Neuropathology, Genetics and Biomarkers of Synucleinopathies


 Synucleinopathies are clinically and pathologically heterogeneous disorders characterized by pathologic aggregates of α-synuclein in neurons and glia, in the form of Lewy bodies, Lewy neurites, neuronal cytoplasmic inclusions, and glial cytoplasmic inclusions (GCIs). Synucleinopathies can be divided into two major disease entities: Lewy body disease (LBD) and multiple system atrophy (MSA). Common clinical presentations of LBD are Parkinson's disease (PD), PD with dementia (PDD), and dementia with Lewy bodies (DLB), while MSA has two major clinical subtypes, MSA with predominant cerebellar ataxia (MSA-C) and MSA with predominant parkinsonism (MSA-P). There are currently no disease-modifying therapies for the synucleinopathies, but elucidation of genetics and mechanisms of α-synuclein conversion to pathologic oligomers and insoluble fibrils offer hope for eventual therapies. It remains unclear how α-synuclein can be associated with distinct cellular pathologies (e.g., Lewy bodies and GCI) and what factors determine neuroanatomical and cell type vulnerability. Accumulating evidence from in vitro and in vivo experiments suggests that α-synuclein species derived from LBD and MSA are distinct "strains" having different seeding properties. Recent advancements in in vitro seeding assays, such as real-time quaking-induced conversion (RT-QuIC) and protein misfolding cyclic amplification (PMCA), not only demonstrate distinct seeding activity in the synucleinopathies, but also offer exciting opportunities for molecular diagnosis using readily accessible peripheral tissues. Cryogenic electron microscopy (cryo-EM) structural studies of α-synuclein derived from recombinant or brain-derived filaments provide new insight into mechanisms of seeding in synucleinopathies. In this review, we describe clinical, genetic and neuropathologic features of synucleinopathies, including a review of classification and staging schemes for LBD. We also review evidence supporting the existence of distinct α-synuclein strains in LBD and MSA.


Historical information
James Parkinson published a short monograph, "An Essay on Shaking Palsy," in 1817 [1]. He described six patients who presented with tremors, slow movements, and falls. His work received little attention for decades until Jean-Martin Charcot introduced it in his lectures in the late 1800's. Charcot added rigidity as a prominent symptom and pointed out that the disability is not caused by palsy; he proposed to call the disease Parkinson's disease (PD). In 1912, Friedrich H. Lewy reported the presence of intracellular inclusion bodies in the dorsal nucleus of the vagus nerve, nucleus basalis of Meynert, and the thalamic paraventricular nucleus of the brains of PD patients. Gonzalo Rodriguez Lafora also examined one patient with PD and con rmed the existence of neuronal inclusions in 1913. Konstantin Tretiakoff reported neuronal loss and neuronal inclusions in the substantia nigra in 1919, and it was proposed that these inclusions be called Lewy bodies [2][3][4]. In 1961, Haruo Okazaki reported two patients with progressive dementia without parkinsonism who had abundant Lewy bodies in the neocortex [5]. Kenji Kosaka subsequently found similar inclusions in the neocortex of patients with cognitive impairment and extrapyramidal symptoms and proposed the term diffuse Lewy body disease for this disorder [6].
Subsequently, an international consortium led by Ian McKeith coined the term dementia with Lewy bodies (DLB) to refer to this clinicopathologic disorder [7]. Lewy bodies are now considered a pathological hallmark of PD, DLB, and PD dementia (PDD).
Multiple system atrophy (MSA) is a distinctive synucleinopathy that includes two major clinicopathologic subtypes previously thought to be separate disorders: olivopontocerebellar atrophy (OPCA) and striatonigral degeneration (SND). The term Shy-Drager syndrome was originally used in 1960 to describe a neurological disorder associated with orthostatic hypotension of unknown etiology, but the pathologic descriptions by Shy and Drager clearly indicated a disorder affecting multiple systems beyond the autonomic nervous system [8]. Accumulating evidence indicates that the majority of patients with SND have some degree of OPCA and vice versa. Moreover, progressive autonomic failure corresponding to Shy-Drager syndrome is common in both SND and OPCA. Therefore, Graham and Oppenheimer considered these variants to be the same disorder and coined the term MSA in 1969 [9]. Although there had been some debate about whether MSA was a single disease, the discovery of argyrophilic glial cytoplasmic inclusions (GCI) in SND and OPCA provided strong evidence that they were part of a disease spectrum [10].
Seminal studies in the early 1990's in the laboratory of Tsunao Saitoh discovered a non-amyloid component in senile plaques (NACP) [11] using biochemical methods, and later showed that NACP was a synaptic protein with genetic locus on chromosome 4q21.3-q22 identical to α-synuclein [12]. Previous studies had de ned α-synuclein as one member of a family of presynaptic proteins (α-, β-and γsynuclein) with a role in membrane-associated processes at the presynaptic terminal (reviewed in [13]). In 1997, Polymeropoulos discovered a mutation in the gene for α-synuclein (SNCA) in Southern Italian families of Greek heritage with PD [14]. Subsequently, Spillantini and colleagues reported that α-synuclein was the major component of Lewy bodies in sporadic PD and DLB patients [15]. These ndings combined pathological and genetic insight and established a disease concept, "synucleinopathy." Wakabayashi and colleagues reported that the major protein component in GCI was also α-synuclein, linking MSA and Lewy body disease as the two major clinicopathologic subtypes of synucleinopathy [16].

Genetics of LBD
A direct causal relationship between α-synuclein and PD was rst established in 1997 with the discovery of a single point mutation in the SNCA gene resulting in a non-synonymous amino acid substitution p.A53T, producing familial early-onset PD [14]. Subsequently, several other substitutions in the αsynuclein protein (e.g., p.A18T, p.A29S, p.A30P, p.E46K, p.G46K, p.H50Q, p.G51D, and p.A53E) have been linked to autosomal dominant familial PD [54][55][56][57][58][59]. Furthermore, genomic multiplications (i.e., duplication and triplication) of SNCA cause familial PD with extramotor features, including dementia [60,61]. This seminal work suggested that overexpression of wild-type α-synuclein could be pathogenic and that the severity of the phenotype may be dose-dependent. For example, the clinical phenotypes tended to be more severe in families with SNCA triplication than SNCA duplication (e.g., more non-motor features in triplication families). Importantly, this observation also nominated reduction of α-synuclein levels as a possible therapeutic strategy for synucleinopathies.
In addition to SNCA pathogenic mutations in other genes have been associated with familial PD, including LRRK2, VPS35, PARKIN, PINK1, DJ-1, and VPS13C [62, 63]. As DNA sequencing technologies have revolutionized population genetics over the last decade, an increasing number of other genes have been nominated for PD (e.g., CHCHD2, LRP10, TMEM230, DNAJC13) but these genetic loci lack de nitive con rmation [63]. Even though the discovery of familial PD-related genes has contributed to dissecting the etiology of the disease (e.g., dysfunction of lysosomal and mitochondrial pathways), they still only account for a relatively small proportion of the genetic risk in PD. A genetic hypothesis for PD is that the disease develops through a complex interplay of low penetrant genetic risk variants and unknown environmental determinants.
Population-based approaches to resolve the genetic architecture of PD have focused on common variant associations at candidate genes including, the SNCA and MAPT loci. Large unbiased studies characterizing patterns of linkage disequilibrium (e.g., the HapMap project) across the genome and high throughput genome-wide genotyping platforms, which simultaneously genotype hundreds of thousands of common single nucleotide variants have been performed. Early genome-wide association studies (GWAS) con rmed associations at the SNCA and MAPT loci, and highlighted novel associations at LRRK2, and PARK16 loci as risk factors for PD in both European and Japanese cohorts [64,65]. Of note, the MAPT locus was identi ed as a risk locus only in European cohorts, while the BST1 locus was identi ed as a risk locus only in Japanese cohorts [64,65]. Additional loci have been nominated in Eastern Asian populations with GWAS methods [66]. The latest PD GWAS in Caucasian populations nominated 90 independent loci covering 78 genomic regions [67], with the SNCA locus showing the strongest signal. Although less well-characterized or studied, recent GWAS and whole-genome sequencing efforts in Lewy body dementia (i.e., PDD and DLB) in Caucasian populations have nominated at least ve loci, including three that are also risk loci for PD (SNCA, GBA and TMEM175) and two other (APOE and BIN1) known risk loci for AD [68,69]. These ndings indicate that Lewy body dementia shares genetic risk factors of both AD and PD, as may be expected from its mixed amyloid and α-synuclein pathology.
Another important genetic determinant of susceptibility to both PD and Lewy body dementia is variation in the GBA gene [70,71]. Recessive mutations in GBA result in a lysosomal storage disorder known as Gaucher's disease, however, astute clinical observations noted that heterozygous carriers were at increased risk of PD and Lewy body dementia [72]. GBA is a GWAS locus for both disorders, but also is an example of a gene that harbors variants of differing penetrance, with rare variants also driving disease risk. GBA is the only signi cant nding in a gene burden analysis in a whole-genome sequence study of Lewy body dementia [39]. The role that GBA susceptibility variants may play in MSA remains controversial [73,74].

Genetics of MSA
MSA is considered a sporadic disease; only a few familial MSA cases have been reported [75]. Several small multiplex families from Japan nominated mutations of the COQ2 gene for MSA [76]. In addition, a common substitution in COQ2, p.V393A, and several rare variants were associated with sporadic MSA [76]. The COQ2 gene encodes an enzyme involved in synthesis of coenzyme Q10, and the p.V393A variant results in lower production of coenzyme Q10. Several studies from non-Asian countries, however, failed to replicate the ndings [77,78]. A meta-analysis of Eastern Asian populations con rmed an association of COQ2 p.V393A variant with MSA (odds ratio [OR] 2.05; 95% CI 1.29-3.25, p = 0.002) [79]. Interestingly, a subgroup analysis revealed that the association was signi cant for MSA-C (OR 2.75, 95 % CI 1.98-3.84, p < 0.001), but not for MSA-P (OR 1.25, 95 % CI 0.64-2.46, p = 0.51). This nding may partially explain the predominant subtypes of MSA in Japan and Western countries. MSA-C is the predominant phenotype in Japan, while MSA-P is the predominant subtype in Western countries [80,81].
Some studies reported decreased levels of coenzyme Q10 in the cerebellum of MSA, suggesting potential association between alterations of coenzyme Q10 activity and cerebellar pathology [82,83].
The largest GWAS, which enrolled 918 MSA patients of European ancestry and 3,864 controls, did not nd any common genetic association [84]. This GWAS identi ed 4 potential risk loci, including single nucleotide polymorphisms in the genes FBXO47, ELOVL7, EDN1, and MAPT; however, these ndings could not be replicated in a GWAS using 906 MSA patients and 941 unrelated healthy controls of the Han Chinese population [85]. These discrepancies could be partially explained by ethnic differences, as GWAS demonstrated some differences in the genetic contribution to PD between the European and Asian populations [66]. Large-scale whole-genome sequencing efforts are underway in MSA, PD and Lewy body dementia, and it is likely that efforts of global consortia will be needed to fully resolve the role of genetics in synucleinopathies.

Neuropathologic features of LBD
In most cases of LBD without cognitive de cits, the macroscopic ndings are comparable to age-and sex-matched controls, except for loss of neuromelanin pigment in the substantia nigra and locus coeruleus (Fig. 2). Dopaminergic neuronal loss in the substantia nigra, particularly in the ventrolateral part, is a pathologic hallmark of PD [86,87]. The severity of neurodegeneration of the substantia nigra correlates with severity of extrapyramidal motor symptoms and the degree of striatal dopaminergic de ciency [88,89]. Neuronal loss is moderate to marked in PD and PDD, but more variable in DLB. In fact, a subset of DLB patients lack parkinsonism and have preserved neuronal population in the substantia nigra. The locus coeruleus is a major noradrenergic nucleus, and neuronal loss leads to de ciency of noradrenaline, which may contributes to various symptoms, including cognitive impairment, affective symptoms, RBD, and gait di culties [90].
Lewy bodies are round, eosinophilic inclusions in neuronal perikarya. There are two types of Lewy bodies: classical (or brainstem) type and cortical type. Classical-type Lewy bodies have a dense hyaline appearance with a peripheral clear halo and are easily visible on hematoxylin and eosin (H&E) stained sections (Fig. 3). Cortical type Lewy bodies have a less compact appearance and are more di cult to detect with histologic methods (Fig. 3) [5]. Both types of Lewy bodies are strongly immunoreactive with antibodies against α-synuclein, especially antibodies targeting phospho-Ser129 (Fig. 3) [53]. In addition to α-synuclein, more than 90 components of Lewy bodies have been reported, based mostly upon immunohistochemical colocalization with brainstem type Lewy bodies, including sequestration of neurotransmitter enzymes of cholinergic and dopaminergic neurons [91][92][93]. Accumulation of phosphorylated α-synuclein also occurs within cell processes (mostly axonal), so-called Lewy neurites (Fig. 3). Lewy neurites in the CA2/3 sectors of the hippocampus (Fig. 3) are a characteristic histopathologic nding in many cases of PD and most cases of PDD and DLB [94]. Deposits of phosphorylated α-synuclein are observed less frequently in oligodendroglia, and rarely in astrocytes in the midbrain and basal ganglia [95].
In addition to Lewy bodies and neuronal loss in the substantia nigra, a subset of cases have spongiform change or neuropil microvacuolation that is often most severe in the amygdala, but also seen in limbic and superior temporal cortices (Fig. 3) [96-98]. The most important co-pathology in LBD is Alzheimertype pathology. In initial reports of DLBD, the main neuropathologic features included not only numerous cortical Lewy bodies, but also numerous senile plaques and neuro brillary tangles in the cerebral cortex [6]. The majority of LBD cases, not exclusively DLBD, have some degree of Alzheimer-type pathology (i.e., neocortical senile plaques and neuro brillary tangles) [99][100][101][102]. Senile plaques in the cerebral cortex are common, and they often are characterized by non-neuritic diffuse amyloid deposits [103]. Of note, 28% of DLB and 10% of PDD cases have su cient pathology for a secondary neuropathologic diagnosis of AD [104].
Lewy bodies are also found frequently in cases with advanced AD, particularly in the amygdala [105][106][107][108]. Hamilton screened α-synuclein pathology in 145 cases of AD and found that 88 cases (61%) had LBD [106]. The amygdala was the most affected region; Lewy bodies were present in all cases of AD with LBD. Interestingly, some of the cases had numerous Lewy bodies in the amygdala but rare or absent in the brainstem in some cases (amygdala-predominant Lewy bodies [ALB]). Uchikado and colleagues also screened Lewy-related pathology in 347 cases of AD and found that 62 cases (18%) were consistent with ALB [108]. Of those, Lewy bodies were only found in the amygdala ("amygdala-only" LBD) in 32 cases (9%). The clinical signi cance of ALB in AD remains uncertain [109], but evidence suggests that they may be associated with increased frequency of visual hallucinations compared to AD without amygdala Lewy bodies [110].
Lewy bodies are widely distributed not only in the central nervous system but also in the peripheral autonomic nervous system, including the nerve terminals and autonomic ganglia in the heart, submandibular glands, enteric nervous system, adrenal glands, skin, and cutaneous nerve [111][112][113][114][115][116][117][118]. Interestingly, phosphorylated-α-synuclein deposits in cutaneous autonomic nerves can be detected by immunohistochemistry in 56-82% of patients with IRBD, a prodromal phase of synucleinopathies [119][120][121][122][123]. The presence of α-synuclein was associated with greater autonomic dysfunction in IRBD [122]. A biopsy taken from these peripheral tissues might be a feasible biomarker in the prodromal phase of synucleinopathies, and increasingly sensitive and speci c methods to detect abnormal α-synuclein in peripheral tissues (e.g., real-time quaking-induced conversion [RT-QuIC] or protein misfolding cyclic ampli cation [PMCA], see below) offer hope for peripheral biomarkers for the disease. [124].

Classi cation of LBD
The term "Lewy body disease" was coined by Kosaka and colleagues to refer to neurodegenerative diseases with numerous Lewy bodies in the central nervous system [20]. He classi ed LBD into three groups: groups A, B, and C, which were later named diffuse type (DLBD), transitional type (TLBD), and brainstem type (BLBD) [125]. BLBD corresponded to PD in their scheme, and DLBD was thought to be an extension of PD pathology into the limbic lobe and the neocortex. DLBD was separated into two forms: a common form and pure form. The common form had not only numerous Lewy bodies, including neocortical Lewy bodies, but also many senile plaques and variable neuro brillary tangles. In contrast, the pure form had few or no Alzheimer-type changes [99]. Kosaka later added a "cerebral type" of LBD, which had numerous Lewy bodies in the cerebral cortex and amygdala, but minimal or no Lewy bodies in the brainstem and diencephalon [126].
The First International Consortium for Lewy Body Dementia (ICDLB) proposed criteria for clinical and pathologic diagnosis of DLB [7]. Subtypes of Lewy-related pathology were categorized based upon the severity and topographical distribution of Lewy bodies [127]: diffuse neocortical, limbic (transitional), and brainstem-predominant, which corresponded roughly to Kosaka's classi cation of DLBD, TLBD, and BLBD. The Third ICDLB report developed a diagnostic scheme that was devised to predict likelihood that the pathology would be associated with DLB. The criteria took into account both the extent of Lewyrelated pathology and Alzheimer's-type pathology to assign a probability that the pathology would be associated with the clinical presentation ( Table 1). The severity of Lewy-related pathology was semiquantitatively assessed on a ve-scale: 0 = none, 1 = mild, 2 = moderate, 3 = severe, and 4 = very severe. Recommended brain regions for assessment included the dorsal motor nucleus of vagus, locus coeruleus, and substantia nigra in the brainstem regions, nucleus basalis of Meynert, amygdala, transentorhinal cortex, and cingulate cortex in the limbic regions, and temporal, frontal, and parietal cortices (Fig. 4). The likelihood of DLB clinical was directly related to severity of Lewy body pathology and indirectly related to severity of Alzheimer pathology. It was recognized from studies of prospective cohorts that when Alzheimer pathology was severe, most patients had Alzheimer type dementia, rather than DLB [128]. A recent study validates this approach in that the diagnostic sensitivity for probable DLB was signi cantly higher in TLBD and DLBD without neocortical tangles than in those with neocortical tangles [129]. These ndings indicate that the phenotypic expression of DLB is associated directly related to the extent of Lewy bodies and inversely related to the extent of neuro brillary tangles [129].   Braak and colleagues proposed a staging scheme for Lewy-related pathology in PD [130], which was anatomically more detailed and speci ed than Kosaka's classi cation of LBD (Fig. 4). Lewy-related pathology initially occurred in the medulla oblongata (dorsal motor nucleus of vagus and the glossopharyngeal nucleus), and in the anterior olfactory nucleus of stage 1. In stage 2, α-synuclein pathology ascends to the pontine tegmentum, while stage 3 is associated with involvement of midbrain, stage 4 with limbic regions, and stages 5 and 6 with neocortical involvement. This staging scheme has largely been con rmed in several studies, but some exceptions have also been pointed out, such as cases with Lewy bodies restricted to the olfactory bulb or to the amygdala, particularly in cases with advanced AD pathology (i.e., ALB). [108, 131,132].
The BrainNet Europe Consortium assessed 33 LBD cases by 22 pathologists and pointed out low interrater agreement (65%) for the Braak staging scheme. Only four cases of Braak stage 6 reached 100% agreement [133]. They proposed a new protocol to improve the inter-rater reliability by (1) selecting nine blocks for α-synuclein immunohistochemistry, (2) using a dichotomous approach to assess Lewy-related pathology (i.e., present vs. absent), and (3) incorporating an amygdala-predominant category (Fig. 4).
This new protocol achieved a high inter-rater agreement for both Braak Lewy body stage, as well as assignment of brainstem, limbic, neocortical, and amygdala-predominant categories.
Beach and colleagues applied Braak Lewy body staging to 216 cases of LBD, and found about half of their cases were unclassi able [132]. About two-thirds of unclassi able cases had limbic system involvement without signi cant brainstem pathology, and the remaining one-third had Lewy-related pathology only in the olfactory bulb. Based on their ndings, they devised a staging scheme to include LBD con ned to the olfactory bulb and limbic predominant cases with variable or no brainstem involvement [132]. Given that the olfactory bulb is often initially affected in both ILBD and ALB, they de ned cases of having Lewy-related pathology con ned to the olfactory bulb as stage I. Stage II was divided along two branches: brainstem predominant (stage IIa) and limbic predominant (stage IIb). In the majority of cases, Lewy-related pathology passed through the brainstem stage prior to the limbic stage (stage IIa), while most cases of ALB showed abundant Lewy-pathology in the limbic system, particularly in the amygdala, before affecting the brainstem. Following the two pathways of stage II, both the brainstem and limbic systems converged at stage III, and the neocortical regions are affected in stage IV.
Subsequently, Adler and colleagues applied this staging system to 280 cases of LBD and found that parkinsonism, cognitive impairment, hyposmia, and RBD were signi cantly correlated with increasing stage [134]. Based on these various schemes, the ICDLB criteria were revised in 2017 to include amygdala-predominant and olfactory bulb only types (Fig. 4) [28].
Semi-quantitative evaluation has been used for diagnosing and classifying neurodegenerative disorders, but there are inherent weaknesses in semi-quantitative measures when it comes to inter-rater reliability [135]. Braak Lewy body stages, ICDLB criteria, and other staging schemes have been widely used, but not all cases t well into the speci ed stages, or they have features that overlap between more than one stage. Inter-rater agreement is not optimal. Recently, Attems  While most patients with PD have BLBD (or TLBD) at autopsy, a "pure form" of DLBD can also be found in a few patients with PD [99,139]. In contrast, most patients with PDD and DLB have DLBD. It has proven challenging to differentiate DLB from PDD based on the neuropathologic ndings alone [104].
Although DLB patients tend to have more severe Alzheimer-type pathology, and PDD patients tend to have more severe neuronal loss in the substantia nigra [140], there are no clear neuropathologic distinctions between DLB and PDD [141]. System-speci c neuronal loss and gliosis are observed in both striatonigral and olivopontocerebellar systems (Fig. 6). In addition to these macroscopically affected brain regions, immunohistochemistry for α-synuclein reveals more widespread α-synuclein pathology, characterized by GCI and variable NCI. GCI are argyrophilic inclusions (Gallyas silver stain positive) in the cytoplasm of oligodendrocytes [10]. They are visible on routine H&E stains, but immunohistochemistry for phosphorylated α-synuclein is far more sensitive for visualization of GCI [16]. The density of GCI correlates with both neuronal loss and disease duration [80]. The neuropathologic diagnostic criteria for MSA require "widespread and abundant GCI in association with neurodegenerative changes in striatonigral or olivopontocerebellar structures" for the de nite diagnosis of MSA. Thus, GCI is the pathologic hallmark of MSA [143]. In addition to accumulation of α-synuclein in the cytoplasm of NCI and dystrophic neurites, many affected neurons also have intranuclear α-synuclein inclusions [18]. The distribution of NCIs is distinct from that of Lewy bodies, and they are most often observed in the putamen, pontine nuclei, and inferior olivary nuclei, which are not susceptible to Lewy bodies. NCI can also be observed in the substantia nigra, cingulate cortex, amygdala, hippocampus, entorhinal cortex, hypothalamus, and neocortex [144,145]. Some MSA cases may have concurrent LBD [137]. In such cases, Gallyas-Braak silver staining is helpful in distinguishing NCI from Lewy bodies; the former are positive in Gallyas-Braak silver staining, but the latter are not [146]. The clinicopathologic signi cance of NCIs has been reviewed by Cykowski and colleagues in a large cohort of MSA cases. They found that presence of NCI in neocortex was associated with cognitive impairment [145]. Other studies have reported that the burden of NCI is associated with cognitive impairment or memory loss in MSA [37,38,147].

Subtypes of MSA
MSA has been divided into two major pathologic subtypes: SND and OPCA, although virtually all cases had microscopic involvement in both systems. Wenning and colleagues proposed a grading system for SND based upon semi-quantitative assessment of atrophy, neuronal loss, astrogliosis, and GCI: In this scheme, Grade 1 = has neuronal loss con ned to the substantia nigra; Grade 2 = has neuronal loss extending to the putamen; and Grade 3 = has involvement of the caudate nucleus and globus pallidus [148]. Ozawa and colleagues proposed three grades of SND and OPCA based upon semi-quantitative assessment of neuronal loss in regions of interest: putamen, globus pallidus, and substantia nigra for SND; pontine nuclei, cerebellar hemisphere and vermis, inferior olivary nucleus, and substantia nigra for OPCA [80]. This classi cation showed good correlation with clinical features; patients with SND had more severe bradykinesia, and those with OPCA had more frequent cerebellar signs. Jellinger and colleagues extended the Wenning grading system of SND for both SND and OPCA to assign MSA-P and MSA-C [149], which correlated with initial symptoms and clinical features of both subtypes [149].
Although the diagnostic criteria for MSA require neurodegeneration in striatonigral or olivopontocerebellar systems, or both, some MSA cases do not have signi cant neurodegeneration in even though they have widespread GCI. This rare subtype is referred to as "minimal change" MSA [150][151][152][153][154]. Cases of minimal change MSA suggest that GCI formation precedes neuronal loss. Clinical presentations of minimal change MSA vary. Some patients are asymptomatic ("preclinical MSA") [151,154], but in a case series from the UK, all minimal change MSA patients had respiratory dysfunction and early orthostatic hypotension [152]. Rarely, minimal change MSA has been reported with limbic-predominant distribution of α-synuclein pathology [153]. . It is not possible to con dently detect α-synuclein oligomers with routine histologic methods, but Roberts and colleagues applied proximity ligation assay (PLA) methods to detect α-synuclein oligomers in histologic sections [167]. The PLA technique was originally developed to increase sensitivity for protein detection and it also has been applied to detect protein-protein interactions [168,169]. In this method, combination of speci c antibody binding to antigen and ampli cation of signal with polymerase chain reaction enables in situ detection of speci c antigens.
The mechanism of homotypic PLA for α-synuclein oligomers is shown in Fig. 7. Two forms of oligonucleotides are linked to α-synuclein antibodies, which interact at a close distance only when αsynuclein oligomerizes. The oligonucleotides are joined by ligase and serve as templates for formation of circular DNA, which serves as a template in rolling-circle ampli cation to produce thousands of singlestranded products at the site of the oligomers. The oligonucleotides are nally labeled with horseradish peroxidase to produce at the site of the oligomers.
Using this emerging technique, Roberts and colleagues showed α-synuclein oligomers in the cingulate cortex and reticular formation of the medulla in PD [167]. They also detected α-synuclein oligomers in morphologically intact neurons and in the periphery of Lewy bodies. Sekiya and colleagues studied PD and MSA brains and showed widespread and abundant α-synuclein oligomers, especially in cortical neurons and Purkinje cells of MSA (Fig. 7) [170]. They also found that most α-synuclein oligomers were localized in neurons and that α-synuclein oligomer accumulation progressed in neurons but in oligodendrocytes. This technique has also been used to detect α-synuclein oligomers in peripheral tissues, such as gastrointestinal or skin biopsies [171,172]. Ruffmann and colleagues eluated gastrointestinal tissues with α-synuclein-PLA and found two staining patterns -cellular and diffuse [171]. Mazzetti and colleagues showed α-synuclein oligomers in skin biopsies of PD [172]. These results suggest that α-synuclein-PLA has potential as diagnostic biomarker. Homotypic PLA for α-synuclein oligomers also has been used to characterize α-synuclein animal experiments [173,174], permitting detection of early pathological changes.

Distinct seeding activity
As noted above, synucleinopathies are clinically and pathologically heterogeneous, but it remains unknown how the same protein can be associated with distinct pathologies (e.g., Lewy bodies and GCI) and what factors determine neuroanatomical and cell type vulnerability. Accumulating evidence suggests that distinct α-synuclein strains may be associated with different disorders. Several studies using αsynuclein extracted from MSA and LBD showed distinct seeding activities in vitro and in vivo [175][176][177].
Prusiner and colleagues reported that extracts from MSA brains, but not from PD brains, induced aggregation of α-synuclein in cultured cells expressing YFP-tagged A53T-mutated human α-synuclein.
Similarly, Woerman and colleagues demonstrated that α-synuclein from MSA postmortem brains induced protein aggregations in cultured cells, whereas α-synuclein isolated from LBD had no effect [175].
Yamasaki and colleagues also demonstrated distinct biochemical properties of α-synuclein from MSA and PD using a FRET biosensor assay based on expression of A53T α-synuclein-CFP/YFP [177]. Both soluble and insoluble fractions of MSA had robust seeding activity, while only insoluble fractions of PD had seeding activity. Moreover, the morphology of induced cellular inclusions was different in MSA and PD. Peng and colleagues demonstrated distinct seeding activity of brain-derived α-synuclein brils from GCI and Lewy bodies [176]. They treated primary oligodendrocytes that expressed α-synuclein with GCIderived α-synuclein and Lewy body-derived α-synuclein and found that GCI-derived α-synuclein was 1,000 times more potent at seeding compared with Lewy body-derived α-synuclein. Injection of GCI-derived αsynuclein induced abundant neuronal inclusions in wild-type mice, but Lewy body-derived α-synuclein did not induce neuronal inclusions at 3 months after the injection. These experiments indicated that αsynuclein derived from GCI had more potent seeding activity in vitro and in vivo than that derived from Lewy bodies, which may correlate with more aggressive disease course in MSA compared to LBD [176].

Distinct conformation of α-synuclein with cryogenic electron microscopy (Cryo-EM)
Cryo-EM has emerged in the last decade as an effective tool for structure determination [178]. The advancement of transmission electron microscope optics and software for data analysis enables threedimensional reconstruction of macromolecular assemblies at near-atomic resolution.
Several studies using cryo-EM have revealed multiple polymorphs from recombinant α-synuclein [179][180][181][182][183]. Li and colleagues investigated the full-length recombinant human α-synuclein and determined two predominant polymorphs, which they termed "rod" and "twister." Both polymorphs were composed of two proto laments with highly conserved kernel structures, but they differed in inter-proto lament interfaces [180]. The interface between the two proto laments in the rod polymorph contained residues H50-E57 from the preNAC region (Fig. 8), while the interface in the twister polymorph contained residues V66-A78 from the NACore. Notably, six missense mutations (p.E46K, p.H50Q, p.G51D, p.A53E, p.A53T, and p.A53V) that cause familial PD are located in the preNAC steric zipper (i.e., residues 46-56) in the rod polymorph. This suggests that PD-associated mutations may disrupt the preNAC zipper of bril cores in the rod polymorph, while having little impact on the stability of the twister polymorph.
Recent cryo-EM studies using recombinant full-length α-synuclein brils with missense mutations have provided more direct evidence on how these mutations affect the conformation of α-synuclein brils. Nterminally acetylated p.E46K mutant α-synuclein brils had conformational changes in the N-terminal region of the bril core [184]. The proto lament interface of p.E46K mutant bril covered V74-Q79, which differed from the interface covering H50-E57 in type 1 polymorph (Fig. 8). The p.H50Q mutation was associated with two new polymorphs: narrow brils and wide brils [182]. The narrow brils were formed from a single proto lament (proto lament A), whereas the wide brils were composed of two proto laments (proto laments A and B). The inter-proto lament interface of the wide brils had only T59 and K60 (Fig. 8). N-terminally acetylated p.A53T mutant α-synuclein brils also had a small proto lament interface, consisting of T59 and K60 with no obvious interaction (Fig. 8) [185]. The proto lament interface of the mutant bril was less stable than the wild type. These results indicate that missense mutations associated with familial PD can alter the conformation of proto laments and interproto lament interfaces, resulting in variable α-synuclein brils with distinct aggregation kinetics, seeding activity, and cytotoxicity.
Although many studies have used recombinant α-synuclein brils, evidence suggests that structures of recombinant laments assembled in vitro may be different from those derived from human brains, as has been observed with tau protein [186][187][188]. Only a few studies have been reported on α-synuclein derived from human brains. One such study determined the atomic structures of α-synuclein brils isolated from MSA brains [189], in which two different types of laments (type I laments and type II laments) were observed (Fig. 8). The ratio of type I to type II differed among MSA patients. Both laments had two different proto laments, which consisted of an extended N-terminal arm and a compact C-terminal body.
Unlike recombinant α-synuclein brils, both laments were asymmetric. The inter-proto lament interface of type I brils contained V37-A53, while that of type II contained V40-A53 (Fig. 8). Mutations in p.G51D and p.A53E, which cause atypical synucleinopathies with features of PD and MSA [58, 59,190], are located in this interface. These mutations increase the negative charge around the central cavity, which may lead to their different molecular composition [189]. The interface between the two different proto laments forms a large cavity surrounded by side chains of K43, K45, and H50 from each proto lament (Fig. 8). This cavity encloses non-proteinaceous molecules. Studies by Puentes and colleagues using computational chemistry hypothesized that non-proteinaceous density in α-synuclein brils may be poly(ADP-ribose) (PAR), a negatively charged polymer generated by PAR polymerase-1 [191]. Previous studies have shown that PAR binds to α-synuclein and accelerates α-synuclein brillization, which results in cell death via parthanatos [192]. Using PLA, they demonstrated PAR-αsynuclein interactions in post-mortem brain tissue from PD PDD, and MSA [191]. Furthermore, they con rmed that PAR and α-synuclein interact via electrostatic forces involving positively charged lysine residues in α-synuclein [191].
Interestingly, the same group who reported the cryo-EM structures of α-synuclein laments from MSA brains investigated whether seeded assemblies of α-synuclein had the same structures as brain-derived seeds [193]. They seeded the in vitro assembly of recombinant wild-type human α-synuclein with αsynuclein derived from MSA brains. The resultant laments showed distinct conformations from the original MSA brains-derived seed, indicating that the products from in vitro seeding assay do not necessarily replicate the atomic structure of the seed.

Machine learning-based protein structure prediction
Although advances in cryo-EM enable structural determination at near-atomic resolution, this technique is labor intensive and not high throughput. As a potential alternative approach, machine learning algorithms can contribute to protein structure predictions [194,195]. One such algorithm, AlphaFold2, has been used to accurately predict structures of proteins. These state-of-the-art technologies hold great promise to provide predictions of the atomic-level structure of proteins; however, limitations must be addressed in predicting the structures of amyloid proteins, such as α-synuclein. As noted above, α-synuclein has multiple polymorphs and various intermediates and aggregates; therefore, the same amino acid sequence can produce different structures, which hinder in silico sequence-based predictions [196]. Nevertheless, further understanding of the atomic structure of α synuclein may elucidate mechanisms of seeding activity and contribute to designing drugs targeting speci c structural features of α synuclein.
RT-QuIC is an ultrasensitive biochemical assay used to detect self-templating amyloidogenic proteins in brain tissue and cerebrospinal uid (CSF). It has also been explored to various peripheral tissue types. It was originally developed to detect pathogenic seeding of prions [203,204], and is currently used in diagnosis of Creutzfeldt-Jakob disease [205]. PMCA is a technology that was originally established to detect misfolded prion aggregates through sonication-based ampli cation of misfolding and aggregation [206,207]. In contrast to RT-QuIC, PMCA relies on long duration shaking instead of more rapid sonication.
Both techniques have been successful as biomarker for aggregation-prone proteins, including α-synuclein [208].
Positive seeding activity was present in all 13 cases of PD and 3 cases of incidental LBD (100% sensitivity) and 1/16 control cases (94% speci city). Despite the high sensitivity and speci city, invasiveness of the submandibular gland biopsy procedure may limit its clinical application. Olfactory mucosa has also been investigated as a potential sampling site for RT-QuIC [221,226]. De Luca and colleagues reported that 10/18 cases of PD (56% sensitivity) and 9/11 cases of MSA (82% sensitivity) showed RT-QuIC seeding activity, while 16% of primary tauopathies showed positive (84% speci city) [221]. Stefani and colleagues demonstrated positive α-synuclein RT-QuIC seeding activity in the olfactory mucosa in 44% of IRBD patients and 46% in PD patients with an overall speci city of 90% [226]. Interestingly, IRBD patients with positive α-synuclein seeding activity had olfactory dysfunction more frequently than those without seeding activity (79% vs. 23%). Although the sensitivity for detecting αsynuclein seeding activity was moderate in these studies [221,226], the olfactory mucosa sampling by nasal swabbing is less invasive than the lumbar puncture, skin biopsy, or submandibular glands biopsy.
More recently, a study compare diagnoses based on RT-QuIC with α-synuclein immuno uorescence on skin biopsies, supporting the value of seeding applications in clinical practice [227].
These new biomarkers will not only assist detection of synucleinopathies at an early stage, which is often a clinical challenge [228,229], but also help recruit patients for future clinical trials of disease-modifying therapies targeting α-synuclein aggregation and propagation.

Conclusions
Synucleinopathies are clinically and pathologically heterogeneous. LBD is a major cause of dementia and Parkinsonism, while MSA is less common, but a progressive disorder affecting extrapyramidal and cerebellar systems. There is no cure for either disorder. To develop disease-modifying therapies, it is imperative to elucidate how α synuclein converts to pathologic oligomers and brils. The hypothesis that clinicopathologic heterogeneity of α-synucleinopathies is linked to different strains of α synuclein is supported by mounting experimental evidence from seeding assays and advanced structural biology. As distinct strains of α synuclein are increasingly used as biomarkers, it is likely that diagnostic accuracy will improve. Accurate and early clinical diagnosis will be increasingly important for intervention early in the disease process for future clinical trials.

Declarations
Ethics approval and consent to participate Not applicable.

Consent for publication
Not applicable.

Competing interests
The authors declare that they have no competing interests.

Funding
This study was supported, in part, by National Institutes of Health grants (P50 NS072187), a Jaye F. and Authors' contributions SK, HS, and DWD conceived the manuscript. SK and HS wrote and NK, OAR, and DWD edited the manuscript. SK and HS prepared tables and gures. All authors read and approved the nal manuscript.