Neuropathological and Biomarker Findings in Parkinson’s Disease and Alzheimer’s Disease: From Protein Aggregates to Synaptic Dysfunction

INTRODUCTION

Partly derived from the fact that dementia is very common in Parkinson’s disease (PD) [1], there is mounting neuropathological evidence that PD and Alzheimer’s disease (AD) share several common features [2, 3]. Traditional post-mortem neuropathological studies are nowadays supplemented by biomarker studies purportedly reflecting the underlying pathology in vivo, ranging from biochemical studies in cerebrospinal fluid (CSF) to molecular imaging of proteins deposition such as amyloid-β (Aβ) or tau. Both types of studies have favoured in recent years the notion that neurofibrillary tangle-type lesions composed of hyperphosphorylated tau and, particularly, Aβ-containing aggregates are common in PD and associated with presence and risk of, as well as timing to, dementia [2, 3]. All this has supposed a paradigm shift of sorts, departing from the general conception at the end of the 20th and early 21st century that cortical Lewy pathology alone accounted for dementia in PD, to conceiving that both Lewy and Alzheimer pathologies are relevant in PD-dementia. Additionally, there is also consistent data as to the concomitant presence of α-synuclein containing Lewy-type aggregates in a significant proportion of both sporadic and familial AD, particularly in the amygdala [4, 5].

However, it remains unknown what is the exact mechanistic role of co-existing Lewy and Alzheimer pathologies observed in post-mortem studies, with common criticism being that these most often reflect findings in end-stage cases (unless the autopsies are performed in patients dying prematurely of an unrelated illness) and that these may not necessarily reflect what originally drove the symptoms (in this case more importantly, but not exclusively, dementia).

As for in vivo biomarker studies-derived evidence of coexistence of both Lewy and Alzheimer pathologies it is yet controversial as it is still not clear whether the used biomarkers actually reflect underlying pathology or rather are the consequence of some other molecular processes. Thus, in studies assessing CSF biomarkers, the main concern is that these might be reflecting non-specific alterations (mostly axonal loss or neuronal degeneration, in the case of tau [6]) or intrinsic processes related to the soluble species of the involved protein (such as synaptic dysfunction in the case of Aβ [7] and α-synuclein [8]), rather than the respective disease protein aggregates.

Here we revisit the neuropathological and bio-marker evidence from recent years focusing in pathology and synaptic dysfunction related to PD and AD-related disease proteins (α-synuclein, Aβ and tau), in order to put these in perspective and suggest future directions.

NEUROPATHOLOGICAL EVIDENCE OF ASSOCIATION OF ALZHEIMER’S DISEASE-TYPE PATHOLOGY WITH LEWY PATHOLOGY IN PARKINSON’S DISEASE

In the pre-α-synuclein era when assessment of cortical Lewy bodies was possible, but more difficult and less reliable, some studies found Alzheimer-type lesions as a correlate of dementia in PD [9–12]. With the discovery of α-synuclein as a key component of Lewy bodies, Lewy neurites and other lesion types [13] the subsequent introduction of α-synuclein immunohistochemistry, the tide turned, and several studies favoured cortical Lewy pathology as the main (and almost sole) neuropathological correlate of cognitive dysfunction in PD. However, in the last decade, a number of clinico-pathological studies consistently showed that coexisting Alzheimer pathology is significantly associated with cognitive dysfunction in PD, both in terms of increased risk and shorter time interval from disease onset to the development of dementia. All these studies have been extensively reviewed [2, 3] and are summarized in Table 1 [9–32]. In short, large studies have identified that, besides cortical Lewy-type pathology, Aβ plaque pathology is a determinant of cognitive impairment in PD as Aβ deposition has been shown to be associated with the risk and timing of developing dementia [22, 29] and with disease duration [30]. Others have identified tau pathology as the determinant of progression to dementia [32]. These discrepancies as to the predominating role of Aβ or tau most probably are the consequence of methodological differences (for instance, including all Aβ plaque forms, such as diffuse and mature plaques [25, 27], vs. only accounting neuritic plaques as Aβ pathology [31]).

NEUROPATHOLOGICAL EVIDENCE OF CO-EXISTING LEWY-TYPE LESIONS IN ALZHEIMER’S DISEASE

Similar to co-existing AD-type pathology in PD, Lewy-type pathology has also been widely studied in AD. Interestingly, the relationship between Lewy pathology and AD attracted the interest of investigators before the actual finding of specific Lewy pathology in AD, since research on the so-called non-amyloid component of plaques (NACP) [33] started long before the identification of alpha-synuclein as the main constituent of Lewy bodies [13]. Subsequently, several studies have consistently shown that both in sporadic and in genetically determined AD (such as in PSEN1 familial AD and in Down’s syndrome) Lewy pathology is common, particularly in the amygdala, but also in the olfactory bulb, as summarized in Table 2 [34–38].

SUMMARY OF CO-PATHOLOGY IN PARKINSON’S DISEASE AND ALZHEIMER’S DISEASE

The concurrence of Alzheimer and Lewy pathologies in structures such as the amygdala and the olfactory bulb, which are commonly affected in both conditions (i.e., PD and AD) is scientifically intriguing, and, as the aforementioned co-existence of Alzheimer and Lewy pathologies, is in keeping with the experimental evidence supportive of pathological synergism. Thus, these proteins have been shown being capable of cross seeding and promoting each other’s aggregation [39], most probably not in all instances, but specifically when some protein strains are present [40]. While these experimental works are not free of criticism (mostly regarding as to what extent they can translate to what actually happens in humans and in disease), they provide a basis for further studies to understand how these proteins form disease-associated aggregates and, ultimately test specific anti-protein-aggregation agents. Discussion of such experimental studies is beyond the scope of this review and we refer to reviews published elsewhere [3].

NEUROPATHOLOGICAL EVIDENCE OF SYNAPTIC DYSFUNCTION IN PD AND AD

Synaptic dysfunction is a relatively new player in the field, since it is not as easily assessable as protein aggregation, for which immunohistochemistry provides a robust tool, albeit not devoid of limitations.

Lewy body disorders can be considered as a clinicopathological spectrum encompassing PD, PD-dementia (PDD) and dementia with Lewy bodies (DLB), rather than a group of truly distinct conditions. Across this spectrum, the use of non-conventional light microscopy techniques, has allowed for sensitive and selective detection of presynaptic α-synuclein aggregates and visualization and semi-quantitation of post-synaptic dendritic spines. For instance, in a study applying the paraffin-embedded tissue (PET) blot and the protein aggregate filtration (PAF) assay, Kramer and Schulz-Schaeffer observed with the PET blot a large amount of very small α-synuclein aggregates, which, using the PAF assay, were most frequently found in presynaptic terminals. This finding was mirrored by an almost complete loss of postsynaptic dendritic spines, in sharp contrast to the relatively small amount of cortical Lewy bodies, particularly compared to the severity of cognitive impairment seen in PDD and DLB [41]. Accordingly, these authors proposed presynaptic α-synuclein aggregates and the loss of dendritic spines as critical events for neurodegeneration in Lewy-related disorders [41, 42].

Also focusing on samples of DLB cases, Colom-Cadena and co-workers applied a microscopy technique called array tomography (which combines ultrathin tissue sections with immunofluorescence to visualize and quantify small structures such as the synapses) to assess presynaptic phosphorylated α-synuclein in the cingulate cortex and striatum from 5 DLB cases and compared them to 5 AD and 5 control cases. These authors found that 19% to 25% of phosphorylated α-synuclein aggregates were in presynaptic terminals with synaptic terminals co-localizing with these small aggregates being larger than terminals without such aggregates. There was also a gradient in the presence of phosphorylated synaptic α-synuclein aggregates, with their greater presence presynaptically suggesting a primary role for the presynaptic compartment [43].

Other authors have aimed at assessing other synaptic alterations such as suboptimal energy metabolism, and oxidative and endoplasmic reticulum stress damage in preclinical PD by means of studying incidental Lewy bodies [44]. Finally, it remains a matter of debate to what extent levodopa influences synaptic dysfunction in PD, as for decades many have made observations supportive of the notion that levodopa is harmful [45], whereas others have not [46].

Synaptic dysfunction is also considered in the pathophysiology of AD. In this vein, loss of dendritic spines has been correlated with loss of synaptic function [47–49]. Intriguingly, Aβ, both in its insoluble (larger aggregates, filaments) and its soluble (oligomers) forms, has been suggested to precede and lead to dysfunction of dendritic spines in experimental and pathological studies by a number of mechanisms ranging from reduced spine formation, stability and plasticity (inhibited long-term potentiation and enhanced long-term depression), to abnormalities in synaptic scaffold proteins and impaired organelle transport [50–56]. Tau hyperphosphorylation and microglia activation, which according to the amyloid cascade hypothesis are events secondary to Aβ pathology, appear to contribute to spine failure in AD as well [57]. Recently the postsynaptic protein neurogranin has been found to be reduced in brain tissue in AD [58].

Therefore, synaptic dysfunction in PD, DLB and AD, appears to be an attractive target both for improving knowledge of disease mechanism and developing new therapies, since preserved synaptic spines have been in turn linked to resilience against neurodegeneration [59]. Should the synaptic failure hypothesis hold true, it would theoretically be possible to revert synaptic dysfunction. However, unlike Aβ and tau pathologies, its assessment neuropathologically is not straightforward, since this requires the aforementioned sophisticated methodologies. In terms of biomarkers (see next sections) it is controversial whether available biomarkers (such as Aβ, tau and α-synuclein) could be used as indirect indicators of synaptic dysfunction or more specific markers (as proper synaptic proteins) should be used. Thus, synaptic dysfunction to date remains investigational and awaits further studies, both neuropathologically and with biomarkers, particularly in terms of the similarities that synaptic dysfunction might have between PD and AD.

BIOMARKER EVIDENCE OF UNDERLYING PROTEINOPATHY AND SYNAPTIC DYSFUNCTION IN PD

One of the main aims of research in biomarkers in neurodegenerative disorders such as PD and AD is to obtain information about the underlying neuropathology in vivo early in the disease process as opposed to traditional post-mortem neuropathological assessments, which most often provide information about end stage disease. There are several different types and sources of biomarkers for both PD and AD, but those that most directly reflect (or at least aim at reflecting) underlying pathology are CSF and positron emission tomography (PET) biomarkers.

In PD the obvious choice as either CSF or PET mar-ker is α-synuclein. Over the last decade the number of studies on the levels of different α-synuclein species in CSF (mostly total and oligomeric) has rapidly increased, albeit with remarkable inconsistencies, most likely related to several pre-analytic and analytic factors. However, overall the trend is that CSF total α-synuclein levels are lowered in PD and other synucleinopathies vs. controls and other neurodegenerative conditions [60, 61], with the opposite occurring with CSF levels of oligomeric α-synuclein [62]. This notwithstanding, the interpretation of CSF markers appears to be more difficult in terms of PD-related cognitive impairment. Thus, few studies have found that CSF levels of oligomeric α-synuclein also tend to increase in PDD and DLB [63, 64] (that is, consistent with its trend as a diagnostic marker), but CSF total α-synuclein has shown conflicting results, with a number of cross-sectional and longitudinal studies having even suggested that high (instead of low) CSF total α-synuclein might be a correlate of cognitive impairment [64–66]. All these findings have led to speculations that low CSF total α-synuclein might be a diagnostic marker in the setting of either sequestration of α-synuclein within the intraneuronal aggregates, or a compensatory reuptake of the protein to maintain the synaptic homeostasis. Conversely, as disease progresses and there is greater neuronal damage and cell death, the levels would increase due to the leakage of the proteins from the intracellular space to the CSF. How this would relate to the CSF levels of the AD-related proteins (tau and Aβ) in PD is not straightforward. CSF total α-synuclein has been reported to correlate positively with both CSF Aβ and CSF tau levels [63, 66], but low CSF Aβ has been consistently associated with poor cognitive outcome [67–69], whereas CSF tau has been reported to be either normal or low [63] in early disease stages, but increased in a proportion of late stage PDD cases [70, 71]. Therefore, in PD low CSF Aβ levels, as in AD, might reflect sequestration of Aβ in extracellular parenchymal Aβ deposits (senile plaques), while CSF total α-synuclein levels would range from low to increased paralleling what happens with CSF tau and reflecting increasing neuronal loss.

Alternatively, all these trends and correlations might be unrelated to aggregation and deposition of these proteins and their trafficking from the intra or extracellular space to CSF, and rather reflect other processes, for instance synaptic dysfunction, as previously mentioned. Yet, this view would be challenged by PET marker studies, which are available and reasonably reliable for Aβ [72] and tau [73], but not yet for α-synuclein. Hence, to date published data of studies on Aβ imaging in PD and DLB have ranged from negligible uptake in PD and moderately increased binding in DLB [74, 75] to more consistently showing a correlation of Aβ imaging and CSF Aβ levels longitudinally with cognitive outcome in PD [76, 77]. More recently, similar data emerged for tau in PD and DLB in two independent studies, albeit the tau PET uptake correlated with amyloid imaging only in one of the studies and not in the other [78, 79]. Therefore, if molecular imaging of Aβ and tau is showing anatomically that there are Aβ and tau lesions in the brains of PD and DLB patients and PET and CSF findings are significantly correlated, it is reasonable to presume that CSF and PET Aβ and tau markers are reflecting, at least partly, the underlying pathology. Few reports of autopsy findings in patients, having previously undergone CSF or PET studies, would also support this notion [68, 81], but caution is still needed with tau imaging, as a recent autopsy report has shown the presence of off-target binding (neuromelanin, choroid plexus, haemorrhages) for the tau PET tracer 18F-AV-1451 [82].

In summary, to date the published CSF and PET studies are overall in keeping with the aforementioned neuropathological studies in that a remarkable proportion of PD patients have conjoint Lewy and Alzheimer pathologies, and that these clinically correlate with cognitive impairment.

This leaves open the question for specific markers of synaptic dysfunction in PD. In this area, the evidence is very limited, with the available information to date coming from proteomic approaches and hypothesis-driven studies [83–86]. In a CSF proteomic study synaptic markers, among other proteins, were detected to differ between different forms of atypical parkinsonism, PD and controls [83]. A subsequent meta-analysis of 27 proteomic studies, which found a total of 500 differentially expressed proteins, concluded that presynaptic proteins involved in vesicle membrane fusion such as SNAP25 could potentially be used as biomarkers for PD [84]. In this vein, a post-mortem study has found associations of cognitive decline in DLB and AD with levels of Rab3A in the inferior parietal lobe and those of SNAP25 in the prefrontal cortex, respectively [85]. The same research group recently published a study of these proteins in CSF and found increased CSF levels of SNAP25 and nejurogranin, which correlated with cognitive and motor symptom severity [86].

A summary of published sensitivities and specificities of α-synuclein markers is provided in Supple-mentary Table 1.

BIOMARKER EVIDENCE OF UNDERLYING PROTEINOPATHY AND SYNAPTIC DYSFUNCTION IN AD

In AD as in PD the accumulated evidence of bio-markers of α-synuclein pathology is indeed restricted to CSF studies, since, as already discussed there is not as yet any validated PET probe specific for α-synuclein. Studies available to date have also displayed discrepancies regarding CSF total α-synuclein. Thus, some studies have found no differences in CSF total α-synuclein between synucleinopathies (PD and DLB) and AD [87–89], whereas others have shown an association between low CSF total α-synuclein levels in AD and scores of a global cognition test such as the mini mental state examination test, suggesting that it constituted a general marker of synapse loss [8]. Yet, several published reports have pointed towards increased levels of CSF total α-synuclein in AD [90–93], linking it to aggressive neurodegeneration in this condition, in a similar way to high levels of CSF tau and 14-3-3 proteins in the setting of aggressive neuronal death as seen in Creutzfeldt-Jakob disease or AD itself.

Regarding CSF indicators of synaptic dysfunction in AD, synaptic proteins partly overlapping with those mentioned above in relation to PD had been extensively in AD before they were investigated in PD. Accordingly, there are several studies which reported increased CSF levels of neurogranin [94–96], synaptotagmin [97], and contactin [98] in AD both in its clinically manifest phase and its prodromal stage as reflected by mild cognitive impairment with biological evidence of underlying AD (that is, CSF tau and Aβ abnormalities), suggesting these might be independent and complementary biomarkers of AD [99–101] Accordingly, a recent meta-analysis recommends including neurogranin in the panel of AD biomarkers [102]. Nevertheless, there are outstanding issues regarding specificity, since as happens with proteins such as tau, increased CSF levels neurogranin might merely reflect neuronal damage in aggressive conditions such as Creutzfeldt-Jakob disease [103].

As for synaptic CSF makers in PD, recently CSF levels of neurogranin have been assessed in parkinsonian disorders, with the finding that these were reduced in PD, PDD, MSA and PSP relative to AD and controls, not correlating with motor or cognitive measures, though [104]. By contrast, in another study increased neurogranin CSF levels mirrored reduced CSF Aβ in PD and in this case a significant correlation with cognition (as measured by MMSE) was reported [105]. Hence more studies are needed to elucidate the actual associations of these synaptic markers in degenerative parkinsonian disorders.

Currently these markers are being explored not only in CSF, but also in blood exosomes, which would provide a more accessible source relative to CSF [106].

An overview of published sensitivities and specificities of τ and Aβ markers is summarized in Supple-mentary Table 1.

OTHER BIOMARKERS IN AD AND PD RESEARCH

Although it is not in the scope of this review, the increasing interest in neurofilaments and markers of neuroinflammation as biomarkers in both AD and PD, needs also to be mentioned. Neurofilament has been identified as a marker of disease progression or prognostic marker in several neurological conditions from multiple sclerosis [107] to amyotrophic lateral sclerosis [108] and, importantly also in both AD [109] and PD [110]. A major breakthrough in the research of this biomarker has been the demonstration that its levels in plasma significantly correlate with those in the CSF [111], making it a much more accessible biomarker. As for markers of neuroinflammation, there is research of both neuronal-specific (YKL-40 [112]) and non-specific markers (cytokines [113]) as diagnostic and progression biomarkers in AD and PD

CONCLUSIONS

There is compelling evidence that PD and AD share neuropathological hallmarks in that Aβ and tau lesions of the Alzheimer-type are common in PD and, vice versa, α-synuclein Lewy-type aggregates are frequent findings in AD. Modern non-conventional techniques overcoming limitations of routine immunohistochemical techniques are promising as to take further our knowledge of the impact of these disease-associated proteinaceous aggregates beyond the neurons’ soma, down to their presynaptic and postsynaptic terminals, with potential mechanistic and even future therapeutic implications.

An even greater challenge is translating this knowledge to the clinic. CSF and PET markers of Aβ and tau work reasonably well in the AD field, but their counterparts in PD are far from being equally reliable, with new promising approaches being those of aggregometric techniques such as real time quaking induced conversion (RT-QuIC) [114]. In terms of PET markers, beside the fact that there is no PET probe available for α-synuclein yet, the AD PET markers range from consistent evidence of their specificity (amyloid imaging) to greater uncertainty of their reliability due to off-target binding (tau imaging). CSF synaptic markers are attractive, but evidence is still scarce and most probably these will be non-specific markers of disease progression. For all of these CSF and PET markers, one should remember that ‘markers are not always makers’, and therefore caution is needed when interpreting associations as causative.

In summary and coming back to the question raised in the title of this review (what are the relevant similarities between PD and AD? the protein aggregates? synaptic dysfunction? or both?), from a neuropathological point of view protein aggregates are there both at the soma and the synapse. Thus, a number of CSF and PET biomarkers might capture these different aspects of protein-related neurodegeneration. More specifically, CSF α-synuclein, tau and Aβ levels might reflect beside underlying protein aggregates also the soluble fractions of these proteins at the synapse level (Table 3).

CONFLICT OF INTEREST

The authors have no conflict of interest to report.