Shaping the future of preclinical development of successful disease-modifying drugs against Alzheimer's disease: a systematic review of tau propagation models

A comprehensive literature retrieval was performed to identify original studies investigating tau propagation upon intracerebral inoculation of various tau inocula in Tg/WT mice. We gathered the pertinent literature from four independent databases: PubMed, Scopus, Web of Science, and Google Scholar. For the literature search, we used “tau” AND “spreading” OR “propagation” AND “injections” OR “seeding” OR “inoculation” AND “animal models” OR “mouse” OR “rat” as the keywords. The search was restricted to the English language with no restrictions on the date of publication. In Google Scholar, only those articles whose titles included the keywords were selected, to achieve a fairly accurate retrieval. We evaluated the qualifications of the animal studies independently according to the inclusion criteria by screening the abstracts and methodology sections of the identified publications.

The inclusion criteria for in vivo modelling of tau propagation were as follows:

The following information from each study is listed in Table 1, 2, and 3:

Through a database search we identified 55 relevant studies that have utilized a wide range of tau propagation models for intracerebral inoculation to model tau propagation in vivo (Fig. 1).

The biochemical and biophysical characterization of tau inoculum is the first vital step toward comprehending one half of the pathological process involved in its propagation. Based on its origin, the tau inoculum utilized in experimental tau propagation models can be categorized as follows:

Human brain-derived inocula were extracted from the isocortex, allocortex, amygdala, or subcortical nuclei of human Alzheimer’s disease [4, 7, 11, 15, 19, 29, 45, 47, 48, 58, 65, 67, 69, 71, 74, 96, 107, 115, 116, 120, 138, 140, 156, 161], Down syndrome (DS) indistinguishable from AD (DSAD) [19], argyrophilic grain disease (AGD) [29, 48], Pick’s disease (PiD) [29, 45, 71], tangle-only dementia (TD) [29], globular glial tauopathies (GGT) [45, 49], corticobasal degeneration (CBD) [19, 29, 71, 115, 116, 171], primary age related tauopathy (PART) [45], aging-related tau astrogliopathy (ARTAG) [45], frontotemporal dementia with parkinsonism linked to chromosome 17 (FTLD-17) [45, 161], and progressive supranuclear palsy (PSP) [29, 45, 71, 115, 116] brains (Fig. 2). The choice of region from which the inoculum was isolated was based on the anatomical distribution of tau inclusions, which varies greatly depending on the disease and its stage. In fact, tau from a single brain can exhibits variation in its proteopathic potency based on the region [82], or within the region whether it is derived from grey or white matter [165], moreover fractions show differences even within a single isolation [102]. Additionally, the tau inclusions have distinct morphological characteristics and cell-specificities [28, 114] and are composed of either 3R, 4R, or both isoforms with disease-specific filament folds [43, 44, 137, 147, 176].

Rodent brain-derived inocula were extracted from the cortices, brainstems, or spinal cords of transgenic mice (P301ST43 [1, 30, 80], rTg4510 [146]), or rats (SHR24 [100], SHR72 [100]), taking into account the regional distribution of tau inclusions. These inclusions are driven by the transgenic expression of either human mutant tau such as P301S [5], or P301L [128, 134] or by truncated tau (aa151-391) found in sporadic AD [51, 178] and are composed of both transgenic and if not knocked out, endogenous rodent tau. The 3R/4R isoform and the filament-fold are either reported to be NFT-like or remains to be determined at large.

From both human and rodent brains, extracellular vesicles, particularly exosomes, were derived from prodromal AD, mild cognitive impairment (MCI), AD, PSP or PiD [99, 131] or from the iPSCs derived from AD patients [12, 163], as well as from brain extracts from Tg rodent models of tauopathy [13]. Given that the neuronal cells are recognized for producing exosomes that transport cargo such as proteins, RNA-binding proteins (RBPs), and RNAs to neighbouring cells, exosomes were initially considered and subsequently confirmed to be vehicles facilitating the cell-to-cell transmission of oligomeric tau [16, 125]. The 3R/4R isoform composition of tau with characteristic filament-folds within these exosomes is specific to the disease or model.

Human/rodent brain-derived insoluble tau inoculums were prepared as either enriched detergent-insoluble protein aggregates or, in some cases, as crude protein extracts (Tables 1, 2, & 3). It is imperative to underscore that within the widely utilized sarkosyl-insoluble protein aggregate fraction, tau comprises a mere 10% of the total proteins [39]. This fraction encompasses additional constituents, including β-amyloid (Aβ), snRNP70 (U1-70 K), apolipoprotein E (ApoE) and complement component 4 (C4-A), among others [39, 64]. With the aim of obtaining a homogenous sample of aggregated tau, certain studies have also incorporated, additional fractionation steps, employing techniques such as immunoprecipitation (IP) or fast protein liquid chromatography (FPLC) [9]. These fractions also consisted of a diverse pool of fibrillary, filamentous, multimeric, and oligomeric units of pathological tau.

The propagation potential of oligomers, ranging from 3 to 100 molecules (< 30 nm long fibrils) [26, 80, 110, 160] along with larger insoluble tau aggregates [80] is ongoing. Although, it remains unclear which molecular entity among these is capable of propagation, tau aggregates were broken down into smaller fragments through differential sonication protocols before inoculation. This step is aligns with the prevailing notion that a soluble, high-molecular-weight (HMW) oligomeric form of tau may exhibit comparable or even heightened bioactivity in terms of propagation across neural networks [106, 146] (Tables 1, 2 and 3).

The comprehensive characterization of brain-derived tau inocula involved a multifaceted approach employing various techniques. Initial assessments utilized Western blot (WB) [15, 29, 45, 48, 69, 74] and enzyme-linked immunosorbent assay (ELISA) [1, 4, 15, 19, 116, 138, 166, 176]. Some studies opted for the bicinchoninic acid (BCA) assay [1, 4, 7, 45‐48, 58, 67, 116], nanodrop spectrophotometry [120, 166] for total protein estimation. Further biophysical characterization of the tau inoculum employed circular dichroism (CD) spectroscopy [96], transmission electron microscopy (TEM) [11, 47, 116, 140, 164], atomic force microscopy (AFM) [95, 120, 146], fast protein liquid chromatography (FPLC) [95], immunoelectron microscopy (IEM) [30], and size exclusion chromatography (SEC) [42, 67, 99]. In the preparation of exosomes, the human/rodent brain homogenates were subjected to sequential differential centrifugation, ultracentrifugation, and ultrafiltration or size exclusion chromatography (SEC) to further purify and concentrate the extracellular vesicles. While the proteomic composition of these vesicles exhibits considerable variability, it has been reported that tau is enriched among the cargo they carry. Furthermore, in order to determine the size and abundance of the exosomes, nanoparticle tracking analysis (NTA) was performed [12, 131, 163], followed by characterization using silver gel staining, electron microscopy, proteomics, and WB/ELISA of the exosome lysate or AFM to validate the presence of tau oligomers.

WB analysis provided insights into the distinctive band patterns of the 3R and 4R tau isoforms, along with disease-specific phospho-tau signature. In the case of AD, the fraction typically exhibited characteristic band patterns with three bands at 68, 64, and 60 kDa, accompanied by a weak upper band of approximately 73 kDa. Contrarily, fractions related to ARTAG, GGT, PSP, and FTDP-17 exhibited two bands of 68 and 64 kDa, which are specific to 4R tau tauopathies [25]. PiD, in contrast, exhibited two bands at 64 and 60 kDa, distinctive of 3R tau tauopathies [62]. Additionally, several lower bands indicated the presence of a fragmented pool of tau spanning between 50 and 25 kDa [62].

The exact concentration of tau was determined through ELISA, while a rough estimation of total protein was made using BCA or nanodrop spectrophotometers. Notably, across the studies, different amounts were inoculated, and as expected, a “dose-dependent” effect on the resulting pathology load was observed [8, 107]. This suggests that the progression of tau pathology is influenced by the quantity of propagation-capable tau in the inoculum, with the understanding that this relationship may not necessarily follow a linear pattern [108]. To confirm the presence of disease-specific fibrillary cores, the monomers-to-oligomers ratio, and the size of fragmented fibrils, techniques such as AFM, IEM, TEM, SEC, and CD spectroscopy were employed. Finally, in vitro validation of the proteopathic potency of tau prior to inoculation involved the use of either Tau RD fluorescence resonance energy transfer (FRET)-based biosensor cells [72] or primary neurons [117].

Synthetic PFFs, have played a pivotal role in demonstrating that tau-aggregation and propagation can be induced by altered tau alone, independent of other proteins, underscoring the role of pathologically altered tau as the causative agent. These fibrils were meticulously characterized and were composed of recombinantly produced full-length (T40) [58, 67, 75, 76, 164] or truncated (K18, K19, dGAE) tau [4, 75, 123, 144, 155, 162, 173], either as wild type [76, 162, 164, 173] or with mutations (P301S [4, 58, 75, 123, 144, 155], P301L [75, 76, 103] or C291S[164]). They were assembled into fibrillary aggregates using polyanionic inducers such heparin [4, 58, 67, 75, 76, 103, 123, 133, 144, 155, 162, 164, 173]. Aggregation confirmation was obtained through Thioflavin T assay [4, 67, 75, 76, 103, 155], or TEM [4, 67, 76, 144, 155, 164, 173]. These PFFs have been validated in both cellular and in diverse in vivo models, inducing robust tau pathology and propagation of tau pathology, thereby offering a comprehensive window for analysis [123, 142, 143, 159]. However, recent cryo-EM studies indicates that the core of the in vitro induced fibrillary structure differs from that of human fibrils [175], potentially limiting their translational value to a certain extent. This disparity might be attributed to the specific fragment used or to the potential lack of some posttranslational modifications, as these have been predominantly produced in E. coli expression system. Nevertheless, Baculovirus or mammalian expression systems could be utilized to counteract the lack of crucial post-translational modification (PTMs) in recombinant tau, such as N-linked or O-glycosylation and widespread Ser/Thr phosphorylation in prokaryotic systems [63].

In summary, while it is essential to acknowledge that the diverse inocula have their unique value and significance in studying different aspects of AD and related tauopathies, among the aforementioned five types of inoculums, the disease-specific human brain-derived tau fibrillary assemblies in either free or encapsulated form exhibit a notably close resemblance to pathological characteristics and the propagation behaviour. To that end, single cycle in vitro amplification of human disease-specific bioactive tau (AD-, PSP- and CBD-lysate) is proposed as an alternative to minimize the reliance on human brain-derived tau and reduce variability induced by the utilization of inoculums from different brains. This would allow extending the scale of the experiments for future studies, and increase comparability [166]. Moving forward, a comprehensive biochemical and biophysical characterization of tau inocula utilized in experiments will enable us to identify the specific attributes such as tau isoforms, hyperphosphorylation patterns, truncations, disease-associated filament-fold changes, and aggregation states (monomers, oligomers, or larger fibrillar aggregates). These characteristics are crucial for establishing comparability and reproducibility across studies, contributing to a more comprehensive understanding of tau pathology.

In tau propagation studies, a recurrent challenge involves recapitulating bona fide pathological inclusions post-inoculation. Although caution should be taken to extrapolate findings based solely on the induction of tau phosphorylation, which does not necessarily indicate dysfunction of tau protein or its relevance for a pathological phenotype [89]. Therefore, it is imperative to develop a standardized protocol for recapitulating fibrillary-like inclusions akin to those observed in human tauopathies. In order to discern the most optimal methodology, we briefly summarize the techniques employed for validating tau inclusions, as well as the diverse types of inclusions observed in various regions with distinct tau inoculum in the studies conducted thus far.

In AD and other tauopathies, immunohistochemistry (IHC) and conventional histological staining techniques are the gold-standard methods for characterizing tau lesions. The primary antibody frequently used in IHC is AT8 (targeting p-tau pS202/pT205) [2, 20, 111]. Additional antibodies also include — RD3 (aa 209–224) [59, 88] and RD4 (aa 275–291) [59, 88] for different tau isoforms, PHF‐1(p-tau Ser396/Ser404) [20, 111] and CP13 (p-tau Ser202) [20] for phospho-tau, MC‐1(aa 312–322) for pathological conformation [20], TauC3 for detecting tau truncated at aspartic acid 421, and Tau‐66 (aa 155–244, 305–314) and Alz‐50 (aa 2–10 and 312–342) [20] for specific tau epitopes. Conventional histological staining methods utilize Gallyas silver stain [2, 3, 23] and Thioflavin S [23, 124], with Bielschowsky [2, 3] and Campbell‐Switzer [23] staining less commonly employed. These staining techniques are crucial for identifying the β-pleated structures that constitute the core of tau fibrils, thus validating the presence of mature pathological aggregates.

In the shortlisted studies, an array of anti-tau antibodies were utilized to characterize pathological inclusions. These antibodies encompassed phosphorylation-dependent variants, notably AT8 (p-tau Ser202/Thr205) [4, 7, 11, 15, 19, 29, 45, 58, 67, 69, 96, 120, 138, 140, 146], PHF 1 [7, 11], AT100 (aa p-tau Thr212 and Ser214) [4, 11, 29, 138], AT180 (p-tau Thr231)[7], T14 (tau and p-tau of A68 polypeptides) [19, 29], and 12E8 (p-tau Ser262 and/or Ser356) [29, 74]. Additionally, conformation-sensitive antibodies such as MC1 (aa 7–9, aa 312–322) [1, 7, 11, 19, 67], isoform-specific antibodies for 3 repeat tau (RD3) [7, 19] and 4 repeat tau (RD4) [7, 11, 19], along with mouse specific tau antibody T49 [19, 58], MT1 [29], and R2295 [15, 58]; and a human-specific pan-tau antibody, HT7 (aa 159 and 163) [15, 67, 96] were employed for comprehensive characterization. The confirmation of fibrillary pathology was conclusively established through Gallyas silver staining [11, 29, 138, 140], thioflavin S staining [19, 67, 69, 74, 120, 146], or a combination of both [1, 96].

In AD, neurofibrillary pathology manifests as pre-tangles, neurofibrillary tangles (NFTs), neuritic plaques (NPs), neuropil threads (NTs), and ghost tangles [2, 21]. To investigate the pathobiology in vivo, it is essential that propagation models mimic the aforementioned disease-associated neurofibrillary inclusions. The induction of tau pathology using AD-tau inoculum successfully recapitulated NFT-like tau inclusions and NTs in mouse models such as ALZ17 [29] and hTau [71, 74]. Similarly, our group previously reported the presence of robust NFT-like tau inclusions and NTs in transgenic rat models, SHR72 and WKY72 [107, 139, 140]. Some studies also noted the presence of NPs [120], globular deposits [7] in WT mice, oligodendroglial tau inclusions in both WT [47] and ALZ17 mice [29], as well as argyrophilic grains in WT [48], THY-T22 [11] and ALZ17 mice [29]. However, several other studies solely noted the presence of aberrantly hyperphosphorylated tau, which does not meet the criteria of neurofibrillary tau pathology.

In CBD, the pathological features encompass corticobasal bodies, astrocytic plaques, neuronal threads, oligodendrocytic coiled bodies, twisted tubules in oligodendroglial cells, and small NFTs [88, 135]. Upon inoculation of CBD-tau, hTau mice developed NFT-like pathology, astrocytic and oligodendroglial tau inclusions [71, 171]. ALZ17, on the other hand, developed NFT-like pathology and astrocytic plaques, but did not recapitulate oligodendroglial tau inclusions [29]. In both WT and PS19 mice, the pathology manifested as astrocytic plaques and oligodendroglial inclusions [29, 71, 116], while with neuronal tau knocked out only oligodendroglial tau inclusions were observed [115].

In PSP, the pathological features encompass the presence of NFTs and threads in subcortical nuclei, tufted shape astrocytes, oligodendroglial coiled bodies, and diffused cytoplasmic structures in neurons [88, 90]. Upon inoculation of PSP-tau in 6hTau, ALZ17, and WT mice, the observed pathology included NFTs, and astrocytic and oligodendroglial tau inclusions [29, 71, 116]. Additionally, mice with knocked out tau expression in neurons developed only oligodendroglial tau inclusions [115].

In PiD, the pathological features encompass neuronal Pick bodies, Pick cells, ramified astrocytes, and globular inclusions in oligodendrocytes [57, 87, 154]. Upon inoculation of PiD-tau, ALZ17 mice [29], hTau and T44mTauKO [71] developed NFTs and oligodendroglial argyrophilic inclusions. WT mice developed a lower pathological load compared to T44mTauKO and hTau mice [71]. The heightened load in T44mTKO can be attributed to two factors: (1) there is no transgenic expression of human 3R tau in ALZ17, and cross inoculation experiments have shown that 4R tau could not efficiently recruit 3R tau [67]; (2) the transgene expression level in T44mTauKO exceeds that of AL17 by several folds [71].

A significant number of studies are confounded by the use of single antibody, or no confirmatory staining. Assertions of inducing tau pathology are widespread, yet caution should be taken to conflate tau phosphorylation, often in the form of a diffuse signal in the cytoplasm, with neurofibrillary pathology. The nomenclature in this context should be clarified, designating diffuse phosphorylation as such, while the term “tau pathology” should be reserved for actual tau aggregates containing beta-sheets. It is noteworthy that the standard confirmatory staining methods, such as Gallyas and Thioflavin S, can serve as robust evidence to validate tau lesions. However, validation can be further enhanced by incorporating techniques such as transmission electron microscopy (TEM) and, ideally, cryo-electron microscopy (cryo-EM) to confirm the presence of filaments [156] and elucidate their structure at atomistic resolution [137].

Taking into consideration the existing evidence, it can be concluded that recapitulating tau pathology is feasible in inoculation-based tau propagation models using either transgenic or non-transgenic models. It is noteworthy that while transgenic models in studies recapitulating CBD, GGT, AGD, and PSP predominantly utilize neuron-specific promoters like Thy1 [5, 126, 149], Thy1.2 [122], murine prion promoter (Prnp) [77, 170], CaMKIIα [85], or human tau promoter [6], they have consistently reproduced tau inclusions in oligodendrocytes and astrocytes in tau propagation models, albeit in small quantities. This suggests that tau overexpression is crucial for robust pathology recapitulation. Employing a model exclusively expressing tau in glial cells could offer a more effective approach to mimic tauopathies with predominant glial tau pathology. The judicious selection of models and histopathological validation of tau inclusions in regions where propagation occurs holds utmost importance. While every model possesses unique strengths, in the context of recapitulating authentic tau pathology, optimal choices include ALZ17, htau, SHR72, and WKY72 for AD; hTau and ALZ17 for PSP; and ALZ17 and PS19 for CBD. It's noteworthy that, none of these models can be used to successfully recapitulate full blown Pick-like pathology.

In light of the data from human research showing that tau propagation follows connectivity patterns in the brain [53, 55], we sought to investigate to what extent the relationships between tau burden and functional connectivity observed in distinct human tauopathies are faithfully recapitulated in the inoculation tau propagation models.

The hippocampus emerged as the preferred site of tau inoculation across the majority of studies [1, 4, 10, 11, 15, 19, 29, 30, 32, 45, 48, 49, 58, 67, 69, 74, 75, 80, 96, 109, 116, 123, 144, 146, 157, 161, 162, 173]. This preference is rooted in the understanding that the neurofibrillary pathology in the AD brain initiates within the hippocampus, subsequently propagating sequentially to the limbic system and ultimately reaching the isocortex [21]. In a few studies, tau inoculation into the hippocampus was followed by inoculation into the overlying cortex [1, 15, 19, 29, 30, 32, 65, 67, 71, 80, 109, 115, 116, 120, 155]. While some studies have focused on isolated cortical injections [100, 164], others explored the propagation from other brain structures such as the locus coeruleus [76], caudate putamen [49], thalamus [47, 116], striatum [75, 171], substantia nigra [144], and corpus callosum [45, 49].

Upon injection in the hippocampus (CA1/CA2/CA3) or dentate gyrus (DG), fibrillary tau pathology propagated throughout the anatomically and functionally connected regions. The hippocampus projects to the mammillary bodies through afferent fibers in the fornix. The mammillary bodies innervate the anterior thalamic nucleus and tegmental nuclei, connecting other structures to the frontal cortex and brain stem, respectively (76). The CA1 and entorhinal cortex (EC) regions heavily innervate the subiculum, which is one of the major sources of efferent projections from the hippocampal formation. In line with this far-reaching network of hippocampal connections, the CA regions [1, 4, 7, 11, 15, 19, 30, 32, 45, 48, 49, 58, 67, 69, 71, 75, 80, 96, 109, 115, 116, 120, 123, 138, 140, 144, 146, 155, 161, 162, 173], DG [1, 7, 11, 15, 32, 49, 58, 67, 69, 71, 75, 80, 109, 115, 120, 123, 138, 142, 146, 173], the fimbria [4, 7, 11, 19, 29, 30, 32, 45, 48, 49, 67‐69, 71, 115, 120], the entorhinal cortex (EC) [1, 7, 15, 19, 58, 67, 71, 75, 116, 120, 144, 173], and subiculum [1, 19, 32, 58, 71, 74, 80] were all reported positive for tau inclusions. Some studies also found associated proximal regions such as fornix [29], and mammillary area [1, 15, 19, 32, 67, 80, 116, 161] being affected.

Upon extended post-inoculation observation time (≥ 9 months), regions connected to the hippocampus across the whole brain were found affected, as evident from the presence of tau lesions in cortical regions [15, 71, 74, 96, 115, 116, 144], along with others such as amygdala [29, 30, 144], thalamic nuclei [19, 29, 30, 71, 96, 144], external capsule [69], internal capsule [29], septal nuclei [32, 74], hypothalamus [30, 96], optic tract [29], olfactory bulb [32, 116], and nuclei in the brain stem [19, 30, 67, 69, 116, 161].

In the studies with unilateral injections, the contralateral side was consistently positive for tau inclusions due to the apparent involvement of the corpus callosum (CC) [1, 4, 7, 11, 19, 38, 45, 48, 49, 58, 67, 71, 80, 96, 109, 115, 116, 120, 138, 140, 144, 155, 161, 162, 171, 173]. Whereas, when tau was inoculated in the cortex, pathology was observed at the site of injection, and the contralateral cortex [100]. Similarly, to show different regional vulnerability, upon tau inoculation in the thalamus or locus coeruleus, propagation was observed throughout the functionally connected regions [47, 76, 116]. On the contrary, no propagation was reported from the caudate putamen following inoculation there [49].

Notably, nuclei that are not directly connected to the hippocampal formation did not show evidence of tau propagation following inoculation into the hippocampus – whether this was due to time, or due to the inability of the inocula to induce pathology in a sequence of linked brain regions is unknown. The mode of propagation of pathological tau isolated from various tauopathies has demonstrated contradictory results. While some reported tau propagate in a similar manner regardless of the disease-specific inoculum [15, 29, 45, 48, 71], others showed different patterns of propagation [19, 161]. In a nutshell, both anterograde and retrograde propagation of tau in both transgenic and WT models are related more to the strength of the connectivity of neuronal networks between the inoculation region and another brain region, rather than to the proximity of the brain region to the injection site [127].

The hippocampal formation and associated areas assist with spatial and episodic memory consolidation and storage. Accordingly, some tau inoculation models described above exhibit behavioral deficits when compared to those in sham-treated animals [11, 15, 69], suggesting that these models are suitable for inclusion in behavioral assessments as an efficacy measure in preclinical efficacy studies (Tables 4 and 5).

In recent decades, a large body of clinical evidence on the modulation of tau pathology has surfaced [33, 40, 132]. In this section, we will discuss those factors that can influence tau propagation in inoculation-based models.

The extracellular accumulation of amyloid-beta (Aβ) represents one of the two key pathological features of AD. While, in the early stages of the disease, the distributions of these two distinct lesions do not overlap in terms of neuroanatomy [21, 150], in the later stages, tau pathology has been observed to propagate from the medial temporal lobe to the neocortex specifically in individuals positive for Aβ [152]. The impact of Aβ on tau propagation was explored in two independent studies employing the APP-KI and 5xFAD mouse models. The presence of Aβ plaques was found to facilitate the rapid amplification of inoculated tau-seeds into large tau aggregates, and promote their distal propagation [70, 156]. Tau inoculation in the 5xFAD/PS19 model that mimics ATN pathology, further confirmed these findings [103]; suggesting that Aβ can partially participate in tau propagation.

In addition to the neuro-centric perspective on tauopathies, the investigation of resilience mechanisms has been broadened to encompass the role of protective blood–brain barriers, the vasculature, and, most notably, glia. Microglia, which have been extensively studied in both AD and experimental models of tauopathies [17, 78, 101, 179], play a complex and sometimes contradictory roles in tau pathology. On one hand, microglia can facilitate the removal of tau by internalizing and processing extracellular tau and synapse housing tau aggregates [104]. On the other, under specific conditions, microglia have also observed to propagate pathological tau [73, 145]. To explore this conundrum, various approaches, such as depleting microglia or modifying their function by suppressing the transcription factor nuclear factor kappa-light-chain-enhancer of activated B cells (NF-κB), impeding autophagy, or deactivating inflammasomes have been employed [79, 103, 120, 142, 143, 158, 159].

In tau propagation models, a significant reduction in tau inclusions is observed following the depletion of microglia and subsequent inoculation with either PSP brain extract or synthetic K18/P301L tau fibrils in PS19 mice or 5xFAD/PS19 [103, 159]. NF-κB, a transcription factor implicated in neuroinflammation, regulates a diverse array of genes [56, 91, 118, 121]. Its activation entails a sequence of events: signal detection, activation of the IκB kinase complex, phosphorylation of IκB mediated by IKKβ, degradation of IκB, and the eventual nuclear translocation of NF-κB, enabling gene transcription [174]. In the context of AD, anomalies in NF-κB expression or function have been reported [27, 84, 141, 148]. Notably, by attenuating IKKβ expression and consequently NF-κB levels, a marked decrease in tau inclusions at the ipsilateral side of inoculation was observed, mirroring the effect of microglial depletion [159]. Complementary experiments revealed that an upsurge in IKKβ expression exacerbated pathology, thereby underscoring the pivotal role of microglial NF-κB activation in the spread of tau [159]. This manipulation of microglia offered insightful perspectives into their paradoxical role in tauopathies.

The autophagy-lysosomal pathway plays a vital role in the homeostasis of tau protein [81, 177]. Autophagy related 7 protein (ATG7), essentially serves as a critical facilitator of autophagosome maturation [31]. Deficiencies or alterations in ATG7 expression or function can result in metabolic disruptions and obstruct the uptake and clearance of extracellular tau, thereby potentially exacerbating the pathological accumulation of tau protein [105]. In ATG7 knockout mice, ATG7 deficiency promoted pro-inflammatory response and inflammasome activation. Inoculation of rTg4520 mice brain-derived tau demonstrated that the microglial ATG7 deficiency enhanced intraneuronal tau propagation [168].

The inflammasome is a multiprotein complex that plays a crucial role in the innate immune system [66, 93, 94]. NLRP3 inflammasome, named after its central component, NOD-, LRR- and pyrin domain-containing protein 3 (NLRP3), is one of the best-studied inflammasome and is known to be activated by a wide variety of stimuli. Once activated, NLRP3 interacts with the adaptor protein, apoptosis-associated speck-like protein containing a CARD (ASC), which in turn recruits pro-caspase-1, resulting in the formation of the NLRP3 inflammasome complex [92, 93]. Aggregated tau can activate the NLRP3–ASC inflammasome, contributing to the disease progression [142]. To study the effect of NLRP3-ASC inflammasome on tau propagation, knock-out of either NLRP3 or ASC based PS19 mice were utilized [142]. Either ASC deficiency or its chronic inhibition (using MCC950) reduced tau propagation post inoculation of pre-aggregated K18 tau fibrils [142]. Similarly, NLRP3 knock-out mice demonstrated reduction in tau propagation post inoculation of the same fibrillary aggregates [143].

Hypoxia is another significant factor that contributes to tau propagation [172]. Chronic intermittent hypoxia (CIH), which is predominantly associated with conditions such as obstructive sleep apnoea (OSA) and apnoea of prematurity best reflects the development of cognitive decline and AD in elderly population [98, 169]. Individuals with OSA have been observed to exhibit a more rapid longitudinal increase in the levels of CSF total-tau and phospho-tau, associated with AD [24]. In tau propagation models, following AD-tau inoculation in P301S mice, CIH exposure significantly exacerbated tau pathology and propagation in the mice. Furthermore, CIH treatment also enhanced burden of phospho-tau and activated microglia in both WT and P301S tau mice [86].

Epidemiological research has shown that cognitive stimulation and physical activities can play significant roles in slowing down the progression of AD [18, 41, 130, 136]. In addition, enriched environment (EE) have been shown to bolster neuronal activity and therefore, improve cognitive abilities such as functional outcomes, learning capacity and spatial memory [14, 52, 97]. In SHR72 rats expressing human truncated 4R tau, following AD-tau inoculation, EE was found to reduce the tangle pathology and improve navigation ability [107].

To summarize, it appears that restraining or impairing certain pathways such as NF-κB or NLRP3-ASC inflammasome can contribute to a reduction in tau propagation. Conversely, alterations in autophagy (through ATG7 deficiency) or conditions of hypoxia seem to enhance tau propagation. While the mechanism by which an EE reduces tau pathology is not yet clear, it can be hypothesized that an EE may shift microglia into the alternative phenotypes that play a role in neuroprotection through various molecular mechanisms that affect synaptogenesis, neurogenesis, and neuronal activity, leading to enhanced clearance of tau aggregates. Additionally, higher levels of tau may be secreted into the interstitial fluid due to increased neuronal activity, where it can be taken up and processed by activated microglia. These results suggest that microglia may represent a crucial modifier of the progression of tau propagation.