Alterations in Self-Aggregating Neuropeptides in Cerebrospinal Fluid of Patients with Parkinsonian Disorders

Abstract

Background: Parkinson's disease (PD), progressive supranuclear palsy (PSP), and multiple system atrophy (MSA) present with similar movement disorder symptoms but distinct protein aggregates upon pathological examination.

Objective: Discovery and validation of candidate biomarkers in parkinsonian disorders for differential diagnosis of subgroup molecular etiologies.

Methods: Untargeted liquid chromatography (LC)-mass spectrometry (MS) proteomics was used for discovery profiling in cerebral spinal fluid (CSF) followed by LC-MS/MS based multiple reaction monitoring for validation of candidates. We compared clinical variation within the parkinsonian cohort including PD subgroups exhibiting tremor dominance (TD) or postural instability gait disturbance and those with detectable leukocytes in CSF.

Results: We have identified candidate peptide biomarkers and validated related proteins with targeted quantitative multiplexed assays. Dopamine-drug naïve patients at first diagnosis exhibit reduced levels of signaling neuropeptides, chaperones, and processing proteases for packaging of self-aggregating peptides into dense core vesicles. Distinct patterns of biomarkers were detected in the parkinsonian disorders but were not robust enough to offer a differential diagnosis. Different biomarker changes were detected in male and female patients with PD. Subgroup specific candidate biomarkers were identified for TD PD and PD patients with leukocytes detected in CSF.

Conclusion: PD, MSA, and PSP exhibit overlapping as well as distinct protein biomarkers that suggest specific molecular etiologies. This indicates common sensitivity of certain populations of selectively vulnerable neurons in the brain, and distinct therapeutic targets for PD subgroups. Our report validates a decrease in CSF levels of self-aggregating neuropeptides in parkinsonian disorders and supports the role of native amyloidogenic proteins in etiologies of neurodegenerative diseases.

Keywords: Neurodegeneration; biomarkers; cerebrospinal fluid; mass-spectrometry; parkinsonism; proteomics.

Fig. 1

Flowchart of the experiment. HAP, High abundant proteins, in this study refers to albumin and IgG.

Fig. 2

Multivariate modeling of CSF protein profiling comparing parkinsonian disorders vs. healthy control (CON). Left panels are OPLS-DA scores scatter plots: x-axis (t[1]) are scores of first predictive component separating individual samples in case vs control classes; y-axis (t₀[1]) represents scores of the orthogonal component of within class differences. Right panels are weights of individual proteins contributing to the optimized models.The positive weights indicate an increase in disease compared to control.A) PD (n = 28) vs. Control (n = 26), R2X = 0.811, R2Y = 0.552 Q2 = 0.465 p = 3.1e-5; B) PSP (n = 12) vs. Control (n = 26), R2X = 0.877, R2Y = 0.626, Q2 = 0.382, p = 0.015; C) MSA (n = 10) vs. Control (n = 26), R2X = 0.851, R2Y = 0.613, Q2 = 0.373, p = 0.025.

Fig. 3

Scatterplots of significant candidates in the MRM validation. A broad range of measurements was detected in CSF from healthy control samples. A-G) Total patient populations compared to controls. While the atypical parkinsonian disorders are generally more similar within their disease group, the PD group generally exhibits a larger range of values, suggesting multiple subgroups within the PD diagnosis.H)Female and male PD patients compared to same sex controls. LRG1_DLL and C9_LSP were quantified label free (no SIS peptide), so the y-axis for these peptidesis a ratio to the optimized internal standard (IS). Other peptides were quantified by a spiked-in standard so the y-axis is the concentration of protein in ng/ml.

Fig. 4

Multivariate modeling of MRM validation comparing parkinsonian diseases to healthy control (CON). Linear regression analysis was employed to identify top candidates from MRM validation to build models for each disease. Further refinement results in optimized models shown. Left panels are OPLS-DA scores plots; right panels are weights of individual peptides contributing to optimized model.The positive weights indicate an increase in disease compared to control. Model statistics: A)PD (n = 59) vs. Control (n = 28): R2X = 0.933, R2Y = 0.339, Q2 = 0.24, p = 0.015; B) PSP (n = 11) vs. Control (n = 28): R2X = 0.921, R2Y = 0.631, Q2 = 0.563, p = 9.4e-5; C) MSA (n = 11) vs. Control (n = 28), R2X = 0.937, R2Y = 0.668, Q2 = 0.533, p = 0.0016.

Fig. 5

Scatterplots of candidate biomarkers among PD subgroups and control. A-D)Candidate biomarker levels in control and PD with or without leukocytosis. Ctr, control; LK, PD with leukocytosis;PDr, remaining PD. E, F)Candidate levels in control (Ctr) and PD subgroups: Pi, postural instability and gait disturbance (PIGD); Tr, tremor dominant; Po,intermediate motor phenotype.FBLN1_TGY was quantified label free (no SIS peptide), so the y-axis for this peptide is the intensity. Other peptides were quantified by a spiked-in standard so the y-axis is the concentration of protein in ng/ml.