Background
The microtubule-associated protein, tau, was first described as a protein that promotes and stabilises microtubule assembly [
43]. It plays a major role in several neurodegenerative diseases called tauopathies, the most common of which is Alzheimer’s disease (AD). Tau is found in both neuronal and non-neuronal cells, has numerous different isoforms and localises to multiple cellular compartments, indicating that it may play many cellular roles [
6]. However, for nearly 30 years, the majority of tau research has focused on its role in microtubule biology (stability/assembly) and the implications associated with tauopathies. In AD, tau becomes hyperphosphorylated and/or truncated and forms paired helical filaments (PHF) that become deposited in neurofibrillary tangles (NFTs) in the cell bodies of neurons. These structures, together with amyloid plaques, constitute the main hallmark of AD. The cellular modifications that accompany the generation of these insoluble, fibrous deposits are believed to play an essential role in neurodegeneration.
A nuclear form of tau has been characterised in several cell lines, primary neurons, mouse brain and human brain tissues (reviewed in [
6]. Nuclear tau species are often visualised distributed throughout the nucleus, depending on the protocol, antibody used and stage of differentiation [
10,
23]. In neurons, non-phosphorylated tau is mostly seen in the nucleus [
42], but can localise to the nucleolus during cellular stress [
39]. In neuroblastoma cells, non-phosphorylated tau appear in puncta that localise to the nucleolar organiser region [
22]. The nucleolus is the major hub for rRNA gene metabolism. Tau has been found to localise with key nucleolar proteins like nucleolin and upstream binding transcription factor (UBF), as well enhance interactions of RNA-binding proteins like T cell intracellular antigen 1 (TIA1) with ribonucleoproteins, suggesting a role for it in rRNA gene metabolism [
4,
37,
41]. Tau has been found to co-localise with the pericentromeric heterochromatin [
37], to play a role in its stability [
26], and regulate transcription [
14]. Tau mutations appear to alter chromosome stability [
35], while tau pathology induces chromatin relaxation [
11,
14].
Non-phosphorylated tau has been found to translocate to the nucleus playing a role in DNA protection during heat stress [
39]. Other reports demonstrate that stress induced by formaldehyde or Aβ42 promotes the nuclear influx of phosphorylated species of tau and this coincides with cellular and DNA damage [
24,
25,
31]. These studies suggest that nuclear species of tau may be affected differently depending on the type or severity of the cellular stress. However, it is not clear whether the nuclear phosphorylated tau accumulates in the nucleolus and whether the species of tau localised to the nucleolus behaves like nucleolar proteins, such as nucleophosmin (B3) and fibrillarin (FBL) which become redistributed during nucleolar stress [
19]. Nucleolar stress is thought to be an early event in cellular dyshomeostasis, preceding apoptosis and occurs in neurodegeneration [
2,
8,
40,
44].
To understand the role of tau on nucleolar function and impact of cellular stress on its nucleolar localisation, here we show that nuclear non-phosphorylated tau localises within the nucleolus in undifferentiated and differentiated human SHSH5Y neuroblastoma cells, where it associates with TIP5, the major subunit of the Nucleolar Remodelling Complex (NoRC) and a key player of heterochromatin stability at constitutive heterochromatin and rDNA [
34]. We reveal that tau knockdown leads to an increase in rDNA transcription and associated destabilisation of the heterochromatin indicating that it plays a role in rDNA transcription. Furthermore, glutamate induced stress causes a redistribution of nucleolar tau associated with nucleolar stress indicating that tau behaves like other nucleolar proteins. Immunogold co-labelling electron microscopy analysis of human brain tissue sections shows tau localised with TIP5 in the nucleolus, highlighting the physiological relevance of our findings.
Methods
Cell culture
Undifferentiated SHSY5Y neuroblastoma cells were maintained in DMEM/F-12 (Life Technologies, UK), supplemented with 1% (v/v) L-glutamine 1% (v/v) penicillin/streptomycin and 10% (v/v) Fetal Calf Serum (FCS). For experiments involving differentiated cells, SHSY5Y cells were incubated for five days in a medium containing 1% FCS supplemented with 10 μM trans-Retinoic acid (Abcam, ab120728), followed by two days incubation with 2 nM brain-derived neurotrophic factor (BDNF) in serum-free media (GF029, Merck Millipore). Cells were treated with 2 mM or 20 mM glutamate (dissolved in DMEM/F-12) or untreated two days post-BDNF incubation.
siRNA transfection
SHSY5Y cells were maintained for 72 h in Accell SMARTpool siRNA against Tau (Tau siRNA) or non-targeting pool (NT siRNA) (Additional file
1: Table S3) at a concentration of 1.5 μM mixed in Accell siRNA Delivery Media (B-005000-100, Dharmacon).
Western blotting
SHSY5Y cells treated or untreated with a test compound were fractionated using 1X RIPA (Abcam, ab156034), supplemented with protease (P8340, Sigma) and phosphatase (P0044, Sigma). A total of 10 μg of protein from each sample were loaded to 4-20% Mini-PROTEAN Protein Gels (4568094, BIO-RAD), for SDS-PAGE at 100 V. The proteins were transferred to PVDF membrane (IPVH00010, Merck Millipore) at 100 V, then blocked in blocking buffer (5% (
w/
v) milk dissolved in washing buffer (TBS-Tween Tablets solution) (524,753, Merck Millipore), and incubated at 4 °C overnight with the different primary antibodies (Additional file
1: Table S1) diluted in the blocking buffer. The membranes were washed in the wash buffer 5× for 10 min each and probed at RT on a shaker for 1 h in the corresponding secondary antibodies diluted in blocking buffer. The membranes were washed 5× for 10 min each and subsequently developed in the darkroom after incubation in Clarity Western ECL substrate for 1 min (1,705,060, BIO-RAD). For loading control antibodies or sequential analyses of other proteins on the same membrane using other antibodies, the membranes were stripped using Restore™ PLUS Western Blot Stripping Buffer (46,428, Thermofisher Scientific), then blocked, and probed as described above. The blots were scanned at high resolution, and then bands were quantified using Image J software.
Immunoprecipitation
SHSY5Y cells were fractionated using RIPA supplemented with protease and phosphatase inhibitors and 1.25 units of Benzonase Nuclease (E1014, Sigma), and used at least 2 h afterwards for immunoprecipitation using Dynabeads protein G according to manufacturers protocol (10007D, Life technologies). At the final step, the beads-antibody-antigen complexes were eluted in 30 μL of 50 mM Glycine (pH 2.8) and 15 μL 1× Laemmli Sample Buffer (1,610,747, BIO-RAD), supplemented with 1:10 dilution of 2-Mercaptoethanol (Sigma, M-6250), and boiled at 80 °C for 10 min. The beads were separated from the magnet and supernatant (containing the eluted protein) and used for SDS-PAGE/Western blotting.
Immunofluorescence labeling
SHSY5Y cells treated or untreated with a test compound, were re-suspended in PBS and spun onto a glass slide at 800 RPM for three min using Cytospin Centrifuge (CellSpin I, Tharmac). Cells were fixed with 4% paraformaldehyde/PBS for 15 min, PBS-washed, permeabilised using 0.5% TritonX-100/PBS for 15 min and PBS-washed. The slides were blocked in blocking buffer [4% BSA/PBS/Tween-20 (0.02%)] for 45 min, incubated with primary antibody for 45 min, PBS-washed three times, incubated in Alexa fluorophore-conjugated corresponding secondary antibody for 45 min. The slides were PBS-washed three times, incubated in 1/1000 DRAQ5 (ab108410, Abcam) diluted in PBS/Tween-20 (0.02%) for 10 min and mounted with coverslips using ProLong Gold Antifade mountant (P36930, Life technologies) or ProLong Gold Antifade mountant with DAPI (P36935, Life technologies). For the labelling of 5-Methylcytosine /(5-mC), cells on the glass slides were fixed with 2.5% PFA/PBS for 30 min at RT, PBS-washed, permeabilised for 1 h at RT with 0.5% Triton X-100/PBS. The cells were next washed in wash buffer [PBS/0.1% Triton X-100 (PBST)] and incubated with 2 N HCl for 30 min at 37 °C to depurinate the DNA, followed by 2 × 5 min wash with 0.1 M borate buffer (pH 8.5). They were then rinsed twice with PBS-T, blocked in blocking buffer (1%BSA/PBS-T) for 1 h at RT, incubated with the primary antibody diluted in the blocking buffer for 2 h at RT and washed three times in PBS-T. Then they were incubated with the corresponding secondary antibody diluted in the blocking buffer for 45 min at RT in the dark and washed three times in PBS-T, then stained with DRAQ5 and mounted.
Confocal microscopy imaging and analysis
Images were taken using a 100× oil objective of LSM510 Meta confocal microscope mounted on Axiovert200M using pinhole size of 1 Airy unit. All images were collected as Z-stacks for all channels using a step size of 1 μm to allow the analysis of the entire signal in the cells. Subsequently, images were Z-projected to sum all signals and then analysed using image J. Five randomly collected images from each experiment and an average of 150 cells per experiment were subjected to the Image J analysis. For the quantification of nuclear foci/cluster, Image J procedure presented by the light microscopy core facility of Duke University was used [
9]. For the quantification of total nuclear fluorescence intensities, the nuclei were first segmented by thresholding using DAPI/DRAQ5 channel, excluding fused nuclei or those at the edges. Subsequently, the multi-measure option on the image J ROI manager was used to measure nuclear fluorescence from all channels in only segmented nuclei. The total corrected nuclear fluorescence (TCNF) was then calculated as TCNF = Integrated Density – (Area of selected cell X Mean fluorescence of background readings [
3]. For the quantification of nucleolar nP-Tau and Fibrillarin redistribution, Z-stack images were Z-projected to maximum intensity, before cells positive for the redistribution were counted.
Immunogold labelling transmission Electron microscopy (TEM)
Brain tissue from the middle frontal gyrus of human brain (see Additional file
1: Table S2) was analysed under local ethics approval and provided by London Neurodegenerative Diseases Brain Bank with informed consent as previously described [
1]. The immunogold labelling for these sections and the SHSY5Y cells were performed by minimal, cold fixation and embedding protocols, as previously described using an established method that employs PBS+ buffer for dilution of all immunoreagents and washes [
1,
38]. Thin sections were collected onto 300-mesh high transmission hexagonal Nickel grids (Agar Scientific), blocked with normal goat serum (1:10 dilution) for 30 min at RT, single or doubled labelled using antibodies for 12 h at 4 °C. The grids were washed three times with PBS+ for 2 min each, then incubated with appropriate gold particle conjugated secondary antibodies for 1 h at RT (see antibody section and results). The grids were next washed three times for 10 min each in PBS+ and four times for 5 min each in distilled water, dried for 5–10 min and then post-stained in 0.22 μm-filtered 0.5% (
w/
v) aqueous uranyl acetate for 1 h in the dark. The grids were finally washed with distilled water five times at 2 min intervals and left to dry for at least 12 h before TEM observation.
TEM imaging and analysis
JEOL JEM-1400 Transmission Electron Microscope with a Gatan OneView® camera was used to image the grids at 120 V. For colocalisation analysis in the human brain, four nuclei per grid, of medium to large size (> 50% of X8000 magnification view), were randomly selected and imaged at X15000-X20000 magnification. Four grids were taken from each case, accounting for one repeat for the two double immunolabelling cases. In all cases, randomised selection was undertaken by identifying nuclei at low magnification (X5000), then imaging at higher magnification. All images were analysed using Image J. For colocalization analysis on brain sections, each observed 15 nm gold particle, signifying a Tau 1 antigen, was checked for colocalisation with 5 nm gold particles, signifying TIP5 antigens. Our definition of colocalisation is; when the number of one antigen (TIP5 particles) within a 5 0 nm radius of the second antigen (Tau 1) is greater than zero (
n > 0). Gold particles were included in our analysis if; Tau 1 particles measured between 11≤x≤19 nm and TIP5 particles measured between 1≤x≤9 nm. The method of colocalisation analysis was roughly based on the cross-K function; we used the number of gold particles of the first type at distances shorter than a given distance from a typical particle of the second type divided by the area of the 50 nm inclusion circle [
29].
CellROX green assay
Oxidative stress was measured in treated or untreated SHSY5Y cells using CellROX Green Reagent (C10444, Lifetechnologies UK).
Nascent RNA and protein synthesis
Nascent RNA and protein synthesis were visualised respectively using Click-iT RNA Alexa Fluor 488 Imaging Kit (C10329, Life technologies) and Click-iT HPG Alexa Fluor 488 Protein Synthesis Assay Kit (C10428, Life technologies) following the manufacturer’s instructions and images were taken using a 100× oil objective of LSM510 Meta confocal microscope mounted on Axiovert200M using pinhole size of 1 AU.
RNA Extraction and Complementary DNA (cDNA) synthesis
RNA was extracted from SHSY5Y cells treated or untreated using the protocol supplied by Lifetechnologies and subsequently used for cDNA synthesis using the High Capacity cDNA Reverse Transcription Kit (4,368,814, Life technologies, UK).
Restriction digest for DNA methylation assays
Whole DNA extract from control or Tau knockdown SHSY5Y cells were digested with 2 U/μL of HpaII (R6311, Promega) or MspI (R6401, Promega), or they were mock-digested. T0 region was amplified using specific primers and samples were run on 10% agarose gel for quantitative analysis. For further details, see Additional file
1: Table S4.
Quantitative polymerase chain reaction (qPCR)
Maxima Probe/ROX qPCR Master Mix (2X) (K0232, Life technologies), 20X TaqMan gene expression assay (Life technologies UK, Table S4) and Nuclease-free water were transferred to a white 96-well semi-skirted PCR plate (I1402–9909-BC, StarLab, UK). A standard curve was prepared using serial dilution of cDNA and qPCR was carried out on all samples using Roche LightCycler 480 II (Roche Diagnostics, Basel, Switzerland). See Additional files for detailed methods.
Statistical analysis
All data were subjected to Kolmogorov-Smirnov (K-S) normality test and then students t-test using GraphPad InStat.
Discussion
Here we reveal a close association between tau and TIP5 in the nucleolus in SHSY5Y cells and in human brain tissue. Based on this association and the widely known role of TIP5 in transcriptional silencing of rDNA, we tested whether nP-Tau plays a role in rDNA transcription. Depletion of tau resulted in increased transcription of 45S-pre-rRNA suggesting a role for nP-Tau in gene silencing and heterochromatin stability. Under conditions of oxidative stress, nucleolar nP-Tau becomes relocalised and the levels of nuclear T-Tau and P-Tau (Thr231) increase in a dose dependent manner.
Tau has been shown to localise with acrocentric chromosomes [
22] and heterochromatin in human fibroblasts, lymphoblasts and HeLa cells [
37], suggesting it may play a role in heterochromatin regulation. A recent study revealed that tau KO transgenic mice harbour pericentromeric heterochromatin instability, which can be rescued by tau overexpression in the nucleus [
26]. Here, we reveal that tau localises to the nucleolus in both SHSY5Y cells and the human brain where it is associated with TIP5. TIP5 has been shown to interact with the nucleolar and constitutive heterochromatin (pericentromeric and telomeric heterochromatin) and plays a vital role in the establishment of these chromatin domains [
13,
34]. Here we revealed that depletion of tau led to a reduction in H3K9me3 foci, H3K9me2 nuclear levels and 5-methylcytosine, indicating heterochromatin instability. These results suggest that similar to TIP5, tau may play a role in the heterochromatin complex, such that its knockdown led to heterochromatin loss, likely leading to the increase in rDNA transcription. Previously, tau KO mice also showed that its absence enhances the transcription of several genes [
32], including the pericentromeric chromatin [
26] and smarce1 gene [
12]. Moreover, tau pathology has been found to induce chromatin relaxation and enhance the transcription of many genes suggesting a role for tau in chromatin remodelling [
11,
14].
How tau is able to affect chromatin conformation remains unclear. However, we found that tau associates with TIP5 at the perinuclear border, and within the nucleus in the heterochromatin and nucleolus. Such an association may suggest that the heterochromatin and rDNA transcriptional silencing roles of tau may be mediated or facilitated by TIP5 or other chromatin remodellers. TIP5, unlike tau, has different domains that facilitate interaction with chromatin remodellers and the DNA, such as AT-hooks, a C-terminal PHD and a bromodomain [
28].
Various cellular stressors are known to induce nucleolar stress, a prominent feature of which is the disruption of the nucleolus and redistribution of nucleolar proteins such as nucleophosmin and FBL to the nucleoplasm or cytoplasm [
19,
44]. Redistributed proteins lose their functional role, resulting to cell death [
40]. Here we showed that glutamate stress induced nucleolar disruption and redistribution of FBL. However, the striking result observed here is redistribution of nucleolar non-phosphorylated tau. Several studies have located tau in the nucleolus of several cell lines [
6,
10], but its disease involvement or impact of cellular stress on its localisation has not been investigated. This study reveals that nucleolar tau also undergoes stress-induced redistribution, similar to other nucleolar proteins, demonstrating a novel involvement of nucleolar tau in nucleolar stress response. Interestingly, several regions of the AD brain have been shown to have a signature of nucleolar stress, associated with the reduction of several nucleolar proteins and nuclear tau [
16]. Given the role of tau in AD and many tauopathies, future studies will investigate the changes and contribution of nucleolar tau to the disease pathology. Interestingly, the nuclear P-Tau (Thr231) levels increased in response to stress, but did not colocalise with FBL or with nucleolar nP-Tau. P-Tau Ser396/Ser404 (PHF-1) has been previously shown to localise in the nucleus but not nucleolus, of a patient with presenile dementia with motor neuron disease [
33]. On the contrary, it was recently shown that inhibition of transcription with
Actinomycin D induces the localisation of AT-positive tau (Phospho-Tau (Ser202, Thr205) to the nucleolus in SK-N-BE neuroblastoma cells [
10]. This generally suggest that different nuclear tau species may exist and play different roles in the nucleus during a stress response.
Conclusions
In this study, we establish the presence of nP-Tau in the nucleolus in undifferentiated and differentiated SHSY5Y, HeLa cells and in human brain tissue. We have revealed a novel association for tau and TIP5 in the heterochromatin and nucleolus in SHSY5Y and brain tissue. Although future studies will address the relationship between Tau and TIP5 in heterochromatin stability and rDNA transcription, we postulate that the Tau/TIP5 association may function to stabilise the repressive epigenetic marks on the rDNA and constitutive heterochromatin. This work establishes nP-Tau is a bona fide nucleolar protein which associates with a key member of the NoRC.
Acknowledgements
The authors thank Dr. Roger Phillips for his advice on fluorescence microscopy data acquisition and analysis, Dr. Youssra Al-Hilaly for valuable help with immunogold labelling studies and Dr. Pascale Schellenberger for help with electron microscopy.