Lack of exon 10 in the murine tau gene results in mild sensorimotor defects with aging
© Gumucio et al.; licensee BioMed Central Ltd. 2013
Received: 21 August 2013
Accepted: 19 November 2013
Published: 22 November 2013
Skip to main content
© Gumucio et al.; licensee BioMed Central Ltd. 2013
Received: 21 August 2013
Accepted: 19 November 2013
Published: 22 November 2013
Complex species-specific, developmental- and tissue-dependent mechanisms regulate alternative splicing of tau, thereby diversifying tau protein synthesis. The functional role of alternative splicing of tau e.g. exon 10 has never been examined in vivo, although genetic studies suggest that it is important to neurodegenerative disease.
Gene-targeting was used to delete exon 10 in murine tau on both alleles (E10−/−) to study its functional role. Moreover, mice devoid of exon 10 (E10+/−) on one allele were generated to investigate the effects of 1:1 balanced expression of 4R-/3R-tau protein, since equal amounts of 4R-/3R-tau protein are synthesized in human brain. Middle-aged E10−/− mice displayed sensorimotor disturbances in the rotarod when compared to age-matched E10+/− and wild-type mice, and their muscular grip strength was less than that of E10+/− mice. The performance of E10+/− mice and wild-type mice (E10+/+) was similar in sensorimotor tests. Cognitive abilities or anxiety-like behaviours did not depend on exon 10 in tau, and neither pathological inclusions nor gene-dependent morphological abnormalities were found.
Ablation of exon 10 in the murine tau gene alters alternative splicing and tau protein synthesis which results in mild sensorimotor phenotypes with aging. Presumably related microtubule-stabilizing genes rescue other functions.
Tau belongs to the family of microtubule-associated proteins (MAPs), which bind to and/or interact with microtubule. It has been suggested that tau and other MAPs serve to promote assembly of microtubule to make them structurally stably, yet dynamic . Disruption of the tau gene led to structural abnormalities of microtubule organization in small-calibre axons , and behavioural deficits with hyperactivity, impaired motor strength and coordination . In contrast with the initial report , there were in vitro phenotypes with delayed development of neuronal polarity and formation of axons in cultured embryonal hippocampal neurons from tau knockout mice . Functional redundancy, particularly between tau and map genes likely exists, since map1a protein expression was enhanced in tau knockout mice [2, 4]. Indeed tau/map1b double knockout mice displayed brain anomalies with severe defects on axon tract and neuronal layer formation, together with abnormal growth cone morphology and pronounced microtubule disorganization in primary neurons derived from these mice . Tau protein synthesis and function is regulated by alternative splicing in a complex species-specific, developmental- and tissue-dependent manner [6, 7]. Depending upon the inclusion or exclusion of exon 10, tau mRNA isoforms with three (3R-tau mRNA) and four (4R-tau mRNA) microtubule binding domains are generated, and in total six tau mRNA isoforms are produced in the human brain. In the adult human brain, splicing is balanced with a 1:1 expression of 4R- and 3R-tau mRNA [8, 9]. This is in contrast to mouse brain in which only 3R-tau mRNA is generated at birth and only 4R-tau mRNA is synthesized at adulthood [7, 10, 11]. Thus equal amounts of 4R- and 3R-tau protein is synthesized in adult human brain, while only 4R-tau protein is produced in adult mouse brain. Interestingly, mutations in exon 10 or in adjacent regulatory sequences, can give rise to neurodegenerative disease with accumulation of filamentous inclusions of tau in the human brain. Regulation of alternative splicing of tau is very complex and partly unknown. In order to investigate the functional role of exon 10 in murine tau we ablated the gene on both alleles, and thereby generated E10−/− mice which should in theory only synthesize 3R-tau protein. Moreover, the effect of a 1:1 balanced 3R-/4R-tau mRNA splicing of human brain was examined by generating mice in which exon 10 in tau was deleted on only one allele (E10+/−). E10+/− mice should in theory synthesize equal amounts of 3R- and 4R-tau protein, as in the human brain (Additional file 1: Figure S1). The effects of fine-tuned regulation of alternative splicing being critical to maintain neuronal functions and viability could thereby be assessed . Here we show that E10−/− but not E10+/− mice, display impaired sensorimotor abilities with aging.
The IntelliCages are equipped with four corners, each with two doors leading to their respective water bottle. In the Place learning protocol, an error was defined as a nosepoke on any of the seven doors where sucrose water (reward) was not available, i.e. the closed side of each corner (=four doors) and the accessible side in the three corners where only water was available (=three doors). Any visit to a door where reward was not provided was defined as an erroneous behaviour. The percentage of errors made by E10−/− mice decreased from the first day of testing to the last day (from 84.2 ± 4.6% to 63.6 ± 8.9%, p = 0.06; chance level is 87.5%). Likewise, the percentage of errors decreased for E10+/− mice from 95.3 ± 2.3% to 79.8 ± 6.0% (p < 0.05). However, spatial learning ability did not depend on genotype (Figure 4B). During reversal learning, when reward was provided behind the diagonally opposite door, E10−/− mice made more errors on first 84.2 ± 5.5% than on the last day of testing 59.7 ± 8.7% (p < 0.05; chance level is 87.5%), but this was not observed for E10+/− mice (82.8 ± 5.7% to 79.7 ± 6.2%, p = 0.72; Figure 4C). Again, reversal learning did not depend on genotype. When mice again could access water behind all eight doors (extinction of Place learning, 12 hours), E10+/− and E10−/− mice preferred the previously rewarded location (t(38) = 2.09, p < 0.05 i.e. above chance level 12.5%; Figure 4D), i.e. they still preferred the location where reward had been provided during the reversal learning. To assess cognitive function, E10+/− and E10−/− mice were also exposed to a passive avoidance test with a memory probe trial . At baseline, before being exposed to an aversive stimulus, neither of the gene-manipulated mice preferred any of the four corners (Figure 4E; left). During training, the mice avoided the corner in which they were exposed to an aversive stimuli (an air-puff), but there was no genotype-dependent difference (training t(18) = 1.07, p = 0.30; probe trial t(18) = 0.35, p = 0.73; chance level is 25%; Figure 4E; middle). In the probe trial, both E10+/− and E10−/− mice avoided the corner where they had previously been exposed to an aversive stimuli (training t(38) = 11.9, p < 0.001; probe trial t(38) = 9.84, p < 0.001; chance level is 25%; Figure 4E; right). These findings were indicative of learning and memory retention abilities in both groups of gene-manipulated mice.
The cytoskeleton helps cells maintain a dynamic cellular morphology in response to the demands of the surrounding environment. Microtubule-binding proteins likely play a major role in cytoskeletal integrity which is highly critical to neuronal functions and viability, since it enables bidirectional axonal transport of vesicles and organelles along extended processes. In the adult human brain, splicing of tau protein gives rise to a 1:1 balanced expression of 3R- and 4R-tau mRNA, while in rodents only 4R-tau mRNA is expressed in adult animals. Interestingly, in mouse brain exon 10 is not utilized for protein synthesis until postnatal stages [11, 17], when the genome might be less able to activate compensatory mechanisms and mask phenotypes. This is the first study in which the splicing pattern of tau, a microtubule-binding protein, has been altered in the genome with the purpose of studying physiological functions of exon 10 in vivo. Gene targeting is typically associated with the creation of knockout mice, often generated by ablating the basal promoter and exon 1. There are a few reports in which the technique has been used to solely eliminate a single exon . In the current study, equal amounts of 3R- and 4R-tau protein were synthesized in 2 months-old E10+/− mice while wild-type littermates only expressed 4R-tau protein. Moreover, immunostaining with a 3R-tau selective antibody confirmed axonal localization of tau in genetically manipulated mice (E10+/−). Altogether, gene targeting technique led us to the successful generation of gene-manipulated mice expressing 3R:4R-tau (E10+/−) and 3R-tau (E10−/−) mRNA and protein with expected subcellular localization.
Functional studies revealed an age-dependent decline of sensorimotor skills only in middle-aged E10−/− mice. Adult and middle-aged E10+/− mice had similar motor skills in the rotarod test as E10+/+ mice, consistent with haploinsufficiency seldom giving phenotypes in knockout studies. Interestingly, adult (5–6 months) E10+/− mice were actually stronger than E10+/+ mice, which could be explained by activation of compensatory mechanisms. It has been suggested that other microtubule-associated proteins, MAP1A and MAP1B, are involved in functional compensation. Ablation of tau resulted in ~2-fold increased MAP1A levels in young mice (~2 weeks old) [2, 4], while levels of MAP1B and other MAPs remained unchanged. Compensatory mechanisms seem more active during embryogenesis and early postnatal development, since MAP1A was unchanged in 12 months-old tau knockout mice . We speculate that functional compensation in adult E10−/− mice was overridden by the effects of aging, resulting in sensorimotor deficits.
Cognitive and emotional functions in the gene-manipulated mice were examined, partly because tau anomalies are related to dementia disorders. We decided to restrict IntelliCage analyses to two groups of mice; E10+/− and E10−/− because we wanted to attain a reasonable statistical power. The decision was also due to availability of IntelliCages and female mice well matched for age. Moreover E10+/− and E10+/+ mice had behaved essentially the same in the sensorimotor tests. These observations were consistent with many other knockout studies, in which haploinsufficiency seldom results in phenotypes. E10−/− mice were less active than E10+/− mice in the IntelliCages, observations which were not reflected in differential anxiety-like behaviours or diminished exploratory behaviour when animals were tested in open field or elevated plus maze. In IntelliCages, mice are housed for days undisturbed in social groups in an environment to which they are allowed to habituate. The procedures of open field and elevated maze testing are markedly different, and this can of course have an impact on the outcome. Both groups habituated well to the new environment and did not show overt cognitive dysfunctions in IntelliCages. We conclude that deletion of exon 10 in tau does not result in cognitive or emotional phenotypes, although it remains to be investigated if this occurs with aging.
Taken together phenotypes in mice genetically manipulated around exon 10 in tau were limited to sensorimotor defects and partly similar to those found in tau knockout mice [3, 19]. Impaired performance might relate to cerebellar dysfunctions, since this brain region to a large extent develops at postnatal stages when splicing of tau-mRNA is abnormal in E10+/− and E10−/− mice . We were unable to find gross morphological changes, but there could be subtle differences in neural tree or synaptodendritic connections leading to network dysfunctions. Electrophysiological and high-resolution quantitative morphology would be needed to investigate such changes, studies which we consider to be outside the scope of an initial characterization of the models.
There is evidence to suggest that in tauopathies, the pathological aggregation of tau can be primary pathogenic e.g. in frontotemporal lobe dementia (FTLD-17) or secondary pathogenic e.g. to Aβ-aggregation in AD . However, it remains unclear why human Aβ accumulation by itself does not generate neurofibrillary tangles in AβPP transgenic mice. Perhaps structural differences make murine tau unable to form fibrils but in vitro experiments contradict this theory . Murine tau can take part in tau aggregation in vivo but so far seems unable to initiate tau pathology by itself [23–25]. An unknown factor or a trigger, such as misfolded or aggregated human tau or perhaps Aβ seems to be critical. The differences in splicing of tau in mouse and human brain, i.e. the 1:1 balanced 3R-/4R-tau mRNA splicing pattern, could directly impact on the stability of the microtubule structure and on aggregation of tau. It remains to be investigated whether the microtubule system of E10+/− and E10−/− mice is more vulnerable to a trigger of tau-misfolding e.g. an aggregate of human Aβ or human tau than wild-type mice harbouring intact murine tau (E10+/+).
This is the first functional study of exon 10 in tau, which is linked to neurodegenerative disease. By deleting exon 10 it was possible to alter alternative splicing of tau such that 3R-tau protein was synthesized instead of 4R-tau protein. The anatomic and subcellular localization of tau protein synthesis was maintained in the gene-manipulated animals. Middle-aged mice lacking exon 10 in tau (E10−/−), expressed only 3R-tau protein and displayed mild sensorimotor deficits with reduced grip strength and motor coordination. Haploinsufficient mice (E10+/−), expressing both 3R- and 4R-tau protein, behaved nearly always equal to wild-type mice (E10+/+) which expressed only 4R-tau protein. There were no genotype-dependent effects on cognition, emotionality or on gross brain structure and tissue morphology. Thus, sensorimotor functions depend on tau exon 10, while perhaps additional functions are masked by related microtubule-binding proteins.
Appropriate experimental procedures were approved by the committee of ethical conduct in research on animal at the Uppsala University (ethical permits C17/7 (2007-03-02 – 2010-03-02), C374/9 (2010-01-29 – 2013-01-29) and C278/12 (2012-12-20 – present) and performed according to guidelines of ethical conduct on science in compliance with local animal care.
A 2.2-kbp genomic fragment in intron 9 of the murine tau gene was amplified with genomic DNA from R1 embryonic stem (ES) cells as template using high-fidelity Expand PCR system (Roche, Basel, Switzerland). Primers started at 15718139 and 15720351 respectively in NT_165773.2. Likewise a 3.6-kbp fragment stretching from intron 10 into intron 11 was amplified with primers starting at 15720889 and 15724540 in NT_165773.2. Primers were designed as to avoid ~200 bp intron sequences located in introns 9 and 10, immediately adjacent to exon 10. Both genomic fragments were inserted into the pBluescript II KS-vector where they flanked a 2 kbp fragment containing a neomycin gene surrounded by loxP-sites that had been excised from the pKK7-vector. The targeting construct was checked by multiple restriction enzyme digests and fully sequenced. Only a few mismatches were found as compared to the NCBI database. None of those were located in or adjacent to protein coding sequences. The targeting construct was introduced into R1 ES cells by electroporation and homologous recombination was verified in 7 out of 375 individual clones by screening with two long-range PCR reactions. In both reactions, one primer annealed to sequences outside the targeted allele (starting at 15717864 and 15724785 respectively) while the other primer annealed to sequences within the neomycin (neo) gene. Male chimeras were obtained by microinjection of ES cell clones into blastocysts and subsequent transplantation of these blastocysts to CD1 pseudopregnant mice. The blastocysts were generated by mating with C57BL/6NCrl females and B6D2F1/Crl males, Male chimeras were bred with female C57Bl/6JBomTac mice (Taconic, Hudson, NY, USA) as to generate female offspring, which were then bred with male transgenic mice expressing phosphoglycerate kinase (pgk)-CRE recombinase (BALB/c × C57B1/B6)F1 . The loxP-flanked neo-cassette was thereby deleted from the targeted allele in some of the offspring.
Genomic DNA from ear-biopsies was prepared as previously described . PCR amplification was done with primers annealing at 15720175 and 15721073 in NT_165773.2 as to frame exon 10 in the murine tau gene and to generate ~900 or ~460 bp bands depending upon the inclusion or exclusion of exon 10. Alternatively the targeted allele was identified by using with one primer located in the ~100 bp sequence adjacent to the remaining loxP-site, and therefore unique to the targeted allele, and the other primer in nearby intron sequences of the murine tau gene. Total RNA was purified with RNA STAT-60 (Tel-Test, Friendwood, TX, USA), which is based on the guanidine thiocyanate method. Universal Riboclone cDNA synthesis system (Promega, Madison, WI, USA) and an antisense primer located at 1281–1262 in NM_010838.3 were used for first strand synthesis. The nascent cDNA was then PCR amplified with primers which annealed at 671–690 (in exon 9) and 975–954 (in exon 11) in NM_010838.3 as to simultaneously assay 3R-tau and 4R-tau mRNA expression. Alternatively cDNA was amplified with primers located at 671–690 (in exon 9) and 893–876 (in exon 10) whereby only 4R-tau mRNA was detected. The PCR products were separated by electrophoresis on ethidium bromide-stained agarose gels, and examined and photographed under UV-light.
Mice were anesthetized with 0.4 ml Avertin (25 mg/ml) and intracardially perfused with 0.9% (w/v) saline solution. Brains were dissected and either directly frozen on dry ice for western blot analyses, or instantly frozen in 2-methylbutane at −30°C with dry ice and used for immunohistochemistry. All tissues were stored at −80°C until use. Sections, 14 μm, were cut in a cryostat, fixed in 4% formalin for 5 min, permeabilized in 0.4% Triton X-100 for 5 min and treated with 0.3% H2O2 for 15 min to eliminate endogenous peroxidase activity. Tissues were then incubated with DAKO-block for 1 h to block unspecific signal and then with primary antibodies 3RD (x40) and Tau-1 (×200) (Millipore, Billerica, MA, USA) over night at +4°C. Sections were then washed and treated with secondary antibodies, ABC-reagents and developed with NOVA RED as described . For Cresyl violet and Luxol fast-blue staining, mice were intracardially perfused with 4% (w/v) formaldehyde solution and neuronal tissues were dissected and immersed in this solution for 3 days, dehydrated in etOH (40% 2×4 h; 70% o.n.; 80% 2×1.5 h; 95% 2×1.5 h; 100% 2×1 h; Xylen 2×30 min), immersed in paraffin o.n at 62°C, embedded and sectioned at 5 μm. Brains used for western blot were extracted in 10 mM Tris, 0.8 M NaCl, 1 mM EDTA, 10% (w/v) sucrose supplemented with Complete® (1:50) protease inhibitor cocktail (Roche, Basel, Switzerland) and centrifuged at 26 000 × g for 20 min at +4°C. The pellet was extracted and centrifuged again at 26 000 × g for 20 min at +4°C. The supernatants were pooled and supplemented with 1% (w/v) sarkosyl and incubated by rotary shaking at 200 RPM for 1 h at RT. The mixture was then centrifuged at 100 000 × g for 1 h at +4°C and the sarkosyl-soluble supernatants were run on 8% Tris-Glycine gels (InVitrogen, Carlsbad, CA, USA) and analyzed with western blot as previously described . A human recombinant tau protein ladder was used as size markers (rPeptide, Bogart, GA, USA). Nitrocellulose filters were incubated with primary antibodies 3RD (x20 000) and Tau-1 (x50 000) in 5% (w/v) non-fat dry milk in 1×TTBS for 1 h at RT.
Motor coordination of wild-type (E10+/+), E10+/− and E10−/− mice was assessed at 5–7 months and 13–17 months of age. The ROTA-ROD apparatus (dimensions 362 × 240 × 400 mm) (model LE8200, Panlab, S.L.U., Spain) consisted of a vertical plastic (methacrylate) rotating rod (dimensions 50 × 30 mm) with a ribbed surface flanked by two large discs (arnite). Each mouse was given three trials with 10 min inter-trial interval for three consecutive days. Trials started at a minimum speed of 4 rpm followed by 5 min acceleration time (maximum speed 40 rpm) with increase of speed with 1 rpm every 8th second. The latency to fall and the rotation speed when the mice fell was recorded. The apparatus was cleaned with 70% ethanol between trials.
Muscular strength of wild-type (E10+/+), E10+/− and E10−/− mice was evaluated at 5–7 months and 13–17 months of age. The grip strength meter (model GS3, BIOSEB, Chaville, France) consisted of a stainless steel grid (dimensions 100 × 80 mm) with a sensor capacity ranging 0–20 Newton (N). Each mouse was held by the tail, placed on the grid with all four paws and gently pulled horizontally by the tail along a straight line until the mice released the grid and the maximum forced was recorded. Each mouse was given three trials with 30 seconds inter-trial interval for four consecutive days. The stainless grid was cleaned with 70% ethanol between trials.
Behavioural analyses in the IntelliCage
Name of the module
1. Free exploration (FE)
3. Place learning (PL)
4. Reversal place learning (RevPL)
5. Extinction of place preference
6. Passive avoidance
Anxiety-like behaviours of 12–16 months-old wild-type (E10+/+), E10+/− and E10−/− mice were assessed in a maze made of grey coloured PVC walls (model LE842, Panlab S.L.U., Barcelona, Spain). It consisted of two open arms and two closed arms (length 29.5 cm and width 6 cm) arranged as a cross. The center was 6 × 6 cm and the closed arms were protected by a 15 cm wall made of grey PVC. Individual mice were placed at the center of the maze facing the open arms for a single 5 min trial. Their behaviours were recorded on video and analysed with Smart Junior program V1.0. The maze was cleaned with 70% ethanol between trials.
A transparent cage (width 37 cm, length 48 cm and height 20 cm) was used to assess locomotor activity, exploratory and anxiety-like behaviours of 12–16 months-old wild-type (E10+/+), E10+/− and E10−/− mice. Individual mice were placed at the center of the cage for 10 min and their behaviour was recorded with a video camera. The testing cage was divided in three zones: periphery, internal and center comprising 43.6%, 48.6% and 7.8% of the total area respectively. Off-line analysis of the records was performed with Smart Junior program V1.0. The area was cleaned with 70% ethanol between trials.
GraphPad (San Diego, CA, USA) and Statistica (Tulsa, OK, USA) were used. Data which had been verified not to deviate from a normal distribution were analyzed with unpaired t-test, one-way or two-way ANOVA, or otherwise with a non-parametric test e.g. Mann–Whitney U-test. Outliners were identified with Grubbs’ test, and post-hoc test of one-way and two-way ANOVA were analyzed with Fisher LSD. Results presented are mean ± s.e.m. unless otherwise stated.
Amyloid-β precursor protein
Cornu ammonis 1
Frontotemporal lobe dementia (FTLD-17)
Glial fibrillary acidic-protein
Light emitting diodes
National center for biotechnology information
Reverse transcriptase polymerase chain reaction
Postnatal day 1
This work was funded by grants from Uppsala University, Landstinget i Uppsala län, the Swedish Research Council [#2009-4567) LL, (#2009-4389) LN], The Swedish Brain Foundation, Alzheimerfonden, Gamla Tjänarinnor, Gun och Bertil Stohne (AG, LN), Magnus Bergvall, Åhlénsstiftelsen, Lundströms Minne, Frimurarstiftelsen, Trolle-Wachmeisters stiftelse (LN). Klas Kullander is greatly acknowledged for giving the pKK7-vector and Åsa Mackenzie for providing access to pgk-CRE recombinase transgenic mice. The Uppsala University Transgenic Facility (UUTF) is greatly acknowledged for their support in developing the E10+/− and E10−/− models. Micaela Vrede is acknowledged for help with functional studies, Erik Sundström and Lena Holmberg for dissecting spinal cord, and Kristina Magnusson for photography.
This article is published under license to BioMed Central Ltd. This is an open access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.