Local cortical overexpression of human wild-type alpha-synuclein leads to increased dendritic spine density in mouse.

Abstract

Lewy body dementias are characterized by deposition of alpha-synuclein (α-syn) protein aggregates known as Lewy bodies and Lewy neurites in cortical regions, in addition to brainstem. These aggregates are thought to cause the death of dopaminergic neurons in the substantia nigra and other vulnerable cell types in patients, leading to parkinsonism. There is evidence from mice that localized overexpression of wild-type α-syn leads to dopaminergic cell death in the substantia nigra. However, it is not known how cortical neurons are affected by α-syn. In this study, we used viral overexpression of α-syn to investigate whether localized overexpression within the cortex affects the density, length, and morphology of dendritic spines, which serve as a measure of synaptic connectivity. An AAV2/6 viral vector coding for wild-type human α-syn was used to target overexpression bilaterally to the medial prefrontal cortex within adult mice. After ten weeks the brain was stained using the Golgi-Cox method. Density of dendritic spines in the injected region was increased in layer V pyramidal neurons compared with animals injected with control virus. Immunohistochemistry in separate animals showed human α-syn expression throughout the region of interest, especially in presynaptic terminals. However, phosphorylated α-syn was seen in a discrete number of cells at the region of highest overexpression, localized mainly to the soma and nucleus. These findings demonstrate that at early timepoints, α-syn overexpression may alter connectivity in the cortex, which may be relevant to early stages of the disease. In addition, these findings contribute to the understanding of α-syn, which when overexpressed in the wildtype, non-aggregated state may promote spine formation. Loss of spines secondary to α-syn in cortex may require higher expression, longer incubation, cellular damage, concomitant dopaminergic dysfunction or other two-hit factors to lead to synaptic degeneration.

Introduction:

Dementia with Lewy body (DLB) and Parkinson’s disease dementia (PDD) are multi-system diseases that present with numerous debilitating symptoms, including executive dysfunction, visual hallucinations, loss of independence, motor slowness and autonomic symptoms. Patients fulfilling criteria for DLB are most likely to have aggregates of alpha-synuclein (α-syn), termed Lewy bodies and Lewy neurites, in a diffuse neocortical or limbic pattern [19]. Similarly, patients with Parkinson’s disease that develop PDD are likely to have cortical α-syn deposits [9]. Under physiologic conditions, α-syn is located predominantly at the presynaptic terminals of neurons and is thought to act as a hub protein together with multiple partners, and aid in the regulation/mobility of synaptic vesicles [4, 16, 17]. Abnormal presynaptic aggregates and loss of dendritic spines have been found in the cortex of patients with DLB [15], but it is not known if cortical cells are directly vulnerable to these synuclein aggregates, or if cortical symptoms in DLB might alternatively be attributed to cholinergic or dopaminergic dysfunction or mixed pathology. Dendritic spines are excitatory postsynaptic connections that protrude from dendritic shafts and act to facilitate communication between presynaptic and postsynaptic neurons. Loss of synapses and synaptic proteins may precede cell loss in Alzheimer’s disease and other dementias and correlates with the severity of cognitive dysfunction [5, 24, 28]. Overexpression of α-syn is used as one model of synucleinopathies because genetic duplication and triplication in patients is associated with Parkinson’s disease and DLB [12]. Viral overexpression of α-syn in the substantia nigra and ventral tegmental area causes pathology, including axonal swelling, dystrophic neurites or neuronal cell loss [2, 10, 21]. α-syn overexpression also leads to an increase in axonal bouton size and number of vesicles [30], but it is unclear what effect this has on the post-synaptic cell. In this study, we aimed to determine the response of cortical neurons to localized wildtype α-syn overexpression. Based on previous literature, we hypothesized that local overexpression of wild-type α-syn would cause dendritic spine loss. In contrast, we found an increase in the density and length of dendritic spines after ten weeks in mice expressing α-syn compared with those expressing a control protein.

Materials and methods

All experiments were approved by the University of Iowa Institutional Animal Care and Use Committee. Four-month-old adult male mice (control: 7, α-syn: 8) were bilaterally injected, targeting layer V of the medial prefrontal cortex (ML = ±0.5 mm, AP = +1.8 mm, DV = −1.5 mm, Fig. 1A). Animals were randomized to injection with either vector coding for expression of a control protein, mCherry: AAV2/6-CAG-mCherry-WPRE (1.7 × 10^12 gc/ml, Vector Biolabs, 1μl bilaterally) or vector coding for human, wild-type alpha-synuclein: AAV2/6-CAG-hSNCA-WPRE (1.0 × 10^13gc/ml, Vector Biolabs, 1 μl bilaterally). After ten weeks, both groups were euthanized for analysis. The brain was rinsed in PBS and immersed in solution A of the FD Rapid GolgiStain Kit, as per the manufacturer’s instructions. FD Rapid GolgiStain is a Golgi-Cox based kit, with a two-week incubation period in solution A/B (containing mercuric chloride, potassium dichromate, and potassium chromate), followed by incubation in proprietary solution C, cryosectioning at 150 μm and development. All layer II/III neurons that met criteria were identified and marked within medial prefrontal cortex (prelimbic, infralimbic, cingulate gyrus 1 (Cg1), from 1.94 mm bregma to 1.54 mm bregma based on a stereotaxic atlas [23], Fig. 1A). Analysis of Layer V neurons was restricted to prelimbic and Cg1. Only neurons with a dominant apical dendrite perpendicular to the pial surface were included. Neurons <200 μm from the surface were categorized as layer II/III and neurons 300–550 μm from the surface were categorized as layer V based on cytoarchitecture in this region [29]. Neurons between 200–300 μm were excluded as they had an intermediate phenotype and could not be classified. Light microscope z-stacks with a 0.25 μm step size were taken of basilar and apical dendrites using an oil 100x, 1.4 N.A objective and then analyzed via ImageJ/FIJI (Fig. 1B). 10 μm bins were selected on basilar and apical dendrites at multiple distances from the pyramidal cell soma, as well as at the tufts for apical dendrites. Dendritic spine density, length, and morphology were recorded by observers blind to the treatment status of each animal.

Figure 1).

A: Viral injection into medial prefrontal cortex. B: human-alpha-synuclein (α-syn) transfection in medial prefrontal cortex (magenta). C: Representative dendrites from basilar layer II/III and Layer V neurons stained using Golgi kit. D-L: Co-staining for human α-syn (magenta, bottom row), phospho-S129 α-syn (cyan, middle row) and composite images (D, G, J). G-I: At the injection site, both total α-syn (I) and phospho-α-syn (H) can be found in the nucleus, soma, dendrites, dendritic spines and axons, although phosphor-α-syn is highest in nucleus (H). J-L: One millimeter away from injection site, only axons are stained with human α-syn (L), and phospho-α-syn is absent (K). L-Q: Phospho-α-syn (cyan) and mCherry (red), from an animal co-injected with both control and α-syn virus (P-R) vs. control virus alone (S-U). PrL (prelimbic), IL (infralimbic) and Cg1 (cingulate).

Immunohistochemistry:

Ten weeks following injection with virus coding for mCherry, h-α-syn, or a mixture of both viruses, mice were transcardially perfused with 1x phosphate buffered saline (1 minute), followed by 50 ml of 4% paraformaldehyde (Thermo Scientific). Primary antibodies used were 1:250 mouse α-syn (LB509, Thermo Scientific) and 1:250 rabbit (phospho S129) alpha-synuclein (EP1536Y, abcam). Secondary antibodies were Alexa 647 goat anti-rabbit, Alexa 647 goat anti-mouse, and Alexa 488 goat anti-mouse (abcam).

Statistics:

Spines were counted in bins at varying lengths from the soma (“distance from soma”). 1–2 dendrites were counted per cell, per dendrite type, with 3–14 cells per mouse for layer V (51–76 total cells) and 1–14 cells per mouse for Layer III (29–42 total cells). All cells in the region that met criteria were included. Not all cells contained all bins. Spine density was predefined as the main outcome measure. The generalized linear mixed modeling (GLMM) framework with a log link function was used to analyze spine density for each dendrite type (apical layer V, basilar layer V, and basilar layer II/III). Dendrite type, distance from soma, and treatment were included as fixed effect variables, while cell was included as a random effect. Clustering on mouse (instead of cell) did not improve the fit of the model or significantly alter the estimates. Spine length and density by morphology were each analyzed separately as secondary measures, also using the GLMM approach with a log link. For each outcome, interactions between fixed effects were screened for inclusion using the Akaike information criterion (AIC, smaller is better). In order to fix the type 1 error at 0.05 for each outcome measure, cutoffs for p values were corrected for multiple-comparisons using a Bonferroni correction; Spine density p < 0.016 (3 dendrite types), Spine length p < 0.025 (2 dendrite types), and spine morphology p < 0.0041 (12 comparisons: morphology(4) x dendrite(3)).

Results

First, we confirmed overexpression of α-syn in our viral model. Immunostaining for human α-syn demonstrated overexpression within the medial prefrontal cortex in the area of injection (Fig. 1 B, G-I), but no transfected cell bodies were visualized outside the target region (Fig. 1 J-L). Human α-syn was localized to soma and dendrites as well as axons, with axonal staining found significantly further from the injection site (Fig. 1 L). We also stained for phospho-α-syn at serine 129, a widely used indicator of pathological synuclein aggregation or pre-aggregation seen in patients with Lewy body disorders [22, 27]. Phospho-α-syn immunofluorescence was more discrete, with expression highest in cell bodies and nuclei in regions of highest α-syn overexpression (Fig. 1H, 1N), and lack of staining in many transfected axons outside the target region, despite the presence of transfected human α-syn (Fig. 1K). Animals transfected with control virus (mCherry) showed little or no immunostaining for phospho-α-syn at the injection site (Fig. 1T).

Next, we quantified spine density using Golgi-Cox staining. We found that spine density in both basilar and apical dendrites of layer V neurons was increased in mice with local overexpression of wild-type human α-syn compared with animals injected with control virus (apical V, p = 0.0004, basilar V, p = 0.0046, Fig 2A-B). There was not a significant difference in basilar dendrites from Layer II/III, although the direction was the same, with higher average density in α-syn group (p = 0.4, Fig 2C). The interaction between dendrite type and distance from soma was included as a fixed effect. It was a significant predictor of density for all dendrite types. The difference between layer II/III and V may be due to lower synuclein concentration, as the injection was targeted to layer V. Quantification of layer II/III also included some infralimbic cells further from the injection site.

Figure 2).

Spine density in Layer V apical (A) and basilar (B) dendrites and Layer II/III basilar (C) dendrites in mice with local overexpression of human alpha-synuclein compared with mice injected with control virus. Density by morphology in Layer V apical (D) and basilar (E), and Layer II/III basilar (F) dendrites. G: Spine morphology categories were defined independent of length. H: Overall spine length. Errors = S.E.M. p-value cutoffs for significance were adjusted for multiple comparisons: *p < 0.016, **p < 0.025, ***p < 0.0004, #p < 0.05 (non-significant due to correction).

To determine if the increased spine density in layer V was due to an increase in filopodia-type spines (long, immature-appearing spines that are less likely to make a synapse) vs. spines with a bulbous head that usually represent synaptic connections [3], we next analyzed morphology in two ways. Each spine was assigned a morphological category (filopodia, thin-headed, stubby, and mushroom) based on the ratio of the head and neck, and irrespective of length (Fig. 2G). We analyzed the density of each spine type by adding morphology to the interaction of dendrite type and treatment. In apical dendrites, filopodia, mushroom and stubby spine types showed increased density in mice overexpressing α-syn, but in this dendrite type, only the difference in filopodia remained significant when corrected for multiple comparisons (Fig 2D). In basilar dendrites, only thin-headed spines showed a significant increase in mice overexpressing α-syn (Fig 2E). Overall these data do not support the hypothesis that the increased spine density is driven only by increases in filopodia-like spines.

We also analyzed spine length independent of morphology in layer V dendrites. In basilar dendrites, animals overexpressing α-syn showed increased spine length compared with mice injected with control virus (Fig. 2H). We did not find any overall significant increase in spine length on layer V apical dendrites.

Discussion

In this study we found that ten weeks overexpression of human wild-type α-syn in adult mice did not lead to spine loss. Surprisingly, we found evidence of spine gain in the targeted region of cortex. A few studies have previously investigated the effect on α-syn on dendritic spines, each with key differences. Global, genetic overexpression of human wild-type α-syn in mice leads to reduced total spine density in cortex by four months of age [6]. Similarly, global expression of the A30P mutation of α-syn leads to decreased spine density and abnormal spine turnover in the olfactory bulb [20]. Alternate models using α-syn pre-formed fibrils have shown the ability to cause α-syn aggregation in cortex, leading to decreased spine density after five months in one study [6, 26].

There are two main differences between prior studies and our own. First, our study examines changes at an earlier time point, prior to evidence of degeneration. This may be important for understanding Lewy body disease, which may have less cortical thinning compared with other dementias, especially compared to those with mixed pathology [14, 31]. Second, prior studies do not distinguish the direct effect of α-syn in cortex from lost inputs from other brain regions, including dopaminergic input. It has been shown, for example, that dopaminergic cell loss via 6-OHDA injection into substantia nigra pars compacta leads to spine loss in the prefrontal cortex of rats [25]. Therefore, spine loss in these models may be due to direct or indirect actions of α-syn in the cortex.

To our knowledge, ours is the first study to investigate the early effect of localized overexpression of α-syn on dendritic spines in the cortex. Overexpression of a similar construct has been used to target the midbrain in rodent models of Parkinson’s disease. Decressac et al (2012) showed that localized overexpression of α-syn via AAV2/6 in the substantia nigra pars compacta (SNpc) led to progressive motor phenotype and loss of tyrosine hydroxylase positive cells starting as early as five weeks after viral injection [10]. Similarly, Giordano et al (2018) also used AAV2/6 injection into SNpc and found alterations in motor learning after 8 weeks. In our study, after ten weeks we saw no evidence of dendritic spine loss using Golgi-Cox analysis, and in contrast saw evidence for increased spine density. There are several possible explanations for these findings. α-syn is known to have dose-dependent effects on synaptic function. Prior to aggregation, modest levels of overexpression appear to promote pore dilation and increase discharge of vesicle cargo, whereas higher levels may inhibit exocytosis [16]. Changes in the shape, size and density of spines are regulated by complex pathways that integrate information about synaptic activity, non-neuronal partners and global states [8, 32]. For example, recent in vivo and cell culture data from medium spiny neurons suggests that α-syn oligomers applied to neurons alone cause spine loss, but when taken up by neighboring astrocytes lead to neuronal spine gain [11]. It is not known whether cortical astrocytes would have a similar effect on cortical neurons. If spine density increases early in the disease it could lead to hyperexcitability or an imbalance in connectivity, thereby causing symptoms. Viral overexpression using AAV6 in rat forebrain during development has been shown to cause cell loss, including of cholinergic interneurons [1]. This type of selective cell loss could also lead to compensatory changes in connectivity and spine density. Additionally, alterations in local dopaminergic terminals might also cause changes in spine morphology and density, as studies have shown dopamine regulates inhibition in the prefrontal cortex [7]. Future studies that determine if cholinergic or dopaminergic terminals are altered by local α-syn overexpression are warranted. Finally, our data show phospho-α-syn can also accumulate directly in the dendrite and post-synapse, creating the potential for a direct effect at dendritic spines. Although further research will be necessary to distinguish between pre-synaptic, post-synaptic and non-neuronal effects of α-syn, our data provide evidence that α-syn can exert a local effect on cortical cells, promoting an increase in spine density in the region.

Limitations:

The spine data in this study was derived from male mice. As PDD and DLB are found with higher prevalence in male patients, it will be essential and informative in future studies to determine the influences of sex on α-syn at the synapse. Secondly, the control protein used in this study was mCherry, injected at a lower titer, but equal volume, compared with α-syn vector. As mCherry is a foreign protein, even at this dosage there may be concern for alterations caused by the control protein, whereas α-syn is endogenously expressed in abundance. Future studies that include mCherry injected into both groups, an empty vector, or uninjected controls could be useful in order to rule out any mCherry regulated cell autonomous changes that may occur.

Finally, this study was limited to evaluation of human wild-type α-syn at a single timepoint and viral dosage. There is evidence that overexpression of mouse alpha-synuclein might be more potent [18]. We chose to evaluate human wild-type α-syn given its role in spontaneous disease. By using this model we were able to examine the effect of local overexpression, with reduced potential for spread. However, it is important to note that overexpression of mouse α-syn or α-syn carrying mutations might cause a different outcome. Furthermore, the effect of alpha-synuclein has been shown to be dose dependent. In this study, we injected AAV2/6 coding for α-syn using 1.0×10^10gc/1 μl, bilaterally). This is intermediate compared to previous studies also expressing human α-syn using this virus, including Giordano et al. (2018) (7.7 × 10^11/1 μl, bilaterally) in mice and Decressac et al. (2012) (3.1 × 10^8 gc/3 μl) unilaterally in rats [10, 13]. Both of these studies found progressive changes in behavior with injection into substantia nigra. However, Decressac et al. reported lack of behavioral phenotype at early time points when the WPRE element to enhance expression was absent from the vector [10]. Although our vector did contain a WPRE element, these findings demonstrate that even in substantia nigra, response can be linked to dosage. Thus, to definitively address differences in brain region susceptibility, direct comparison using identical virus in the same environmental setting could further validate these results.

Conclusions

PDD and DLB are pathologically defined by the presence of α-syn aggregates in the cortex, yet the effect of α-syn pathology on cortical cells is poorly understood. Here we show that ten weeks of local overexpression of wild-type α-syn caused an increase in average dendritic spine density in layer V neurons compared with a control protein. These data have important implications for other studies using viral overexpression of α-syn as a model system. Furthermore, these data have implications for studying region and cell-type-specific responses to α-syn, which are vital to our understanding of these multi-system diseases.

Highlights.

• Lewy Body dementias are defined by cortical alpha-synuclein.

• It is unclear how alpha-synuclein affects cortical cells.

• Local overexpression of alpha-synuclein leads to increased spine density after ten weeks.

• These findings may have important implications for modeling early stages of cortical dysfunction.