While neurodegeneration of the nigrostriatal basal ganglia circuitry may be responsible for the motor symptoms of PD, the neuropathologic underpinnings of cognitive decline in DLB and PDD remain less clear, as studies have highlighted the importance of cortical LB and Lewy-neurite pathology, concurrent Alzheimer’s disease-like pathology, and involvement in the brain’s cholinergic system in the development of cognitive decline (Braak et al., 2003; Compta et al., 2011; Irwin et al., 2013). Parkinson’s disease is traditionally characterized by neurodegeneration and accumulation of α-syn¹ in the substantia nigra, and α-syn pathology can be observed throughout the brain (Braak et al., 2003; Bohnen and Albin, 2011; Irwin et al., 2013). Post-mortem neuropathologic studies have revealed the importance of cortical LBs in development of PDD, (Hurtig et al., 2000, Mattila et al., 2000, Harding and Halliday, 2001, Kövari et al., 2003) with specific involvement of the entorhinal cortex, temporal neocortex, anterior cingulate gyrus, and the pre-frontal cortices (Irwin et al., 2012, 2013). Cortical LB burden has also been shown to predict dementia onset and severity in DLB patients (Ruffmann et al., 2016; Ferman et al., 2018). Studies of α-syn aggregates in animal models have long raised the increasingly accepted concept of a prion-like ability for aggregates to spread throughout the nervous system and even from the periphery, although mechanisms of spread are an active area of research (Luk et al., 2012; Masuda-Suzukake et al., 2013; Kovacs et al., 2017; Henderson et al., 2019). Meticulous pathologic analysis of the brains of PD patients, conducted by Braak and others, have led to theories that PD pathogenesis may follow a caudal-to-rostral pattern of spread in at least some patients (Braak et al., 2003). In these patients, pathology was found to start in the brainstem, spread to limbic, sub-cortical structures, before eventually propagating to the cerebral cortex. In fact, the original description of this abnormality by Lewy noted considerable pathology in the basal forebrain, where cholinergic neurons project to regions responsible for many cognitive functions (Lewy, 1913; Bohnen and Albin, 2011) (Box 1). A challenge for modeling these forms of human disease is thus to recreate where possible this concept of pathology propagation in a manner that is consistent with human observations.

The presence of concurrent AD-like pathology, extracellular amyloid-β (Aβ) plaques and intracellular neurofibrillary tau-tangles, are predictive of PDD (Kalaitzakis et al., 2008; Compta et al., 2011) and LBD (Jellinger and Attems, 2008; Gomperts, 2016). In fact, it is estimated that between one-quarter and one-half of PDD patients have enough AD pathology to meet post-mortem pathologic criteria for diagnosis (Mattila et al., 1998; Galvin et al., 2006; Sabbagh et al., 2009). Some have demonstrated that time onset to dementia in parkinsonian patients is correlated with cortical deposition of both α-syn and Aβ (Ballard et al., 2006). Both α-syn and AD-pathology are likely involved in development of the Lewy-Body Dementia, possibly in an additive or synergistic manner (Irwin et al., 2013; Ferman et al., 2018).

In addition to these pathologic hallmarks of neurodegeneration in DLB and PDD, impairment of the brain’s cholinergic systems may play a significant role in development of the diseases. Specifically, cholinergic neurons of the basal forebrain (BF), which have projections to the hippocampus, olfactory bulb, cortical mantle, and amygdala, (Bohnen and Albin, 2011; Pasquini et al., 2021) are implicated in PDD and DLB. Neurodegeneration of BF cholinergic neurons (Arendt et al., 1983; Candy et al., 1983; Nakano and Hirano, 1984; Braak et al., 2003) and their targets in the hippocampus (Perry et al., 1994; Tiraboschi et al., 2002; Hall et al., 2014) have been associated with PDD and DLB. This has been supported by positron-emission tomography (PET) studies revealing decrease in cortical cholinergic activity, (Asahina et al., 1995; Bohnen et al., 2003; Hilker et al., 2005; Shimada et al., 2009; Kanel et al., 2020) and the fact that acetylcholinesterase inhibitors, which prevent acetylcholine degradation, have been shown to improve cognitive outcomes in PDD and DLB (McKeith et al., 2000; Reading et al., 2001; Bergman and Lerner, 2002; Fabbrini et al., 2002; Giladi et al., 2003; Beversdorf et al., 2004; Dujardin et al., 2006; van Laar et al., 2011; Dubois et al., 2012; Mori et al., 2012; Wang et al., 2015). While the exact relationship between cholinergic neurodegeneration and the other neuropathologic findings of Lewy Body Dementia is unknown, there is likely a complex interplay of factors resulting in the ultimate clinical phenotype.

While there is certainly overlap in the pathology seen in PDD and DLB, subtle differences may explain the disparities in cognitive and overall clinical presentation. There is less nigral neurodegeneration in DLB patients compared to PDD patients, which may explain the motor differences seen between the disorders (Tsuboi and Dickson, 2005). Compared with PDD patients, DLB patients have higher α-syn burden in the hippocampus and amygdala (Kalaitzakis et al., 2009; Jellinger and Korczyn, 2018). Additionally, DLB patients have an increase in cortical and subcortical Aβ when compared to PDD patients, specifically with increased Aβ load in the entorhinal cortex, amygdala, and putamen (Hepp et al., 2016). DLB patients also have higher amounts of cortical atrophy compared to PDD patients, most notably in the temporal, parietal, and occipital lobes (Beyer et al., 2007). Collectively, these may account for the cognitive differences seen between patients with PDD and DLB. These findings also emphasize the challenges faced when attempting to model cognitive dysfunction from synucleiopathies in animal models, since not only must the specific mechanism of generating α-syn pathology be considered but also co-pathologies can be critical to the models and may differ in consequences depending upon both levels of expression and location of expression within specific brain regions.

To date, the most efficient way of achieving dopaminergic denervation is through the use of the catecholaminergic selective toxins MPTP and 6-hydroxydopamine (6-OHDA). When administered systemically (MPTP) or intracerebrally (6-OHDA), these toxins cause an accumulation of oxidative species resulting in death of dopaminergic neurons. While not involving Lewy pathology specifically, due to their selective targeting of the dopaminergic pathway, these models capture the motor impairments associated with midbrain degeneration well. Toxin-treated mice fail to remain on the rotarod as long as their sham treated littermates and display amphetamine desensitization. When 6-OHDA is injected unilaterally, mice display rotation behaviors as an effect of the asymmetric burden of neurodegeneration. These toxin models are also useful in understanding the contribution of the nigro-striatal dopaminergic pathway to cognition. While both models have been shown to induce deficits in memory and attention, there is some evidence that MPTP is better at impairing short-term memory, and that 6-OHDA primarily affects long-term memory (Fan et al., 2021). One drawback to these models, however, is that the cognitive impairment induced by the toxins is not stable long-term, with animals showing recovery a few months post injection (p.i.) (Bezard et al., 2000; Fan et al., 2021). This could indicate the measured cognitive decline is related to the brain’s inflammatory response rather than the neurodegeneration itself. In a rotenone study, depleting or inactivating microglia (with PLX397 or minocycline, respectively) attenuated the cognitive impairments as measured by the MWM, NOR, and passive avoidance tests. The microglial inhibition however, also attenuated the pSyn pathology and neuronal loss, making it difficult to ascertain whether the neuroinflammation itself caused the impairment (Zhang D. et al., 2021). In contrast, inducing neuroinflammation in wildtype mice by systemic administration of lipopoylysaccharide (LPS), activates microglia and causes hippocampal neuronal loss as well as sickness behavior and impairments in the MWM and passive avoidance tests (Zhao et al., 2019).

Viral vector-mediated overexpression of SNCA, the gene coding for α-syn, has been shown to result in a local, progressive accumulation of the protein and severe neurotoxicity around the area of injection (Huntington and Srinivasan, 2021). The use of viral vectors to stably overexpress the protein allows for control of the types of cells affected and the degree of pathology by modulating the injection site, serotype, volume and titre of virus delivered as well as promoter and α-syn form (human vs mouse; full-length vs truncated; wildtype vs mutated) expressed. Even though these models have been reported to display cognitive and motor impairments, (Fan et al., 2021) the focal overexpression, the acute neurotoxicity, and the ensuing inflammatory response are not thought to be representative of the neuropathology of synucleinopathies (Volpicelli-Daley et al., 2016). These models have therefore fallen out of favor in recent years. Nonetheless, if the level of α-syn expression influences susceptibility to pathology or co-pathologies, the viral model still can be attractive if dosing, promoter strength or regulation and cell-type specificity can be adjusted to specific applications.

In contrast, transgenic mouse models overexpressing some form of SNCA, often display a time-dependent accumulation of α-syn aggregates which more closely mimic the disease. These models rarely achieve levels of neurodegeneration comparable to the toxin models, but they do reflect a genetic form of human disease, characterized by triplication or mutations of the SNCA gene leading to overexpression and eventual pathogenesis. Depending on the promoter and SNCA variant used, the spatial and neuropathologic signature of the accumulation can be manipulated (Giasson et al., 2002; Freichel et al., 2007; Zhou et al., 2008). The M83 line, which overexpresses the human familial PD-associated A53T form of SNCA under the control of the prion promoter, is one of the most popular transgenic models of synucleinopathy (Giasson et al., 2002). When bred homozygous, these mice develop severe motor phenotypes such as trembling, limb paralysis and inability to stand. Prior to motor symptom onset, which is typically described at 7–8 months of age, these mice display diminished nesting behavior and impaired spatial memory as tested by the Y-maze (Paumier et al., 2013). These mice, however, accumulate α-syn in CNS regions not typically associated with human synucleinopathy, such as the cerebellum and the spinal cord, which may be explained by the use of the prion promoter.

Another well characterized transgenic mouse line, line 61, which overexpresses human full-length wild-type SNCA under the control of the Thy-1 promoter, also displays progressive accumulation of α-syn, with the highest fold increase described in the thalamus when compared to wildtype littermates (Chesselet et al., 2012). These mice display reduced motility in the open-field test at 14-months of age, around which time a 40% decrease in striatal TH levels are observed. Prior to motor symptom onset, these mice display mild cognitive impairments as measured by the Y-maze and NOR tests. Similar to many other transgenic mouse models, these mice do not display TH cell loss in the nigra, even at 22-months of age. Similarly, other Thy-1-asyn mice also display cognitive impairments in the NOR (Subramaniam et al., 2018).

Another approach, in an effort to more closely mimic accumulation of α-syn in relevant brain regions, is to overexpress the human form of the protein under the control of the mouse’s endogenous SNCA promoter, on either a wildtype (Hansen et al., 2013) or mouse SNCA knockout background (Janezic et al., 2013). These mice also display a progressive accumulation of α-syn, very mild neurodegeneration, and progressive motor and cognitive phenotypes starting at around 9-months of age. Besides SNCA, other genes have been implicated in PD and LBD in GWAS studies. Among the most commonly reported are PINK1 and LRRK2. PINK1 KO mice have been developed and though they show impaired synaptic plasticity, (Kitada et al., 2007) cognitive impairment hasn’t been as widely reported (Gispert et al., 2009). LRRK2G2019S mice on the other hand, have been reported to show impairments in executive function in the touchscreen operant task tests (Cheng et al., 2022; Hussein et al., 2022). A comprehensive review of transgenic models used and the cognitive impairments reported can be found elsewhere (Fan et al., 2021).

A model that has been gaining popularity in recent years is the preformed fibril (PFF) model, in which preformed fibrils of human or mouse α-syn are injected into mice (Chung et al., 2019). Alpha-synuclein can be recombinantly expressed and fibrillization induced, commonly by shaking at 37°C. The resulting α-syn fibrils closely resemble those found in Lewy bodies when examined by electron microscopy (Cremades et al., 2012; Marotta et al., 2021). When added to cells or injected into animals, PFFs have been shown to rapidly induce the templating of endogenous α-syn, resulting in a spatio-temporally defined spread of α-syn aggregates (Polinski et al., 2018). PFFs also cause neurotoxicity, with the severity of the toxicity determined by the α-syn PFF ‘strain’ (Bousset et al., 2013) and cellular subtype, (Alegre-Abarrategui et al., 2019) with nigral dopaminergic neurons showing particular vulnerability. One main advantage of this model is that the targeted pathways and the signature of α-syn pathology can be altered with the injection site. As such, the contributions of different pathways to behavioral deficits can be dissected. Summaries of the cognitive impairments induced with PFF injections are listed in Figure 1 and Table 2.

Considering the prevalence of co-pathology in patients, another strategy for PFF-induction of synucleinopathy is the use of α-syn PFFs in Aβ, tau, or APOE mouse models. Injecting A53T PFFs unilaterally into the right frontal cortex has been shown to improve the spatial memory deficits observed in APP/PS1 mice with Aβ accumulation. Surprisingly, in that same study, wildtype mice injected with the PFFs also showed a statistically significant improvement in the MWM compared to PBS-injected controls (Hao et al., 2018). In contrast, in a study using a different model of Aβ pathology, 5xFAD mice injected unilaterally in the hippocampus with mouse α-syn PFFs displayed cognitive impairment, as measured by the Y-maze, starting at 3-months p.i. and continued to worsen over time. These mice showed a progressive increase in pSyn accumulation compared to wildtype mice injected with the PFFs. While the pSyn was primarily contained to the hippocampus, hypothalamus, and cortex at 9-months p.i. in wildtype mice, the pSyn load was already ubiquitous throughout the brain at 4.5-months p.i. in the 5xFAD mice. The PFF injected 5xFAD mice also displayed an increase Aβ and hyperphosphorylated tau burden compared to PBS injected 5xFAD mice (Bassil et al., 2020). Even though tau has been shown to co-localize with α-syn at presynaptic terminals, there have been conflicting reports on the neuroprotective effects of knocking out tau in the PFF model (Stoyka et al., 2021; Pan et al., 2022) with one study suggesting a unidirectional relationship with α-syn as a modulator of tau pathology (Bassil et al., 2020). In the A53T transgenic model, knocking out tau not only reduced neuronal loss but attenuated the memory deficits associated with the model (Singh et al., 2019). Both phosphorylated tau and pSyn have been shown to be upregulated in the MPTP model of neurodegeneration (Hu et al., 2020). Finally, two separate studies have demonstrated a modulatory effect of APOE genotype on α-syn pathology. While APOE2 showed a neuroprotective effect compared to APOE3 and even APOE KO, APOE4 exacerbated α-syn pathology burden, both in the A53T mouse (Davis et al., 2020) and in the AAV-induced overexpression model (Zhao et al., 2020). The α-syn APOE4 mice also demonstrated a faster cognitive decline compared to the other APOE genotypes.

The discrepancies in results between studies may be attributed to one of several factors. The strain of PFFs used, which is dictated by the aggregating conditions and specific α-syn variant (mouse vs human, wildtype vs familial-associated point-mutations (A53T, A30P, etc.), may influence outcome. Although many groups use the same behavioral tests, the protocols for these tests (for instance, whether they are performed during the light or dark cycle) and therefore also the readouts vary between groups. The number and sex of animals used also affects whether any potential impairment reaches statistical significance, which in turn affects how differences in behavior are reported. Additionally, the timepoint at which the cognitive behaviors are evaluated can influence what is observed. If a late timepoint is used, many areas of the brain may be burdened with pSyn and as such conclusions about the studied pathway can be confounded. If an early timepoint is used, however, the pSyn pathology may be restricted to the area of injection and closely connected pathways, but the acute inflammation accompanying the injections may confound any observed cognitive impairment. While cholinergic pathology is thought to underlie cognitive impairments in humans, many mouse models have failed to recapitulate cholinergic degeneration in mice. This could be due to the key differences in cholinergic neurons in terms of distribution, functionality and structural roles in rodents as compared to primates (Disney and Robert, 2019). Acute systemic inflammation is known to cause reversible cognitive impairments (Skelly et al., 2019) but inflammation has also been suggested as a key driver in α-syn pathology (Kim et al., 2022).