Low plasma serotonin linked to higher nigral iron in Parkinson’s disease

A growing body of evidence suggests nigral iron accumulation plays an important role in the pathophysiology of Parkinson’s disease (PD), contributing to dopaminergic neuron loss in the substantia nigra pars compacta (SNc). Converging evidence suggests this accumulation might be related to, or increased by, serotonergic dysfunction, a common, often early feature of the disease. We investigated whether lower plasma serotonin in PD is associated with higher nigral iron. We obtained plasma samples from 97 PD patients and 89 controls and MRI scans from a sub-cohort (62 PD, 70 controls). We measured serotonin concentrations using ultra-high performance liquid chromatography and regional iron content using MRI-based quantitative susceptibility mapping. PD patients had lower plasma serotonin (p < 0.0001) and higher nigral iron content (SNc: p < 0.001) overall. Exclusively in PD, lower plasma serotonin was correlated with higher nigral iron (SNc: r(58) =  − 0.501, p < 0.001). This correlation was significant even in patients newly diagnosed (< 1 year) and stronger in the SNc than any other region examined. This study reveals an early, linear association between low serotonin and higher nigral iron in PD patients, which is absent in controls. This is consistent with a serotonin-iron relationship in the disease process, warranting further studies to determine its cause and directionality.

Parkinson's disease (PD) is a neurodegenerative disorder characterized clinically by bradykinesia, rigidity, and/ or tremor and pathologically by dopaminergic neuron loss in the substantia nigra pars compacta (SNc) and the presence of ⍺-synuclein-containing inclusions in cell bodies or neurites (Lewy pathology) 1,2 . PD also involves varying extents of extra-dopaminergic pathology in regions beyond the basal ganglia, and patients can suffer from many non-motor symptoms, not all of which are directly attributed to loss of dopamine 3 . The etiology and pathogenesis of PD remain poorly understood, but multiple genetic, environmental, and aging-related factors are thought to contribute to its pathophysiology 4 .
Iron accumulation in the brain is common with aging and consistently reported to be increased in PD 5 , particularly in the substantia nigra [6][7][8] (and specifically the SNc [9][10][11][12][13][14]. This accumulation has been suggested to contribute to dopaminergic neuron loss via multiple mechanisms. Although iron is essential to many biochemical processes, iron overload can predispose cells to oxidative stress 15 , ferroptosis 16 , and neuroinflammation 17,18 . In dopaminergic neurons, excess iron can be particularly damaging as iron-dopamine interactions can additionally facilitate the formation of highly neurotoxic dopamine intermediates 18 . Iron also interacts directly with ⍺-synuclein, and a complex interplay between these two factors in PD has been proposed to exacerbate ⍺-synuclein aggregation and iron-mediated damage 19,20 . Nigral iron content can be estimated in vivo using MRI-based susceptibility imaging, including state-of-the-art quantitative susceptibility mapping (QSM) 6,21 . Recent studies

Results
Demographic and clinical characteristics of subjects. We obtained plasma samples and clinical data from n = 186 participants in the NINDS PD biomarker program (97 PD patients, 89 controls), whose demographic and clinical characteristics are summarized in Table 1. MRI scans were obtained from a subset of these subjects who consented and were able to complete the scan with good image quality (n = 62 PD, 70 controls). In both the full cohort and MRI sub-cohort, PD patients and controls were similar in age, sex, BMI, and years of education and showed expected differences in PD-related metrics. Notably, PD patients had a four-fold higher rate of SSRI/SNRI use, indicating treatment for depression/anxiety, than controls overall, with approximately one-third of patients reporting use of an SSRI or SNRI (32% vs. 8%; p < 0.001).
Group differences in plasma serotonin concentrations. PD patients had lower mean plasma serotonin concentrations than controls [in log scale, 5.53 ± 1.40 vs. 6.25 ± 0.85, p < 0.0001; Fig. 1a]; however, the PD distribution appeared distinctly bimodal, and the group difference was driven by approximately one-third of patients with extremely low concentrations (the lowest tertile, n = 32; < 150 nM; Supplementary Fig. S1). In the remaining two-thirds of PD patients, plasma serotonin concentrations were distributed relatively widely and within the normal range of controls. Contamination from platelet serotonin can affect measured plasma concentrations, so we compared mean platelet counts between PD patients and controls. There was no significant difference in platelet counts between PD patients and controls (222 ± 50 vs. 230 ± 68; units = 109 platelets/L; p = 0.405).
Group differences in regional iron content in the brain (QSM and R2*) and serum iron status. PD Fig. S1). PD patients also had higher R2* values than controls in both nigral sub-regions [SNc, controls: 26.0 ± 3.7 ppb vs. PD: 27.6 ± 4.0 ppb; p = 0.018 and SNr, controls: 37.0 ± 6.8 ppb vs. PD: 40.6 ± 9.2; p = 0.016; data not shown]. In all other regions of interest, there were no statistically significant differences between PD patients and controls in either QSM or R2* values (p-values > 0.05; Supplementary Fig. S2). There also were no significant differences in serum iron markers between the groups (Supplementary Table 1 Association between plasma serotonin and QSM in other brain regions and with serum iron markers. To gauge the regional specificity of the correlation between plasma serotonin and QSM values in the SNc, we explored this association in several other regions of interest. In PD patients, the correlation was strongest in the SNc; however, lower serotonin also was associated with higher QSM values in the putamen  Table 2). Loess curves for each region (Fig. 3) demonstrated that plasma serotonin and QSM were correlated most robustly in the SNc and that, distinctly in this region, the correlation was strongest among the patients with the lowest serotonin concentrations. In other regions, the correlations seemed to be influenced by a few outliers (Fig. 3). In controls, no correlations were observed between plasma serotonin and QSM in any region (Supp. Table 2). There also were no significant correlations between plasma serotonin and serum iron markers in either PD subjects (Supp.  Of the depression-related outcomes, low plasma serotonin was strongly associated with SSRI/SNRI use, in both PD patients [age-and sex-adjusted mean: 3.88 (3.62, 4.14) in users vs. 6.30 (6.12, 6.48) in non-users; p < 0.0001; Fig. 4a] and controls [4.35 (3.80, 4.90) in users vs. 6.41 (6.25, 6.57), p < 0.0001 in non-users; data not shown)]. There was no relationship between plasma serotonin and depression or anxiety scores (HDRS or HARS) in either group, even after controlling for SSRI/SNRI use (p-values > 0.05).

Associations between nigral iron and other clinical outcomes. Similar to plasma serotonin, QSM
values in the SNc were not associated with age or sex in PD patients. In contrast, higher SNc QSM values were associated with longer disease duration (r = 0.465, p < 0.001). Clinically, higher QSM values in the SNc were associated with the same outcomes as low plasma serotonin, including higher LEDD (r = 0.450, p < 0.001) and worse scores on UPDRS-I (r = 0.304, p = 0.016), FOG-Q (r = 0.353, p = 0.005), and PDQ-39 (r = − 0.310, p = 0.014).
Higher SNc QSM values also were associated with SSRI/SNRI use (p = 0.003; Fig. 4b). All clinical associations with plasma serotonin remained significant after controlling for age, sex, and disease duration (Model 1a; Supplementary Table S3) but were weaker in the MRI sub-cohort (Model 1b; Supplementary Table S3) and disappeared or were further weakened by controlling for QSM in the SNc (Model 2; Supplementary Table S3).
Low plasma serotonin, SSRI/SNRI use, and the influence of SSRI/SNRI use on the serotonin-iron association. Because SSRI/SNRI use was more frequent in PD patients and strongly associated with both lower plasma serotonin and higher QSM values in the SNc, we investigated its potential influence on the serotonin-iron association. Partial correlation analysis showed that the association between plasma serotonin and QSM values in the SNc in PD patients remained significant after controlling for age, sex, disease duration, and SSRI/SNRI use [r (56)

Discussion
The present study investigated the potential relationship between low serotonin and increased nigral iron accumulation in PD. We first demonstrated, consistent with previous studies 8,26 , that PD patients had lower plasma serotonin concentrations and higher nigral QSM values (reflecting higher nigral iron content) than controls overall. Both features, however, showed wide inter-individual variation. Exclusively in PD patients, we observed a negative correlation between plasma serotonin concentrations and iron content in the SNc. Importantly, this correlation existed regardless of correction for age, sex, disease duration, and motor symptom severity (LEDD, UPDRS II and III), suggesting low serotonin and higher nigral iron are not simply clustering together in males or females or worsening in parallel as patients get older or the disease progresses. In fact, the correlation appeared strongest in patients during the first year of diagnosis but waned in later stage disease, when an increasing number of influences on these features presumably come into play. We also observed correlations between plasma serotonin and QSM in several other brain regions; however, none of these correlations were as robust as the one involving the SNc. Clinically, worse outcomes on several metrics were weakly associated with both low serotonin and higher nigral iron. These included freezing of gait, a symptom independently tied to each of these features in past studies 27,44,45 . Notably, SSRI/SNRI use, indicating treatment for depression or anxiety, was strongly associated with both lower plasma serotonin and higher nigral iron, but did not explain the correlation between these two features. Taken together, our findings reveal an early, robust association between low serotonin and higher iron in the SNc in PD and weaker associations between low serotonin and higher iron content in several other regions. We propose this lends further support for a serotonin-iron relationship in the disease process. To our knowledge, this is the first evidence linking low serotonin to higher nigral or other regional iron content in the brain in PD. Our findings are consistent mechanistically with prior reports of both an inverse www.nature.com/scientificreports/ serotonin-transferrin correlation in CSF in PD patients 37 and a correlation between higher depression and anxiety scores and higher nigral iron content in PD, even in patients with overall mild Parkinson's symptom severity 43 .
Neither the current nor previous cross-sectional studies allow inference of causality; however, collectively they suggest three intriguing possibilities: (1) that low serotonin contributes to nigral iron accumulation, either early in PD or as a risk factor to developing the disease, (2) that early nigral iron accumulation in the disease contributes to low serotonin (and depression), and/or (3) that another factor (or factors) mediates the relationship between the two. Exploring these possibilities could lend insight into serotonin and iron regulation in the disease and possible novel therapeutic targets.
An important observation in the current study was the absence of any serotonin-iron association in healthy controls, suggesting a PD-specific or disease-specific phenomenon. Interestingly, however, in the general population, major depression is associated with both lower plasma serotonin in older adults 46 and higher iron content in the putamen (reflected by QSM; the SN was not included as a region of interest) 47 . In the current study, the putamen of PD patients had the second strongest inverse correlation between plasma serotonin and iron content. It is possible that our observation of the serotonin-iron association only in PD patients simply reflects increased www.nature.com/scientificreports/ power to detect this association due to the relatively high rate of depression (and low plasma serotonin) in the PD population 33,34 . If so, the association could indicate a more general relationship between low serotonin/major depression and iron accumulation in the SNc, putamen, and other regions of the brain. Notably, depression in PD patients often emerges in the few years prior to PD diagnosis and is a risk factor (and possible causal risk factor) for the disease in older adults 48 . That increased iron accumulation in the brain in individuals with low serotonin and depression could mediate this risk is an intriguing possibility. Another interesting observation was that the serotonin-QSM association was most robust in the SNc, the major site of dopaminergic neuron loss and also the region where increased iron is reported most widely and consistently in PD 8 . Although iron accumulation in PD has been reported in other regions, including the putamen 49 , this occurs more variably. A recent systematic review reports that extra-nigral iron accumulation in PD is less consistently observed across QSM studies, with only a minority of studies reporting increases in extra-nigral regions and some studies reporting decreases 8 .
Although we cannot determine causality for the serotonin-nigral iron association, several plausible scenarios merit consideration. The most enticing, given the association even in early stage disease, is that serotonergic dysfunction contributes to iron accumulation in the brain, particularly in the nigra. Murine models have been used to show that the serotonin transporter (SERT) regulates ventral midbrain iron concentrations 39,50 . There are several factors increased in PD that could cause SERT dysfunction (⍺-synuclein pathology 41 , TNF-⍺ 42,5152 , SSRI/SNRI use 53 ), and SERT dysfunction could, in turn, affect plasma serotonin levels; however, the effects of SERT dysfunction on plasma serotonin are nuanced 54,55 , and the mechanisms connecting SERT to iron remain unclear. An interesting possibility is that SERT is only one of multiple points of intersection between the serotonin system and iron regulation in the brain. Serotonin synthesis is iron-dependent 56 , and early iron deficiency in rodent models causes persistent changes in brain serotonin levels 57,58 . A serotonin-iron feedback loop, therefore, is plausible, and there is initial evidence that serotonin regulates iron directly, as activation of 5HT1C receptors increases transferrin production in choroid plexus epithelium 59 .
It also is possible that brain iron accumulation might lead directly to serotonergic neurodegeneration and dysfunction 37 ) or that the relationship is indirect or mediated by a third or multiple factors. For instance, ⍺-synuclein and inflammation each have intricate, bidirectional ties to both serotonin and iron regulation and are suggested as pathogenic factors in PD 19,31,42,[60][61][62] Inflammation in particular stands out as a potential causal mechanism linking peripheral serotonin deficiency and brain iron accumulation. Serotonin has peripheral immunomodulatory actions contributing to both innate and adaptive immunity, and the immune system communicates with the brain via both humoral and neuronal mechanisms 63 . In addition, there is evidence that serotonin and other neuromodulators decrease microglial activation in the CNS 64 , the latter a process reported to be increased in both PD and Alzheimer's disease and linked directly to regional iron dyshomeostasis and iron transport into the brain 17 . It would be useful to determine whether low serotonin and higher nigral iron cluster together with more extensive ⍺-synuclein pathology or signs of inflammation in a subset, or subtype, of patients.
This study had several strengths, including the large, clinically well-characterized PD Biomarker Program cohort 65 and use of QSM, currently the most sensitive and reproducible MRI-based method for iron www.nature.com/scientificreports/ estimation 6,9,[21][22][23]66,67 . One potential limitation was the use of plasma serotonin concentrations as a surrogate for brain or CSF levels. Serotonin is synthesized both centrally and peripherally, and plasma and CSF pools are separated by the blood-brain and blood-nerve barriers; thus, the relationship between plasma serotonin and nigral iron is likely to be indirect. Nevertheless, peripheral and central serotonin share common regulatory mechanisms (e.g., tryptophan hydroxylase, SERT), and plasma (or serum) and CSF serotonin concentrations are correlated 68 . Both low plasma and CSF serotonin concentrations have been associated with depression in PD in independent studies 26,37 , suggesting that a mechanistic link is possible. We suggest the association between low plasma serotonin and higher nigral iron in PD reflects a broader, more general relationship between systemic serotonin deficiency (or its upstream factors) and iron accumulation in the SNc and several other brain regions, but further studies are needed to test this hypothesis. In our analysis, we understood that the source of a blood-derived sample (i.e., whole blood, platelets, plateletrich or platelet-poor plasma) has a marked influence on serotonin concentrations, with the vast majority of serotonin in the blood being stored in platelets [68][69][70] . Although we used rigorous, standardized PDBP protocols for total plasma collection, we did not measure serotonin in these various compartments or completely deplete the plasma of platelets. Conversely, all samples were collected in identical fashion, and our whole-blood platelet counts did not differ between PD patients and controls (consistent with 71 ) or influence the association between plasma serotonin and iron in the SNc. In the future, it might be useful to assess serotonin concentrations in the various blood compartments.
Future studies should also assess the high rate of SSRI/SNRI use in PD patients and its effect on plasma serotonin. About a third (35%) of PD patients have clinically significant depression 34 . In line with this, in the present cohort, approximately one-third of PD patients showed abnormally low plasma serotonin concentrations and/or were being treated with SSRI or SNRI antidepressants. There was significant overlap between these two groups. Due to the cross-sectional nature of this study, we could not determine whether low serotonin in any given individual existed prior to the use of antidepressant medications and might have contributed to depression and thus higher likelihood of antidepressant treatment or if low plasma serotonin was caused or exacerbated by SSRI/SNRI use. In prior studies, low serotonin (or 5-HIAA) in both plasma and CSF has been demonstrated in PD patients either not taking SSRI or SNRI medications 28,35,36 or not taking any serotonergic agents 26,37,38 . However, two recent studies in the general population report that long-term SSRI treatment lowers plasma serotonin concentrations, and lower pretreatment plasma serotonin concentrations (and less change) predict poor treatment response 54,55 . Importantly, regardless of cause, the association of low plasma serotonin with higher nigral iron remained significant even after controlling for SSRI/SNRI use. Moreover, the correlation between plasma serotonin and nigral QSM values even within SSRI/SNRI users as a subgroup argues against the association being a simple group effect. Nevertheless, future studies should account for SSRI/SNRI use and investigate the underlying causes of low plasma serotonin in PD patients.
Aside from SSRI/SNRI use, other possible causes of low serotonin should be considered in future work. These include (1) increased inflammation-induced kynurenine pathway activation that leads to reduced amounts of tryptophan available for serotonin synthesis (recently reported in PD 72 ), (2) changes in the microbiome that have been suggested to play a role in PD 73 and might alter serotonin synthesis in enterochromaffin cells of the intestine 74 , and (3) altered catabolism of serotonin, a factor recently associated with hyperserotonemia in autism 75 .
This study also opens the door to mechanistic studies using animal or cellular models aimed at determining whether there is a causal relationship between low serotonin and iron accumulation in the brain in PD and, if so, any therapeutic opportunity. It would be interesting to explore the effects of various serotonergic agents, such as SSRIs, SNRIs, and long-acting 5HTP 76 , on regional iron content in the brain. Similarly, one could explore the effects of iron chelators on serotonin levels. Future human studies could also investigate the relationship between increased nigral iron and dopaminergic neuron loss in low serotonin patients (e.g., using PET, SPECT).

Conclusions
This study revealed a robust correlation between low plasma serotonin and higher iron content in the SNc in PD, which was absent in controls. This correlation was present even in patients within one year of diagnosis, suggesting it might also be present in the prodromal phase, and was stronger in the SNc than any other brain region examined. Low serotonin and higher nigral iron were associated with a common set of clinical outcomes, and both were most strongly associated with SSRI/SNRI use, or treatment for depression/anxiety. Importantly, their correlation with each other persisted after controlling for this factor. We propose that these findings lend further support for a serotonin-iron relationship in the PD process and indicate a potential biochemical basis for the link between depression and nigral iron accumulation in PD patients recently reported 43 . Further studies in both patients and carefully selected and rigorously used animal models can determine direct mechanistic links if they exist as suggested by our data.

Methods
Participants. The study included 89 control and 97 PD subjects who participated in the National Institute of Neurological Disorders and Stroke PD Biomarker Program (NINDS PDBP) at the Pennsylvania State University between 2012 and 2015. PD patients were recruited from a tertiary movement disorders clinic, whereas controls were recruited from spouses and the local community. PD diagnosis was determined by a movement disorder specialist according to UK brain bank criteria. The enrollment criteria for PD subjects included history of adequate response to dopaminergic therapy, history of asymmetric onset, and lack of neurological disorders other than PD. PD duration was defined as the date since first PD diagnosis by a physician. Disease stages were set as early (< 1 year), mid (< 5 years), and late (5 + years) in accordance with previously established cutoffs 77 . www.nature.com/scientificreports/ All controls were free of known major medical and/or neurological conditions. All subjects provided written informed consent. The study was carried out in accordance with the Declaration of Helsinki. Ethical approval for the study was provided by the Pennsylvania State University Hershey Institutional Review Board.
Blood sample collection and processing. Blood samples were obtained and processed in compliance with the PDBP laboratory manual. Blood samples were collected before 10:00 AM following an 8-12 h overnight fast using BD Vacutainer glass blood collection tubes with K 3 EDTA. Within 30 min of blood draw, samples were centrifuged at 1,500 × g for 15 min at 4 °C followed by aspiration and division of plasma into 1 mL aliquots prior to freezing at − 80° C. Care was taken not to disturb the buffy coat. Plasma samples were maintained at − 80° C for 4-7 years (with CO 2 backup and 24/7/365 monitoring) prior to thawing once for sub-division of sample to send to the Van Andel Institute on dry ice for analysis. For analysis of peripheral serum iron metrics (red blood cell count, hemoglobin, hematocrit, serum iron, transferrin, total iron binding capacity (TIBC), transferrin saturation) and platelet counts, blood was collected into a BD PST II (plasma separator) tube and analyzed by the Penn State Hershey clinical lab using standard assays. This tube is routinely used in the clinical setting to measure iron, as these tubes lack EDTA, which is a known divalent metal chelator.
Plasma serotonin measurement. Plasma serotonin was measured as described in 72 . Plasma serotonin concentrations were measured using a Waters Acquity ultra-high-performance liquid chromatography (UPLC) I class and Xevo triple-quadrupole mass spectrometry (TQ-S MS/MS) system. 10 µL samples were injected into an Acquity HSS T3 column conjugated with a Vanguard HSS T3 guard column and eluted using a mobile phase, 0.6% formic acid in Milli-Q water (Solvent A) and 0.6% formic acid in analytical grade methanol (Solvent B), at an isocratic flow rate of 0.3 mL/min. The intra-assay coefficient of variability for serotonin was 5.5%. Estimation of regional iron content using QSM and R2*. Brain MRI scans were offered to all subjects but completed by only a subset (70 controls, 62 PD patients). For each subject, T1-weighted, T2-weighted, and multi-gradient-echo MRIs were acquired using a 3 T Siemens scanner. Iron content was estimated using QSM (primary estimate of iron) and the transverse relaxation rate (R2*) since R2* has been suggested to capture brain tissue properties beyond iron 23 . QSM and R2* values were generated for the purposes of a prior study; detailed scan parameters, segmentation methods, and quantification methods can be found therein 78 . Mean QSM and R2* values were quantified for both sides of each of the following iron-rich or basal ganglia structures: SNc (for brevity referred to as the nigra), substantia nigra pars reticulata (SNr), putamen, globus pallidus, subthalamic nucleus, dentate nucleus, caudate nucleus, red nucleus, and thalamus. Regional QSM values were used to assess group differences in brain iron content and its relationship with plasma serotonin and other clinical variables. Statistical methods. Statistical analyses were performed using IBM's SPSS (version 26) and SAS (version 9.4). All figures were produced using the ggplot2 package 80 in R (version 3.5.3) 81 . Plasma serotonin and nigral QSM and R2* distributions first were tested for normality, with plasma serotonin distributions failing this test (Shapiro-Wilk, p < 0.001). Plasma serotonin concentrations thus underwent logarithmic transformation (natural log) prior to parametric analyses. Control and Parkinson's subjects were compared using independent t-tests, ANCOVA with age and sex as covariates, or chi-square tests, as appropriate. Two-tailed tests were used to determine the statistical significance at ⍺ = 0.05. All mean values are reported ± standard deviation. Because the measurement of iron in brain regions other than the SNc and clinical metrics was exploratory, multiple comparison corrections were not performed. Associations between plasma serotonin and QSM, R2*, and serum iron markers were determined using Pearson's correlation coefficients and partial Pearson's correlations, controlling for potential confounders. Associations between plasma serotonin and clinical measures within PD patients were first assessed using Pearson's correlations, then in more depth using linear regression models (Type III). Two models were used to compare these associations. Model 1a and 1b tested the association between plasma serotonin and each clinical measure in all PD patients (n = 97) and in the subset with MRI data (n = 62). Model 2 tested this association in the subset with MRI data also controlling for SNc QSM (n = 62). All models controlled for age, sex, and disease stage as factors, and were run with only main effects and also with main effects plus the plasma serotonin and sex interaction. Since no interaction was observed, we only presented main effects models.