Introduction

Alzheimer’s disease (AD) is a progressive neurodegenerative disease that presents with debilitating memory and cognitive impairments, and accounts for between 60 and 80% of all dementias worldwide [100]. Whilst the prevalence and the consequent economic and social impacts of AD are predicted to increase each year as the global population ages, the exact aetiology of AD is still unknown, making the development of effective treatments extremely challenging [20]. Accumulations of β-amyloid (Aβ) exhibited as extracellular plaques and cerebral amyloid angiopathy (CAA), and accumulations of hyperphosphorylated tau present as paired helical filaments (PHFs) and neurofibrillary tangles (NFTs) are accompanied by significant neuronal loss and constitute the primary histopathological hallmarks of the disease [11, 20].

One particular challenge of using rodent species for preclinical studies is that rodents do not naturally exhibit either Aβ or tau pathology as they age [26]. To overcome this obstacle, it has been necessary to create transgenic mouse models overexpressing Aβ [68]. Such models have been used to elucidate the significance of mutations in the amyloid precursor (APP), and presenilin proteins (PSEN1 and PSEN2) that cause familial AD (FAD) [26]. Mutations in these proteins affect APP metabolism, leading to an increased production of the amyloidogenic form of Aβ (Aβ1–42), and impaired Aβ clearance [38]. These events form the basis of the ‘amyloid cascade hypothesis’ to explain the pathogenesis of AD, in which it is proposed that accumulation of Aβ1–42 is the primary pathological event that drives all other associated pathologies (including tau pathology, inflammation, vascular damage, and neuronal loss) [36]. This hypothesis, however, does not explain the mechanisms by which soluble and/or insoluble forms of intracellular and/or extracellular aggregates of Aβ and tau differentially affect one another [50, 75]. Indeed, while transgenic mouse models using APP and PSEN mutations present with significant Aβ pathology between three and six months of age they often fail to exhibit significant tau pathology or neuron loss [26]. Consequently, tau pathologies within these rodent models are achieved only by introducing tau mutations that cause tau pathology in other dementias, namely those associated with frontotemporal dementia with parkinsonism-17 (FTDP-17) [26]. This, combined with the relatively short lifespan of rodents means that the capacity of transgenic mouse models to fully reflect the aetiological mechanisms and temporal progression of familial AD is limited [81]. Furthermore, given the lack of naturally occurring age-related neuropathology, the ability of mouse models to recapitulate mechanisms associated with ‘sporadic’ AD (that is not associated with mutations in any of the FAD genes and accounts for more than 95% of AD cases), is also extremely restricted [22].

Increasing recognition of the limitations associated with rodent models of AD has led to the investigation of species that have a longer lifespan, a more physically and functionally differentiated brain, and a propensity to naturally develop both Aβ and tau pathology with age [23, 80]. Non-human primates such as the chimpanzee (Pan troglodytes), rhesus macaque (Macaca mulatta) and common marmoset (Callithrix jacchus) are of particular interest, because they naturally develop some AD-like pathology [41]. For example, diffuse and dense-cored Aβ plaques, and CAA have been detected in aged chimpanzees [25, 28, 30, 78], rhesus macaques [79, 96] and marmosets [35, 58], with quantities of Aβ in aged individuals comparable to levels seen in AD patients. NFTs have also been described within the entorhinal cortex of aged chimpanzees and rhesus macaques [4], while in marmosets, abnormally phosphorylated tau has been identified as early as adolescence [77]. A range of other species have been found to exhibit spontaneous AD-like pathology. Such species include the gorilla (Gorilla gorilla gorilla [71] and Gorilla beringei beringei [72]), Ursidae species [16, 88, 95], cetacean species [88], and pinniped species [92]. The domestic dog (Canis familiaris) [17, 18, 80, 86] and cat (Felis catus) [12] also develop cognitive decline alongside diffuse Aβ plaques in old age. However, tau pathology rarely accompanies Aβ deposition and dense-cored Aβ plaques and NFTs are not consistently detected. Whilst non-human primates, dogs and cats have the potential to mitigate the limitations of current rodent AD models, they are limited in terms of their ethical use and, therefore, their numbers available to facilitate robust experimental design.

We chose to focus on sheep as a potential animal model of human AD for a number of reasons. Sheep have a moderate lifespan (15–20 years), an extensively annotated reference genome [31, 54, 60, 73, 97]. Thus, the S396 antibody is commonly used to study early tau hyperphosphorylation in a range of species and AD models [1, 29]. In contrast, the phosphorylation of residues serine202 and threonine205, detected by the AT8 antibody, occurs in the latter stages of tau pathology [32]. Although it is likely that the absence of mature NFTs in the stock sheep reflects the relatively young age of the sheep, the presence of pre-tangles is indicative of early-stage tau pathology in sheep as young as 5–8 years of age [74]. As the preliminary study of a 21-year-old sheep detected numerous AT8-positive mature NFTs, it is likely that older sheep display advanced stage tau pathology as observed in AD, and that the pre-tangle pathology observed in younger sheep may eventually progress to this advanced stage. The variation in spontaneous tau and Aβ pathology observed between the very aged sheep in the pilot study also reflects the significant variability of tau and Aβ pathology observed in aged humans, further pointing to the sheep as a relevant animal model.

Phosphorylation of the serine396 residue was also investigated. We found phosphorylated S396 immunoreactivity was present in some sheep 2 years of age and that phosphorylation increased with age. While the observation of this pathology at such an early age may be surprising, it is consistent with human data where it has been shown that abnormal tau changes occur in the brain as early as childhood or puberty [8, 9]. Recent evidence also suggests that neurons undergo a dying back pattern of degeneration, where tau pathology begins in the axonal compartment before progressing to the somatodendritic compartment [14, 51]. The differential patterns of immunostaining within the CA3 region of the hippocampus observed using different anti-Aβ1–42 and anti-tau antibodies in this study may indicate early hippocampal intraneuronal Aβ deposition, accompanied by early-stage axonal tau phosphorylation. The presence of mOC64 and S396 immunopositivity may also indicate that these changes occur in sheep as young as approximately 2 years of age. The pattern of S396 labelling within the CA3 region of the hippocampus is remarkably similar to labelling detected using the same antibody in the hippocampus of cows with idiopathic brainstem neuronal chromatolysis (IBNC) [48]. While the exact aetiological cause of IBNC is unknown, studies have indicated that it is a complex proteinopathy characterised by significant hyperphosphorylated tau with associated secondary accumulations of alpha-synuclein and ubiquitin, without associated NFT formation or amyloid deposition [48]. Interestingly, while cases of IBNC are predominantly reported in cows > 6 years old, the youngest recorded case was in a cow just 4 years old. Therefore, further research into tau phosphorylation and tauopathy in sheep and large ruminants would be beneficial to determine if these two pathologies are related.

The current study identified diffuse plaque-like tau-positive deposits that were not characteristic of AD-like tau pathology. We found no background staining in the tissue surrounding these focal regions. All control slides were also negative, and the pattern of staining with the AT8 antibody on both AD-positive and AD-negative human control tissue revealed staining only of the AD brain. Given that the regional and age-dependent distribution of these AT8-positive elements is consistent with a progressive pathogenesis, it is possible that the AT8 antibody in this instance was accurately labelling phosphorylated tau. However, equally, while the use of the AT8 antibody is renowned for achieving specific phospho-tau labelling with little or no background signal [7, 30], this pattern of atypical staining may have resulted from non-specific antibody staining, tissue treatment artefacts, or cross reactivity with another species of tau. Further investigations into possible sources of this atypical AT8 labelling in sheep are necessary.

Tau isoform expression

In humans, all six of the tau isoforms expressed in the brain become hyper-phosphorylated and involved in tau pathology. MAPT splicing and tau isoform expression, however, differs between species, affecting the subsequent formation of tau pathology [40, 84, 92]. This study confirmed the expression of the four distinct ovine tau isoforms described by Janke et al. [47], formed from the splicing of exons 3 and 10. This pattern of ovine tau expression is consistent with the phylogenetic expression of tau in other ruminants [43, 47]. While exon 8 has previously been identified in transcripts from the rhesus macaque and the cow [66] the ovine tau transcripts generated in this study confirmed that exons 4a, 6 and 8 are not transcribed in sheep (for comparisons of tau expression between species, see Fig. 10). Janke et al. [47] also described two additional isoforms in sheep, interpreted as the two 0N tau isoforms, consistent with the two shortest tau isoforms expressed in humans. However, we found no evidence for the expression of these isoforms, resulting from the alternative splicing of exon 2. The confirmation that sheep express the 3R and 4R tau isoforms that are expressed in humans is an important step in assessing of their suitability as an AD model, particularly given that mice naturally express only 4R tau [3]. Possessing both 3R and 4R endogenous tau isoforms confers considerable potential to the sheep as an AD model with a highly translatable value.

The variations in the tau N-terminal domains of different species have been cited as reasons why tau pathology differs between species [3]. The human N-terminal domain contains an eleven amino acid motif (residues 17–28) not present within the N-terminal domain of murine tau [40]. This additional peptide sequence is thought to affect the intramolecular interactions between the N- and C-terminals, and the microtubule-binding domains of tau, increasing the likelihood of normal tau undergoing pathological conformational changes [3, 40]. The lack of this N eleven amino acid motif in murine tau has been suggested as one reason why mice do not naturally develop tau pathology [40]. Additionally, within the first 190N-terminal amino acids, human tau contains 33 serine, threonine, and tyrosine residues. Murine tau shares 22 of these residues. Many of the residues, found in humans (but lacking in mice) are phosphorylated in AD, or are phosphorylated by kinases that exhibit dysregulation in AD [37]. However, an alignment of the human, murine and ovine tau proteins show that sheep lack the eleven amino acid sequence, and many of the same phosphorylation sites as murine tau, even though they naturally develop tau pathology [7, 64, 65, 74]. We speculate that while the presence of the 11 amino acid motif and the phosphorylation of these residues may accelerate pathological tau formation in humans, their involvement is not essential for the development of PHF tau.

Proline-rich regions facilitate the phosphorylation of serine and threonine residues via kinases thought to have key roles in the pathogenesis of AD [44]. There are seven distinct variations between the human and ovine proline-rich regions of tau that may be implicated in tau hyperphosphorylation in the sheep. For example, proline176, alanine178 and proline182 in human tau are substituted with threonine, threonine and serine, respectively, in ovine tau. However, proline residues that are important for the directed phosphorylation of tyrosine residues such as proline213, proline216 and proline219 are conserved between humans and sheep [40, 56]. Additionally, each of the four KXGS motifs which can be phosphorylated by multiple kinases, and the four PGGG sequences which facilitate the formation of type II β-turns and β-hairpin structures, are conserved between humans and sheep [13, 21].

Future work

As only three other studies have applied phosphorylation dependent antibodies to study the ovine brain, further work is required to elucidate the residue specific patterns of tau phosphorylation in sheep as they age. Although this study focused on the hippocampus and entorhinal cortex, critical regions for the development of AD pathology, future studies would benefit from analysing other cortical and subcortical structures that have been implicated in the early tau pathology of both humans and aged canines, such as the thalamus [1]. The investigation of other tau post-translational modifications, such as truncation would also be valuable as truncation has been associated with increased tau aggregation in AD [55]. A complete evaluation of other protein pathologies, markers of astrogliosis and neuroinflammation, such as those involving ubiquitin, APOE and GFAP would also be beneficial since this would provide a more complete picture of AD-related pathology in sheep [57]. Environmental factors that could increase the likelihood of sheep develo** AD-like pathologies also need to be investigated. For example, both copper (to which sheep are sensitive), and rumen-protected feed ingredients such as formaldehyde are factors that have been implicated in AD pathogenesis [48, 87, 99]. They may contribute to the AD-like pathology observed in other young ruminants. Future studies will also benefit from a complete evaluation of potential heterozygous substitutions and single nucleotide polymorphisms of the ovine MAPT transcript. Finally, screening sheep for relevant alleles, such as APOEε4 may help to enhance the detection of AD-like pathological characteristics in the commercial sheep population, increasing their potential suitability of being a natural AD model [74, 94].

Conclusion

We identified intracellular and vascular Aβ deposition in the brains of sheep as young as 2 years of age. We also identified AT8-positive pre-neurofibrillary tangles in sheep > 5 years of age. One of the early stages in the conversion of normal to pathological tau is the phosphorylation of multiple residues. We show that even by 2 years of age, some phosphorylation of the serine396 residue is present in sheep, and the relative amount of tau phosphorylated at this residue increases with advancing age. Given that these findings are consistent with early-stage AD-like pathology, sheep could be used to investigate the early biochemical and structural changes that eventually result in advanced AD-like pathology, without the complications of late-stage pathology. This would be an important step, given that the disease course of rodent transgenic AD models often progresses too rapidly for a comprehensive evaluation of this pre-clinical phase of AD aetiology to be completed [23]. Our pilot study detected numerous NFTs in a sheep 21 years of age, indicating that the development of late-stage tau pathology is possible in older sheep. This study has also confirmed that sheep express four distinct tau isoforms in their CNS, composed of three and four binding repeats, which result from the alternative splicing of exons 3 and 10. As a result, the sheep, either as a spontaneous model with genetic predisposition or, as a transgenic model, has the potential to provide a valuable model of early-stage AD-like tauopathy, reflective of that seen in humans.