Several key findings are presented in this study: 1) pS129 and pS129/α-syn ratio increased in longitudinally collected CSF from untreated early PD patients, particularly those who progressed to requiring dopamine therapy; and 2) the relationship between pS129 and PD severity differed at earlier and later disease stages (high pS129 reflected less severe symptoms at earlier stages, but not at later stages). We examined this possibility using reanalysis of a previously studied cohort, after stratification by disease duration, and found that this progression appeared as a U-shaped curve when the two cohorts were combined. Moreover, this interpretation was supported by examination of a cohort of LRRK2 carriers with early or preclinical PD.
Longitudinal changes in pS129 in PD progressions
α-Syn in LBs is highly phosphorylated at S129 [5,6,10], particularly in more severe stages of Lewy pathology [7,20], suggesting a progressive increase in pS129 as PD advances. Additionally, cross-sectional studies in brain tissue and CSF indicate increased phosphorylated α-syn in PD [6,17], suggesting that it may be useful as a PD biomarker. However, whether it increases longitudinally in the CSF of individual subjects has been less well examined. Here, we found that pS129 increased over the approximately two-year DATATOP study, while α-syn, though it trended to be lower in more advanced stages, did not achieve statistical significance in the placebo group. The ratio of pS129/α-syn showed a greater increase, as anticipated given their opposite expected directions of change. This investigation is, to our knowledge, the first study of pS129 in living subjects with CSF collected longitudinally over a reasonably long period of time (about two years). The observation is entirely consistent with our previous cross-sectional study demonstrating that pS129 and the pS129/syn were higher in PD than in controls, and pS129 significantly positively correlated with severity of motor symptoms [17]. Our data, however, contrasted with another recent study of post-mortem ventricular fluid, which observed no difference in pS129 between PD patients and controls [30]. The lack of significance between advanced PD and controls, when assessed using ventricular fluids collected at autopsy, could be attributed to a number of factors, including cohort size, antibodies used, or use of autopsy ventricular fluid, which has a different protein composition from lumbar CSF [31]. However, later studies by the same group found increased plasma pS129 in PD, but no longitudinal increase in follow-up periods between 3 months and up to 4 years [32,33]. In light of our CSF results, the lack of further increase in plasma pS129 in longitudinal samples could be explained by a combination of different sample types, as well as by the follow-up period, particularly considering the modest magnitude of the change observed in CSF here, even over approximately two years. Based on our CSF investigations, it is likely that a much longer period of follow-up is needed to detect biochemical alterations in slow neurodegenerative processes [20,22].
pS129 and disease severity: evidence for a non-linear pattern in PD progression?
The most intriguing observation in the current study is that pS129 negatively correlated with PD severity at baseline, i.e. at a relatively early stage of the disease, in the DATATOP cohort. We hypothesized that this paradoxical observation might depend on disease stage, and separation of two cohorts by stage showed a trend, in both cohorts, toward more positive correlation with increasing severity. The effect of this change would be reflected in clinical trials as contradictory associations between early and late cohorts, as was observed when comparing the baseline DATATOP and multi-center collaborative cohorts. We considered that this relationship could be explained if pS129 decreases in the initial or preclinical stages of PD (leading to the negative correlation at early but post-diagnosis time points such as DATATOP baseline), then progressively increases at later stages (leading first to a loss of any correlation, followed by the introduction of a positive correlation, as well as longitudinally observed increase in concentration). We therefore sought to determine if a non-linear relationship appeared when a wider range of disease stages are studied together, particularly by including subjects at very early stages. Remarkably, the trend of a negative correlation between pS129 and PD severity at early stages was maintained in a cohort consisting of LRRK2 mutation carriers exhibiting early signs of PD. Although UPDRS scores were typically very low in these subjects, TBZ imaging, which reflects nigrostriatal damage robustly [34], showed that higher pS129 levels were associated with better disease states.
Of four groups (DATATOP baseline and final, multi-center collaborative, and LRRK2), only the DATATOP final cohort, at intermediate disease stage, showed no association between pS129 and UPDRS, in contrast to the late multi-center collaborative cohort, where a clear positive correlation is shown. This lack of correlation in a portion of DATATOP cohort may be explained by a mixture of the limited number of the early stage subjects at the final time point and the comparatively earlier status of even the “late” subjects, most of which just reached the point of requiring PD medication. This leads to the hypothesis that these subjects are at the stages where the pS129/severity relationship is undergoing a negative-to-positive transition (i.e., near the vertex of a U-shaped curve). It should also be noted that, although the reported severity (UPDRS scores) in the multi-center collaborative cohort overlapped substantially with that of the DATATOP final group, in the former, all subjects were on anti-parkinsonism medications, partially masking the true severity of their symptoms. In other words, subjects in this cohort can be considered to have more severe PD than DATATOP subjects with equivalent scores. For the reasons discussed above, it is not surprising that neither the levels of the markers at baseline, nor their changes over the duration of the study, predicted longitudinal progression in UPDRS of the DATATOP cohort (data not shown).
Taken together, these data suggest the possibility that the relationship between pS129 and disease severity may alter with progression. The mechanisms by which such a phenomenon occurs, and what it means for the role of pS129 in PD pathogenesis, remain to be examined. One possible interpretation of the current study is that phosphorylation increases as a compensatory mechanism, explaining both its increasing levels and apparently contradictory association with less severe symptoms at early-mid stages, but that the benefits are eventually overcome by accumulating negative effects of this or other PD-related changes, explaining how pS129 could continue to increase, and become associated with worse outcomes in patients with more advanced disease. Regardless of the mechanisms involved, however, these observations might provide some insight to the conflicting results obtained in various animal models of PD. For example, experiments in fly [10] and rat models [11,15] found opposite effects of phosphomimetic mutant α-syn on neuronal degeneration, and differing effects of phosphorylation on the propensity of α-syn to form aggregates have also been reported [6,10,11,16,35]. Further, differences have been observed between the phosphomimetic and genuine phosphorylated proteins [16,35], complicating interpretation of these studies. If alterations in pS129 depend on the stage of the disease, it is likely that other disease-related molecular interactions influence the effects of pS129 in the cell, and, for future studies, the effects of pS129 must be investigated in model systems that recapitulate the human condition as closely as possible. An important caveat that must be considered is that, while patients in the early/preclinical cohorts (DATATOP and LRRK2) were untreated, those in the multi-center collaborative cohort, in addition to being at more advanced stages, are also undergoing therapeutic treatment. With these datasets, one could argue that the differing relationship between disease severity and pS129 with progression is driven by effects of drug treatments (for example, treatment could result in increased pS129 levels, such that those with worse symptoms requiring increased therapy exhibit the highest levels, and masking a negative relationship such as was observed in both the DATATOP and LRRK2 cohorts). While this remains a theoretical possibility, the fact that a further increase in pS129 occurred in the longitudinal DATATOP cohort (Table 2), where all subjects are unmedicated, indicates PD therapy is unlikely the primary cause of this observation.
Additional considerations
Several additional caveats must also be considered. One obvious concern is the lack of neurologically normal controls in the DATATOP study, meaning that no group exists for comparison of PD-related changes in pS129 with its natural course with aging. Therefore, whether the increasing levels of pS129 observed in the DATATOP cohort are due to PD progression or aging cannot be definitively determined by this dataset alone. However, some data exists to suggest that aging is not the primary factor. First, we previously examined this question in our multi-center collaborative cohort, and found no relationship between age and pS129 in older PD patients or control subjects [31]. Further, we examined the cross-sectional relationship between pS129 and age in the DATATOP cohort, and again found no association. Together, these data suggest that CSF pS129 is not dependent on age in the older patients included here. However, this must be confirmed in longitudinally collected control subjects in later studies. Additionally, when using UPDRS to assess the utility of CSF pS129 and/or pS129/α-syn in monitoring the central nervous systems of PD patients, it is important to consider that biochemical and clinical measures of disease reflect different things: CSF α-syn and pS129 are measures of brain-wide pathology while UPDRS motor score reflects largely the degeneration of nigrostriatal system. Despite these caveats, sensitive and objective detection of alterations in CSF pS129 and total α-syn, which both alter over two years of progression, [20] would greatly aid clinical trials of novel, disease-modifying treatments targeting α-syn-related systems.
In summary, in this study of pS129 using longitudinal samples in addition to two cross-sectional cohorts, we provide evidence for pathological alterations of pS129, along with total α-syn, in the natural course of PD progression. These observations, though obtained in a large cohort, need to be further validated in an independent investigation, e.g. in prospective studies like the ongoing Parkinson’s Progression Markers Initiative. The significance of this study also goes beyond a biomarker research, informing future mechanistic studies of the disease, which should consider this natural course of pS129 in humans when modeling PD experimentally.
Compliance with ethical standards
The authors report no conflicts of interest. Protocols including human subjects were approved by all participating institutions, and performed in accordance with the ethical standards of the institutional research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards. Informed consent was obtained from all subjects prior to any procedures.