulation, where they undergo a new round of metabolic modification (such as reconjugation) and become detectable in systemic venous plasma [57]. The increase of unconjugated bile acids in plasma levels that we observed in PD may be explained by an increased bacterial degradation of conjugated bile acids and/or a less efficient removal of unconjugated bile acids from the peripheral circulation. Interestingly, recent results suggested a bidirectional communication between the gut and the brain [58,59], and PD patients indeed exhibit a high prevalence of AGING small intestinal bacterial overgrowth (SIBO) [60]. Therefore, a possible method to understand the influence of intestinal dysbiosis on these results would be through differential bacteria proportions, by performing a thorough analysis of fecal microbiota and measuring metabolite levels after the removal of the bacterial influence.
Another pathway that is significantly modified in both human and murine PD concerns purine metabolism. UA is one of the products of purine metabolism and many studies have shown a correlation between high levels of plasma UA and reduced prevalence of idiopathic PD [35], suggesting a protective role for this antioxidant. We observed a trend towards lower UA levels in plasma from PD patients, and also an important increase in hypoxanthine levels, an UA precursor (Supplementary Figure 3B). Similarly, significant changes were observed in this pathway in the PD mouse model, where hypoxanthine and xanthine plasma levels increased (Supplementary Figure 3C). The mechanisms leading to changes in UA associated to PD are not known, and the hypothesis that aberrations in the purine pathway occur in PD has to be examined in more detail. In 2009, Johansen et al. observed that differences in the UA precursors could be responsible, at least in part, for the final decrease in UA levels observed in PD patients [51]. It is interesting to note that in asymptomatic LRRK2 p.G2019S carriers, hypoxanthine levels were significantly lower, but changes in UA levels were not significant. However, in advanced PD, the hypoxanthine levels were not significantly modified, and the levels of UA were notably reduced [51]. Thus, as the disease progresses, there appears to be a tendency to increase the levels of UA precursors when UA level decrease.
The strength of this study is the reproducibility of similar results in several different PD models: plasma from idiopathic, familial (p.G2019S and p.R1441G LRRK2) PD patients, and from a 6-OHDA-induced mouse model of PD, being the first complete metabolomic study carried out with these characteristics. The main limitation of our study lies in the number of individuals being tested. Increasing the number of patients studied would most likely strengthen the results. Moreover, it might be interesting to analyze asymptomatic LRRK2 mutation carriers to understand whether the metabolomic shifts observed here occur before the disease becomes clinically apparent.
In summary, the present study identified robust changes in bile acids and purine pathways that may constitute biomarkers for both idiopathic and familial PD and potentially reflect pathogenesis-relevant metabolic alterations.

PD patients:
Clinically established PD patients according to UK Bank criteria and healthy controls were divided into four groups (N=8/group): 1) healthy control individuals, 2) idiopathic PD patients, patients with 3) LRRK2 p.G2019S mutation, and 4) LRRK2 p.R1441G mutation. All the patient information (including age, gender and treatment received) is shown in Table 1. No interactions between gender or age and metabolic changes were observed ( Table 2). All PD patients were treated with different antiparkinsonian medications (mainly levodopa/carbidopa or statins+levodopa/|carbidopa), but no significant differences in the metabolic changes between the two sub-groups within the PD patients were reported ( Table 2). After informed consent approved by the Ethical Board of Hospital Donostia (ALM-LRRK2-2016-01) a blood sample was taken from antecubital vein after overnight fast and immediately processed to separate plasma and cells and stored at -80ºC until further experiments.

Mouse strains and housing
All animal experiments were approved by the "Ethics Committee for Animal Experimentation" of the Biomedical Research Institute "Alberto Sols" (CSIC-UAM) in Madrid (Spain) and carried out in accordance with the European Communities Council Directive (2010/63/EEC) and National regulations (normative RD1386/2018). Adult, male wild type C57/BL6 mice were obtained from Jackson Laboratories. The animals were housed in a cage (2-3 animals per cage) with free access to food and water under a 12 h light/dark cycle. Special care was taken to minimize pain or discomfort of animals.

Animal model of PD
Parkinsonism was induced as previously described [61]. Briefly, anaesthetized mice were placed in a stereotaxic apparatus (Kopf Instruments, CA) and 6-OHDA (5µg in 2µL saline with 0.02% ascorbic acid) was unilaterally injected into the substantia nigra pars compacta at the following coordinates from bregma: posterior, -3.2 mm; lateral, +2.0 mm; and ventral, +4.7 mm, with the skull flat between lambda and bregma, according to the atlas of Paxinos and Franklin [62]. Mice were then housed to recover.

Plasma sample preparation
A volume of 25 µL of plasma were mixed with 250 µL a cold solvent mixture with ISTD (MeOH/

Widely-targeted analysis of metabolites gas chromatography (GC) coupled to a triple quadrupole (QQQ) mass spectrometer
GC-MS/MS method was performed on a 7890B gas chromatography coupled to a triple quadrupole 7000C (Agilent Technologies, Waldbronn, Germany) equipped with a High sensitivity electronic impact source (EI)  AGING operating in positive mode. Front inlet temperature was 250°C, injection was performed in splitless mode. Transfer line and ion-source temperature were 250°C and 230°C, respectively. Septum purge flow was fixed at 3 mL/min, purge flow to split vent operated at 80 mL/min during 1 min and gas saver mode was set to 15 mL/min after 5 min. Helium gas flowed through column (J&WScientificHP-5MS, 30m x 0.25 mm, i.d. 0.25 mm, d.f., Agilent Technologies Inc.) at 1 mL/min. Column temperature was held at 60°C for 1 min, then raised to 210°C (10°C/min), followed by a step to 230°C (5°C/min) and reached 325°C (15°C/min), and be hold at this temperature for 5 min.

AGING
Collision gas was nitrogen. Scan mode used was MRM for biological samples. Peak detection and integration of analytes were performed using Agilent Mass Hunter quantitative software (B.07.01).

Untargeted analysis of metabolites by ultra-high performance liquid chromatography (UHPLC) coupled to a Q-Exactive mass spectrometer. Reversed phase acetonitrile method
The profiling experiment was performed with a Dionex Ultimate 3000 UHPLC system (Thermo Scientific) coupled to a Q-Exactive (Thermo Scientific) equipped with an electrospray source operating in both positive and negative mode and full scan mode from 100 to 1200 m/z. The Q-Exactive parameters were: sheath gas flow rate 55 au, auxiliary gas flow rate 15 au, spray voltage 3.3 kV, capillary temperature 300°C, S-Lens RF level 55 V. The mass spectrometer was calibrated with sodium acetate solution dedicated to low mass calibration.
10 μL of sample were injected on a SB-Aq column (100 mm × 2.1 mm particle size 1.8 μm) from Agilent Technologies, protected by a guard column XDB-C18 (5 mm × 2.1 mm particle size 1.8 μm) and heated at 40°C by a Pelletier oven. The gradient mobile phase consists of water with 0.2% of acetic acid (A) and acetonitrile (B). The flow rate was set to 0.3 mL/min. Initial condition is 98% phase A and 2% phase B. Molecules were then eluted using a gradient from 2% to 95% phase B in 22 min. The column was washed using 95% mobile phase B for 2 minutes and equilibrated using 2% mobile phase B for 4 min.
The autosampler was kept at 4°C. Peak detection and integration were performed using the Thermo Xcalibur quantitative software (3.1.).

Quantification and statistical analysis
Data are reported as Box and whisker plots (mean, first and third quartiles, and maximum and minimum values) or mean ± standard error of the mean (SEM). The number of independent data points (n) is indicated in the corresponding figure or in the Supplementary Tables 1 and 2 2 and 3) and Pearson's correlation coefficients with their 95% confidence interval was also used (Pearson's correlation coefficient (R) (Figure 1 and Supplementary Figure 1) and their p-value are shown). Clusterings and heatmaps have been performed using ''dist'' and ''hclust'' functions, using Euclidean distance method. (Prism version 7, GraphPad Software). Differences were considered statistically significant when p-values: * (p<0.05), ** (p<0.01), *** (p<0.001), and n.s. = not significant (p>0.05).

Ethics approval
Approval from the "Ethical Board" of Hospital Donostia

ACKNOWLEDGMENTS
Authors are grateful to patients and donors. They also thank Maria Pura Delgado-Luceño, Dr. Valentina Sica and FUNDESALUD for helpful assistance.

CONFLICTS OF INTEREST
The authors declare that they have no conflicts of interest.