Introduction
Hyperphosphorylation and aggregation of tau is a key feature of a number of dementia-causing diseases. These include both primary tauopathies, such as progressive supranuclear palsy (PSP), corticobasal degeneration (CBD) and frontotemporal dementias (FTD) with tau pathology, and secondary tauopathies, such as Alzheimer’s disease (AD), where accumulation of amyloid-beta (Aβ) is thought to initiate the disease cascade [
30,
63]. Whilst tau plays a number of important physiological roles in the brain, including regulating microtubule function, myelination, neuronal excitability and DNA protection [
34], there is strong evidence that hyperphosphorylated, oligomeric tau disrupts synaptic function and may be an important driver of synapse loss and neurodegeneration in dementia [
4,
58]. Tau can be phosphorylated at up to 85 different sites (45 serine, 35 threonine and 5 tyrosine residues), with the levels of phosphorylation regulated by an ever expanding list of kinases and phosphatases [
25,
35,
66]. There is increasing focus on whether specific forms of phosphorylated tau are key drivers of downstream pathology, and whether targeting upstream kinases could be an effective therapeutic tool to mitigate tau pathology in dementia [
37,
45,
66].
Of recent interest is the possibility that increased activity of the AMP-activated protein kinase (AMPK)-related kinase, NUAK1, in AD and primary tauopathies, results in the specific phosphorylation of tau at Ser356 [
37]. Ser356 is located in repeat 4 of the microtubule binding domains, so phosphorylation of this site is likely to disrupt key aspects of physiological tau function [
26]. Interestingly, studies in
Drosophila melanogaster have highlighted that phosphorylation of tau at Ser356 could be an important catalyst for further downstream phosphorylation and aggregation [
2,
46]. NUAK1-mediated phosphorylation of tau at Ser356 has also been identified as a mechanism for regulating
total tau levels, with the chaperone C-terminus of Hsc70-interacting protein (CHIP) unable to bind tau phosphorylated at Ser356, thus preventing ubiquitination of tau and its subsequent degradation by the proteasome [
16,
17,
37]. Previous work reports that NUAK1 protein expression is increased in postmortem AD and PSP middle frontal gyrus and is found to co-localise with neurofibrillary tangles [
37]. By crossing NUAK1
+/− mice to the tauopathy P301S mouse model, Lasagna-Reeves et al. found that reduction of NUAK1 lowered both p-tau Ser356 and total tau levels and rescued aspects of tau pathology [
37]. This work highlighted NUAK1 as an attractive target for therapeutic development in primary tauopathies, opening important questions about whether similar strategies could be applicable to secondary tauopathies such as AD.
Whilst there are reports that NUAK1 levels may be upregulated [
37] and tau is phosphorylated at Ser356 in end-stage AD [
3,
18,
26,
27,
64], the progression of p-tau Ser356 accumulation over the disease time course, its representation in tangles and its association with synapses (synaptic tau has been found to be important for both tau toxicity and trans-synaptic tau spread [
9,
36,
47,
68]) have not been fully characterised. Of the few studies that do examine appearance of this epitope in AD brain tissue, many use the 12E8 antibody [
21,
60], which shows considerable preference for p-tau Ser262, complicating interpretation of the unique involvement of p-tau Ser356 [
55]
. In this work, we characterise specific accumulation of p-tau Ser356 in AD brain using biochemical and histological methods including sub-diffraction limit resolution microscopy to examine synapses [
9,
33,
47,
52].
Based on the likely pathological involvement of p-tau Ser356, we further explore the effects of pharmacological inhibition of phosphorylation of tau at this residue. We characterise the impact of the commercially available NUAK inhibitor WZ4003, which has been previously shown to inhibit NUAK1 activity in vitro [
5] and reduce p-tau Ser356 in neuroblastoma cells [
37]. In this study, we look to examine the impact of WZ4003 treatment under a number of physiological and pathological conditions using both mouse and human organotypic brain slice cultures, which retain physiologically relevant neuronal architecture, supporting cell types and synaptic connections for several weeks in vitro [
12,
19,
20,
28,
56]
. We explore the impact of WZ4003 in both wildtype and APP/PS1 cultures, to assess whether NUAK1 inhibition has differential impacts on models of Aβ pathology, partially replicating changes seen in Alzheimer’s disease, where dysregulated Aβ processing is thought to drive downstream changes to tau [
63]. Our results reinforce the importance of p-tau Ser356 in AD and highlight potential biological differences in mouse and human brain in terms of how NUAK1 regulates tau turnover.
Discussion
This work combines multiple experimental tools to better characterise the timing and location of p-tau Ser356 in AD and assess the impact of pharmacological NUAK inhibition under a range of physiological and pathological conditions. Our work highlights p-tau Ser356 as a highly disease-associated form of tau in postmortem AD human brain. We show that p-tau Ser356 is not readily detectable in protein extracts from control postmortem brains, but can localise to dystrophic neurites surrounding areas of sporadic pathology in control tissue paraffin sections. We find an effect of Braak stage on the accumulation of p-tau Ser356, and potential downstream phosphorylation sites (p-tau Ser202/Thr205), in postmortem temporal lobe (BA20/21). In contrast to previous reports [
37], we do not find an increase in NUAK1 levels in our AD brain cases, but we do find a subtle shift in NUAK1 protein expression into synaptoneurosome compartments in Braak III–IV and Braak VI samples. When examining paraffin sections from AD brain, we find that almost all (93%) of ThioS-positive tangles are dual-labeled with p-tau Ser356, indicating this epitope may be phosphorylated early in the tangle formation process. This finding is in agreement with studies in
Drosophila melanogaster [
2,
46] and neuroblastoma cells [
37] suggesting that phosphorylation at this site can promote downstream phosphorylation of multiple other sites. Indeed, we find here that WZ4003 treatment also results in reduced levels of p-tau Ser202/Thr205 in MOBSCs. The consistent and early appearance of p-tau Ser356 in the AD disease course once again highlights this epitope as a potential therapeutic target. Interestingly a mutation in this site on the
MAPT gene has been linked to a very-early onset form of FTD with Parkinsonism linked to chromosome 17 (FTDP-17) [
44,
65,
67], so it seems likely that early changes to tau at Ser356 may be relevant across multiple dementia-causing diseases.
For the first time, we used array tomography, a microscopy method permitting sub-diffraction limit resolution characterisation of protein composition of individual synapses [
33]
, to assess whether p-tau Ser356 is present at the synapse in AD brains. We found that, whilst p-tau Ser356 is almost undetectable in control brain synapses, there is a small but significant proportion (~ 1 to 3%) of pre- and post-synapses that co-localise with p-tau Ser356 in AD brain with positive FRET signals indicating co-localisation of p-tau356 with either synaptophysin or PSD95 within the synaptic compartments. Interestingly, whilst a similar proportion of synapses co-localise with AT8 (p-tau Ser202/Thr205), the proportion of synapses that contain both epitopes is considerably lower (~ 0 to 1%), raising the possibility that the order of tau phosphorylation may be different in individual synapses. Alternatively, it may be that detection of both epitopes together is under-represented through technical limitations, such as reduced antibody binding when both epitopes co-localise. Building on recent reports that astrocytes engulf synapses in AD brain [
59,
62], we find that p-tau Ser356-containing pre-synapses are around five times more likely to be ingested by astrocytes than the general pre-synapse population, potentially highlighting a role in targeting synapses for phagocytosis. Recent work has highlighted potentially important roles of synaptic tau for both toxicity and involvement of trans-synaptic tau spread [
9,
36,
47,
68]. Future work exploring the impact of synaptic p-tau Ser356, in contrast to tau phosphorylated at alternative sites, could further elucidate its role in AD pathology.
Given the potential importance of NUAK1 in the phosphorylation of tau at Ser356, and our findings that p-tau Ser356 is highly associated with disease progression in AD, we sought to characterise the impacts of pharmacological NUAK inhibition under a range of physiological and disease-model conditions. In this study, we used WZ4003, which has previously been shown to be a potent inhibitor of NUAK1, and to a lesser extent NUAK2, and with no inhibitory activity on a panel of 139 other related kinases [
5]. Previous studies using WZ4003 have used simple in vitro systems such as primary culture [
8] or cell lines [
5,
37] which may oversimplify the impacts of NUAK inhibition on brain tissue containing multiple cell types, and functionally relevant neuronal architecture [
19]. In addition, prior work looking at the effect of NUAK1 knockdown in animal models (
Drosophila melanogaster and mouse) focussed on models with tau pathology exclusively, leaving a gap in our understanding of how Aβ pathway dysregulation, or elevated Aβ production, may impact response to NUAK inhibition [
37]. Here, we used MOBSCs from the APP/PS1 mouse model of Aβ pathology, alongside wildtype littermates, to model first, whether we see changes to p-tau Ser356 expression in this AD model, and then the implications of targeting NUAK activity, using WZ4003, under physiological (wildtype) or Aβ pathology (APP/PS1) conditions. Whilst previous work has found that MOBSCs can show accelerated pathological changes compared to in vivo [
12,
13,
20,
28]
, and APP/PS1 mice are found to show increased tau phosphorylation with age [
43], we did not find any differences between the genotypes in this study up to 4 weeks in culture. Nevertheless, the APP/PS1 cultures served an important purpose to establish whether conditions of elevated Aβ production alter biological responses to WZ4003 treatment. In our work, the response of APP/PS1 MOBSCs was not statistically different to their WT littermates.
A unique aspect of this study is the use of both MOBSCs and HBSCs to examine the impact of WZ4003 treatment. In MOBSCs, we found that, whilst WZ4003 treatment lowered both total tau and p-tau Ser356 protein, this reduction coincided with a loss of PSD95, and was proportional to a loss of the neuronal tubulin marker Tuj1. Interestingly, the effects of WZ4003 treatment on tau, synaptic and neuronal protein levels were strongest in the 0–2 week culture period, indicating the early stage cultures were especially sensitive to negative impacts of NUAK inhibition, loss of tau protein or any potential off-target effects of WZ4003. This possibly reflects differential involvement of NUAK1/2, altered processing of tau, or differential vulnerability to off-target (non-tau) effects of WZ4003/NUAK inhibition in different culture phases. By contrast, WZ4003 treatment in HBSCs resulted in a specific reduction in p-tau Ser356, whilst preserving total tau, that occurred alongside
increased levels of neuronal and synaptic protein. These findings could demonstrate important species differences in how NUAK1 regulates tau and highlight the benefits of using human experimental systems to assess impacts of pharmacological agents [
6,
53,
61]. However, another key difference between MOBSCs and HBSCs is the age of the brain tissue used to generate slices. MOBSCs are taken from postnatal (P6–P9) animals, whilst the age of human brain in this study ranged from 37 to 76 years old. Therefore, another interpretation of the different response to WZ4003 in MOBSCs versus HBSCs is potential differences in the role of NUAK1/2 during development versus ageing [
7]. Indeed, a number of studies have identified key roles of NUAK1 in regulating a number of developmental processes including axon elongation [
8,
10,
11], axon branching [
10], and cortical development [
11]. It may be that the postnatal slice cultures are negatively affected by either NUAK1/2 inhibition or the loss of total tau during this period, whilst adult human tissue is less dependent on NUAK activity (or benefits from the relative preservation of total tau levels, which may be important for physiological function [
34]). Indeed, our results here indicate NUAK1 inhibition may
increase neuronal and synaptic protein levels in adult human brain tissue.
Human slice cultures represent a translationally powerful new tool for neuroscience research [
1,
6,
22,
38,
41,
42,
50,
51,
53,
54,
61]. Although live human tissue has historically been difficult to obtain, with close collaboration with neurosurgery units, research nurse teams and the laboratory scientists, we have established an efficient pipeline to obtain and culture human brain slices. We show here that they can be an effective tool to examine the impact of pharmacological compounds in live human brain tissue. Previous work has shown benefits of using human cerebrospinal fluid to boost longevity of HBSCs, particularly in regards to electrophysiological activity [
53,
54]. In our work, using an enriched stem-cell-like medium [
41], we find HBSCs retain MAP2 positive neuronal cell bodies and intact neurites, and we are readily able to detect tau, neuronal and synaptic proteins via Western blot for at least 2 weeks in vitro. By comparing control and treatment conditions in slices taken from the same individual, we are able to detect biologically relevant responses, even on a background of unavoidable variations in patient age, sex, brain region taken and variations in patient lifestyle and genetic factors. It is worthy of comment that all of our HBSCs, despite none being clinically diagnosed with AD, had detectable levels of p-tau Ser356 in protein extracts, in contrast to our postmortem study which found very little p-tau Ser356 in Braak 0–I control protein extracts. As we are able to detect p-tau Ser356 in acute samples taken from surgery, prior to culturing, this could indicate that p-tau is susceptible to degradation in the post-mortem interval, and thus small levels of p-tau in control postmortem brain in our samples were rendered undetectable. Alternatively, this could represent rapid increases in tau phosphorylation in response to damage during brain surgery [
32,
34], exposure to anaesthetic [
23,
49], or fundamental differences between peri-tumour and post-mortem “control” tissue. We should also be mindful that whilst the tissue itself is non-tumour, we cannot rule out alterations to NUAK1 expression or activity in these samples, which have originated from individuals with brain tumours [
39]. Such differences will be important to reflect on as the tool becomes more widely used. The use of HBSCs as a research tool is expanding and comparison between mouse tissue, primary cultures, postmortem human and live human tissue models is likely to be highly valuable when assessing the translational viability of future therapies under development.
One limitation of the present study is that it uses a single small molecule tool which has activity at both NUAK1 and NUAK2, and whilst the published kinase-selectivity data suggest it is a relatively clean inhibitor [
5], we cannot rule out activities at other kinases. Future work should explore further compounds with selectivity for NUAK1 over NUAK2 and more complete profiles. We should also consider that NUAK1 also acts on other, non-tau targets, such as the TGF-β pathway, so cannot rule out that some of the impacts we see on synaptic/neuronal proteins may be due to alterations in NUAK function beyond phosphorylating tau [
7]. Future work to develop specific tools to target p-tau Ser356 directly, such as humanised antibodies, could be highly informative.
In summary, the work in this study further highlights p-tau Ser356 as a potential target of interest in developing AD therapeutics, with increased p-tau Ser356 strongly correlating with Braak stage, being a near-ubiquitous presence in NFTs and co-localising with synapses in AD brain. Whilst NUAK inhibition via WZ4003 treatment of postnatal MOBSCs results in tau lowering that is proportional to loss of synaptic and neuronal protein, we find WZ4003 effectively and specifically lowers p-tau Ser356 in live, adult human brain tissue, highlighting the importance of using complementary experimental systems in pre-clinical work. Future work should further explore the impact of pharmacological NUAK inhibition in vivo using a range of tau and Aβ pathology models and with NUAK1 inhibitors optimised for increased potency and drug-like properties with more fully characterised selectivity profiles.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.