The characterization of AD/PART co-pathology in CJD suggests independent pathogenic mechanisms and no cross-seeding between misfolded Aβ and prion proteins

Creutzfeldt-Jakob disease (CJD) and Alzheimer’s disease (AD) are protein-misfolding disorders characterized by auto-propagation and tissue deposition of protein aggregates leading to progressive neuronal dysfunction and neurodegeneration.

CJD, the most common human prion disease with an estimated incidence of 2 cases per million, comprises several clinical-pathological phenotypes and uniquely occurs in three forms with apparent distinct etiologies [6, 38]. They include sporadic cases of unknown origin, a genetic form linked to mutations in the prion protein gene, PRNP, and an infectious form acquired through medical procedures or contaminated food. Current classification of sporadic (s) CJD, the most common form, includes six major disease subtypes that are primarily determined by the combination of the genotype at the polymorphic codon 129 of PRNP, encoding for methionine (M) or valine (V), with either one of two types (1 and 2) of misfolded prion protein (PrP^Sc) that differ in the size (21 and 19 kDa) of their protease-resistant core [55, 56, 58, 59]. Notably, experimental transmissions to syngenic animals demonstrated that prion isolates from 5 of the 6 sCJD subtypes maintain distinctive phenotypic features upon serial passages, supporting the view that they represent different strains of prions [7, 34, 54, 74].

AD, the most common cause of dementia in humans with an estimated prevalence of about 10% in the elderly over 65 years of age [3], involves the misfolding of two proteins. Accordingly, the pathological hallmarks of AD comprise extracellular amyloid plaques consisting of aggregated amyloid beta (Aβ) peptide and intracellular neurofibrillary degeneration (NF), made of paired helical filaments of hyperphosphorylated tau (p-tau) forming neurofibrillary tangles (NFTs), neuropil threads (NThs) and abnormal neurites (Nts).

The amyloid cascade hypothesis, the most widely accepted theory for the molecular sequence of pathological events in AD, postulates that the initial tissue changes in the AD brain involve the aggregation and deposition of the Aβ peptide [33], while the hyperphosphorylation and polymerization of tau into NFTs, NThs, Nts and, ultimately, neuronal loss would occur later. However, the fact that a significant proportion of elderly people accumulates tau earlier or even in the absence of Aβ, and that the first site of brain deposition of the two proteins does not match is apparently in contrast with the primacy of Aβ over tau in AD pathogenesis [9]. To accommodate this apparent discrepancy, a novel neuropathological condition named primary age-related tauopathy (PART) has been recently introduced [18]. The unbound investigation of the pathogenesis and relative risk factors of NF and Aβ pathology allowed by the independent categorization of brains which accumulate NFTs in association with minimal or no Aβ will, hopefully, shed light on the fundamental question of whether PART is indeed a distinct entity unrelated to AD.

Recent experimental evidence suggests that both Aβ and tau aggregates become self-propagating during disease in a prion-like manner [16, 40, 50, 66, 77]. This notion, combined with the frequent coexistence of AD pathology in CJD brains [25, 36], raises the question of the potential cross-seeding between PrP^Sc, Aβ and tau. Interest into the issue has also been brought up by the finding that the normal cellular prion protein, PrP^C, can bind different Aβ species, especially oligomers [23, 67], and act as a receptor that internalizes the Aβ aggregates facilitating their neurotoxicity. Nevertheless, current evidence indicating a pathogenic role of PrP in Aβ formation or a synergistic effect between Aβ and prion pathology remains controversial [51, 61, 69, 70].

Conflicting results also concern the frequency of the association between the two pathologies [25, 27, 28, 36, 45]. Moreover, the question of whether the apolipoprotein E (APOE) and PRNP genotypes, the most significant genetic modifiers of Aβ and PrP^Sc pathologies, also affect other proteinopathies, including tau neurofibrillary degeneration, remains largely unsolved [4, 10, 11, 13, 17, 19, 21, 24, 26, 37, 47, 52, 62, 68, 70, 78].

To contribute knowledge on these issues, we thoroughly characterized the AD/PART pathology spectrum in a series of 450 cases with definite CJD, analyzed the effect the CJD molecular subtype, the M/V polymorphism at codon 129 of PRNP, and the presence of PRNP mutations on the co-occurring neurodegenerative pathology, and evaluated the effect of APOE genotype on both CJD and AD/PART pathologies.

This is the largest study to date on the characterization of AD/PART co-pathology in a CJD cohort representative of both sporadic and genetic forms and all major disease subtypes. Several previous studies provided evidence of a significant association between CJD and AD in both humans and animal models [25, 36, 79]. These data, combined with others indicating that PrP^C acts as a receptor mediating the internalization of Aβ aggregates, facilitating their neurotoxicity, are often taken as evidence of shared pathogenic pathways between the two disorders [35, 65]. Similarly, the recent observation of an unexpected significant Aβ accumulation in the brain of young people affected by iatrogenic CJD, following dura mater graft or treatment with cadaveric-derived human growth hormone (hGH) [12, 32] raised speculations about the possibility of a cross-seeding between PrP^Sc and Aβ fostering the formation of Aβ plaques, although the recent demonstration that Aβ deposition is equally high in cadaveric hGH recipients without CJD, suggests that Aβ deposition is triggered by the iatrogenic inoculum independently from the prion pathology [22, 63, 64].

Overall, our results indicate that AD and CJD behave as independent disease processes even when present in the same brain, and strongly argue against shared pathogenic mechanisms between these disorders, which is in line with the conclusions of other investigators based on case-control [17, 60, 80], experimental [62], in-vitro [15], and neuropathological studies [28]. As the only possible exception, we observed a trend towards an increased tau- and Aβ-related pathology in gCJD patients carrying the V210I variant. However, the effect of a shared genetic background related to a common founder, which would fit with the clustering of the mutation in the Italian population, provides an alternative plausible explanation. Furthermore, we detected a significantly lower degree (i.e. lower mean Braak stage) of NF pathology in the CJD subtypes VV2 and MV2K (i.e. those linked to the V2 strain) in comparison to typical CJDMM(V)1, despite the comparable Aβ pathology. The finding may suggest a specific host-genotypic effect on tau pathology, possibly related to PRNP codon 129. Given that the CJD MV2K and VV2 subtypes show, on average, more severe pathological changes, including the formation of dot- or stub-like tau deposits in the MTL in comparison to CJDMM(V)1 [39, 56], it is reasonable to believe that the more severe neurodegeneration and the more widespread formation of prion-related tau deposits, possibly causing the sequestration of tau proteins, might have prevented the formation of NF degeneration or made it less visible in such cases. Interestingly, the previous finding that CJDVV2 cases often demonstrate an atypical sub-regional distribution of tau NF degeneration in the hippocampus, and that this occurs more frequently than in other subtypes [35], seems to support this interpretation. Thus, the reduced level of NF pathology in prion strain V2-related subtypes likely reflects the difference in the primary prion pathology. Nevertheless, the possibility that genetic factors linked to codon 129 genotype influence the burden of tau pathology in the MTL in the CJD affected brains cannot be excluded.

The analysis of tau neurofibrillary co-pathology in CJD deserves a further comment. Indeed, at variance with Aβ pathology, tau pathology not only includes the NF degeneration in the cerebral cortex that is spatially linked to senile amyloid plaques (i.e. Braak stages IV-VI), which is unanimously considered AD-related, but also the NF pathology fitting the definition of PART. This adds further complexity since PART and AD pathologies distribute differently in the elderly population, and there is initial evidence indicating that they might be driven by different genetic risk factors [42]. Consequently, they may also differ with respect to their association with CJD. In the present study, we found that PART is equally distributed between cases stratified according to PRNP codon 129 genotype, CJD subtype and strain, which support the conclusion of PART and CJD being independent pathological processes.

The strong association between APOE ε4 and AD risk [29, 41, 73], and the protective effect of APOE ε2 are well established. Our data demonstrating a significant positive association between APOE genotype and the level of AD pathology, the Aβ phase and the presence of CAA in the analyzed CJD population are in full agreement with these notions. As in previous studies [20, 44, 71, 72], we also investigated the association between APOE genotype and tau pathology (Braak stage) independently from Aβ accumulation (Thal phase). We found no differences in the relative frequency of APOE ε4 and ε2 genotypes between tau pathology stages. This result is likely influenced by the fact that, in the large majority of our cases, tau pathology was restricted to the hippocampal region, which as discussed above, might be largely unrelated to AD.

Previous studies also analyzed the influence of APOE on CJD, again with discordant results [4, 11, 14, 68, 75, 76]. Recent meta-analysis data involving 1001 CJD patients and 1211 controls suggested an increased risk of developing CJD that is proportional to the number of APOE ε4 alleles [78]. Furthermore, proteomic studies pointed to APOE as a potential biomarker for prion disease [30, 48] by showing significantly elevated levels of the protein in prion-infected mice [49] and its co-localization with PrP^Sc deposits in vivo [53]. In the present study, we systematically analyzed the APOE genotype across the whole spectrum of CJD subtypes. We found no differences in the distribution of APOE ε4 and ε2 genotypes between CJD subtypes.

Moreover, the presence of a specific APOE genotype did not influence the age at onset or the disease duration in our cohort. These observations suggest that APOE has no role in the pathogenic mechanism determining the CJD strain and the disease subtype and make the hypothesis of a conformation-specific interaction between PrP^Sc and APOE unlikely. Finally, the finding that APOE is strongly correlated to Aβ pathology but not to CJD pathology further corroborates the view supported by the whole data of the present study, of largely independent pathogenic mechanisms between AD and CJD.

From the clinical point of view, our results indicate that the AD co-pathology does not have a major impact on the clinical presentation of CJDMM(V)1, likely because of the subacute onset and rapid progression of the disease, although it appears to slightly increase the frequency of cognitive symptoms. The possibility that the AD co-pathology might more consistently favor an earlier appearance of cognitive disturbances in other sCJD subtypes, especially those lacking cognitive decline at onset (e.g., VV2 subtype) or manifesting a slower course (e.g. MV2K subtype) remains a possibility.

By showing that the age-related prevalence and profile of AD/PART co-pathology in CJD is comparable to those described in the normal aging population and demonstrating the lack of any correlation between variables known to affect CJD pathogenesis and those defining the AD/PART pathology spectrum, the results of the present study strongly argue that PART, AD and CJD represent independent disease processes even when present in the same brain.