Alzheimer’s disease tau is a prominent pathology in LRRK2 Parkinson’s disease

LRRK2 mutation carriers with available brain tissue were identified among deceased individuals with LRRK2 genotyping results at the Center for Neurodegenerative Disease Research (CNDR) at the University of Pennsylvania. Ten individuals were identified who carried the p.G2019S mutation, two of whom were homozygous for this mutation. Three of these carriers were previously described [23]. Two other individuals had been previously described [24] who had variants of unknown pathogenicity. While genotyping has historically been applied on a project-by-project basis, we assessed the likelihood of ascertainment bias in our database by evaluating the percentage of all cases for which LRRK2 had been genotyped for at least the most common p.G2019S mutation. Remarkably, only 76/1484 (5.1%) of all cases for which DNA was available had not been genotyped, making it unlikely that patients with LRRK2 mutations were missed due to a lack of LRRK2 genotyping.

Brains were removed and regions of interest were dissected and fixed in either neutral buffered formalin (NBF) or 70% ethanol, 150 mM sodium chloride. Tissue was infiltrated with paraffin, embedded in blocks, cut into 6 μm sections and mounted on glass slides. Slides were then stained using standard immunohistochemistry as described below. Slides were de-paraffinized with 2 sequential 5-min washes in xylenes, followed by 1-min washes in a series of descending ethanol concentrations: 100, 100, 95, 80, 70%. Slides were then incubated in deionized water for one minute prior to antigen retrieval as noted. After antigen retrieval, slides were incubated in 5% hydrogen peroxide in methanol to quench endogenous peroxidase activity. Slides were washed for 10 min in running tap water, 5 min in 0.1 M Tris, then blocked in 0.1 M Tris/2% fetal bovine serum (FBS). Slides were incubated in primary antibodies overnight. The following primary antibodies were used. For pathologically-phosphorylated α-synuclein, pS129 α-synuclein (EP1536Y; Abcam ab51253, RRID:AB_869973) was used at 1:20,000 with microwave antigen retrieval using citric acid-based antigen unmasking solution (Vector H-3300, RRID:AB_2336226). For phosphorylated tau, AT8 (pS202/T205; Thermo Fisher MN1020, RRID:AB_223647) was used at 1:10,000 without retrieval. To selectively detect AD tau, GT-38 (CNDR, [21]) was used at 1:1000 without antigen retrieval. For Aβ, NAB228 (CNDR, RRID:AB_2314850) was used at 1:10,000 with formic acid antigen retrieval.

Primary antibody was rinsed off with 0.1 M Tris for 5 min, then incubated with goat anti-rabbit (Vector BA1000, RRID:AB_2313606) or horse anti-mouse (Vector BA2000, RRID:AB_2313581) biotinylated IgG in 0.1 M Tris/2% FBS 1:1000 for 1 h. Biotinylated antibody was rinsed off with 0.1 M Tris for 5 min, then incubated with avidin-biotin solution (Vector PK-6100, RRID:AB_2336819) for 1 h. Slides were then rinsed for 5 min with 0.1 M Tris, then developed with ImmPACT DAB peroxidase substrate (Vector SK-4105, RRID:AB_2336520) and counterstained briefly with Harris Hematoxylin (Fisher 67–650-01). Slides were washed in running tap water for 5 min, dehydrated in ascending ethanol for 1 min each: 70, 80, 95, 100, 100%, then washed twice in xylenes for 5 min and coversliped in Cytoseal Mounting Media (Fisher 23–244-256).

Finally, slides were scanned into digital format on a Lamina scanner (Perkin Elmer) at 20x magnification so the digitized slides could be analyzed by quantitative pathology.

All annotation and quantification was done blinded to disease and genotype. All quantitation was performed in HALO quantitative pathology software (Indica Labs). After annotation of all sections for a given stain, all sections were analyzed simultaneously using the same optical density threshold. Optical density thresholds were empirically determined to not include any background signal and were as follows: EP1536Y (0.192), AT8 (0.355), GT-38 (0.213) and NAB229 (0.244). This signal was then normalized to the total tissue area and values were expressed as percentage area occupied by the immunostained pathologies. A minimal tissue optical density of 0.02 was used to exclude any areas where tissue was damaged.

Linear regressions were all performed in R (https://www.R-project.org/↗) [25] as described in the figure legends and results section, and the results of these tests are reported in the figure legends. Comparisons of pathology in subgroups (non-demented versus demented; pSyn positive versus pSyn negative) brains was done in GraphPad Prism 7 and results are reported in the figure legends.

Previous studies have described heterogeneous pathologies in LRRK2 mutation carriers, depending on the areas surveyed and antibodies used [13, 14]. In the current study, all autopsied cases in our brain bank with a documented LRRK2 variant were selected for quantitative pathology analysis. We present 12 cases: 10 p.G2019S variant carriers and 2 rare variants with unclear LRRK2 pathogenicity—p.L1165P and p.R793M. The two rare variants were previously described as possibly pathogenic [24], although subsequent research also found p.R793M in two healthy individuals [26], suggesting that R793M may be a rare polymorphism. Of the 10 p.G2019S carriers, 7 had a clinical diagnosis of PD, 2 had a clinical diagnosis of PDD, and 1 had a history of schizophrenia. While the schizophrenia case was included in histological analyses, it was not included in subsequent determination of pathology prevalence since the patient did not suffer from neurodegenerative disease. Both of the rare variant cases had a clinical diagnosis of PDD.

Seven brain regions (entorhinal cortex, Cornu Ammonis (CA) regions, dentate gyrus, visual cortex, amygdala, midbrain, and cingulate cortex) were selected to encompass regions that typically accumulate α-synuclein LBs, tau neurofibrillary tangles (NFTs), other tau inclusions, and Aβ plaques. We first examined the prevalence of α-synuclein pathology using an antibody targeting pS129 α-synuclein (pSyn, EP1536Y, Fig. 1). Excluding the schizophrenia case (p.G2019S-5), which had no pathology of any type, 63.6% (7/11) of cases showed prominent pSyn pathology, including 55.5% (5/9) of the p.G2019S cases. pSyn pathology was abundant in the substantia nigra, amygdala, hippocampus and cingulate cortex in cases where it was found and completely absent in the 5 negative cases. While α-synuclein pathology is typically a necessary neuropathological criterion for PD diagnosis, a small percentage of cases present clinically with PD and have nigral degeneration, but do not have LBs or another pathology to which the neuron loss can be ascribed. We found that 8.3% (9/109) of all iPD cases or 3.7% (9/242) of the iPD-PDD cases in our database met these criteria and therefore were classified as “atypical PD,” similar to the LRRK2 mutation carriers lacking LBs.

Next, we sought to determine the prevalence of tau pathology in LRRK2 mutation carriers using a pathogenic (pS202/T205) anti-tau antibody (AT8, Fig. 2). Surprisingly, 100% (11/11) cases had some tau pathology, including neurofibrillary tangles, neuritic plaques and neuropil threads, with only one PD case (p.G2019S-1) having rare inclusions. The schizophrenia case (p.G2019S-5) had no pathology. Areas with prominent pathology included the amygdala, hippocampus, and entorhinal cortex. The abundant hippocampal pathology in most cases and inclusion morphology suggest that the tau in these cases is AD-type tau, but we sought to unambiguously characterize the type of tau using an AD-selective anti-tau antibody.

Our lab has recently generated a series of monoclonal antibodies against an AD tau immunogen [21]. Many of these antibodies showed conformational selectivity for misfolded tau over monomeric tau, and intriguingly, two antibodies showed selectivity for AD tau over PSP, CBD or Pick’s disease tau. This selectivity seems to arise from the presence of both 3 and 4 microtubule-binding repeat isoforms (3R and 4R, respectively) in the tau aggregates as seen in AD. The other tauopathies are primarily composed of a single repeat isoform: PSP (4R), CBD (4R), and Pick’s (3R). This allows the selective detection of AD tau pathology only, even in the presence of abundant non-AD tau inclusions [21, 22].

We first confirmed the selectivity of one of these antibodies (GT-38) in PSP, Pick’s, CBD, and AD cases (Fig. 3). Staining of these cases with AT8 shows abundant tau pathology in all four cases and in various regions (Fig. 3a), consistent with presentation of these diseases. GT-38, in contrast, shows no staining of PSP, Pick’s, or CBD tau, but prominent staining of AD tissue (Fig. 3b). It should be noted that while GT-38 only recognizes AD tau, it seems to recognize a subset of that pathology, resulting in reduced overall staining (See Fig. 3b, especially entorhinal cortex and amygdala). Pathological α-synuclein and Aβ staining in these cases can be found in Additional file 1: Figure S1.

In order to determine whether the tau pathology observed in LRRK2 mutation carriers was indeed AD tau, adjacent sections to those stained with AT8 were stained with GT-38 (Fig. 4). As previously noted, GT-38 staining is not as strong, even in pure AD tissue, but all regions that were recognized by AT8 were also recognized by GT-38. Therefore, tau inclusions observed in LRRK2 mutation carriers are consistent with AD tau pathology. Aβ pathology was also abundant in most of the LRRK2 mutation carriers, consistent with comorbid AD pathogenesis (Additional file 1: Figure S2). Intriguingly, the two rare variant cases, p.L1165P and p.R793M had no Aβ pathology, despite having abundant AD tau. This finding is explored further in the discussion.

While pathology observations allowed us to make general conclusions about the pathology in LRRK2 PD cases, we sought a greater understanding of pathology and associations with clinical measures through quantitative pathology. All sections were digitized and gray matter regions were manually annotated in HALO software for automated quantitation. Once annotated, algorithms were developed that thresholded pathology detection by optical density. Thresholds were optimized for each immunohistochemical stain to provide broad detection of pathology in regions with both high and low density pathology, without picking up any background stain (Additional file 1: Figure S3). Pathology values are reported as the percentage of area occupied, and ranged over several log-fold (Fig. 5). Pathology values both by region (Fig. 5a) and mean pathology by case (Fig. 5b) are reported (Additional file 2: Table S1) and were used for subsequent analyses.

To understand quantitatively the degree to which the tau pathology observed in LRRK2 mutation carriers was AD tau, the percentage area occupied with GT-38 pathology was plotted against the AT8 pathology burden in the same areas (Fig. 6). Cases with primarily AD tau are expected to show a good correlation between the two pathology measures, while primary tauopathies would not. Indeed, the AT8 and GT-38 measures correlated very well in the AD case, but not in CBD, Pick’s or PSP cases (Fig. 6a). The p.G2019S, p.L1165P and p.R793M cases all clustered with the AD case, but not with the other tauopathy cases, suggesting that the majority of tau in these cases is AD tau. To further investigate the type of tau on a case-by-case basis, linear regressions were fit individually for each case (Fig. 6b). LRRK2 mutation carrier cases all continued to cluster and show positive correlation between AT8 and GT-38, with the exception of p.G2019S-1. This case has minimal tau pathology (Figs. 2, 4), and all points clustered in the bottom left of the graph, with an overall negative correlation. Even this case was distinct from the high AT8, low GT-38 clusters of CBD, Pick’s, and PSP cases.

While on average LRRK2 mutation carriers clustered with AD pathology, it is possible that some regions had non-AD pathology that was obscured by the high AD pathology in other regions. Having quantitative AT8 and GT-38 pathology measures provides the unique opportunity to examine regions in an unbiased manner for potential non-AD pathology. Regions that fell closer to the CBD, Pick’s and PSP clusters were more closely examined to determine whether any substantial non-AD tau was present. Several regions showed substantial AT8 immunoreactivity, but minimal GT-38 immunoreactivity (Additional file 1: Figure S4). Interestingly, these regions did not appear to have non-AD tau, but instead abundant neuritic plaques and low neurofibrillary tangles. This pattern is consistent with the preferential recognition by GT-38 of tangle pathology, and is a caveat of using this stain to rule out the presence of non-AD tau. Using this strategy, we were unable to identify substantial non-AD tau in the regions and cases surveyed.

One of the questions we sought to address is whether tau is the neuropathological substrate of neuron loss in the substantia nigra of LRRK2 mutation carriers who lack LB pathology (p.G2019S-3, 4, 6, 7). Each of these cases showed minimal tau inclusions in the midbrain by both AT8 and GT-38 independently, suggesting that tau is not likely to be causing neuron loss in this region (Fig. 2). However, midbrain tau pathology is more often associated with non-AD tauopathies, so we chose to investigate the type of tau present in these cases more closely. Co-immunofluorescence of AT8 and GT-38 in sections from the midbrains of LB-negative LRRK2 mutation carriers was used to assess the type of tau in these cases (Fig. 7). The AD and PSP cases were used as controls. Tau in the AD midbrain was mostly recognized by both AT8 and GT-38, suggesting that a similar strain of tau is present throughout the brain of this AD case. In contrast tau in the PSP midbrain was AT8-positive, but showed no immunoreactivity with GT-38. The sparseness of tau pathology in the midbrain of LRRK2 mutation carriers was confirmed in this experiment, but intriguingly most of the tau inclusions in these brains were recognized by both AT8 and GT-38, suggesting that it is predominantly AD tau. One GT-38-negative tau inclusion was observed in p.G2019S-4, but this type of inclusion was rare.

While LRRK2 PD resembles idiopathic PD in many ways, the presence of a substantial percentage of cases without α-synuclein pathology has led to some confusion about the neuropathological substrate of disease in these mutation carriers. Our broad survey of brain regions in LRRK2 mutation carriers has shown that α-synuclein pathology is absent not only in the substantia nigra, but throughout the brain. This suggests that LBs are not simply cleared from this region after neuron death, but that LBs were likely not ever present in these brains. This discrepancy is one of the major reasons tau has been investigated as a neuropathological substrate of disease in these cases. If tau is indeed responsible for motor symptoms in these patients, higher tau pathology might be expected in LB-negative cases. However, we found that while none of the co-pathologies is associated with Aβ pathology (Fig. 8 a-c; Additional file 1: Figure S5a-c), both AT8 and GT-38 tau pathology burden are positively correlated with α-synuclein pathology (Fig. 8 d, e; Additional file 1: Figure S5d, 5e), suggesting that tau and α-synuclein pathology burden are associated. This association could be due to a common factor precipitating both pathologies, the two pathologies enhancing each other indirectly, or the misfolded proteins directly cross-seeding each other. To assess cross-seeding, cases which had both α-synuclein and tau pathologies were visualized by co-immunofluorescence (Additional file 1: Figure S6). While both pathologies were abundant in many of the same regions, they were rarely in the same cell, suggesting that the association of these two pathologies is not through direct cross-seeding.

To further investigate tau pathology burden specifically in pSyn-positive or pSyn-negative cases, average pathology was grouped by pSyn pathology level (Fig. 8 f-i). Contrary to tau acting as an independent neuropathological substrate, tau pathology was higher on average in cases with pSyn co-pathology (Fig. 8 f-h). However, pSyn pathology was not the only indicator of tau burden, since some cases without pSyn pathology had moderate tau pathology and some cases with pSyn pathology had low tau pathology. Aβ pathology was variable and had no relationship with pSyn pathology (Fig. 8i).

LRRK2 mutation carriers have been reported to have a more prominent tremor [27] and slightly slower disease progression [18], although the presentation of disease is largely overlapping with iPD. In a recent clinicopathological survey of 37 LRRK2 mutation carriers, progression to dementia was associated with elevated LB pathology [14]. We therefore sought to determine how pathological measures were related to clinical features, including progression to dementia, age at death, and disease duration. Of the 12 LRRK2 mutation carriers in our cohort, 11 presented clinically with PD, and 4 of those developed dementia over the disease course. Previous research has found that this progression to dementia in idiopathic patients is associated with elevated α-synuclein and tau pathology, especially in cortical regions [4, 9]. LRRK2 mutation carriers show a similar pattern, with elevated AT8 (Fig. 9a), GT-38 (Fig. 9b) and pSyn (Fig. 9c) in carriers with dementia. In contrast, Aβ levels in these patients showed no association with dementia (Fig. 9d). While age at death (Additional file 1: Figure S7a), age of disease onset (Additional file 1: Figure S7b), and disease duration (Additional file 1: Figure S7c) showed no significant association with dementia, there was a positive correlation between age at death and AT8 (Fig. 9e, Additional file 1: Figure S7e), GT-38 (Fig. 9f, Additional file 1: Figure S7f), and pSyn (Fig. 9g, Additional file 1: Figure S7 g) pathologies. This suggests that tau and α-synuclein pathologies increase with age, but additional factors are responsible for the elevated pathology in individuals with dementia. Aβ pathology showed a negative correlation with age (Fig. 9h, Additional file 1: Figure S7 h). Tau and Aβ pathologies were not associated with disease duration (Fig. 7i, 9j, l, Additional file 1: Figure S7i, 7j, 7 l), and pSyn showed mild positive correlation with disease duration (Fig. 9k, Additional file 1: Figure S7k).

AD neuropathology has recently been described as a major co-pathology in PD-PDD and is associated with progression to dementia [4, 9]. Therefore, we sought to determine whether the AD tau pathology observed in LRRK2 PD-PDD is similar to that observed in iPD-PDD. Using the AD-selective tau antibody GT-38, LRRK2 mutation carriers were staged using Braak AD tau stages 0-VI (Fig. 10a). Most LRRK2 mutation carriers had mild to high levels of AD tau pathology, although none reached Braak stage VI. When compared to tau staging in 133 previously published PD-PDD cases, LRRK2 PD had a similar prevalence of low to intermediate tau pathology (Braak I-IV) and high tau pathology (V-VI, Fig. 10b). The breakdown in stages I-II versus III-IV is slightly different in LRRK2 mutation carriers versus iPD-PDD, although this could be partially due to the low number of LRRK2 mutation carriers.

The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.