Generation of htau-KI
The htau-KI mouse line was generated together with Taconic Bioscience. The bacterial artificial chromosome (BAC) RP11-111I23 containing the whole human tau sequence was used to generate a targeting vector, which has the mouse genomic sequence from the translation initiation codon in exon 2 to the termination codon in exon 10 replaced with its human counterpart. This vector was transfected into mouse ES cells from the C57BL/6NTac strain. Two ES cell lines, Mapt 20955-A-C08 and Mapt 20955-A-F07, were obtained. Line Mapt 20955-A-C08 resulted in offspring with a chimerism >50% after injection into blastocysts of BALB/c females. Blastocyst injection of line Mapt 20955-A-F07 did not result in chimeric offspring. Therefore, all generated mice resulted from ES cell line Mapt 20955-A-C08. Based on coat color, the mice with highest chimerism were selected and bred to C57BL6/NTac females. Germline transmission was confirmed by birth of black offspring. Mice were subsequently genotyped to verify the presence of the htau gene (See Fig. 1 for a scheme of the targeting strategy). Eight males and ten females were used as founders for establishment of the colony. Mice were backcrossed for two generations to C57BL/6J background and analyzed for a homozygous deletion in the Nnt gene as major hallmark of the C57BL/6J strain.
Animals
All animal experiments were approved by the responsible animal ethics committee of the state of Saxony-Anhalt, Germany (Landesverwaltungsamt Sachsen-Anhalt, Department of Consumer Protection and Veterinary Affairs, Halle (Saale), Saxony-Anhalt, Germany) under the following approval number: 42502-2-1371 MLU. For electrophysiology, mice were bred in Halle, Germany, and transported to KU Leuven, Belgium, a minimum of 2 weeks ahead of testing. Housing and animal procedures were approved by the KU Leuven Ethical Committee (project P181/2017). We confirm that all animal handling procedures were carried out in accordance with directive 2010/63/EU of the European Parliament and of the Council on the protection of animals used for scientific purposes, the Belgian and German Animal Protection Acts and recommendations from the Federation of European Laboratory Animal Science Associations (FELASA).
After establishing the line, htau-KI+/+ and 5xFADxhtau-KI+/+ mice were generated from a mating between heterozygous 5xFADxhtau-KI+/− mice and homozygous htau-KI+/+ mice. C57BL/6J WT controls and 5xFAD mice were obtained by mating 5xFAD mice with C57BL/6J WT mice. All mice were kept on a C57BL/6J background, and all 5xFAD mice were bred heterozygous. Age-matched phenotyping groups of 6 and 12 months were generated using timed mating. Mice had access to water and food ad libitum and were kept on a regular 12/12-h light/dark cycle. After concluding behavioral analyses, the mice were sacrificed at an age of 7 and 13 months, respectively.
Genotyping
At the age of 19–21 days, ear punches were taken to individually mark the mice. The resulting material from each mouse was used for genotyping. After sacrificing the mice at the end of the respective studies, tail tips were taken for confirmation of the genotype. DNA was isolated from the ear punches/tail tips using the 1-Step kit (Nexttec, Leverkusen, Germany) according to the manufacturer’s protocol. The DNA was analyzed for the presence of 5xFAD and htau genes using the GoTaq DNA Polymerase (Promega, Madison, Wisconsin). Primers were specific for the different genes, and sequences are shown in Supplementary Tab. 4. Samples were afterwards loaded onto a 1.6% agarose gel in TAE buffer with ethidium bromide to visualize the PCR products.
Organ collection
After completion of all behavioral experiments, mice were euthanized using CO2 inhalation and subsequently perfused with 20 mL PBS. The brain was collected and divided into its two hemispheres; the left hemisphere was fresh frozen on dry ice, and the right hemisphere fixed in 4% PFA. All fresh frozen samples were subsequently stored at −80°C.
Biochemical analyses
Fresh frozen left-brain hemispheres were homogenized in cell extraction buffer (Invitrogen) using the following protocol: frozen brain was transferred to a Dounce tissue homogenizer, after which 1 mL of cell extraction buffer (Invitrogen, Carlsbad, California) was added. Brain tissue was mechanically homogenized, followed by incubation on ice for 30 min. Every 10 min, the homogenate was vortexed and resuspended. Subsequently, the samples were centrifuged for 30 min at 13,000×g and 4°C. The supernatant was collected in a fresh Eppendorf tube, and both supernatant and pellet were stored at −20°C until further use.
Western blot
Brain homogenates were thawed, and protein concentration was measured using the Pierce Bradford assay (Thermo Fisher Scientific) according to the manufacturer’s protocol. Selected brain extracts were also dephosphorylated using lambda phosphatase essentially as described elsewhere [7]. SDS-PAGE was performed using the NuPage system and reagents (Invitrogen), following the protocol of the manufacturer. The Pageruler™ Plus Prestained Protein Ladder, 1.0 mm Bis-Tris gels with 12 or 17 wells, and MES SDS running buffer were used. Dephosphorylated tau was analyzed on 12.5 % SDS-PAGE Tris gels. For subsequent Western blotting, the semi dry system by Galileo Biosciences (Cambridge, Massachusetts) was used according to the manufacturer’s instructions. Proteins were blotted on 0.2 μm nitrocellulose membranes (Amersham Biosciences, Little Chalfont, UK) for 90 min at 38 mA. Afterwards, the membrane was blocked while floating for 30 min in a box with 5 % milk 0.05 % Tween TBS, under continuous shaking. The membrane was then placed in a 50-mL tube and incubated over night with the primary antibody (see Supplementary Tab. 3; dissolved in 5 % milk 0.05 % Tween TBS) at 4 °C on a roller bank. The following day, the membrane was washed three times 5 min in 0.05 % Tween TBS, after which it was incubated with the secondary antibody (see Supplementary Tab. 3), also diluted in 5 % milk 0.05 % Tween TBS (1 h, RT, on roller bank). Subsequently, the membrane was washed twice in 0.05 % Tween TBS, and once in TBS (all washes 5 min, RT, on a roller bank). For protein detection, the SuperSignal West Pico or Femto kit (Thermo Fisher Scientific) was used according to the manufacturer’s protocol. Bio1D software (Vilber Lourmat, Collégien, France) or ImageJ was used for quantification of signal intensity. For absolute quantification of 3R and 4R tau, the tau protein ladder (Sigma Aldrich, St. Louis, MO, USA) has been used as a standard. Protein amounts were calculated densitometrically in ImageJ by comparing GAPDH normalized signal intensities of samples with bands of the tau ladder.
RNA isolation and qRT-PCR
After removal of the olfactory bulb, a small piece of prefrontal cortex was removed from the fresh frozen left-brain hemisphere for RNA isolation. The Nucleospin RNA kit (Macherey Nagel, Oensingen, Switzerland) was used according to the manufacturer’s protocol with some minor adjustments. TCEP was used instead of β-mercaptoethanol in the lysis step. Furthermore, RNA was eluted in 40-μL RNase-free H2O instead of 60 μL, and the eluate was loaded onto the spin column a second time to retrieve higher RNA yield. RNA concentrations were measured using a Nanodrop 2000 (Thermo Fisher Scientific). The cDNA was synthesized using the Superscript II kit (Invitrogen), according to the manufacturer’s protocol. Three hundred nanograms total RNA was used, and random primers were added during the first step.
A qRT-PCR was performed for quantification of 3R and 4R tau isoforms. Therefore, a standard curve of double-stranded DNA (Metabion, Planegg, Germany) was prepared. The sequences were specific for 3R and 4R and are shown in Supplementary Tab. 4. A standard curve from 30 to 300,000 molecules/μL was prepared. Two biological replicates of the standard curve were prepared, and the standard curve best fitting a linear equation (R2 > 0.98) was used to calculate the DNA concentration of the other samples. Primers were specific for 3R and 4R tau isoforms, as described by Ingelsson et al. [21]. The specificity was confirmed by a lack of signal when 3R primers where used with 4R tau DNA and vice versa. All primer sequences are shown in Supplementary Tab. 4. The Rotorgene 3000 system (Corbett research, Mortlake, Australia) was used, and the SYBR Green kit (Applied Biosystems, Foster City, California) according to the manufacturer’s protocol. For each biological replicate, three technical replicates were performed, and a maximum difference of 0.4 CT was allowed between technical replicates.
Gene expression analysis of AD-specific genes
Isolated RNA of 12 mice per genotype (7- and 13-month-old males and females, three each) was used for the nCounter Mouse AD gene expression panel developed by NanoString Technologies (Seattle, Washington). One hundred to one hundred fifty nanograms of RNA was used to set up the hybridization reaction, after which the experiment was performed according to the manufacturer’s protocol. The prep station was run with high-sensitivity settings, and the Analyzer with MAX data resolution (555 FOV). Data normalization and analysis was performed using the nSolver 4.0 software (NanoString Technologies) according to the manufacturer’s protocol. Normalization was done by dividing counts within a lane by geometric mean of the housekeeping genes from the same lane. For the downstream analysis, counts were log-transformed from normalized count values.
Mouse-human co-expression data comparison
Data on 30 harmonized human co-expression modules, significantly enriched for differentially expressed genes in AD post mortem brain samples, were obtained via the AD Knowledge Portal (https://www.synapse.org/#!Synapse:syn11914606). The data derives from over 2000 human brain samples from three independent AD cohorts (ROSMAP, Mayo Clinic, Mount Sinai Brain Bank) and includes seven distinct brain regions. Details on post mortem brain sample collection, tissue and RNA preparation, and sequencing can be found elsewhere [22,23,24]. Among the 30 human co-expression modules, five specific consensus clusters have been previously described by Wan et al. [25]. These consensus clusters consist of overlapping co-expression modules, which are associated with specific AD-related changes that were highly preserved across studies and brain regions. In order to align human with mouse data, a Nanostring gene expression panel in combination with a recently described systems biology approach was used [26]; https://www.synapse.org/#!Synapse:syn23569889) to assess LOAD-relevance of the novel mouse models. Briefly, gene expression changes (log fold change) across a set of human key genes within AD relevant co-expression modules were correlated with expression changes in the novel mouse model to determine opposing and concordant effects driven by htau and amyloid. To obtain log fold changes associated with htau in the htau-KI and amyloid effects in the 5xFAD model, differential gene expression analysis was performed for each mouse model and sex using the voom-limma package in R [27]. All log fold change values for human transcripts were obtained via the AD Knowledge Portal as previously reported (https://www.synapse.org/#!Synapse:syn23569889). The correlation between changes in expression (log fold change) was computed for each gene in a given expression module with each mouse model, sex, and age. Correlation coefficients were obtained using cor.test function built in R as: cor.test( LogFC(h), LogFC(m)), where LogFC(h) is the log fold change in transcript expression of human AD patients compared to control patients and LogFC(m) is the log fold change in expression of mouse transcripts compared to control mouse models. Significance of cross species post mortem brain module correlations were determined after FDR multiple testing correction, and age-, strain-, and sex-specific effects were assessed using a multiple linear regression analysis.
Gene set enrichment analysis in mouse and human data
Gene set enrichment analysis based on the method proposed by Subramanian et al. [28] was performed using the cluster profiler package in the R software environment [29] for the Kyoto Encyclopedia of Genes and Genomes (KEGG) database. Briefly, human data with log fold changes for the seven AMP-AD brain regions were obtained through the AD Knowledge Portal (https://www.synapse.org/#!Synapse:syn14237651). Only orthologous genes on the NanoString Mouse AD panel were selected and KEGG pathway enrichment was performed for each brain region independently to identify significantly up- and downregulated gene sets. For the mouse data, differential expression analysis between the 5xFAD and 5xFADxhtau-KI models was performed to obtain a list of fold changes highlighting genes that are either up- or downregulated in the presence of htau on the 5xFAD background. Enrichment scores for all significantly associated KEGG pathways were computed to compare relative expression on the pathway level between post mortem brain samples and the 5xFADxhtau-KI mouse model. Finally, we extracted the leading-edge gene set, which identified the genes that appeared in the ranked list at or before the point, at which the running sum reaches its maximum deviation from zero. This leading-edge subset can be interpreted as the core that accounts for the gene set’s enrichment signal.
Immunohistochemistry
The right hemisphere was fixed in 4% PFA for 2 days at 4°C, followed by incubation in 30% sucrose solution at 4°C. After a minimum of 2 days, the organs were washed in PBS, and subsequently frozen in Tissue-Tek O.C.T. (Sakura Finetek, Alphen aan den Rijn, the Netherlands) in plastic molds. Sagittal slices (30 μm) of the hemisphere were prepared using a cryomicrotome (Cryostar NX70, Thermo Fisher Scientific, Waltham, Massachusetts), starting in the center of the brain. Slices containing the hippocampal formation were collected in 24-well plates and stored at 4°C in PB buffer containing 0.025% sodium azide until staining.
For free-floating 3,3’-diaminobenzidine (DAB) staining, the slices were washed in TBS (5 min, RT), after which the endogenous peroxidase was blocked by incubation with 1% H2O2 in 60% MeOH (30 min, RT). Subsequently, the slices were washed three times in TBS, followed by blocking with 5% goat serum in 0.3% Triton-X TBS (30 min, RT). For mouse-on-mouse staining, M.O.M. reagent (Vector Laboratories, Burlingame, California) was added during the blocking step. The primary antibodies (see Supplementary Tab. 3) were diluted in 5% goat serum 0.1% Triton-X TBS, and the slices incubated overnight at 4°C in a humidified chamber. The following day, slices were washed in TBS three times, followed by incubation with the respective secondary antibodies (see Supplementary Tab. 3), which were diluted in 2% BSA TBS (1h, RT). After another three washes in TBS, slices were incubated with ExtrAvidin® Peroxidase, diluted 1:1,000 in 2% BSA TBS (1h, RT), followed by two times washing in TBS, and one time in Tris buffer. DAB staining was performed by incubating slices in 0.05% DAB 0.015% H2O2 in Tris buffer for 4 min (Iba-1, GFAP, 3A1) or 8 min (all other applied antibodies) at RT. Subsequently, the slices were washed once in Tris buffer and once in TBS, followed by placing the slices on glass slides, coated with protein glycerol (Carl Roth, Karlsruhe, Germany). After the sections were air dried, the slides were dehydrated by immersing them twice in ethanol followed by immersion in ROTI®-Histol (Carl Roth). Finally, the slices were embedded using Permount mounting media (Fisher Scientific, Hampton, New Hampshire) and closed with cover slips. Slides were stored at RT.
After air drying, Congo red amyloid staining was performed according to Uhlmann et al. [30]. In brief, sections were incubated in alkaline saturated NaCl solution for 20 min followed by incubation in filtered alkaline Congo red solution (0.2 %) for 30 min at room temperature. After dipping the slides eight times in 95% ethanol and 2 × 30s in 100% ethanol the sections were dehydrated and embedded as described above. Slices were recorded with a BZ-9000E microscope (Keyence, Osaka, Japan), and subsequent quantification of plaques was performed manually or using the BZ-II-Analyzer software (Keyence).
Electrophysiology
Acute slice preparation
Mice were sacrificed by cervical dislocation, after which the brain was rapidly dissected and immersed in ice-cold artificial cerebrospinal fluid (ACSF), saturated with carbogen (95% O2, 5% CO2) and containing (all in mM) 124.0 NaCl, 4.9 KCl, 1.3 MgSO4, 2.5 CaCl2, 1.2 KH2PO4, 25.6 NaHCO3, and 10.0 glucose (pH 7.4). From the left hemisphere, the hippocampus was immediately isolated and cut into 300-μm-thick slices using a custom-made tissue chopper for field potential recordings. The part of the right hemisphere containing the medial hippocampus was cut into 400-μm-thick slices using a vibratome (Microm HM 650 V, Thermo Fisher Scientific) for whole-cell recordings. The ages of the mice used for electrophysiology were (mean ± SD) as follows: 5xFAD = 8.3 ± 1.7 months, htau-KI = 7.6 ± 0.9 months, 5xFADxhtau-KI = 7.7 ± 1.1 months, WT = 8.1 ± 1.7 months.
Extracellular recording of field potentials
Hippocampal slices were placed in an incubation chamber (Warner Instruments, Hamden, Connecticut) with carbogenated ACSF (same composition as during slice preparation) and left to recover at RT for at least 1 h. For the recording of extracellular field potentials, a multi-electrode array (MEA) system (Multi Channel Systems, Reutlingen, Germany) was used as described before [31]. After incubation, a single slice was placed on a MEA chip (60 TiN electrodes in 8×8 layout and 100 μm spacing; imec, Heverlee, Belgium) [32] and perfused with ACSF at 2.5 mL/min and a constant temperature of 32°C. Stimulation and recording was performed using a stimulus generator (SG 4002), MEA1060-BC amplifier, temperature controllers (TC, PH01), and software (MEA_Select, MC_Stimulus, MC_Rack) from Multi Channel Systems. A single electrode located at the level of the Schaffer collaterals in the CA1 region was selected for biphasic constant voltage stimulation. The analysis of the evoked field excitatory post-synaptic potentials (fEPSPs) was focused on the signals from the electrode adjacent to the stimulation electrode (in anterograde direction of the Schaffer collaterals). Data streams were sampled at 10 kHz. First, an input/output curve was established using stimulation intensities ranging from 0.5 to 4 V (at 0.5 V intervals). The intensity leading to 35% of the maximum of the input/output curve was used for all subsequent stimulations. Next, a series of paired-pulse stimulations with intervals of 10, 20, 50, 100, 200, and 500 ms was applied. Last, after obtaining a stable baseline for at least 30 min (3 sweeps with 15 s interval, repeated every 3 min), LTP was induced by 3x theta-burst stimulation (TBS) (each TBS consists of 10 trains of 4 pulses at 100 Hz, with 200-ms intervals and 200 μs pulse-width; 10-min interval between subsequent TBS). fEPSPs were recorded for at least 120 min after LTP induction (3 sweeps with 15 s interval, repeated every 5 min). For analysis, raw data were extracted using MC_Rack software in replayer mode (Multi Channel Systems). fEPSP slope values were obtained with the region of interest set from peak-to-peak (10–90%), and all groups of 3 subsequent sweeps were averaged. Paired-pulse ratios were calculated by dividing the slope of the second fEPSP by the slope of the first. LTP recordings were normalized to the average baseline slope.
Whole-cell recordings
After cutting, slices were placed in an incubation chamber (Warner Instruments) with carbogenated ACSF (all in mM: 124.0 NaCl, 4.9 KCl, 1.2 NaH2PO4, 25.6 NaHCO3, 2.0 CaCl2, 2.0 MgSO4, 10.0 glucose, pH 7.3–7.4) and left to recover at RT for at least 90 min. Whole-cell voltage clamp recordings of CA1 pyramidal cells were performed at RT using a MultiClamp 700B patch-clamp amplifier and pClamp™ software (Molecular devices, San José, California), as previously described [33]. Recording electrodes, pulled from borosilicate glass (World Precision Instruments, Sarasota, Florida), were filled with a solution containing the following (in mM): 135 CsMeSO4, 4 NaCl, 4 Mg-ATP, 0.5 EGTA-Na, 0.3 Na-GTP, 10 K-HEPES, 5 QX-314; pH 7.3 (pipette resistance 3–5 MΩ). Access resistance was 10–20 MΩ and was then compensated to 75%. If the input resistance changed more than 25% during the recordings, the neuron was excluded from the study. Miniature excitatory and inhibitory post-synaptic currents (mEPSCs and mIPSCs) were measured, mostly consecutively from the same neurons. First, mEPSCs were recorded at the reversal potential for GABAA receptor-mediated events (−60 mV), after which mIPSCs were recorded at the reversal potential for glutamatergic currents (+10 mV) with tetrodotoxin (1 μM) present in the bath medium. To verify that mEPSCs were indeed glutamatergic, they were blocked at the end of the experiments by applying 20 μM 6-cyano-7-nitroquinoxaline-2,3-dione (CNQX) and 10 μM d-aminophosphonovalerate (d-APV). Similarly, mIPSCs were verified by blocking them using 100 μM picrotoxin, a GABAA receptor antagonist. Data were low-pass filtered at 2 kHz and acquired at 10 kHz using Digidata 1440 and pClamp™ 10 software. Offline analysis of mEPSCs and mIPSCs was performed using MiniAnalysis software (v.6.0.7, Synaptosoft, Decatur, Georgia).
Behavioral studies
Mice were housed on a light regime of 12 h dark and 12 h light with lights on at 7 AM. Animals subjected to behavioral experiments were tested in the light phase of the day. The experimenter was blinded to the genotype of the mice for the complete duration of the tests, as well as the subsequent analysis of the videos.
Primary screening
Before starting full behavioral analysis, all mice were screened for general health according to the SHIRPA protocol [34]. Vision, hearing, grip strength and mobility were tested. The weight of each mouse was also documented.
Elevated plus maze
The setup was a plus shaped maze with arms of 40 cm × 7.5 cm, placed 70 cm above the ground. Two of the arms had walls with a height of 20 cm (closed arms), the other 2 arms were without walls (open arms). Mice were placed in one of the closed arms and allowed to explore freely for 10 min. Behavior was recorded with a camera attached to the ceiling above the maze and analyzed using Biobserve Viewer software (Biobserve GmbH, Bonn, Germany). The time spent in open arms was used as measure for exploratory and anxiety-like behavior.
Morris water maze
The Morris water maze consisted of a round pool (diameter 120 cm) with white walls, filled with water. A platform (diameter 10 cm) covered with white sports tape (for grip) was placed inside, 8 mm below water level. The water temperature was 26°C (±0.5°C). Four differently colored/shaped 3D environmental cues were attached to the side of the pool, to allow the mice to orientate. The protocol consists of 4 days, with 4 swimming trials each day. For each trial, a mouse was placed in the pool and allowed to swim and to find the platform within 60 s. If the mouse did not find the platform within 60 s, the experimenter led the mouse to the platform. When the mouse was on the platform, it was allowed to sit there for 10 s to orientate. The mice were subsequently dried with paper towels and warmed under a heat lamp for 5 min, followed by the next trial. Each trial, the mouse started in a different quadrant of the pool. The platform was always located in quadrant 3. During trials 1 and 4 of each test day, the mice started in quadrant 1, opposite from the platform. In the second trial, the mice started in quadrant 2, and in the third trial in quadrant 4. Movement of the mice was recorded using a camera, and the animal was tracked using the Biobserve Viewer software (Biobserve GmbH).
Statistical analysis
Statistical analysis was done using GraphPad Prism 8 (GraphPad Software). One-way ANOVA with Tukey post hoc analysis was used. For electrophysiology, time series were tested in GraphPad Prism 8 using two-way repeated measures analysis of variance (RM-ANOVA) and Tukey’s or Dunnett’s multiple comparisons test. Because of significant differences in variance between samples of mEPSCs and mIPSCs, they were analyzed by Welch’s one-way ANOVA test using the two-stage linear step-up procedure of Benjamini, Krieger, and Yekutieli to correct for multiple testing. Cumulative distribution functions of mEPSCs and mIPSCs were compared by the Kolmogorov-Smirnov test. For analysis of Morris water maze data, RM-ANOVA were performed using the lmerTest package in R with Tukey post hoc test. For the gene expression analysis of the NanoString Mouse AD panel, counts were log-transformed from normalized count values. Differential gene expression analysis was performed for each mouse model and sex using the voom-limma package in R. For the human to mouse comparison, the correlation between changes in expression (log fold change) across species was computed for each gene in a given AMP-AD expression module with each mouse model, sex, and age based on an established systems biology method to assess disease relevance of novel LOAD mouse models [26, 35, 36]. Briefly, Pearson’s correlation coefficients based on log fold changes differences were obtained for all co-expression module comparisons between human and mouse. Significant positive or negative correlations (FDR-adjusted P < 0.05) across all orthologous genes within each of the 30 human AMP-AD expression modules were assessed after multiple testing correction for all modules using the cor.test function built in R. Correlation plots were visualized using the corrplot package [37]. Gene set enrichment analysis was performed using the clusterprofiler package in R for the KEGG pathway database.