Human brain samples
Hippocampal and frontal lateral cortex brain samples from sporadic Alzheimer’s disease patients and control subjects were kindly provided by Dr. A. Rabano from Banco de Tejidos (Fundación CIEN, Instituto de Salud Carlos III, Madrid, Spain). Based on quantitative pathological features, the Alzheimer’s brain specimens were classified according to Braak stages I (n = 3), II (n = 6), III (n = 4), IV (n = 1), V (n = 10) and VI (n = 8), and non-demented control subjects (n = 10) (see Supplementary table 1, online resource for further subject information). Written informed consent premortem was obtained from all patients.
Classical nomenclature  for MAPT gene exon/intron numbers has been used consistently throughout the paper, although some of the databases (such as ENSEMBL: https://www.ensembl.org/index.html) employ a different nomenclature assigning different numbers to exons and introns (according to ENSEMBL, intron retention would take place in intron 13).
As for the new Tau isoform described in this work, the transcript including the translation of intron 12 up to the first stop codon has been named TIR-MAPT (Truncated by Intron Retention MAPT), while the protein generated from such transcript has been termed W-Tau, due to the appearance of two characteristic tryptophan residues (W) in this isoform, an amino acid that does not appear elsewhere within the Tau molecule.
In addition, Tau truncated by asparagine endopeptidase, as previously described by Zhang et al.  is referred to as ET-Tau (Endopeptidase Truncated Tau) throughout the text.
HEK293T (CRL-11268, ATCC), SK-N-MC (HTB-10, ATCC), and SH-SY5Y (CRL-2266, ATCC) cells were cultured in DMEM or MEM supplemented with 10% fetal bovine serum, 2 mM glutamine, non-essential amino acids, 10 U/ml penicillin and 10 µg/ml streptomycin, at 37 °C and 5% CO2.
TIR-MAPT and ET-MAPT cloning
TIR-MAPT transcript was obtained from an SH-SH5Y RNA extract, using specific oligos (TauNt and TauD, see Supplementary table 2, online resource) and was cloned into pBlueScript-SK + (212,205, Agilent Technologies), thanks to a TA-cloning strategy . Using specific oligos that include the appropriate restriction sites (A1 and TIR-T-BglII and ET-T-BglII, see Supplementary table 2, online resource) TIR-MAPT and ET-MAPT were subcloned into a eukaryotic expression vector pSG5 (216201, Agilent technologies). Similarly, Tau isoforms were cloned into a prokaryotic expression vector pRK172 using specific oligos (TAU-PRK172 fw, TIR-T-pRKpWPI rv and ET-T-pRKpWPI rv; see Supplementary table 2, online resource). After cloning, all vectors were sequenced using the described oligos (Supplementary table 2, online resource).
RNA-seq data analysis
RNA-seq raw data from 363 samples of 3 brain regions (frontal cortex, dorsolateral prefrontal cortex and hippocampus) of 180 human brain healthy donors were retrieved from the Genotype-Tissue Expression (GTEx) project  (Supplementary table 3, online resource). SRA files were converted to FASTQ files and reads were re-mapped to human genome GRCh38 using STAR version 2.5.2a . Gene expression quantification was performed with RSEM version 1.3.1 .
The annotation file was retrieved from GENCODE (gencode.v23.annotation.gtf) and was modified to include a new MAPT-related gene (TIR-MAPT) whose genome coordinates are chr17:45894382–46018851, which includes part of the intron 12 (chr17:46018731–46018851) as the 3′ end of the gene in the region mapping the oligonucleotide sequence TauD (Supplementary Table 2, online resource and Fig. 1). Gene expression levels were obtained from RSEM as transcripts per kilobase million (TPM) values. Expression levels of MAPT and TIR-MAPT genes were analyzed per brain region. For those donors with more than one sample in the same brain region, analysis was done only in the sample with highest MAPT expression.
Bacteria culture and Tau purification
Upon cloning, pRK172 vectors encoding different Tau isoforms were transformed in BL21 E. coli competent cells by electroporation, from which Tau was purified, as described elsewhere . These bacteria were cultured in LB medium with 100 ng/ml of ampicillin at 37 °C overnight. Bacterial suspensions were transferred to 1L of LB with 100 ng/ml of ampicillin and further incubated at 37 °C, up until optical density readings at 600 nm ranged between 0.6 and 0.8. At this point, 0.4 mM of IPTG was added to each sample so as to trigger transcription, and incubated once again at 37 °C for 2 h. The samples were centrifuged at 2950×g at 4 °C for 20 min and pellets were resuspended in Tau buffer, which consists of buffer A (0.1 M MES pH 6.4, 0.5 mM MgCl2 and 2 mM EGT) supplemented with 1 mM PMSF; 0.5 NaCl and 5 mM β-mercaptoethanol. Resuspended pellets were sonicated on ice at 24 microns peak to peak (mpp) five times for 1 min, waiting 10 s between repetitions. Upon sonication, samples were centrifuged at 13,850×g at 4° C for 10 min and the resultant supernatants were subjected to 5 min at 100° C and 5 min on ice before centrifuging one more time at 13,850×g at 4 °C for 30 min. Supernatants were kept and Tris 1 M was added until pH reaches a level of 11, to get rid of residual DNA. Ammonium sulfate 50% was added to the samples and they were incubated agitating at 4 °C for at least 1 h before centrifuging again at 13,850×g at 4 °C for 1 h. Pellets containing purified Tau were resuspended in buffer A and the purification process was checked by SDS-PAGE and Coomassie blue staining to check if any other protein bands appeared upon Coomassie blue staining.
Tau protein quantitation
Protein concentration for each Tau isoform was first determined by means of absorbance measurements corrected by individually calculated extinction coefficients (ε), based on the specific amino acid sequence of each isoform  (Supplementary Table 4, online resource). Namely, we measured Tau concentration as indicated by Kundel and collaborators : absorbance at 280 nm (A280) was measured for each Tau isoform and concentration was determined as the ratio between A280 and ε. Extinction coefficients can be estimated from the amino acid sequence of the protein , according to the equation: ε280 = number of tryptophan residues (W) × 5500 + tyrosine residues × 1490 . Using this equation and the amino acid composition of each Tau isoform, the extinction coefficients were calculated for each one (Supplementary Table 4, online resource). Calculated values for T42 and T30 coincide with those calculated by Kundel , which has been typically used for all CNS human Tau isoforms , but are not valid for W-Tau isoforms that contain 2 extra tryptophan residues nor for ET-Tau isoforms, which lack 1 tyrosine that appear on the rest of the isoforms. Coomassie blue staining of samples containing the same amount of protein was performed to confirm their concentrations.
RNA extraction and purification
Either total RNA or cytosolic enriched fraction RNA was purified from cells using RNAeasy Mini Kit (74104, Qiagen) following the protocols described in Qiagen handbook. For brain tissue, previous homogenization using a TissueLyser (Retsch MM300, Qiagen, Hilden, Germany) (30 Hz, 5 min) with 5-mm stainless steel beads (69989, Qiagen) in 700 µl QIAzol Lysis Reagent (79306, Qiagen) was performed. RNA integrity numbers (RIN) were calculated using the Agilent 2100 Bioanalyzer system (Agilent Technologies), and only RNAs with RIN > 5 were used for RT-qPCR.
RNA was purified using RNAeasy Mini Kit (74104, Qiagen) with the following modifications. Cells were pelleted and frozen at -80 °C. Pellets were carefully resuspended in 175 μl of precooled (4 °C) buffer: 50 mM TrisHCl pH 8, 140 mM NaCl, 1.5 mM MgCl2, 0.5% (v/v) Nonidet P40 (1.06 g/ml) plus 1 mM DTT just before use and incubated on ice for 5 min. Lysates were centrifuged at 4 °C for 2 min at 300 × g. Supernatants were kept as cytoplasmic-enriched fraction. To each fraction, 600 μl of Buffer RLT were added. After vortexing, 430 μl of ethanol 100% were added to the homogenized lysate. Samples were transferred to RNeasy spin columns, centrifuged for 15 s at 9000×g and the flow-through was discarded. Samples were treated with 10 µl of DNase in 70 µl of buffer RDD (RNase-free DNase Set 79254, Qiagen) for 15 min at room temperature. The rest of the extraction was performed following the protocol and RNA was collected in 30 µl of RNAse free water twice.
Total RNA was purified using RNAeasy Mini Kit (74104, Qiagen) following provider’s guidelines. Cytoplasmic RNA fractions were purified as previously described. Retrotranscription was performed using 40 ng/µl of RNA with the Transcriptor First Strand cDNA Synthesis Kit (04379012001, Roche) using oligo(dT)18 primer. Semi-quantitative PCR was performed using 1 µl of cDNA (0.5 ng/µl) supplemented with 2.5 mM MgCl2, 0.2 mM each dNTP, 1.3 M betaine, 0.5 mM of each primer (described in supplementary table 2, online resource) and 0.025 U/ml of GoTaq® Flexi DNA Polymerase (M829, Promega), in the following conditions: 95 °C for 2 min, and 35 cycles (PCR1-8) or 30 cycles (PCR9-13) of 95 °C for 45 s, 58.6 °C for 45 s and 72 °C for 45 s, followed by a final extension of 10 min at 72 °C. PCR combinations are described in supplementary table 5, online resource.
Detection of Tau levels by quantitative RT-PCR
RNA was retrotranscribed with the Transcriptor First Strand cDNA Synthesis Kit (04379012001, Roche) using 20 ng/µl RNA with oligo(dT)18 primer. Quantitative PCR was performed in a LightCycler480 (Roche) in the following conditions: 50 °C for 2 min, 95 °C for 10 min, and 40 cycles of 95 °C for 15 s and 60 °C for 1 min. Specific intron-spanning oligonucleotides against MAPT or TIR-MAPT transcript were designed (MAPT-E11-E13-fw/rv and TIR-T-fw/rv, respectively; see Supplementary table 2, online resource). Gene expression was normalized to GAPDH expression using TaqMan primer human GAPDH (Hs02758991_g1, Applied Biosystems). For every RT-qPCR experiment, only samples with RIN above 5 were used.
Detection of Tau levels by Western blot
Frozen brain tissue was homogenized using a TissueLyser (Retsch MM300, Qiagen, Hilden, Germany) (30 Hz, 5 min) with 5-mm stainless steel beads (69,989, Qiagen) in total extraction buffer: 50 mM Tris–HCl pH 7.5, 300 mM NaCl, 0.5% SDS (sodium dodecyl sulphate) and 1% Triton X-100 (Fig. 2d, e and Supplementary Fig. 6b, online resource). The same buffer was also used for cells (Figs. 2b, c, 3b, 4a, 5b and Supplementary Fig. 4b, online resource and Supplementary Fig. 5a, b, online resource). For human cortex brain extracts, a strong lysis buffer (50 mM Tris–HCl pH 7.6, 400 mM NaCl, 1 mM EDTA, 1 mM EGTA, 1% SDS) was also used in Fig. 7b. The homogenates are incubated for 15 min at 95 °C, centrifuged (16,100×g, 10 min) and the supernatant was considered the brain extract. Protein concentration was measured using the DC protein assay kit (500-0111, Bio-Rad). From each sample, equal amount of total protein was resolved on a 10% Bis–Tris gel and transferred to nitrocellulose membranes. Blots are probed with the corresponding primary antibodies (see below), followed by horseradish peroxidase-conjugated anti-mouse or anti-rabbit antibody (Dako, Glostrup, Denmark). Protein expression was quantified by measuring the ECL signal with ImageJ software (Fig. 7 and Supplementary Fig. 6, online resource) (http://rsbweb.nih.gov/ij/) or with Quantity One® 1-D analysis software from Bio-Rad.
Different antibodies were used for the different essays, as indicated in each one. Altogether, Tau 12 (Millipore, MAB2241; amino acids 6-18, diluted 1/1000), Tau 5 (Calbiochem, D00139295, amino acids 210-241, diluted 1/1000), Tau antibody (NOVUSBIO, NB100-82247, epitope around amino acid 231, diluted 1/500) and Tau 7.51 (, amino acids 315-376, diluted 1/100) were used as antibodies recognizing all Tau isoforms; Tau 1 (Chemicon, MAB3420, diluted 1/1000) recognizing dephosphorylated residues Ser195, 198, 199 and 202 and AT8 (phospho-Ser202/Thr205; Innogenetics, MN1020, diluted 1/100), AT180 (phospho-Thr231; Innogenetics, MN1040, diluted 1/100), Tau 404 (phospho-Ser404; Life Technologies, 44758G, diluted 1/1000), Tau 396 (phospho-Ser396; Life Technologies, 44752G, diluted 1/1000) and PHF1 (phospho-Ser396/Ser404, kind gift of Peter Davies , diluted 1/100) as antibodies recognizing phosphorylated Tau isoforms. In addition, β-actin (SIGMA A5441, diluted 1/20000) or GAPDH (Cell signalling, 2118, diluted 1/1000 antibodies were used as loading control.
A peptide composed of the W-Tau unique sequence (KKVKGVGWVGCCPWVYGH) (W-Tau peptide) was synthesised and a polyclonal IgG antibody against that peptide was obtained from Abyntek (Bizkaia, Spain), using New Zealand rabbit as host strain. This antibody was named W-Tau antibody and its specificity was tested (diluted 1/1000) to ensure it reacts exclusively with TIR-Tau but not with other Tau isoforms (Fig. 2a–c).
Immunoprecipitation was performed with W-Tau and total Tau (NOVUSBIO, NB100-82247) antibodies using SantaCruz Immunoprecipitation Kit (SC-2003) using 500 µg of whole cell extracts and following provider’s guidelines.
Tau phosphorylation determination
HEK293T cells were transfected with pSG5 plasmids (2 µg/p60) encoding different Tau isoforms using Lipofectamine and Plus reagents following instructions of the supplier (#18324 and #11514, respectively, Life Technologies). 48 h after transfection, cells were collected, pelleted and homogenized in 200 µl of total extraction buffer. Protein concentrations were measured using the DC protein assay kit (500-0111, Bio-Rad) and samples containing equivalent amounts of protein were analyzed by Western blot, as previously described, using nitrocellulose membranes. Immunodetection was performed with antibodies Tau 7.51 and Tau 5 for total Tau, Tau 1 for dephosphorylated Tau and AT8, AT180, Tau 404, Tau 396 and PHF1 for phosphorylated Tau.
Tau solubility determination
Once again, HEK293T cells were transfected with pSG5 plasmids (2 µg/p60) encoding different Tau isoforms using Lipofectamine and Plus reagents (Life Technologies). After 48 h, cells were collected, pelleted and homogenized in 150 µl of lysis buffer (50 mM Tris–HCl (pH 7,4), 150 mM NaCl, 20 mM NaF, 1 mM Na3VO4, 0.5 mM MgSO4; supplemented with protease inhibitors cocktail, 04693159001 Roche) at 4 °C. Homogenates were centrifuged at 27,000×g for 20 min at 4 °C. Pellets were discarded and 1% sarkosyl was added to the supernatants, being incubated for 1.5 h agitating at room temperature. Upon incubation, samples were centrifuged at 150,000×g during 45 min at 4 °C. Supernatants were kept as sarkosyl-soluble fraction and pellets were resuspended in 100 µl of a mix of total extraction buffer and loading buffer (1:1) and considered sarkosyl-insoluble fraction.
Results were further confirmed by carrying out a similar protocol to measure Triton X-100 solubility. Namely, pelleted cells were homogenized in 500 µl of lysis buffer (1% Triton X-100, 50 mM Tris–HCl pH 7, 100 mM NaCl and 1 mM EDTA) and incubated 20 min at 4 °C. The samples were centrifuged at 18,000×g for 5 min and the supernatants were kept as soluble fraction. Pellets were resuspended in total extraction buffer (see point 11 of methods) and considered Triton-insoluble fraction.
In both cases, equivalent volumes of each sample were then analyzed by Westernblotting.
Tau in vitro aggregation determination
Tau aggregates were grown for the different Tau isoforms by vapor diffusion in hanging drops in the standard way used for protein crystallizations . The different Tau isoforms were purified from bacteria as described above and their concentration was estimated by measuring absorbance at 280 nm and taking into account individually calculated extinction coefficients for each isoform (Supplementary table 4, online resource) and confirmed by means of Coomassie Blue staining. Equivalent quantities of each isoform were added to the corresponding volume of buffer A (0.1 M MES pH 6.4, 0.5 mM MgCl2 and 2 mM EGT) plus 50 mM NaCl in the presence of heparin, reaching a concentration of 1 mg/ml. The reservoir contained 0.2 M NaCl in buffer A. Aggregates were obtained after incubation for 10 days at room temperature (see also ). An aliquot was kept for electron microscopy analysis, and the rest of the samples were centrifuged in Airfuga at 28 lb per square inch (psi) for 30 min at room temperature to separate soluble and aggregated protein. The fractionated proteins were characterized by electrophoresis and Western blot.
Tau microtubule-binding capacity assay
For microtubule preparation, the brains of 12 2-month-old C57BL/6J mice were homogenized with a potter in isotonic buffer (0.32 M sucrose, 1 mM EGTA, 1 mM MgCl2, 10 mM phosphate buffer pH 7 and 1 mM PMSF) at 4 °C. Supernatant was collected after 40 min ultracentrifugation at 100,000×g at 4 °C in an Optima L-100 XP Ultracentrifuge (Beckman Coulter). 100 μl of this supernatant were incubated with equal quantity of protein for the different Tau isoforms, purified from bacteria as mentioned above during 30 min at 37 °C, in the presence of 30% glycerol, 1 mM PMSF and 1 mM GTP to promote tubulin polymerization. After that, samples were ultracentrifuged for 60 min at 100,000×g at 25 °C.
Equivalent volumes of pellets containing microtubules and microtubule-bound Tau were analyzed by Western Blot, using Tau 5 and tubulin antibodies and the ratio between Tau 5 and tubulin signal was calculated for each isoform as their microtubule-binding ratio, normalized to the corresponding full-length isoform.
SK-N-MC human neuroblastoma cells (ATCC® HTB-10™) were treated with GSK3 inhibitors SB216763 (25 µM, GlaxoSmithKline compounds ) and AR-14418 (10 µM, AstraZeneca compound ) or Aβ1–42 (1.1 µM; Neosystem Laboratoire, Strasbourg, France) for 24 h and whole cell extracts were used for Western blot analysis, probing the blots with W-Tau antibody. From those cells treated with AR-14418 and Aβ1-42, RNA-enriched cytoplasmic fraction was also retrieved and TIR-MAPT RNA levels were assessed by quantitative RT-PCR as previously described.
Splicing factor binding site prediction analysis
Exonic and intronic human MAPT sequences were retrieved from ENSEMBL (ENSG00000186868). Exon 12–exon 13 and exon 12–intron 12 sequences (Exon 13–exon 14 and exon 13–intron 13, according to ENSEMBL nomenclature of Tau exons/introns) were analyzed using ESEFinder 3.0 (http://exon.cshl.edu/ESE/) . ESEFinder settings were adjusted to show binding sites with score 4.0 or higher. The analysis was completed by a detailed bibliographic search comparing SRSF2 binding sites described in Masaki et al., 2019 and Cavaloc et al., 1999 [11, 38] with junctions MAPT Exon 12–exon 13 and exon 12–intron 12 sequences.
Quantitative data, represented as mean ± SD or SEM, were compared between groups using the two-tailed Student’s t test. For multiple comparisons, one-way ANOVA and Dunnett's test were performed to compare each isoform and the correspondent full-length isoform. When the distribution of the data was not Gaussian a nonparametric Kruskal–Wallis test was used. The differences are given with their corresponding statistical significance or p value, which is the probability that the difference occurred merely by chance under the null hypothesis (*p ≤ 0.05; **p ≤ 0.01; ***p ≤ 0.001; ****p ≤ 0.0001; N.S. not significant). The method used for each experiment is specified in the corresponding figure legend.