Altered ribosomal function and protein synthesis caused by tau

K3 mice [29] and control wild-type (WT) littermates of mixed gender were used. Mice were maintained on a 12 h light/dark cycle and provided access to food and water. All experiments were approved by and carried out in accordance with the guidelines of the Animal Ethics Committee of the University of Queensland [AEC QBI/554/17].

For primary cultures, cortices were dissected from individual K3 embryos at embryonic day 17, dissociated in a mixture of dissection media with papain, and then titrated in Neurobasal medium (Gibco, 21,103,049) supplemented with 5% fetal bovine serum (FBS), 2% B27 (Gibco, 17,504,044), 2 mM Glutamax (Gibco, 35,050,079) and 50 U/ml penicillin/streptomycin. The same number of cortical neurons was then plated into poly-D-lysine (PDL)-coated wells at 500,000 cells/well in a 12-well plate. After 72 h, the medium was replaced with Neurobasal medium with 2% B27 without FBS, and half of the medium was changed twice a week until the cells were collected. Cultures were maintained throughout at 37 °C in a humidified 5% CO₂ incubator. At days 17 in vitro (DIV17), neurons were treated with 4 mM azidohomoalanine (AHA) (Click chemistry tools, 1066) for 16 h for fluorescent non‐canonical amino acid tagging western blot (FUNCAT-WB) analysis.

HEK293 cells were cultured at 37 °C in a 5% CO₂ saturated humidity incubator in Dulbecco's modified Eagle's medium (DMEM) (Thermo Fisher, 11,965‐092) supplemented with 10% FBS and 50 U/ml penicillin/streptomycin. The same number of cells was plated for each experiment for 24 h, followed by transfection with equal amounts of plasmid using lipofectamine LTX (Thermo Fisher, 15,338,100), as per the manufacturer's instructions. All plasmids used either the pEGFP-N1 (Addgene, 6085–1) or the pCMV (Addgene, 11,153) plasmid backbone. Tau was inserted into these vectors via PCR-digestion-ligation using either XhoI (NEB, R0146) and BamHI (NEB, R3136) for pEGFP-N1, or EcoRV (NEB, R3195) and NotI (NEB, R3189) for pCMV. To examine protein synthesis, all cells were treated with 4 mM AHA for a period of 16 h before being collected. For analysis after 7 days of expression, the cells were grown in neomycin as selection marker. Cells which were analyzed via polysome profiling were not treated with AHA.

Cells were extracted in equal volumes of 1X radioimmunoprecipitation assay (RIPA) buffer (Cell Signaling, 9806), with protein concentrations determined using the bicinchoninic acid (BCA) assay (Thermo Fisher, 23,225). Newly synthesized proteins were then detected by incubating 15 µg of protein from each sample with IRDye800-DIBO (LI‐COR, 929-50,000, 1:200,000) for one hour at room temperature. Samples were then denatured via boiling 1 × Laemmli buffer, separated via SDS–PAGE and transferred to a PVDF membrane using a Turbo Transfer System (Bio‐Rad). For total protein visualization, REVERT total protein stain (LI‐COR, 926‐11,010) was used. For analysis of samples from 5 month old K3 and WT mice, proteins were extracted from one hemisphere using RIPA buffer as previously described [30]. For the detection of specific proteins, membranes were first blocked with Odyssey Tris-buffered saline (TBS) blocking buffer (LI‐COR, 927‐50,000). Membranes were then separately incubated overnight at room temperature with the following primary antibodies: Tau12 (kind gift from Dr Nicholas Kanaan, Michigan State University; 1:10,000), Tau 5 (Millipore, MAB361, 1:2,000), RPL5 (Abcam, ab86863, 1:1,000), RPS14 (ProteinTech, 16,683-1-AP, 1:1,000), RPS6 (Cell signaling, 5402, 1:500), and RPL22 (NOVUS Biologicals, NBP1-06,069, 1:1,000). Primary antibodies were added into Odyssey TBS blocking buffer. Proteins were detected using either IRDye680 anti‐rabbit IgG (LI‐COR, 926–68,071, 1:15,000), IRDye680 anti‐mouse IgG (LI‐COR, 926–68,070: 15,000) or IRDye680 anti-goat IgG (LI‐COR, 926–68,024, 1:15,000), and imaged and quantified using a LI‐COR Odyssey FC scanner. Western blots were quantified using the LI‐COR Light Studio software, with the total protein stain REVERT used for normalization.

Polysome profiling was performed as previously described but with minor modifications [32]. Briefly, transfected HEK293 cells were treated with 10 mg/mL cycloheximide (CHX) (Sigma-Aldrich, 01,810) for 3 min. Taking care to avoid RNAse contamination, cells were then placed on ice and washed in PBS with 10 mg/mL CHX, after which they were lysed in fresh lysis buffer (50 mM KCl, 20 mM Tris.HCl pH 7.5, 10 mM MgCl₂, 1% Triton X100, 1 mM 1,4-dithiothreitol (DTT), 0.5% w/v sodium deoxycholate, 1X protease and phosphatase inhibitors, 10 mg/mL CHX, 1:1000 RNAse OUT inhibitor (Invitrogen, 10,777–019)). Samples were then centrifuged for 5 min at 13,000 × g at 4 °C. The supernatant was then loaded onto a 10–50% sucrose linear gradient (50 mM KCl, 20 mM Tris-HCl pH 7.5, 10 mM MgCl₂) created using the BioComp gradient station (BioComp, 153). The 40S and 60S ribosomal subunits, along with monosomes and polysomes were then separated via centrifugation at 235,000 × g at 4 °C for 2 h and detected using the BioComp TRIAX flow gradient collection system (BioComp, FC-1-26). Abundance of the various ribosomal complexes was quantified by calculating area under the curve (AUC) for these regions of the polysome profile.

In preparation of analysis via nano-LC MS/MS, samples were reduced with 5 mM DTT at 60 °C for 30 min, followed by alkylation in 10 mM iodoacetamide (IAA) for 15 min in the dark at 25 °C, with excess IAA being quenched with an equivalent amount of DTT. Samples were then acidified with 12% orthophosphoric acid, then with S-trap loading buffer (90% methanol, 100 mM tetraethylammonium bromide (TEAB), pH 7.1). Proteins were then loaded onto an S-trap micro (Protifi), before being digested with 1.4 μg Tryspin in 100 mM TEAB for 1.5 h at 37 °C. After digestion, peptides were eluted from the trap with 100 mM TEAB, 0.2% formic acid, then 50% acetonitrile 0.2% formic acid, followed by lyophilization. Peptides were reconstituted with 30 μL of 0.1% formic acid, with 6μL of sample being loaded for injection. Peptide samples were injected (6 μl) onto the peptide trap column and washed with loading buffer for 10 min. The peptide trap was then switched in line with the analytical nano-LC column. Peptides were eluted from the trap onto the nano-LC column and separated with a linear gradient of 3.5% mobile phase B to 25% mobile phase B over 70 min at a flow rate of 600 nl/min and then held at 85% B for 8 min prior to re-equilibration.

The column eluent was directed into the ionization source of the mass spectrometer operating in positive ion mode. Peptide precursors from 350 to 1850 m/z were scanned at 60 k resolution. The 20 most intense ions in the survey scan were fragmented using a normalized collision energy of 33 with a precursor isolation width of 1.3 m/z. Only precursors with charge state + 2 to + 5 were subjected to MS/MS analysis. The MS method had a minimum signal requirement value of 4.3 × 10⁴ for MS2 triggering, an AGC target value of 3 × 10⁶ and a maximum ion injection time of 45 ms. MS2 scan resolution was set at 3 × 10⁴, an AGC target value of 1 × 10⁵ and a maximum injection time of 70 ms. MS/MS scan resolution was set at 3 × 10⁴ and dynamic exclusion was set to 30 s. The mass spectrometric data files were searched using Proteome Discoverer (Thermo, Version_2.1) embedded with search engine SequestHT against Mus musculus (17,002 sequences, accessed November 2019, https://​www.​uniprot.​org/​) protein sequences downloaded from the UniProt database. Label free quantification proteomic data were normalised to total protein abundance in each sample. Technical replicates were averaged for each sample. Fold-change to WT samples were then compared via Student’s t-test with proteins having a p value ≤ 0.05 and an absolute fold-change ≥ 1.5 being classified as being differentially expressed.

Network analysis was performed as previously described [24], with minor modifications. Briefly, data from the label-free quantitative mass spectrometry were mapped to the STRING protein query database for Mus musculus using Cytoscape (v3.8.2) and the edge-weighted spring-embedded layout. A STRING confidence of interaction score cut-off of 0.7 was used. Clusters of regulated proteins were identified using Molecular Complex Detection (MCODE) [33]. The proteins in these clusters where then analyzed using the REACTOME database.

All statistical analysis was performed on samples run at least in experimental triplicate, as detailed in the Figure legends. Statistics was performed in GraphPad Prism 7.0 software, using either one‐way ANOVA or Student's t‐test, with Tukey's multiple comparison test (MCT), as appropriate. All values are given as mean ± standard error of the mean (SEM). Significance was defined as *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

To explore the effect of FTD-mutant hTau on both global translation and ribosomal subunit abundance, we utilized primary cortical neurons cultured from the K3 mouse model of FTD [29, 34]. We previously reported decreased protein synthesis in these mice that was correlated with tau pathology [17]. Here, we extend these findings and show that overall de novo protein synthesis is also decreased in K3 primary neuronal cultures (Fig. 1a). This was achieved by using non-canonical amino acid (NCAA) labelling, in which de novo synthesised proteins were labelled with the methionine surrogate azidohomoalanine (AHA). Newly synthesized proteins containing AHA were then fluorescently labelled through the reaction of the azide-group of AHA with an DIBO-conjugated fluorophore before being analyzed via western blot in a technique known as fluorescent non‐canonical amino acid tagging followed by western blot analysis (FUNCAT-WB) [35]. Cortical neurons were cultured from individual pups until DIV17, before being treated with AHA for 16 h before FUNCAT-WB analysis, with the human tau-specific Tau12 antibody being used to distinguish K3 and WT pups. The FUNCAT-WB signal was significantly reduced in K3 samples compared to WT controls (Fig. 1a), showing that protein synthesis was decreased in K3 primary neurons.

Next, we used label-free quantitative tandem mass spectrometry to determine if ribosomal protein abundance was altered in K3 primary neurons. This allowed us to quantify the levels of 80 ribosomal proteins, including 72 of the ≈80 eukaryotic cytosolic ribosomal proteins and 8 mitochondrial RPs (Fig. 1b, Additional file 4). Of these proteins, 11 were altered in abundance (|FC|≥ 1.5,  p ≤ 0.05), with 10 showing decreased abundance (RPS2, RPS5, RPS14, RPS28, RPL5, RPL18, RPL23a, RPL35, RPL36 and MRPL12) in K3 primary neurons compared to WT littermates (Fig. 1b). RPS6 was the only ribosomal protein that was increased in K3 compared to WT samples (Fig. 1b). Interestingly, these alterations in RP abundance in K3 mice appeared to be evenly distributed across both the 40S and 60S ribosomal subunit (Fig. 1c). Additionally, we observed that the abundance of many eukaryotic initiation factors involved in translation (eIF3E, eIF4G2, eIF4B, eIF3G and eIF5) was also decreased in K3 primary neurons (Additional file 1: Fig S1, Additional file 5).

To determine if these changes in RP abundance were also found in vivo and, thus, associated with a more advanced pathology, we quantified the abundance of a subset of representative RPs in brain homogenates from K3 and WT mice at 5 months of age, when tau pathology is robust [36] (Fig. 1d). By western blotting, we found that in K3 brain, the abundance of RPS14 and RPL5 was also decreased (Fig. 1d), recapitulating our findings from primary neuronal experiments. Interestingly, unlike in primary neurons, the abundance of RPS6 was decreased in K3 mice (Fig. 1d). Furthermore, the abundance of RPL22 was decreased in K3 mice (Fig. 1d), unlike in K3 primary neurons, where its abundance was unaltered (Fig. 1b). This suggests that the extended exposure to pathogenic tau in in vivo models of tauopathy may lead to a more robust decrease in select RP levels.

Building upon our findings in K3 primary neurons, we next sought to determine if expression of FTD-mutant hTau, initially over a period of 24 h, was sufficient to decrease protein synthesis and alter RP abundance. We therefore utilised a HEK293 cell model in which we induced the expression of either non-mutant hTau or mutant forms of hTau associated with inherited tauopathy (K369I-hTau and P301L-hTau) [5] for 24 h, with newly synthesised proteins again being labelled with AHA (Fig. 2a). For non-mutant hTau and P301L-hTau expressing cells, the full length (2N4R) isoform of hTau was used, whereas in cells transfected with K369I-hTau, the 1N4R isoform was used as this matches what is expressed in K3 mice [29]. All forms of hTau were expressed by fusion to EGFP, with EGFP-only expressing cells serving as an expression control.

FUNCAT-WB analysis revealed that 24 h of FTD-mutant hTau expression resulted in significantly decreased overall de novo protein synthesis, with both P301L-hTau and K369I-hTau expressing cells showing a decreased FUNCAT-WB signal compared to both hTau and EGFP expressing cells (Fig. 2a). Western blots were performed to examine the abundance of a subset of RPs in these transfected cells. Interestingly, despite the large decrease in global protein synthesis, only the abundance of RPS14 was decreased in the FTD-mutant hTau expressing cells, with the abundance of both RPL5 and RPS6 being unchanged at this time-point (Fig. 2a). The abundance of RPL22 was also not altered in hTau or FTD-mutant hTau expressing cells, a finding similar to what was observed in K3 primary neurons (Fig. 2a). Semi-quantitative real-time polymerase chain reaction (qRT-PCR) analysis revealed that the abundance of mRNAs encoding RPS14, RPL5 and RPL22 was unchanged after 24 h of hTau or mutant tau expression (Additional file 2: Fig. S2A).

Given that we observed a large decrease in protein synthesis in HEK293 cells expressing K369I-hTau or P301L-hTau, we next sought to examine if FTD-mutant hTau expression altered ribosomal complex formation or ribosomal biogenesis. For this we used polysome profiling, a technique in which polysomes, monosomes, and the 60S and 40S ribosomal subunits are separated on a linear sucrose gradient and their abundance is quantified by measuring their absorbance at 260 nm, which detects the rRNA bound within these ribosomal complexes. We found that 24 h post-transfection, the abundance of both polysomes and monosomes were decreased in K369I-hTau and P301L-hTau expressing cells compared to the EGFP control (Fig. 2b), which is consistent with our previous observation that protein synthesis is decreased by FTD-mutant hTau expression. Interestingly, the abundance of the 60S ribosomal subunit was also decreased in these cells, suggesting that FTD-mutant hTau expression may impair ribosomal biogenesis (Fig. 2b), which precedes ribosome-mediated translation. We also demonstrated that K369I-hTau expression decreased polysome, monosome, and 60S subunit abundance regardless of whether the 1N4R or 2N4R isoform was used (Additional file 3: Fig S3).

Given the relatively long half-lives of ribosomal proteins [37] and the limited changes observed in select RP abundance after 24 h of FTD-mutant hTau expression, we next sought to examine whether RP abundance was altered after 7 days of FTD-mutant hTau expression in HEK293 cells. We found that changes in RP abundance were more pronounced at this time point, with RPL5 and RPS14 levels being decreased by FTD-mutant hTau and RPS6 abundance being increased in P301L-hTau expressing cells (Fig. 3a). Interestingly, unlike at the 24 h time-point, after 7 days of expression non-mutant hTau also decreased levels of RPL5, RPS14, as well as decreasing overall de novo protein synthesis as measured by FUNCAT-WB (Fig. 3a). These decreases, however, were less pronounced than for FTD-mutant hTau expressing cells, suggesting that the FTD-associated mutations analysed in this study have an accentuated effect on the interaction between hTau and ribosomal function (Fig. 3a). No significant change was observed of RP transcript levels in FTD-mutant hTau expressing cells after 7 days (Additional file 2: Fig S2B).

With these findings in mind, we next sought to determine if non-mutant hTau expression also altered ribosomal complex formation after 7 days. Using polysome profiling, we demonstrated that the abundance of polysomes, monosomes, and the 60S ribosomal subunit was decreased in all hTau expressing groups compared to EGFP transfected controls (Fig. 3b). Interestingly, we found that FTD-mutant hTau had a greater effect on monosome abundance compared to non-mutant hTau (Fig. 3b). Together, these data suggest that, as with FTD-mutant hTau, elevated levels of hTau also alter the biogenesis of the 60S ribosomal subunit.

Given our observation that hTau and FTD-mutant hTau can decrease protein synthesis and the abundance of particular ribosomal complexes, we next sought to determine which regions of hTau were involved in this. Thus, we expressed various truncated domains of hTau in HEK293 cells, including the N-terminal projection domain of hTau (Proj-dom hTau, aa1- 224, which contains tau’s phosphatase activating domain, the amino-terminal inserts and proline-rich region); the microtubule-binding domain (MTBR hTau, aa225-380); and the carboxy-terminal region of tau (C-term hTau, aa381-441) (Fig. 4a). Cells transfected with these constructs were analysed using FUNCAT-WB and polysome profiling.

Interestingly, whereas full-length non-mutant hTau required longer periods of expression to alter protein synthesis, Proj-dom hTau was able to considerably decrease protein synthesis after only 24 h of expression (Fig. 4b). However, unlike FTD-mutant hTau, Proj-dom hTau expression did not alter RP abundance at this time point (Fig. 4b). Using polysome profiling, we also observed that Proj-dom hTau expression resulted in a decreased abundance of polysomes, monosomes and the 60S ribosomal subunit when compared to an Emerald-expressing control (Fig. 4c). Expression of MTBR hTau and C-term hTau, on the other hand, had no observable effect on protein synthesis, RP abundance or ribosomal complex formation (Fig. 4b, c).