doi:10.1093/brain/awy193 BRAIN2018:Page1of15 | 1 Induction of a transmissible tau pathology by traumatic brain injury Elisa R. Zanier,1 Ilaria Bertani,1 Eliana Sammali,1,2 Francesca Pischiutta,1 Maria Antonietta Chiaravalloti,1 Gloria Vegliante,1 Antonio Masone,1 Alessandro Corbelli,3 Douglas H. Smith,4 David K. Menon,5 Nino Stocchetti,6,7 Fabio Fiordaliso,3 Maria-Grazia De Simoni,1 William Stewart8,9 and Roberto Chiesa1 Traumatic brain injury is a risk factor for subsequent neurodegenerative disease, including chronic traumatic encephalop- athy, a tauopathy mostly associated with repetitive concussion and blast, but not well recognized as a consequence of severe traumatic brain injury. Here we show that a single severe brain trauma is associated with the emergence of widespread hyperphosphorylatedtaupathologyinaproportionofhumanssurvivinglateafterinjury.Inparallelexperimentalstudies,in a model of severe traumatic brain injury in wild-type mice, we found progressive and widespread tau pathology, replicating the findings in humans. Brain homogenates from these mice, when inoculated into the hippocampus and overlying cerebral cortex of na¨ıve mice, induced widespread tau pathology, synaptic loss, and persistent memory deficits. These data provide evidence that experimental brain trauma induces a self-propagating tau pathology, which can be transmitted between mice, and call for future studies aimed at investigating the potential transmissibility of trauma associated tau pathology in humans. 1 Department ofNeuroscience, Istituto diRicerche FarmacologicheMarioNegri IRCCS, 20156Milan, Italy 2 Department ofCerebrovascular Diseases,Fondazione IRCCS – Istituto Neurologico Carlo Besta,Milan, Italy 3 Department ofCardiovascular Research, Istituto di Ricerche FarmacologicheMarioNegri IRCCS, 20156Milan, Italy 4 Penn Centre forBrainInjuryand Repair andDepartment ofNeurosurgery, Perelman School ofMedicine,University of Pennsylvania, Philadelphia, PA19104,USA 5 Division ofAnaesthesia, Department ofMedicine, UniversityofCambridge, Cambridge CB22QQ, UK 6 Department ofPathophysiology andTransplants, University ofMilan, 20122Milan, Italy 7 Fondazione IRCCS Ca` GrandaOspedale Maggiore Policlinico di Milano, 20122Milan, Italy 8 Institute ofNeuroscience andPsychology, University ofGlasgow, Glasgow G128QQ, UK 9 Department ofNeuropathology, Queen Elizabeth UniversityHospital, Glasgow G514TF, UK Correspondence to: Elisa R.Zanier Department ofNeuroscience, Istituto di Ricerche Farmacologiche MarioNegri IRCCS,Via G.LaMasa19,20156 Milan, Italy E-mail: [email protected] Correspondence mayalso beaddressed to: Roberto Chiesa E-mail: [email protected] Keywords: traumatic braininjury; taupathology; prion;neurological dysfunction; neurodegeneration Abbreviations: CTE = chronic traumatic encephalopathy; P-tau = hyperphosphorylated tau; TBI = traumatic braininjury ReceivedJanuary15,2018.RevisedMay28,2018.AcceptedJune6,2018. (cid:2)TheAuthor(s)(2018).PublishedbyOxfordUniversityPressonbehalfoftheGuarantorsofBrain.Allrightsreserved. Forpermissions,pleaseemail:[email protected] Downloaded from https://academic.oup.com/brain/advance-article-abstract/doi/10.1093/brain/awy193/5062642 by Open University Library (PER) user on 01 August 2018 2 | BRAIN 2018: Page 2 of 15 E. R. Zanier et al. Introduction inducedtaupathologycouldbetransmittedbetweenmiceand cause neurological deficits, satisfying the requirements for a Traumatic brain injury (TBI) is a recognized risk factor for bona fide tau prion. delayed neurodegeneration and dementia, particularly chronic traumatic encephalopathy (CTE). The neuropathological picture that is emerging with late survival from TBI is of a Materials and methods complex of pathologies featuring abnormalities in tau, amyl- oid-beta, neuronal loss, axonal degeneration, neuroinflamma- Study design tionandblood–brainbarrierdisruption(Johnsonetal.,2012, 2013; Hay et al., 2016). Of particular note is the emergence The aim of the study was to test whether a single severe TBI of hyperphosphorylated forms of the microtubule-associated was able to induce a self-propagating tau pathology. First, we protein tau in a stereotypical pattern and distribution re- analysed the deposition of P-tau late after single TBI in garded as pathognomonic of CTE (McKee et al., 2016). humans. Next, we characterized the temporal evolution of Specifically, in its early stage, CTE is recognized through the tau pathology in wild-type mice subjected to severe controlled demonstration of clusters of perivascular neurons and glia corticalimpact.Finally,wetestedwhetherauthentictauprions containing hyperphosphorylated tau (P-tau), typically with are generated after trauma by attempting the transmission of tauopathy to naı¨ve mice by intracerebral inoculation of brain patchy distribution and concentrated towards the depths of homogenates from TBI mice. cortical sulci (Geddes et al., 1999; McKee et al., 2014, 2016; Male C57BL/6J mice (8 to 9 weeks of age) from Envigo Mez et al., 2017). There is a suggestion of hierarchical pro- were used in this study. The animals were housed four per gression of pathology from these initially focal and patchy cage at controlled temperature (22(cid:2)2(cid:3)C) and relative humid- cortical P-tau deposits to more widespread cortical, medial ity (60(cid:2)5%) with a 12/12-h light/dark cycle and free access temporal and subcortical grey matter in late-stage disease to pelleted food and water. (McKee et al., 2014). However, the mechanisms by which Allanimalexperimentsweredesignedinaccordancewiththe this initially localized cortical pathology progresses to more ARRIVE guidelines (Kilkenny et al., 2010), with a commit- widespread and generalized disease are still not known. ment to refinement, reduction, and replacement, minimizing In Alzheimer’s disease, tau pathology appears to progress the numbers of mice, while using biostatistics to optimize along neural networks, suggesting transcellular spread of tau mouse numbers [as in our previously published work using the mouse TBI model (Zanier et al., 2016) and prion inocula- through neural connections (Goedert et al., 2017). In vivo tion in mice (Bouybayoune et al., 2015)]. Thus, for statistical studies have described the spreading of Alzheimer’s disease- validity, we used 8 to 12 mice for the behavioural tests, and related tau pathology from local sites to distant regions of three toeightmice forbiochemical analysis andhistology. The the brain (Clavaguera et al., 2009, 2013; de Calignon et al., mice were assigned to the different experimental groups using 2012; Ahmed et al., 2014; Guo et al., 2016), and it has been randomization lists (www.randomizer.org). Inoculations, be- suggested that tauaggregates—or seeds—serve as the agent of havioural and histological assessments were done by re- this spread, transmitting the aggregated/hyperphosphorylated searchers blinded to the experimental groups. state from cell to cell in a prion-like fashion (Prusiner, 2012; Clavaguera et al., 2013; Iba et al., 2013, 2015; Woerman et al., 2016). Prions are proteinaceous infectious agents con- Ethics statement sisting of misfolded prion protein (PrPSc) that propagate by Human tissue was obtained from the Glasgow TBI Archive of imposing their abnormal structures on host-encoded cellular theDepartmentofNeuropathology,QueenElizabethUniversity prionprotein(PrPC)(Prusiner,1982).AsPrPScprionscolonize Hospital,Glasgow,UK.Tissuesampleswereacquiredatroutine the affected patient, they perturb the function of the nervous diagnostic autopsy, with approval from the Greater Glasgow system, leading to motor and neurophysiological deficits, cog- and Clyde Biorepository Governance Committee for their use nitive decline, and ultimately death (Chiesa, 2015). Whether in research. Procedures involving animals and their care were TBI generates misfolded or aggregated forms of tau protein conducted in conformity with the institutional guidelines at the that propagate by inducing misfolding of native tau in injured IRCCS–MarioNegriInstituteforPharmacologicalResearchin tissue and can be transmitted from one individual to another compliance with national (D.lgs 26/2014; Authorization n. 19/ (tauprions),isnotclear(Prusiner,2012;Mudheretal.,2017). 2008-AissuedMarch6,2008byMinistryofHealth)andinter- national laws and policies (EEC Council Directive 2010/63/UE; We hypothesized that late P-tau deposition after severe TBI the NIH Guide for the Care and Use of Laboratory Animals, wouldshowprionbehaviour,andaddressedthishypothesisin 2011 edition). They were reviewed and approved by the Mario sequential steps. First, we set out to demonstrate and charac- Negri Institute Animal Care and Use Committee that includes terize the deposition of P-tau late after severe TBI as distinct ad hoc members for ethical issues, and by the Italian Ministry from P-tau deposition in ‘normal’ aged human controls. ofHealth(Decretono.D/07/2013-Band301/2017-PR).Animal Second, we sought the emergence of self-propagating P-tau facilities meet international standards and are regularly checked pathology inwild-type mice subjectedtosevere TBI that repli- by a certified veterinarian who is responsible for health moni- cated P-tau pathology observed in humans long after a single toring, animal welfare supervision, experimental protocols and severe TBI. Finally, we went on to establish whether this TBI- review of procedures. Downloaded from https://academic.oup.com/brain/advance-article-abstract/doi/10.1093/brain/awy193/5062642 by Open University Library (PER) user on 01 August 2018 Brain trauma induces tau prion BRAIN 2018: Page 3 of 15 | 3 Human studies tangles and neurites of sparse to moderate density in the trans- entorhinal cortex with or without small numbers in the CA1 At the time of the original diagnostic autopsy, whole brains sector of the hippocampus; score 2, PHF1-immunoreactive were immersion-fixed in 10% formal saline for a minimum of neurofibrillary tangles and neurites as described for score 1 to- 3 weeks, then the specimens were examined, sampled and pro- gether with neurofibrillary tangles extending into the fusiform cessed to paraffin tissue blocks. For this study, 15 patients sur- gyrus (lateral to entorhinal cortex) and sparse tangles in the viving a year or more from single moderate or severe, closed cortex (cingulate/insular blocks); and score 3, contained TBI (defined as Glasgow Coma Scale at presentation 513) and PHF1-immunoreactive neurofibrillary tangles and neurites in 15 age-matched controls with no history of TBI or of known more widespread sectors of the hippocampus and subiculum, neurological disease were randomly selected from the compre- with extensive medial temporal, cingulate and insular cortical hensive databases of the Glasgow TBI Archive by a laboratory involvement. researchassistantwithnoknowledgeofstudypurposeordesign or of the pathologies in each of the TBI cases and controls Mouse model of traumatic brain selected. Full demographic information for late TBI cases and controls is provided in Supplementary Table 1. Information on injury TBIandcontrols’historywasobtainedfromreviewofdiagnos- Thecontrolledcorticalimpactbraininjurymousemodelusedin tic post-mortem and/or forensic reports. The TBI survival cases this study replicates both the mechanical forces and the main andtheage-matchedcontrolshadnorecordedhistoryofexpos- secondary injury processes observed in severe TBI patients with ure to contact sports or military service. The mechanism of brain contusionandisassociated with clinically-relevant behav- injury is summarized in Supplementary Table 1. None were ioural and histopathological outcomes (Smith et al., 1995; penetrating. From review of the original hospital records, at Pischiutta et al., 2018). Mice were anaesthetized by isoflurane presentation, intracerebral haemorrhage was present in five inhalation (induction 3%; maintenance 1.5%) in an N O/O cases, subdural a further seven, with three cases having no evi- 2 2 (70%/30%) mixture and placed in a stereotaxic frame. Rectal dence of intracranial haemorrhage. Following the original TBI, temperature was maintainedat 37(cid:3)C.Mice were then subjected all cases were discharged from hospitalization following recov- to craniectomy followed by induction of controlled cortical ery and,ultimately,diedfrom causesofdeathunrelatedtoTBI. impact brain injury as previously described (Zanier et al., None were in a persistent vegetative state prior to death. Each 2016). Briefly, the injury was induced using a 3-mm rigid im- case was sampled and assessed for this study as standardized pactor driven by a pneumatic piston rigidly mounted at an regions, no matter the injury severity or location. As this study angleof20(cid:3) fromtheverticalplaneandappliedto the exposed accesses archival tissue blocks, there were necessary restrictions dura mater, between bregma and lambda, over the left parieto- on availability of sampled regions. Nevertheless, the standar- temporal cortex (antero-posteriority: (cid:4)2.5mm, laterality: dizedsamplingprotocolsusedattheoriginaldiagnosticautopsy (cid:4)2.5mm), at an impactor velocity of 5m/s and deformation includedacoronalsliceofthecerebralhemispheresatmid-thal- depth 1mm, resulting in a severe level of injury (Smith et al., amiclevel from whichtissueblocks frommultiplebrainregions 1995). The craniotomy was then covered with a cranioplasty were selected for the purposes of this study to include: the and the scalp sutured. Sham mice received identical anaesthesia corpus callosum with adjacent cingulate gyrus; the medial tem- and surgery without brain injury. poral lobe to include the hippocampus and entorhinal cortex; the insular cortex and the thalamus. From these tissue blocks 8mm sections were prepared for immunohistochemistry proced- Transmission studies ures. After deparaffinization and rehydration, endogenous per- oxidase was quenched by immersing sections in 3% aqueous Brainhomogenates(10%w/vinPBS)fromtheipsi-andcontra- H O for15min.Thereafter,heat-inducedantigenretrievalwas lateralcerebralcortex(includingallthecorticaltissueabovethe 2 2 performed using a microwave pressure cooker for 8min in pre- rhinalfissure)ofmice12monthsfromTBIorshaminjurywere heated0.1MTrisEDTAbuffer(pH8),followedbyblockingin inoculated in the hippocampus (A/P, (cid:4)2.5mm from bregma; L, 50ml of normal horse serum (Vector Labs) per 5ml of (cid:2)2.0mm; D/V, (cid:4)1.8mm) and overlying cerebral cortex (A/P, OptiMaxTM buffer (BioGenex) for 30min. The sections were (cid:4)2.5mm from bregma; L, (cid:2)2.0mm; D/V, (cid:4)0.8mm). Inocula then incubated in antibody to phosphorylated tau (PHF1; were from single brains. The TBI brain homogenates showing 1:1000; gift from P. Davies, Feinstein Institute for Medical the highest amounts of P-tau by western blot were chosen for Research,Manhasset, NY,USA)at 4(cid:3)C for20hpriortowash- inoculation.Micewereinjectedbilaterally(2.5mleach,injection ing and application of biotinylated secondary antibody for speed of 1.25ml/min). 30min,followedbyanavidinbiotincomplex,asperthemanu- facturer’s instructions (Vectastain Universal Elite Kit, Vector Novel object recognition Labs). The target antigen was visualized using the DAB perox- idase substrate kit (Vector Labs), with the sections then coun- Mice were tested in an open-square grey arena (40(cid:5)40cm), terstained with haematoxylin. Control sections (positive and 30cm high, with the floor divided into 25 squares by black negative) for each antibody were run in parallel with the test lines. The following objects were used: a black plastic cylinder sections to confirm antibody specificity. Sections were viewed, (4(cid:5)5cm), a glass vial with a white cap (3(cid:5)6cm) and a blindtowhethertheywereTBIorcontrol,usingaLeicaDMRB metal cube (3(cid:5)5cm). The task started with a habituation light microscope (Leica Microsystems) and the extent and dis- trial during which mice were placed in the empty arena for tribution of tau pathology was assessed and semiquantitatively 5min and their movements were recorded as the number of scoredas:score0,wherenotaupathologywaspresent;score1, line-crossings. The next day, mice were placed in the same PHF1 pathology comprising immunoreactive neurofibrillary arena containing two identical objects (familiarization phase). Downloaded from https://academic.oup.com/brain/advance-article-abstract/doi/10.1093/brain/awy193/5062642 by Open University Library (PER) user on 01 August 2018 4 | BRAIN 2018: Page 4 of 15 E. R. Zanier et al. Exploration was recorded in a 10-min trial by an investigator positive pixels/total pixels, and reported as the percentage blinded to the experimental group. Sniffing, touching and stained area for statistical analysis. stretching the head toward the object at a distance of no For immunofluorescence, brain sections on glass slides were more than 2cm were scored as object investigation. Twenty- boiledincitratebufferandincubatedinblockingsolutionwith four hours later (test phase) mice were placed in the arena the following primary antibodies: anti-drebrin (Fitzgerald, containing two objects: one identical to one of the objects 1:200) an actin-binding protein enriched in dendritic spines, presented during the familiarization phase (familiar object), (Sekino et al., 2017) and anti-vesicular glutamate trans- and a new, different one (novel object), and the time spent porter-1 (VGLUT-1, Synaptic System, 1:6000), for analysis exploring the two objects was recorded for 10min. Memory of the presynaptic compartment (Senatore et al., 2012). Slices was expressed as a discrimination index, i.e. the time spent were then incubated with biotinylated secondary antibody exploring the novel object minus the time spent exploring (Vecstain, 1:500) and streptavidin 647 (Molecular Probes, the familiar one, divided by the total time spent exploring 1:500), and examined by confocal microscopy (FV-500 both objects; the higher the discrimination index, the better Olympus) acquiring images in two different areas of CA1 the performance. field of the hippocampus using a 60(cid:5) oil objective. Z-stack images were acquired: 0.2mm Z-distance, 3-mm thick confocal z-stacks (optical section depth 0.2mm, 21 sections/z-stack) and Immunohistochemistry and 1024(cid:5)1024 pixel resolution. Confocal images were deconvo- immunofluorescence luted (CellP) and puncta were identified with Imaris software using the ‘quality’ parameter (Heise et al., 2016). Only puncta Mice were euthanized by cervical dislocation, brains were stringently conforming with the quality parameter entered removed and fixed in 10% formalin for 24h, dehydrated in analysis. graded ethanol solutions, cleared in xylene, and embedded in paraffin. Serial sections (5mm thick) were cut onto glass slides and boiled in citrate buffer (10mM pH 6.0, Dako) for antigen Biochemical analysis retrieval, followed by incubation with the following anti-tau monoclonal antibodies: AT8 (phospho-tau Ser202/Thr205; Frozen brain tissue was homogenized in 10 volumes of phos- Thermo Fisher; 1:1000), AT180 (phospho-tau Thr231; Thermo phate-bufferedsaline(PBS).Homogenateswerealiquotedand Fisher; 1:1000), PHF1 (Ser396/Ser404; from P. Davies; 1:4000) stored at (cid:4)80(cid:3)C until use. Western blot analysis of total andMC1(aa312–322;fromP.Davies;1:100).AT8andAT180 brain homogenates with monoclonal antibody AT8 was immunostaining was visualized using the ARK kit (Dako). For done on heat-stable protein fractions prepared according to secondarystainingofMC1andPHF1,sliceswereincubatedwith Petry et al. (2014) to avoid the interfering signal due to en- biotinylated goat anti-mouse IgG1 in Superblock Solution dogenous immunoglobulins. Proteins (50mg) were separated (Pierce), followed by visualization with the TSA Biotin System onNuPAGE4–12%gels(ThermoScientific)andblottedonto (Perkin Elmer) and DAB (diaminobenzidine) Substrate nitrocellulose membranes (Hybond, GE Healthcare). After Chromogen System (Dako). blocking with 5% dry milk for 1h at room temperature, Slices were examined by an Olympus BX-61 Virtual Stage membranes were incubated overnight at 4(cid:3)C with anti-total microscope, interfaced with VS-ASW-FL software (Olympus). tau(1:10000,Dako) orAT8(1:500,ThermoScientific),then For quantification of the AT8-positive staining in TBI mice with HRP-conjugated secondary antibodies (Santa Cruz). regions of interest were selected according to the Paxinos Peroxidase activity was detected using an ECL (Iluminata atlas (Paxinos and Franklin, 2004) in coronal (bregma Forte) kit (Millipore) and visualized with a ChemiDoc ima- (cid:4)1.6mm) and sagittal (lateral 1.5mm) sections. In TBI mice ging system (Bio-Rad). For RIPA insolublepreparation 100ml the ipsilateral cortex was analysed over an area 1mm deep of brain homogenate were incubated for 20min on an orbital from the edge of the contusion (Pischiutta et al., 2018) and shaker with 1% RIPA buffer (50mM Tris-HCl, pH 7.4, 1% the contralateral from midline to the entorhinal cortex. In the NP-40, 150mM NaCl, 0.25% Na-deoxycholate, 1mM inoculated mice, regions of interest included whole cortex, EDTA, 1mM Na VO , 1mM NaF, 1mM PMSF and prote- 3 4 hippocampus, thalamus and cerebellum. The same regions of ase inhibitors). Brain extract was ultracentrifuged for 60min interest were selected in sham mice. Images were acquired at100000gatroomtemperature. Thepelletwasresuspended using a 60(cid:5) oil objective with pixel size 0.115mm. in 450ml of RIPA buffer and ultracentrifuged at 100000g for Acquisition was done over 5-mm thick stacks, with step size 30min at room temperature. The supernatant was carefully 1mm. The different focal planes were merged into a single removed and the pellet was resuspended in 100ml of 1(cid:5) stack by mean intensity projection to ensure consistent focus Laemmli sample buffer and proteins analysed by western throughout the sample. Images were analysed using Fiji soft- blot. ware (http://fiji.sc/Fiji). For assay of protease resistance brain lysates prepared in The digital separation of DAB staining from haematoxylin 0.05% TritonTM X-100 in PBS were centrifuged at 300g for signal was done by colour deconvolution (http://www. 5min. Proteins (20mg) in the supernatants were digested mecourse.com/landinig/software/cdeconv/cdeconv.html) and with pronase (0–10mg/ml; Sigma) in a final volume of the DAB channel was used for quantification. To ensure con- 20ml for 1h at 37(cid:3)C, then resolved by NuPAGE 4–12% sistency of the segmentation in the image processing, we sub- gels and analysed by western blot with primary anti-tau tracted the intensity background from the image calculated by antibody BR135 (1:1000, gift from M. Goedert, MRC averaging non-specific signal in identified region of interest. Laboratory of Molecular Biology, Cambridge, UK; Goedert Specific signal after the image processing was computed as et al., 1989). Downloaded from https://academic.oup.com/brain/advance-article-abstract/doi/10.1093/brain/awy193/5062642 by Open University Library (PER) user on 01 August 2018 Brain trauma induces tau prion BRAIN 2018: Page 5 of 15 | 5 Electron microscopy more,PHF1-immunreactiveprofilesappearedasneurofibrillary tangles and immunoreactive neurites, astrocytes and plaques Three controls and three TBI inoculated mice were deeply within all cortical layers, although often with higher density anaesthetized and perfused through the ascending aorta with in the superficial cortical layers (Fig. 1A). Elsewhere, where PBS (0.1M; pH 7.4) followed by a 5-min perfusion with 4% hippocampal profiles were present, all sectors appeared paraformaldehyde.Hippocampuswasexcisedandcutinsagittal involved, CA2 less frequent than other sectors in this series. planewitharazorbladeandpostfixedin4%paraformaldehyde Sixcasesshowedinvolvementofstructuresintheregionofthe and2%glutaraldehydeinphosphatebuffer0.12M,thenfor2h thalamusassparseimmunoreactivethalamicornigralneurons, in OsO . After dehydration in graded series of ethanol, tissue 4 samples were cleared in propylene oxide, embedded in epoxy occasional neurites and immunoreactive astrocytes. Although medium (Epoxy Embedding Medium kit; Sigma-Aldrich) and the proportions of patients with PHF1-immunoractive tau polymerized at 60(cid:3)C for 72h. From each sample, one semi- pathology were similar in controls and TBI (12 cases with thin (1mm) section was cut with a Leica EM UC6 ultramicro- pathology in each group), the extent and distribution of the tome and mounted on glass slides for light microscopic inspec- pathologieswereconsiderablygreaterinsurvivorsofTBIthan tiontoidentifythestratumoriensofthehippocampus.Ultrathin in non-injured controls (Fig. 1B). Specifically, in controls, (70-nm thick) sections of the areas of interest were obtained, PHF1-immunoreactive neurofibrillary tangles and neurites counterstained with uranyl acetate and lead citrate, and exam- were largely restricted to sparse profiles in the entorhinal ined with an Energy Filter Transmission Electron Microscope cortex, with just four cases showing more extensive patholo- (Libra120, Carl Zeiss NTS) equipped with an yttrium alumin- gies extending to the hippocampus and cortex. In contrast, in ium garnet scintillator slow-scan charge-coupled device camera (Sharpeye,TRS).Aregionofinterestincludingthepostsynaptic late survivors of TBI, 9 of 12 PHF1-immunoreactive cases density was manually traced and its area and thickness mea- displayedhigherstagetaupathology,withPHF1-immunoreac- sured by iTem software (Olympus Soft Imaging Solutions, tive profiles in moderate to high density across the medial Germany) of 263 synapses in control inoculated mice and 262 temporal lobe, including the hippocampus, and extending to inTBIinoculatedmicefittingthefollowingtwocriteria:(i)pres- the cortex. Neuritic P-tau-positive profiles were found in the ence of a cluster of presynaptic vesicles (at least three of them); neuropil, many with a truncated, granular appearance as illu- and (ii) presence of a defined synaptic cleft. strated in Fig 1A. In 3 of 12 TBI cases with P-tau pathology, immunoreactive neuronal and glial profiles showed focal clus- Statistical analysis tering around cortical vessels, in two of these at the depths of sulci (Fig. 1A); appearances consistent with the pathogno- All graphs represent the mean(cid:2)standard error of the mean monic pathology of CTE described in the preliminary NIH (SEM). GraphPad Prism 7 was used for statistical analyses. diagnostic criteria (McKee et al., 2016). One further TBI A categorical assessment score was used to assess human tau deposition.Resultsweredichotomizedineither‘low’score(0–1) case, an 87-year-old male with a 5-year survival following a or ‘high’ score (2–3) and the chi-square test was used to assess roadtraffic accident, showedclustersof P-tauimmunoreactive differencesbetweengroups.Datainmicewereanalysedbyone- subpialastrocytesin the medialtemporal lobe, consistent with way ANOVA, followed by Tukey’s test to assess differences in ageing-relatedtauastrogliopathy(Kovacsetal.,2016).Bothin case of three or more groups (behavioural tests and temporal late TBI cases and controls the extent of P-tau pathology ap- changesoftaudeposition)orbyunpairedt-test.P-values50.05 peared to increase with increasing age; the higher score path- were considered to be statistically significant. ologyencounteredmoreoftenwithincreasingage.Specifically, the four controls with higher tau scores (2 or 3) were among Data availability the older aged (74 years or above) controls. Of note, while again typically olderwithinthe cohort, the nine late TBI cases The authors confirm that the data supporting the findings of this with higher tau scores ranged from a younger age (age 59 studyareavailablewithinthepaperanditsSupplementarymaterial. years upwards) than controls. No clear association between survival duration and extent of tau pathology could be ascer- Results tained in this small series (Supplementary Fig. 1). In keeping with injury severity, evidence of healed contusion was near ubiquitous; present in all but one case. There was no associ- Phosphorylated tau deposition is ationbetweenpresenceornature ofintracranialhaemorrhage, widespread in patients surviving a including contusions, and tau pathology. year or more from single TBI Single severe TBI induces a Representative tissue sections from the brains of patients sur- progressive tau pathology that viving a median of 5 years (range: 1–18 years) following a single moderate or severe, closed TBI [n=15; age 19–89 spreads from the site of injury (median 60) years], or control subjects with no history of in wild-type mice TBI or known neurological disease [n=15; age 20–92 (median 60) years] (Supplementary Table 1) were stained for We tested whether single focal brain trauma triggered propa- P-tauusingantibodyPHF1.InTBIpatientssurvivingayearor gating tau pathology in wild-type mice. Two-month-old Downloaded from https://academic.oup.com/brain/advance-article-abstract/doi/10.1093/brain/awy193/5062642 by Open University Library (PER) user on 01 August 2018 6 | BRAIN 2018: Page 6 of 15 E. R. Zanier et al. Figure1 TaupathologyinsurvivorsofasinglemoderateorsevereTBI.(A)Typically,controlsshowedsparseP-tauimmunoreactive neurofibrillarytanglesandneuriteslocalizedtotheentorhinalcortex(i;59-year-oldmale,nohistoryofTBI).Incontrast,patientssurvivingayear ormorefromsinglemoderateorsevereTBIshowedmoreextensiveneurofibrillary,neuriticandglialP-tauimmunoreactiveprofiles,with occasionalclustersaroundcorticalvesselsina‘CTE-like’manner(ii;59-year-oldmale17-yearsurvivalfromsinglesevereTBI),andextending beyondtheentorhinalcortextoinvolvewidergreymatterregionsincludingthehippocampus(iii;87-year-oldmale5-yearsurvivalfromsingle severeTBI).(B)WhiletheproportionofpatientsinTBIandcontrolswithtaupathologieswassimilar,followingTBItheextentanddistributionof pathologywereconsiderablygreater.Thus,usingasemiquantitativeschemetoscoretaudistribution,whileallbutfourcontrolsshowednoor onlylocalizedtaupathology(score0or1),inamajorityofTBIcases(9of15)P-taupathologywaswidespread(score2or3;P=0.0035;(cid:2)2).All sectionsstainedforPHF-1.Scalebar=100mm. C57BL/6J mice were subjected to severe TBI (16 mice) or to to injury, withconcentration tothe regions adjacent to injury sham injury (14 mice). Eight TBI and seven sham mice were [Fig. 2A (cf. i and ii), B, and Supplementary Fig. 2]. In con- sacrificed at 3 months and used for immunohistochemistry trast, at 12 months, bothipsi- andcontralateral P-tau pathol- (four TBI and four sham) or biochemical analysis (four TBI ogies were present, supporting progressive spreading of tau and three sham). One TBI mouse was found dead at pathologyfromtheinjuredhemisphere[Fig.2A–D(cf.Aiand (cid:6)4 months, and the brain was not collected due to prolonged Aiii;cf.CivandCvi)andSupplementaryFig.2].Westernblot post-mortem delay. The remaining mice were culled at analysis found increased levels of total and AT8-reactive tau 12 months and used for immunohistochemistry (four TBI inonefrozenTBIbraincomparedtosham-injuredcontrolsat andfivesham)andforbiochemistry(threeTBIandtwosham). 3 months and in two of three TBI brains at 12 months (an Phospho-tau immunopositivity, assessed by AT8, AT180 example is shown in Supplementary Fig. 3). Thus, three of and PHF1 antibodies, was evident in two of four mice at seven TBI mice ((cid:6)40%) developed pathological tau at 3monthspost-TBIandrestrictedtothehemisphereipsilateral 3 months, and six of seven ((cid:6)85%) at 12 months. Downloaded from https://academic.oup.com/brain/advance-article-abstract/doi/10.1093/brain/awy193/5062642 by Open University Library (PER) user on 01 August 2018 Brain trauma induces tau prion BRAIN 2018: Page 7 of 15 | 7 Figure2 EvidenceoftaupathologyinmouseTBI.AT8immunostainingandquantificationoftheAT8-positiveareaintheipsilateraland contralateralcortexofshamandTBImiceat3and12monthspost-injury.(AandB)AT8immunoreactivityisdetectedintheipsilateral(ipsi) pericontusionalcorticalareaintwooffourmiceat3monthspost-TBI,andinallmiceat12months.(CandD)InthecontralateralcortexAT8 immunoreactivityisdetectedonlyafter12monthspost-TBI.Dataaremean(cid:2)SEM;n=4–5;*P50.05;**P50.01byone-wayANOVA,Tukey posthoctest.Scalebar=100mm(BandD). P-tau was observed as neuronal cytoplasmic, neuritic and prion, causing brain dysfunction, we intracerebrally inocu- glial profiles throughout the cerebral cortex, hippocampus lated 10% brain homogenates from TBI and sham-injured (CA1andCA3 fields)andthalamus,with pathologyalso in mice into na¨ıve C57BL/6J mice. Brain homogenates of con- the striatum and zona incerta (Fig. 3). In addition, a not- tused and contralateral hemispheres (TBI and TBI , ipsi contra able observation was numerous small to medium sized, respectively) showing increased levels of total and AT8-posi- rounded, ‘grain-like’ profiles, particularly within the neuro- tive tau by western blot at 12 months post-TBI pil (Fig. 3C). Occasionally there was faint clustering of (Supplementary Fig. 3) were inoculated bilaterally into the these granular immunoreactive profiles (Fig. 3G), with hippocampus and overlaying cerebral cortex of the recipient still other examples showing apparent localization along mice (Fig. 4A). The mice were examined in the novel object cortical vessels (Fig. 3H). Elsewhere, glial cytoplasmic P- recognition test 8 and 12 months post-inoculation (m.p.i.). tau immunoreactivity was present within white matter of There was an overt memory deficit in the TBI - but not in ipsi the corpus callosum. the TBI -inoculated mice at 8m.p.i. (Fig. 4B). At 12 contra m.p.i the memory deficit persisted in the TBI - and was ipsi also detected in the TBI -inoculated mice (Fig. 4C). No Tau prions are generated in TBI contra memory deficit was seen in mice inoculated with sham- mice, causing transmission of injured brain tissue. tau pathology, synaptic loss and This was confirmed in an independent experiment in which we tested the effect of the inoculation of TBI brain behavioural deficits homogenate, obtained from the injured hemisphere The evidence that TBI induced a P-tau pathology that pro- 12 months post-injury, on recognition memory in the gressively spread throughout the brain suggested forma- inoculated animals at an earlier time (4m.p.i). Memory tion of a self-templating tau conformation. To test was already significantly impaired at 4 months, and this whether this could be horizontally transmitted like a was confirmed at 8m.p.i. (Fig. 5B and C). Locomotor Downloaded from https://academic.oup.com/brain/advance-article-abstract/doi/10.1093/brain/awy193/5062642 by Open University Library (PER) user on 01 August 2018 8 | BRAIN 2018: Page 8 of 15 E. R. Zanier et al. Figure3 Widespreadtaupathologyinmice12monthspost-TBI.ExamplesoftauimmunostainingwiththeAT180(AandF),PHF1(B), andAT8(C–E,GandH)antibodiesinthecerebralcortex(AandB),CA1(C)andCA3(D)fieldsofthehippocampus,andinthezonaincerta (EandF).Examplesofclustersofsmalltomediumsized,rounded,‘grain-like’profiles,inthezonaincerta(G),andgranularprofilesalonga corticalvessel(H;arrows).Scalebars:20mm(A–DandG);10mm(EandF);50mm(H). Downloaded from https://academic.oup.com/brain/advance-article-abstract/doi/10.1093/brain/awy193/5062642 by Open University Library (PER) user on 01 August 2018 Brain trauma induces tau prion BRAIN 2018: Page 9 of 15 | 9 Figure4 Inoculationofcontusedtissuesinducesmemorydeficitsinwild-typemice.(A)Groupsofmiceusedintheinoculation studies.C57BL/6Jmalemicewereeitherleftuntreated(openbar)orinoculatedbilaterallyintothehippocampusandoverlayingcerebralcortex with10%brainhomogenatesfromshammice(grey)orfromthecontused(TBI ,indigo)orcontralateral(TBI ,magenta)hemisphereat ipsi contra 12monthspost-TBI.(BandC)Recognitionmemorywasinvestigatedbythenovelobjectrecognition(NOR)testat8(B)and12(C)months afterinoculation.Performanceinthenovelobjectrecognitiontaskwasexpressedasdiscriminationindex(thehigherthediscriminationindexthe bettertheperformance).(DandE)Locomotoractivitywassimilarinna¨ıve,shamandTBIinoculatedmiceat8(D)and12(E)monthsafter inoculation.Dataaremean(cid:2)SEM;*P50.05,versusna¨ıve;#P50.05versusshambyone-wayANOVA,Tukeyposthoctest. activity was similar in sham and TBI inoculated mice, indi- with impaired synaptic efficacy. Analysis of P-tau deposition catingthatthedeficitinthenovelobjectrecognitiontestwas by antibody AT8 showed a significant difference between not due to motor abnormalities (Figs 4D, E, 5D and E). sham and TBI inoculated mice not only at the site of injec- The memory deficit was associated with hippocampal tion, but also widespread throughout the brain (Fig. 7). synaptic pathology. Analysis of pre- and postsynaptic com- Specifically, P-tau neurons, neurites and glia were identi- partments by quantitative VGLUT-1 and drebrin immuno- fied in the cerebral cortex, hippocampus (CA1, CA3, and fluorescence showed marked synaptic loss in the stratum hilus of the dentate gyrus) and thalamus, with morpholo- oriens of the CA1 region at 12m.p.i. (Fig. 6A and B). gies similar to those observed in mice exposed to TBI, Electron microscopy examination showed a significant de- including widespread ‘grain-like’ profiles (Fig. 8B–I). In the crease in postsynaptic density thickness and area in hippo- cerebellum, P-tau immunoreactive cytoplasmic and neuritic campal synapses in TBI inoculated mice (Fig. 6C) consistent inclusions were noted in association with Purkinje cells and Downloaded from https://academic.oup.com/brain/advance-article-abstract/doi/10.1093/brain/awy193/5062642 by Open University Library (PER) user on 01 August 2018 10 | BRAIN 2018: Page 10 of 15 E. R. Zanier et al. notable pattern and distribution. Further, a single severe TBI induced persistent and evolving tau pathology in wild-type mice, with characteristics similar to late post- TBI tau pathology in humans, which progressively spread from the site of injury to ipsi- and contralateral brain re- gions. This TBI generated tau pathology could be trans- mitted to wild-type recipient mice by intracerebral inoculation, inducing tau aggregation, memory deficits and synaptic alterations. These results indicate that tau prions are generated in TBI, providing a mechanistic ex- planation for how a biomechanical insult might trigger self-sustained neurodegeneration. In the past decade, there has been growing recognition that repetitive mild TBI might set in motion brain patho- logical processes that place exposed individuals at risk for neurodegenerative disorders in mid-to-late life, particularly CTE. While there is also evidence that tau pathology may beseenmanyyears afterasingle severe TBI,mostattention has focused on the association between repeated mild TBI and CTE, with the common impression tau pathology is uniquely associated with repetitive mild TBI, which dictates a distinct neurodegenerative disease. Herewe confirm that widespread CTE-like tau pathology is present in survivors of a year or more after just a single moderate to severe TBI in humans, and in mice exposed to severe TBI surviving up to 1 year from injury. In our series of 15 patients who survived a year or more post-TBI, 12 Figure5 Memorydeficitsarealreadydetectedat4 showed tau pathology, the extent and distribution of which monthspost-inoculation.(A)Schemeofthegroupsofmice usedintheinoculationstudies.C57BL/6Jmalemicewereeitherleft was considerably greater than that seen in age-matched, untreated(openbar)orinoculatedbilaterallyintothehippocampus non-TBI controls. Among this small series of late post- andoverlayingcerebralcortexwith10%brainhomogenatesfrom TBIpatientsthistaupathologyappearedtofollowastereo- sham(grey)orTBI(contusedhemisphere,indigo)mice12months typic pattern of deposition, reminiscent of hierarchical P- post-injury.(BandC)Recognitionmemorywasinvestigatedby tau deposition in other neurodegenerative diseases (Braak novelobjectrecognition(NOR)testat4(B)and8(C)months and Braak, 1995; Kovacs, 2015). These data provide one afterinoculation.Histogramsindicatethediscriminationindex(the possible neuropathological substrate for the increased inci- higherthediscriminationindexthebettertheperformance).(Dand dence of neurodegenerative disease in patients who survive E)LocomotoractivitywassimilarinshamandTBIinoculatedmice a single moderate or severe TBI. Although the extent and at4(D)and8(E)monthsafterinoculation.Dataaremean(cid:2)SEM, n=11–14.***P50.001,**P50.01versusna¨ıve;##P50.01versus distribution of P-tau pathology was greater in late TBI sur- vivors, it was not possible to establish a clear association sham;byone-wayANOVA,Tukeyposthoctest. between survival interval and extent of tau pathology. Further, our data in this limited human study show similar percentages of cases with tau pathology in TBI and control within the granule cell layer (Fig. 8J–K). Consistent with groups, making it hard to establish whether TBI accelerates widespread tau pathology, there was an increase in the a pre-existing age-related tau pathology, or triggers de amount of aggregated tau in the thalamus of TBI-inoculated novo tau deposition. mice(SupplementaryFig.4AandB).Thetauconformerthat A better understanding of the biology that underlies this accumulated in the brains of TBI-inoculated mice was par- process calls for tractable experimental models. Previous tially resistant to protease digestion, yielding a pronase-re- investigations have shown that TBI accelerates genetically sistant fragment of (cid:6)10kDa (Supplementary Fig. 4C), determined tau deposition in transgenic mice overexpres- consistent with tau prion formation (Sanders et al., 2014; sing mutant or wild-type human tau (Yoshiyama et al., Kaufman et al., 2016). 2005; Tran et al., 2011a, b; Ojo et al., 2016), and can lead to tau pathology in wild-type mice (Goldstein et al., 2012; Kondo et al., 2015; Meabon et al., 2016). A few Discussion studies reported formation of tau aggregates shortly after severe TBI in rats (Hawkins et al., 2013; Gerson et al., The present study found that a single moderate or severe 2016), with only one study showing persistent tau path- TBI in humans resulted in widespread tau pathology with a ology in wild-type mice (Kondo et al., 2015). Our data Downloaded from https://academic.oup.com/brain/advance-article-abstract/doi/10.1093/brain/awy193/5062642 by Open University Library (PER) user on 01 August 2018
Description: