Association between CSF alpha-synuclein seeding activity and genetic status in Parkinson’s disease and dementia with Lewy bodies

The clinicopathological heterogeneity in Lewy-body diseases (LBD) highlights the need for pathology-driven biomarkers in-vivo. Misfolded alpha-synuclein (α-Syn) is a lead candidate based on its crucial role in disease pathophysiology. Real-time quaking-induced conversion (RT-QuIC) analysis of CSF has recently shown high sensitivity and specificity for the detection of misfolded α-Syn in patients with Parkinson's disease (PD) and dementia with Lewy bodies (DLB). In this study we performed the CSF RT-QuIC assay in 236 PD and 49 DLB patients enriched for different genetic forms with mutations in GBA, parkin, PINK1, DJ1, and LRRK2. A subgroup of 100 PD patients was also analysed longitudinally. We correlated kinetic seeding parameters of RT-QuIC with genetic status and CSF protein levels of molecular pathways linked to α-Syn proteostasis. Overall, 85% of PD and 86% of DLB patients showed positive RT-QuIC α-Syn seeding activity. Seeding profiles were significantly associated with mutation status across the spectrum of genetic LBD. In PD patients, we detected positive α-Syn seeding in 93% of patients carrying severe GBA mutations, in 78% with LRRK2 mutations, in 59% carrying heterozygous mutations in recessive genes, and in none of those with bi-allelic mutations in recessive genes. Among PD patients, those with severe GBA mutations showed the highest seeding activity based on RT-QuIC kinetic parameters and the highest proportion of samples with 4 out of 4 positive replicates. In DLB patients, 100% with GBA mutations showed positive α-Syn seeding compared to 79% of wildtype DLB. Moreover, we found an association between α-Syn seeding activity and reduced CSF levels of proteins linked to α-Syn proteostasis, specifically lysosome-associated membrane glycoprotein 2 and neurosecretory protein VGF. These findings highlight the value of α-Syn seeding activity as an in-vivo marker of Lewy-body pathology and support its use for patient stratification in clinical trials targeting α-Syn. Supplementary Information The online version contains supplementary material available at 10.1186/s40478-021-01276-6.


Introduction
The current gold-standard diagnosis for Parkinson's disease (PD) is based on clinicopathological criteria that include neuronal loss in the substantia nigra pars compacta and Lewy-body (LB) pathology related to alpha-synuclein (α-Syn) aggregation [1][2][3]. However, histopathological findings in some genetic forms of PD challenge this definition. While PD patients with mutations in the GBA gene (PD GBA ) show extensive LB pathology, most PD patients with bi-allelic mutations in the recessive gene parkin (PD recessive_bi-allelic ) show nigral degeneration without LBs [4]. The histopathology in PD patients with LRRK2 mutations (PD LRRK2 ) is also remarkably variable, including typical LB pathology, misfolded tau deposition or nigral degeneration without LBs [5][6][7]. This heterogeneity highlights the need for pathology-driven biomarkers that distinguish between the different underlying pathologies in-vivo. With disease-modifying treatment options targeting α-Syn underway, patient stratification according to α-Syn-specific enrichment strategies is a much-needed prerequisite to introduce patients to clinical trials.
As we have no reliable imaging marker to assess the cerebral load of α-Syn in-vivo, research has focused on CSF. Yet, it is unclear whether CSF profiles of α-Syn species reflect α-Syn brain pathology. Analyses in sporadic PD and PD GBA demonstrated decreased CSF levels of total α-Syn compared to controls with the highest decrease in PD GBA patients carrying severe mutations [8][9][10]. However, a substantial inter-individual variability and overlap with controls are seen so that CSF levels of total α-Syn are not ideal as a single biomarker. Recently, the ultrasensitive assays real-time quaking-induced conversion (RT-QuIC) and protein misfolding cyclic amplification have been successfully implemented. These assays exploit the seeding capacities of prion or prion-like proteins as an amplification strategy to reveal minute amounts of disease-specific protein aggregates in CSF [11,12]. Both methods showed a high sensitivity of 88-96% and 83-98% specificity for sporadic LBD such as PD and DLB compared to controls [13,14]. However, only one study on a limited number of patients has applied the α-Syn RT-QuIC assay to patients with genetic PD [15].

Aim, design and setting of the study
To further asses the value of α-Syn RT-QuIC, as pathology-driven biomarker for LBD, we assessed CSF α-Syn seeding activity in two large cohorts with LBD enriched for genetic forms. We included PD and DLB patients with GBA mutations as a proxy for pronounced α-Syn pathology, and PD patients with parkin, PINK1 and DJ1 mutations as representatives for nigral degeneration with sparse or no α-Syn aggregation. Moreover, PD patients with LRRK2 mutations were assessed. As no study has yet analyzed α-Syn seeding capacities longitudinally, we also evaluated α-Syn seeding profiles throughout the disease course in repeatedly collected CSF samples.
Besides, as several molecular pathways associated with α-Syn proteostasis have been identified over the last decade [16], we evaluated associations between α-Syn seeding profiles and CSF levels of proteins representative of these pathways.
Repeated lumbar punctures allowing longitudinal CSF measurements were performed in 100 PD patients (61

Clinical investigations
All participants were examined by a movement disorders specialist. Diagnosis of PD was defined according to UK Brain Bank Society Criteria [18]. Diagnosis of DLB was made according to the DLB consortium revised consensus criteria; only patients with probable DLB were included [19]. Patients were assessed in dopaminergic ON. We assessed severity of motor symptoms using part III of the Unified Parkinson's disease Rating Scale (UPDRS-III), from 2006-2008 the old version, from 2009 on the MDS-UPDRS [20]. Disease stage was categorized by modified Hoehn and Yahr Scale (H&Y) [21]. Cognitive function was tested with Montreal Cognitive Assessment (MoCA) [22] and/or Mini Mental Status Examination (MMSE) [23]. Since the MoCA was available only from 2009 on, previously obtained MMSE scores were converted into MoCA equivalents [24].

CSF collection
Spinal tap was performed between 9.00 am and 1.00 pm. Samples were centrifuged within 60 min and frozen at − 80 °C within 90 min after collection. Samples with abnormal routine CSF diagnostics (erythrocytes > 1/µl, white blood cell count > 5 cells/µl, immunoglobulin subtype G index > 0.7) were excluded.

Alpha-synuclein real-time quaking-induced conversion assay (RT-QuIC)
We performed purification of recombinant wildtype human α-Syn and the various steps of the RT-QuIC assay as described [14]. In addition to the negative control, we ran the same positive sample throughout all experiments to optimize the comparison between fluorescent responses in different plates. To overcome batch-to-batch variations and intrinsic plate-to-plate variability, we normalized the relative fluorescent units (RFU) for every time point to the maximum intensity reached by the positive control within each plate and expressed it as percentage. Intra-assay and inter-assays CVs before and after normalization are reported in Additional file 3: Table S2.
We calculated the threshold as the average normalized fluorescence value of negative control repeats during the first 10 h of recording, plus 30 standard deviations. The cut-off was set at 30 h. When only one of the four replicates crossed the threshold, the analysis was considered "unclear" and repeated up to three times. In those participants who showed a positive RT-QuIC α-Syn seeding profile (at least 2 out of 4 runs), we measured the area under the curve (AUC), the peak of the fluorescence response (Imax), and the lag phase (LAG) (time required to reach the threshold).
RT-QuIC experiments were performed at the Institute of Neurological Science of Bologna (ISNB). Results were reported blinded of the clinical diagnosis and genetic status. The assay showed a high specificity (98.7%) for LB pathology in a series of 121 CSF samples from individuals referred to the laboratory of neuropathology at ISNB for dementia of various etiologies in which the presence of LBs and abnormal α-Syn deposits was excluded by neuropathological examination [25].

CSF measurement of total alpha-synuclein
CSF levels of total α-Syn were assessed using an ELISA kit for human α-Syn (Roboscreen GmbH, Germany). Intraassay imprecision of 4.4% was calculated from duplicate analyses and expressed as median of the range to average of the duplicates. Inter-assay imprecision of < 10% was determined using two quality control CSF pool samples.

CSF measurements of proteins related to alpha-synuclein proteostasis
Measurements of the following protein concentrations were performed using parallel reaction monitoring mass spectrometry (PRM-MS) as previously described [26,27].

Parallel reaction monitoring mass spectrometry (PRM-MS)
Samples were dissolved by addition of 50μL 50 mM NH4HCO3 and shaken at room temperature for 1 h. Forty μL of sample were injected and separated using a Dionex ™ UltiMate ™ 3000 standard-LC system (Thermo Fisher Scientific Inc., Waltham, MA, USA) and a Kinetex ® EVO C18 column (length 150 mm; inner diameter 2.1 mm; particle size 1.7 μm; Phenomenex Inc., Torrance, CA, USA) with a SecurityGuard ™ ULTRA cartridge prefilter (Phenomenex Inc.). On a 60 min method, with solvents A (0.1% formic acid in H2O (v/v)) and B (84% acetonitrile and 0.1% formic acid in H2O (v/v)), using a flow rate of 300μL/min, the gradient went from 3 to 5% B over one minute followed by 5-26% B over 48 min. The column temperature was set to 50 °C. Separation by high-performance liquid chromatography was performed in online mode coupled to a Q Exactive ™ Hybrid Quadrupole-Orbitrap ™ mass spectrometer (Thermo Fisher Scientific Inc.). Using a HESI-II ionization probe (Thermo Fisher Scientific Inc.) electrospray ionization was performed in positive ion mode with the following settings: spray voltage + 4.1 kV, heater temperature 400 °C, capillary transfer tube temperature 380 °C, sheath gas flow rate 25, auxiliary gas flow rate 10, and S-Lens RF level 60. Acquisition of data was performed using single microscan in PRM mode with an isolation window of m/z 2 centred on the second isotope of the precursor ion. The resolution setting was 70 k with an AGC target of 1 × 106 and a 256 ms injection time. Fragmentation was performed using higher energy collision-induced dissociation.

Data extraction
Skyline v.19.1 was used to calculate and export fragment ion peak areas and to monitor fragment ion traces and ratios. The ratio between tryptic peptide and isotopelabelled peptide peak area was used for quantification.

Statistical analysis
Statistical analysis was performed using IBM SPSS 26.0 software. Group comparisons of dichotomous data were analyzed using the likelihood-ratio chi-square test as it is suited for unequally distributed samples sizes. Intergroup comparisons (disease groups stratified by genetic mutation; α-Syn seeders vs. non-seeders) of demographics, clinical markers, and CSF parameters were calculated using ANOVA/ANCOVA including sex, age, age at onset, and disease duration as co-variates as appropriate.
Pearson's correlation was used to evaluate associations between RT-QuIC seeding profiles and clinical markers and CSF parameters. Longitudinal analysis of RT-QuIC α-Syn seeding was done using linear mixed model with fixed factors group (seeders vs. non-seeders) and time (time of follow-up in years), their interaction and the random variable subject, modelled by random intercepts. Hypothesis testing was 2-sided and p values ≤ 0.05 were considered statistically significant.

Results
Demographic and clinical characteristics of all groups and subgroups are given in Table 1 and 2 and Additional file 4: Table S3, Additional file 5: Table S4, and Additional file 6: Tables S5.

RT-QuIC seeding activity is associated with disease status and genetic mutations Association with disease
Eighty-five percent of PD and 86% of DLB patients showed a positive RT-QuIC α-Syn seeding activity compared to 8% of controls (Table 1). Of the asymptomatic mutation carriers, 14% (1 GBA, 1 LRRK2) were α-Syn seeding positive. A single PD patient showing 1 out of 4 positive replicates by RT-QuIC in three runs was excluded from analyses given the inconclusive result. There were no significant differences in RT-QuIC seeding kinetics (AUC, Imax, and LAG) between PD and DLB patients with positive α-Syn seeding (p > 0.05).
In PD patients with positive α-Syn seeding, AUC was highest in PD GBA_severe (median 858 RFU) and lowest in PD recessive _heterozygous (median 702 RFU; overall p = 0.048). Accordingly, PD GBA_severe had the shortest LAG phase (19 h; overall p = 0.009) ( Table 2). PD GBA_severe showed also the highest proportion of samples with 4 out 4 positive replicates among PD groups (overall p ≤ 0.001).
Higher age and female sex were associated with higher CSF levels of AP2B1, CHGA, GM2A, LAMP2, SCG2, ubiquitin, and VGF (p < 0.01, respectively). No associations were found with disease duration.

RT-QuIC seeding activity is associated with motor and cognitive impairment
Positive α-Syn seeding and higher numbers of positive RT-QuIC replicates were associated with higher   Higher RT-QuIC AUC and shorter LAG were associated with higher UPDRS-III (r = 0.150, p = 0.034; r = − 0.165, p = 0.020) and lower MoCA scores (r = − 0.189, p = 0.009; r = 0.186, p = 0.011) (Additional file 5: Table S4). There were no associations with disease duration.
DLB patients with positive α-Syn seeding activity had a higher prevalence of REM-sleep behaviour disorder (RBD) (50% vs. 0%, p = 0.049) whereas sex, age, age at onset, disease duration, prevalence of parkinsonism and the interval between onset of parkinsonism and onset of dementia were not statistically significant when compared to non-seeders. DLB RT-QuIC non-seeders had higher mean CSF levels of t-Tau (533 vs. 294; p = 0.012) and NFL (3586 vs. 1630; p = 0.010) compared to RT-QuIC seeders (Additional file 6: Table S5).

RT-QuIC seeding profiles remain stable over time
Out of 100 PD patients included in this longitudinal analysis, 86 were RT-QuIC seeding positive. Mean time between first and last lumbar puncture was 3.5 years in RT-QuIC seeding positive and 3.4 years in RT-QuIC seeding negative patients. Linear mixed model analyses revealed that the number of positive RT-QuIC replicates did not change over time, neither in the overall longitudinal PD cohort (p = 0.670) ( Fig. 2A) nor in those PD patients with first LP at disease duration < 2 years (p = 0.949). Similarly, α-Syn seeding positives showed stable results of AUC (p = 0.576), Imax (p = 0.558) and LAG phase (p = 0.324) over time (Fig. 2B, C).

Discussion
The present results show a significant association between RT-QuIC α-Syn seeding activity and mutation status across the LBD spectrum. Remarkably, mutations in GBA, especially those classified as severe, showed the highest percentage of positive α-Syn seeding activity in both PD (93%) and DLB (100%). In contrast, PD patients with bi-allelic mutations in recessively inherited genes like parkin or PINK1 did not show CSF α-Syn seeding at all, whereas those carrying heterozygous mutations in these genes showed less α-Syn seeding than wild-type cases with a reduced positivity rate of 59%. Also, PD  = 0.670). B, C Similarly, RT-QuIC seeding positives showed stable results of Imax (p = 0.558) and LAG phase (p = 0.324) over time patients with LRRK2 mutations showed a reduced rate of α-Syn seeding (78%), although not to the extent previously reported in a Spanish cohort carrying the p.G2019S mutation [15]. The broader spectrum of mutations in our cohort, along with differences in clinical features such as older age and longer disease duration, might have contributed to these different results. Larger cohorts of patients carrying other mutations are needed to comprehensively evaluate the effect of LRRK2 on α-Syn pathology.
Overall, 85% of PD and 86% of probable DLB patients showed α-Syn seeding activity, which is slightly lower than the percentage obtained in previous studies [11][12][13]. The characteristics of our cohort, enriched for different genetic mutations, including those not harbouring LB pathology, best explain the results in PD patients. In contrast, the relatively low number of patients combined with the purely clinical diagnosis (akinetic-rigid parkinsonism, dementia and hallucinations in all cases) without supportive biomarkers (MIBG, DaTScan) possibly explain the lower frequency of positive cases in our wildtype DLB cohort. Moreover, previously analyzed DLB cohorts might have included some patients carrying a mutation in GBA given that they were not characterized genetically. Interestingly, patients with DLB that were RT-QuIC seeding negative had fewer clinical core features related to α-Syn pathology such as RBD [28]. These results suggest that most of the negative α-Syn RT-QuIC outcomes we obtained in patients with a clinical diagnosis of DLB did not indicate false-negative assay results. Indeed, it is well known from post-mortem and CSF studies that heterogeneous pathologies may sustain this clinical syndrome [19]. Consistent with this interpretation is the finding that DLB patients carrying GBA mutations, who were all positive by α-Syn RT-QuIC, showed normal mean CSF profile of t-Tau, p-Tau and Aβ 1-42, whereas the group of sporadic DLB patients without GBA mutations presented higher mean CSF levels of t-Tau, p-Tau and NfL as well as reduced Aβ 1-42 levels indicating a typical Alzheimer's disease profile. The heterogeneity in α-Syn seeding activity among the different genetic forms of PD mirrors histopathological findings in these cases. While PD patients with GBA mutations show extensive α-Synpositive LB pathology, those with bi-allelic mutations in parkin or PINK1 show nigral degeneration without LB, whereas those carrying LRRK2 mutations show LB pathology in most cases, but also misfolded tau deposition or nigral degeneration without LB pathology [4,29].
Additionally, we demonstrated an association between RT-QuIC parameters expressing higher α-Syn seeding activity (higher AUC, shorter LAG) with lower CSF levels of proteins related to α-Syn proteostasis, including autophagy, endocytosis, lysosomal function, ubiquitin-proteasome system and neurosecretion. Accordingly, PD and DLB patients with positive RT-QuIC seeding showed reduced CSF levels of lysosomalassociated membrane glycoprotein 2 and neurosecretory protein VGF compared to patients with negative seeding. Recent studies reported lower CSF protein levels of AP-2-complex subunit beta, cathepsin F, ganglioside GM2 activator, lysosomal-associated membrane glycoprotein 1 and 2, and ubiquitin in PD compared to patients with Alzheimer's disease and controls, and decreased CSF levels of chromogranin-A and B, neurosecretory protein VGF, and secretogranin species in PD and DLB compared to controls [30,31]. Interestingly, lower CSF levels of VGF were associated with lower CSF levels of α-Syn while no association was found with Aβ 1-42 [32,33]. These data suggest that amplification assays such as RT-QuIC might provide new tools to study in-vivo the mechanistic link between proteins linked to α-Syn proteostasis and seeding, which was previously shown only in postmortem studies and cell models [34][35][36][37][38][39].
Besides the capacity to accurately discriminate patients with LB pathology in-vivo, it will be essential to determine whether α-Syn RT-QuIC may have a role in predicting clinical trajectories. Increased α-Syn seeding capacities might predispose to accelerated/widespread α-Syn aggregation, thereby promoting faster progression to clinical disease milestones such as cognitive impairment. Our finding of an association between RT-QuIC kinetics parameters (AUC, LAG) with more severe motor impairment and cognitive dysfunction suggests that this might be the case. However, additional studies in prospective longitudinal cohorts where CSF and clinical data are available in de-novo PD patients with several years of follow-up, combined with further refinement of the quantitative capabilities of the assay are needed to fully address the value of the α-Syn RT-QuIC as a prognostic marker in LBD patients.
Most asymptomatic mutation carriers did not show RT-QuIC α-Syn seeding. The two RT-QuIC α-Syn positive mutation carriers (1 GBA, 1 LRRK2) were older and had lower MoCA scores than the negative ones. The observation is in line with a recent study in LRRK2 mutation carriers where only 19% showed a positive seeding compared to 20% of the control group. Those with positive seeding activity had higher likelihood ratios for probable prodromal PD [15]. These findings warrant further longitudinal investigations of larger cohorts with mutation carriers enriched for prodromal symptoms specific for α-Syn such as RBD.
Repeated longitudinal measurements revealed stable RT-QuIC α-Syn seeding profiles. This finding implicates that seeding activity represents a trait marker for α-Syn pathology rather than a marker that simply increases with time or disease duration. Further analysis in longitudinal cohorts with de-novo patients is needed to evaluate α-Syn seeding activity in different disease stages prospectively. The full reproducibility of the α-Syn RT-QuIC positive vs. negative outcome in samples taken from the same patients longitudinally also supports the assay's value as a specific pathology-driven biomarker of α-Syn pathology.
Strengths of our study include the large monocentric standardized collection of CSF samples minimizing variance in sample collection and processing, the validation of findings in longitudinal samples from the same patients, and the reproducibility in two alpha-synucleinopathies.
Limitations are the lack of longitudinal correlation between CSF RT-QuIC α-Syn seeding activity and postmortem brain histopathology and the small/imbalanced sample sizes of genetic subgroups and those stratified by RT-QuIC seeding profiles.

Conclusions
Our finding of heterogeneous seeding profiles across genetic forms of LBD suggest that CSF RT-QuIC α-Syn seeding activity is highly representative of and mirrors LB brain pathology in-vivo. We also show that RT-QuIC α-Syn seeding activity is associated with reduced CSF levels of proteins related to α-Syn proteostasis supporting mechanistic links suggested by in-vitro models. Importantly, this assay allows patient stratification according to α-Syn-specific enrichment strategies in preparation for clinical trials targeting α-Syn.