Transmission of α-synucleinopathy from olfactory structures deep into the temporal lobe

Mice infused in the right OB/AON with fibrils sonicated for 1 h in a waterbath displayed dense pSer129⁺ inclusions in the olfactory peduncle (defined here as the AON complex and the tenia tecta [27], Fig. 1, Additional file 1: Table S2). Mice infused with fibrils sonicated for 1 min with a probe exhibited far less α-synucleinopathy (Additional file 1: Table S2, Figure S1). PBS infusions did not result in any α-synucleinopathy. pSer129⁺ inclusions were formed in structures known to harbor anatomical connections with the OB/AON, such as the amygdala, the frontal, piriform, and entorhinal cortices, the nucleus of the lateral olfactory tract, the subiculum, and hippocampal CA1 [28‐30] (Fig. 1a-c). Additional sparse pSer129 label was observed in some animals in CA2/CA3, the dentate gyrus, and the ectorhinal/perirhinal cortex (Additional file 1: Table S2; Fig. 1a-b), perhaps reflecting limited transynaptic transport to these sites from CA1, the subiculum, and the entorhinal cortex, all of which harbored dense inclusions (Fig. 1a-d) and project directly to the OB/AON [28, 31, 32]. We invite the reader to download Fig. 1 in high resolution and zoom in and out of the images in Fig. 1b as with a street map, in order to appreciate the full extent of the pSer129⁺ pathology and verify its anatomical sublocalization. Note that the image takes an extremely long time to fully populate the computer screen with high-resolution photos, not only upon opening the file, but also upon zooming in and navigating across brain regions. Most α-synuclein-containing inclusions were wrapped around Hoechst⁺ nuclei or found in processes (Fig. 1d). A blinded analysis revealed that the number of pSer129⁺ structures in the olfactory peduncle, the piriform and entorhinal cortices, amygdala (posteromedial cortical amygdala), and hippocampal formation (subiculum and CA1) were dramatically higher in fibril-injected mice (Fig. 1e). The SEM values in Fig. 1e reveal significant variability in hippocampal and entorhinal inclusion counts, perhaps because of slight differences in stereotaxic placements of infusate or inter-animal differences in vulnerability. The most robust pathology was evident in the AON, and this was far denser than in the OB itself (Additional file 1: Figure S1, S2, Fig. 1a, cb, e, Additional file 1: Table S2), consistent with reports that the AON is one of the main olfactory structures bearing Lewy pathology [3, 4, 15]. For AON subdivisions, please examine the higher magnification images provided beneath Additional file 1: Table S2 and the elegant work of Brunjes, Haberly, and colleagues [27‐30]. pSer129 labeling was strikingly absent from the olfactory tubercle (Fig. 1a), which receives dense first-order projections from the OB/AON but does not send reciprocal projections back to olfactory structures [28‐30], suggesting lack of anterograde transmission of α-synucleinopathy.

The development of dense perinuclear and neuritic inclusions following fibril infusions was confirmed with two independent (monoclonal and polyclonal) pSer129 antibodies (see polyclonal anti-pSer129 staining in Fig. 1f, Additional file 1: Figure S3). The staining with the polyclonal and monoclonal antibodies overlapped well when applied simultaneously (Additional file 1: Figure S3). For preadsorption controls for primary antibody specificity, we used the polyclonal pSer129 antibody because preadsorption controls with monoclonal antibodies are always expected to lead to loss of labeling regardless of which proteins they bind in tissue [33]. Preincubation of pSer129 antibodies with pSer129 blocking peptide led to dramatic loss of immunoreactivity relative to adjacent sections exposed to free, unbound primary antibodies (Fig. 1f). Omission of primary antibodies also led to loss of immunofluorescent signal (not shown).

Next we characterized the nature of the inclusions formed three months following fibril infusions into the OB/AON. First, we confirmed that most perinuclear pSer129⁺ inclusions were formed within cells expressing the neuronal nuclear marker NeuN (Fig. 2a). This was further verified by a confocal analysis (Fig. 2b) and then by three-dimensional video analyses of the confocal images (see two videos in Additional file 2). The appearance of the somal α-synuclein inclusions—fibrous perinuclear strands—is similar to that reported in mouse tissue and some human Lewy bodies by Osterberg and colleagues [34]. Second, we examined whether the inclusions exhibited characteristics of Lewy pathology. The Thioflavin S stain for amyloid was applied to OB/AON tissue and densely labeled cellular structures at the infusion site but faintly or not at all elsewhere (Fig. 2c, Additional file 1: Figure S4). Ubiquitin staining, a traditional method for detecting Lewy-like pathology [22, 35], was present in some but not all pSer129⁺ structures, as shown with two antibodies against K48-linked ubiquitin (Fig. 2d, Additional file 1: Figure S5). These findings are consistent with the work of Osterberg et al. showing pSer129⁺ inclusions at progressive stages of maturity, with mature inclusions bearing hallmarks of Lewy pathology [34]. In addition, our ubiquitin staining pattern is consistent with the classic work of Spillantini et al. demonstrating that ubiquitin only stains a portion of all α-synuclein⁺ Lewy bodies and neurites [22].

Previous work has shown that the fibrils themselves are not phosphorylated [36]; the pSer129⁺ structures in the present study therefore likely demark endogenous, aggregated α-synuclein and not direct transport of injected material as in the Rey study [13]. This interpretation is consistent with the findings that 1) there was unexpectedly sparse pSer129 labeling in the OB itself—the epicenter of the fibril infusions—and 2) that there were no pSer129⁺ inclusions in animals sacrificed 1.5 h after fibril infusions (Fig. 2e). However, the latter animals exhibited extremely dense total (non-phosphorylated) α-synuclein immunoreactivity at the injection site relative to PBS-infused controls, effectively delineating the center of the infusion site in the caudal OB (Fig. 2e).

The topography of the pSer129 label three months following fibril infusions into the OB/AON suggested that brain regions projecting directly to the site of infusion eventually developed inclusions. Thus, the retrograde tracer FluoroGold was infused into the OB/AON at the same stereotaxic coordinates in order to label first-order afferent neurons that project directly to the injection site (Fig. 3a). FluoroGold labeling of first-order projection neurons was evident at all inclusion-bearing sites, with the exception of the caudoputamen, accumbens, CA2/CA3, and dentate gyrus (Additional file 1: Table S2). Furthermore, dense and brightly stained FluoroGold⁺ neurons were additionally present in some areas not harboring any pSer129⁺ pathology, such as the horizontal limb of the diagonal band of Broca (Fig. 3aa, af-g , Additional file 1: Table S2), which is known to project densely to the AON [28‐30] but resisted the development of α-synucleinopathy (Fig. 1a, also see Fig. 4), suggestive of selective vulnerabilities. No retrogradely labeled FluoroGold⁺ cells were present in the olfactory tubercle (Fig. 3aa, af-g, Additional file 1: Table S2), consistent with the lack of pSer129⁺ structures at this location (Fig. 1a, also see Fig. 4b, d, f) and with previous reports that the olfactory tubercle does not project to the OB/AON [28‐30]. These findings suggest a lack of anterograde transmission of α-synucleinopathy, as argued above, or additional selective vulnerabilities.

In order to confirm that our sonication protocol was reproducible in a different set of neuroanatomical circuits, we examined the spread of α-synucleinopathy through the perforant path from CA2/CA3. These animals were infused with vehicle or fibrils sonicated for 1 h or 24 h in a waterbath, as 1 min-long probe sonication was less successful in Study 1. Animals were sacrificed three months later. Fibril infusions centered in CA2/CA3 of the hippocampus led to dense pSer129⁺ inclusions at the site of injection as well as CA1 and the dentate gyrus (Fig. 3b-c). The entorhinal cortex was densely labeled in fibril-infused animals, suggestive of successful α-synucleinopathy transmission through the perforant path, as in previous studies [24]. Additional pathology was evident in the ectorhinal/perirhinal region, which also projects into the hippocampal formation [37] and has been highlighted by Braak as an important node for the spread of Lewy pathology from the mesocortex into the neocortex in Braak stages 5 and 6 [3, 38]. Similar patterns were observed with the polyclonal pSer129 antibody, and this was again significantly reduced upon preadsorption (Additional file 1: Figure S6). The antibody control experiments—including colabeling with polyclonal and monoclonal antibodies—were performed on at least two separate occasions, on coronal hippocampal tissue from CA2/CA3 infusions and sagittal tissue from OB/AON infusions, and the results were in agreement.

Hippocampal pathology was denser in the 1 h waterbath-sonicated group relative to 24 h sonication (Additional file 1: Figure S7), possibly because of overheating and degradation of α-synuclein over the course of 24 h (the entire waterbath became hot). In order to visualize fibril morphology under different sonication conditions, we performed electron microscopy and observed reductions in fibril size with 1 min probe sonication, 1 h waterbath sonication, and 24 h waterbath sonication (Fig. 3d), as in previous work [39].

In order to ensure that the transmission of α-synucleinopathy from the OB/AON into deeper rhinencephalic structures was reproducible, we injected a second set of animals in the OB/AON with PBS or fibrils. Based on the findings of the CA2/CA3 study (Study 3), the fibrils were sonicated for 1 h. As an additional control, the investigator performing the stereotaxic surgeries was blinded to the identity of the injected solutions. Animals were infused at 5 or 17 months of age and sacrificed three months later. The extent of pathology in older animals is illustrated in Fig. 4. The pSer129 label in these fibril-infused animals can be contrasted with the lack of inclusions in large stitched montages of PBS-infused animals in Additional file 1: Figure S8. These images from PBS mice can also be deeply examined for background staining with the pSer129 antibody. For the semi-quantitative analysis shown in Additional file 1: Table S2, all the two month-old animals from the first study were restained with pSer129 antibodies in parallel with all the older animals in order to control against inter-experimental variability in immunofluorescence. The relative degree of staining in each brain region was then rated by a blinded observer. Relative to young animals, some aged (20 month old) animals exhibited more inclusions in the ventral striatum (nucleus accumbens), which was not labeled retrogradely with FluoroGold and therefore does not project to the OB/AON injection site. These findings are consistent with prior reports, as Rey and colleagues also reported striatal and frontal cortical uptake of human α-synuclein following OB infusions [13]. Two additional areas had sparse and variable labeling in some animals and were not included in the table, such as the bed nucIeus of stria terminalis, which also projects to the OB/AON [40], and the brainstem immediately caudal to the ventral portion of the fasciculus retroflexus, in or near the parabrachial nucleus. Both of these regions also harbored some FluoroGold⁺ cells.

Three important caveats of our study must be conceded. First, rodents are macrosmatic animals (i.e., they depend heavily upon olfaction) and therefore may exhibit denser rhinencephalic interconnectivity than humans. It would therefore be premature to conclude that hippocampal Lewy pathology can be traced to the OB/AON in humans based only on rodent work. Second, recent studies suggest that labeling with the pSer129 antibody is subject to cross-reaction with neurofilament-L [39]. Indeed, we see abundant evidence of background staining, especially in fiber tracts (see edges of olfactory peduncle in Fig. 1ca-b, Additional file 1: Figure S1, photos below Additional file 1: Table S2, polyclonal pSer129 staining in Additional file 1: Figure S3a). To demonstrate that this background staining did not confound our interpretations, we have included enormous stitched montages of the whole brain in fibril and PBS groups (Figs. 1b and 4, Additional file 1: Figure S8). Although we cannot rule out all cross-reactivity with certainty [24], the background label is distinct from high-intensity labeling of inclusions, which is absent in PBS animals, as is evident from the stitched montages and the blinded quantification in Fig. 1e. It is also worth mentioning that the antibody specificity controls were successful (Fig. 1f, Additional file 1: Figure S3, S6). The third limitation of the present study is that our OB/AON model differs from the human condition in many important ways—for example, OB/AON infusions do not result in pathology in the olfactory tubercle or dense hippocampal pathology in CA2/CA3. In Parkinson’s disease, α-synucleinopathy is hypothesized to emerge first in enteric and olfactory structures and to spread from the OB/AON regions to the olfactory tubercle, piriform cortex, periamygdalear cortex, and entorhinal cortex (as observed in our study, with the exception of the olfactory tubercle), but “without advancing into non-cortical areas” [41]. Instead, most α-synucleinopathy is hypothesized to enter the human brain from the dorsal motor nucleus of the vagus [3, 38]. It is worth noting, however, that most of the sites exhibiting inclusions in Studies 1 and 4 are affected in Braak stages 1–4 of Parkinson’s disease [3, 38]. Despite the obvious differences between the models presented here and the human condition, the fibril model may be useful to test Heiko Braak’s hypotheses that neurons with long, thin, unmyelinated axons, and perhaps high metabolic rates, are more vulnerable to α-synucleinopathy than neurons with sturdy, myelinated axons [38, 42]. Such neurons may also exhibit impairments in proteasomal and autophagic degradation systems.

It is unlikely that diffusion and non-specific uptake of fibrils throughout all the interstitial space of the brain confounded our interpretations, as so many regions close to the OB/AON remained free of inclusions (e.g., Fig. 4) and the pSer129 label closely overlapped that of FluoroGold. Had non-specific uptake following fibril diffusion through the interstitial space been responsible for the pSer129 labeling, we would have expected a gradient of dense labeling close to the injection site and sparse labeling with increasing distance from the injection site, but this was not the case; for example, in the OB/AON studies there was dense label in distant locations such as the lateral and caudal entorhinal cortex but absence of label in closer sites such as the hypothalamus and thalamus (Studies 1 and 4). Similarly, in the CA2/CA3 study, the injection site in the hippocampus lies close to many areas that were free of pSer129 label (e.g., see thalamus and overlying cortex in Fig. 3c, Additional file 1: Figure S7). Instead, the inclusion topographies are most parsimoniously explained by preponderant transmission through anatomically linked circuits. Furthermore, the pSer129 antibodies are not simply labeling the injected fibrils following diffusion, because pSer129⁺ label was scarce at the OB infusion epicenter in all fibril-treated animals and the inclusions were absent 1.5 h following fibril injections, a timepoint at which there was a dramatic increase in total α-synuclein labeling in the OB/AON relative to PBS controls.

In conclusion, the present short report demonstrates that 1) α-synucleinopathy can spread from the rostrally and medially situated OB/AON all the way to the caudal recesses of the temporal lobe (hippocampal formation and entorhinal cortex) within three months, 2) the spread of pathology is consistent with retrograde transmission through neuroanatomical networks, although this obviously remains to be confirmed, 3) some first-order afferent sites retrogradely transport FluoroGold from the OB/AON but resist the development of α-synucleinopathy, and 4) age at fibril infusion does not make a robust difference in α-synucleinopathy transmission in the first three months. Although our findings are consistent with the olfactory vector hypothesis of Parkinson’s disease [2] and may be especially relevant to diffuse Lewy body disease or dementia with Lewy bodies, it is difficult to extrapolate from rodent data directly to humans. If such extrapolation is eventually validated, then our findings may have implications for neurological outcomes, such as those controlled by the piriform and entorhinal cortices, amygdala, hippocampus proper, and subiculum. For example, memory loss in Parkinson’s disease is associated with α-synuclein pathology in CA2, the amygdaloid complex, and the entorhinal cortex and with tissue atrophy in CA1, CA3, and the subiculum [43, 44]. Indeed, loss of olfaction is a harbinger of cognitive dysfunction and memory loss [45‐47] and predicts mortality in senior citizens; 22 % to 40 % of older adults with anosmia are dead within 5 years, as opposed to 10 % of elderly subjects with intact olfaction [48, 49].